Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (2024)

Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (1)

1. Introduction

This target article describes how medial temporal lobe–medial diencephalic interactions contribute to episodicmemory. Previous models have focussed on neural circuitrywithin the temporal lobe. This earlier focus on “temporallobe memory systems” arose from a number of assumptionsabout amnesia and models for amnesia. This target articlequestions these assumptions, and from this emerges a dif-ferent way of considering the neural substrates of episodicmemory. At the centre of this revision is the notion that thelink from the hippocampus to the mamillary bodies and an-terior thalamic nuclei, via the fornix, is critical for normalepisodic memory (Gaffan 1992a). Moreover, damage to thisaxis is responsible for the core deficits in anterograde am-nesia, as was originally proposed by Delay and Brion (1969).To understand why this view became unpopular and why ithas now reemerged, it is necessary to describe how a num-ber of past findings have been interpreted.

We will first describe the main features of the proposedmodel. This is followed by a section summarising relevantevidence from studies of amnesia and animal models ofamnesia, describing the way this evidence has often beeninterpreted. Section 4 examines certain assumptions un-derlying previous interpretations and shows that existingevidence can be reinterpreted in a different way. Section 5

BEHAVIORAL AND BRAIN SCIENCES (1999) 22, 425–489Printed in the United States of America

© 1999 Cambridge University Press 0140-525X/99 $12.50 425

Episodic memory, amnesia, and the hippocampal–anteriorthalamic axis

John P. AggletonSchool of Psychology, Cardiff University, Cardiff, CF1 3YG, [emailprotected] www.cf.ac.uk/uwc/psych/

Malcolm W. BrownDepartment of Anatomy, University of Bristol, Bristol, BS8 1TD, United Kingdom [emailprotected]

Abstract: By utilizing new information from both clinical and experimental (lesion, electrophysiological, and gene-activation) studieswith animals, the anatomy underlying anterograde amnesia has been reformulated. The distinction between temporal lobe and dien-cephalic amnesia is of limited value in that a common feature of anterograde amnesia is damage to part of an “extended hippocampalsystem” comprising the hippocampus, the fornix, the mamillary bodies, and the anterior thalamic nuclei. This view, which can be tracedback to Delay and Brion (1969), differs from other recent models in placing critical importance on the efferents from the hippocampusvia the fornix to the diencephalon. These are necessary for the encoding and, hence, the effective subsequent recall of episodic mem-ory. An additional feature of this hippocampal–anterior thalamic axis is the presence of projections back from the diencephalon to thetemporal cortex and hippocampus that also support episodic memory. In contrast, this hippocampal system is not required for tests ofitem recognition that primarily tax familiarity judgements. Familiarity judgements reflect an independent process that depends on a dis-tinct system involving the perirhinal cortex of the temporal lobe and the medial dorsal nucleus of the thalamus. In the large majority ofamnesic cases both the hippocampal–anterior thalamic and the perirhinal–medial dorsal thalamic systems are compromised, leading tosevere deficits in both recall and recognition.

Keywords: amnesia; fornix; hippocampus; memory; temporal cortex; thalamus

John Aggleton has been Professor of Cognitive Neu-roscience at Cardiff University, Wales, since 1994. Priorto that he worked at Durham University for 11 years.His research has focussed on the role of limbic brain re-gions in memory and emotion, both in humans and in arange of animal species. He is the author of over onehundred scientific publications, which include a paperon his favourite soap opera (The Archers) and a paperon dead cricketers! He has edited a book on the amyg-dala (The Amygdala, 1992, Wiley-Liss, New York) andis currently editing a follow-up book on the same brainstructure.

Malcolm Brown is Professor of Cognitive Neuro-science and Anatomy and a member of the MRC Cen-tre for Synaptic Plasticity at the University of Bristol,England. He took a first degree in theoretical physicsbefore a Ph.D. in neuroscience. His research centres onstudies of the neural basis of memory, particularlyrecognition memory. He is the author of over one hun-dred scientific publications in the area of neuroscience,including papers on hippocampal synaptic plasticity, im-printing, and benzodiazepine- and pregnancy-inducedhuman amnesia, as well as recognition memory.

Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (2)

describes new evidence from behavioural studies, humanclinical studies, single-unit recording studies, and brainactivation studies that provide further support for the pro-posed model of medial temporal–medial diencephalicinteractions. The final sections consider some of the impli-cations of the model. Throughout this review we havedrawn on evidence from studies of animals when the clini-cal data lack sufficient anatomical resolution. Great care isneeded when transposing results across species (Tulving &Markowitsch 1994), and biases can be introduced by the re-liance on one particular research method. For these reasonswe have tried, wherever possible, to present complemen-tary data from an array of techniques and from more thanone species.

2. Main features of the proposed model

1. The anatomical focus of the model concerns the con-nections between the hippocampus, the mamillary bodies,and the medial thalamus. (As a matter of terminology theterm hippocampus is used here to refer to the hippocampalfields CA1-4, the dentate gyrus, and the subicular complex.The mamillary bodies and the medial thalamus are bothmedial components of the diencephalon, which is com-posed of the thalamus, hypothalamus, epithalamus, andsubthalamus). The hippocampal efferents to the medial di-encephalon are regarded as vital for normal hippocampalactivity and are, hence, seen as functional extensions of thehippocampus (Fig. 1). The principal thalamic targets in thissystem are the anterior thalamic nuclei. These nuclei re-ceive direct hippocampal projections via the fornix, and in-direct hippocampal projections via the mamillary bodiesand the mamillothalamic tract. Other thalamic nuclei thatmay contribute to this system are the rostral midline nucleiand the lateral dorsal nucleus.

2. The system beyond the anterior thalamic nuclei be-comes more diffuse, but one component projects back fromthe anterior thalamic nuclei to the hippocampus and to ad-jacent temporal cortical regions. These return connections,which mainly use the cingulum bundle, form part of a cir-cuit that permits these diencephalic regions to influencetemporal lobe processing. Other important outputs are tothe cingulate and prefrontal cortices. A consequence of thediffuseness of the system beyond the anterior thalamic nu-clei is that damage in the relevant tracts or regions (e.g., cin-gulum bundle and prefrontal cortex) has a less profoundimpact upon episodic memory.

3. This extended hippocampal–diencephalic system iscritical for the efficient encoding and, hence, normal recallof new episodic information. As a consequence, damage tothe component structures can result in anterograde amne-sia; a common feature of all diencephalic and temporal lobeamnesias is the bilateral involvement of part of this “ex-tended hippocampal–diencephalic system” (i.e., the hip-pocampus, fornix, mamillary bodies, anterior thalamus,and, possibly, cingulum bundle). Furthermore, damage todifferent parts of this system produces similar memory im-pairments.

4. In contrast, this extended hippocampal–diencephalicsystem need not be vital for efficient recognition. This is be-cause recognition is regarded as being composed of at leasttwo independent processes (Mandler 1980), only one ofwhich is hippocampally dependent. Thus item recognition

occurs through recollection of the stimulus (“remember-ing”), a process that is hippocampally dependent, and bydetecting stimulus familiarity (“knowing”), which does notrequire the hippocampus. The latter process is especiallydependent on the perirhinal cortex in the temporal lobes.

5. Although the hippocampus and perirhinal cortex areanatomically linked, they are not necessarily dependent oneach other for their respective roles in the encoding ofepisodic information and familiarity-based recognition. Inparticular, both the hippocampus and perirhinal cortexhave independent links with other association cortical areas.

6. Although the hippocampus is closely linked to the an-terior thalamic nuclei, the perirhinal cortex is connectedwith the medial dorsal thalamic nucleus. These two paral-lel temporal–thalamic systems (Fig. 2) make qualitativelydifferent contributions to learning and memory. The en-torhinal cortex has attributes of both systems.

7. The traditional distinction between temporal lobe anddiencephalic amnesics is misleading; both groups havedamage to the same functional system. Nevertheless, thelarge majority of amnesics have additional pathology in cer-tain subcortical and cortical sites, and this can extend thenature of the memory loss so that it involves other aspectsof memory.

8. The proposed hippocampal–diencephalic system is re-quired for the encoding of episodic information, permittingthe information to be set in its spatial and temporal context(“episode”), so aiding subsequent retrieval and reducing in-terference (i.e., heightening discriminability).

9. The prefrontal cortex interacts with both of these sys-tems at a variety of levels, engaging efficient encodingstrategies that can then aid subsequent recall.

3. Studies of amnesia and animal models of amnesia, and their interpretation

3.1. Neuropathological evidence

Anterograde amnesia is typified by a failure to acquire or re-tain “episodic” information (Tulving 1983) that occurred af-

Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Figure 1. Schematic diagram of the principal pathways that al-low the encoding of episodic information and underlie recollec-tive aspects of recognition. The relative thickness of the lines in-dicates the putative importance of the various connections.

Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (3)

ter the onset of brain injury. Damage in more than one brainregion can result in anterograde amnesia, and neuropatho-logical studies have repeatedly highlighted the medial tem-poral lobes and the medial diencephalon. Identifying thecritical structures has, however, proved to be surprisinglydifficult. Although it is often assumed that temporal lobeamnesia is principally a consequence of damage to the hip-pocampus, it remains to be confirmed whether such dam-age is sufficient to induce amnesia. Relevant evidence hascome from amnesic cases with discrete unilateral hip-pocampal damage in one hemisphere combined with moreextensive temporal lobe damage in the other hemisphere(Penfield & Mathieson 1974; Woods et al. 1982). If bilateraldamage is required to induce amnesia, such cases stronglyimplicate the hippocampus. Other evidence has come fromamnesics with confirmed bilateral pathology restricted tothe hippocampus and the adjacent parahippocampal gyrusor uncus (DeJong et al. 1969; Glees & Griffiths 1952). Someof the most convincing evidence has, however, come fromthe discovery that hypoxia can produce both a permanentanterograde amnesia and discrete bilateral hippocampalpathology (Cummings et al. 1984; Rempel-Clower et al.1996; Victor & Agamonolis 1990; Zola-Morgan et al. 1986).There is, however, debate over whether these patients suf-fer “hidden” pathology (see sect. 4.1), so there is still a needto confirm whether discrete, bilateral hippocampal damagecan induce anterograde amnesia.

Diencephalic amnesia appears to be even more complex;neuropathological evidence has implicated several struc-tures, namely the mamillary bodies, the anterior thalamicnuclei, the medial dorsal thalamic nucleus, and the paratae-nial thalamic nucleus (Aggleton & Sahgal 1993; Clarke etal. 1994; Dusoir et al. 1990; Mair et al. 1979; Markowitsch1982; Parkin & Leng 1993). A number of adjacent tracts(the mamillothalamic tract and the internal medullary lam-ina) have also been implicated (Markowitsch 1988; Savageet al. 1997). Unfortunately, there are still no amnesic caseswith confirmed, circ*mscribed damage in just one of thesestructures. Furthermore, the proximity of these nuclei toone another, along with the likelihood of damage to fibresof passage and adjacent tracts, makes it extremely unlikelythat unambiguous cases will be discovered.

3.2. Testing recognition to assess anterograde amnesia in animals

The lack of unambiguous clinical evidence has led re-searchers to model anterograde amnesia in animals, andthus test unusually selective lesions. A prerequisite, how-ever, is the development of behavioural tasks that tax thesame classes of memory that are lost in amnesia. This needis underscored by the many examples of spared learningabilities in amnesia, which include classical conditioning, vi-suomotor skill tasks, and priming (Parkin & Leng 1993;Schacter et al. 1993; Weiskrantz 1990).

Studies with animals have, in fact, relied very heavily onbehavioural tests of recognition. This is because a loss ofrecognition is a striking feature of anterograde amnesia andhas been regarded as a core deficit (Haist et al. 1992; Parkin& Leng 1993; Squire & Knowlton, 1995; Squire & Shima-mura 1986). Furthermore, the use of forced-choice designsmakes it relatively easy to test animals. In contrast, examin-ing the recall of episodic information by animals has provedmuch more problematic. As a consequence the favoured testof recognition, delayed nonmatching-to-sample (DNMS),has become the litmus test for models of anterograde am-nesia.

In DNMS the animal is first shown a sample stimulus(often a “junk” object). After a delay the animal is shownthat same object along with a novel or less familiar object.Selection of the novel object (nonmatching) is rewarded inDNMS, whereas in delayed matching-to-sample (DMS)selection of the familiar object is rewarded. In the “trial-unique” version of DNMS and DMS both the novel andthe familiar objects are then discarded so that new itemscan be used for the next sample and the next novel alter-native. Early studies using the trial-unique version of theDNMS task with monkeys soon confirmed that, as in peo-ple, large medial temporal lesions (Mishkin 1978; Zola-Morgan et al. 1982) and large medial diencephalic lesions(Aggleton & Mishkin 1983a; 1983b) produce very severerecognition deficits. The apparent validity of these recog-nition tests was further strengthened by studies showingthat people with either temporal lobe or diencephalic am-nesia are markedly impaired on forced-choice recognitiontasks designed to be analogous to the DNMS and DMStasks given to monkeys (Aggleton et al. 1988; Squire et al.1988). It is therefore not surprising that these tasks havebeen used to assess the effects of selective bilateral dam-age in a number of key sites.

3.3. Testing the contribution of the fornix

One site of special interest has been the fornix. Among itscomponents this tract contains the cholinergic innervationto the hippocampus from the medial septum, as well as hip-pocampal efferents to the diencephalon, striatum, and pre-frontal cortex. These efferents include dense projections tothe mamillary bodies and the anterior thalamic nuclei,which in monkeys are conveyed solely in the fornix (Aggle-ton et al. 1986a; Aggleton & Saunders 1997). As a conse-quence, the fornix forms a vital bridge between medialtemporal and medial diencephalic regions implicated inanterograde amnesia.

Although the first study to use the DMS task to assessthe effects of fornix transection reported an impairment(Gaffan 1974), a series of later DNMS and DMS studiesfound that fornix transection produced little or no recog-

Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Figure 2. Schematic diagram of the principal pathways underly-ing the detection of item familiarity. The relative thickness of thelines indicates the putative importance of the various connections.

Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (4)

nition deficit in monkeys (Bachevalier et al. 1985a; 1985b;Gaffan et al. 1984; Zola-Morgan et al. 1989a) and sparedDNMS performance in rats (Aggleton et al. 1990; Roth-blat & Kromer 1991; Shaw & Aggleton 1993). Similarly,fornix lesions were found to have no effect on spontaneoustests of object recognition (Ennaceur & Aggleton 1994;Ennaceur et al. 1996; 1997). Indeed, in one study, mon-keys with fornix lesions eventually performed the DNMStask significantly better than control animals (Zola-Mor-gan et al. 1989a); in another study, monkeys with fornix le-sions showed enhanced preference for perceptual novelty(Zola-Morgan et al. 1983). Similarly, rats with fornix le-sions were able to acquire a DNMS task more rapidly thancontrol animals (Shaw & Aggleton 1993). The immediateconclusion was that fornix damage did not disrupt recog-nition and, hence, was not sufficient to induce antero-grade amnesia (Squire & Zola-Morgan 1991; Zola-Mor-gan et al. 1989a).

This conclusion was consistent with a review of mem-ory loss and fornix damage in humans (Garcia-Bengochea& Friedman 1987). Among 142 patients thought to havebilateral fornicotomy for the treatment of epilepsy, nonehad persistent memory problems. A further 13 cases withfornix damage associated with third ventricle colloid cystswere also considered. Four of them had persistent mem-ory loss (Carmel 1985; Garcia-Bengochea & Friedman1987; Sweet et al. 1959), but the likelihood that the cystshad caused additional diencephalic damage weakened thevalue of these individual cases. Similar constraints can beapplied to other patients in whom surgery for cysts or tu-mours resulted in both fornix damage and memory loss(Cameron & Archibold 1981; Geffen et al. 1980; Heilman& Sypert 1977; Tucker et al. 1988). Additional problemsof interpretation arise with those patients in whom thehippocampal commissures as well as the fornix were cutor disconnected (Heilman & Sypert 1977; Tucker et al.1988). Although Hassler (1962) described a woman inwhom stereotaxic coagulation of the fornices led to an am-nesic state, the woman survived only a few days aftersurgery, severely limiting assessment. Taken together, thecases with presumed fornix damage and apparently un-changed memory (Garcia-Bengochea & Friedman 1987;see also Woolsey & Nelson 1975) far outnumbered the fewsingle case studies in which fornix damage appeared to beassociated with amnesia.

Other evidence has come from studies on the mamillarybodies, which the fornix innervates. It had long been ap-preciated that mamillary body degeneration is a consistentfeature of Korsakoff ’s disease and that it might contributeto the anterograde amnesia. A comprehensive neuropatho-logical study by Victor and his co-workers (1971) concluded,however, that thalamic damage (and in particular damageto the medial dorsal thalamic nucleus) was a better predic-tor of the memory loss. Consistent with this was the findingthat mamillary body lesions in animals did not disruptDNMS performance (Aggleton & Mishkin 1985; Aggletonet al. 1990; Zola-Morgan et al. 1989a), whereas lesions inthe medial dorsal thalamic region impaired both the acqui-sition and performance of the DMS and DNMS tasks (Ag-gleton & Mishkin 1983b; Hunt & Aggleton, 1991; Mumbyet al. 1993; Parker et al. 1997; Zola-Morgan & Squire1985a). Insofar as these findings failed to support a role forthe mamillary bodies in anterograde amnesia they accordedwith similar evidence for the fornix.

3.4. Comparing the effects of lesions in the hippocampus and lesions in adjacent cortical regions

Studies with animals also provided the opportunity to ex-amine the effects of increasingly selective lesions within thetemporal lobe. Aspiration lesions of the hippocampus con-sistently produced a modest, but significant, DNMS deficit(Murray & Mishkin 1986; Zola-Morgan & Squire 1986;Zola-Morgan et al. 1989a; 1993), supporting the contribu-tion of this structure to amnesia. More discrete temporallobe lesions also revealed that the amygdala was not critical(O’Boyle et al. 1993; Zola-Morgan et al. 1989b). Muchmore surprising was the discovery that the cortex immedi-ately lateral to the amygdala and hippocampus is of vital im-portance for DNMS performance. Thus lesions involvingthe rhinal region (comprising the perirhinal and entorhinalcortices) or more extensive lesions involving the rhinal re-gion and the parahippocampal gyrus produce extremely se-vere and persistent DNMS deficits (Meunier et al. 1993;Mumby & Pinel 1994; Murray 1996; Murray & Mishkin1986; Suzuki et al. 1993; Zola-Morgan et al. 1989b). Morediscrete lesions within the rhinal region have since high-lighted the special importance of the perirhinal cortex (Me-unier et al. 1993; 1996). In contrast, entorhinal lesions pro-duce only a very mild or transient impairment (Leonard etal. 1995; Meunier et al. 1993). Similarly, removal of para-hippocampal cortex does not contribute to the DNMSdeficit (Meunier et al. 1996; Ramus et al. 1994). These find-ings, along with those from single unit recording studies(see sect. 5.3), have forced a fundamental reappraisal of thecontribution of individual temporal lobe structures tomemory (Murray 1996).

Anatomical studies have shown that the perirhinal andparahippocampal cortices project densely upon the en-torhinal cortex, and, in fact, they provide nearly two thirdsof the cortical inputs to the entorhinal cortex (Insausti et al.1987; Suzuki & Amaral 1994). The entorhinal cortex itselfis the major source of afferents to the hippocampus. As aconsequence these indirect connections, along with a num-ber of direct perirhinal–hippocampal projections (Suzuki& Amaral 1990; Witter & Amaral 1991), ensure that theperirhinal and parahippocampal cortical areas are a majorsource of hippocampal inputs. Additionally, the hippocam-pus has extensive reciprocal connections with the ento-rhinal, perirhinal, and parahippocampal cortices (Suzuki1996a; Suzuki & Amaral 1994; Witter et al. 1989). These in-terconnections help to reinforce the view that the hip-pocampus along with the perirhinal, parahippocampal, andentorhinal cortices function as a closely integrated unit subserving aspects of memory, including recognition(Squire & Zola-Morgan 1991). It should be emphasized,however, that the DNMS deficit following perirhinal re-moval is not due simply to a disconnection of hippocampalinputs; the severity of this deficit is considerably greaterthan that found after hippocampectomy (Meunier et al.1996; Murray 1996; Zola-Morgan et al. 1993). Thus theperirhinal region must have independent mnemonic capa-bilities.

3.5. The temporal lobes and episodic memory: Current models

These new findings have been integrated with growing clin-ical evidence suggesting that hippocampal damage is suffi-

Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (5)

cient to induce amnesia, and they have led to a number ofinfluential models of temporal lobe function. A commonfeature of these models is that the perirhinal, entorhinal,and parahippocampal cortices, along with the hippocam-pus, form the key components of a closely integrated tem-poral lobe memory system. This system is reciprocal; theplentiful projections back from the hippocampus to the en-torhinal cortex and the perirhinal/parahippocampal cor-tices are seen as instrumental in setting up long-term rep-resentations (i.e., memories) in neocortex (Eichenbaum etal. 1994; Squire & Knowlton 1995; Squire & Zola-Morgan1991; Suzuki 1996a; 1996b). One important consequenceof the reciprocal nature of these interactions is that the pro-posed systems are largely self-contained within the tempo-ral lobes. This has served to distance other structures suchas the fornix, anterior thalamic nuclei, and mamillary bod-ies and implies that the involvement of these regions in di-encephalic amnesia will reflect a qualitatively different syn-drome.

In one of the most often cited models (Squire & Zola-Morgan 1991) the parahippocampal, perirhinal, and en-torhinal cortices form a reciprocal network with the hip-pocampus to create a “medial temporal memory system.”This system is crucial for the rapid acquisition of new in-formation about facts and events, which then gradually be-comes consolidated in the neocortex and eventually be-comes independent of the hippocampus (Squire & Alvarez1995; Squire & Zola-Morgan 1991). The role of the hip-pocampus is to bind together different components of thememory. Later expansions of this model have acknowl-edged some linkage with medial thalamic regions, but noapparent role has been provided for hippocampal outputsto the mamillary bodies and anterior thalamus via the fornix(Squire & Knowlton 1995; Zola-Morgan & Squire 1993).This exclusion stems from the failure of either fornix ormamillary body lesions to disrupt DNMS performance, andthe assumption that there is a close relationship betweenrecognition and recall (Haist et al. 1992; Squire & Knowl-ton 1995). It is therefore presumed that these connectionsare not necessary for the recall of episodic (declarative)memory.

A related model (Eichenbaum et al. 1994) proposes a“hippocampal memory system” formed by the hippocam-pus and the “parahippocampal region” (comprising the en-torhinal, perirhinal, and parahippocampal cortices). Thishippocampal memory system contributes both to the tem-porary maintenance of memories and to the processing ofa particular type of memory representation. In particular,the parahippocampal region supports intermediate-termstorage of individual items, whereas the hippocampal for-mation is concerned with organizing memories accordingto relevant relationships between items, including spatialrelationships (Eichenbaum et al. 1994). This “hippocampalmemory system” is seen to be critical for episodic memory,so dysfunction of the system can lead to anterograde am-nesia.

4. A critical examination of key assumptionsunderlying these models of the neuralsubstrates of recognition and recall

In developing these models of temporal lobe involvementin episodic memory, a number of different assumptionshave proved very influential. These are: (1) that hippocam-

pal damage is sufficient to impair recognition; (2) that test-ing recognition (i.e., using DNMS or DMS) provides a validassay for the core deficits in anterograde amnesia; and (3)that hippocampal function is critically dependent on affer-ents from the perirhinal region. There are now, however,good grounds for questioning all these assumptions, and indoing so a quite different view of temporal lobe–dien-cephalic interactions emerges.

4.1. Is hippocampal damage sufficient to impair recognition?

The importance of the perirhinal cortex highlights the needto reexamine the effects upon DNMS of hippocampectomyusing techniques that spare rhinal regions. Interestingly, le-sions of the rat hippocampus are possible via a dorsal routethat avoids the rhinal cortices. Hippocampectomies per-formed in this manner have little or no effect on DNMStests (Aggleton et al. 1986b; Duva et al. 1997; Mumby et al.1996; Steele & Rawlins 1993). Another approach has beento induce ischaemic lesions, which can produce seeminglyselective pathology in the hippocampus. Such lesions areaccompanied by persistent DNMS deficits in both monkeys(Bachevalier & Mishkin 1989; Zola-Morgan et al. 1992) andrats (Wood & Phillips 1991; Wood et al. 1993). A problemis that the neural dysfunction caused by the ischaemia maybe much more extensive than the region of gross pathology(Bachevalier & Meunier 1996; Gaffan & Lim 1991; Nunn& Hodges 1994). Occlusion of the posterior cerebral artery,for example, results in a DNMS deficit greater than that ex-pected from the grossly apparent brain damage (Bacheva-lier & Mishkin 1989). Similarly, discrete ischaemic hip-pocampal lesions in rats produce marked DNMS deficits(Wood & Phillips 1991; Wood et al. 1993), yet neurotoxiclesions intended to match the extent of the apparent is-chaemic damage have no effect on DNMS performance(Duva et al. 1997). Extensive conventional hippocampal le-sions not only spare DNMS performance (Mumby et al.1996; Wood et al. 1993) but, remarkably, can attenuate theeffects of ischaemia (Mumby et al. 1996). This result notonly highlights the mismatch between the observed pathol-ogy and the functional pathology, but also indicates that theischaemia resulted in extrahippocampal dysfunctions sub-sequent to the initial hippocampal pathology (Mumby et al.1996). Finally, a recent positron emission tomography(PET) study (Markowitsch et al. 1997) has highlighted thelimitation of relying on magnetic resonance imaging (MRI)to uncover functional damage in cases of anoxia. This is sobecause PET revealed widespread regions of hypoactivityin an amnesic patient that could not be predicted from MRIscans (Markowitsch et al. 1997).

The possibility that ischaemia can lead to more extensivedysfunction than that apparent by standard pathologicalmeasures has, however, been disputed (Squire & Zola1996). It has been argued that the DNMS deficits follow-ing posterior artery occlusion (Bachevalier & Mishkin1989) were exagerated by reference to unusually high scor-ing controls, and that monkeys with hippocampal lesionsproduced by sterotaxy (Alvarez et al. 1995) perform at alevel comparable to those with ischaemic lesions (Squire &Zola 1996). The first of these points requires additional con-trol data to resolve. The second criticism is, however, po-tentially misleading insofar as the comparison included datafrom other tests, that is, those not testing recognition. When

Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (6)

the data are taken only from comparable DNMS tests (de-lays 15 sec to 10 min), it is found that three of the four is-chaemic monkeys performed at more than 2.7 standard deviations below the mean score of the stereotaxic hip-pocampectomy cases (Squire & Zola 1996), whereas thecontrol animals for the two studies performed at equivalentlevels. The DNMS scores of the ischaemic animals were,however, comparable to those of monkeys with hippocam-pal lesions made by techniques that also damage adjacentperirhinal cortex (Bachevalier & Meunier 1996). Althoughthe balance of evidence indicates that anoxia can producemore extensive recognition dysfunction than that predictedfrom an assessment with standard histological methods, itis also clear that this key issue requires further examination(Nunn & Hodges 1994).

For these reasons it is preferable to focus on studies thathave examined selective, stereotaxic lesions within the hip-pocampus. In one of the few such studies, radio frequencylesions were placed bilaterally within the hippocampus (Al-varez et al. 1995). The lesions did not disrupt DNMS per-formance significantly until there was a delay of 10 min be-tween sample presentation and test (Alvarez et al. 1995). Ina number of other stereotaxic studies a neurotoxin (ibotenicacid) was injected into the monkey hippocampus, sparingfibres of passage and adjacent fibre tracts (Beason-Held etal. 1993; Murray 1996; Murray & Mishkin 1996; O’Boyle etal. 1993). Although the first of these studies reportedDNMS deficits (Beason-Held et al. 1993), the remainingstudies showed normal levels of performance even thoughthe hippocampal fields CA1-4, along with the amygdala,were destroyed. In one of these studies the retention inter-val was extended to 40 min, but, unlike the case in an ear-lier study that had found an impairment with such delays(Alvarez et al. 1995), the animals were not removed fromthe apparatus during testing (Murray & Mishkin 1996).These animals showed no DNMS impairment (Murray &Mishkin 1996). It therefore appears that selective hip-pocampal lesions can often spare DNMS performance, al-though for some of these reports the histology remains tobe published in a comprehensive form. It is also still neces-sary to examine the performance of monkeys with neuro-toxic hippocampal lesions that include the subiculum.

The effects of these selective hippocampal lesions nowclosely correspond to the effects of fornix lesions onDNMS; that is, they typically have little or no effect. This isnoteworthy insofar as fornix transection often mimics hip-pocampal dysfunction, most obviously for tests of spatialmemory (Aggleton et al. 1986b; 1992; 1995a; Barnes 1988;Olton et al. 1982; Saunders & Weiskrantz 1989). It had ap-peared that DNMS presented an important exception tothis general rule, but these recent stereotaxic studies showthat the effects of hippocampectomy and fornicotomy arein accordance for DNMS as well.

It has been argued that the lack of a clear hippocampallesion deficit in DNMS tasks might be due to the trainingprior to surgery, which can then mask any subsequent le-sion deficit (Alvarez et al. 1995). Because learning the non-matching rule per se cannot help the animal solve any indi-vidual problem, it is difficult to see how greater trainingcould obscure a deficit unless there are ceiling effects. Nev-ertheless, this claim has led to a number of studies of spon-taneous recognition based upon preferential viewing ofnovel visual stimuli. With such tasks it has been reportedthat lesions not only of the perirhinal cortex (Clark et al.

1997) but also of the hippocampus (Clark et al. 1996) candisrupt performance at delays as short as 10 sec. Such tasksoften use complex visual stimuli, and previous lesion stud-ies have demonstrated that hippocampal system lesions im-pair the ability to use “scenes” that are composed of an ar-ray of different features (Gaffan 1994b). Thus the abnormalbehaviour following hippocampal lesions may reflect a fail-ure to associate the component elements. It is also the casethat spontaneous tests of recognition are more prone to dis-ruption by other factors such as hyperactivity or increaseddistractability. In an ingenious variant on such tasks, Honeyet al. (1998) showed that neurotoxic lesions of the rat hip-pocampus do not affect orientation and subsequent habit-uation to novel visual and auditory stimuli. It was, however,found that animals with these lesions failed to orient whenfamiliar combinations of these cross-modal stimuli were rearranged (mismatched). Thus the hippocampal lesionsspared novelty detection per se, but the mismatch condi-tion revealed a failure to detect or respond to changes in thelearned association between the pairs of cross-modal stim-uli (Honey et al. 1998).

The evidence showing that extensive, but selective, hip-pocampal damage can often spare DNMS raises the ques-tion of whether there is comparable, clinical evidence. Onesource of potential evidence comes from amnesic peoplewith hypoxic damage, who are very likely to suffer hip-pocampal damage (but may also suffer “hidden pathology”;see above). Such amnesics can show apparently normalrecognition performance in spite of impaired recall (Volpeet al. 1986). Consistent with this are the findings from a re-cent survey of amnesics (Aggleton & Shaw 1996), whichanalysed results from a standard test of recognition, theWarrington Recognition Memory Test (RMT). The RMT(Warrington 1984) consists of two subtests, one testing facerecognition, the other testing word recognition. From asample of 112 amnesics placed in 11 distinct pathologicalgroupings, it was found that three groups of amnesics didnot differ from their age-matched norms (Aggleton & Shaw1996). One of these groups comprised patients with re-stricted hippocampal damage following hypoxia, anothercontained patients with fornix damage (Aggleton & Shaw1996; see also McMackin et al. 1995), and a third group hadselective diencephalic damage. These groups not onlyfailed to differ from the normal subjects, they also per-formed significantly better than some of the other amnesicgroups.

Although these RMT results closely match the findingsfor DNMS performance by nonhuman primates (i.e., littleor no effect following hippocampal or fornix damage), thereare a number of important constraints. The first is that theRMT data come from just one test of recognition and, as wasindicated in section 2, it is to be predicted that hippocam-pal damage will have more impact on some tests of recogni-tion than on others. The second is that cases with anoxicdamage may have variable covert pathology. Both of theseconsiderations apply to a recent review of recognition fol-lowing anoxic hippocampal damage in humans (Reed &Squire 1997), which convincingly shows that this aetiologycan lead to recognition deficits across a wide range of tests.Even so, compared to test norms, performance on the stan-dard version of the RMT is apparently preserved in some ofthese patients and deficient in others (Reed & Squire 1997).A related case concerns an amnesic who performed verypoorly on the RMT, even though MRI studies indicated that

Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (7)

the subject had circ*mscribed lesions confined to areas CA1and CA2 (Kartsounis et al. 1995). This same person did,however, show very severe retrograde amnesia suggestive ofmore extensive cortical damage (Kapur et al. 1992; Zola-Morgan et al. 1986). In view of the fact that the amnesiaarose from repeated ischaemic episodes, this apparent dis-crepancy might be related to the issue of hidden pathology(Bachevalier & Mishkin 1989; Mumby et al. 1996).

Other relevant evidence comes from a recent study of104 epileptic patients who had been tested on the RMT andhad unilateral temporal lobe pathology confirmed by MRI(Baxendale 1997). Patients with combined cortical and hip-pocampal damage performed significantly worse than thosewith selective hippocampal damage. Furthermore, the groupmean score of those with selective left hippocampal dam-age on the test of word recognition (the subtest on whichthey should be most impaired) was in the normal range, aswas the group mean score for those with right hippocampaldamage on the face recognition test (Baxendale 1997). Theconclusion, that unilateral hippocampal damage had noconsistent effect on this test of recognition, reinforced aprevious study showing that hippocampal sclerosis had noapparent effect on either of the RMT subtests (tested pre-operatively), although deficits on delayed recall were found(Miller et al. 1993). These conclusions appear to contrastwith those of a recent study using event-related potentials,which showed a loss of reactivity to novel stimuli in five sub-jects with combined unilateral pathology in the hippocam-pus (Knight 1996). In all five cases, however, the pathologyinvolved the parahippocampal gyrus and the entorhinalcortex (Knight 1996), so the resulting deficit could be pre-dicted.

Even if it is accepted that hippocampectomy can inducea subtle DNMS deficit (Alvarez et al. 1995; but see Murray& Mishkin 1996), this is apparent only after very lengthy de-lays, for example, 10 min. This contrasts with amnesic sub-jects who are typically impaired on DNMS and DMS tasksafter delays of only 40 sec between sample presentation andtest (Aggleton et al. 1988; Holdstock et al. 1995; Squire etal. 1988). Furthermore, amnesic subjects show significantlyfaster rates of forgetting over these relatively short delays(Holdstock et al. 1995), whereas monkeys with selectivehippocampectomy do not. These differences suggest thathippocampal damage in monkeys is not sufficient to repro-duce the recognition deficit typically found in amnesia.

A final factor concerns the type of stimulus being tested.Studies with rats have shown that both fornix transectionand hippocampectomy can disrupt recognition when large,relatively featureless stimuli (test boxes) are used instead oftrial-unique discrete objects (Cassaday & Rawlins 1995;Rawlins et al. 1993). This impairment is most evident whenthe plain boxes are used repeatedly within a session, that is,are not trial unique (Rawlins et al. 1993), but deficits arealso observed when discrete objects are placed in theselarge test boxes (Cassaday & Rawlins 1997). A plausible ex-planation of these results is that the animal encodes thelarge box or the stimuli inside the large box as part of a spa-tial (scenic) array rather than as a discrete stimulus (Cassa-day & Rawlins 1995), thus rendering it sensitive to hip-pocampal dysfunction. When the boxes are small they areencoded as objects, and no deficit is seen (Cassaday &Rawlins 1997). Similarly, studies with monkeys have shownthat fornix lesions can reliably disrupt the recognition of“scenes” in which common elements are repeated but oc-

cur in different spatial configurations (Gaffan 1991; 1992b;1994b). These impairments can be directly related to thewidely accepted view that the hippocampus is vital for theefficient encoding of allocentric space (O’Keefe & Nadel1978). The importance of stimulus type is further empha-sized by the spontaneous orientation task used by Honey etal. (1998) and by recent activation studies (see sect. 5.4).

4.2. Does testing recognition provide a valid assay for anterograde amnesia?

A closely related debate concerns whether tests such asDNMS and DMS are a valid assay for amnesia. One view isthat recognition is an integral part of declarative memory(Haist et al. 1992; Knowlton & Squire 1995) insofar as peo-ple can subjectively evaluate their memory and either re-trieve items (recall) or make judgements regarding theirprevious occurrence (recognition). This model tightly linksthe two processes and so predicts that anterograde amne-sia will impair both recall and recognition and that thedeficits will be related. Alternate views hold that recogni-tion and recall depend, in part, on different processes. Onesuch view is that recognition benefits from an additionalcomponent of processing that is based on “perceptual flu-ency” or “feelings of familiarity” (Gardiner 1988; Gardiner& Parkin 1990; Jacoby 1991; Mandler 1980; see sect. 6).This process is regarded as being additive to and separatefrom the explicit memory of an event (Mandler 1980) andcorresponds to feelings of “knowing” that something is fa-miliar rather than “remembering” (i.e., recalling) its previ-ous occurrence (Gardiner 1988; Gardiner & Parkin 1990).As a consequence it may be predicted that a loss of episodicmemory need not be accompanied by a comparable loss ofrecognition.

In fact, a number of reports have described individualamnesic cases (Dusoir et al. 1990; Gaffan et al. 1991; Han-ley et al. 1994; Parkin & Hunkin 1993; Parkin et al. 1993)or even groups of amnesics (McMackin et al. 1995; Volpeet al. 1986) with relatively well-preserved recognition. Forexample, a group of subjects with bilateral fornix damagefollowing third ventricular cysts (McMackin et al. 1995) wasable to perform the RMT tasks within normal limits, eventhough they were clearly impaired on tests of episodicmemory. Individual cases of interest include a man who suf-fered bilateral traumatic injury to the mamillary body re-gion (Dusoir et al. 1990), in whom PET studies revealed ad-ditional hypoactivity in the left hippocampus (Kapur 1995).A clear and persistent anterograde amnesia developed, yethe performed well within normal limits on a series of recog-nition tests including the RMT (Dusoir et al. 1990). He alsoperformed very well on a DMS task using single abstractpatterns (Holdstock et al. 1995; Fig. 3). This is of interestin that the task avoided ceiling effects even though it useda DMS procedure to assess the retention and recognitionof single stimuli. Furthermore, all of the other amnesic sub-jects tested on the same DMS task were markedly im-paired, even though their delayed recall deficits (as mea-sured by the Wechsler Memory Scale Revised; WMSr)were comparable to those of the mamillary body case(Holdstock et al. 1995).

Other individual cases include a person who had suffereda hypothalamic tumour close to the mamillary bodies andwho displayed a severe anterograde amnesia (Parkin &Hunkin 1993). This patient achieved scores in the 83rd

Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (8)

(words) and 94th (faces) percentiles on the two RMT sub-tests (in stark contrast to a score of 56 on the WMSr De-layed Recall Index). Of similar interest was a young womanwho displayed Wernicke’s encephalopathy following a rela-tively brief history of alcoholism (Parkin et al. 1993). Sheshowed a chronic, profound impairment on tests of recallbut her recognition memory was remarkably well preservedacross a variety of tests. These included the RMT on whichshe scored in the 75th percentile for both words and faces(Parkin et al. 1993), as well as showing normal performanceon a more difficult RMT variant in which the face stimuliwere presented upside down.

These examples of spared recognition do not arise sim-ply because tests of recognition are easier to perform thantests of recall, nor because the individual patients sufferedonly from very mild amnesic syndromes. The former can beexcluded because a number of studies have taken specialcare to preclude ceiling effects (Hanley et al. 1994; Hold-stock et al. 1995; Parkin et al. 1993; Volpe et al. 1986). Sim-ilarly, differences in the severity of the anterograde amne-sia can also be discounted because performance on othermemory tests has been carefully documented in individualpatients with spared recognition (Aggleton & Shaw 1996;Dusoir et al. 1990; Hanley et al. 1994; Parkin & Hunkin1993; Parkin et al. 1993).

4.3. Is hippocampal function dependent on afferentsfrom the perirhinal region?

One of the more surprising aspects of the current model isthe supposed extent to which some hippocampal functionsare independent of their perirhinal inputs. This appearssurprising for at least two reasons. First, there are many di-rect and indirect connections between the two regions. In-

deed, the perirhinal and parahippocampal cortices com-bined provide approximately two-thirds of the inputs to theentorhinal cortex (Insausti & Amaral 1987; Suzuki 1996a;Suzuki & Amaral 1994), which is the cortical gateway to thehippocampus. In addition, the hippocampus projects di-rectly upon the perirhinal cortex and entorhinal cortex; thelatter projects, in turn, to the perirhinal and parahippo-campal cortices (Saunders & Rosene 1988; Suzuki 1996;Suzuki & Amaral 1994). Second, the type of informationthat appears to be present in the perirhinal cortex (see sect.5.3) could provide the elemental fragments upon which anepisodic “memory system” might operate (Brown 1990;Eichenbaum et al. 1994; Gaffan & Parker 1996; Squire &Zola-Morgan 1991). Not surprisingly, both classes of evi-dence are featured in previous models of medial temporallobe function (Eichenbaum et al. 1994; Squire & Zola-Mor-gan 1991). As a consequence these models predict thatperirhinal damage should disconnect the hippo-campusand so mimic the effects of hippocampal removal. For thisreason it should not be possible to produce a double disso-ciation between these two regions.

Recent studies on the perirhinal cortex do, however, sug-gest that this cortical region has a relationship with the hip-pocampus different from that proposed in previous models(Eichenbaum et al. 1994; Squire & Zola-Morgan 1991).Most striking is lesion evidence showing that functions ofthe hippocampus and the perirhinal cortex can be doublydissociated from one another. In one study fornix lesions inmonkeys produced severe deficits on a spatial discrimina-tion and reversal task (Gaffan 1994a). This accords withprevious studies showing the sensitivity of this spatial taskto lesions in the hippocampus and mamillary bodies as wellas in the fornix (Aggleton & Mishkin 1985; Jones & Mishkin1972; Mahut 1971; 1972). In contrast, perirhinal lesions hadno apparent effect on the same spatial task (Gaffan 1994a).The same study also tested recognition for visual scenes andfound that on this task the perirhinal lesions produced a se-vere deficit, whereas the fornix lesions resulted in a muchmilder impairment (Gaffan 1994a). This double dissocia-tion shows that the perirhinal cortex is not a critical way sta-tion for all hippocampal inputs and suggests that themnemonic contributions of the two regions can differ sub-stantially.

Evidence for a similar double dissociation has recentlybeen uncovered in rats (Ennaceur & Aggleton 1997; En-naceur et al. 1996). Although fornix lesions severely im-paired tests of spatial working memory (T-maze alternation,radial-arm maze nonmatching, and delayed nonmatching-to-position in a Skinner box), cytotoxic perirhinal lesionshad no apparent effect (Table 1). In contrast, only theperirhinal lesions disrupted a test of object recognition (En-naceur et al. 1996). Although they were not tested simulta-neously, lesions of the hippocampus and the anterior thala-mic nuclei have been shown in other experimentsconsistently to disrupt these same spatial tasks (Aggleton etal. 1986b; 1995a; 1996), but have no apparent effect on ob-ject recognition (Aggleton et al. 1986b; 1995a). Conversely,large neurotoxic lesions including the perirhinal cortex,postrhinal cortex and area TE impaired object recognitionbut had no apparent effect on T-maze alternation (Aggle-ton et al. 1997). This last result is especially surprising inthat the postrhinal cortex offers an alternative route for spa-tial information to reach the hippocampus (Naber et al.1997). Additional evidence for a double dissociation be-

Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Figure 3. Performance of an amnesic subject with bilateralmamillary body damage (BJJ) showing spared performance com-pared to normal controls and a group of mixed amnesics on a de-layed matching-to-sample task using visual patterns (data fromHoldstock et al. 1995).

Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (9)

tween hippocampal and perirhinal functioning has comefrom a series of c-fos activation studies (see sect. 5.5). Theseindicate that exposure to novel visual stimuli increases neu-ronal activity in the perirhinal cortices but not in the hip-pocampus (Zhu et al. 1995b; 1996). In contrast, exposure toa novel environment can raise hippocampal activity but notperirhinal activity (Zhu et al. 1997).

These results lead to the prediction that selective dam-age to the human hom*ologue of the perirhinal cortex willimpair some memory functions, including recognition, butneed not produce a full amnesia. Kapur et al. (1994) de-scribed a person with extensive bilateral damage to tempo-ral neocortex combined with apparent sparing of the hip-pocampus and amygdala. This pathology spared parts of theentorhinal cortex but appeared to invade much of the peri-rhinal cortex (Kapur et al. 1994). In spite of some everydaymemory difficulties, the patient did not suffer from antero-grade amnesia (e.g., his WMSr delayed memory quotientwas 99). He did, however, show a retrograde amnesia and a“semantic” memory loss. Furthermore, his recognition per-formance on the faces test of the RMT was severely im-paired although word recognition appeared normal, sug-gesting a hemispheric difference in the extent of pathology.

Other evidence concerns a woman who suffered bilateraldamage to the rostral temporal cortex, while subcorticalregions appeared intact (Kapur et al. 1992; see also Marko-witsch et al. 1993). She displayed a severe retrograde am-nesia but only a very mild loss of new learning. Of specialinterest was the finding that her performance on the facessubtest of the RMT was impaired, yet on recall tests of vi-sual nonverbal memory her performance was excellent (Ka-pur et al. 1992). Damage to the parahippocampal gyrus wasthe best predictor of the recognition memory deficit, but itwas not associated with anterograde amnesia. Other rele-vant evidence comes from a description of five cases with aprofound loss of semantic information associated with focaltemporal lobe atrophy (Hodges et al. 1992). One of the keyfeatures of these subjects with “semantic dementia” was therelative preservation of episodic memory (Hodges et al.1992). Another striking feature of semantic dementia is thefinding that the loss of past autobiographical (episodic) in-formation can show a reverse Ribot effect, that is, relativesparing of recent memories (Graham & Hodges 1997). This

is the opposite of what is seen in amnesic syndromes and inAlzheimer’s disease (Graham & Hodges 1997), and hencepoints to dissociable functions played by the hippocampalsystem and its neighbouring cortices.

Further evidence for this dissociation comes from a se-ries of three adolescents who suffered bilateral hippocam-pal pathology at birth or aged 4 or 9 years (Vargha-Khademet al. 1997). All three show a clear anterograde amnesia af-fecting episodic memory, yet, remarkably, they have at-tained levels of language competence and factual knowl-edge that are within the low to average range. Not only dothey show an apparent dissociation between semantic andepisodic memory, they also show evidence of relativelywell-preserved recognition (forced-choice) memory in theface of deficient spatial and temporal memory (Vargha-Khadem et al. 1997). Although potentially important fac-tors such as developmental reorganisation must be consid-ered, these cases provide further evidence against the viewof an interdependent relationship between the hippocam-pus and the temporal (perirhinal) cortices.

5. Recent support for the proposed model of hippocampal–diencephalic interactions

5.1. Behavioural evidence from lesion studies inanimals: Spatial memory and scene memory

In recent years Aggleton and his collaborators have sys-tematically examined the involvement of various limbicbrain sites in the performance of tasks dependent on nor-mal hippocampal function (Table 1). These experiments,which have used rats, have focussed on tests of allocentricspatial memory (O’Keefe & Nadel 1978). Studies usingforced spatial alternation in a T maze have revealed thatnormal performance depends on the integrity of the ante-rior thalamic nuclei, the mamillary bodies, and the cingu-lum bundle as well as the hippocampus and fornix (Aggle-ton & Sahgal 1993; Aggleton et al. 1986b; 1995a; 1995b;1996; Neave et al. 1997). Furthermore, probe tests haveconfirmed that normal rats use allocentric cues to solve thisspatial alternation task (Aggleton et al. 1996; Neave et al.1997). Not surprisingly, lesions in these same sites (i.e., theanterior thalamic nuclei, the mamillary bodies, the cingu-

Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Table 1. Effects of lesions in rats showing double dissociations between spatial memory (DNMP, T-maze, radial arm maze, swim maze) and object recognition (spontaneous object recognition, DNMS)a

Anterior Mamillary Cingulum Medial PPRHippocampus Fornix thalamus bodies bundle dorsal th. cortex

DNMP 3 3 3 u u u uT-maze 3 3 3 3 3 u uRadial arm maze (3) 3 3 3 3 u uSwim maze latency (3) 3 3 (3) 3 (u) uSpontaneous object — u u — u — 3

recognitionDNMS u u — u — 3 (3)

aA cross indicates a deficit, a check mark indicates no effect, and a dash indicates that the results have not been reported. All data aredrawn from published research in the laboratory of the authors, with the exception of those in parentheses, which are from Kolb et al.(1982), Morris et al. (1982), Mumby and Pinel (1994), and Sutherland and Rodriguez (1989). Abbreviations: PPR, perirhinal/postrhi-nal cortices.

Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (10)

lum bundle, and the fornix) can disrupt other spatial tasksthought to tax allocentric spatial processing. These includethe radial arm maze and the Morris water maze (Aggletonet al. 1996; Byatt & Dalrymple-Alford 1996; Neave et al.1997; Sutherland & Rodriguez 1989; Warburton & Aggle-ton 1999; Warburton et al. 1997; 1998). These deficits areselective, however; lesions in the same sites (i.e., fornix, an-terior thalamic nuclei, mamillary bodies, and cingulumbundle) do not disrupt a comparable egocentric spatial taskin which the animals are rewarded for turning in a constantdirection while allocentric cues became irrelevant (Aggle-ton et al. 1996; Neave et al. 1997).

By using a standard alternation task, it has been possibleto compare the severity of the spatial deficits following var-ious limbic lesions. These comparisons show that the alter-nation deficit is greatest after lesions in the hippocampus,fornix, and anterior thalamic nuclei and least after mamil-lary body damage or cingulum bundle damage (Fig. 4). Al-though some of these comparisons are affected by floor ef-fects, they do serve to underline the importance of theanterior thalamic nuclei. Furthermore, because the ante-rior thalamic lesion effects are significantly greater thanthose observed after mamillary body lesions (Aggleton &Sahgal 1993; Aggleton et al. 1995) the results point to a sys-tem subserving allocentric spatial memory that involves thedirect hippocampal–anterior thalamic projections as well asthe indirect hippocampal–mamillary body–anterior thala-mic projections (Fig. 5, Table 1). This is consistent with thefinding that complete or near-complete neurotoxic lesionsof all three anterior thalamic nuclei produce an impairment

as severe as that observed after fornix transection (Warbur-ton et al. 1997).

This focus on the anterior thalamic nuclei raises thequestion of whether any of the three component nuclei (an-terior ventral, anterior dorsal, anterior medial) is especiallycritical for spatial memory. All three nuclei have substantialconnections with the hippocampus, mamillary bodies, andcingulate cortices (Shibata 1992; 1993a; 1993b), althoughthere are some distinct differences in the detailed patternsof these connections. Most notably, the anterior dorsal nu-cleus receives afferents from the lateral mamillary nucleus,whereas the anterior ventral and anterior medial nuclei re-ceive their afferents from the medial mamillary nucleus(Cruce 1975). The anterior dorsal nucleus also receives thefewest hippocampal inputs (Aggleton et al. 1986a). Singleunit recording studies also point to differences within theanterior thalamic nuclei as the distribution of “head direc-tion” cells varies within the anterior thalamic nuclei (Taube1995), but this has yet to be matched to any particularanatomical boundary or projection zone.

To investigate possible functional differences within theanterior thalamic nuclei the effects of lesions in the ante-rior medial nucleus have been contrasted with those ofmore lateral lesions involving both the anterior ventral andthe anterior dorsal nuclei (Aggleton et al. 1996; see also By-att & Dalrymple-Alford 1996). Both sets of lesions pro-duced mild, but significant, deficits on the T-maze alterna-tion task, but it was only when the lesions were combinedthat the full effect of anterior thalamic damage became ev-ident. These results not only suggest that all three anteriorthalamic nuclei are integral to the proposed system, but alsoshow that attempts to assess fully the effects of anteriorthalamic damage should involve all three nuclei. Thesefindings may therefore help to account for those studies inwhich small, subtotal anterior thalamic lesions had little orno apparent effect on tests of spatial working memory (Be-racochea & Jaffard 1995; Beracochea et al. 1989; Greene &Naranjo 1986). They also help to explain some of thedeficits reported with lesions of the internal medullary lam-ina following pyrithiamine-induced thiamine deficiency(Langlais & Savage 1995). This is because damage to theanterior thalamic nuclei and mamillary bodies is a consis-tent feature of this animal model (Langlais & Savage 1995)and could account for many of the spatial deficits.

Damage to a number of other sites can disrupt T-mazealternation, including the prelimbic (medial prefrontal) andcingulate cortices (Brito et al. 1982; Markowska et al. 1989;Shaw & Aggleton 1993; Sutherland & Hoesing 1993;Sutherland et al. 1988; Thomas & Brito 1980). Both regionsare of interest insofar as they have connections with the hip-pocampus and anterior thalamic nuclei as well as with themedial dorsal nucleus of the thalamus. Although both cor-tical regions presumably contribute to the normal process-ing of these spatial tasks, their importance might have beenoverestimated. This is because most lesion studies havedamaged fibres of passage and adjacent tracts (e.g., the cin-gulum bundle). With use of cytotoxins to produce selectivelesions in these cortical areas, evidence is emerging thateven extensive damage to the cingulate cortices has little, ifany, effect on spatial tasks such as T-maze alternation (Ag-gleton et al. 1995b; Neave et al. 1994) or the Morris watermaze (Warburton et al. 1998). Similarly, more selective pre-frontal lesions often produce only transient deficits on stan-dard tasks thought to assess allocentric spatial memory (Ag-

Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Figure 4. T-maze alternation: Combined summary data fromthree experiments (No. 1, Aggleton et al. 1996; No. 2, Neave et al.1997; No. 3, Warburton et al. 1997a) showing the effects of selec-tive limbic lesions on spatial alternation. The histograms show themean percentage of correct scores for each of the groups over 15acquisition sessions (90 trials). The lesion locations, from left toright are: sham controls; anterior thalamic nuclei plus lateral dor-sal (ANTLD1); fornix (FX); anterior thalamic nuclei (ANT); bilat-eral cingulum bundle (CCB2); mamillary bodies (MB2); anteriorventral/anterior dorsal thalamic nuclei (AD1); anterior medialthalamic nucleus (AM1). The numbers 1– 3 refer to the number ofthe experiment.

Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (11)

gleton et al. 1995b; Shaw & Aggleton 1993; Thomas & Brito1980; Thomas & Spafford 1984). Furthermore, when morepermanent deficits are observed after prefrontal damage,they often appear qualitatively different from those ob-served after hippocampal damage; for example, they reflecta loss of behavioural flexibility rather than a loss of spatialmemory (Aggleton et al. 1995b; Bruin et al. 1994; Granonet al. 1994).

A further diencephalic region that might prove to be ofimportance is the thalamic midline. Nuclei in this region,most especially the more rostral portions of nuclei such asreuniens, paraventricular, and parataenialis, all have recip-rocal connections with the hippocampus. These connec-tions are found across a range of species, although they ap-pear particularly dense in the rat. Although highly selectiveanterior thalamic lesions show that damage to these mid-line nuclei is not necessary to induce a spatial alternationdeficit (Aggleton et al. 1996), this does not show whetherthese nuclei contribute to the effects of more extensive le-sions. Evidence of a possible involvement of the midlinenuclei in the amnesia associated with paramedial thalamicinfarcts has come from a PET study showing widespreadcortical hypometabolism that might be attributable to theloss of midline nuclei (Levasseur et al. 1992). The contri-butions of these nuclei clearly require systematic investiga-tion, though this will prove technically difficult.

The studies so far cited have concerned rats or mice;comparable tests of allocentric spatial memory have rarelybeen conducted with monkeys. It has been shown, how-ever, that fornix lesions impair T-maze forced alternation bymonkeys, whereas cingulate gyrus lesions have little or noeffect (Murray et al. 1989). Related evidence comes fromstudies showing that hippocampal and fornix lesions canboth impair tasks that require the animal to remember theposition of a given object (Gaffan & Harrison 1989; Parkin-son et al. 1988) or to perform a place discrimination and

subsequent reversals (Gaffan 1994a; Jones & Mishkin 1972;Mahut 1972). The effects on such tasks of discrete anteriorthalamic lesions have yet to be assessed in monkeys, but thepattern of deficits following lesions in the mamillary bodies(Holmes & Butters 1983) strongly suggests that spatialdeficits will be found. This overall pattern of results is inclose accordance with findings from rats.

Other relevant evidence has come from an ingenious se-ries of experiments showing how fornix damage in monkeysdisrupts discrimination tasks in which task performance isaided by the ability to remember the spatial disposition ofthe elements that make up the stimulus. Initial evidencecame from the finding that fornix lesions disrupt condi-tional tasks that tax the ability to identify a particular place(Gaffan & Harrison 1989). It is important that this deficitwas evident only when the places to be distinguished con-tained common elements that were spatially rearranged(Gaffan & Harrison 1989). It was proposed that the fornixis necessary for the creation of a snapshot memory thatstores the spatial arrangements of the items in a “scene”(Gaffan 1991). This conclusion was supported by the find-ing that lesions of the fornix will disrupt the acquisition ofconcurrent discriminations when the stimuli to be discrim-inated are complex scenes that often contain common ele-ments (Gaffan 1992b; 1994a). This finding has recentlybeen explored in more detail, and it appears that the criti-cal feature is whether task performance (in this case, con-current discrimination) is aided if the animals can remem-ber the background and location of the stimuli to bediscriminated (Gaffan 1994b).

These results have been taken to indicate that the fornix,and hence the hippocampus, is important for the scene-specific memory of objects (Gaffan 1992a; 1994a; 1994b).From use of the same “object-in-place” task there is nowevidence that the mamillary bodies are involved in thissame process (Parker & Gaffan 1997a). Furthermore, this

Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Figure 5. Schematic diagram showing the main group of interconnections underlying pure allocentric spatial memory performance inthe rat. The thickness of the line corresponds to the relative importance of the connection as determined by lesion experiments.

Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (12)

mamillary body involvement appears to be via its afferentsfrom the fornix (Parker & Gaffan 1997a). Consistent withthe main proposals in this target article, this object-in-placetask also depends on the integrity of the anterior thalamicnuclei in monkeys (Parker & Gaffan 1997b). These studiesprovide important evidence that the hippocampal-fornix–anterior thalamic system might aid the normal recall ofepisodic information as it permits the subject to distinguishor recreate the unique scene associated with the item to beremembered (Gaffan 1992a; 1994b; Tulving 1983), a processthat will reduce interference from other similar events(Gaffan 1994b). Interestingly, surgical ablations of the cin-gulate cortex produced only a mild impairment (Parker &Gaffan 1997b) and so echo the effects of selective cingu-late/cingulum bundle lesions on spatial memory tasks inrats (Neave et al. 1994; 1996; 1997).

It has been pointed out that the anterior thalamic nucleiand the lateral dorsal nucleus contain “head direction” units(Mizumori & Williams 1993; Taube 1995). These inform theanimal of the direction in which it is pointing, irrespective ofactual position in space. In that a loss of thalamic “head di-rection” information appears to disrupt “place” cells in thehippocampus (Mizumori et al. 1994), this could account forsome of the similarities between the effects of damage inthese two regions on tests of spatial navigation. This couldalso explain how anterior thalamic lesions can disrupt theMorris water maze to a greater extent than do fornix lesions(Warburton & Aggleton 1998), the additional deficit re-flecting the loss of head-direction information. This role inproviding head-direction information is not, however, suffi-cient to explain the full array of similarities between the ef-fects of hippocampal and anterior thalamic damage.

Perhaps most important are the findings that lesions ofthe mamillary bodies and anterior thalamic nuclei in mon-keys disrupt the “object-in-place” discrimination task andthat this effect is dependent on inputs from the fornix(Parker & Gaffan 1997a; 1997b). It is difficult to imaginehow a loss of head-direction units could account for a deficitin such a discrimination because the animal is pointing tothe whole scene. Other evidence comes from the findingthat selective lesions of the anterior medial thalamic nu-cleus in rats are sufficient to disrupt tasks sensitive to hip-pocampal dysfunction, for example, T-maze alternation orradial-arm maze (Aggleton et al. 1996; Byatt & Dalrymple-Alford 1996), even though there is no evidence that this di-vision of the anterior thalamic nuclei contains head directionunits (Taube 1995). Single unit recordings also highlight thecontribution of the anterior thalamic nuclei to discrimina-tive avoidance tasks that have minimal spatial demands(Gabriel 1993). Finally, studies of neuronal activation inmonkeys using the 2-deoxyglucose method have revealedsimilar increases in activity in the hippocampus, mamillarybodies, and anterior thalamic nuclei on a variety of tests ofworking memory (Friedman et al. 1990), even though per-formance on some of the tasks is unlikely to involve headdirection information.

5.2. Reconsidering clinical evidence in light of the proposed model

A central aspect of the current proposal is that bilateraldamage to the fornix, mamillary bodies, or anterior thala-mic nuclei is sufficient to induce anterograde amnesia. Fur-thermore, the more selective the damage, the greater the

disparity between the loss of episodic memory and the spar-ing of recognition.

5.2.1. Fornix damage and amnesia. A highly influential re-view on the effects of fornix section (Garcia-Bengochea &Friedman 1987) concluded that damage to this tract wasnot sufficient to induce amnesia. This review has, however,been strongly criticised (Gaffan & Gaffan 1991). One prob-lem concerns the need to separate those cases with bilateraland unilateral fornix surgery. This is because unilateral hip-pocampal ablation in cases of epilepsy does not bring aboutanterograde amnesia; hence, neither should unilateralfornix section. Gaffan and Gaffan (1991) pointed out thatmany of the cases regarded as having bilateral fornix sec-tions (Garcia-Bengochea & Friedman 1987) in fact, hadonly unilateral surgery. Other problems with this materialinclude the lack of pre- and postsurgical psychometric data.This information is required because some of the relevantcases were psychotic or mentally retarded (Gaffan & Gaffan1991; Sugita et al. 1971), making it difficult to measure anychange in memory.

There are now a growing number of reports that havelinked fornix damage with a loss of episodic memory. Mostof them concern the outcome of cysts or tumours in thethird ventricle (Calabrese et al 1995; Cameron & Archibold1981; Gaffan et al. 1991; Geffen et al. 1980; Heilman &Sypert 1977; Hodges & Carpenter 1991; McMackin et al.1995; Sweet et al. 1959; Tucker et al. 1988). Of special in-terest is a report by McMackin et al. (1995) in which six pa-tients who had undergone surgical removal of a third ven-tricular cyst were assessed on a variety of memory tasks.The status of the fornix was examined using MRI, and thisrevealed a clear association between bilateral fornix dam-age and a loss of recent verbal and nonverbal memory. Thiswas found even in patients with no signs of ventricular di-lation. The only patient to show relatively normal perfor-mance on the verbal tests was a patient in whom the leftfornix was intact (McMackin et al. 1995). This comple-ments other evidence of an association between left forni-cal damage and poor performance on tests of verbal mem-ory (Cameron & Archibold 1981; Hodges & Carpenter1991; Tucker et al. 1988). A very small number of reportsexist of bilateral fornix damage associated with tumours orsurgery for cysts that did not appear to disturb memory(Cairns & Mosberg 1951; Woolsey & Nelson 1975), but theimpact of these studies is limited by a lack of psychometricinformation.

5.2.2. Mamillary bodies and amnesia. Although mamillarybody pathology has repeatedly been associated with mem-ory loss, it has long been uncertain whether damage in thisstructure is sufficient to induce amnesia. Initial evidencecame from the consistent necrotic state of the medialmamillary bodies in Korsakoff ’s syndrome (Delay & Brion1969; Gudden 1896; Mair et al. 1979; Rigges & Boles 1944;Victor et al. 1971). This disease typically affects multiplebrain sites (Victor et al. 1971), but there are a small num-ber of amnesic cases in which the pathology appeared tohave been restricted to the mamillary bodies (Colmant1965; Delay & Brion 1969; Remy 1942; Torvik 1987). Cau-tion is required, however, as it is often unclear whether allother regions have been studied (e.g., frontal or temporallobe areas). Nevertheless, the consistency of mamillarybody damage in well-characterised cases of Korsakoff ’s dis-

Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (13)

ease is striking (Mair et al. 1979; Mayes et al. 1988). Otherrelevant evidence comes from reports of amnesia followingtumours located in the region of the mamillary bodies (As-sal et al. 1976; Benedek & Juma 1941), but interpretationis hindered by the possible effects of raised intraventricu-lar pressure on other diencephalic regions.

There are a few examples in which pathology appears tobe confined to the mamillary bodies, yet no amnesia is re-ported. The most influential of these negative examples arethe five Wernicke cases listed by Victor et al. (1989). Theseare clearly very important exceptions in that they run counterto the current proposals. Unfortunately no psychometricdata for these five cases have been published, leaving it un-certain how it could be determined that their memories wereintact (Victor et al. 1989). Furthermore, although Victor etal. (1989) argue that damage to the medial dorsal nucleus ofthe thalamus is consistently associated with amnesia, otherstudies have failed to find this precise linkage (Mair et al.1979; Torvik 1987). In view of the widespread dysfunctionsrevealed by PET studies of Korsakoff’s disease (Fazio et al.1992; Paller et al. 1997), it is most unlikely that studies of Kor-sakoff’s disease can resolve this issue.

Other evidence comes from the amnesia that can followthalamic vascular lesions. A consistent feature of the pathol-ogy is damage to the mamillothalamic tracts (Castaigne etal. 1981; Cramon et al. 1985; Gentilini et al. 1987; Golden-berg et al. 1983; Graf-Radford et al. 1990; Hodges & Mc-Carthy 1993; Mori et al. 1986; Parkin et al. 1994; Stuss etal. 1988). This pathway carries projections from the mamil-lary bodies to the anterior thalamic nuclei and so is integralto the circuitry in this proposal. The thalamic pathologydoes, however, often invade other regions, including themedial dorsal nucleus and the internal medullary lamina,both of which have also been implicated in memory dys-function. The conclusions of two surveys that compared theextent of diencephalic pathology and the presence andseverity of amnesia are therefore of particular value (Cra-mon et al. 1985; Gentilini et al. 1987). Both studies failedto find an association with damage to the medial dorsal thal-amus (Cramon et al. 1985; Gentilini et al. 1987; see alsoDaum & Ackerman 1994; Graf-Radford et al. 1990; Mar-kowitsch 1982). One of these surveys also found no consis-tent link with internal medullary lamina damage (Gentiliniet al. 1987). In all of these studies, however, the presenceof bilateral mamillothalamic tract damage was a reliablepredictor of amnesia (Cramon et al. 1985; Gentilini et al.1988).

The corollary of this conclusion is that cases in which themamillothalamic tract and the anterior thalamic nuclei arespared will not become amnesic. This is supported by thefailure of lesions restricted within the medial dorsal nucleusto produce anterograde amnesia (Kritchevsky et al. 1987;Markowitsch 1982). Exceptions to this prediction are case5 of Castaigne et al. (1981) and a single case described byCalabrese et al. (1993). In the latter case the pathology isbased on CT scans (Calabrese et al. 1993) and hence maylack sufficient resolution. In the former case the thalamicpathology is confirmed by postmortem examination (Cas-taigne et al. 1981), but the fact that this patient sufferedfrom hypertension and loss of consciousness prior to the in-farct raises the possibility of pathology in other key brainsites not reported.

Penetrating brain injuries in the region of the mamillarybodies have been described in two amnesic patients (Du-

soir et al. 1990; Squire et al. 1989). One of these patients(BJ) appeared to suffer bilateral damage closely confined tothe mamillary body region (Dusoir et al. 1990), and al-though the subject showed some recovery he has a perma-nent impairment for the recall of verbal and nonverbal ma-terial (Dusoir et al. 1990; Kapur et al. 1995). The severityof his amnesia, as measured by the WMSr, is typical for an-terograde amnesia (Butters et al. 1988). The nature of BJ’sinjury does, however, mean that other diencephalic path-ways could have been damaged (Kapur et al. 1995). Indeed,a PET study of BJ revealed evidence of unilateral hip-pocampal hypoactivity in the same hemisphere that re-ceived the largest amount of mamillary body damage (Ka-pur et al. 1995). It is not, however, possible to tell whetherthis hypoactivity reflects a primary pathology or a secondaryresponse to the mamillary body damage. Taken together,there is still no single example of an amnesic with discrete,bilateral mamillary body damage. Nevertheless, the weightof evidence strongly indicates that damage to this region issufficient to impair episodic memory.

5.2.3. The anterior thalamic nuclei and amnesia in hu-mans. Direct evidence implicating the anterior thalamicnuclei in diencephalic amnesia is very limited. In Kor-sakoff ’s syndrome, damage to the anterior nuclei occursonly in about one-third of all cases (Victor et al. 1971),whereas the lateral dorsal nucleus shows degeneration inthe majority of cases (Brion & Mikol 1978; Victor et al.1971). Although infarcts of the paramedian thalamic arter-ies often spare the anterior thalamic nuclei, they do consis-tently deafferent the anterior thalamic nuclei by disruptingthe mamillothalamic tract.

More direct evidence comes from a handful of cases withanterior thalamic damage. For example, a recent MRIstudy described a patient with anterograde amnesia associ-ated with an infarct largely confined to the left anterior thal-amic nuclei and the left mamillothalamic tract in whom aloss of verbal memory was most pronounced (Clarke et al.1994). A PET study of the same subject revealed decreasedmetabolic rate in the posterior cingulate cortex (Clarke etal. 1994), a result consistent with pathology in the anteriorthalamic nuclei. Complementary evidence comes from thecase of an amnesic man who suffered a lesion involving theretrosplenial cortex and cingulum bundle (Valenstein et al.1987). A subsequent PET study revealed evidence of hy-pometabolism in the thalamus, although the medial tem-poral region appeared normal (Heilman et al. 1990). It wasconcluded that the amnesia in this patient was a conse-quence of the disconnection of anterior thalamic pathways(Heilman et al. 1990). A similar explanation could accountfor a patient who suffered a circ*mscribed haemorrhage inthe left retrosplenial area and cingulum bundle (Von Cra-mon 1992). A mild, but perceptible, loss of verbal learningand memory was observed, which disappeared after 6months (Von Cramon 1992).

Other evidence comes from a person who displayed apersistent impairment in the recall of verbal material, but asparing of recognition (Hanley et al. 1994). The impairmentfollowed an anterior communicating artery aneurysm,which resulted in left hemispheric damage in the anteriorthalamus and caudate nucleus (Hanley et al. 1994). In an-other informative case, a woman suffered an infarct centredon the anterior thalamic nuclei and the genu of the right in-ternal capsule. She developed a severe anterograde amne-

Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (14)

sia for verbal and nonverbal information (Schnider et al.1996), which was characterised by a failure to use contex-tual information so that her recognition performance wasmarred by high levels of false positives. Her performanceimproved dramatically to within normal limits when testedwith nonsense words and nonsense designs, stimuli forwhich contextual information (i.e. remembering) would beof less value but familiarity might be sufficient (Schnider etal. 1996). Although this pattern of performance clearly fitswith the current proposals it should be noted that the in-farct probably disconnected other potentially important fi-bre tracts.

Finally, the production of bilateral radio frequency lesionsin the anterior nuclei induced a loss of memory in a patientwith chronic depression (Mark et al. 1970). It was possibleto confirm the placement of the lesions following the suicideof the patient 6 weeks after surgery. The lesion destroyed thewhole of the anterior nuclear complex “but essentially nocells of nucleus ventralis anterior or nucleus dorsomedi-anum” (Mark et al. 1970). Although the patient refused totake memory tests, she expressed concern about her recentmemory loss and she had difficulty in remembering the lo-cation of her hospital room. Although the authors regardedthe memory loss as transient, this could not be verified.

5.3. Electrophysiological studies

Neuronal recording studies provide strong support for theidea of a division of function within the medial temporallobe between the perirhinal and the hippocampal cortices.Indeed, electrophysiological evidence of the importance ofperirhinal cortex rather than the hippocampus in judgingstimulus familiarity preceded lesion evidence (Brown et al.1987). A number of studies have now examined neuronalresponses in anterior inferior temporal cortex, includingperirhinal cortex, during the performance by monkeys ofrecognition tasks using large sets of stimuli (Brown et al.1987; Eskander et al. 1992; Fahy et al. 1993a; Li et al. 1993;Lueschow et al. 1994; Miller et al. 1993; Riches et al. 1991;Sobotka & Ringo 1993; 1994; for reviews, see Brown 1996;Brown & Xiang 1998; Ringo 1996). The tasks employedhave been either variants of DMS or variants of a serial(running) recognition memory task (Gaffan 1974). Thesestudies have confirmed that many neurones in perirhinalcortex respond maximally to first presentations of visualstimuli, but less so to subsequent presentations. Hence, thereoccurrence of a particular stimulus is signalled by a de-crease in the neuronal response to that stimulus. Similar re-sponses have been found in the perirhinal cortex of the rat(Zhu et al. 1995a). It should be noted that, in monkeys, neu-rones that increase their responses when a stimulus reoc-curs are found even less often than might be expected bychance (Xiang & Brown 1998).

These changed responses with stimulus repetition aresufficient to solve recognition memory tasks such as DMSand DNMS as commonly tested in monkeys. Thus: (1) a sin-gle exposure to a stimulus is sufficient to cause a change inresponsiveness, that is, one-trial learning; (2) for many neu-rones, the response change is found even when a long pe-riod of time (for example, 24 hr) has elapsed and/or many(hundreds of) presentations of other stimuli have inter-vened between the first and the next presentation of a stim-ulus; and (3) the effect is highly stimulus specific; that is, aneurone that responds weakly to a stimulus that has been

seen before still responds strongly to novel stimuli, thoughthere may be a limited amount of stimulus generalisation(Brown 1996; Lueschow et al. 1994; Ringo 1996). More-over, individual neurones independently signal differenttypes of information of potential use to the judgement ofprior occurrence (Fahy et al. 1993a; Zhu et al. 1995a). Thus,there are “recency neurones” that encode whether a stim-ulus has been seen recently irrespective of whether it hasbeen seen many or few times previously. Other, “familiarityneurones” encode the relative familiarity of stimuli (i.e.,whether they have been seen many or a few times previ-ously) without regard to whether they have been seen re-cently. A third category responds best to novel stimuli orfirst presentations of unfamiliar stimuli that have not beenencountered recently (Xiang & Brown 1997; 1998). Allthree types of neurones are found in anterior and medialtemporal cortex (area TE) and in entorhinal cortex as wellas in perirhinal cortex (Xiang & Brown 1997; 1998). Neu-rones have been reported (Li et al. 1993) whose responsechanges progressively as initially unfamiliar stimuli are suc-cessively repeated (and presumably become increasinglyfamiliar). Other types of neuronal activity changes such asincremental responses (Miller & Desimone 1994) and in-creased firing after stimulus offset (delay activity; Fuster &Jervey 1981) are also seen in these regions, and they mayfacilitate or provide alternative means of solving specifictypes of recognition memory tasks (Brown 1996).

Carefully controlled experiments have established thatthe described response changes are not artefactual: theycannot be explained by alterations in eye or body move-ments, or motivational or attentional factors (see Brown1996). Importantly, they are found not only during the per-formance of recognition memory tasks, but also when stim-uli are shown without the animal being required to make aparticular behavioural response and using types of stimulinot employed in the animal’s training (Brown 1996; Richeset al. 1991). Accordingly, the response changes are an en-dogenous rather than a training-induced property of theneurones.

Not all visually responsive neurones recorded in perirhi-nal cortex change their response with stimulus repetition;many respond equivalently even after the same stimulushas been shown many times. The latter type of neurone en-codes information concerning the physical characteristicsof the stimuli and so may contribute to stimulus identifica-tion. Nevertheless, the incidence of neurones whose re-sponses do change is about 20–25% of the whole neuronalpopulation (Brown 1996), a finding that has been exploitedin c-fos studies (see sect. 5.4). Neurones with changing andneurones with unchanging responses are found intermin-gled, so comparisons between cells with the two types of re-sponse could provide information concerning the previousoccurrence of a stimulus. Furthermore, as the responsive-ness of both types of neurones varies with the physical char-acteristics of the stimulus, and the degree of responsedecrement and the time for which it lasts vary amongst neu-rones whose responses change, population measures canpotentially provide a sensitive measure of the past occur-rence of a particular stimulus (both of its recency of occur-rence and of its familiarity).

There is also electrophysiological and ablation evidencethat the perirhinal region is involved in the learning of vi-sual paired associates (Higuchi & Miyash*ta 1996; Miya-sh*ta et al. 1996; Murray et al. 1993; Okuno & Miyash*ta

Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (15)

1996). Thus electrophysiological evidence supports a rolefor perirhinal cortex in recording knowledge concerning in-dividual stimuli and of simple associations between suchstimuli. Information concerning the prior occurrence of in-dividual stimuli is also available in the inferior temporal cor-tex adjacent to perirhinal cortex, but in that region it is likelyto be restricted to the visual modality. Perirhinal cortex re-ceives multimodal sensory input (Burwell et al. 1995) andso may serve a similar function for information not confinedto a single modality, although present evidence is largelyconfined to the visual modality. Information about prior oc-currence is also available within entorhinal cortex (Fahy etal. 1993b; Suzuki et al. 1995; Xiang & Brown 1998). In con-trast, neurones that change their response after a single oc-currence of an individual stimulus are uncommon in thehippocampus of monkeys (Brown et al. 1987; Miyash*ta etal. 1989; Riches et al. 1991; Rolls et al. 1993; Xiang & Brown1998) and rats (Zhu et al. 1995a), a finding confirmed in c-fos studies in the rat (Wan et al. 1997; Zhu et al. 1995b;1996). Thus such perirhinal and entorhinal information isnot necessarily passed on to hippocampal neurones in spiteof the numerous interconnections between these regions.

Contrasting with the findings for individual stimulus rep-etition is much evidence that the activity of hippocampalneurones reflects features of the spatial environment orspatial arrangements of stimuli (Eichenbaum et al. 1994;Eif*cku et al. 1995; Gothard et al. 1996; O’Keefe 1993;O’Keefe & Nadel 1978). In particular, individual neuronesincrease their firing when a rat is located in a particular partof an environment (“place fields”). Such place fields are in-fluenced by the presence of defining features of that envi-ronment, including its size and shape (O’Keefe & Burgess1996). In the monkey, increases in firing have beenrecorded when the animal directs its gaze to a particularpart of the environment (“view fields”; Rolls & O’Mara1995). Additionally, changes in hippocampal responses re-lated to a combination of the relative familiarity of a stimu-lus and the place in which it was shown have been described(Miyash*ta et al. 1989), as they have in entorhinal cortex(Suzuki et al. 1995). Although hippocampal neurones alsorespond to discrete stimuli and during paradigms that arenot spatial (see, e.g., Brown 1982; Eichenbaum et al. 1994;Eif*cku et al. 1995; Otto & Eichenbaum 1992b; Thompson1990), electrophysiological recordings have not yet estab-lished that the response characteristics of hippocampalneurones could provide a satisfactory explanation for theproperties of episodic memories. Spatial information is im-portant to episodic memory, but it does not encompass thewhole of the information contained in the context and con-tent of all episodic memories. Nevertheless, the anatomicalconnectivity of the hippocampus is ideal for making thewidely distributed associations (Brown 1990; Lorente deNo 1934) between the many different, contemporaneouslyexperienced stimuli making an event. Such associations arenecessary to the formation of an episodic memory.

Recordings in the monkey medial thalamus have re-vealed neurones whose responses signal information con-cerning the prior occurrence of stimuli. Neurones in boththe medial dorsal nucleus and the paraventricular midlinenucleus responded more strongly to first presentation thanto subsequent presentations of stimuli during the perfor-mance of a recognition memory task (Fahy et al. 1993b).Such findings are consistent with the suggested functionallink between perirhinal cortex and the medial dorsal nu-

cleus of the thalamus, although the numbers of such neu-rones were small both in absolute and in percentage terms.There are no reported recordings from the anterior thala-mic nuclei during the performance of recognition memorytasks. However, neuronal responses in the anterior thalamicnuclei in the rat can code spatial information, particularlyabout the direction of the head (Mizumori & Williams1993; Taube 1995). Thus, thalamic recordings again sup-port the apportionment of function between the hip-pocampal–anterior thalamic and the perirhinal–medialdorsal thalamic axes, with the latter underlying the detec-tion of the prior occurrence of a stimulus (its sense of fa-miliarity, “knowing” that something has occurred previ-ously) and the former being involved in the processing ofimportant components, including the contextual informa-tion concerning events as is necessary for “remembering”or “recollecting” a prior occurrence.

5.4. Brain activation studies: c-fos

The expression of the immediate early gene c-fos providesa potential marker for changes in neuronal activity (Herde-gen 1996) and so can be used to identify regions that re-spond to specific experimental manipulations. We haveused this technique to help map brain regions involved inrecognition (Zhu et al. 1995b; 1996). In the first of thesestudies, counts of fos-stained nuclei were compared in thebrains of rats shown novel objects and in the brains of ratsshown highly familiar objects (Zhu et al. 1995b). The ob-jects were placed behind a glass screen, limiting the task tovisual processes. Four of the eight sites examined showedsignificantly higher levels of fos protein in the brain exposedto novel stimuli. These sites were the occipital cortex, areaTE, perirhinal cortex, and anterior cingulate cortex. No dif-ferences were observed in the entorhinal cortex, the diago-nal band of Broca, the medial dorsal nucleus of the thala-mus, or the hippocampus (Zhu et al. 1995b). These corticalsites of increased activation correspond closely to those de-scribed in a human PET study comparing exposure to fa-miliar and unfamiliar visual stimuli (Vandenberghe et al.1995). Increased blood flow was found in the lateral ante-rior temporal neocortex, the medial temporal pole, and theanterior cingulate cortex following exposure to the unfa-miliar stimuli (Vandenberghe et al. 1995).

A shortcoming with the c-fos result concerns the need tocontrol for any differences in behaviour induced by thepresence of novel objects (e.g., eye movements, alertness).For this reason a second study used a within-subjects de-sign (Zhu et al. 1996). On each trial two objects were shownsimultaneously to the rat, so that one eye saw only novel ob-jects while the other eye saw only familiar objects (Fig. 6).This “paired-viewing” procedure makes it possible to com-pare fos levels in the two hemispheres of the same brain.Novel stimuli activated significantly more neurones (asmeasured by fos) in the perirhinal cortex, area TE, and ven-tral lateral geniculate nucleus of the thalamus. No differ-ences were observed in the hippocampus or any other areasampled. Moreover, the number of stained (activated) neu-rones in the hippocampus was low.

The failure of novel stimuli to evoke significantly in-creased c-fos products in the hippocampus might be be-cause neurones in this structure do not show activity-re-lated expression of c-fos. This possibility was excluded by athird study (Zhu et al. 1997), in which exposure to novel ob-

Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (16)

jects took place in either a familiar or a novel context. Rel-ative to the perirhinal cortex, the hippocampus (fields CA1-4) showed a fourfold increase of activity for the novelcontext compared to the familiar context. This result, whichcontrasts with that for novel compared to familiar objects inperirhinal cortex, provides a double dissociation betweenthe hippocampus and the perirhinal cortex. As a conse-quence the result concurs with lesion and electrophysio-logical studies pointing to a qualitative difference betweenthe contributions of the two regions to recognition and con-textual (spatial) memory.

To explore this difference further a subsequent studycompared c-fos activation when rats were exposed either tonovel or familiar visual stimuli or to novel or familiar spatialrearrangements of familiar stimuli (Wan et al. 1999). As inprevious studies the former condition led to higher c-fos ex-pression for novel than for familiar stimuli in the perirhinalcortex but not in the hippocampus or postrhinal cortex. Incontrast, the novel spatial compared to familiar spatial re-arrangements resulted in increased c-fos activation in thepostrhinal cortex, and, whereas some regions of the hip-pocampus showed increased (CA1) activation, others (den-tate gyrus and subiculum) showed decreased activation(Wan et al. 1999). In view of the supposed hom*ology be-tween the postrhinal cortex and the parahippocampal gyrus(Burwell et al. 1995), these findings provide an insight intohow novelty per se need not engage the hippocampus, butnovelty arising from the mismatch of certain learnt associ-ations in space may do so. This conclusion matches thatfound in lesion studies of novelty detection (Honey et al.1998) and conditional learning (Gaffan & Harrison 1989).It also accords with a recent functional MRI study showingraised activation in the parahippocampal gyrus during thepresentation of novel words or scenes but increased activa-tion in the subiculum during retrieval (Gabrieli et al. 1997).

6. The consequence of multiple process modelsof recognition

Until now recognition has largely been considered asthough it were a single process, but many psychologists ar-gue that it involves at least two distinct processes (Gardiner& Parkin 1990; Jacoby & Dallas 1981; Mandler 1980). Oneof these permits a recognition judgement to be made on thebasis of stimulus familiarity (sometimes regarded as “per-ceptual fluency”). Using this information, subjects will have

a feeling of “knowing” that they have experienced the testitem but may remember no other associated details. Thesecond process is regarded as elaborative; it involves recol-lecting (“remembering”) the experience of the test item. Asa consequence the item may be recalled along with associ-ated contextual information (Perfect et al. 1996). Familiar-ity is regarded as an automatic process, whereas recollec-tion is an effortful retrieval process. An alternative view ofrecognition is that “know” responses merely reflect a weakerrecognition process than “remember” responses and thatthey are part of the same mechanism.

Evidence for this two-process account comes from ma-nipulations, such as changing levels of processing and in-creasing the interval between sample and test, that differ-entially affect the two classes of response (Gardiner 1988;Rajaram 1993; Tulving 1985a). Most importantly, there areconditions that affect “know” responses but not “remem-ber” responses, and vice versa, producing a double dissoci-ation between the two classes of response (Gardiner & Java1990; Rajoram 1993). In spite of these dissociations, it hasbeen argued that a two-process model might not be neces-sary and that it is possible to model these dissociations us-ing a single process model within which different criteriaare set for “remember” and “know” responses (Donaldson1996; Hirshman & Master 1997). It should also be notedthat some of this debate relates not so much to the notionof a distinction between recollective and nonrecollectivememory but more to the way that recognition memory ismeasured (Donaldson 1996). Furthermore, recent analysesof receiver operator characteristics (ROCs) provide furthersupport to dual-process models (Yonelinas 1994; 1997).This final approach is of special interest in that it has led toa comparison (Yonelinas 1997) of recognition judgementsfor item information and associative information (learnedpairs). This revealed evidence that item recognition and as-sociative recognition rely differentially on recollection andfamiliarity. Whereas item recognition reflected principallyfamiliarity but also recollection, the recognition of associa-tive information relied on recollection (Yonelinas 1997).This distinction is especially relevant because it maps ontothe proposed differences between the contributions of theperirhinal cortex to recognition (item familiarity) and thehippocampus to recognition (associative recollection).

It is therefore assumed that the hippocampal–anteriorthalamic system supports recollective-based recognition(remembering) whereas cortical (perirhinal) systems sup-port familiarity-based recognition (knowing). As a conse-quence, all amnesics will show a marked fall in recollective-based recognition although the loss of familiarity-basedrecognition will vary according to the extent of cortical (ex-trahippocampal) dysfunction. Thus, very selective hip-pocampal system damage may spare those tests of recogni-tion that can be performed effectively by using just relativefamiliarity. In contrast, more extensive cortical damage, asfound in the majority of cases, will disrupt both processes.The proposal that the perirhinal contribution to this formof memory is distinct from that of the hippocampus (be-cause double dissociations can be found) has further impli-cations. Namely, that recollection could occur without fa-miliarity. There is good evidence that this can indeed occur(Jacoby & Dallas 1981; Jones & Gardiner 1990; Mandler1980), but it would be expected to be unusual given the in-terlinked flow of afferent sensory information to the rhinalregion and hippocampus (Gaffan & Parker 1996; Mishkin

Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Figure 6. Testing arrangement for the c-fos “paired-viewing”procedure, in which one hemisphere receives direct visual infor-mation from an eye seeing novel stimuli while the other eye viewsfamiliar stimuli (Zhu et al. 1996).

Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (17)

& Murray 1994; Suzuki 1996b). From this proposal it canbe seen that the two components of recognition are re-garded as distinct processes that are independent of one an-other (Jacoby 1991; Yonelinas & Jacoby 1995) but not mu-tually exclusive. This latter point is important; tests basedon “remember” and “know” decisions carry with them anassumption that the processes are mutually exclusive (sub-jects can never respond both “know” and “remember”) andfor this reason provide an imprecise measure of the twoprocesses. This is because “know” responses reflect famil-iarity in the absence of recollection, whereas “remember”responses will contain some items that are both recollectedand familiar (see also Donaldson 1996).

One method for testing these possible dissociationswithin memory has been to assess whether manipulationsthat lead to similar levels of recognition in amnesics andcontrols also result in matched levels of recall. In one studyKorsakoff subjects received an extended study time so thattheir word recognition performance was equated with thatof control subjects (Hirst et al. 1986). The same extendedperiod of study did not, however, abolish the recall deficit,and it was concluded that recall and recognition are differ-entially affected by amnesia (Hirst et al. 1986). A secondstudy equated recognition performance in amnesics andcontrols by testing the controls with longer retention delays(Hirst et al. 1988). At the same delays, the recall perfor-mance of the amnesics was still significantly poorer thanthat of the controls (Hirst et al. 1988). Similar studies havesince been repeated, however (Haist et al. 1992), and theyfailed to show a difference between recall and recognition.This led to the conclusion that recall and recognition aretightly linked functions of declarative memory (Haist et al.1992). The difficulty with these experiments is that theytreat recognition as a unitary process and assume thatequating groups across different retention delays is valid.One problem is that “know” and “remember” responseshave differential rates of forgetting (Tulving 1985a), limit-ing such comparisons.

A second study (Knowlton & Squire 1995) used theknow/remember distinction to examine recognition per-formance in a group of 13 amnesics. The amnesics showeda significant loss of both remember and know responses butrevealed a disproportionate loss of remember responsescompared to control subjects tested with the same reten-tion delay. Although this pattern is consistent with the pre-sent proposals, the same study did include some amnesicswith relatively selective pathology. The study reported thatthese cases did not show a different profile of impairment(Knowlton & Squire 1995) as might be predicted (but notethe debate over anoxia and covert pathology; sect. 4.1). It is the case, however, that the data for these individual sub-jects were not provided, and shortcomings of the know/remember distinction have already been discussed.

Single-case information comes from an atypical Wer-nicke-Korsakoff subject who showed remarkable sparingon tests of recognition (Parkin et al. 1993), suggestive of aselective loss of recollective-based recognition. She wastested using the know/remember subjective distinction,and, although she produced the predicted increase in knowresponses, she showed only a very modest drop in remem-ber responses. Other evidence comes from an amnesic pa-tient whose performance on different tests of recognition,following anterior communicating artery aneurysm, wasbest described by the extent to which familiarity informa-

tion could be used (Parkin et al. 1994). Although this casesupports the current proposals, it was not possible to de-termine the locus or extent of the brain injury. Other sup-port comes from studies showing that the recognition per-formance of Korsakoff amnesics depends on judgements oftrace strength, that is, familiarity (Huppert & Piercy 1978).

Taken together, these studies provide general support forthe current proposals, but they fail to reveal the pattern ofrecognition loss following selective hippocampal systemdamage. The finding that amnesics with selective pathologycan perform the Warrington Recognition Memory Test(RMT) within normal limits (Aggleton & Shaw 1996) ac-cords with the present views only if it is assumed that theRMT can be performed accurately using just familiarity in-formation. Some support for this assumption comes fromthe discovery that performance of the Faces Subtest of theRMT (the Words Subtest was not examined) is not influ-enced by extreme switches in context between the samplephase and the test phase (Parkinson & Aggleton 1994) eventhough the same switches markedly impair recall (Godden& Baddeley 1975; Martin & Aggleton 1993) and, hence,should disrupt recollective processes.

At this point it is necessary to consider the relationshipbetween familiarity and priming, particularly repetitionpriming. Because familiarity (a feeling of “knowing”) is anessentially explicit, conscious form of memory, whereaspriming is fundamentally implicit, these two processes can-not have identical substrates; that is, “know” responses can-not be a direct consequence of priming mechanisms andnothing else. Nevertheless, many descriptions of “knowing”regard it as reflecting increased perceptual fluency, that is,priming. Moreover, in the present model familiarity as wellas priming is regarded as an automatic consequence of pas-sive exposure to stimuli. Accordingly, processes leading topriming and to familiarity will normally co-occur, yet the twocan be dissociated. Thus, priming is often preserved in am-nesics even though they are typically poor at tests of recog-nition (Cermak et al. 1985; Hamann & Squire 1997; Schac-ter et al. 1991; Warrington & Weiskrantz 1974), includingthe RMT. This distinction is highlighted in an amnesic, EP,who showed chance performance on a range of recognitiontasks (i.e., could not use familiarity information) yet displayedintact priming on matched tasks (Hamann & Squire 1997).Other evidence comes from the loss of know responses inmany amnesics (Knowlton & Squire 1995), which contrastswith the preservation of other forms of implicit memory.

These findings leave open the possibility that primingand familiarity share the same initial stages of processingbut that familiarity requires additional, separate mecha-nisms that give rise to subjective appreciation (“knowing”)and allow explicit guidance of choice behaviour. Two fur-ther possibilities concerning these additional mechanismsare suggested by the data available. One is that processinginvolving the integrity of perirhinal cortex is necessary forfamiliarity but not for priming: this is consistent with thestudies reviewed above, including those of patient EP(Hamann & Squire 1997). The other requires that the ab-sence of contextual information (remembering) rendersvery difficult the use of primed information to aid recogni-tion, even though priming is intact as measured by othermeans. However, this second suggestion leaves unansweredthe problem of why this implicit information cannot beused even in a forced-choice paradigm (Hamann & Squire1997). Similarly, it does not explain why monkeys with hip-

Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (18)

pocampal system damage can still perform at normal ornear-normal levels on DNMS tasks, nor how some am-nesics can perform within normal limits on certain tests ofrecognition (insofar as both have lost contextual informa-tion). Thus, it seems necessary to adopt the former of thesesuggestions and conclude that, unlike the case with repeti-tion priming, feelings of familiarity depend on processinginvolving perirhinal cortex (Wagner et al. 1998).

This conclusion may well be relevant to a study using the“process-dissociation procedure” (Jacoby 1991), which isdesigned to separate recollective (explicit) processes fromimplicit aspects of recognition. It was found that amnesicsshowed a disproportionate loss in the use of recollection asa basis for recognition (Verfaellie & Treadwell 1993). Thiscomponent of the study supports the current proposals;such a loss would reflect the core loss of episodic memory.The amnesic subjects were, however, as likely as controlsubjects to use perceptual fluency (Verfaellie & Treadwell1993). This latter finding is also consistent with the currentproposals if the measurement of “familiarity” in the process-dissociation procedure corresponds to implicit memory(which may be spared). If, on the other hand, this measurereflects perirhinal familiarity, then the results are at oddswith the current proposals, unless the pathologies were un-usually selective. It should also be added that the findingsof Verfaellie and Treadwell (1993) have been challenged(Roediger & McDermott 1993; but see Verfaellie 1994), ashas the validity of the process-dissociation procedure(Dehn & Engelkamp 1997; Ratcliff et al. 1995).

A final consequence of the multiple process model ofrecognition is that it can unify different descriptions of thedeficit that follows selective hippocampal–anterior thala-mic damage. The current proposals argue that such dam-age will spare familiarity-based recognition but impair rec-ollection-based processes in recognition and recall. Thiswill result in a loss of associative memory (Yonelinas 1997)and, hence, source memory. This prediction is supported bya study confirming that contextual knowledge for items in arecognition test is consistently higher for “remember”items than for “know” items (Perfect et al. 1996). Thus thecurrent proposals are consistent with those of Gaffan (1991;1992a; 1994b), that is, that the hippocampus–anterior thal-amic axis is required for the creation of episodic (associa-tive) scenes that can heighten discriminability and so aid re-trieval of the item to be remembered.

7. Diencephalic amnesia and recognition memory

Amnesia associated with diencephalic damage typically re-sults in a severe loss of recognition. That there is a loss ofboth familiarity and recollective-based recognition is there-fore assumed. Whereas the recollection deficit arises fromdysfunctions in the hippocampal–anterior thalamic system,the source of the familiarity deficit must be elsewhere. Ex-perimental evidence from behavioural studies of both mon-keys (Aggleton & Mishkin 1983b; Parker et al. 1997; Zola-Morgan & Squire 1985a) and rats (Hunt & Aggleton 1991;Mumby et al. 1993) points to the importance of the medialdorsal nucleus of the thalamus. Lesions in this region im-pair DNMS acquisition and performance, which heavilytaxes familiarity judgements. Cells in the medial (magno-cellular) portion of the medial dorsal nucleus and in theadjacent midline nuclei show decremental responses to fa-

miliar visual stimuli (Fahy et al. 1993; see sect. 5.3). More-over, the medial dorsal nucleus is very often affected in di-encephalic amnesic syndromes; gliosis is consistently ob-served in Korsakoff ’s disease (Victor et al. 1971), whereasparamedial thalamic infarcts disconnect both frontal andtemporal interactions (Graf-Radford et al. 1990; Malamutet al. 1992). There have been two cases of lesions largelyconfined within the medial dorsal nucleus that had no ap-parent effect on memory (Kritchevsky et al. 1987), butthese involved only about 15% of the nucleus and so do notprovide a conclusive test of the present proposals.

The medial dorsal thalamic nucleus also receives inputsfrom the perirhinal cortex. Direct projections run to themagnocellular portion of the medial dorsal nucleus via theinferior thalamic peduncle (Aggleton et al. 1986a; Russ-chen et al. 1987). In light of the current proposals, it is note-worthy that these connections do not use the fornix, nordoes the perirhinal cortex appear to project to the anteriornuclei of the thalamus (Aggleton & Saunders 1997). Themagnocellular portion of the medial dorsal nucleus hasdense projections to the medial and orbital frontal cortices,and removal of these cortical regions also impairs DNMSperformance in monkeys (Bachevalier & Mishkin 1986). Incontrast, removal of the dorsolateral prefrontal cortex, whichreceives its inputs from the lateral (parvocellular) portionof medialis dorsalis, does not affect DNMS performance(Bachevalier & Mishkin 1986). Furthermore, lesions of theuncinate fascicle, which connects the temporal associationcortex with the prefrontal cortex, have no effect on DMS(Gaffan & Eacott 1995). These findings all suggest that oneset of key perirhinal outputs for recognition are those to thethalamic nucleus medialis dorsalis (Parker et al. 1997) andthence to the prefrontal cortices. These perirhinal outputsare unlikely, however, to be the sole route; the magnitudeof the medialis dorsalis recognition deficit appears less thanthat associated with perirhinal damage (Aggleton & Mish-kin 1983b; Parker et al. 1997). This indicates that otherperirhinal outputs contribute to recognition but that noneof these other routes is individually critical. Candidates in-clude the projections to prefrontal cortex to entorhinal cor-tex (Meunier et al. 1993) and those back to the inferior tem-poral cortex (in the case of visual recognition).

In view of the fact that the major ouput from medialisdorsalis is to the prefrontal cortex, it is relevant that frontaldamage in humans can sometimes disrupt recognition andalso that this is not an invariable finding (Aggleton & Shaw1996; Schacter 1987; Shimamura et al. 1990). This is high-lighted by a recent survey (Wheeler et al. 1995) concludingthat frontal lobe damage could impair recognition, al-though the effects are smaller than those for free recall. Itshould be noted, however, that dorsolateral prefrontal le-sions in monkeys do not disrupt DNMS performance, andit is this region that is involved in many of the human cases.Clinical cases with extensive medial and orbital damage arerarer, and it is possible that such damage is sufficient to im-pair recognition consistently. The relative mildness of thefrontal recognition impairment might also reflect the pres-ence of projections from the medial dorsal nucleus back tothe rostral perirhinal cortex (Markowitsch et al. 1985) thatmight aid recognition. The mild frontal deficit in the judge-ment of previous occurrence can be contrasted with themore robust failure to discriminate relative recency orsource information (Schacter 1987; Shimamura et al. 1990).Evidence that recency judgements and recognition judge-

Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (19)

ments are not based on the same information has been re-ported elsewhere (Brown 1996; Shaw & Aggleton 1995),and taken together these findings suggest that, although theprefrontal cortex may receive information concerning fa-miliarity (either directly from the temporal lobes or fromthe thalamus), this need not be critical for recognition. In-deed, recent PET studies provide evidence of a dissociationbetween temporal lobe and dorsolateral prefrontal activity,with temporal lobe activity highest for novel stimuli andprefrontal activity highest for novel rearrangements of fa-miliar stimuli (Dolan & Fletcher 1997; see also Tulving etal. 1996).

The prefrontal cortex (like the entorhinal cortex) is, infact, in a privileged position in that it receives both famil-iarity information and source information. This source in-formation is presumably associated with the direct inputs tothe prefrontal cortex from the hippocampus, along with in-puts from the medial portions of the thalamus, includingthe anterior thalamic and midline nuclei (Kievit & Kuypers1977). Brain activation studies also indicate that the frontallobes have a specific involvement in recollective (Dolan &Fletcher 1997; Wilding & Rugg 1996) and retrieval (Rugget al. 1996) aspects of recognition. The key difference is thatfrontal damage does not reproduce the temporal lobe or di-encephalic amnesic syndromes, so its involvement is oftennot critical. To account for this it is assumed that the hip-pocampal–anterior thalamic axis has both a diffuse frontalextension as well as an important reciprocal componentfrom the thalamus back to the temporal lobe and hip-pocampus. Activity in the latter system is sufficient to ame-liorate some of the effects of prefrontal damage on standardtests of recall and associative recognition. At the same time,the prefrontal cortex allows the most effective recall strate-gies to be employed.

8. Subdivisions of amnesia

The present proposals run contrary to the traditional divi-sion between temporal lobe and diencephalic amnesia andpredict that the core symptoms should appear very similar.Indeed, when the problems of matching the severity of theamnesia among different cases and allowing for differentialdamage to additional structures (e.g., frontal cortex) aretaken into consideration, the evidence for qualitative dif-ferences in the core features of these amnesias seems weak(Weiskrantz 1985; Zola-Morgan & Squire 1993). Initial ev-idence of differences in forgetting rates between temporallobe and diencephalic amnesia (Huppert & Piercy 1979;Squire 1981) failed to survive later scrutiny (Freed &Corkin 1988; Freed et al. 1987; McKee & Squire 1992). Al-though some diencephalic amnesics do show a greater lossof short-term memory and contextual cues (Parkin 1984;Parkin et al. 1990), these differences often relate to Kor-sakoff ’s syndrome, which results in frontal dysfunction.This may also account for the loss of short-term memory(Cave & Squire 1992; Leng & Parkin 1989). Evidence of agreater failure to use temporal context information is moredifficult to resolve (Parkin & Hunkin 1993), but it may wellprove to be a combined effect of frontal dysfunction (Shi-mamura et al. 1990) and a loss of recollective informationconcerning the learning episode.

Until now we have sought to emphasize the distinctionbetween a hippocampal–fornix–anterior thalamic system

and a perirhinal–medial dorsal thalamic system. It is, ofcourse, the case that the perirhinal cortex is a major affer-ent source to the entorhinal cortex and thence to the hip-pocampus. In view of evidence showing that the perirhinalcortex is important for knowledge concerning objects(Suzuki 1996), principally their familiarity and whetherthey have been associated with other discrete visual inputs,it can be assumed that this route normally provides object(item)-related information to the hippocampus that may beretained in episodic memory, although other routes can beused following brain pathology (see sect. 4.3). The normalprocess then involves setting the to-be-remembered item(or items) within its episode or context, and for this associ-ation to be possible the hippocampus must receive spatial/contextual information. In the primate brain the most plau-sible route is via the parahippocampal cortex (Habib &Sirugu 1987; Maguirre et al. 1996; Suzuki & Amaral 1994),which permits item-place representations to be formed.The situation in the rat brain might be different; thepostrhinal cortex does not appear to be necessary for somespatial tasks (Aggleton et al. 1997). If, as is assumed, the an-terior thalamic nuclei are vital for episodic memory, thenthese nuclei must interact with those classes of informationdisrupted by amnesia, including memory for discrete items.Thus, it is to be expected that the pathway connecting theperirhinal cortex–entorhinal cortex–hippocampus–fornix–anterior thalamic nuclei forms the route by which discrete-item information is made available for recall.

The assumption that the hippocampus receives object-related information from the perirhinal cortex fits with thenotion that damage to the hippocampal–anterior thalamicsystem can disrupt recognition when familiarity informa-tion is not available. Further evidence comes from the ad-ditive effects of anterior thalamic damage upon posteriormedial thalamic damage to DNMS performance by mon-keys (Aggleton & Mishkin 1983a; 1983b). Similarly, fornixlesions on their own have very little effect on the standardDNMS task, but they can exacerbate the effects of cuttingthe ventral amygdalofugal pathway/temporal stem, whichdisconnects rhinal projections to the medial thalamus(Bachevalier et al. 1985b). Similar evidence has come fromstudies of rats showing that fornix damage can accentuatethe recognition deficit associated with perirhinal damage(Ennaceur & Aggleton 1998; Wiig & Bilkey 1995). In all ofthese instances, the fornix/anterior thalamic lesion willhave involved efferents from the entorhinal cortex as wellas the subiculum. The loss of this entorhinal informationmay contribute to these additive effects.

9. Final comments

This target article has been concerned with extending thefunctional hippocampal system and showing how this re-lates to the pathology underlying amnesia. What has notbeen discussed is why this additional diencephalic process-ing is required. Lesion studies have been relatively unhelp-ful in this regard because damage to the anterior thalamicnuclei seems to mimic the effects of hippocampectomy soclosely. Other approaches are required, and, among these,electrophysiological studies might prove to be especiallyvaluable. An example of this is the work of Gabriel and hiscolleagues (Gabriel 1993). Using a discriminative avoid-ance task in which one tone predicts an avoidable shock

Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (20)

(S1) and another tone does not predict shock (S2), train-ing-induced changes in neural activity have been found inboth the anterior thalamic nuclei and the posterior cingu-late cortex (Gabriel 1993). These consist of increased firingto the conditioned stimuli and greater discriminative firingto S1 than to S2. Because both changes are found in theanterior thalamic nuclei and the upper layers of the poste-rior cingulate cortex (Gabriel 1993), regions that are closelyinterconnected, it has been proposed that they reflect an in-terlinked system. Because this training-induced neuronalplasticity is not observed until late in training, when the as-sociation is well learned, it has been described as a “pri-macy” system, which holds primary or original encodingseven after more recent information has been obtained. Asmight be predicted, lesions of the anterior thalamus–pos-terior cingulate cortex leave initial acquisition of the avoid-ance task intact but affect final levels of performance(Gabriel 1993). In contrast, the adjacent medial dorsal nu-cleus and the interconnected anterior cingulate cortex showdiscriminative activity from the first session in which be-havioural discrimination is observed and are seen as moreimportant in the initial learning stages (a “recency” system).

This model system of avoidance learning offers a meansof assessing how the anterior thalamic nuclei can interactwith cortical regions in a mnemonic capacity, but the basicavoidance task is unlikely to provide a direct measure ofepisodic memory and hence is of limited application. Con-sistent with this is the finding that rabbits with hippocam-pal lesions can show normal acquisition and performance ofthe avoidance task (Gabriel 1993), and hippocampal lesionsdo not disrupt training-induced plasticity in the anteriorthalamic nuclei. Conversely, the cingulate cortices are seenas vital for the avoidance task yet are not critical compo-nents of the episodic memory system outlined in this targetarticle. There is evidence, however, that the hippocampusmodulates overall levels of activity in the anterior thalamicnuclei during the avoidance task. Furthermore, these in-teractions may be influenced by context, and this may de-pend on hippocampal activity. Evidence comes from thefinding that hippocampal lesions in rabbits attenuate the ef-fects of a context shift on extinction of the conditionedavoidance response (Gabriel 1993). It can be seen that suchpreparations are moving closer to the demands of an epi-sodic memory system and so may help to address more di-rectly the important issue of why diencephalic relays arerequired in normal episodic memory.

An underlying assumption in this target article is that thecritical contribution of the anterior thalamic nuclei toepisodic memory will involve not only its hippocampal andmamillary body inputs but also its other afferents. A similarassumption applies to the mamillary bodies: that is, thesestructures are contributing something new and not merelypassively processing hippocampal outputs. In the case ofthe anterior thalamic nuclei one potentially important in-put is the ascending cholinergic projection from the teg-mentum. Preliminary evidence comes from the finding thatacquiring the discriminative avoidance learning task(Gabriel 1993) leads to increased muscarinic acetylcholinebinding in the anterior thalamic nuclei, and this correlateswith the appearance of training-induced changes in neuralactivity (Vogt et al. 1991). A future task will be to look sys-tematically at these and other inputs to the anterior thala-mic nuclei and mamillary bodies and thus better determinethe nature of the contribution of these structures to the hip-

pocampal-anterior thalamic axis. As a consequence, theanalysis of lesions will continue to refine our understandingof the critical pathologies underlying aspects of amnesia,but quite different techniques will be required to identifyhow these different structures contribute to the encodingand maintenance of episodic memory.

ACKNOWLEDGMENTSThe research of both the authors has been supported by the Med-ical Research Council (U.K.) and the Wellcome Trust. The au-thors thank W. Macken and the helpful comments of the referees.

Open Peer Commentary

Commentary submitted by the qualified professional readership of thisjournal will be considered for publication in a later issue as ContinuingCommentary on this article. Integrative overviews and syntheses are es-pecially encouraged.

Perirhinal cortex: Lost in space?

David K. BilkeyDepartment of Psychology, University of Otago, Dunedin, New [emailprotected] psy.otago.ac.nz:800/academic/bilkey.html

Abstract: Aggleton & Brown argue that the function of the hippocampusand perirhinal cortex can be dissociated along a spatial/nonspatial dimen-sion. They further suggest that this division corresponds to a distinctionbetween episodic and recognition memory. An analysis of the data, how-ever, fails to support the underlying dissociation.

Aggleton & Brown (A&B) provide a timely reappraisal of the roleof temporal and diencephalic structures in amnesia. As part oftheir reassessment, however, they draw a division between thehippocampus and the rhinal cortex that minimises the influenceof the stream of neocortical information that is channeled throughperirhinal and entorhinal cortex into this former structure (sect.4.3). This division is adopted on the basis of evidence from animallesion studies, where it has been claimed that the function of theperirhinal cortex can be doubly dissociated from that of the hip-pocampus (fornix/fimbria) on a spatial/nonspatial dimension. Be-cause a major thrust of the A&B target article is that this spatial/nonspatial dissociation reflects an episodic/recognition memorydichotomy it is appropriate that we examine this evidence care-fully.

Recent experimental findings demonstrate that, whereas le-sions of the hippocampus have minimal effects on object recogni-tion memory, lesions of perirhinal cortex produce marked deficitsin these processes. These findings, which are summarised in thereviews of Jarrard (1995) and Murray (1996), suggest that this as-pect of the dissociation is well founded. The evidence for the re-verse aspect of the putative double dissociation, that hippocampallesions disrupt spatial memory but perirhinal lesions do not, is,however, somewhat more equivocal. In the study of Gaffan(1994a), for example, monkeys were trained in a visually-guided,delayed match-to-sample task, and then received either perirhinalcortex lesions or fornix transections. The animals with perirhinal– but not fornix – damage suffered a decrement in performanceon this object memory task. When these animals were subse-quently trained in a spatial discrimination and reversal task, how-ever, fornix-, but not perirhinal-lesioned animals were deficient.Although at face value these findings appear to support the spa-tial/nonspatial dichotomy, it should be noted that these animals

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (21)

acquired the visual memory task prior to the lesion procedure, butacquired the spatial task in the post-lesion state. This procedureleaves open the possibility that the “spatial/nonspatial” dissocia-tion may actually reflect the differential effects of the two types oflesion on acquisition versus retention processes. We do not know,for example, whether the perirhinal cortex lesion would have pro-duced a “spatial” deficit if this task had also been acquired preop-eratively.

In studies by Ennaceur et al. (1996) and Aggleton et al. (1997)perirhinal cortex lesions did not produce a deficit when ratslearned a spatial, delayed nonmatch-to-position task conducted inthe T-maze. Although this finding appears to support A&B’s pro-posal, we have recently shown that excitotoxic lesions of perirhi-nal cortex will elicit a deficit in this T-maze task, provided that thememory retention interval is increased to beyond one minute (Liu& Bilkey 1998a). Critically, in the latter study it was demonstratedthat animals were unlikely to have used olfactory cues, or a simpleleft/right response rule, to solve the task. Rather, the most parsi-monious explanation of the results was that both lesioned and con-trol animals were solving the task by making reference to a spatialframework, and that lesioned animals displayed faster forgettingof this spatial information. In separate studies we have alsodemonstrated that lesions of perirhinal cortex disrupt perfor-mance in both radial arm maze (Liu & Bilkey 1998b) and watermaze (Liu & Bilkey 1998c) procedures. Although the former find-ing differed from that of Ennaceur and Aggleton (1997), raisingthe possibility that it was caused by damage to fibres of passage,we have recently observed similar deficits in animals with ibotenicacid lesions (Liu & Bilkey, unpublished data). It is more likely,therefore, that the differences stem from variations in methodol-ogy. In the study of Ennaceur and Aggleton (1997), for example,animals received three days of habituation prior to the onset oftesting. Each habituation session involved 10 minutes of maze ex-posure, during which time food was available at the end of eachmaze arm. Because this procedure is very similar to the actual ex-perimental protocol, it may have allowed time for the lesioned an-imals to develop a non-mnemonic strategy with which to solve thetask prior to the actual testing. It is interesting that when we useda habituation procedure during which food was scattered through-out the maze (thus providing less information about the subse-quent test) we observed deficits in the performance of perirhinalcortex-lesioned animals during the first few days of the testingprocedure (Liu & Bilkey 1998b). By the fourth day of the test se-ries, however, we found that lesioned animals began to use astereotyped, non-mnemonic strategy that appeared to allow theirperformance to improve to the level of controls.

In summary, recent findings indicate that lesions of the perirhi-nal cortex produce deficits in tasks that place demands on spatialmemory. Further, the results of Liu and Bilkey (1998a) indicatethat these deficits are not merely secondary to a deficit in objectrecognition. Because these findings suggest that spatial informa-tion may be processed within several subareas of the temporal re-gion (including both the hippocampus and rhinal cortex), it ap-pears that space/nonspace does not map neatly onto the episodic/recognition memory dichotomy proposed by A&B. An alternativeapproach to the spatial/nonspatial question is to downplay the lo-calisationist aspects of the distinction in preference to a task-ori-ented approach. This will require an understanding of how prob-lems with a spatial component are decomposed into subtasks, andwill involve investigations of the strategies that different brain re-gions adopt to generate solutions to these problems.

The gap between episodic memory and experiment: Can c-fos expressionreplace recognition testing?

Jan Bures and Andre A. FentonInstitute of Physiology, Academy of Sciences, 14220 Prague, Czech Republic. [emailprotected]

Abstract: The effort to identify the neural substrate of episodic recall,though ambitious, lacks experimental support. By considering the data onc-fos activation by novel and familiar stimuli in recognition studies, we il-lustrate how inadequate experimental designs permit alternative inter-pretations. We stress that interpretation of c-fos expression changes shouldbe supported by adequate recognition tests.

Aggleton & Brown (A&B) have organized the recent literature onthe functions of hippocampal-diencephalic-neocortical circuitryby fractionating recognition into two distinct processes: an effortfulelaborative reconstruction of contextual information (“remem-bering”) and a feeling of familiarity-based recognition (“know-ing”). It is important that they emphasize that whereas a hip-pocampal-diencephalic circuit subserves episodic encoding and“remembering,” a distinct perirhinal-medial dorsal thalamic cir-cuit underlies “knowing.”

Our commentary highlights the lack of experimental supportfor the neural basis of episodic memory. Clinical and experimen-tal evidence reveals that the perirhinal-medial dorsal thalamic axisis critical for “knowing”; however, without an animal model ofepisodic memory, evidence for the hippocampal-diencephalic cir-cuit’s role in “remembering” is restricted to clinical cases and theirinherent limitations. To make their case for the hippocampal-di-encephalic circuit, A&B decompose recognition into “remember-ing” and “knowing” and then argue that the hippocampal-dien-cephalic function can be dissociated from the “knowing” process.This does not mean, however, that the hippocampal-diencephalicaxis implements “remembering.” This can at best be supported byspeculating that spatial cells in the hippocampus (“place” and“view” cells), and “head-direction” cells in the anterior thalamusand mammillary bodies could contribute to episodic memory. Infact, directional cells have been found in related areas, includingthe lateral dorsal thalamus, posterior parietal, and retrosplenialcortices, and lesion studies suggest that the positional and direc-tional systems are independent (reviewed by Muller et al. 1996).The episodic memory hypothesis should be entertained, but onemust admit that spatial cells and many properties of the rodenthippocampus have been productively interpreted as a spatial nav-igation system (O’Keefe & Nadel 1978). An important merit ofA&B’s target article is that it draws attention to the need to char-acterize the relationship between episodic memory and spatial in-formation processing.

Evidence from the rat supporting the role of the perirhinal (andtemporal) cortex in “knowing” during the discrimination of noveland familiar objects is based on the increased expression of c-fosin these areas after viewing novel but not familiar stimuli. Thisweaker c-fos response to familiar objects contradicts the assump-tion that recognition is an active process that activates the circuitsmediating the “familiarity feeling.” The c-fos activation elicited bynovel stimuli more likely reflects the formation of new engramsrather than the readout of memories representing familiar stim-uli. Because the recognizable (i.e., familiar) objects elicit no sig-nificant reactions in the perirhinal and TE cortices there is no rea-son to believe that these regions contribute to “knowing” morethan other nonreactive brain structures.

Independent of using relative c-fos inactivity as evidence of“knowing” several experimental issues are critical for interpretingthese results.

(1) Using an ingenious within-subject design Zhu et al. (1996)simultaneously exposed one eye to the familiar stimuli and theother eye to the novel stimuli. Because the principle of eye equiv-alence also applies to the essentially crossed visual pathways of the

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (22)

rat, it is probable that the novel objects were recorded in bothhemispheres via direct and commissural projections. The absenceof c-fos activation in the perirhinal and TE cortices contralateralto the eye seeing the familiar objects, rather than indicating its rolein recognition, may indicate either that the callosally transmittednovel stimulus record is too weak or that its formation is preventedby the projection of the familiar stimuli, the processing of whichmay block the callosal input. These possibilities can be distin-guished by covering the eye exposed to familiar stimuli and thenexamining the effect of familiar or novel objects on the open eyewithout the masking effect of contralateral input.

(2) Should the two objects simultaneously presented to differ-ent eyes be interpreted as a complex scene, the scene would benovel for novel combinations of familiar objects. Zhu et al. (1996)are not explicit about such combinations and allow the reader toassume that the presentation of familiar objects to both eyes is ac-companied by (weak) symmetrical c-fos activation. Did the novelcombination of familiar objects, contrary to A&B’s model, elicitadditional c-fos activation? Wan et al. (1999) distinguish the “ob-ject novelty per se” from the “novelty arising from the mismatchof certain learnt associations in space.” However, the latter con-textual novelty elicited c-fos activity in various hippocampal fieldsas well as in perirhinal cortex.

(3) A recognition task should be used to study recognition, butthis is not done in the c-fos experiments. The rat was exposed tofamiliar and novel stimuli, but because the reward was indepen-dent of the animal’s response, it is possible that the rat habituatedto the behaviorally irrelevant stimuli. The differential c-fos activa-tion may therefore reflect habituation rather than recognition. Aconvincing experiment would relate brain changes to behavioralresponses showing that the animal can distinguish novel from fa-miliar stimuli (e.g., whereas familiar stimuli require a prolongednosing response to get additional reward, novel stimuli indicatethat no additional reward is available). The same considerationsapply to studying episodic memory.

Whereas the perirhinal-medial dorsal thalamic axis is clearly re-sponsible for familiarity-based recognition in primates, the case isnot so straightforward in rodents. Similarly, whereas humanepisodic memory is well characterized, it has proven difficult tostudy in animals. We hope that Aggleton & Brown’s target articlewill generate novel experiments clarifying the role of the hip-pocampal system in the mechanism of episodic memory.

What’s new in animal models of amnesia?

Rebecca D. Burwella and Howard Eichenbaumb

aWalter S. Hunter Laboratory of Psychology, Brown University, Providence,RI 02912; bLaboratory of Cognitive Neurobiology, Boston University, Boston,MA 02215. rebecca [emailprotected] [emailprotected] www.bu.edu

Abstract: In general, we endorse Aggleton & Brown’s thesis that the neu-roanatomy of amnesia comprises two functionally distinct systems, but weare disappointed in the lack of detail regarding the critical functional con-tribution of the hippocampus. We also take issue with the characterizationof the cortical areas surrounding the hippocampus, particularly the de-creased emphasis on the cortical input to the hippocampus.

Aggleton & Brown (A&B) present a reformulation of the neu-roanatomy of amnesia, one that focuses more on subcortical ef-ferents of the hippocampus and less on its cortical afferents thanprior models have done. They further suggest the view that twofunctional components of amnesia, familiarity judgment and rec-ollection, have different neuroanatomical substrates and can bedoubly dissociated. Whereas the reformulation is to be admiredfor the careful analysis of the neuroanatomy and function of dien-cephalic contributions to amnesia, we think a few points in theirmodel deserve further consideration.

On the one hand, we enthusiastically endorse the notion thatthere exist two functionally distinct components of the medialtemporal lobe memory system. We are also in agreement with thegeneral distinction between a simpler “recognition” function thatis sometimes sufficiently mediated by the cortex surrounding thehippocampus versus a more complex “recollective” function thatrequires the hippocampus. Indeed, these two points were the cen-tral argument of the BBS target article, “Two functional compo-nents of the hippocampal memory system” (Eichenbaum et al.1994). Furthermore, we have no argument with the anatomical ex-tension of the hippocampus component to include the dien-cephalic connections via the fornix. The idea that subcortical path-ways through the fornix are vital for normal hippocampal activitywas also a part of that BBS target article, and other earlier models(Mishkin et al. 1984; Papez 1937; Squire et al. 1993).

On the other hand, we were disappointed in A&B’s vaguenessregarding the critical functional contribution of the hippocampus,wherein they seem satisfied that “scene memory” and “spatialmemory” are equivalent to “episodic memory.” They are not, ofcourse, and it would help if they could elaborate on the funda-mental cognitive mechanism that relates all these notions. Theyshould include an explanation of why animals with selective hip-pocampal or fornix damage are so deficient on nonepisodic tasks,for example, learning the water maze (Eichenbaum et al. 1990;Morris et al. 1982) and learning relationships among odors (Bun-sey & Eichenbaum 1996; Dusek & Eichenbaum 1997).

We also take issue with A&B’s characterization of the corticalareas surrounding the hippocampus. In any formulation of thistype, boundaries must be drawn somewhere. A&B have chosennot to include the cortical regions surrounding the hippocampusin their model of hippocampal-diencephalic contributions to am-nesia. Given recent insights into the organization of the pathwaysby which sensory information is conveyed through the hippocam-pus, it seems to us a particularly inopportune time to de-emphasizethe cortical input to the hippocampus and related brain regions.In the rat and the monkey, the perirhinal and parahippocampal(postrhinal in the rat) cortices provide input to the hippocampalformation via the entorhinal cortex (Burwell & Amaral 1998b;Naber et al. 1997; Suzuki & Amaral 1994b). In both species, theperirhinal and parahippocampal cortices project to different sub-divisions of the entorhinal cortex. These regions also project di-rectly to the CA1 field of the hippocampus (Kosel et al. 1983;Suzuki & Amaral 1990). Preliminary observations indicate that, inthe rat, the perirhinal and parahippocampal cortices exhibit dif-ferent patterns of termination in the CA1 (Burwell & Amaral, un-published observations). Thus, there are multiple and parallelpathways by which sensory information is conveyed through thehippocampus (Fig. 1). This segregation of input to the hippocam-pus, both direct and via the entorhinal cortex, is especially inter-esting because the perirhinal and parahippocampal cortices re-ceive neocortical input from different unimodal and polymodalassociational regions (Burwell & Amaral 1998a; Suzuki & Amaral1994a). Taken together, the new anatomical findings suggest thatthe multiple, parallel pathways by which sensory information isconveyed to the hippocampus may have important functional im-plications for information processing in the hippocampus and be-yond. It seems that this circuitry should not be ignored in a refor-mulated model of hippocampal function.

The target article does not entirely ignore all cortical input to the hippocampus, only that directed through the entorhinal andpostrhinal cortices. A&B propose that, of the cortical regions sur-rounding the hippocampus, only damage to the perirhinal cortexcontributes to the core deficits observed in amnesia. In construct-ing their models, A&B have largely ignored the parahippocampaland entorhinal cortices, as if their interconnections with the perirhi-nal cortex and hippocampus have no functional relevance. In some sections, the perirhinal cortex and the collected cortical regions surrounding the hippocampus (perirhinal, parahippo-campal, and entorhinal cortices) are presented as being function-ally interchangeable. This is especially evident as A&B describe

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (23)

double dissociations between the perirhinal cortex with its subcor-tical efferents and the hippocampal-diencephalic region. In an ear-lier work, Eichenbaum et al. (1994) proposed that normal func-tioning of the hippocampus is dependent on its sensory inputarriving from the parahippocampal region. Thus, a double dissoci-ation of the parahippocampal region and the hippocampal regionwould not be possible. A&B take issue with this proposal, citing le-sion evidence that the perirhinal cortex and hippocampus can bedoubly dissociated. We are compelled to point out that experimen-tal lesion studies of the perirhinal cortex versus the hippocampusare not a test of the Eichenbaum et al. (1994) proposal. Because ofthe convergence of sensory information in the parahippocampal re-gion and the organization of the parallel pathways by which it is con-veyed to the hippocampus, substantial sensory information reachesthe hippocampus when the perirhinal cortex is damaged (Fig. 1).This input includes somatosensory and olfactory information arriv-ing via the entorhinal cortex, visual and visuospatial information viathe postrhinal cortex and the medial entorhinal area, and auditoryinput arriving via the postrhinal cortex (Burwell & Amaral 1998a;Suzuki & Amaral 1994a). Thus, an appropriate test of the Eichen-baum et al. (1994) proposal would compare the effects of hip-pocampal lesions with those of a combined lesion of the perirhinal,parahippocampal, and entorhinal cortices. The difficulty of makinga selective and complete lesion of this large cortical area should notbe underestimated, and indeed has never (to our knowledge) beenaccomplished.

Raising the profile of the anterior thalamus

John C. Dalrymple-Alford, Anna M. Gifkins, and Michael A. ChristieDepartment of Psychology, University of Canterbury, Christchurch, NewZealand. [emailprotected]{psyc002; amg55}@psych2.psyc.canterbury.ac.nzwww.psyc.canterbury.ac.nz/staff/jda/jda.htm

Abstract: Three questions arising from Aggleton & Brown’s target articleare addressed. (1) Is there any benefit to considering the effects of partiallesions of the anterior thalamic nuclei (AT)? (2) Do the AT have a separaterole in the proposed extended hippocampal system? (3) Should perirhinalcortex function be restricted to familiarity judgements?

Aggleton & Brown’s (A&B’s) target article provides a timely per-spective on the potential contributions to memory of the limbicdiencephalon because the relative influence of the anterior thal-amic nuclei (AT) and the dorsomedial thalamic nucleus (DM)have been a major source of conjecture in the recent literatureon the neuropsychology of memory. The focus on the AT is per-haps of special importance to the human domain because the hu-man AT, in particular, has more neurones than expected by rela-tive brain size or the number of sensory thalamic and corticalneurones (e.g., Armstrong 1986; Armstrong et al. 1987). As in therecent work by Aggleton and his colleagues (e.g., Steckler et al.1998b), the idea of an extended hippocampal-AT axis, which me-diates the acquisition and normal recall of new episodic informa-tion, is a commendable focus on neural networks rather than onisolated brain structures. Hence A&B’s approach provides a wel-come addition to our understanding of episodic memory and am-nesia. As with any new proposal, however, unanswered questionsremain and the evidence that supports the theory has limitations.Three of the issues stimulated by the target article are outlinedhere.

Contrary to earlier reports (e.g., Beracochea et al. 1989;Greene & Naranjo 1986), recent evidence from several researchgroups has now accumulated to show that AT lesions impair per-formance on tasks that are sensitive to hippocampal system le-sions (e.g., Aggleton et al. 1996; Beracochea & Jaffard 1995; By-att & Dalrymple-Alford 1996). We still need to be cautious aboutthis proposed relationship, however, as much remains to be doneto test the precise nature of the comparison. One important con-sequence of this new evidence concerns whether minor damageto the AT, for example within subcomponents such as the an-teromedial (AM) and especially the anteroventral (AV) nucleus,is sufficient to disrupt episodic recollection (sect. 5.1, para. 4).Although Aggleton and colleagues imply that such an effect re-quires substantial damage to the entire AT region, more minordamage can in fact cause clear impairments (Aggleton et al.1996; Byatt & Dalrymple-Alford 1996) even when inadvertentAT damage occurs as a consequence of an intended DM lesionthat otherwise has no effect on delayed nonmatching-to-position(Hunt & Aggleton 1991). That the same appears to be less trueof mammillary damage serves to highlight the importance of theAT region as a nodal point within the extended hippocampal sys-tem (Aggleton & Sahgal 1993). Conversely, there is little evi-dence that the laterodorsal nucleus deserves the same status asthe conventional parts (i.e., AM; AV) of the AT (see Mizumori etal. 1994; Warburton et al. 1997). Whatever the outcome of thelatter issue, we appear to have a situation where relatively minordysfunction to the AT especially may exacerbate the severity ofamnesia caused by an insult elsewhere in the system (for exam-ple, with alcoholic or pyrithiamine-induced damage; Beracochea& Jaffard 1991; Kopelman 1995; Langlais & Savage 1995). Whatfollows from this is the possibility that disruption to other partsof the hippocampal system can diminish the functional status ofthe AT region – or vice versa. Thus, an appreciation of the in-fluence of minor damage to the AT leads one to speculate thatsome recovery of mnemonic function might be produced by ma-nipulating changes to the functional status of either the hip-

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Figure 1 (Burwell & Eichenbaum). Wiring diagram showing theflow of information through corticohippocampal pathways in therat. Multiple parallel pathways result in segregation of cortical in-put to the hippocampus (depicted in light and dark grey). This or-ganization suggests specialization of function between these twosets of parallel pathways. Abbreviations: perirhinal (PR), postrhi-nal (POR), lateral and medial entorhinal areas (LEA and MEA),dentate gyrus (DG), CA fields of the hippocampus (CA1 and CA3),and the subiculum (sub).

Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (24)

pocampus or the AT, as key regions of an episodic recollectionsystem.

A related issue is whether the AT acts as a simple relay in an ex-tended hippocampal system or has a specific functional role in thecontext of encoding new information into episodic memory. Onthe basis of Gabriel’s (1993) recordings of training-induced neu-ronal activity, A&B speculate that the AT may be involved in latestages of acquisition in perhaps a different manner than that of thehippocampal system (sect. 9). We share their reservations with theshock-avoidance task employed by Gabriel, but the main point isthat the hippocampal system may be involved in consolidationprocesses that extend beyond a possibly more transient role for theAT. The problem for their model is that Gabriel’s evidence sug-gests an earlier, rather than different, role during training for theDM, and functional overlap between an anterior circuit (DM) anda posterior circuit (AT). It may be that under normal (intact brain)circ*mstances, the AT’s influence extends beyond just “episodicmemory,” but only the latter is dependent on normal AT-hip-pocampal function.

When recognition is based on familiarity, according to themodel, a separate neural system along a perirhinal-DM thalamicaxis is proposed as the principal neurological substrate. Thisknight’s move is another welcome break with tradition: The tem-poral/diencephalic dichotomy is discarded, and we now have anew dichotomy across both diencephalic and temporal lobe struc-tures. Of particular relevance to this idea is the relationship be-tween the hippocampus and the perirhinal cortex, which consti-tutes a major issue for their model, in which the precise role of theperirhinal cortex seems ambiguous (sect. 4.3). That the mnemoniccontributions of the two regions can differ substantially is usuallynot in question (but see Colombo & Gross 1996). The problem isthat, irrespective of lesion method, the effects of perirhinal lesionsmay not always neatly fit the suggested distinctions of recollection/familiarity, spatial/nonspatial (Liu 1998; Liu & Bilkey 1998b). In-stead, lesions here usually produce reliable delay-dependentdeficits as their core feature (Eichenbaum et al. 1994); this is nota characteristic of DM lesions even in object-recognition tasks(Hunt & Aggleton 1991; Parker et al. 1997). In fact, the distinc-tion between a perirhinal axis and an AT axis may be weaker thansuggested, as the perirhinal area receives direct AM efferents(Shibata 1993a). An alternative proposal is that perhaps there aremultiple, but overlapping, mnemonic systems in which theperirhinal cortex provides a critical temporal lobe interface forboth an AT-hippocampal system (encoding for contextual infor-mation and relational representations) and a DM-amygdala sys-tem (encoding the salience of item-specific information, subserv-ing relative familiarity of both individual items as well as specificemotional stimuli).

That old familiar feeling: On uniquelyidentifying the role of perirhinal cortex

M. J. EacottDepartment of Psychology, University of Durham, Durham DH1 3LE, England. [emailprotected]/staff/mje/mje.htm

Abstract: Perirhinal cortex contributes to judgements about stimulus fa-miliarity, but its role is far greater than this. Impairments on tasks that donot involve familiarity judgements attest to the fact that perirhinal cortexis involved in the greater role of knowing about objects, including, but notlimited to, their relative familiarity.

In their summary of the main features of their proposed model,Aggleton & Brown (A&B) contrast the recollection process, whichdepends on the hippocampal-thalamic system, with the detectionof stimulus familiarity, dependent on the perirhinal cortex. Muchof this idea will be familiar to some (Buckley & Gaffan 1998a;

Gaffan 1992a; 1994a; Parker & Gaffan 1998c), although the targetarticle brings together many facets of it in an extremely cogent ac-count of the hippocampal/perirhinal dissociation and the roleof the hippocampal-thalamic system. Yet it is less clear what theperirhinal cortex contributes to the memory processes. Under theheading, “Main features of the proposed model,” the role of theperirhinal cortex is described as “detecting stimulus familiarity”(sect. 2, para. 4). Later, it is acknowledged that some of the neu-rons within perirhinal cortex may “encode information concern-ing the physical characteristics of the stimuli and so may con-tribute to stimulus identification” (sect. 5.3, para. 4). Yet the roleof the perirhinal cortex is repeatedly described as “the detectionof the prior occurrence of a stimulus” (sect. 5.3, para. 7) or “fa-miliarity judgements” (Abstract). This view is easy to take on thebasis of the electrophysiological evidence that neurons in perirhi-nal cortex respond most vigorously to novel stimuli (Brown et al.1987). When it was demonstrated that lesions in this region re-sulted in deficits in learning or performing delayed match (or non-match) to sample (Eacott et al. 1994; Gaffan & Murray 1992;Horel et al. 1984: Meunier et al. 1993), it became easy to viewperirhinal cortex as a region that signalled novel stimuli. However,it is a leap from this evidence to the position that the neurons aresimply judging stimulus familiarity.

A neuron that responds more vigorously on first presentation ofa stimulus than subsequent presentations could be signalling thenovelty of the stimulus, but could equally be doing a number ofother things. Although the neurons do respond differentially tonovel and familiar stimuli, it is noteworthy that the visual qualitiesof the stimulus play an equal or greater role in the pattern of theresponse of these neurons (Li et al. 1993). Moreover, whereas theneurons that respond differentially to novel stimuli represent lessthan a third of the total population of perirhinal neurons in themonkey, the remaining neurons code the physical properties ofthe stimuli (Fahy et al. 1993b; Li et al. 1993). Thus, the familiar-ity of the stimulus is not the most important determinant of re-sponding in perirhinal cortex.

Equally, animals with lesions to perirhinal cortex are not alwaysimpaired in tasks that require familiarity judgements. Rhinal orperirhinal lesions may result in severe impairments in perfor-mance of trial-unique DMS (or DNMS) in monkeys (Eacott et al.1994; Gaffan & Murray 1992) and in rats (Mumby & Pinel 1994;Otto & Eichenbaum 1992a). However, Eacott et al. (1994) foundthat DMS performance could be completely unimpaired in mon-keys with rhinal lesions if the stimulus set used was very small. Thesame animals were severely impaired when trial-unique stimuliwere used, even when the delay component of the task was re-moved (Eacott et al. 1994). Thus, these animals were impairedwhen, and only when, novel stimuli had to be processed, not justwhen judgements regarding their familiarity had to be made. Oneinterpretation of this finding was that the impairment is one ofunique stimulus identification. With trial-unique DMS, the chal-lenge of uniquely encoding stimuli so that they can be discrimi-nated from a great number of potential foils is great and ever pres-ent. With the small-set DMS, the same stimuli repeatedly occurso that the identification problem is both less challenging, and,once solved, able to be repeatedly reused.

This interpretation of these data allows an explanation of otherbehavioural findings that cannot easily be reconciled with the viewthat perirhinal cortex is merely judging stimulus familiarity. Suchan explanation would predict unimpaired performance on tasksthat do not rely heavily on judgements of stimulus familiarity, forexample, discrimination learning and visual-visual associativelearning. Yet reports of impairments in associative learning (Buck-ley & Gaffan 1998b; Bunsey & Eichenbaum 1993; Murray et al.1993) and discrimination learning (Astur et al. 1995; Buckley &Gaffan 1997; Eacott 1998; Eacott et al. 1994; Gaffan 1994a;Gaffan & Murray 1992; Horel et al. 1987; Wiig et al. 1995) fol-lowing lesions in this region continue to appear. It is importantthat impairments in discrimination learning are seen where thereare a great many discriminanda (Gaffan 1994a), or where they can

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (25)

be visually confused (Buckley & Gaffan 1998c; Horel et al. 1987),thus providing a challenge to identify each one uniquely. Thusperirhinal cortex is not simply signalling that the stimulus has oc-curred before, but is reflecting knowledge about the object gainedin this previous encounter.

An area involved in producing a unique identification of stim-uli would be most active when stimuli are presented that have nocurrent representation in the system, that is, when novel stimuliare presented. The activity of such a putative area would reflectthe active processing of novel stimuli. Such activity could certainlyserve to signal novelty, yet it is not its primary purpose to do so.The presence of a majority of neurons in perirhinal cortex that donot actively engage in processing novel stimuli, but code the defin-ing physical characteristics of a stimulus, might be part of this sys-tem. The knowledge about stimuli held by such a system couldalso include associations with other stimuli (Buckley & Gaffan1998b; Bunsey & Eichenbaum 1993; Murray et al. 1993; Sakai &Miyash*ta 1991) and crossmodal associations that bind togetherdifferent aspects of the same stimulus event to form a full repre-sentation of an object.

The idea that perirhinal cortex contributes far more than thesimple detection of stimulus familiarity (Buckley & Gaffan 1998a;Eacott & Heywood 1995; Murray 1996) fits well within the frame-work outlined by Aggleton & Brown’s target article. The role ofperirhinal cortex is to identify objects as complete and unique en-tities so that they can be recognised.

Hippocampus, recognition, and recall: A new twist on some old data?

Jonathan K. FosterDepartment of Psychology, University of Manchester, Manchester M13 9PL,England. [emailprotected] www.psy.man.ac.uk

Abstract: This commentary attempts to reconcile the predictions of Ag-gleton & Brown’s theoretical framework with previous findings obtainedfrom experimental tests of laboratory animals with selective hippocampallesions. Adopting a convergent operations approach, the predictions of themodel are also related to human neuroimaging data and to other comple-mentary research perspectives (cognitive, computational, psychopharma-cological).

In their target article, Aggleton & Brown (A&B) present a timelyand clearly articulated framework for explaining the range ofepisodic memory impairments that can occur after neurologicaldamage. In so doing, they build and elaborate on the views origi-nally expressed by Delay and Brion (1969). They reinforce theemerging viewpoint that the hippocampus is closely involved inmediating relational or associative memory (see Mayes & Downes1997).

We can now reevaluate such data thanks to A&B. More specif-ically, A&B provide a valuable opportunity to reappraise some oldand hitherto rather puzzling data. Foster and Rawlins (1992a;1992b) reported two extended experiments in which, contrary toa number of then prevailing views of hippocampal memory func-tioning (for example, Olton et al. 1979; Rawlins 1985), rats withcomplete aspiration lesions of the hippocampus were able to per-form as well as controls on both unsignalled (Foster & Rawlins1992a) and signalled (Foster & Rawlins 1992b) Go/No-Go tasks.On both tasks, the performance of hippocampal animals was char-acterised by nonspecific, disinhibited responding (Gray 1982).However, across intertrial intervals ranging from 1 sec to 80 sec,hippocampal animals demonstrated unimpaired memory, evenwhen repeated stimuli were used (as in the signalled Go/No-Gotask; cf Cassaday & Rawlins 1995; see sect. 4.1, para. 5 of the tar-get article), and when interference was introduced between trials.At the time, these unexpected sets of findings were interpreted interms of lesion differences, task discrepancies, the adoption of un-

usual response strategies, and/or an interaction between thesefactors.

On the signalled Go/No-Go task, rats perform efficiently by de-termining whether the stimulus presented on the present trial isdifferent from the stimulus presented on the previous trial. Thiscan therefore be construed as a test of stimulus recognition. How-ever, on the unsignalled Go/No-Go task, rats ostensibly have to re-call whether responding was reinforced on the previous trial (orthe trial preceding that, if they are performing optimally and thecurrent trial is a Go trial), to determine whether to respond on thecurrent trial. Therefore, unless one attempts to explain perfor-mance on this unsignalled task in terms of familiarity (in whichcase, one wonders just which subtypes of recall can be mediatedvia familiarity processes and which cannot), it seems challengingto account for the data from the unsignalled Go/No-Go taskwithin A&B’s framework. (Taken together, these observations alsoindicate the importance of publishing statistically nonsignificantdata, about which journal editors and referees have been knownto be somewhat less than enthusiastic!)

In humans, the predictions of A&B’s framework would seem tobe consistent with at least some of our findings using structuralMRI to evaluate neurobehavioural relationships in the healthy el-derly and in patients with Alzheimer’s disease. Thus, Foster et al.(1997) noted a significant positive association between hippocam-pal volume and individual performance on delayed reproductionof verbal and nonverbal materials in a combined sample of healthyelderly and Alzheimer patients. In a second study, conducted byKoehler et al. (1998), in which a larger sample size permitted datafrom different participant groups to be analysed separately, theoriginal finding with verbal materials was replicated in Alzheimerpatients. However, a significant positive relationship was observedin Alzheimer patients in this second study between the volume ofthe parahippocampal cortex (but not the hippocampus) and de-layed visual reproduction (a nonverbal recall rather than a recog-nition task). This finding appears to be problematic for A&B. (Itis interesting that in the second study, a negative relationship wasobserved between hippocampal volume and memory perfor-mance in the healthy elderly when the data from these individu-als was analysed separately. This may be related to the degree ofneural pruning that has previously taken place [i.e., during child-hood and adolescence] in healthy old brains, although on the sur-face this finding would also appear to be problematic for at leastthe simple version of A&B’s framework.)

A&B’s theory satisfies at least two of the three “desiderata” re-cently specified when evaluating theoretical models of long-termmemory and amnesia (see Mayes & Downes 1997). Furthermore,the authors marshal an impressive array of data in support of thetheory: Their position is constrained by neuroanatomical knowl-edge, as well as relevant psychological data obtained from testingboth patients and laboratory animals. However, the authors areless sure-footed when considering the specific cognitive and com-putational component processes subserved by particular elementsof their proposed anatomical framework.

When considering recognition memory, the authors begin byassuming a trace strength model of recognition. However, I didnot find the final paragraph of section 6 very clear. In particular, Iwas not sure why, in the A&B model, “more extensive corticaldamage” would impair “both processes” (i.e., familiarity and rec-ollection), if recollection is deemed to be subserved by the hip-pocampus, rather than by cortical regions, and given the evidencecited in section 4.3 that the hippocampus is at least partially inde-pendent of anatomical inputs from the perirhinal region. It wouldalso have been helpful if A&B had made clearer the working def-initions they were using when referring in the final paragraph ofsection 6 both to the “recollection”/“familiarity” distinction (forwhich they state double dissociations exist) and to the “remem-ber”/“know” dichotomy (for which they imply that true doubledissociations cannot be demonstrated).

Moreover, A&B do not appear to be sure about whether theyregard “know” responding as a form of implicit memory (see

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (26)

sect. 6, end of para. 8, “the loss of ‘know’ responses in many am-nesics . . . which contrasts with . . . other forms of implicit mem-ory” [my emphasis], whereas in the second sentence of the sameparagraph, A&B state that “familiarity (a feeling of ‘knowing’) isan essentially explicit, conscious form of memory”).

The next paragraph is also not very clear, at least to this reader.A&B assert that damage to the extended hippocampal system “willresult in a loss of associative memory . . . and, hence, source mem-ory,” yet other researchers believe that (at least temporal) sourcejudgement is more closely associated with frontal lobe function-ing, and indeed we have some unpublished data obtained fromtesting schizophrenic patients to support this position (Foster etal., in preparation). A&B do go some way in acknowledging this“frontal” position at the bottom of section 7, paragraph 4. How-ever (cf sect. 7, para. 1), data obtained in Manchester indicate thatthe evaluation of trace strength is in fact a poor index for efficientperformance of at least some forms of stimulus recency judge-ment, and that contextual information needs to be encoded effi-ciently for these temporal judgement tasks to be performed well.The respective roles of the frontal lobes and structures within themedial temporal lobe and diencephalon in mediating recency andother forms of (temporal) source judgement clearly need to be ex-plored further in future research.

A&B usefully explore the structural and functional connectivitybetween medial temporal, diencephalic, and prefrontal regions.This approach seems especially worthwhile for those of us whohave been somewhat bemused in the past by the use of conceptssuch as inhibition as explanatory principles in both the hippocam-pal and frontal lobe literatures (see Fuster 1989; Gray 1982), with-out (at least regarding the concept of inhibition) much apparentcrossover or interaction between these two literatures to date. Iwould, however, have welcomed some reference to the importantrole of the prefrontal cortex in memory deficits such as confabu-lation (which is not an inevitable feature of amnesia), with respectto the postulated role of the prefrontal cortex in retrieval strate-gies (i.e., initiation and directions of memory search), and/or inthe monitoring and verification of search outcome (sect. 2). In-deed, taken as a whole, the focus of the model seems to be on en-coding of new episodic information, so that I was left to reflect onhow the hippocampal-anterior thalamic system was judged to in-teract with neocortical regions during reproduction (and consoli-dation?) of old, fully elaborated memories, and whether in theA&B model remote memories were thought to be hippocampus-dependent or -independent (cf Nadel & Moscovitch 1997; Squire1992). (Of note in this context, the frontal lobes have also beenlinked to the retrieval of remote memories [see Stuss et al. 1994].)It is also not clear to me whether the authors would predict a di-encephalic/perirhinal lesion effect that is equivalent in immediateand delayed recognition, or greater for the latter; nor am I clearon the time scale of the role attributed to the hippocampus in me-diating recall (in terms of forgetting rate and performance on mea-sures of immediate versus delayed recall).

There is now a developing consensus that different structureswithin brain circuits, such as the circuit of Papez, may mediate dis-sociable functional capacities: that is, even when there are pre-dominantly serial connections, individual structures or regionsmay be involved in separate and independent computational func-tions. A&B seem to support this general view (sect. 9: “[T]hesestructures are contributing something new and not merely pas-sively processing hippocampal outputs”), although they do notspeculate on the statistical algorithms or computational architec-tures embodied by the anatomical elements cited in their model.These details will need to be fleshed out in future simulations andexperimental investigations. In addition, when reflecting on therange of structures implicated by A&B and other researchers insubserving episodic memory, one is entitled to wonder where toplace the structural boundary or outer limit beyond “core mem-ory structures” (as ongoing anatomical projections become morediffuse), when considering the level of interconnectivity betweenbrain regions associated with amnesia.

More generally, I find it puzzling that in the past many re-searchers either tacitly or explicitly assumed that the anatomicalstructures within such neural circuits or loops are computationallyrelatively undifferentiated and apparently passively subserving acommon functional or computational goal. This view, which mayhave represented an inappropriate application of Lashley’s prin-ciple of equipotentiality, is now fortunately being questionedthrough the articulation of models such as that put forward byA&B. Similarly, we now accept that amnesia is not a hom*ogeneousentity, but reflects the family resemblance of a varied group of dif-ferent memory-related disorders, so that what we should be seek-ing to establish is a theoretically driven and empirically substanti-ated taxonomy and detailed explanatory framework of memorydisorders and phenomena. In this context, it is notable that in thesections of their target article dealing with human patients, A&Bfocus on single case studies to test the validity of their theoreticalframework, rather than adopting a syndrome-based approach to“organic amnesia” (see Parkin, 1997, for further examples of thesingle-case approach to amnesia).

In adopting a convergent operations approach, psychopharma-cological and neurochemical approaches, as well as neuroanatom-ical circuit mapping, must constrain our thinking about memoryand amnesia. For example, we have recently demonstrated thatthe administration of glucose (which may act via the hippocam-pus) significantly enhances performance on verbal recall but notverbal recognition in healthy young subjects (Foster et al. 1998).This finding would seem to be consistent with A&B’s framework.They acknowledge (sect. 9) that one “potentially important” inputto the anterior thalamic nuclei is the ascending cholinergic pro-jection from the tegmentum, although, given that the cited evi-dence is reliant on correlations, it is not entirely clear whether theauthors anticipate that these projections represent critical infor-mation-containing inputs or are more nonspecific and modulatory(see Rolls 1996). One way of addressing these sorts of issues in thefuture would be to implement neuroimaging investigations (suchas fMRI) in conjunction with precise time-locked EEG measure-ments.

More generally, anatomically driven theories such as that ad-vanced by A&B also need to explain the findings from neu-roimaging studies of healthy controls, which typically use cogni-tive challenge to probe the involvement of different brain regionsin episodic memory (see Foster, in press). Further developmentsof the model will therefore need to take into account these com-plementary perspectives, and in particular, consider the involve-ment of structures such as the precuneus and the prefrontal cor-tex, which have been associated with the mediation of long-termepisodic memory in several in vivo human neuroactivation stud-ies. By contrast, it has proven more challenging using these neu-roimaging techniques to demonstrate activation in the medialtemporal and diencephalic brain regions, on which the authors’ in-terest is focused in the target article. Conversely, the evidence ad-duced by A&B should demonstrate to computational and con-nectionist researchers that they cannot attempt to simulateepisodic memory effectively through computational modelling ofmedial temporal lobe structures alone, but that other regions andstructures also need to be considered and integrated (see Fosteret al. 1997).

Finally, A&B’s caution concerning cross-species comparisons ispertinent (see sect. 1, para. 2), but it also raises the spectre of theextent to which one can legitimately generalise across species,given the central role of language in many instances of humanepisodic memory. The issue of generalisation is, of course, alsocentral when (language-related issues aside) one is consideringthe extent to which anatomical hom*ology of brain regions mapsonto functional analogy across different species (e.g., from rats tohumans).

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (27)

What does the limbic memory circuit actually do?

Michael Gabriela and David M. Smithb

aDepartment of Psychology and Beckman Institute, University of Illinois,Urbana, IL 61801; bNeuroscience Program and Beckman Institute, Universityof Illinois, Urbana, IL 61801. {mgabriel; dmsmith}@uiuc.edu

Abstract: We applaud Aggleton & Brown’s affirmation of limbic dien-cephalic-hippocampal interaction as a key memory substrate. However,we do not agree with a thesis of diencephalic-hippocampal strict dedica-tion to episodic memory. Instead, this circuitry supports the production ofcontext-specific patterns of activation that subserve retrieval for a broadclass of memory phenomena, including goal-directed instrumental behav-ior of animals and episodic memory of humans.

Support marshaled by Aggleton & Brown (A&B) for the essentialinvolvement of hippocampal-diencephalic (H-D) interactions inrecollection (but not recognition) has potential to confer on thelimbic diencephalon its long-deserved (and widely ignored) statusas a legitimate participant in the limbic memory circuit. Yet, wefind problematic the main thesis: H-D dedication to episodicmemory. We argue instead that the H-D interactions subserve abroader mnemonic function, cue-, and context-encoding in theservice of retrieval. This function is used in multiple memory par-adigms including basic instrumental learning of goal-directed be-havior of animals and episodic memory.

Our studies of the neurophysiological bases of discriminativeinstrumental learning in rabbits (cited in the target article; Gabriel1993) indicate an essential involvement of the limbic dien-cephalon, and a clear relevance of H-D interactions to this learn-ing. The results are in accord with A&B’s general thesis that theH-D axis is involved in memory-relevant processing. We appreci-ate their acknowledgment of our work (sect. 9, “Final com-ments”). Yet, A&B were ambivalent concerning the relevance ofour data to their account, as indicated by their comment that “thebasic avoidance task is unlikely to provide a direct measure ofepisodic memory and hence is of limited application.” This arisesfrom the authors’ conviction that the hippocampus, clearly im-portant in mediation of episodic memory, is not involved in basicinstrumental learning. Admittedly, the hippocampus is not essen-tial for acquisition of discriminative avoidance behavior. However,it modulates the behavior as well as the task-relevant neuronal ac-tivity of the anterior thalamus, as indicated below.

Extensive recordings of neuronal activity during behavioral ac-quisition demonstrate unique topographic distributions of brieflatency, S1elicited neuronal activity across distinct nuclei of theanterior thalamus and the layers of posterior cingulate cortex. Cer-tain thalamic nuclei and cortical layers were maximally activatedby the S1 in the initial session of training, others in intermediatetraining stages, and still others as the rabbits attained asymptoticdiscriminative performance (Gabriel et al. 1991). The distribu-tions of activation changed systematically, not only across time(training stage) but also with respect to the spatial context. Thesame physical cues elicited different patterns of activation, de-pending on whether the subjects were engaged in a moderately-learned discriminative avoidance task, or (in a separate trainingapparatus) a well-learned discriminative approach task (Freemanet al. 1996). Thus, the distribution patterns coded both the spatialand temporal context. These context-specific patterns, elicited atbrief latency (80 msec) by the cue (S1) that calls forth the learnedresponse, are arguably the brain’s earliest sign (in the millisecondseries) of context-specific retrieval.

Other evidence suggests that the context-specific patterns de-pend on the integrity of hippocampal efferent flow to the limbicdiencephalon and cingulate cortex. Subicular, hippocampal, andentorhinal cortical lesions alter anterior thalamic and cingulatecortical task-related neuronal activity in ways that are likely to de-grade the value of the patterns as a spatio-temporal context code(Freeman et al. 1997; Gabriel et al. 1987; Kang & Gabriel 1998).Moreoever, the lesions impair contextual processing. Rabbits with

lesions failed to reduce avoidance responding when the familiartraining context was altered (reviewed by Gabriel 1993; see alsoFreeman et al. 1997). This failure of context-specificity was notsimply a “response inhibition” problem as the rabbits inhibitednormally when S1 quality was altered instead of the context(Gabriel 1993, p. 515).

These and many other findings implicate the hippocampus andthe anterior thalamus in mediation of context-specific retrieval ofdiscriminative, instrumental, goal-directed behavior of rabbits.We propose further that these areas are involved in the retrievalof episodic memory, a point on which we and the authors are inagreement. However, our account does not place responsibility forthe whole of episodic memory on the hippocampal-diencephaliccircuitry. The circuitry does not, in our view, store items or con-tents of memory as A&B seem to suppose (sect. 7). Rather, asso-ciative synaptic modifications are stored, which allow familiar andsignificant stimuli to elicit context-specific patterns of activation.The patterns, in turn, retrieve context-appropriate contents ofmemory. The contents of memory, be they goal-directed behav-iors or experiential recollections (episodic memories), are storedelsewhere in the brain (in different functional circuitries) and arerecalled by virtue of their association with the limbic retrieval pat-terns. Thus, the more general issue concerns the appropriate map-ping of components of a complex construct (e.g., episodic mem-ory) onto the brain’s distinct functional circuits.

Our view integrates many findings in addition to the foregoing.For example, the limbic diencephalon’s mediation of discrimina-tive avoidance behavior is time-limited. Retention of the behaviorwas severely impaired when anterior and MD thalamic lesionswere made after training to criterion, but no impairment wasfound when the lesions were made after 10 days of post-criterialovertraining (Hart et al. 1997). Just as for episodic and declarativememory, discriminative avoidance learning undergoes gradual“consolidation,” that is, a progressively acquired independencefrom limbic circuit processes. (Given the strong consensus thatepisodic memory is subject to consolidation, we were surprisedthat A&B did not explicitly address this issue.)

The fact that hippocampal lesions do not impair acquisition ofdiscriminative behavior (whereas they do impair episodic mem-ory) is not a compelling justification for regarding these forms oflearning as products of entirely separate memory systems. Dis-criminative avoidance behavior is acquired normally by subjectswith damaged hippocampi because original acquisition poses little challenge to the limbic retrieval circuit. Retrieval during acquisition does not require or use the exquisite thalamic spatio-temporal patterns endowed by hippocampal efferents for disam-biguating inputs or for response selection. It occurs instead as aresult of direct information flow from the limbic thalamus and cin-gulate cortex to areas involved in priming and execution of the lo-comotor response.

The absence of a hippocampus is noticed when there is poten-tial for substantial mnemonic interference, such as when multiple,similar tasks are learned concurrently, when retrieval cues are sim-ilar to cues associated with already-stored items, or when subjectsare required to use different configurations of contextual stimulifor selection among response alternatives. It is therefore not sur-prising that a hippocampal contribution to instrumental perfor-mance is detected when transfer-of-training tests are given, suchas testing in a novel context, training with cue-reversal, or requir-ing subjects to base their behavior on complex spatial configura-tions. Substantial intereference is commonplace, however, instudies of episodic memory, which typically involve complex stim-uli and multiple response items.

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (28)

Episodic memory in semantic dementia:Implications for the roles played by theperirhinal and hippocampal memory systems in new learning

Kim S. Graham and John R. HodgesMRC Cognition and Brain Sciences Unit, Cambridge, CB2 2EF, England.{kim.graham; john.hodges}@mrc-cbu.cam.ac.ukwww.mrc-cbu.cam.ac.uk

Abstract: Aggleton & Brown (A&B) propose that the hippocampal-ante-rior thalamic and perirhinal-medial dorsal thalamic systems play indepen-dent roles in episodic memory, with the hippocampus supporting recol-lection-based memory and the perirhinal cortex, recognition memory. Inthis commentary we discuss whether there is experimental support for theA&B model from studies of long-term memory in semantic dementia.

One of the central premises underlying Aggleton & Brown’s(A&B’s) reformulation of the neuroanatomy of episodic memoryis that the hippocampal-anterior thalamic and perirhinal-medialdorsal thalamic systems have dissociable functions: whereas theformer is critical for the efficient recall of true episodic informa-tion, the latter supports recognition memory based primarily onjudgement of familiarity. This provocative reformulation predictsthat patients with damage to the hippocampal-anterior thalamicversus the perirhinal-medial dorsal thalamic system will show adouble dissociation on tests of recognition- and recall-basedepisodic memory. In support of this hypothesis, A&B draw atten-tion to the three patients described by Vargha-Khadem and col-leagues (1997) who suffered isolated bilateral hippocampal dam-age early in life: All three patients showed disproportionatelybetter recognition memory, despite poor spatial, temporal, andepisodic memory. In support of the opposite neuropsychologicalpattern, A&B cite the patients described by Kapur et al. (1992;1994) who showed impaired performance on the faces version ofthe Recognition Memory Test (Warrington 1984), but normal re-call on tests of visual nonverbal memory.

A&B also suggest that data from patients with the neurodegen-erative disorder semantic dementia (Graham & Hodges 1997;Hodges & Patterson 1996; Hodges et al. 1992; Snowden et al.1989) are relevant to their model. Patients with semantic demen-tia provide memory researchers with a unique opportunity to in-vestigate the organisation of episodic and semantic memory in thehuman brain because the neuroanatomical damage is comple-mentary to that typically seen in patients with amnesia: focal atro-phy to the inferior and middle temporal gyri with preservation ofmedial temporal lobe structures, including the hippocampus (Ha-rasty et al. 1996; Mummery et al., in press).

Although A&B describe the reverse retrograde amnesia re-ported in semantic dementia (see below), they do not discuss thepotentially more relevant findings from studies of new learning insemantic dementia (Diesfeldt 1992; Graham et al. 1997; Warring-ton 1975). For example, recent studies from our research grouphave demonstrated spared recognition memory at early stages inthe disorder: Graham et al. (1997) showed normal two-alternative,forced-choice recognition memory using a real/nonreal object de-cision task in a group of five patients with semantic dementia com-pared to control subjects. A comparison group of amnesic patientsin the early stages of Alzheimer’s disease showed impaired recog-nition memory on the same test. A further study in a group of eightpatients with semantic dementia has replicated this finding (Gra-ham et al., submitted): A double dissociation between perfor-mance on a three-alternative, forced-choice recognition memorytest was found in patients with semantic dementia (preservedmemory) and early Alzheimer’s disease (impaired memory). De-ficient recognition memory was only demonstrable in semanticdementia if “unknown” targets seen at study (i.e., stimuli that wereno longer familiar to the patient as measured by picture namingand/or word-picture matching) were replaced by a different ex-ample of the same item in the recognition memory task (i.e.,

changing a round-dial telephone seen at study with a touch-but-ton phone at test). We interpreted this result as evidence for mul-tiple inputs to medial temporal lobe structures, and suggested thathigher-order perceptual information is sufficient to supportrecognition memory in semantic dementia, even when the item tobe remembered is no longer familiar to the patient.

The relevance of the neuropsychological data from semanticdementia to A&B’s model hinges on whether there is pathologicalinvolvement of the perirhinal cortex in these patients. If patientsdo possess damage to this area of the temporal lobe, the preser-vation of recognition memory in semantic dementia would be con-trary to much of the animal literature in which perirhinal damageresults in impaired recognition memory, even for perceptuallyidentical stimuli. On the other hand, if the perirhinal cortex is rel-atively unaffected in the disease, the preserved recognition mem-ory seen in such patients would concur with A&B’s model. Clearlythere is a need for accurate quantitative structural neuroimagingin semantic dementia to demonstrate preservation or damage toperirhinal, entorhinal, and hippocampal areas. The exact locationof the perirhinal cortex in man remains somewhat controversial,however. Current neuroanatomical studies indicate that whereasthe rostral portion occupies the banks of the collateral sulcus, thecaudal part extends onto the medial temporal pole (Corkin et al.1997; Van Hoesen 1997). Clinical MRI studies in semantic de-mentia suggest consistent atrophy of the temporal pole but rela-tive sparing of the parahippocampal gyrus and collateral sulcus.Preliminary findings from a voxel-by-voxel structural analysis ofregions of significant difference in grey matter density in six pa-tients with semantic dementia, compared to a control group, alsoindicate that the pathology may not extend into either the perirhi-nal or entorhinal cortices (Mummery, unpublished data).

There is reason to believe, therefore, that further neuropsy-chological studies of new learning in semantic dementia, com-bined with detailed structural analyses of individual areas withinthe patients’ temporal lobes, will provide additional support forA&B’s model. A further question remains: How does the data onnew learning in semantic dementia equate with A&B’s distinctionbetween familiarity- and recollection-based episodic memory?The integrity of recognition memory, considered in conjunctionwith the preservation of recent autobiographical memories, sug-gests that both types of episodic memory may be preserved earlyin the disorder. Of course, we are not suggesting that new learn-ing remains normal in semantic dementia throughout the courseof the disease. In fact, three of the patients described in Grahamet al. (submitted) showed a significant, although mild, impairmenton recognition memory for perceptually identical stimuli (see alsoGraham et al. 1997). On the basis of A&B’s proposal about the roleof the perirhinal cortex in recognition memory, there may be aclear stage in the disease when perirhinal-dependent recognitionmemory may be disproportionately more affected than the recol-lection of temporal and spatial information (as demonstrated inKapur et al.’s [1992; 1994] patients).

Turning now to the issue of retrograde amnesia: As cited by A&B,experiments have demonstrated that patients with semantic de-mentia show a reverse Ribot gradient, whereby there is better re-trieval of recent memories compared to those from the more dis-tant past, on tests of autobiographical memory and knowledge offamous personalities and public events (Graham & Hodges 1997;Graham et al., in press; Hodges & Graham 1998; Snowden et al.1996). These studies highlight the time-limited role played by thehippocampal complex in the encoding and storage of autobio-graphical and semantic memories: Typically, patients with semanticdementia are able to produce autobiographical memories and se-mantic knowledge only for the last two to three years of their lives(Graham & Hodges 1997; Hodges & Graham 1998). Although A&Bhave undertaken a comprehensive review of anterograde amnesia,it is not clear from their target article how their reformulated modelcorresponds to the double dissociations seen on tests of remotememory in patients with selective damage to hippocampal or neo-cortical areas of the brain. More specifically, how do the different

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (29)

memory systems (perirhinal versus hippocampal) contribute tolearning of semantic and episodic memories? We presume thatA&B would adopt a position similar to that of Vargha-Khadem andcolleagues (1997), who suggested that the perirhinal and entorhinalcortices can support context-free, long-term semantic learning (asseen in their three developmental cases), but that the hippocampalsystem is critical for normal acquisition and storage of context-dependent autobiographical memories.

In summary, despite our current lack of understanding con-cerning the next neuroanatomical pattern of damage in semanticdementia, and the neuropathological evolution of the disease overtime, it seems that there is preliminary support from semantic de-mentia for Aggleton & Brown’s model. Furthermore, there areclear theoretical predications regarding the integrity of recogni-tion and recollection memory over the course of the disease thatcould be addressed in further experimental studies. The authorsare to be congratulated on a clear and compelling review of thesometimes confusing and contradictory literature on anterogradeamnesia in humans and animals.

Retrieval dynamics and brain mechanisms

Douglas L. HintzmanDepartment of Psychology, University of Oregon, Eugene, OR [emailprotected]

Abstract: Experiments on memory-retrieval dynamics support the hy-pothesis that different mechanisms mediate processes of familiarity andrecall (recollection). The minimal retrieval time for accurate recognitionand frequency judgments is about 100 msec earlier than the minimal timeto judge details such as modality or location. The difference is consistentwith neurophysiological evidence. Retrieval dynamics data are also rele-vant to other aspects of Aggleton & Brown’s target article.

Aggleton & Brown’s (A&B’s) proposal that different brain circuitsmediate judgments of familiarity on the one hand, and the recallor recollection of experiential details on the other, may help ex-plain certain experimental findings on the dynamics of memoryretrieval in normal humans. In the response-signal (or speed-ac-curacy tradeoff) method, the experimental subject is shown a re-trieval cue such as a word, and after a variable lag is given a signaldemanding an immediate binary (yes-no) response. The responsemight indicate whether the cue is old or new, or reflect some otherdecision about the cue that is based on memory. If the signal to re-spond comes at a very short lag after cue onset, decision accuracyis at chance (hit and false alarm rates are the same). If the signalcomes after a long lag, accuracy is at asymptote. In between, ac-curacy typically traces out a curve that rises from chance abruptly,and can be closely fit by an exponential function. The point atwhich the function first starts to rise above chance (the intercept)is an estimate of the minimal time required to retrieve and evalu-ate relevant information and execute a response.

Several experiments have compared the response-signal func-tions from different memory-judgment tasks, holding subjects andmaterials constant. In Experiment 2 of Hintzman et al. (1998), forexample, subjects saw two successive word lists. Some words ap-peared once and some twice in the same list. Next, subjects weretold whether the response-signal tests to follow would require oldversus new decisions (recognition) or list 1 versus list 2 decisions(list discrimination). Figure 1 shows discrimination performancein the two tasks, as a function of processing time (lag plus responselatency). Whereas asymptotic performance was affected by boththe task and the number of repetitions, the intercept was deter-mined primarily by the task. Specifically, for about 100 msec earlyin the retrieval episode, information was available that could beused in making accurate recognition decisions, but not in judgingwhether the test word came from list 1 or list 2.

The response-signal method has been used to compare other

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Figure 1 (Hintzman). List-discrimination and recognition-memory performance as a function of lag plus mean reaction time (Hintz-man et al. 1998); dL 5 accuracy; 1x 5 1 study trial; 2x 5 2 study trials.

Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (30)

discrimination tasks with recognition memory, with similar results.Recognition intercepts are shorter for recognition than for judg-ing the auditory versus visual study modality of a word (Hintzman& Caulton 1997), for judging whether the test word is in the samespatial location in which it was studied (Gronlund et al. 1997), andfor detecting whether the word’s plurality (FROG vs. FROGS) haschanged on the test (Hintzman & Curran 1994, Exp. 3). By con-trast, frequency judgments may be based on the same informationas recognition. In Hintzman and Curran (1994) Experiment 1, theintercept for judging whether a test word had been seen one timeor two times was as early as that for judging whether a test wordwas old or new. With the exception of frequency judgments, eachof these memory-judgment tasks has an intercept significantlylonger than that of recognition memory. The size of the obtaineddifference is usually around 100 msec.

The early and late intercepts may reflect minimal retrieval timesfor the two memory circuits proposed by A&B (see also Eichen-baum et al. 1994). The perirhinal-thalamic circuit supports retrievalof familiarity, and could mediate judgments of both recognition andfrequency. The hippocampal-thalamic circuit supports recall of thekinds of contextual and structural information that would be neededto make judgments of list membership, modality, spatial location,and plurality. Retrieval from the hippocampal system might be ex-pected to take a longer time, because it is computationally andanatomically more complex. Studies using monkeys indicate that afamiliarity decrement in neural responding first shows up at 70–80msec after stimulus onset in inferotemporal cortex (Miller et al.1993), and at 140–260 msec in hippocampus (Rolls et al. 1993).Comparisons across species (and across labs) can be problematic,but the difference between these figures is intriguingly similar tothat between the response-signal intercepts that I have described.

Incidentally, response-signal results also suggest that the famil-iarity that underlies recognition judgments should not be identifiedtoo closely with perceptual fluency. For purposes of timing control,response-signal test cues are presented visually. When some testwords were studied auditorily and some were studied visually – ex-actly matching the test cue – no difference was found in their recog-nition intercepts. Although this is a null result, it has been replicatedat least three times (Hintzman & Caulton 1997; Mulligan & Hirsh-man 1995). It suggests that perceptual fluency caused by exact stim-ulus repetition is not an important factor in recognition.

Enhanced fluency does show up, however, in the response-signalintercept for lexical decisions – particularly in the case of low-fre-quency words (Hintzman & Curran 1997). The lexical-decision in-tercepts for low-frequency words that were primed by repetition,and for primed and unprimed high-frequency words, are signifi-cantly earlier than that for recognition. The recognition interceptseems to be unaffected by study modality, repetition, or frequencyof the word in the language, all of which affect priming or fluency.These retrieval differences between recognition and lexical decisionsuggests there is more than one kind of familiarity, a conclusion thatis also consistent with the literature reviewed by A&B.

Neuropsychological assumptions and implications

Narinder KapurWessex Neurological Centre, Southampton General Hospital, Southampton,England S016 6YD. [emailprotected]

Abstract: Some of the assumptions underlying the arguments in Aggleton& Brown’s target article are reviewed; discrepancies/predictions arepointed out in relation to human lesion studies. A&B’s proposal is inter-esting, but it may require harder, confirmatory evidence before it can beconsidered to be all-encompassing.

Aggleton & Brown’s (A&B’s) is an interesting and stimulating tar-get article. However, the authors seldom make it clear whether

and to what extent they are relying on human neurobiologicalfindings as bases for their arguments. For example, in the case ofneuroanatomy, Crick and Jones (1993) have powerfully pointed toour ignorance in this field. In addition to species differences in hu-man morphology, there may also be possible differences in bloodflow, neurochemistry, anatomical connectivity, inhibitory-excita-tory harmony, and so forth. In an article such as A&B’s there is in-evitably a jumping from one species-platform of data to another,and it is important to note the assumptions that go along with suchconceptual leaps.

One of our problems in amnesia research is that we do notknow, or at least have not agreed on, what we mean by amnesia.That is, there is no agreed set of criteria by which we can classifya patient as amnesic, or – more importantly – give an indicationof the severity of amnesia. This applies both to anterograde andretrograde amnesia. A further issue is of course the purity of am-nesia. This is also important, though not as difficult a problem toresolve. To take one or two examples from the types of cases re-viewed by A&B: the patient RB (Zola-Morgan et al. 1986) and ourown patient BJ (Kapur et al. 1994) were both moderately amnesicby most standards, and our own patient has in fact gradually improved over the years so that he might now be classified as “mildly amnesic.” Can one build the same theoretical argumentsaround these patients, and include them in the same database aspatients such as HM (Corkin 1984) or CW (Wilson et al. 1995)who are by all accounts among the most severely amnesic patientsever seen?

In the case of the mammillary bodies, it is – as A&B point out– very difficult to find cases of discrete, isolated damage to thesestructures in the human domain. We have recently come acrosstwo cases of suprasellar tumours (Kapur et al. 1999) where thereappeared to be minimal additional damage apart from that to themammillary bodies. In both cases, the memory loss was moderaterather than severe.

Similarly, there are few human cases of discrete, isolated lesionsof the rhinal cortex. There is one paper, not cited by A&B (Yonedaet al. 1994), which pointed to a role for the rhinal cortex in retro-grade but not anterograde memory. In the case of the thalamiccomponent of their argument, the authors should note and incor-porate negative cases of thalamic damage but normal memory,such as that of Kritchevsky et al. (1987).

I would have liked to see a few comments about whether in thelimbic-diencephalic memory system proposed by A&B we aredealing with an “analogue” system or a “digital” system. That is,is it that the greater the number of components in the system thatare damaged, the more severe the amnesia, or is there a thresh-old point after which amnesia suddenly develops and remains atthat level of severity regardless of additional damage? Similarly,is the pattern of memory loss (e.g., recall versus recognition) thesame after each component is damaged, or do different compo-nents of the limbic-diencephalic system have a unique, distinc-tive contribution to make? The answers may be present in the tar-get article, but they do not come across as clearly as one wouldhave liked.

If A&B are arguing that the perirhinal-medial dorsal thalamicsystem selectively subserves familiarity recognition memoryjudgements, with little role in recall, then would one predict hu-man cases with damage to this system to show impaired recogni-tion memory (or perhaps impaired priming) but intact recallmemory? In the few studies purporting to show such an unusualpattern of memory loss (e.g., Delbecq-Derousne et al. 1990), thepathology has mainly been in or near the frontal lobes.

The single-cell recording studies reviewed by A&B provide fas-cinating clues as to the bases of memory processes such as famil-iarity recognition memory, but it is important to remember thatwithout simultaneous multiple recordings elsewhere in the brainit is not possible to be certain about the significance of these spe-cific sets of recording data, in particular, the stage of any sets ofneuronal firing within an overall network of related cell assem-blies.

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (31)

Perirhinal cortex and hippocampus mediateparallel processing of object and spatiallocation information

Raymond P. KesnerDepartment of Psychology, University of Utah, Salt Lake City, UT [emailprotected] www.utah.edu

Abstract: An alternative to Aggleton & Brown’s interpretation is presentedsuggesting that the perirhinal cortex and hippocampus mediate differentattribute information, but use the same processes, supporting the idea ofparallel processing based on attribute (visual object and spatial location)rather than process characteristics (item recognition and familiarity).

Aggleton & Brown (A&B) have provided a scholarly review, in-cluding relevant data from both the animal and human literatureto support the idea that the perirhinal cortex and the hippocam-pus mediate two different systems labeled as item recognition andfamiliarity. The distinction is based on assumed differences in pro-cessing of information. An alternative interpretation of the data isthat the differences between the hippocampus and perirhinal cor-tex reflect the involvement of the hippocampus in processing spa-tial attribute information and reflect the involvement of theperirhinal cortex in processing of visual object information. Tosupport the aforementioned alternative, one needs to questionwhether it is mandatory to apply the concept of item recognitiononly to visual object information and not to spatial location infor-mation. The concept of item recognition has been more readilyapplied to visual object information, because one can identify thecritical stimulus to be remembered more easily than is the case forspatial locations, which are often assumed to be based on rela-tional representations. I would suggest, however, that both visualobject and spatial location information are represented by an in-tegration and relational representation of many features and thusone should be able to apply the concept of item recognition to spa-tial locations as well as to visual objects. Even Aggleton has statedin a different article that spatial location recognition memory canbe as important as visual object recognition memory (Steckler etal. 1998a). Also, familiarity judgments are not likely to depend onone specific circuit; rather, they may represent a process that sub-serves different attributes of memory. To support this idea I haveused a continuous recognition memory paradigm using either spa-tial locations or visual objects. This is an ideal paradigm in that onecould easily point to the operation of familiarity as well as itemrecognition.

For the visual object version, the continuous recognition mem-ory task involves sequential presentation of 12 three-dimensionalobjects within a session selected from a set of 120 objects. Fromthe 12 objects, 8 were novel and 4 of the 8 were presented twice.Repeated objects have lags ranging from 0 to 4 (from 0 to 4 dif-ferent objects are presented between the first and the repeatedpresentation). Rats are reinforced for approaching the novel ob-ject, but they are not reinforced for a repetition. Rats learn thistask quickly and show significantly longer latencies to approachand move the repeated compared to the novel objects (familiar-ity). Furthermore, the shorter the lag the greater the latency dif-ference.

For the spatial version, the continuous recognition memory taskinvolves sequential presentation of 12 maze arms within a session.Of the 12 presentations, 3 or 4 of the arms are repeated. Rats arereinforced for approaching the new arm for that session, but theyare not reinforced for a repetition. Repeated arms are presentedwith lags ranging from 0 to 6 (from 0 to 6 different arms are pre-sented between the first and the repeated presentation). Ratslearn this task quickly and show significantly longer latencies toapproach the repeated arm compared to the nonrepeated arm (fa-miliarity). Furthermore, the shorter the lag the greater the latencydifference.

It is important to point out that these two tasks are analogous,so that comparisons can be more readily made concerning the op-

eration of familiarity and item recognition. The results indicatethat perirhinal cortex lesions disrupt visual object-based continu-ous recognition memory across all lags, and that hippocampal le-sions disrupt spatial location continuous recognition memoryacross all lags, but hippocampal lesions do not impair visual objectcontinuous recognition memory across any lags (Jackson-Smith etal. 1993; Kesner et al., submitted). Because the deficits appear forall lags, the data suggest that the perirhinal cortex supports short-term visual object recognition memory as well as visual object fa-miliarity and that the hippocampus supports short-term spatial lo-cation recognition memory as well as spatial location familiarity.Thus, this alternative interpretation is based on the assumptionthat there are separate neural circuits for memory for spatial lo-cation (hippocampus and interconnected neural circuits) andmemory for visual objects (perirhinal cortex and interconnectedneural circuits). Yet these two neural circuits are part of the samememory system, which has been labeled as data-based memory ordeclarative memory.

Further support for this alternative interpretation is based onthe assumption that specific operational characteristics are asso-ciated with each attribute as part of the data-based memory sys-tem. These include a number of processes, such as pattern sepa-ration based on selective filtering or attenuation of interferenceassociated with temporary memory representations of new infor-mation, short-term memory or working memory of new informa-tion, short-term memory or working memory of new information,consolidation or elaborative rehearsal of new pattern associationinformation, and retrieval or pattern completion of new informa-tion based on flexibility and action. It can be shown that the hip-pocampus mediates pattern separation, short-term memory, con-solidation, and retrieval of new spatial information, whereas theperirhinal cortex can be shown to mediate pattern separation,short-term memory, consolidation, and retrieval of new visual ob-ject information (for a review see Kesner 1998). Thus, the twoneural regions (perirhinal cortex and hippocampus) mediate dif-ferent attribute information, but use the same processes, sup-porting the idea of parallel processing based on attribute ratherthan process characteristics.

Recall, recognition, and the medial temporal lobes

Barbara J. KnowltonDepartment of Psychology, University of California, Los Angeles, LosAngeles, CA 90089. [emailprotected]

Abstract: Aggleton & Brown predict that recognition and episodic recalldepend on different brain systems and can thus be dissociated from oneanother. However, intact recall with impaired recognition has not yet beendemonstrated if the same type of information is used in both tests. Cur-rent evidence suggests that processes underlying familiarity-based recog-nition are redundant with processes underlying episodic memory.

Aggleton & Brown (A&B) put forth an intriguing view that placesmedial temporal lobe structures subserving declarative memoryin their anatomical context. Their view makes a very straightfor-ward prediction, that recall and recognition are independent andcan therefore be doubly dissociated from one another in neu-ropsychological patients. Although it is clear that there are pa-tients who exhibit deficits in recall in the face of normal recogni-tion, it is not yet clear that there are patients who can consistentlyrecall items despite an inability to recognize them.

It is well known that patients with prefrontal damage exhibitdeficits in free recall despite good recognition abilities (see Shi-mamura 1995 for a review). These patients appear to have diffi-culty with the strategic aspects of recall – effectively generatingcues and using them to build a memory. This deficit seems espe-cially pronounced for episodic memories, which by definition are

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (32)

tied to a particular place and time. The deficit frontal patients ex-hibit on tests of source memory or memory for temporal order canalso be seen as related to a specific deficit in episodic memory. De-spite these deficits, prefrontal patients perform well on tests ofrecognition, especially forced-choice, and thus differ from pa-tients with medial temporal lobe damage who exhibit deficits inboth recall and recognition. The fact that prefrontal damage leadsto specific recall deficits is a concern for studies comparing recalland recognition, because many common etiologies of amnesia,such as Korsakoff ’s disease, head trauma, and rupture of the an-terior communication artery, lead to prefrontal damage. I agreewith A&B that the hippocampus is critical for episodic recall, andI would add that prefrontal cortex is also a crucial part of this sys-tem.

To support the idea that recognition deficits can occur despiteintact recognition, A&B present data from nonhuman primatestudies demonstrating double dissociations between objectrecognition and spatial memory (sect. 4.3). However, not only dothese tasks differ in terms of recognition versus recall, they alsodiffer in terms of visual object versus spatial memory. Perhaps thisdissociation could be viewed in terms of the type of informationlearned, and not the mnemonic process involved.

The data from neuropsychological patients also fails to makethe case that recognition deficits can exist alongside intact recall.A&B describe patients with focal temporal cortical atrophy thatdisplay preserved episodic memory but have deficits in seman-tic memory. However, the dissociation in these patients is be-tween remote memories (both episodic and semantic) and newlearning (Graham & Hodges 1997). In addition to intact episodiclearning, these patients exhibit intact recognition, consistentwith the idea that they are losing stored memories yet are ableto form new memories. In other cases, because of its role in ven-tral visual stream processing, temporal lobe damage may resultin visual agnosia that could impair visual recognition. However,it has not been clearly demonstrated that these patients wouldbe able to recall newly acquired memories of the same type ofinformation used in recognition tests. The deficit is not one ofrecognition per se, but a more basic deficit in visual processing.For example, a prosopagnosic patient would have difficulty rec-ognizing faces, but no problem with the recall or recognition ofwords.

A&B make the point that recognition itself is composed of mul-tiple processes (sect. 6). Individuals may recognize items becausethey are consciously recollected (Remembered), a process thatshares many properties of episodic recall. Conversely, subjectsmay recognize items based on familiarity, in the absence of spe-cific recollections (Knowing). A&B predict that these two types ofrecognition could be doubly dissociated because they depend onindependent brain systems. However, the evidence to date on theremember/know (R/K) distinction suggests that the relationshipbetween the two is more likely to be redundancy than indepen-dence. First, although R response accuracy can be manipulatedwithout affecting K response accuracy, changes in K response ac-curacy are always accompanied by changes in R response accu-racy. When a manipulation does affect K responses, it tends to in-crease both hits and false alarms, suggesting that it affects only theresponse threshold, not memory (Rajaram 1993). Knowing hasonly been manipulated independently of remembering undervery shallow study conditions when subjects must make rapidjudgments and memory is weak (Gregg & Gardiner 1994). Underthese circ*mstances, knowing may reflect perceptual fluency and,like priming, may not depend on the medial temporal lobe at all.

Remembering and knowing were first conceptualized as indicesof episodic and semantic memory, and thus knowing would be re-dundant with remembering (Tulving 1985a). One could not re-member an episode without also knowing the component factsthat make up that episode. The fate of items receiving R and K re-sponses over time argues for redundancy as well. In this study,subjects were tested after a 10-minute delay, and again after oneweek (Knowlton & Squire 1995). For each subject, we found that

a large proportion of items receiving R responses at 10 minuteswent on to receive K responses after 1 week. There was virtuallyno conversion of K items to R items over the week. Also, the R-to-K conversion rate was so great that it could only be accountedfor if all items that were given an R response also had the capac-ity to elicit a K response (Knowlton 1998).

If recall or recollection cannot occur independently of pro-cesses underlying familiarity, it suggests that the brain structuresthat subserve recognition are also critical for encoding for subse-quent recall. However, it is likely that different subsystems of themedial temporal lobe subserve memory for different modalities(visual/object vs. spatial/contextual). Recall, especially episodicrecall, depends additionally on other brain systems, including pre-frontal cortex, that are important for the reconstruction of mem-ories from one’s personal past.

Consideration of the drive properties of the mammillary bodies solves the “fornix problem”

E. E. KrieckhausDepartment of Psychology, New York University, New York, NY [emailprotected]

Abstract: Fornix problem: Why do lesions of the fornix, which connectsthe hippocampus (HF) to the medial mammillary nucleus (MMN), oftencause no deficits in tasks severely affected by lesions of HF or MMN? So-lution: The direct HF feedback to antero ventral (AV) thalamus (MMN⇒ AV ⇔HF), which is blocked by MMN lesions but not fornix lesions, is suf-ficient for nonscene-relevant consolidation.

As a long-time proponent of the importance of the medial mam-millary nucleus (MMN) in higher cognitive processes (Krieck-haus 1964), I was pleased to see the conclusion of Aggleton andBrown (A&B) that MMN is necessary for explicit consolidation ofmemory. As they acknowledge, if one argues that MMN is impli-cated in consolidation, either because MMN receives its only cog-nitive input from hippocampus (HF) or because of MMN pathol-ogy in Korsakoff psychosis, the crucial fornix problem emerges:Why are there often no consolidation deficits with lesions of fornixthat connects HF with MMN?

Because of fractionation of disciplines, the intrinsic importanceof MMN in terms of physiological psychology and drive (D) hasbeen overlooked. More generally, as A&B recognize, it is currentlydifficult to conceptualize how MMN plays any definitive role incognition. I hope to show that only by the resurrection of the land-mark work of the great American psychologist Clark Hull (1943)can the role of MMN in cognition in general, and consolidation ofmemory in particular, be understood. Specifically, to incorporatedrive (D) into physiological psychology we must use Hull’s equa-tion (E 5 H 3 D), which synthesizes cognitive habit strength (H)and motivational D, to compute the strength of an action/reactionpotential (E). By considering the hypothalamic nature of MMN asmediating Hull’s D, the difficulty in understanding the function ofMMN in consolidation of explicit memory is clarified in a way thatmay solve the fornix problem.

1. Neural instantiation of Hull’s learning equation. In Figure1, H, as long-term memory, is instantiated as neural connectionsin parietal temporal occipital cortex (PTO). Furthermore, H islargely focused into MMN, the hypothalamic node of the Papezcircuit (Pap) loop: subiculum (SUB) of HF ⇒ MMN ⇒ AV ⇒ cingu-late ⇒ SUB (Krieckhaus et al. 1992; Ungerleider & Mishkin 1982).E is instantiated in cingulate cortex, which directs action at all lev-els of the motor system (Dum & Strick 1993), and is broadcast outof Pap. Now that we see that Pap lies between H (PTO) and E(cingulate), and that D has not yet been introduced, we concludethat Pap should mediate D.

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (33)

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Figure 1 (Krieckhaus). Structure and function of Papez circuitand its computation of E from H and D.

The neural solution of the Hull equation, E 5 H 3 D, is in theflow of information through the mammalian forebrain, in particu-lar through Pap, which is depicted – synapse by synapse – in thelower portion of the Figure. Pap consists of: [SUB ⇒ MMN ⇒ AV ⇒M4 ⇒ SUB...]

At the top of the picture is the external WORLD as it affects theprimary sensory input systems, [IN], to the left. [IN] projects intothe PTO posterior association cortex, which embodies H. At the up-per right is output from the forebrain. M4 of cingulate cortex pro-jects to M3 of cingulate. In turn, M3, as E, directs action at all lev-els of the motor system, [OUT], shown at the upper right. To closethe classical connections of Pap as a circuit, SUB and M4 (as all lim-bic cortices) are significantly interconnected, as indicated with thedouble-headed arrow between them. The hypothalamic function ofMMN as mediating D is depicted at the bottom. Opiate neuronscomprising MMN change their firing pattern as a function of con-centrations of various humors, particularly 5-HT, the alarm humorfrom MR, which is inhibitory, as shown at the lower right.

Pap’s thalamus, AV, in the center, receives its subcortical cogni-tive signal exclusively from MMN, via MTT, and provides “interest”to SUB to consolidate memory as H in PTO. MMN, via MTT, pro-vides “hope” for voluntary action through AV and M4. The otherMMN function is to provide relief from alarm, via MTgT, by in-hibiting MR from releasing 5-HT. With the depiction of scene ar-riving at MMN from fx, the function of each connection of Pap isspecified, as shown. Note in Figure 2 that, quite unlike any otherthalamic nucleus, AV receives its inputs (on both proximal and dis-tal dendrites) from the same cortex – SUB.

Abbreviations: E 5 reaction potential; H 5 habit strength; D 5drive; Pap 5 Papez circuit; SUB 5 subiculum; MMN 5 medialmammillary nucleus; AV 5 antero ventral thalamic nucleus; PTO5 parietal temporal occipital cortex; 5-HT 5 serotonin; MR 5 me-dial raphe; MTT 5 mammillo thalamic tract; MTgT 5 mammillotegmental tract; fx 5 fornix.

Moreover, because D is naturally associated with hypothalamicfunctions, and MMN is clearly hypothalamic, then MMN medi-ates D.

2. Four features of thalamic anatomy necessary for solutionof the fornix problem. (1) The architecture of the mammalian thal-amus is well known: Feedback from cortex is largely on the distalportion of the dendrites of the principle neurons of thalamus. Inmost cases, principal neurons project back to the same discreteportion of cortex that supplies their cortical input. The external in-put (e.g., retina) is largely excitatory on the proximal dendrites ofprincipal neurons within glomeruli. These proximal glomeruli alsoenclose dendrites of GABAergic inhibitory interneurons.

(2) Functionally, the morphology of thalamic neurons allows dis-tal dendritic input to gain access to the soma easily (Bloomfield &Sherman 1989; Cai & Lo 1996).

Assumption: Proximal gates distal. Unlike the statements above,which are descriptions of well-documented neural connections orfunctions, the following gating assumption – though quite plausi-ble from (2) – has not been proposed, as far as I know. Given theease with which distal potentials can reach the soma, I assume thatoften, the function of proximal input is not to transmit its input pat-tern to drive thalamus and thus cortex, but rather only to selectivelyregulate, modulate, or gate the potential of the distal cortical feed-back to invade the soma of principal neurons, and thus be sent backto cortex.

(3) To understand the unique properties of Pap’s AV, note thatthe thalamus may be simply dichotomized: Posterior nuclei receivedirect sensory input (e.g., retina) and project to posterior cortex(PTO), whereas anterior nuclei receive indirect feedback from the

forebrain via four loops, and project to anterior cortex (frontal andcingulate).

(4) Of these four indirect forebrain feedback loops informing an-terior cortex, only one, Pap, reveals the following highly eccentricthalamic architecture: Via fornix and MMN, AV receives its “ex-ternal” input from just that same bit of cortex to which AV projects,namely SUB of HF. In contrast, the other three loops receive theirfeedback indirectly from large portions of cortex. Thus AV receivesthe same input on both its distal (directly from SUB) and proximal(indirectly from SUB, via fornix and MMN) dendrites.

The fornix problem is now resolved by the fortuitous simplicityof the inputs into AV – just two, SUB feedback distally and indirectSUB input via fornix and MMN proximally (see Fig. 2). From A&Bwe know that lesions of MMN or MTT cause severe deficits in con-solidation, and from AV architecture, we know that the lesionseliminate most of the excitation on AV’s proximal dendrites. Thisallows the now dominant inhibition from the interneurons of AV tochoke off the potential of distal SUB feedback to drive AV. There-fore, we have the critical question: Is the consolidation deficit at-tributable to loss of SUB feedback or to loss of input from SUB viafornix and MMN with a combination of scene and D? Becausefornix lesions cause no deficits in consolidation in general, it is ap-parently SUB feedback that is critical, which explains why fornixlesions do not generally cause deficits in consolidation. However,as stressed by A&B, fornix lesions do cause severe deficits in scene-relevant consolidation. Thus, for such consolidation, SUB requiresthat its own output (indirectly via fornix and MMN) modulate orgate the transmission of its distal feedback into AV.

In line with other hypothalamic outputs of MMN, the term “in-

Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (34)

terest” best captures the drive-like functional significance of thishypothalamic MMN input (via AV) onto SUB, which is necessaryfor scene-sensitive consolidation – to pay attention to what is ofinterest and let the rest go by.

In conclusion, consideration of the drive properties of hypo-thalamic MMN in Hull’s formulation, taken with the remarkablearchitecture of AV, solves the fornix problem raised by A&B andsuggests that the gating assumption (proximals gate distals) is true,at least for Pap – and perhaps for primary sensory input as well.

Thalamic amnesia and the hippocampus:Unresolved questions and an alternativecandidate

Robert G. Mair, Joshua A. Burk, M. Christine Porter, and Jessica E. LeyDepartment of Psychology, University of New Hampshire, Durham, NH03824. {rgm; jaburk; mcporter; jeley}@christa.unh.edu

Abstract: Aggleton & Brown have built a convincing case that hippocam-pus-related circuits may be involved in thalamic amnesia. It remains to beestablished, however, that their model represents a distinct neurologicalsystem, that the distinction between recall and familiarity captures theroles of these pathways in episodic memory, or that there are no other sys-tems that contribute to the signs of amnesia associated with thalamic dis-ease.

Aggleton & Brown (A&B) have synthesized an impressive body ofevidence to make a case for their theory relating episodic memoryto two parallel pathways linking hippocampus and perirhinal cor-tex to thalamus. They have avoided some of the pitfalls of reifica-tion by focusing on the localization of lesions that impair episodicmemory rather than episodic memory per se. Nevertheless, theiremphasis on a construct like episodic memory and its relationshipto an extended neural circuit raises several questions that are notfully addressed by their review.

1. Are the pathways outlined reasonably construed as a dis-crete neurological system? The hippocampal projections throughfornix to thalamus are well established and (as the authors note)represent pathways that once dominated speculation about theneurological basis of diencephalic amnesia. The main contributionof A&B’s model is the emphasis placed on a projection fromperirhinal cortex to the mediodorsal nucleus and its supposed in-volvement in judging the familiarity of sensory stimuli. It is thisprojection that represents the most questionable part of theirmodel. Although this projection has been described in monkeys,it is not clear how prominent it is compared to other connectionsof the mediodorsal nucleus or of perirhinal cortex (Amaral 1987;Jones 1985; Suzuki 1996a), and it apparently has not been demon-strated in the rat (a species cited for much of the supporting be-havioral evidence). In describing the transport of wheatgerm ag-glutinin-horseradish peroxidase from the rat mediodorsal nucleus,Groenewegen (1988) described signs of anterograde and retro-grade transport in prefrontal cortex and retrograde transport inprepiriform cortex, periamygdaloid cortex, the lateral entorhinalarea, the infralimbic area, and subicular cortex, but not in perirhi-nal cortex. Although A&B emphasize the parallel and separate na-ture of the thalamic projections of hippocampus and perirhinalcortex, the hippocampal formation in fact has well-establishedconnections with the mediodorsal nucleus that travel directly fromsubiculum and indirectly through striato-pallido-thalamic path-ways (Groenewegen 1988). We are aware of no evidence for theprojection from the mediodorsal nucleus to perirhinal cortex in-dicated in A&B’s Figure 1.

2. Does the construct of “episodic memory” and the abstractdistinction between recall and familiarity capture the essence ofwhat is impaired by lesions affecting the hippocampal- andperirhinal-related pathways? There is, of course, a great deal ofcontroversy and active research interest in this issue. Althoughthere are other examples that seem difficult to reconcile withA&B’s model, we will focus on recent work in our lab. We havefound two tasks, DMS trained with retractable levers and olfac-tory continuous DNMS, that would seem to require episodicmemory but were unaffected by lesions destroying hippocampusor the mediodorsal nucleus (Burk & Mair 1998a; Mair et al. 1998;Zhang et al. 1998). It would seem to follow that either these tasksare mediated equally well by both the hippocampal- and perirhi-nal-systems (and are thus little affected by lesions of only one) orthey are mediated by other pathways in the brain. If the first pos-

Commentary/Aggleton & Brown: Episodic memory, amnesia,


Figure 2 (Krieckhaus). Effect of lesions of FX or MTT.Figure 2 is a simplification of Pap’s connections presented in

Figure 1. For abbreviations see caption of Figure 1. Added here isthe detailed anatomical relationship between direct cortical feed-back from SUB on distal dendrites of AV and indirect SUB input,via fornix and MMN, on proximal dendrites of AV. The GABAer-gic, inhibitory interneurons (IN) are shown antagonizing the exci-tatory MMN input proximally on AV. Lesions of MTT, which ma-terially decrease proximal excitation, allow the inhibition frominterneurons to choke off the distal inputs from SUB. Because thelesions that block SUB feedback from penetrating the soma of AVproduce severe deficits in consolidation, this feedback is appar-ently necessary to determine the firing pattern of AV. Because fxlesions, which eliminate the only other source of cognitive input toAV (via MMN), cause no deficits in consolidation (in general), it isclear that SUB feedback, via AV, is also sufficient for consolidation.However, because fornix lesions cause severe deficits in scene-rel-evant consolidation, such consolidation by SUB apparently re-quires SUB’s distal AV thalamic input to be gated or modulatedproximally by SUB’s own indirect output via fornix and MMN.

Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (35)

sibility is true, then one must wonder why DNMS trained in theradial maze is impaired by comparable hippocampal lesions (Mairet al. 1998). Why would perirhinal-related pathways be sufficientto mediate DMS trained with retractable levers but not DNMStrained in an automated radial arm maze?

3. Are there other candidate systems that might also con-tribute to the signs of anterograde amnesia that have been as-sociated with diseases affecting the diencephalon? There are atleast two other systems in thalamus that have been linked to am-nesia: the mediodorsal nucleus and its projections to prefrontalcortex (Markowitsch 1982; Victor et al. 1989) and the intralaminarnuclei (ILn) and related nonspecific nuclei within the internalmedullary lamina and midline areas of thalamus (Mair 1994; Mairet al. 1979; Malamud & Skillicorn 1956; Mennemeier et al. 1992;von Cramon et al. 1985).

Studies of animal models of amnesia have provided several linesof evidence implicating the ILn as an important site of pathologyin the Wernicke-Korsakoff Syndrome, a common cause of dien-cephalic amnesia thought to be caused by thiamin deficiency. Ear-lier studies showed that the ILn are vulnerable to the effects ofpyrithiamine-induced thiamin deficiency and that the extent ofILn damage correlates with the extent of behavioral impairmentin rats recovered from this treatment (Langlais et al. 1996; Mairet al. 1988). Thus rats that recover from the acute effects ofpyrithiamine-induced thiamin deficiency with lesions involvingextensive areas of the ILn tend to perform poorly on DMS andDNMS tasks used to measure memory function (Knoth & Mair1991; Langlais & Savage 1995; Robinson & Mair 1992). The roleof the ILn in these impairments has been verified by studies show-ing comparable behavioral deficits in rats with electrolytic or ex-citotoxic lesions affecting these nuclei (Burk & Mair 1998a; Mair1994; Mair et al. 1998; Zhang et al. 1998). The involvement of hip-pocampal-related pathways in these deficits seems inconsistentwith evidence that similar impairments are not observed follow-ing lesions of the mediodorsal nuclei, fornix, mammillary bodies,hippocampus, or the laterodorsal nuclei (Burk & Mair 1998a;1998b; Mair 1994; Zhang et al. 1998). We recently completed a se-ries of studies that demonstrated distinctive patterns of impair-ment in rats with lesions of the ILn, the hippocampus, or the pre-frontal cortex on different measures of remembering: DMStrained with retractable levers, several versions of DNMS trainedin automated radial arm mazes, and go/no go olfactory continuousDNMS (Burk & Mair 1998a; Koger & Mair 1994; Mair et al. 1998;Porter & Mair 1997; Zhang et al. 1998). We interpret these resultsas evidence of functionally distinct impairments of memory asso-ciated with thalamic, hippocampal, and prefrontal pathology.

The questions raised in this commentary should only reinforcethe importance of understanding the effects of lesioning hippo-campal-related pathways in thalamus and their possible contribu-tions to diencephalic amnesia. A&B are well justified in pointingout the potential significance of these pathways for memory andthey have presented a model that should help focus thinking aboutthis issue. In our view, the available evidence is insufficient to sup-port the claims of the model or to rule out the possible involve-ment of other systems in diencephalic amnesia.

Gestalt view of the limbic system and the Papez circuit – another approach to unity and diversity of brain structures and functions

Hans J. MarkowitschDepartment of Physiological Psychology, University of Bielefeld, D-33501Bielefeld, Germany. hjmarkowitsch@post.uni-bielefeld.dewww.psychologie.uni-bielefeld.de/ae/ae14/index.htm

Abstract: The idea of distinct brain systems for the processing of episodicand other forms of memory is welcome. The two brain systems actuallyproposed however, appear to be stripped of further existing connectionsand could be integrated with one another. If integrating them, it seemsmore logical to propose one enlarged system of limbic structures whoseindividual components make partly different contributions to the forms ofmemory under discussion.

Aggleton & Brown’s (A&B’s) proposal is welcome, as it emphasizesthat the discussion of anterograde amnesia should not focus onlyon the hippocampal region and that there may be important dif-ferences on the anatomico-behavioral levels between the episodicand other memory systems (semantic, declarative memory) (Tul-ving & Markowitsch 1998). Nevertheless, there seems to be some-thing artificial about A&B’s major distinction between an “ex-tended hippocampal system” (which is in fact nothing other thanthe old Papez [1937] circuit) for encoding of episodic informationand a “perirhinal-mediodorsal thalamic system” for familiarityjudgments. The division into these two systems (exemplified byA&B’s Figs. 1 and 2) ignores the existence of a multitude of addi-tional connections with these structures and consequently, theirintegrating role in memory processing (e.g., basal forebrain struc-tures, amygdala). A&B make their proposed division immune toattack by acknowledging that in the large majority of cases of am-nesia, both systems are affected, and by limiting their second sys-tem to the thalamic mediodorsal nucleus (proper) and the perirhi-nal cortex (proper).

From the clinical literature it is hardly possible to excludeperirhinal damage in cases with medial temporal lobe pathologyor midline nucleus or mamillothalamic tract damage in cases withmedial thalamic pathology. On the other hand, it is also nearly im-possible to find patients with selective bilateral mediodorsal thal-amic damage (Markowitsch 1982; 1988; 1992) or selective bilat-eral perirhinal damage.

Moreover, the circuitry in Figures 1 and 2 either ignores the ex-istence of further prominent connections (e.g., between the amyg-dala and the mediodorsal thalamus) or simplifies and hides themby mentioning (for example) a fornix pathway from the hippo-campus to the prefrontal cortex. If so, this pathway would be bi-directional and would not involve major portions of the prefrontalcortex but basal forebrain regions such as the medial septum andthe hippocampal formation (subicular cortex). Though the com-parability (in degree and severity) of the memory deficits after bi-lateral basal forebrain system damage and bilateral damage of thetwo other conventionally defined memory systems (medial dien-cephalic and medial temporal lobe system) is disputed (DeLuca& Diamond 1995; Diamond et al. 1997), there are at least sub-groups of basal forebrain lesioned patients with amnesias of aseverity comparable to that seen after major damage to one or theother (Böttger et al. 1998; Diamond et al. 1997; Von Cramon etal. 1993).

Consequently, the issue becomes less about the differences be-tween the two somewhat artificially divided systems (cf. Fig. 1),and more about the degree and diversity of involvement of limbicsystem structures as such, and the consequences of disconnectionthat have been repeatedly emphasized in so-called global amne-sia. First, we know that a brain region never acts on its own, butdepends on input and output from other regions. This statementis trivial and is often taken as a persiflage by arguing that no brain

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (36)

region is more than six synapses away from another one. Such aview throws the baby out with the bath water. Functional imaginghas revealed that the pathology after a local injury, infarct, or de-generation as demonstrable by static imaging may extend to re-mote, diverse, and dispersed additional loci (Markowitsch et al.1997). Nevertheless, these further affected regions usually haveintimate connections to the local focus.

Second, amnesia for episodic information is to a great extentamnesia for emotionally laden information (Cahill et al. 1995;Markowitsch 1994; Markowitsch et al. 1993). That is, structuressuch as the amygdala (which projects to the mediodorsal nucleus)and the septal nuclei (Von Cramon et al. 1993) contribute sub-stantially to a successful encoding of episodic information. Giventhis interconnectivity, it is quite likely that the perirhinal-mediodorsal memory system is centrally engaged in episodicmemory encoding. In fact, seen from another level, Tulving andMarkowitsch’s (1998) model would likewise predict that theperirhinal-mediodorsal system must be engaged before the ex-tended hippocampal system, as the model assumes (see theirFig. 1) that episodic information must pass through the semanticmemory system before becoming successfully encoded. Hencethe division into the memory systems proposed by A&B does havesome face validity, especially based on findings obtained from an-imal research; but it is certainly inadequate in light of the intimateinterdependencies of the many structures subsumed under theheading “limbic system” (Markowitsch 1999) and engaged in epi-sodic memory functions.

ACKNOWLEDGMENTSI thank Eva Böcker for help with the drawing of Figure 1. My work wassupported by the DFG through grants Ma 795/24 and Ma 795/25.

What are the functional deficits produced byhippocampal and perirhinal cortex lesions?

A. R. Mayesa, R. van Eijkb, P. A. Goodingb, C. L. Isaaca,and J. S. Holdstocka

aDepartment of Clinical Neurology, Royal Hallamshire Hospital, University of Sheffield, Sheffield S10 2JF, England; bDepartment of Psychology,University of Strathclyde, Glasgow G1 1QE, Scotland. {a.mayes; c.isaac; j.holdstock}@[emailprotected]

Abstract: A hippocampal patient is described who shows preserved itemrecognition and simple recognition-based recollection but impaired recalland associative recognition. These data and other evidence suggest thatcontrary to Aggleton & Brown’s target article, Papez circuit damage im-pairs only complex item-item-context recollection. A patient with perirhi-nal cortex damage and a delayed global memory deficit, apparently in-consistent with A&B’s framework, is also described.

This comment focuses on Aggleton & Brown’s (A&B’s) proposalthat selective Papez circuit lesions in humans disrupt free recalland leave item recognition intact and A&B’s interpretation of thiseffect. We also comment briefly on evidence that selective perirhi-nal cortex system lesions have different effects on human mem-ory.

Selective Papez circuit lesions are extremely rare in humans andnormal or relatively preserved item recognition is not always seenfollowing apparently selective damage. It remains to be shownwhether such patients have additional cryptic damage outside thePapez circuit. We have examined extensively one patient with rel-atively selective bilateral hippocampal damage who often showscompletely normal recognition in the face of impaired recall. Onmany item recognition tests, she scores as well as or better thanmatched control subjects, although she has occasionally been im-paired. Her performance has been within the normal range onYes/No and Forced Choice item recognition tests at both shortand long delays of one day or more. Relative preservation is not asimple function of difficulty because she has performed above thecontrol mean on a hard object recognition test matched for diffi-culty with an object free recall test at which she was very impaired.

Several features of this patient’s performance warrant com-ment. First, she is impaired at associative, but not item, recogni-tion. Thus, she was greatly impaired at an object-location recog-nition test that normal subjects found easier than the objectrecognition test at which she was unimpaired. She was also im-paired at temporal order, face-voice, word-meaning, and animalpicture-occupation name associative recognition. Second, herstory recall, but not her story recognition ability, declined abnor-mally fast between delays of 15 seconds and 10 minutes. Third, us-ing a modified process dissociation procedure (PDP) with bothfamous faces and names tasks, we found not only normal itemrecognition, but normal recollection.

A&B argue that patients like ours have selective item-contextmemory (recollection) deficits. This argument assumes that nor-mal performance on some recognition tasks depends on familiar-ity alone, which is contrary to the widely held view that item recog-nition depends on both recollection and familiarity. The targetarticle suggests that Papez circuit damage disrupts the context-item retrieval underlying free recall. If this is identical to the item-context retrieval manifested in recognition, recollection cannotcontribute to item recognition tasks at which our patient is nor-mal. However, Hasselmo and Wyble’s (1997) recent model of hip-pocampal function shows that this identity may not hold. Theymodelled recall as the retrieval of items from context cues andrecognition as the retrieval of context from item cues (familiarity

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Figure 1 (Markowitsch). Realignment of the structures and con-nections of the two memory systems proposed by A&B. The figurecontains the “perirhinal-mediodorsal system” together with the“extended hippocampal system,” which can be seen as embeddedin the circuitry of the perirhinal-mediodorsal system. The connec-tivity follows exactly that of A & B’s Figures 1 and 2 but it omits afew structures and connections (along with the occasional empha-sis on the relative strength of a given pathway). Regions that be-long to the “perirhinal-mediodorsal system” are rectangular; re-gions belonging to the “extended hippocampal system” are oval;regions belonging to both circuits contain both shapes.

Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (37)

played no role). Only aspects of the study situation appearing re-peatedly were counted as context, so features appearing uniquelywith a studied item were excluded. Giving the model a “biochem-ical lesion” disrupted recall, but not recognition, because such re-peated context was well learned. The model shows that recollec-tion involving repeated context features can break down for recall,when it does not for item recognition.

A&B suggest that recollection does not contribute to recogni-tion because this is not usually affected by study-test context shifts,whereas recall is. Although this was not modelled by Hasselmoand Wyble, it is plausible that such modelling would show mini-mal effects of context shifts on item recognition as these resemblelesions of the model. Hence, stronger evidence for the assump-tion that recollection does not contribute to recognition needs tobe provided.

We are also unconvinced that the familiarity, preserved afterhippocampal lesions, cannot be mediated by enhanced item flu-ency (priming) being automatically used to attribute familiarity topreviously encountered items. This is consistent with some am-nesics performing at chance on recognition tests because at-tributing familiarity may depend not only on enhanced fluency ofitem processing, but also on being able to recollect sufficientlyabout the study context when they are unable to do this. Forcedchoice tasks are no more likely than Yes/No tasks to force subjectsto use fluency to make oldness attributions when memory for thestudy context is largely absent.

In contrast to global amnesics, our hippocampal patient makesoldness attributions normally (familiarity) because she can alsorecognize associations between items and repeated contextualfeatures normally as shown by our PDP results. However, she doesnot recognize more complex associations between two or moreitems and these repeated contextual features as is suggested byher impaired performance on association recognition tasks. De-tailed analysis of what she recollects when she recognizes itemssuggests that she rarely recalls specific associations between a tar-get item and one or more other items (or thoughts) together withthe repeated context features.

We think that our patient and similar ones show preserved itempriming and item-repeated context feature memory, but do notconsolidate complex item-item-repeated context feature associa-tions into long-term memory normally after single presentations.Support for this view is the accelerated loss of story free recallshown by our patient. Unlike story recognition, which is mediatedby simple item-repeated contextual feature association retrievaland familiarity, recall requires retrieval of these more complexkinds of association. The deficit may be confined to the consoli-dation of complex associations where the components are repre-sented in distinct neocortical regions.

Selective perirhinal cortex lesions have never been reported inhumans. We have, however, studied a patient with extensive bi-lateral damage to what should be this region in humans. This pa-tient has an intact hippocampus although damage extends into thetemporal association cortex at the temporal pole. She shows pre-served recognition and recall except for faces at delays of 30 min-utes, but forgets rapidly over delays of a few weeks, at which timerecall and recognition deficits appear. Such cases suggest that thisbrain region may not have the functions one might expect fromthe animal studies referred to in the target article. Assuming thatface recall, like face recognition, is impaired early, our patientshows no recall/recognition dissociation, but a delayed globalmemory deficit. Future work must identify the conditions thattrigger this syndrome and its functional basis.

How do animals solve object-recognitiontasks?

Dave G. MumbyDepartment of Psychology, Concordia University, Montreal, Quebec, CanadaH4B 1R6. [emailprotected]

Abstract: This commentary reviews recent evidence that some hippo-campal functions do not depend on perirhinal inputs and discusses howthe multiple-process model of recognition may shed interpretive light onprevious reports of DNMS reacquisition deficits in pretrained subjectswith hippocampal damage. Suggestions are made for determining whethernonhuman subjects solve object-recognition tasks using recollective mem-ory or familiarity judgments.

Aggleton & Brown (A&B) have abandoned the widely held viewthat there is a single “temporal-lobe memory system,” suggestinginstead that the perirhinal cortex and hippocampal formation arefunctionally independent. Their observation that this may beconsidered a “surprising” feature of their model gives insight tothe kinds of pretheoretical assumptions that have helped tomaintain the unitary-system view for so long in the first place –for it has not been on the basis of behavioural evidence that theperirhinal cortex and hippocampal formation are functionally in-terdependent, but rather on anatomical evidence of strong reci-procal connections between these two structures. Perhaps whatshould surprise us is how much has been previously inferredabout the functional relationship the perirhinal cortex shareswith the hippocampus from this one feature of its extrinsic cir-cuitry. The existence of the requisite connections does not tell uswhether two structures depend on each other for their respec-tive functions. Nor is there any reason to assume that the degreeof functional similarity between two structures is proportional tothe density of connections they share. The latter assumptionseems to be the main reason why many believe information-pro-cessing within the hippocampal formation requires inputs fromthe perirhinal cortex, despite a dearth of behavioural evidence tosupport this view.

We recently reported a double-dissociation following lesionsof the perirhinal cortex and hippocampal formation, consistentwith A&B’s conclusion that these two structures are functionallyindependent: Rats with perirhinal lesions were impaired on anobject-based delayed nonmatching-to-sample (DNMS) task butperformed normally on delayed matching-to-place (DMTP) in awater maze, whereas rats with hippocampal lesions were im-paired on DMTP, but not DNMS (Glenn & Mumby 1996). Eachrat was tested on both tasks, so the dissociations were apparentwithin subjects, and therefore, cannot be attributed to differ-ences in lesion extent – a potential caveat whenever ostensibledissociations occur across separate studies, as was the case withour previous demonstrations of this particular double-dissocia-tion (Glenn & Mumby 1998; Mumby & Pinel 1994; Mumby etal. 1992). The DNMS and DMTP tasks employed the same re-tention delays (ranging from 4 to 300 sec), and (quasi) trial-unique stimuli were used in both tasks, similarities that suggestthat the dissociations occurred because one task required mem-ory of object identity, whereas the other required memory of aspatial location.

Although our findings are consistent with A&B’s conclusion thatspatial-information processing by the hippocampus does not re-quire inputs from perirhinal cortex, there are reports that perirhi-nal lesions disrupt performance of some allocentric-spatial tasks(Liu & Bilkey 1998c; 1998b; Wiig & Bilkey 1994). But demon-strating that either perirhinal or hippocampal lesions can impairperformance of the same spatial-memory task does not reveal whythe task is sensitive to perirhinal damage. Any task can fail for avariety of reasons. Given the lack of perirhinal lesion effects on al-locentric tasks in our experiments and in those of A&B, it is likelythat the deficits observed by Bilkey and his colleagues were causedby some other feature of their tasks. Do perirhinal lesions cause

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (38)

deficits by depriving the hippocampal formation of critical inputs?This question can potentially be addressed by assessing the effectsof crossed-unilateral lesions of these two structures, combinedwith commissurotomy, on performance of the task that is dis-rupted by bilateral lesions of either structure (assuming the typi-cal situation where unilateral lesions of either structure spare per-formance of the task).

The suggestion that subjects may solve the DNMS task by us-ing either recollective memory or a familiarity judgment raises thequestion of whether members of a particular species are predis-posed to use one strategy rather than the other. The answer is im-portant if we are to compare findings across species, as the super-ficial similarities in the DNMS tasks for humans, monkeys, andrats do not justify the assumption that all three species solve themthe same way. If a lesion impairs a subject’s capacity for solving thetask using the preferred strategy, how might this affect postsurgeryperformance? Some of our results are consistent with the hypoth-esis that hippocampal lesions disrupt a preferred strategy for solv-ing the DNMS task, while leaving intact the capacity for an alter-native solution. Following hippocampal ablation, rats that hadreceived presurgery training were impaired in reacquisition at the4-sec retention delay (Mumby et al. 1992; 1996). We originally in-terpreted this as reflecting a nonmnemonic procedural impair-ment because all rats soon overcame the deficits, and thereaftercontinued to perform as well as controls at much longer delays.But another possibility that stems from the multiple-processmodel of recognition is that these transient deficits occurred ashippocampal rats discovered their inability to solve the task thesame way as before surgery and had to switch to a less-preferredbut still effective alternative.

How could one go about determining whether a subject is solv-ing the DNMS task using recollective memory or familiarity judg-ments? Two points are worth noting here: First, A&B’s distinctionbetween familiarity and priming on the basis of conscious aware-ness being part of the former but not the latter is of little utilitywhen considering how nonhuman animals solve a task because wecan neither confirm nor disconfirm conscious awareness in non-humans. Second, in the absence of conscious awareness there maybe little or nothing to distinguish the processes underlying famil-iarity judgments from those that occur in instances of repetitionpriming. A general strategy for asking whether a recognition taskfor nonhuman animals engages processes that are similar to thoseof recollective memory is to examine the effects of task manipula-tions that are known to have different effects on recollective mem-ory and familiarity judgment in humans. For example, distractionduring the retention interval disrupts the performance of humanson explicit-memory tests but not implicit-memory tests (Graf &Schacter 1985). DNMS performance in rats (Mumby, unpub-lished data) and monkeys (Zola-Morgan & Squire 1985b) is dis-rupted when the subjects must perform a distraction task duringthe retention interval. Does this mean that rats and monkeys useexplicit memory to solve the DNMS task? The effect of distrac-tion by itself might not be very convincing, but the evidence couldbecome more compelling if additional manipulations were foundto have similar effects on explicit memory in humans and DNMSperformance in animals. The same general approach could beused to probe whether two groups of subjects performing at sim-ilar levels of accuracy are actually solving the task the same way,or whether brain-damaged subjects that received presurgerytraining are still solving the task the same way they did before thelesions.

A “presence/absence hypothesis”concerning hippocampal function

David J. MurrayDepartment of Psychology, Queen’s University, Kingston, Ontario K7M 3L6,Canada. [emailprotected] pavlov.psyc.queensu.ca

Abstract: According to a “presence/absence hypothesis,” the hippocam-pus is not necessary for the formation of learned associations between cur-rently present stimuli and responses (as in classical conditioning), but isnecessary whenever a stimulus, if it is to activate a particular response,must first activate a memory-representation of something not present inthe here-and-now. The distinction between responses made to presentstimuli as opposed to (memories of) absent stimuli was first stressed byRomanes (1889), but we find evidence in the target article that supportsthe relevance of this distinction to our understanding of hippocampalfunctioning.

In this commentary I shall offer a hypothesis concerning the rolethe hippocampus plays in memory tasks given to animals or hu-mans. I shall take three findings about the hippocampus, one wellknown, and two that were new to me until I read Aggleton &Brown’s (A&B’s) target article, and offer these as preliminary evi-dence for the hypothesis. According to “presence/absence hy-pothesis,” the hippocampus is not necessary for mediating learnedresponses when the retrieval cue is present in the here-and-now;but if the learned responses require the use of memory represen-tations of absent things or situations, then the hippocampus doesbecome necessary.

(1) The well-known finding is that, in animal subjects, hippo-campal lesions do not destroy the ability of the subject to learnclassically conditioned responses, disrupt spontaneous recovery,or affect the processes of extinction. Evidence for these state-ments has been provided in the reviews of hippocampal function-ing by O’Keefe and Nadel (1978, pp. 316–24) and, more recently,by Cohen and Eichenbaum (1993). In classical conditioning an an-imal (or human) learns to make novel responses to what had pre-viously been a stimulus of neutral emotional value; for example,Pavlov taught his dogs to salivate on hearing the clicks of ametronome. Since Pavlov’s time, Rescorla (1967) and his col-leagues have indicated that the activation of the salivary responseappears to depend on the dog’s having learned to “expect” food af-ter the metronome has sounded. I find it difficult to see what iswrong with rephrasing this last sentence as follows: “The doglearns to recognize the metronome as a signal that food is aboutto arrive.” All pet owners know the speed with which their cat ordog comes running when it “recognizes” the sound of a can beingopened as a signal that food is on the way. What we wish to em-phasize here, however, is that the sound of the metronome or ofthe can being opened is a conditioned stimulus that is “present”and is not imagined by the animal. In contrast with that of animals,much human behaviour is initiated by imagining a stimulus that isnot present, but absent. A hungry person can salivate at thethought of a hamburger. Hence, when hippocampal lesions fail toeradicate the formation or elicitation of classically conditioned re-sponses, we must at least consider the possibility that it is becausethe conditioned stimuli are present in the animal’s environmentimmediately before the conditioned response is activated and thatthe hippocampus is not necessary for processing information con-cerning “recognition as” responses to present stimuli, at least inanimals.

(2) In section 4.1, third paragraph, A&B reported that DNMSdeficits had followed localized destruction (by the injection ofneurotoxins) of hippocampal fields CA1-4, along with the amyg-dala, in experiments in which monkeys were removed from the ap-paratus during the retention-interval (Alvarez et al. 1995). On theother hand, Murray and Mishkin (1996) have reported that, if an-imals with hippocampal lesions similar to those induced by Al-varez et al. were left in the apparatus during the retention inter-val, they could pick out a new stimulus (from two stimuli, one ofwhich had been seen before) after a delay of as long as 40 minutes.

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (39)

A large number of apparatus cues would have remained presentfor all of the retention interval during Murray and Mishkin’s ex-periment, but would have been absent in the case of the experi-ment by Alvarez et al. (1995). This finding, which obviously mer-its further investigation, is consistent with the notion that if thespatial context is always available to the subject during a retentioninterval, then adequate performance on this task does not requirean intact hippocampus to the extent demanded in tasks where thesubject is removed, then put back into the apparatus. Agreed, thismight be because positional stances may be adopted in the inter-val, but it may also be because the contrast of a novel stimulus ob-ject with the apparatus background is more striking if the back-ground has been available for many minutes beforehand than if ithad been absent and suddenly reinstated.

(3) In the first column of Table 1, A&B indicate that hippo-campal ablations cause little damage in DMNS tasks but cause sig-nificant damage in DMNP tasks. The content of the previousparagraph may indicate that this generalization needs qualifying,but here I wish to focus on the fact that in delayed memory tasksin which the subject must remember objects (DMNS) as opposedto places (DMNP), there is a reinstatement of a spatial back-ground at the time of delayed nonmatching to sample in which twoobjects (one old, one new) are both present. One of the objectshas to be responded to as “new.” But in the cases of DNMP tasks,or of spatial alternation in T-mazes, the animal at the choice pointhas to retrieve from memory information concerning the directionin which to turn. In the DMNS task the animal has to recognize apresent stimulus object as new, whereas in the DMNP task it doesnot have to recognize a present sensory stimulus (a choice point)as new, but only as a starting point, for the self-initiation of a newset of movements (see also Poucet 1993).

The idea that the hippocampus may have evolved for the pur-pose of facilitating the representation, to the subject, of things,places, and emotions that are not actually “present” in the here-and-now is consistent with the general notion that human intelli-gence differs from animal intelligence chiefly insofar as humanscan imagine absent objects far more easily than animals can, par-ticularly when humans have language to help them do so. Theclearest and most detailed statement of this idea has been pro-vided, in my opinion, by Romanes (1889).

A secondary account of Romanes’s theory has been given byMurray (in press), but an important aspect of it consists in thedistinction made by Romanes between what he called “recep-tual” and “conceptual” memory. Receptual memory is possessedby many animals, and is used whenever a present stimulus is rec-ognized as having been linked in the past with pleasure or dis-pleasure. Conceptual memory is a more advanced stage in theevolution of memory in which absent things can be representedby sensory imagery (including kinesthetic imagery) or by lan-guage.

I propose, therefore, that what really matters in determiningthe role played by the hippocampus in any task is whether the taskinvolves a present stimulus (to which a “receptual” response frommemory can be given) or is one in which the subject must pro-vide, from memory, some intermediate (“conceptual”) responseto a present stimulus. The hippocampus, it is proposed, plays amore important role in conceptual memory than in receptualmemory.

Episodic memory: It’s about time (and space)

Lynn Nadel, Lee Ryan, Katrina Keil, and Karen PutnamDepartment of Psychology, University of Arizona, Tucson, AZ 85721. {nadel; ryant; keil; kputnam}@u.arizona.edu

Abstract: Aggleton & Brown rightly point out the shortcomings of the me-dial temporal lobe hypothesis as an approach to anterograde amnesia.Their broader perspective is a necessary corrective, and one hopes it willbe taken very seriously. Although they correctly note the dangers of con-flating recognition and recall, they themselves make a similar mistake indiscussing familiarity; we suggest an alternative approach. We also discussimplications of their view for an analysis of retrograde amnesia. The no-tion that there are two routes by which the hippocampus can reactivateneuronal ensembles in the neocortex could help us understand some cur-rently puzzling facts about the dynamics of memory consolidation.

Aggleton & Brown (A&B) have done the field a considerable ser-vice in pulling together many strands of research and clinical ex-perience to refocus our attention on the broader picture in thestudy of amnesia. This is not to say that their model is completelycorrect, but we believe their argument is a much needed correc-tion to the overly narrow emphasis on the “medial temporal lobesystem” that has dominated the field in the past 10–15 years. It isinteresting to recall that only 20 years ago the connections be-tween hippocampus and its temporal lobe neighbors such as theentorhinal cortex were largely ignored, and all attention wasplaced on outputs exiting via the fornix. A&B do a good job ofhelping us understand just why the focus switched to the caudalprojections, and more importantly, how this change in focus hascreated more confusion than clarity in explanations of memoryand memory disorders. Indeed, the past 10 years can be charac-terized as a series of debates and controversies that have eruptedwhenever the “medial temporal lobe system” has met with appar-ent empirical refutation. In replaying these debates, A&B clarifythe flimsiness of the evidence in favor of the MTL hypothesis andthe desperate need for a framework that can encompass the com-plexity of memory and amnesia. A&B point the way to what mightbe such a new approach.

Much of the force behind the MTL hypothesis derives fromwork with primate models, and what the proponents of this viewheralded as the definitive development of an animal model of am-nesia. Indeed, some went so far as to characterize this “accom-plishment” as one of the major triumphs of recent cognitive neu-roscience. One result of A&B’s argument is the demolition of thispiece of puffery. It is clear from their review of the relevant workthat the animal model, focused as it is on recognition memory (andtasks such as delayed nommatch to sample) falls far short of a re-alistic model of the complex syndrome of human amnesia. By forc-ing tasks as disparate as free recall, recognition, and semantic rec-ollection into the same “declarative” framework, the MTL theoryhas impeded our understanding of the tasks themselves and theirneural bases. A similar point can be made regarding the proposeddistinction between “explicit” and “implicit” forms of memory:Recognition and perceptual identification, as examples of thesetwo, are distinct in some ways but also share common processes(Graf & Ryan 1990). Taxonomies of tasks based on a single at-tribute such as “conscious recollection” necessarily result in over-simplifications.

There are, however, some weaknesses in the A&B approach.We would like to discuss a few of them briefly: the current lack ofclarity concerning the impact of damage restricted to the hip-pocampal formation; how best to think about notions such as nov-elty and familiarity; and the retrograde portion of the amnesic syn-drome.

There remains a lack of clarity concerning the impact of dam-age restricted to the hippocampal formation. Although there arecompelling reasons to be concerned about hidden pathology inclinical cases, one must take care not to turn this idea into agarbage can for findings inconsistent with one’s views. The devel-opment of techniques for measuring in vivo brain activity provides

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (40)

a new source of data that might help clarify this issue. For exam-ple, although recent neuroimaging studies that focus on the en-coding of novel stimuli point to potentially important functionaldifferentiations among medial temporal lobe structures, and sug-gest that hippocampus and parahippocampal regions may play arole in encoding relational or configural aspects of complex scenes(Stern et al. 1996; Tulving et al. 1994), the entorhinal cortex maydetect and encode singular novel objects. These studies are con-sistent with A&B’s review of both the c-fos activation studies in an-imals and recent rat hippocampal lesion studies (Honey et al.1998). Given this convergence of results, a finding of impaired ob-ject novelty detection in a patient with brain damage apparentlylimited to the hippocampus itself would cause one to be con-cerned about hidden pathology. In the absence of evidence con-firming more extensive damage, however, one is hard-pressed todisregard these inconsistent cases, unless one chooses to disregardneuropsychological evidence all together.

Although A&B do an excellent job of pointing out the inaccu-racy of lumping recognition and recall together, they fall into asimilar trap when they attempt to tackle the notion of familiarity.Their definition of it is too narrow, limited as it is to tasks involv-ing an explicit judgment of “knowing.” In fact, the term “familiar-ity” has been used in many contexts, including the accumulationof multiple similar experiences (as in the case of high-frequencywords; Glanzer & Bowles 1976), single-trial learning (as in thecase of recognition; Mandler 1980), and even semantic learning(as in the case of semantic priming; Jacoby 1983). Instead of thislimited view, we would suggest that the feeling of knowing (FOK)involves at least two important, and separable, components. First,there must be a representation of the thing to be known, and sec-ond, there must be a process of monitoring and/or accessing thisrepresentation. Under this conception there could be multipleneural systems involved in representation, each subserving differ-ent kinds of knowledge, and each contributing to performance onvarious tasks, including tasks such as item priming. The explicitfeeling of knowing, however, would require monitoring and com-paring the attributes of these representations, and might dependcritically on involvement of the hippocampal formation.

Finally, there is the matter of retrograde amnesia, about whichA&B say very little. This is a pity in that their reformulated viewof anterograde amnesia suggests a very intriguing way to thinkabout retrograde amnesia. There is widespread consensus that thehippocampal formation and neocortex interact over time aftermemories are initially stored. One popular view (e.g., McClellandet al. 1995) is that memory traces are rapidly formed in the hip-pocampus and then replayed afterwards to allow incrementalchange within the neocortex. According to some, everything to beremembered is established within neocortex during this “consoli-dation” period (e.g., Squire & Alvarez 1995). Others argue thatonly some aspects of permanent memory are established withinneocortex, and that the hippocampal system is always importantin retrieval of the contextual aspects of episodic memory (Mosco-vitch & Nadel 1998; Nadel & Moscovitch 1997; 1998). In thecourse of the replay process, the hippocampal trace is used to re-activate dispersed neocortical representations, as a consequenceof which two different outcomes can occur: Categorical knowl-edge can be extracted from episodic experience, and associativefacts can be learned. It was never clear in the McClelland et al.treatment just how these two different outcome emerge from thesame reactivation mechanism.

A&B’s perspective suggests that there are two routes by whichhippocampus might reactivate neocortical representations. Thesetwo routes could make quite different contributions to the reacti-vation process. One might suppose that reactivation by way of thetemporal cortex and its representations of things in the world canlead to the formation of concepts/categories, whereas reactivationby way of the frontal cortex, and its representation of the tempo-ral order in which things within an episode occurred, can lead tothe learning of associations. This analysis maintains the separationbetween processes involving recognition memory (temporal neo-

cortex) and episodic memory (frontal cortex), and shows how theymight work together during memory consolidation. One very ex-citing speculation it immediately leads to is the following: What ifonly one part of the reactivatiion process, say the fronto-hip-pocampal part, is impaired? One might reactivate the various partsof an episode, but in the wrong temporal sequence. This couldlead to incorrect conclusions about the causal relations inherentin prior experience. Could this be anything like what happens inschizophrenia, a disorder now known to involve damage in thefronto-hippocampal axis?

Such speculations only hint at the rich possibilities opened upby A&B’s approach. They are to be commended for their correc-tive action.

ACKNOWLEDGMENTWe thank Christina Charbonneau-Perry for her comments.

Memory systems, frontal cortex, and the hippocampal axis

Amanda ParkerDepartment of Experimental Psychology, University of Oxford, OX1 3UD, England. [emailprotected]

Abstract: Three comments are made. The proposal that recollection andfamiliarity-based recognition take different thalamic routes does not fit re-cent experimental evidence, suggesting that mediodorsal thalamus acts inan integrative role with respect to prefrontal cortex. Second, the role offrontal cortex in episodic memory has been understated. Third, the role ofthe hippocampal axis is likely to be the computation and storage of ideo-thetic information.

Aggleton & Brown’s (A&B’s) target article presents a timely reviewof the substantial body of evidence that supports the circuit viewof anterograde amnesia first proposed by Delay and Brion (1969).This commentary uses the target article as a starting point for fur-ther exploration of: (1) A&B’s proposed anatomical substrate forthe fractionation of recognition memory, (2) the role of frontal cor-tex as a likely endpoint of the hippocampal axis, and (3) the ideo-thetic nature of the processing undertaken by the hippocampal-frontal system.

(1) A&B propose that recollection-based recognition is pro-cessed by the hippocampal-anterior thalamic system, whereasfamiliarity-based recognition is the product of the perirhinal-mediodorsal thalamic system. Although this is an appealing pro-posal, recent experimental evidence suggests that the situation ismore complex than this, and that magnocellular mediodorsal thal-amus (MDmc) is more likely to be acting in an integrative rolewith respect to the prefrontal cortex. A large impairment in recog-nition memory performance results from disconnection of peri-rhinal cortex from the mediodorsal thalamus (Parker & Gaffan1998a), which at first sight seems to be consistent with the pro-posal by Aggleton and Mishkin (1983a; 1983b) and Bachevalier etal. (1985a) that visual object recognition memory depends on in-formation flow from temporal cortex to frontal cortex via the thal-amus. However, several further considerations argue against thisinterpretation. First, the route of information flow from temporalcortex to thalamus as originally envisaged by these authors was viathe fornix and amygdala, and we have demonstrated that with pre-operative training in delayed matching-to-sample, aspirationamygdalectomy combined with fornix transection need not pro-duce an impairment (Parker & Gaffan 1998a). Second, as Gouletet al. (1998) recently showed, although there is a small direct pro-jection from the perirhinal cortex to the mediodorsal thalamus, in-dependent of synaptic relays in the hippocampus or amygdala, thisdirect projection passes through or very near to the amygdala andis therefore interrupted by aspiration amygdalectomy carried outin the same way. Third, bilateral lesions restricted to the medial

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (41)

portion of the magnocellular part of the mediodorsal nucleus,which is the portion that receives limbic and perirhinal afferents,had only a very mild effect on delayed matching-to-sample whencompared with rhinal ablation (Parker et al. 1997). For these rea-sons the impairment from disconnection of perirhinal cortex fromthe whole of the magnocellular part of the mediodorsal nucleuscannot be explained by the direct or limbic interactions betweenthese two structures. Instead, it seems likely that because of theextensive efferent and afferent connections of the magnocellularpart of the mediodorsal nucleus with prefrontal cortex, a unilat-eral lesion of the whole magnocellular mediodorsal nucleus is suf-ficient to produce widespread dysfunction of the ipsilateral pre-frontal cortex. Thus, when crossed with a unilateral perirhinalcortex ablation, a unilateral lesion of this nucleus produces a func-tional effect similar to the effect of a large unilateral prefrontalcortex lesion, although somewhat milder (Parker & Gaffan 1998a).Furthermore, bilateral removal of the entire MDmc thalamus alsoaffects object-in-place memory, suggesting that frontal cortex, andby extension MDmc, becomes particularly important when an in-tegration of object and spatial information is necessary (Parker &Gaffan, in press).

(2) A substantial body of the work using rats reviewed in the tar-get article has stressed the contribution of the hippocampal axisto spatial memory. The conclusion often drawn from this evidenceis that the anterior thalamic efferents to cingulate cortex and tohippocampus via the cingulum bundles complete the circuit,rather than the efferents to prefrontal cortex. As rats have littlecortex that is a hom*ologue of primate prefrontal cortex, manipu-lations that clarify the role of frontal cortex have not produced re-sults that are readily interpretable. Furthermore, consideration ofthe effects of frontal lesions on human memory have often beeninconclusive. However, we have shown in the monkey that bilat-eral ablations of the prefrontal cortex abolish the ability to associ-ate even a single visual stimulus with reward (Parker & Gaffan1998b). This suggests that in humans with extensive bilateralfrontal lesions far more severe memory impairments would beseen than those found with frontal patients of the type that havetypically been studied, with much smaller and often unilateral le-sions. In the monkey, the results of a series of experiments that ex-amined the effects of lesions of fornix, mamillary bodies, and an-terior thalamus on object-in-place memory have supported theproposal that the hippocampal system is preferentially importantfor object/spatial configural memory (Gaffan 1994b; Parker &Gaffan 1997a; 1997b). We did not find a similar impairment aftercingulate cortex ablation, however, and histological examinationshowed degeneration of cingulum bundles bilaterally in two casesand unilaterally in one. This led us to propose that, for object-in-place memory, connections from anterior thalamus to prefrontalcortex were critical for choosing and reaching toward a specifictarget on the basis of an episodic representation (Parker & Gaffan1997b).

(3) Recent work with both rats and monkeys suggests that thespecific role of the hippocampal-anterior thalamic system in spa-tial memory may be in the computation and storage of ideotheticinformation, rather than spatial processing in more global terms,whereas extrahippocampal temporal cortex may provide a moregeneral representation of the environment. This path integrationhypothesis proposed in rats (McNaughton et al. 1996; Wishaw &Jarrard 1996; Wishaw et al. 1997) can also be applied to monkeys(Gaffan 1998). For example, in the proposal for the object-in-place performance deficit after anterior thalamic lesion discussedabove, the computation of arm movement toward the previouslyrewarded object on the screen is guided in part by ideothetic cues,generated by the hippocampal axis and being used by prefrontalcortices to guide voluntary movement. That prefrontal cortex andthe hippocampus both play a role in the integration of movementresponses to visual stimuli is also suggested by the finding thatmonkeys show a deficit in visuomotor conditional learning afterboth hippocampal and ventral frontal lesions (Murray & Wise1996; 1997).

Conclusions that can be drawn from the points raised in thiscommentary are that the hippocampal-anterior thalamic axis pro-vides one component of memory, ideothetic information, that, to-gether with allothetic and object information processed in othercortical regions, aids the prefrontal cortex in selecting voluntaryactions on the basis of stored strategies (Parker & Gaffan 1997c).Prefrontal cortex is particularly crucial in this process when novelitems are being encoded (Parker et al. 1998), and modulation ofposterior cortical areas by prefrontal cortex on the basis of thislearning is likely to be via prefrontal afferents to basal forebrain,which then modulates posterior cortices via efferents through theanterior temporal stem and fornix (Gaffan et al. 1998).

The neural bases of recollection and familiarity: Preliminary tests of the Aggleton–Brown model

Alan D. PickeringDepartment of Psychology, St. George’s Hospital Medical School, Universityof London, London SW17 0RE, England. [emailprotected]

Abstract: Aggleton & Brown suggest that whereas familiarity is computedin perirhinal cortex, the hippocampus contributes to recollection. This ac-count raises issues about the definition of amnesia, clarifies confusionabout dual-process models of recognition, and sits comfortably with ac-counts of hippocampal function from outside the amnesia literature. Themodel can – and should – be tested. Some preliminary data suggest thatit may need changes.

Modern behavioural neuroscience often takes an experimentalmeasure of relevant processes from one species (e.g., the rat ormonkey) and creates an analogous measure in another species(e.g., man). Parallel drug studies, or the comparison of patientswith lesioned animals, may indicate that the animal preparationand measures can serve as a model of normal and abnormal hu-man psychological processes. This cross-species approach, how-ever, carries with it a number of dangers, including a tendency to-wards a “functional autonomy of methods.” Simply put, theproblem is that research becomes focused on the tools, assays, andindices being employed, rather than on the bigger picture (forwhich the tools, etc. were initially developed).

The target article by Aggleton & Brown (A&B) neatly illustrateshow a valid question for amnesia research: “Is hippocampal dam-age sufficient to cause amnesia?” became transformed into: “Ishippocampal damage sufficient to impair DNMS (delayed non-matching to sample) test performance?” They argue that answersto the latter question may be relatively uninformative in address-ing the former. When the particular forced-choice format used inDNMS (and DMS) tests with animals is also used with humansubjects (e.g., in the Warrington Recognition Memory Test,RMT), patients with damage restricted to the hippocampusproper show no deficits (Aggleton & Shaw 1996).

When research methods (such as a particular behavioural test)start to become functionally autonomous there is also a danger ofequating critical underlying processes directly with test perfor-mance (see also Jacoby 1991). A&B show that relating (tests of)“recognition memory” to a particular neural substrate is an inad-equate level of analysis for a proper understanding of amnesia be-cause recognition tests are likely to involve multiple processes.Although the components of recognition that they address (famil-iarity and recollection) represent only a modest increase in speci-ficity, this work is moving in the right direction.

A&B’s work raises the following issues:1. Defining amnesia. Mayes and Downes (1997) claim “good

general agreement” for a definition of amnesia in which the an-terograde deficit comprises impaired recognition and recall. A&Bclaim that their patients with selective hippocampal damage are am-nesics and yet they may show impaired recognition only under cer-

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (42)

tain circ*mstances. A&B’s position seems to imply that the defininganterograde deficit in amnesia is one affecting episodic memorybased on recollection. I suspect such a definitional shift will provecontroversial, although it follows from the earlier comments that Iapplaud definitions based more on processes than on tests.

2. Testing the model. A&B’s account also has the merit of testa-bility and refutability. For example, any patient with selectivemamillary body lesions should have memory deficits similar to theselective hippocampal patients of Aggleton and Shaw. If such a pa-tient demonstrated preserved recollection-based remembering(as crudely indexed by normal Wechsler Memory Scale scores),then A&B’s account would be invalid. Patients with damage to therecollection circuitry should show impaired recognition memoryunder conditions (e.g., yes-no testing) in which recollection con-tributes to the performance of healthy subjects (Yonelinas 1994;Yonelinas et al. 1996). Systematic testing is needed in which se-lective hippocampal (and other selective recollection loop dam-age) amnesics are evaluated in varying recognition test formats torelate their performance to the level of recollection employed byhealthy controls. Finally, A&B’s model would be seriously com-promised if healthy subjects did not carry out forced-choice recog-nition tasks (similar to the RMT) using familiarity only.

Unfortunately, the latter tests of A&B’s model rely on soundmethods for estimating recollection and familiarity. A&B them-selves review some of the difficulties with the remember-knowprocedure and the process dissociation procedure has been widelycriticised (e.g., Curran & Hintzman 1995). Following Yonelinas,we have just completed a series of studies with healthy subjectsusing direct fitting of models to recognition memory receiver op-erating characteristics. In each of the half-dozen separate yes-norecognition studies we have carred out, the fit of a familiarity-onlymodel was numerically superior to a dual-process model (famil-iarity plus recollection), in direct contrast to the results of Yoneli-nas et al. (1996). Moreover, neither model produced a statisticallyacceptable fit to the data obtained. A&B’s neat biological mappingof the dual-process accounts of recognition memory will fail if, asour data tend to indicate, normal yes-no recognition memory isnot well described by dual-process models. More directly, in onepilot experiment we had healthy subjects study eight word lists,each being followed by a recognition memory test. Half of thesetests (unpredictably) were yes-no, the remainder were two-alter-native forced choice (2AFC) format. As already indicated, a fa-miliarity-only model fit the yes-no data only modestly; however,the fit to the 2AFC data was much worse. If these early findingsare reliable, then the explanation for good 2AFC recognition per-formance by selective hippocampal amnesics cannot lie in A&B’ssuggestion that 2AFC tests can be performed accurately using fa-miliarity information only.

3. The relationship between familiarity and priming. One fur-ther problem with dual-process accounts of recognition is thatsome of their proponents (e.g., Jacoby 1991) have claimed that fa-miliarity in recognition equates with perceptual fluency/implicitmemory. As A&B point out (sect. 6) this seems wrong phenome-nologically (as shown by the alternative description of familiarityas a “know” state). Although they stop short of strong statements,A&B’s model clearly separates familiarity (perirhinal cortex) frompriming (which involves posterior cortical regions). Objections todual-process accounts of recognition based on contradictory evi-dence from experiments manipulating fluency (see Squire 1992,pp. 205–207) are neatly resolved by A&B’s framework.

4. Links to other theories of hippocampal function. As notedpreviously (Pickering 1997) it seems unparsimonious that the hip-pocampus is widely acknowledged to have other functions outsidethe memory domain. In fact, the hippocampus has been describedas part of a system of associative prediction and hence mismatchdetection (Gray 1982). A&B cite direct evidence for this in thework of Honey et al. (1998), and the involvement of the hip-pocampus specifically in recollection, but not familiarity, is con-sistent with these more general formulations. Neural networkmodelling by Levy (e.g., Prepiscus & Levy 1994) has shown that

the architecture of the hippocampus is suited to learning (long)chains of associations between items in a way that could supportboth recollection and mismatch detection. A&B’s model shouldprompt a search for a detailed understanding of the processes sup-porting recollection and familiarity.

Mere functional characterization is notenough to understand memory circuits

Alessandro TrevesProgramme in Neuroscience, SISSA, 34014 Trieste, [emailprotected] www.sissa.it/

Abstract: What exactly is going on via fornical connections? Aggleton &Brown’s target article correctly stresses their importance, but a detailedunderstanding of their role in memory appears to require fresh researchapproaches.

After having eaten with Malcolm Brown in a Kyoto restaurant in1995, I was trying to remember whether he had spelled out his an-terior thalamic view.

I turned to my hippocampus for help: “Can you recall thescene?”

Hipp: “Sure, he was wearing that nice green jacket.”“Do you remember what he was saying?”Hipp: “Yes. He was trying to get that ever-smiling waiter to un-

derstand that you would not eat any meat.”“But was he emphasizing your interaction with the mamillary

bodies?”Hipp: “Interaction? What does that mean?”“It means, what do you do with the mamillary bodies?”Hipp: “Oh, well, we talk to each other.”“You mean, same as with the entorhinal cortex?”Hipp: “Sure, with entorhinal too, we just talk!”Aggleton & Brown’s (A&B’s) is a very convincing and useful view

based largely on neuropsychological evidence. It is helpful to pointout the sometimes neglected connections via the fornix, which arecertainly there to play some important role; at the very least, theyallow the structures at the two ends of the line to “just talk.” I amnot sure, though, that conventional neuropsychological analyses ofhumans or animals, examining qualitative deficits in behaviour, cango beyond an initial clarification of the sort of messages that are ex-changed along that line, and why they need to be exchanged.

A very important approach to understanding memory opera-tions, if it were possible, would ideally be to wiretap all conversa-tions through the fornix, and analyze them with date, sender, re-ceiver and an interpreter to translate them from neurospeak.Neurophysiological research has made some progress in this di-rection (Rolls et al. 1998), exploring the information content ofneuronal messages from these parts of the brain not only qualita-tively but quantitatively. One perspective afforded by hippocam-pal recordings so far is the relative increase in the heterogeneityof significant correlates of neuronal activity from rodents(Gothard et al. 1996; O’Keefe & Burgess 1996) to primates (Ei-f*cku et al. 1995; Miyash*ta et al. 1989; Rolls & O’Mara 1995; Rollset al. 1993). If we were able to record from human hippocampalneurons, not at the gross scale allowed by PET and fMRI but in-dividually, and not sporadically but systematically, we might expecteven wider heterogeneity of neuronal concerns (Mirsky et al.1997). It is difficult to believe that such variety (yet to be demon-strated, I admit) could be encapsulated into simple, clear-cut cat-egories such as recognition and recall. A differential involvementof distinct units and connections in operations that “tax primarilyfamiliarity judgements” and in those that “retrieve episodic items”is quite conceivable, and the evidence reviewed is convincing.However, to go from this to the notion of two parallel systems, onedesigned for recognition and one for recall, one must jump quitea conceptual distance.

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (43)

Analyses of the requirements for internal consistency of the rel-evant network operations, such as those we have proposed (Treveset al. 1996b) are limited in that they do not relate directly to hu-man clinical categories, but this limitation can also be considereda virtue in that they invoke only intrinsic as opposed to anthropo-morphic information processing notions. Such analyses have, forexample, indicated the importance of quantitative variations in in-formation content-related parameters in determining the perfor-mance of a memory system. Thus, the differential implication ofdistinct circuits in recognition and recall tasks are more likely tobe one of the consequences of quantitative differences in, for ex-ample, activation sparsity or connectivity parameters, than theoutcome of evolutionary pressure to develop qualitatively sepa-rate systems for recognition and for recall.

Beyond a qualitative description of the correlates of neural ac-tivity, single and multiple single unit recording likewise yields ac-cess to the structure of neuronal representations, bringing out forexample quantitative differences in the way the same spatial corre-lates are encoded by neighboring populations (Treves et al. 1996a;1998). These structural differences may also be revealed by non-conventional use of neuropsychological tests, such as examining thedistribution of performance errors by memory patients of differentcategories in a Famous Faces test (Lauro-Grotto et al. 1997). Theactivation and connectivity patterns that underlie such differencesprobably satisfy internal consistency requirements in the organiza-tion of the networks involved (Fulvi-Mari & Treves 1998).

In conclusion, reviews like A&B’s illustrate the need to inte-grate more traditional approaches to memory research, those re-flecting the point of view of the subject, with all the approachesnow available, examining memory transactions from the point ofview of the actors – the neurons. Such contamination has in myview helped assess the perforant path, and it should prove no lessfertile once fornically extended.

The hippocampus and path integration

Ian Q. WhishawDepartment of Psychology and Neuroscience, University of Lethbridge,Lethbridge, Alberta, Canada, T1K 3M4. [emailprotected]

Abstract: Recent studies of the contribution made by the hippocampusto spatial behavior suggest that it plays a role in integrating and double in-tegrating distance and direction information using cues generated by self-movement. This and other evidence that the hippocampus plays a centralrole in spatial behavior seems inconsistent with proposals that it is pri-marily involved in episodic memory.

It is well known that damage to the hippocampus and/or fimbriafornix impairs the ability of rats to learn to swim to a hidden plat-form (place) in a swimming pool but not their ability to learn toswim to a visible platform (cue). The place/cue dissociation is cen-tral to theories that suggest a special spatial role for the hip-pocampus. Aggleton & Brown (A&B) however, propose that thedissociation supports their contention that the extended hip-pocampus plays a special role in episodic memory (sect. 5.1). Wequestion whether this claim advances our understanding of thefunction of the hippocampus.

We have been investigating an important feature of the placedeficit in hippocampal rats. Such rats, after training to swim to avisible cue in the swimming pool, will swim directly to the cue’sprevious location and search that location after the cue is re-moved. Four aspects of their performance suggest that they havelearned a place response as accurately as do control rats, eventhough the demands of the task did not require that they do so: (1)They swim accurately from start points located at different dis-tances and in different directions from the correct location(Whishaw & Jarrard 1995; 1996; Whishaw et al. 1995); (2) theysearch in the correct location indicating that they expect to find

the platform there; (3) they make an accurate swim from a novellocation (Whishaw 1998); and (4) they are more persistent thancontrol rats in returning to that location (Whishaw & Tomie1997b). Cuing the location of the platform does not simply allowthe hippocampal damaged animals to overcome a problem of slowlearning because if cued and hidden place trials are intermixed,hippocampal acquisition matches that of control rats (Whishaw &Tomie 1997a). Nor is the learning nonvisual, because if the ratsare blindfolded, they, like control rats, are impaired (Whishaw1998). These results indicate that the place deficit in hippocam-pal rats is not in the episodic domain of “knowing where” but inthe spatial domain of “getting there.” Independent studies usinga shrinking hidden platform provide support for the generality ofthis conclusion (Day & Schallert 1996).

We propose that hippocampal rats have an impairment in usingself-movement cues (idiothetic cues) and path integration (Whi-shaw et al. 1997). To know where they came from, animals canmake a record of their own movements and then integrate thatrecord to link the starting point to the goal. On a subsequent trial,they can double-integrate the record to generate a more direct re-sponse. This strategy of navigation is referred to as dead reckon-ing or sense of direction (Barlow 1964). To investigate the role ofthe hippocampus in path integration, we designed a task in whicha rat leaves a hidden burrow to forage for a piece of food placedat an indeterminate location. The piece of food is large enoughthat when found, it is carried back to the burrow for consumption.Once the rats find and carry food to a home base, they are blind-folded and started from a novel location. Control rats return ac-curately to the new starting location, whereas hippocampal rats donot (Whishaw & Maaswinkel 1998).

Information obtained from path integration can be used not onlyfor dead reckoning, but also to create a spatial frame of referenceand to calibrate the distances and directions between objects. Ac-cordingly, information from path integration can be used to solvelaboratory problems such as spatial reversals presented to rats byA&B. When spatial cues that once signaled a place suddenly be-come meaningless, the animal can maintain its spatial orientationusing self-movement cues until it discovers which distal cues sig-nal the problem’s new solution (Whishaw & Tomie, 1997a).

Our conclusion that the hippocampal formation has a funda-mental role in spatial behavior, perhaps even in a single computa-tional form of spatial behavior, is consistent with a growing litera-ture inspired by the spatial hypothesis (O’Keefe & Nadel 1978).We argue that the hippocampal rat’s episodic abilities combinedwith the poverty of the explanation for making predictions limitsthe episodic theory’s utility. In other words, the hazard in arguingthat brain regions play fundamental mnemonic roles either inde-pendently or as systems, detracts from the goal of explaining howthose same brain regions contribute to adaptive behavior.

The medial dorsal nucleus of the thalamus is not part of a hippocampal-thalamicmemory system

Menno P. Witter and Ysbrand D. Van der WerfGraduate School of Neurosciences Amsterdam, Research Institute-Neurosciences, Vrije University, Department of Anatomy and Embryology,1081 BT Amsterdam, The Netherlands. {mp.witter.anat; yd.van der werf.anat}@med.vu.nl

Abstract: Aggleton & Brown propose that familiarity-based recognitiondepends on a perirhinal-medial dorsal thalamic system. However, con-nections between these structures are sparse or absent. In contrast, theperirhinal cortex is connected to midline/intralaminar nuclei. In a human,a lesion in this thalamic domain, sparing the medial dorsal nucleus, im-paired familiarity-based recognition while sparing recollective-based rec-ognition. It is thus more likely that the intralaminar/midline nuclei areinvolved in recognition.

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (44)

Aggleton & Brown (A&B) are to be complimented for their impor-tant conceptual contribution about memory formation and re-trieval. Their arguments for a dissociation between recollective- andfamiliarity-based recognition are impressive. Their claim that forthe former the hippocampal-fornix-anterior thalamic system is of vi-tal importance accords well with our conclusions based on the neu-ropsychological assessment of a recently collected series of patientswith restricted thalamic infarctions, taken in conjunction with a re-cent meta-analysis of the literature (van der Werf et al., unpublishedobservations). We concluded that the fibre tract connecting themammillary bodies with the anterior nuclei of the thalamus (themammillo-thalamic tract) is necessarily affected in those subjectswhere damage to the thalamus is associated with an amnesic syn-drome (cf. Graff-Radford et al. 1990; Von Cramon et al. 1985).

In contrast, we doubt whether A&B’s second proposition, thatfamiliarity-based recognition is mediated by a perirhinal-medialdorsal system, stands scrutiny. Based on two lines of evidence, weparticularly question the participation of the medial dorsal nu-cleus of the thalamus in this recognition system. First, monosy-naptic projections from perirhinal cortex to the medial dorsal nu-cleus appear either sparse or nonexistent in animal species.Although the authors refer to two tracing reports in the monkey(sect. 7, para. 2; Aggleton et al. 1986a; Russchen et al. 1987), fromthose papers it is clear that perirhinal-medial dorsal projectionsare sparse. More detailed studies in the rat (Burwell et al. 1994;Chen & Burwell 1996; our own unpublished observations) and thecat (Room & Groenewegen 1986; Witter & Groenewegen 1986)indicate that the perirhinal cortex projects massively to the poste-rior complex of the thalamus and to certain components of the in-tralaminar/midline thalamic nuclei, including the reuniens andcentral medial nuclei. In the monkey, a comparable projectionpattern prevails (Suzuki, personal communication).

Second, we have recently described a patient suffering fromcognitive deficits after a thalamic infarction (Van der Werf et al.1998). This patient shows a dissociation between the two types ofrecognition described in the target article. It is interesting, how-ever, that this dissociation takes the form of a sparing of recollec-tive-based recognition with a deficit in familiarity-based recogni-tion. This is evident from normal scores on the Visual AssociationLearning Task (Lindeboom 1989; Rombouts et al. 1997), a testprobing the recall of visual associations. On the other hand, resultson the classical Recognition Memory Test for faces (Warrington1984) are deficient. If one takes the viewpoint of A&B (sect. 6,para. 7), the latter finding indicates an impaired familiarity-basedrecognition. The lesion in this patient was shown to spare the me-dial dorsal and the anterior nuclei as well as the mammillothala-mic tract, but to involve the region of the right lateral internalmedullary lamina (IML). In contrast, patients with similarly sizedlesions just outside the lateral IML, involving the right medial dor-sal nucleus, do not show an impairment of familiarity-based recog-nition (Kritchevsky et al. 1987; Shuren et al. 1997). This observa-tion corroborates our conclusion that the IML and/or nucleiwithin it, rather than the medial dorsal nucleus, is involved in thistype of recognition. It is of interest that SPECT (single photonemission computer tomography) imaging and extensive neu-ropsychological testing of our patient led us to conclude that thispatient had a dysfunction of “executive” or prefrontal” cognitiveprocesses rather than hippocampal, formation-dependent pro-cesses. This combination of familiarity-based recognition and pre-frontal symptoms is in line with the account given in section 7,paragraphs 2–4.

That the perirhinal cortex may contribute in a unique way to thefunctioning of the medial temporal lobe memory system, as pro-posed earlier by Eichenbaum et al. (1994), is in line with our re-cent electrophysiological and anatomical findings that the perirhi-nal cortex has direct reciprocal connections to CA1 and thesubiculum. Those connections show a topographical organizationthat differentiates them from the more widespread projectionsthat originate from the entorhinal cortex. Based on these obser-vations, we proposed that the perirhinal inputs to the hippocam-

pal formation may be crucially involved in ongoing comparisonsof newly acquired information as carried out by the hippocampalsystem (Naber et al., submitted). As such, this proposition wouldbe complementary to a familiarity-based recognition process asproposed by A&B. Whether the perirhinal cortex has specificfunctional relations with thalamic domains, different from thoseassociated with the hippocampal system, remains to be estab-lished. If so, likely candidates are the midline/intralaminar nuclei(see also sect. 5.1, para. 6), instead of the medial dorsal nucleus.

ACKNOWLEDGMENTSThis research has been supported by grants 903-47-008 and 970-10-012 ofthe Netherlands Organization for Scientific Research (NWO).

The neural substrates of recollection and familiarity

Andrew P. Yonelinasa, Neal E. A. Krolla, Ian G. Dobbinsa,Michele Lazzaraa, and Robert T. KnightbaDepartment of Psychology, University of California at Davis, Davis, CA95616; bDepartment of Psychology, University of California at Berkeley,Berkeley, CA 94720. {apyonelinas; neakroll; igdobbins; mmlazzara}@[emailprotected]

Abstract: Aggleton & Brown argue that a hippocampal-anterior thalamicsystem supports the “recollection” of contextual information about previ-ous events, and that a separate perirhinal-medial dorsal thalamic systemsupports detection of stimulus “familiarity.” Although there is a growingbody of human literature that is in agreement with these claims, when rec-ollection and familiarity have been examined in amnesics using the processdissociation or the remember/know procedures, the results do not seemto provide consistent support. We reexamine these studies and describethe results of an additional experiment using a receiver operating charac-teristic (ROC) technique. The results of the reanalysis and the ROC ex-periment are consistent with Aggleton & Brown’s proposal. Patients withdamage to both regions exhibit severe deficits in recollection and smaller,but consistent, deficits in familiarity.

Aggleton & Brown (A&B) argue that a hippocampal-anterior thal-amic system supports the “recollection” of contextual informationabout previous events, and that a separate perirhinal-medial dor-sal thalamic system supports detection of stimulus “familiarity.”There is a growing body of human literature that is consistent withthese claims, showing that recognition memory judgments reflectthe separate contributions of recollection and familiarity pro-cesses. These two processes are functionally independent (e.g.,Atkinson & Juola 1974; Jacoby 1991; Mandler 1980) and they ex-hibit separate electrophysiological correlates (e.g., Düzel et al.1997), suggesting that they reflect partially distinct cortical gen-erators. Some studies of human amnesia support A&B’s proposedmapping of these processes to areas within the medial temporallobes. Other recent reports appear to be in conflict, however. Re-examining the human literature on recollection and familiarity inamnesia finds that the current evidence is, in fact, quite consistentwith their proposal.

According to A&B’s proposal, patients with damage to both thehippocampal and perirhinal systems – which includes most of thehuman amnesic patients that have been studied – should exhibitdeficits in both recollection and familiarity. Direct tests of recog-nition memory provide some support for this prediction in show-ing that amnesics typically exhibit recognition memory deficits.However, because recognition memory judgments can be basedon either recollection or familiarity, these results do not show ifthe deficits are in both recollection and familiarity, or restricted toa single process. Therefore, it is necessary to look to proceduresthat provide separate measures of these two processes.

A&B describe a study by Knowlton and Squire (1995) using theremember/know procedure (Tulving 1985a) that found amnesics

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (45)

were less likely to respond that they “knew” on the basis of famil-iarity that test items were studied, indicating that familiarity wasdisrupted. However, A&B also describe a study by Verfaellie andTreadwell (1993) using the process dissociation procedure (Ja-coby 1991), that concluded that familiarity was preserved in am-nesia. More recent studies using the remember/know procedurehave led to different conclusions than either of the earlier studies.For example, Schacter et al. (1996) found that the proportion ofknowing responses actually increased significantly for amnesic pa-tients relative to healthy controls, indicating that amnesia was as-sociated with an increase in familiarity. Given that all three ofthese studies included patients with widespread damage to themedial temporal lobes, if A&B are right, these patients shouldhave exhibited deficits in both recollection and familiarity. On thesurface, therefore, human amnesia data does not appear to pro-vide consistent support for their new proposal.

However, a closer examination of these studies finds that theprevious interpretations of these results were incomplete in thatthey did not fully account for response biases or, in the case of theremember/know studies, did not correct their measures to ac-count for the mutual exclusivity between the remember/know re-sponses. That is, in all of the reported studies the amnesics exhib-ited higher false alarm rates than the controls and this biased theestimates of recollection and familiarity. Moreover, the probabil-ity of a “know” response in the remember/know procedure ismathematically constrained by the proportion of remember re-sponses, so that “knowing” responses by themselves do not pro-vide an accurate measure of familiarity. In a recent article (Yoneli-nas et al. 1998), we reanalyzed the results of these earlier studiesusing a dual-process signal-detection model (Yonelinas 1994) thatallowed for the independent contribution of recollection and fa-miliarity and incorporated signal detection theory to accommo-date differences in false alarms. We also tested additional am-nesics using a receiver operating characteristics (ROC) analysis tovalidate the model and to derive estimates of recollection and fa-miliarity in these patients.

The ROC analysis showed that the model provided an accurateaccount of recognition performance. Most important, however,was that all three estimation procedures converged in showingthat both recollection and familiarity were disrupted in amnesics.Figure 1 summarizes the results of the remember/know, processdissociation, and ROC experiments. The results provide consis-

tent support for A&B’s proposal, in that patients with damage toboth the hippocampal and perirhinal systems exhibit the expecteddeficits in both recollection and familiarity.

The results are also consistent with previous studies showingthat amnesia is associated with a disproportionate disruption ofrecollection compared to familiarity (e.g., Kroll et al. 1996). Al-though familiarity was consistently disrupted in the amnesics, rec-ollection was disrupted to a much greater extent. This dispropor-tional deficit was observed in every condition of every experimentwe examined. Note that recollection and familiarity are measuredon different scales in Figure 1. The disproportional deficit in rec-ollection is also observed, however, when the familiarity d9 scoresare converted to probabilities.

These results join a growing body of studies that show recollec-tion and familiarity to be functionally dissociable memory pro-cesses, and suggest that they rely on distinct anatomical regions.However, a critical test of A&B’s specific proposal will be to de-termine whether patients with damage that is restricted to the hip-pocampal system or to the perirhinal system will exhibit selectivedeficits of recollection and familiarity.

Remembering the hippocampus

Stuart M. Zolaa,b and Larry R. Squirea,b,c

aVeterans Affairs Medical Center, San Diego, CA 92161; bDepartments of Psychiatry and Neurosciences, University of California, La Jolla, CA92093; cDepartment of Psychology, University of California, La Jolla, CA92093. {szola; lsquire}@ucsd.edu

Abstract: The proposal that the hippocampus is important for the encod-ing of episodic information, but not familiarity-based recognition, is in-compatible with the available data. An alternative way to think about func-tional specialization within the medial temporal lobe memory system issuggested, based on neuroanatomy.

Damage to the hippocampus and adjacent, anatomically relatedstructures impairs memory in rats, monkeys, and humans (Bunsey& Eichenbaum 1996; Mishkin & Murray 1994; Squire 1992;Squire & Zola-Morgan 1991). In both humans and monkeys, it isestablished that memory impairment becomes more severe as

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Figure 1 (Yonelinas et al.). Estimates of recollection and familiarity in amnesics and controls derived using the remember/know (R/K), the process dissociation (PDP), and the receiver operating characteristic (ROC) procedures. The R/K data reflect the average esti-mates derived from Knowlton and Squire (1995), Schacter et al. (1996 and 1997); the PDP data are from Verfaellie and Treadwell (1993);and the ROC data are from Yonelinas et al. (1998). All three procedures showed that amnesics exhibited a reduction in recollection anda smaller but consistent reduction in familiarity.

Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (46)

more of these structures are damaged (Rempel-Clower et al.1996; Zola-Morgan et al. 1994). As we have pointed out previously(Zola-Morgan et al. 1994), this fact is consistent with the possibil-ity that the different structures make different contributions tomemory functions. The target article by Aggleton & Brown (A&B)is the most recent of several efforts to consider what these differ-ent contributions might be (Eichenbaum et al. 1994; Gaffan1994a; Vargha-Khadem et al. 1997). A&B propose that the hip-pocampus itself (and its diencephalic targets) is important for theencoding of episodic information. By contrast, the judgment of fa-miliarity (simple recognition memory) does not require the hip-pocampus, but depends instead on the perirhinal cortex (and itsdiencephalic targets).

The attempt to find functional specialization within the medialtemporal lobe is worthwhile, but the view proposed by A&B en-counters a number of problems. First, they view the distinctionbetween episodic and semantic memory (Tulving 1991) and thedistinction between remembering and knowing (Tulving 1985b)as closely related to their own. These distinctions have been use-ful and important for separating medial temporal lobe functionfrom frontal lobe function, but it is not at all clear that they parti-tion medial temporal lobe function itself (Knowlton & Squire1995; Squire & Zola 1991). Second, the specific distinction pro-posed by A&B (episodic memory versus recognition memory) isnot supported by the data. Here we focus on A&B’s view thatrecognition memory does not require the integrity of the hip-pocampus. We first mention findings from humans and rats andthen consider in more detail the findings from monkeys.

First, amnesic patients R. B. and G. D., who had lesions restrictedto the CA1 region of the hippocampus, were impaired at recogni-tion memory (Reed & Squire 1997; Zola-Morgan et al. 1986). In pa-tients with damage involving all the cell fields of the hippocampusand some cell loss in entorhinal cortex (patients L. M. and W. H.,Rempel-Clower et al. 1996), recognition memory was also unmis-takably impaired (Reed & Squire 1997). Second, rats with hip-pocampal lesions were impaired on the delayed nonmatching-to-sample task after a 2-minute delay (Mumby et al. 1995).

Third, the weight of evidence is that monkeys with hippocam-pal lesions have impaired recognition memory. There have beenseven studies involving lesions limited to the hippocampus that as-sessed performance on either the delayed nonmatching-to-sam-ple task or the visual paired-comparison task. The lesions wereproduced by ischemia, radio-frequency, or ibotenic acid. Four ofthe studies are published or in press (Alvarez et al. 1995; Beason-Held et al. 1998; Murray & Mishkin 1998; Zola-Morgan et al.1993), and three are preliminary reports published in Abstract

form (Clark et al. 1996; Teng et al. 1998; Zola et al. 1998). Of these,only one study (Murray & Mishkin 1998) found no deficit. Thisstudy was the only one that used two-stage surgery and the onlyone in which the nonmatching rule was trained preoperatively (fordiscussion, see Zola & Squire 1998).

Recent results from our laboratory (Fig. 1) show that, as mea-sured by the delayed nonmatching-to-sample task, hippocampallesions made by ischemia, radio-frequency, or ibotenic acid impairrecognition memory in monkeys to a similar degree (Zola et al., inpress). For all three groups, surgery was done in a single stage, andthe nonmatching rule was trained postoperatively. The deficit waspresent whether animals were removed from the test apparatusduring the delay intervals or allowed to remain in the apparatus(Teng et al. 1998).

Questions have been raised as to whether ischemic lesions canprovide a valid method for assessing hippocampal function (A&B;Bachevalier & Meunier 1996; Nunn & Hodges 1994). If ischemiacaused cell loss in the hippocampus but also caused neuronal dys-function (not cell death) in other structures important for mem-ory, then the hippocampal cell loss that is detected in histologicalexams need not be the cause of memory impairment. This issuecan be approached experimentally by comparing the effects onmemory of ischemic lesions and similarly-sized stereotaxic lesionswithin the hippocampal region.

When this was done, the effects of ischemia (Zola-Morgan et al.1992) closely approximated the effects of radio-frequency lesions(Alvarez et al. 1995; also see Squire & Zola-Morgan 1996) (Fig. 1).A&B point to a trend in our data, whereby ischemic animals per-formed worse across the delays of the performance test than theanimals with radio-frequency lesions (79.3% correct vs. 84.6%correct). At the longest delay at which both groups were tested (10min), the corresponding scores were 68.1% correct and 68.5%correct, respectively (p > 1.0). We do not know how to evaluatesmall numerical differences that do not approach statistical sig-nificance. In any case, the nature of ischemic lesions could befruitfully pursued in the rat. What is needed are new parametricstudies that assess memory performance as a function of the sizeof stereotaxic lesions (for this approach, see Moser et al. 1993;1995). One could then ask whether ischemic lesions of varying sizeaffect performance to the same extent as similarly sized stereotaxiclesions, or whether the impairment following ischemia is more se-vere.

A consideration of functional specialization within the medialtemporal lobe should be guided by the neuroanatomy of the re-gion. The medial temporal lobe region is a convergence zone forunimodal and polymodal inputs from neocortical association ar-

Commentary/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Figure 1 (Zola & Squire). Monkeys with lesions limited to the hippocampal region (the dentate gyrus, the cell fields of the hippocam-pus, and the subiculum) made by ischemia (ISC, n 5 4), radio-frequency (RF, n 5 4), or ibotenic acid (IBO, n 5 5) were similarly im-paired on the delayed nonmatching-to-sample task. When the scores at all delays $ 15 sec were combined, each lesion group was im-paired (N 5 10 normal control monkeys; all ps ,0.05). (From Zola & Squire, in press).

Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (47)

eas, areas important for information processing and ultimately forlong-term memory storage. The first clue to how the componentsof the medial temporal lobe function comes from the fact that dif-ferent parts of the neocortex access the medial temporal lobe atdifferent points (Suzuki & Amaral 1994a). For example, thehigher visual areas TE and TEO project preferentially to theperirhinal cortex. Conversely, input about spatial information thatcomes to the medial temporal lobe from parietal cortex arrives ex-clusively at the parahippocampal cortex. With this anatomy inmind, one might expect perirhinal cortex to be especially impor-tant for visual memory and the parahippocampal cortex to be im-portant for spatial memory. Findings to date are consistent withthis expectation (Malkova & Mishkin 1997; Parkinson et al. 1988;Ramus et al. 1994; Teng et al. 1997).

The hippocampus itself is the final stage of convergence withinthe medial temporal lobe, receiving input from both the perirhi-nal and parahippocampal cortices, as well as from the entorhinalcortex. The entorhinal cortex receives about two-thirds of its cor-tical input from the parahippocampal and perirhinal cortices. Rea-soning again from anatomy, one might expect the hippocampus tobe involved in both visual and spatial memory, combining all theoperations of memory formation that are carried out indepen-dently by the more specialized structures that project to it. Again,the data seem to bear this out. In patients with damage restrictedto the CA1 region of hippocampus, the memory impairment ex-tends across all sensory modalities and types of material. Thememory impairment is only modest because considerable mem-ory function can be supported by the cortical components of themedial temporal lobe system.

There has been an eagerness to identify a separate and distinctsubfunction for the hippocampus itself. If there is some indepen-dent function that is not shared at all by adjacent cortex, then forsome memory operation the CA1 patient should exhibit a deficitas severe as that observed in a patient like H. M., whose lesion in-cludes the hippocampus as well as most of the medial temporallobe. Yet the CA1 patient presents as a mild version of H. M. In-deed, regardless of how memory is measured, the impairment isless severe following a small lesion within the hippocampus (am-nesic patients R. B. and G. D.) as compared to a large medial tem-poral lobe lesion (patient H. M.). A&B’s proposal provides no ac-count of this simple fact.

Finally, it is unclear to us why evidence that hippocampal le-sions do impair recognition memory has sometimes been consid-ered unwelcome on theoretical grounds. Even a recognition (orfamiliarity) decision requires that a link be made between thestimulus presented in the retention test and what was presentedduring learning. The recognition test does not ask the subjectwhether an item has ever been seen or whether one knows whatan item is. It asks whether an item that appeared recently is fa-miliar. It is this association, this relational work, that many havesupposed is at the heart of declarative memory and hippocampalfunction in both humans and nonhuman animals (Eichenbaum1997; Squire 1992; Sutherland & Rudy 1989).

Authors’ Response

Thanks for the memories: Extending thehippocampal-diencephalic mnemonic system

John P. Aggletona and Malcolm W. Brownb

aSchool of Psychology, Cardiff University, Cardiff, CF1 3YG, Wales;bDepartment of Anatomy, University of Bristol, Bristol, BS8 1TD, [emailprotected] www.cf.ac.uk/uwc/psych/[emailprotected] www.bris.ac.uk/depts/anatomy

Abstract: The goal of our target article was to review a number ofemerging facts about the effects of limbic damage on memory inhumans and animals, and about divisions within recognition mem-ory in humans. We then argued that this information can be syn-thesized to produce a new view of the substrates of episodic mem-ory. The key pathway in this system is from the hippocampus tothe anterior thalamic nuclei. There seems to be a general agree-ment that the importance of this pathway has previously been un-derestimated and that it warrants further study. At the same time,a number of key questions remain. These concern the relationshipof this system to another temporal-lobe/diencephalic system thatcontributes to recognition, and the relationship of these systemsto prefrontal cortex activity.

One of the purposes of a BBS target article is to elicit de-bate by raising a potentially contentious viewpoint and thenseeing how it stands up under the scrutiny of expert com-mentary. After the dust has settled it is easier to determinewhether the consensus view in the field has altered andwhere the key areas of debate are. We feel that the insight-ful commentaries contributed have helped to achieve bothgoals. Thus it seems apparent that there has been a shift inour understanding of the neuroanatomical basis of the am-nesias, leading to a reappraisal of what that information cantell us about neural systems underlying aspects of memory(Nadel et al.). This is most evident in the general accep-tance of the importance of the direct outputs from the hip-pocampus to the rostral diencephalon, and the associatedloss of the division between temporal lobe and diencephalicamnesias. At the same time, there is much uncertainty overthe relationships of these pathways to cortical areas in thetemporal and frontal lobes. To address these issues it is per-haps simplest to restate the “main features of the proposedmodel” and to relate these to the various commentaries.

R1. The hippocampal efferents via the fornix to the mamillary bodies and medial thalamus,most especially the anterior nuclei, are vitalfor normal hippocampal activity

The fornix contains a variety of pathways that project bothto and from the hippocampus. That cutting the fornix oftenmimics the effects of hippocampectomy is obviously a con-sequence of this fact. Less straightforward has been theidentification of those pathways that are most critical, andwhether they are afferent or efferent to the hippocampus.The target article sought to emphasize the importance ofthe hippocampal projections to the anterior thalamic nu-clei, both those that are direct and those that are indirectvia the mamillary bodies. This emphasis was approved ei-ther explicitly (Burwell & Eichenbaum, Dalyrymple-Alford et al., Foster, Gabriel & Smith) or tacitly by the

Response/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (48)

large majority of commentators. This is noteworthy as itmarks a clear shift away from the influential notion of sep-arate temporal lobe and diencephalic systems in normalepisodic memory.

Knowlton made the point that our model overempha-sized the hippocampal/diencephalic connections to thedetriment of those between the hippocampus and the pre-frontal cortex. Although we did acknowledge the likely im-portance of these prefrontal connections (see sect. R9),they were not included in Figure 1, as this would have dis-tracted from the chief purpose of the article, which was toconsider the hippocampal/anterior-thalamic axis. More-over, a recent series of disconnection studies in our labora-tory (Warburton et al., unpublished findings) has furtheremphasized the importance of these hippocampal-thalamicconnections. We examined the effects on rat behaviour ofcrossed unilateral lesions involving the fornix in one hemi-sphere and the anterior thalamic nuclei in the other hemi-sphere. Such lesions, which disconnect the hippocampusfrom the anterior thalamic nuclei, led to clear deficits ontests of spatial working memory. This was not found for uni-lateral lesions in the same hemisphere. It is interesting thatthis disconnection effect was observed only when the fornixlesion included the hippocampal commissure (Olton et al.1982). We have also examined crossed lesions between thehippocampus and the anterior thalamic nuclei and foundsimilar results. These results do not show that other forni-cal connections (such as those with the basal forebrain andseptum) are not important, but they do strongly indicatethat the similarities between the spatial memory deficits ob-served after anterior thalamic lesions and hippocampal le-sions are more than just superficial.

Krieckhaus has discussed a possible anomaly with themodel whereby mamillary body lesions could lead to agreater deficit than fornix lesions. He offers an ingeniousaccount to explain how this could occur, but we are unawareof relevant examples of animal studies where this unusualpattern of results is found. Nevertheless, anterior thalamicdamage that involves the lateral dorsal nucleus could bemore disruptive than fornix damage as this nucleus receivesnonfornical inputs from the subiculum (Aggleton & Saun-ders 1997; Warburton & Aggleton 1999; Warburton et al.1997). Although earlier clinical reviews indicated thatfornix damage might have little or no effect on memory(e.g., Garcia-Bengochea & Friedman 1987), subsequentclinical studies have consistently linked fornix damage withmemory loss (see sect. R3).

R2. This extended hippocampal-diencephalicsystem becomes more diffuse beyond theanterior thalamic nuclei, but includes twomajor components. Whereas one componentreturns to the temporal lobe, principally viathe cingulum bundle, the other projects toprefrontal and cingulate cortices

Little was said about these proposals, presumably becausethey are largely drawn from established anatomical knowl-edge. Parker did, however, report a lesion study in whichcingulum bundle damage was without effect, even thoughthe task is sensitive to hippocampal/fornix/anterior-thala-mic damage. This is potentially important and may point tothe fact that partial damage to these more diffuse, multiple

pathways is unlikely to produce the full behavioural deficit.Surprisingly little is known about the functions of the cin-gulum bundle and it is hoped that these proposals will helpdraw attention to this structure.

R3. This extended hippocampal-diencephalicsystem is critical for episodic memory anddamage to the component structures canresult in similar memory impairments

The closer linking of temporal lobe and diencephalic am-nesias was commended by a number of commentators(Gabriel & Smith, Markowitsch, Nadel et al., Picker-ing, Treves). Further support was provided by a review ofthalamic infarcts where it was found that anterograde am-nesia was best correlated with damage to the mamillothal-amic tract (Witter & Van der Werf). The failure of lesionsrestricted within the medial dorsal thalamic nucleus to pro-duce anterograde amnesia (Kritchevsky et al. 1987, cited byKapur) provides more indirect support for the current pro-posals.

A key test of this proposal is that fornix damage shouldbe sufficient to produce an anterograde amnesia. To inves-tigate this further we have examined a series of 10 people,all of whom have had surgery to remove colloid cysts in thethird ventricle (Aggleton, unpublished findings). This con-dition is of interest as it is sometimes, but not always, asso-ciated with loss of the fornix. Clear and quite severe deficitsin the learning of new verbal and nonverbal material werefound, but this was only in the subgroup of three caseswhere the fornix was interrupted bilaterally (Fig. R1).Other factors, such as ventricular enlargement, could not

Response/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Figure R1. Mean scores (1s.e.) on the subscales of the Wechs-ler Memory Scale revised (WMSr) and mean total score on theWarrington Recognition Memory test (RMT) for a group of sevensubjects (white bars) who required surgical removal of a colloidcyst but in whom the fornix remained intact. The comparisonscores (shaded bars) are for three colloid cyst cases where thefornix was interrupted bilaterally. Normal scores on the WMSr are100, whereas 100 represents the maximum score on the RMT.

Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (49)

be correlated with memory loss. These results provide fur-ther support for other clinical reports linking fornix damagewith amnesia (Calabrese et al. 1995; Gaffan & Gaffan 1991;McMackin et al. 1995).

The significance of the mamillary bodies was questionedby Kapur in light of two new cases where only moderatememory problems were associated with selective mamillarybody pathology. Such cases are clearly of enormous interest,but in view of the direct projections from the hippocampusto the anterior thalamic nuclei it is perhaps not surprisingthat less severe memory problems might be found. The is-sue of mildness of amnesia was also raised by Zola &Squire, who questioned how the severity of the amnesia incases like H. M. could be placed alongside the less severeamnesias in cases such as R. B. with anoxia and associatedgross pathology in the CA1 subfield of the hippocampus.This concern arises because both suffered from hippocam-pal dysfunction and this should be the key predictor of theirloss of episodic memory. It does not seem appropriate, how-ever, to expect such cases to be comparable in this regard asthe extent of hippocampal loss is not equivalent and animalstudies have shown that this is potentially a very importantfactor (Moser et al. 1995; Nunn & Hodges 1994). There isnothing in our model that precludes this simple conclusion.In addition, the loss of adjacent regions, such as the en-torhinal cortex (in H. M.), will further exacerbate the mem-ory loss in such subjects (Corkin et al. 1997).

Perhaps more critical is the suggestion that we haveoveremphasized the anterior thalamic nuclei at the expenseof other thalamic nuclei, in particular the intralaminar nu-clei (Mair et al., Witter & Van der Werf). That damageto the intralaminar nuclei can affect mnemonic perfor-mance is now becoming increasingly accepted (Bentivoglioet al. 1997): the work of Mair and his group has been veryimportant in this regard. Nevertheless, the extent to whichthese regions are important for the encoding and consoli-dation of new episodic information is far less certain. Manyof the animal lesion studies suffer the inevitable problem ofeither direct damage to adjacent nuclei (including the an-terior thalamic nuclei) or damage to tracts passing withinthe thalamus (including those between the prefrontal cor-tex and the medial dorsal thalamic nucleus). There are alsoimportant questions over the extent to which certain be-havioural tests that are sensitive to damage in this regioncan be regarded as assessing episodic memory (see sect.R4.1).

Although the intralaminar nuclei contribute to memory,and receive direct inputs from the perirhinal cortex (Wit-ter & Van der Werf), they lack direct hippocampal inputs.This is in marked contrast to the anterior thalamic nuclei.This fact alone points to a qualitatively different contribu-tion to memory. This view is reinforced by Mair et al., whor*port examples of dissociations between the effects of hip-pocampal lesions and lesions of the intralaminar region inrats. This can be contrasted with the effects of anterior thal-amic lesions, which often resemble hippocampal lesions(see sect. R1). It has been suggested, for example, that theintralaminar and midline nuclei may have much more gen-eral regulatory effects on the storage and retrieval of re-sponse habits and sensory memories, respectively (Ben-tivoglio et al. 1997). Such damage may also contribute to theextent of retrograde amnesia (Paller et al. 1997). Thus, al-though damage to the intralaminar and midline nuclei willexacerbate the memory problems associated with anterior

thalamic dysfunction, we regard the intralaminar nuclei aspart of an essentially different system and not one that hasintegral links with the hippocampus.

We proposed that damage to different parts of this hip-pocampal-diencephalic system will produce similar mem-ory impairments. This is because key components – thehippocampus and the anterior thalamic nuclei – can be re-garded as acting in series (Kapur). This is not to imply thatthe anterior thalamic nuclei do not have their own specificcontribution, for example, through ascending cholinergicor reticular innervations, or through connections with thecingulate cortices. Gabriel, who has pioneered much re-search on the anterior thalamic nuclei and their interactionwith the cingulate cortices, argues (Gabriel & Smith) thatour description of the contributions of these nuclei is toonarrow (see also Dalrymple-Alford et al.). We readily ac-cept these views, as it was not our goal to provide a com-prehensive account of all of the functions of the anteriorthalamic nuclei (or the hippocampus for that matter), butto consider the particular issue of how they might con-tribute to episodic memory in humans. In this regard Dal-rymple-Alford et al. make the intriguing suggestion that thecontribution of the anterior thalamic nuclei to consolida-tion might differ from that of the hippocampus in beingmore transient, and this warrants further investigation.

A different approach is taken by Markowitsch who ar-gues that the proposed system is too narrow as it ignores thecontribution of structures such as the amygdala, which ap-pears to be important for aiding the memory of emotion-laden material. Although we cannot agree with his assertionthat amnesia is to a large extent the failure to rememberemotion-laden information, we of course recognise thatsuch mechanisms exist. In a fully comprehensive model onewould have to consider how such arousing mechanisms in-teract with the systems highlighted in our proposal. Perhapsthe most pertinent question is whether regions such as theamygdala exert their influence on emotion-laden episodicinformation via the extended hippocampal-diencephalicsystem or at some other level (e.g., association cortex). Infact, the amygdala and hippocampus are interconnected,both directly and indirectly (Aggleton 1986), and there isevidence that at least some of the effects of emotionalarousal are via this interaction (Cahill & McGaugh 1996;Packard et al. 1994).

R4. The hippocampal-diencephalic system neednot be vital for efficient recognition, as thissystem is only required for recollection-basedrecognition (“remembering”) so thatfamiliarity-based recognition (“knowing”) can still function. The perirhinal cortex, in contrast, is especially important for familiarity-based recognition

This seemingly simple proposal has many ramifications andso it is not surprising that it was addressed in many of thecommentaries. These can be broken down into four mainareas: (1) implications for behavioural testing, (2) evidencefor this division within recognition in normal and clinicalpopulations, (3) evidence of spared recognition in humansand animals following hippocampal lesions, and (4) evi-dence that the perirhinal cortex has a specific role in recog-nition in humans.

Response/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (50)

R4.1. Behavioural testing and episodic memory. A num-ber of commentators applauded the division within mem-ory, as it highlighted the weakness of relying on tests ofrecognition to uncover substrates for episodic memory(Mumby, Nadel et al., Pickering). Others suggested thatwe had implied that spatial memory and episodic memorywere interchangeable (Bures & Fenton) and thereforequestioned the implications of some of our findings con-cerning selective diencephalic lesions and spatial memory.In fact, we were at pains never to use the terms spatialmemory and episodic memory in an interchangeable man-ner, while acknowledging that spatial information is oftencrucial to episodic memory. Although tests of spatial mem-ory in rats are of enormous value in examining functionallinks with the hippocampus, they do not represent directtests of episodic memory (see Whishaw).

This important point is underlined in a recent, ingeniousstudy by Clayton and Dickinson (1998), which shows thatto demonstrate anything closely akin to episodic memory itis necessary to show that the animal has learned not only thewhere, but also the what and when of an event. This theysuccessfully achieved in scrub jays that were hoarding dif-ferent kinds of foods that differed in the rates at which theydecayed or disappeared once hoarded. To our knowledge,there have been no comparable demonstrations with ratsand it would be an enormous advance if such tests could bedevised (see also Bures & Fenton). Mumby offers someuseful thoughts on this issue when he discusses the fasci-nating question of how it might be possible to distinguishtwo processes within recognition in rats. The study by Clay-ton and Dickinson (1998) also reveals the limitations of re-lying on delayed matching or nonmatching tasks to assessepisodic memory (Mair et al.), as they do not capture thefull extent of episodic processes and can be solved by sim-pler means.

A closely related issue was raised by Foster when con-sidering the effects of hippocampal lesions on go/no gotasks. He described how hippocampectomised rats wereable to perform an unsignalled go/no go task, and arguedthat because this task has a recall element our proposalshould have predicted a deficit. Apart from the potentialproblems of motor mediation that rats can use to bridge de-lays, we are reluctant to assume that this task must beepisodic in nature, given the stringent demands for such atask (Clayton & Dickinson 1998). At the same time we areanxious not to fall into the circular trap of implying that be-cause task performance is spared the task does not taxepisodic or episodic-like memory (see also Murray).

R4.2. The two-process model of recognition in humanneuropsychology. Two of the commentators provide di-rect support for this model. Hintzman points out how re-sponse time measures accord with this model and goes onto suggest that the longer response times for recall-typetasks might be linked to extra processing. The latter pro-posal is consistent with measurements of neuronal responselatencies in recognition tasks in monkeys (Xiang & Brown1998). Yonelinas et al. provide further support and go onto address a particular anomaly in the clinical literature. Itwas pointed out that amnesics with extensive temporal lobedamage may show apparent preservation of familiarity(Verfaellie & Treadwell 1993), a finding that conflicts withour proposals. By reexamining this issue using signal de-tection models, evidence has emerged that amnesics with

nonfocal temporal lobe damage exhibit deficits in both rec-ollection and familiarity (Yonelinas et al.). Although Pick-ering has adopted a similar signal detection approach hehas failed to find evidence of a division within recognitionin normal subjects. It is unclear whether this reflects aweakness in his procedure or a more general weakness inthe approach, and there is a need for more studies thatadopt this method of analysis.

Other critical comments have come from Knowlton.She argues that “remember” (R) and “know” (K) responsescannot reflect independent processes because of the ap-parent conversion over time of R to K responses. The ex-tent of this conversion indicates that every R item could alsohave a K response and so these are redundant rather thanindependent processes. In fact, Knowlton and Squire (1995)reported 29% of items going from R to K, but only 10% go-ing from K to R during a retention interval of between 10min and 1 week. The difficulty with this approach lies in themethod of distinguishing R from K, because when usingself-report it is almost impossible to give an R response thatdoes not also assume a K response, that is, the probabilityof a “know” response is mathematically constrained by theproportion of “remember” responses (Yonelinas et al.).

A related issue is our claim that familiarity cannot simplybe regarded as equivalent to repetition priming. This is po-tentially important, as priming is often intact in amnesicseven though recognition is severely impaired. This rela-tionship was considered by Hintzman and Pickering, whoboth saw the merits of regarding these processes as distinct(see also Wagner et al. 1998). Indeed, it might be specu-lated that there are even more component processes torecognition (Nadel et al., Brown & Xiang 1998). In con-trast, Mayes et al. argued that priming does mediate fa-miliarity. This assertion seemed to be based largely on thegrounds of parsimony and although we applaud this guid-ing principle we feel that the present evidence does notsupport this direct overlap. Foster felt that we were unsureas to whether “familiarity” was a form of implicit memory.This is largely a matter of definition, but given that this in-formation can be used consciously to direct a recognitionjudgement and to make a “remember”/“know” judgement,we would have to argue that it is not implicit.

R4.3. The effects of hippocampal damage on recognition.The clear prediction from our model is that selective hip-pocampal damage will spare those recognition tasks thatcan be performed effectively using familiarity information.In the case of animals the degree of this sparing will dependon the extent to which it is possible for animals to use otherforms of information (see R4.1 and Mumby). For rats, thesituation appears reasonably consistent, with many studiesfailing to find delay-dependent deficits, or indeed anydeficits, after hippocampectomy (Kesner, Mumby). Al-though a study by Mumby (cited by Zola & Squire) didfind evidence of an impairment after a retention interval of120 sec, the use of a more difficult condition with a similardelay in the same study failed to reveal any evidence of animpairment (Mumby et al. 1995). Furthermore, in his owncommentary Mumby reports a double dissociation in whichrats with hippocampal lesions performed normally on a de-layed nonmatching-to-sample task in spite of an impair-ment on a spatial task.

The situation with monkeys is far more complicated be-cause different researchers have used different techniques

Response/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (51)

and approaches, factors that are likely to be critical giventhe sensitivity of the adjacent perirhinal cortex (Gaffan &Lim 1991). Zola & Squire present evidence of a mild butsignificant recognition deficit after selective hippocampec-tomy (Alvarez et al. 1995) that is comparable to that ob-served after ischemia (Zola-Morgan et al. 1992). In supportof their view they note that the performance of these twosets of animals after a 10 min delay is virtually indistin-guishable (means of 68.5% and 68.1%, respectively). Infact, the mean score for the ischemic group published inZola-Morgan et al. (1992, Table 1) is 61% (not 68.1%).These findings can be directly contrasted with those of a re-cent study that examined the effects of neurotoxic lesionsof the hippocampus on delayed nonmatching to sample(Murray & Mishkin 1998). Even though there was exten-sive hippocampal damage there was no evidence of adeficit, even after delays of as long as 40 min. A possiblecomplication with assessing the effects of ischemic damageto the hippocampus is that there may be “cryptic” pathol-ogy, but Nadel et al. sensibly warn about relying on this ex-planation in the absence of other corroborative evidence.

In fact, our model clearly predicts that a recognitiondeficit should emerge following hippocampectomy if an an-imal uses episodic recall to aid recognition. This fact mightlead to phylogenetic differences because different animalsmight naturally use different strategies to solve the task.Furthermore, when an animal is confronted by two scenescomposed of many common items it may be supposed thatfamiliarity alone becomes less effective and deficits may ap-pear (e.g., Gaffan 1992a; 1992b). Unfortunately, we lack anindependent means of establishing when an animal is usingepisodic recall. Although evidence of episodic memory inbirds (Clayton & Dickinson 1998) must strengthen the like-lihood of episodic memory in monkeys, it is premature toassume that tasks such as DNMS (delayed nonmatching-to-sample) ever tax this aspect of memory.

Zola & Squire suggest that pretraining may aid post-surgical performance following hippocampectomy and thismay account for the lack of deficit in some studies (e.g.,Murray & Mishkin 1998). Although this may be a factor, itmerely indicates that there are nonspecific factors that candisrupt performance and that pretraining helps amelioratethese effects. The key issue is whether hippocampec-tomy in monkeys can reproduce the consistent recognitiondeficits observed in amnesia. Amnesics typically show a de-lay-dependent deficit that is not an artifact of baseline dif-ferences (Holdstock et al. 1995). This deficit is apparentafter intervals as short as 40 sec. There is no evidence thathippocampectomy has comparable effects on delayed non-matching (or matching) to sample in monkeys.

The proposed model predicts that humans with selectivehippocampal damage will show relative sparing of recogni-tion based on familiarity judgements. In a study of peoplewith fornix damage following colloid cysts of the third ven-tricle (Aggleton, unpublished findings) evidence for thispattern of deficits was found. Although recognition was notnormal, it was disproportionately spared (Fig. R1). It is in-teresting that when subjects were given the Warringtonrecognition memory test (Warrington 1984) twice so that onthe second test (immediately after the first) both stimuliwere familiar, a greater deficit was apparent in those sub-jects with fornix damage. As a result, the overall scores ofthe groups overlapped on the first test but clearly differedon the second test, leading to a significant interaction.

Mayes et al. also described the recognition performanceof a person with selective, but subtotal, hippocampal dam-age. The predicted sparing of recognition set against im-paired recall was observed. Mayes et al. nevertheless arguethat the pattern of results is not in strict accord with themodel. This is because the person performed poorly onwhat are described as associative tests of recognition.Clearly, we are in danger of circularity if we simply arguethat these exceptions must therefore be episodic in nature,yet the fact that such tasks appear difficult to solve by familiarity alone means that this must be a realistic possi-bility.

R4.4. The perirhinal cortex and human recognition. Onthe basis of lesion, electrophysiological, and brain activationdata from animals we have assumed that the perirhinal cor-tex is integral to making familiarity judgements of objects.We have also assumed that the human perirhinal cortex hasthe same function. This point was considered in detail byGraham & Hodges who carefully spelled out the currentlimitations of the clinical evidence, but also provided evi-dence that is consistent with this view. Mayes et al., how-ever, have described a patient with perirhinal damage whodoes not fit the proposed model as recognition is relativelyintact. This is a finding of great potential importance andwe eagerly await a fuller report as to the extent and locationof the pathology. Kapur also cites an example of rhinaldamage associated with retrograde amnesia rather thanrecognition loss (Yoneda et al. 1994). As Graham & Hodgescorrectly point out, there are uncertainties in identifyingthe location and extent of perirhinal cortex in the humanbrain, and until this is resolved the interpretation of suchclinical cases will remain difficult.

R5. Although the hippocampus and perirhinalcortex are anatomically linked they are notnecessarily dependent on each other for theirrespective roles in the encoding of episodicinformation and familiarity-based recognition

In some respects this might seem the most contentious pro-posal, as anatomical evidence highlights the extent of thepotential interactions between the perirhinal cortex, theparahippocampal cortex, and the hippocampus (Burwell& Eichenbaum; Knowlton, Witter & Van der Werf). Atthe outset we should make it clear that we did not proposethat these regions are typically independent of one another,but that there is a continuum of situations where they in-teract to different degrees. Normally, when information isbeing learned about complex arrays of individual events thetwo regions will closely cooperate. At the same time, theremay be situations where they can be dissociated and evendoubly dissociated. Such double dissociations need not re-flect the normal situation, but they highlight the weaknessof theories of medial temporal lobe memory systems thatassume an automatic interdependency. These views aresupported by Graham & Hodges who cite evidence fromstudies of the progression of semantic dementia. It was notour intention to “de-emphasize” the cortical inputs to thehippocampus (Burwell & Eichenbaum), but to show thatthere are other routes to learning, and that the separate re-gions can function in an effective, independent manner.This appears to include context-free, long-term semantic

Response/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (52)

learning that may occur independent of the hippocampus(Vargha-Khadem et al. 1997).

In the rat brain we have found evidence for such doubledissociations between object recognition and spatial mem-ory. Mumby supports this view with reports of similar dou-ble dissociations. In contrast, whereas Bilkey argues thatperirhinal lesions do lead to consistent spatial memorydeficits in rats, Burwell & Eichenbaum point out a pos-sible explanation for these dissociations as the perirhinal le-sions spare the postrhinal cortex, so providing an intactroute for spatial information to reach the hippocampus. Wehave examined these issues at some length and, to our sur-prise, have consistently found that neurotoxic lesions thatessentially eliminate the postrhinal cortex, the perirhinalcortex, and area TE in rats have no effect on T-maze alter-nation, radial arm maze performance, or Morris water mazeperformance (Aggleton et al. 1997; Bussey et al. 1999). Notonly do they have no effect on standard spatial workingmemory or reference memory tasks, but we have also foundnormal radial arm maze performance when a delay of 30min was interposed after the first 4 choices and the arms ofthe maze were rotated to nullify intramaze cues. Indeed,the same lesions have resulted in significantly enhancedperformance on some spatial tasks (Bussey et al. 1999).These striking results appear inconsistent with the views ofBurwell & Eichenbaum, although it should be noted thatwe have not included the medial entorhinal cortex in thesesurgeries.

We remain uncertain as to why Bilkey has consistentlyobserved spatial deficits following perirhinal lesions in rats.It is suggested that rats in our studies may have developednon-mnemonic strategies, but this seems impossible inview of the radial arm maze results with arm rotation andlengthy delays (Bussey et al. 1999). It should also be re-membered that the magnitude of the perirhinal deficits ob-served by Bilkey are very modest when compared to thoseobserved after hippocampectomy. Moreover, the inconsis-tencies between these results and those of other groups(Aggleton et al. 1997; Ennaceur & Aggleton 1997) are mostlikely to reflect procedural differences such as lesionmethod and extent and type of pretraining. These factorscannot hide the qualitatively different effects of hippocam-pectomy from joint removal of the perirhinal and postrhi-nal cortices in rats.

We agree with all of those commentators who noted thatthe perirhinal (and parahippocampal) cortices have otherprobable functions (Burwell & Eichenbaum, Dalrym-ple-Alford et al., Eacott, Parker, Zola & Squire). In de-scribing the perirhinal cortex we deliberately chose to em-phasize its role in stimulus recognition. A full account of theperirhinal cortex was never intended, and object identifica-tion is undoubtedly part of its functions. Instead, our mainobjective was to consider the hippocampal-diencephalicaxis and how this could be linked to both normal episodicmemory and disturbances of episodic memory. Eacott does,however, make the important point that the recognitiondeficit associated with perirhinal damage is markedly re-duced when a smaller set size is used and then argues thatthis region may therefore be specialised for object identifi-cation and not familiarity. There is, however, behaviouralevidence that animals solve the DNMS task with small setsize in ways that differ from those used with trial uniquestimuli (Mishkin & Delacour 1975). Furthermore, with fre-quently repeated stimuli the task is most easily solved using

rapidly decaying recency information, and such informa-tion is available in neurons in area TE of the temporal cor-tex (Baylis et al. 1987; Fahy et al. 1993a; 1993b). The find-ing of an impairment at even the shortest delays when usinga large set size (Eacott et al. 1994) again indicates a per-ceptual component to the deficit that is consistent with aperirhinal involvement in stimulus identification, as well asstimulus familiarity (Brown & Xiang 1998). A further sig-nificance of the findings concerning set size is that theyserve to distinguish the possible contributions of this corti-cal region from those of the fornix, where familiarity withthe stimuli may exacerbate rather than spare the recogni-tion deficit (Owen & Butler 1984). This contrasting patternof results can be accommodated in the present proposal ifit is assumed that fornix damage spares familiarity, and thatstimulus repetition in primates lessens the value of this cue.In contrast, perirhinal lesioned animals cannot use famil-iarity and so stimulus repetition will not heighten task dif-ficulty, as it would in normal animals.

R6. There are two parallel temporal-thalamicpathways, one from the hippocampus to the anterior thalamic nuclei, the other from perirhinal cortex to the medial dorsalthalamic nucleus. The second of thesepathways underlines familiarity-basedrecognition

Evidence concerning the first of these pathways has alreadybeen discussed (see sect. R3) but the nature of the involve-ment of the medial dorsal thalamic nucleus (MD) in recog-nition remains more a matter of debate. Clinical studieshave proved to be of little help, as circ*mscribed damageto this region is extraordinarily rare (Markowitsch). Ex-perimental studies have repeatedly shown, however, thatdamage to the medial dorsal nucleus can impair recognitionin monkeys (Aggleton & Mishkin 1983b; Parker et al. 1997;Zola-Morgan & Squire 1985a), but the underlying reasonsfor this deficit are more difficult to determine. The keyquestion is whether the involvement of MD depends on in-puts from the perirhinal (and entorhinal) cortices orwhether the recognition deficit is qualitatively different(Eacott, Parker, Mair et al.), in particular, whether theMD deficit may be more closely linked with frontal lobedysfunction rather than perirhinal cortex dysfunction.

Various commentators noted that the current evidencefor a crucial perirhinal-medial dorsal circuit is not com-pelling. Although reciprocal pathways do exist in primates(Aggleton et al. 1986a; Markowitsch et al. 1985; Russchenet al. 1987), these are only light (Witter & Van der Werf).Furthermore, the severity of the recognition deficit aftermedial dorsal damage is markedly less than that observedafter medial temporal damage (Parker), a fact first notedby Aggleton and Mishkin (1983b). For this reason we pos-tulated that the projections from the perirhinal cortex to theprefrontal cortex were also involved in familiarity recogni-tion (see Fig. 2, target article), even though they need notbe essential (Gaffan & Eacott 1995). Other relevant evi-dence has come from recent comparisons between therecognition deficits seen after medial thalamic lesions andperirhinal lesions (Parker et al. 1997). It was found that MDlesions, like perirhinal lesions, lead to a deficit for large butnot small set sizes. This is clearly consistent with the postu-

Response/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (53)

lated link between the two regions. It should also be notedthat it is hard to argue that this thalamic set size effect is theresult of a perceptual deficit, and so shows a weakness withthe assumption that the same effect following perirhinal le-sions must be perceptual (Eacott).

Another approach is to disconnect the medial dorsal thal-amic nucleus by combining a unilateral lesion in MD witha unilateral lesion in the perirhinal cortex in the contra-lateral hemisphere. A recent study using this approachshowed that this surgery impaired DMS (Parker & Gaffan1998a), again implicating the perirhinal-MD link. An alter-native means of disconnecting the perirhinal cortex fromMD is to cut the efferent pathway from the perirhinal andentorhinal cortices as it passes through the external capsulein the rostral temporal stem (Goulet et al. 1998). A consid-eration of relevant experiments reveals an uncertain pic-ture. In both studies (Bachevalier et al. 1985a; Parker &Gaffan 1998a) a dorsal amygdala cut, that is presumed tohave disrupted inputs from perirhinal cortex to MD, hadonly mild effects on tests of recognition. This has been in-terpreted as indicating that the key connections are thosewith the prefrontal cortex (Parker), even though the directconnections may not be critical (Gaffan & Eacott 1995).One problem is that null results with disconnection studiesare typically very difficult to interpret unless it can bedemonstrated that the key connections have been fully sev-ered. In the case of crossed lesion experiments (Parker &Gaffan 1998a) there is an additional assumption that thefunctional pathways are essentially ipsilateral, but it is likelythat there are light crossed projections and these might sup-port performance.

The important point was made that it might be more ac-curate to regard the contribution of MD as more a conse-quence of its connections to the prefrontal cortex than tothe temporal lobe (Parker). This is a view that we havegrowing sympathy for, and it has been reinforced by stud-ies that highlight the similarity between the effects of MDlesions and prefrontal lesions in animals (Bachevalier &Mishkin 1986; Hunt & Aggleton 1998). The difficulty withthis view comes from the clinical literature, which showsthat the recognition deficit associated with diencephalicamnesias is typically as severe as that observed in temporallobe amnesias (Aggleton & Shaw 1996), whereas the effectsof prefrontal cortex damage on recognition are typicallymild. One suggestion is that the importance of the pre-frontal cortex has been masked by clinical cases that em-phasize the effects of partial cortical damage and that theconsequences of extensive, bilateral prefrontal cortical lossremain poorly understood (Parker). We are aware of theseproblems, and it is the case that interactions between theperirhinal cortex and prefrontal cortex and between MDand prefrontal cortex figure prominently in our putativesystem underling familiarity recognition (Fig. 2, target arti-cle). There is clearly growing evidence that these interac-tions should be emphasised even more, and this may proveto be partly at the expense of the connections betweenperirhinal cortex and MD.

Finally, Witter & Van der Werf make the case that it isthe intralaminar and midline nuclei, rather than MD, thatare linked with the perirhinal cortex in signalling familiar-ity. They present preliminary data from a patient who showsdeficits on face recognition but is able to preserve visual as-sociation learning following a thalamic infarct. We awaitmore details on this case, in particular, information con-

cerning the location of the infarct and its possible effect onfibres of passage. It will also be important to ensure that thepreserved learning was not a result of verbal mediation.

R7. The traditional distinction between temporallobe and diencephalic amnesias is misleadingas both have damage to the same functionalsystem

This change of emphasis was welcomed by a number ofcommentators (Kapur, Pickering) and it is seen as an in-evitable consequence of our better understanding of thecontributions of the fornix to memory. The issue of how am-nesias may be of different severity was noted in some com-mentaries (Dalrymple-Alford et al., Kapur, Zola &Squire), and the growing evidence concerning the possiblecontributions of the prefrontal cortex (Parker, Knowlton)and the intralaminar thalamic nuclei (Mair et al.) helps toemphasize how difficult it may be to interpret the effects ofpathologies in multiple sites. The potential impact of evensubtotal pathology in the anterior thalamic nuclei was noted(Dalrymple-Alford et al.) and this may serve as a usefulprompt to reexamine the status of this nuclear group in avariety of conditions that disrupt memory (e.g., various de-mentias and Korsakoff ’s syndrome). Recent imaging stud-ies have helped to show how pathologies in one region canlead to extensive abnormalities in other, distal sites (Mar-kowitsch et al. 1997; Paller et al. 1997) and this is likely toprove a valuable means of addressing this highly complexissue. Finally, Pickering and Kapur both commented on theneed for an agreed definition of amnesia, and whether im-paired recognition is a necessary feature of this syndrome.Clearly, we do not think this is a necessary feature, but weagree that there needs to be greater awareness of the prob-lems of definition.

R8. The hippocampal-diencephalic system is required for the encoding of episodicinformation permitting it to be set in itsspatial and temporal context

A number of commentators have offered alternative ac-counts of the contributions of the hippocampus to mne-monic functions (Burwell & Eichenbaum, Gabriel &Smith, Mayes et al., Murray, Nadel et al., Parker,Whishaw). We should, at the outset, reiterate the fact thatthe target article was focussed on the hippocampal-rostralthalamic pathways and was not intended as a comprehen-sive review of hippocampal function. Two of these accountsemphasized the importance of this region for rememberingmovements in space (Parker, Whishaw), and this view canreadily be used to help explain the head direction informa-tion that passes from the anterior thalamic group to the hip-pocampus (Taube 1995). It also accords with other studiesindicating a role in the memory for instrumental responses(Gaffan et al. 1984; Rupniak & Gaffan 1987). A number ofstudies into the effects of fornix lesions in animals do reveallimitations with this as a complete account, however. Thusalthough deficits are found for conditional and discrimina-tion tasks where the key manipulations are the spatial dis-position of the stimuli (Gaffan 1991; Gaffan & Harrison1989), the animal’s movements provide little information.Furthermore, the recent report that hippocampectomy

Response/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (54)

does not disrupt the ability of monkeys to perform accuratepositional responses that differ from trial to trial (Murray &Mishkin 1998) appears incompatible with the suggestion(Parker) that deficits on the object-in-place discriminationreflect a failure of the hippocampal-thalamic axis to provideideothetic (self-motion) information that can help distin-guish the different stimuli. If on the other hand Parker issuggesting that the hippocampus is required for remem-bering both the movements and the particular situations,then the links between scene memory, ideothetic informa-tion, and episodic memory become considerably stronger.

Other commentators have argued that a role in episodicmemory is too limiting a description of hippocampal-ante-rior thalamic function (Burwell & Eichenbaum; Gabriel& Smith), and does not account for the full array of deficitsobserved after hippocampal damage in rats. It has to beremembered that we still lack convincing evidence of epi-sodic memory in rats, and so for this species the descriptionmust be in terms of functions that can ultimately form a ba-sis for episodic memory. Thus the creation of “scenes” bothaccounts for the deficits at the level of this species and per-mits phylogenetic development into something we callepisodic memory. In a similar vein, Murray argued that thehippocampus is important for associations with memoryrepresentations of stimuli not present in the “here andnow.”

Whishaw argues that the principal function of the hip-pocampus is to permit accurate navigation in space bymeans of path integration. Critically, it is argued that hip-pocampectomized rats still have place memory, and so thelink from place memory in rats to scene memory in pri-mates is seriously weakened. One of the problems with thisproposal is that rats have a number of different ways of solv-ing a spatial task such as finding a hidden platform in awater maze. The most obvious method is to identify the po-sition with respect to a configuration of landmarks (a “cog-nitive map” approach), but another is to learn its positionwith respect to a single landmark. There is compelling evi-dence that both strategies exist, and there is recent evi-dence that this second strategy is independent of the hip-pocampus (Pearce et al. 1998). If the training methodsadopted by Whishaw encourage this landmark strategy,then the pattern of performance he describes does not con-flict with our initial proposals.

R9. The prefrontal cortex interacts with both the hippocampal-anterior thalamic and theperirhinal-medial dorsal thalamic systems at a variety of levels, thereby engagingefficient strategies for recall

A number of commentators wished that interactions withprefrontal cortex had been more detailed (Foster, Knowl-ton, Parker). To provide a comprehensive review of thecontributions of the prefrontal cortex to learning wouldhave been beyond the goals of the target article. Neverthe-less, there is recent evidence (see sect. R6) to suppose thatthe prefrontal cortex has an important role in recognition aswell as recall. There is also considerable evidence fromfunctional imaging studies implicating different regions ofthe prefrontal cortex in various aspects of memory. Indeed,it seems highly likely that the contributions of this regionhave been systematically underestimated as extensive, but

circ*mscribed pathologies are rarely if ever described inhumans. Furthermore, animal studies, and especially thoseusing rats, will inevitably fail to reveal the full potential im-portance of the prefrontal cortex given its phylogenetic de-velopment. Although summary Figures 1 and 2 (target ar-ticle) both included prefrontal cortex their interactionsremain to be examined and understood in much greater de-tail. This represents one of several areas for research thatwe hope might be stimulated by our target article.

R10. Concluding remarks

A very deliberate aspect of our review was the use of datafrom a variety of methodologies and species. There are, ofcourse, well-recognised problems in moving across species,and there may have been occasions when we failed to pointout the implications of this practice (Kapur). We feel thatit is important, however, to show how these data interlink,as certain methodologies are much more appropriate forsome species than others. An example of this is the use ofimmediate early gene activation studies in rats. Bures &Fenton commented on this approach and correctly notedthat to complete the link between increased c-fos activationand recognition it is important to have simultaneous be-havioural indices of novelty detection. We are currently ad-dressing this issue. Bures & Fenton also argued that thepaired viewing procedure might be flawed because thecombined view from the two eyes might produce a novelconfiguration, so making all conditions “novel.” This is aningenious notion, but the fact that we find very similar re-sults when the same studies are run in a between-subjectsdesign (Zhu et al. 1995b) argues against this view. Further-more, the c-fos results are highly consistent with the find-ings from single unit recording studies in both rats andmonkeys (Zhu et al. 1995a).

Several commentators (Kesner, Knowlton, Mumby)suggested that we had misinterpreted the apparent differ-ences between the perirhinal cortex and the hippocampusbecause the dissociations between these regions reflect dif-ferences in the nature of the afferent information and notthe form of processing that occurs within these structures.Thus it is because the hippocampus receives a considerableamount of spatial information, in contrast to the perirhinalcortex, that these two structures differ functionally. This isan appealing notion, but it suffers a number of seriousdrawbacks. To take the specific example provided above, itdoes not account for the failure of combined perirhinal andpostrhinal lesions to affect a variety of spatial tasks in rats(Aggleton et al. 1997; Bussey et al. 1999), even though thecombined cortical regions undoubtedly receive much spa-tial information. At a more general level, this proposal fliesin the face of the established notion that different cyto-architectures reflect different modes of processing. A con-sideration of the intrinsic connections of the ventral tem-poral cortex reveals a system of information convergence(e.g., in columns) that can be contrasted with the diver-gence of information that characterises the hippocampus(Brown 1990). Thus it seems more appropriate to regardthese structures as having qualitatively different processingcapabilities.

Finally, Nadel et al. make a number of imaginative spec-ulations concerning how our proposals might be applied torecall processes and retrograde amnesia. These include the

Response/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (55)

notion that there might be two routes by which the hip-pocampus can reactivate cortical representations, and theconsequences of dysfunctions in just one of these routes.We deliberately did not consider retrograde amnesia in ourreview, principally because there is sufficient evidence thatanterograde amnesia can (occasionally) occur in the ab-sence of retrograde amnesia. Nevertheless, any attempt toexplain the nature of new learning will have implicationsboth for the fate of old memories and for the subsequentreactivation of new memories. This is just one example ofhow our initial proposals may lead to implications that wehad not anticipated, and we look forward to other aspectsof our model being explored and tested.


Letters “a” and “r” appearing before authors’ initials refer to target articleand response, respectively.

Aggleton, J. P. (1986) A description of the amygdala-hippocampal interconnectionsin the macaque monkey. Experimental Brain Research 64:515–26. [rJPA]

Aggleton, J. P., Desimone, R. & Mishkin, M. (1986a) The origin, course, andtermination of the hippocampo-thalamic projections in the macaque. Journalof Comparative Neurology 243:409–21. [arJPA, MPW]

Aggleton, J. P., Hunt, P. R., Nagle, S. & Neave, N. (1996) The effects of selectivelesions within the anterior thalamic nuclei on spatial memory in the rat.Behavioural Brain Research 81:189–98. [aJPA, JCD-A]

Aggleton, J. P., Hunt, P. R. & Rawlins, J. N. P. (1986b) The effects of hippocampallesions upon spatial and non-spatial tests of working memory. BehaviouralBrain Research 19:133–46. [aJPA]

Aggleton, J. P., Hunt, P. R. & Shaw, C. (1990) The effects of mammillary body andcombined amygdalar-fornix lesions on tests of delayed non-matching- to-sample in the rat. Behavioural Brain Research 40:145–57. [aJPA]

Aggleton, J. P., Keen, S., Warburton, E. C. & Bussey, T. J. (1997) Extensivecytotoxic lesions of the rhinal cortices impair recognition but spare spatialalternation in the rat. Brain Research Bulletin 43:279–87. [arJPA, DKB]

Aggleton, J. P., Keith, A. B., Rawlins, J. N. P., Hunt, P. R. & Sahgal, A. (1992)Removal of the hippocampus and transection of the fornix producecomparable deficits on delayed non-matching to position by rats. BehaviouralBrain Research 52:61–71. [aJPA]

Aggleton, J. P. & Mishkin, M. (1983a) Visual recognition impairment followingmedial thalamic lesions in monkeys. Neuropsychologia 21:189–97. [aJPA, AP]

(1983b) Memory impairments following restricted medial thalamic lesions inmonkeys. Experimental Brain Research 52:199–209. [arJPA, AP]

(1985) Mamillary-body lesions and visual recognition in monkeys. ExperimentalBrain Research 58:190–97. [aJPA]

Aggleton, J. P., Neave, N., Nagle, S. & Hunt, P. R. (1995a) A comparison of theeffects of anterior thalamic, mamillary body and fornix lesions on reinforcedspatial alternation. Behavioural Brain Research 68:91–101. [aJPA]

Aggleton, J. P., Neave, N., Nagle, S. & Sahgal, A. (1995b) A comparison of theeffects of medial prefrontal, cingulate cortex, and cingulum bundle lesions ontests of spatial memory: Evidence of a double dissociation between frontal andcingulum bundle contributions. Jounal of Neuroscience 15:7270–81. [aJPA]

Aggleton, J. P., Nicol, R. M., Huston, A. E. & Fairbairn, A. F. (1988) Theperformance of amnesic subjects on tests of experimental amnesia in animals:Delayed matching-to-sample and concurrent learning. Neuropsychologia26:265–72. [aJPA]

Aggleton, J. P. & Sahgal, A. (1993) The contribution of the anterior thalamic nucleito anterograde amnesia. Neuropsychologia 31:1001–19. [aJPA, JCD-A]

Aggleton, J. P. & Saunders, R.C. (1997) The relationships between temporal lobeand diencephalic structures implicated in anterograde amnesia. Memory5:49–71. [arJPA]

Aggleton, J. P. & Shaw, C. (1996) Amnesia and recognition memory: A reanalysis ofpsychometric data. Neuropsychologia 34:51–62. [arJPA, ADP]

Alvarez, P., Zola-Morgan, S. & Squire, L. R. (1995) Damage limited to thehippocampal region produces long-lasting memory impairment in monkeys.Journal of Neuroscience 15:3796–807. [arJPA, DJM, SMZ]

Amarel, D. G. (1987) Memory: Anatomical organization of candidate brain regions.In: Handbook of physiology: The nervous system, ed. V. B. Mountcastle, F.Plum & S. R. Geiger. American Psychological Society. [EEK]

Armstrong, E. (1986) Enlarged limbic structures in the human brain: The anteriorthalamus and medial mamillary body. Brain Research 362:394–97. [JCD-A]

Armstrong, E., Clarke, M. R. & Hill, E. M. (1987) Relative size of the anteriorthalamic nuclei differentiates anthropoids by social system. Brain Behaviorand Evolution 30:263–71. [JCD-A]

Assal, G., Probst, A., Zander, E. & Rabionwicz, T. (1976) Syndrome amnesique parinfiltration tumorale. Archiv Suisse Neurology Neurochirurgica Psychiatry119:317–24. [aJPA]

Astur, R. S., Mumby, D. G. & Sutherland, R. J. (1995) Perirhinal cortex damage:Effects of acquisition and retention of object and place discriminations in rats.Society for Neuroscience Abstracts 21:1935. [MJE]

Atkinson, R. C. & Juola, J. F. (1974) Search and decision processes in recognitionmemory. In: Contemporary developments in mathematical psychology: I.Learning, memory and thinking, ed. R. C. Atkinson, R. D. Luce, D. H. Krantz& P. Suppes. Freeman. [APY]

Bachevalier, J. & Meunier, M. (1996) Cerebral ischemia: Are the memory deficitsassociated with hippocampal cell loss? Hippocampus 6:553–60. [aJPA,SMZ]

Bachevalier, J., & Mishkin, M. (1986) Visual recognition impairment followsventromedial but not dorsolateral prefrontal lesions in monkeys. BehaviouralBrain Research 20:249–61. [arJPA]

(1989) Mnemonic and neuropathological effects of occluding the posteriorcerebral artery in Macaca mulatta. Neuropsychologia 27:83–105. [aJPA]

Bachevalier, J., Parkinson, J. K. & Mishkin, M. (1985a) Visual recognition inmonkeys. Effects of separate vs. combined transection of fornix andamygdalofugal pathways. Experimental Brain Research 57:554–61. [arJPA,AP].

Bachevalier, J., Saunders, R. C. & Mishkin, M. (1985b) Visual recognition inmonkeys. Effects of transection of fornix. Experimental Brain Research57:547–53. [aJPA]

Barlow, J. S. (1964) Inertial navigation as a basis of animal navigation. Journal ofTheoretical Biology 6:76–117. [IQW]

Barnes, C. A. (1988) Spatial learning and memory processes: The search for theirneurobiological mechanisms in the rat. Trends in Neuroscience 11:163–69.[aJPA]

Baxendale, S. A. (1997) The role of the hippocampus in recognition memory.Neuropsychologia 35:591–98. [aJPA]

Baylis, G. C., Rolls, E. T. & Leonard, C. M. (1987) Functional subdivisions of thetemporal lobe neocortex. The Journal of Neuroscience 7:330–42. [rJPA]

Beason-Held, L. L., Rosene, D. L., Killiany, R. J. & Moss, M. B. (in press) Ibotenicacid lesions of the hippocampal formation produce memory deficits in therhesus monkey. Hippocampus. [SMZ]

Beason-Held, L., Rosene, D. L. & Moss, M. (1993) Memory deficits associatedwith ibotenic acid lesions of the hippocampal formation in rhesus monkeys.Society for Neuroscience Abstracts 19:438. [aJPA]

Benedek, L. & Juma, A. (1941) Korsakowsyndrom bei den Geschwulsten desZwischenhirns. Archive Psychiatry 114:366–76. [aJPA]

Bentivoglio, M., Aggleton, J. P. & Mishkin, M. (1997) The thalamus and memoryformation. In: Thalamus, ed. M. Steriade, E. G. James & D. A. McCormick.Elsevier. [rJPA]

Beracochea, D. & Jaffard, R. (1991) Effects of chronic ethanol consumptionassociated or not with experimental anterior thalamic lesions on spontaneoussequential alternation in mice. Neuroscience Letters 134:45–48. [JCD-A]

(1995) Effects of anterior thalamic lesions on spatial memory in mice.Neuroreport 5:917–20. [aJPA, JCD-A]

Beracochea, D. J., Jaffard, R. & Jarrard, L .E. (1989) Effects of anterior ordorsomedial thalamic ibotenic lesions on learning and memory in rats.Behavioral and Neural Biology 51:364–76. [aJPA, JCD-A]

Bloomfield, S. A. & Sherman, S. M. (1989) Dendritic current flow in relay cellsand interneurons of the cat’s lateral geniculate nucleus. Proceedings of theNational Academy of Sciences USA 86(10):3911–14. [EEK]

Böttger, S., Prosiegel, M., Steiger, H. J. & Yassouridis, A. (1998) Neurobehaviouraldisturbances, rehabilitation outcome, and lesion site in patients after ruptureand repair of anterior communicating artery ancurysm. Journal of Neurology,Neurosurgery and Psychiatry 65:93–102. [HJM]

Brion, S. & Mikol, J. (1978) Atteinte du noyou lateral dorsal du thalamus etsyndrome de Korsakoff alcoolique. Journal of the Neurological Sciences38:249–61. [aJPA]

Brito, G. N. O., Thomas, G. J., Davis, B. J. & Gingold, S. I. (1982) Prelimbiccortex, mediodorsal thalamus, septum, and delayed alternation in rats.Experimental Brain Research 46:52–58. [aJPA]

Brown, M. W. (1982) Effect of context on the responses of single units recordedfrom the hippocampal region of behaviourally trained monkeys. In: Neuronalplasticity and memory formation, IBRO Monograph Series Vol. 9, ed. C.Ajmone-Marsan & H. Matthies. Raven Press. [aJPA]

(1990) Why does the cortex have a hippocampus? In: Learning andcomputational neuroscience: Foundations of adaptive networks, ed. M.Gabriel & J. Moore. MIT Press. [arJPA]

(1996) Neuronal responses and recognition memory. Seminars in theNeurosciences 8:23–32. [aJPA]

References/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (56)

Brown, M. W., Wilson, F. A. W. & Riches, P. (1987) Neuronal evidence thatinferomedial temporal cortex is more important than hippocampus in certainprocesses underlying recognition memory. Brain Research 409:158–62.[aJPA, MJE]

Brown, M. W. & Xiang, J.-Z. (1998) Recognition memory: Neuronal substrates ofthe judgement prior occurrence. Progress in Neurobiology 55:149–89.[arJPA]

Bruin, J. P .C. de, Sanchez-Santed, F., Heinsbroek, R. P. W., Donker, A. &Postmes, P. (1994) A behavioural analysis of rats with damage to the medialprefrontal cortex using the Morris water maze: Evidence for behaviouralflexibility, but not for impaired spatial navigation. Brain Research 652:322–33.[aJPA]

Buckley, M. J. & Gaffan, D. (1997) Impairment of visual object discriminationlearning after perirhinal cortex ablation. Behavioral Neuroscience 111:467–75.[MJE]

(1998a) Perirhinal cortex ablation impairs visual object identification. Journal ofNeuroscience 18:2268–75. [MJE]

(1998b) Perirhinal cortex ablation impairs configural learning and paired-associate learning equally. Neuropsychologia 36:535–46. [MJE]

(1998c) Learning and transfer of object-reward associations and the role of theperirhinal cortex. Behavioral Neuroscience 112:15–23. [MJE]

Bunsey, M. & Eichenbaum, H. (1993) Critical role of the parahippocampal regionfor paired associates learning in rats. Behavioral Neuroscience 107:740–47.[MJE]

(1996) Conservation of hippocampal memory function in rats and humans.Nature 379:255–57. [RDB, SMZ]

Burk, J. A. & Mair, R. G. (1998a) Thalamic amnesia reconsidered: Excitotoxiclesions of the intralaminar nuclei, but not the mediodorsal nucleus disruptplace DMTS performance in the rat (Rattus norvegicus). BehavioralNeuroscience 112:54–67. [RGM]

(1998b) Delayed matching-to-sample trained with retractable levers is impairedby lesions of the intralaminar or ventromedial but not the laterodorsalthalamic nuclei (submitted). [RGM]

Burwell, R. D. & Amaral, D. G. (1998a) Cortical afferents of the perirhinal,postrhinal, and entorhinal cortices. Journal of Comparative Neurology398(2):179–205. [RDB]

(1998b)The perirhinal and postrhinal cortices of the rat: Interconnectivity andconnections with the entorhinal cortex. Journal of Comparative Neurology391(3):293–321. [RDB]

Burwell, R. D., Caballero, M., Witter, M. P. & Amaral, D. G. (1994) A PHA-Lstudy of the perirhinal projection to the thalamus in the rat. Society forNeuroscience Abstracts 20:1415. [MPW]

Burwell, R. D., Witter, M. P. & Amaral, D. G. (1995) Perirhinal and postrhinalcortices of the rat: A review of the neuroanatomical literature and comparisonwith findings from the monkey brain. Hippocampus 5:390–408. [aJPA]

Bussey, T. J., Muir, J. & Aggleton, J. (1999) Functionally dissociating aspects ofevent memory: The effects of combined perirhinal and postrhinal cortexlesions on object and place memory in the rat. The Journal of Neuroscience19:495–503. [rJPA]

Butters, N., Salmon, D. P., Munro Cullum, C., Cairns, P., Tröster, A. I., Jacobs, D.,Moss, M. & Cermak, L. S. (1988) Differentiation of amnesic and dementedpatients with the Wechsler Memory Scale–Revised. Clinical Neuropsychology2:133–48. [aJPA]

Byatt, G. & Dalrymple-Alford, J. C. (1996) Both anteromedial and anteroventralthalamic lesions impair radial-maze learning in rats. Behavioral Neuroscience110:1335–48. [aJPA, JCD-A]

Cahill, L., Babinsky, R., Markowitsch, H. J. & McGaugh, J. L. (1995) Involvementof the amygdaloid complex in emotional memory. Nature 377:295–96.[HJM]

Cahill, L. & McGaugh, J. L. (1996) Modulation of memory storage. CurrentOpinion in Neurobiology 6:237–42. [rJPA]

Cai, H. B. & Lo, F. S. (1996) Passive cable properties of relay cells in dorsal lateralgeniculate nucleus of the rabbit. Journal Sheng Li Hsueh Pao 48(6):577–80.[EEK]

Cairns, H. & Mosberg, W. H. (1951) Colloid cyst of the third ventricle. SurgeryGynecology, Obstetrics 92:546–70. [aJPA]

Calabrese, P., Haupts, M., Markowitsch, H. J. & Gehlen, W. (1993) The cognitive-amnesic performance profile of a patient with bilateral asymmetrical thalamicinfarction. International Journal of Neuroscience 71:101–106. [aJPA]

Calabrese, P., Markowitsch, H .J., Harders, A. G., Scholz, M. & Gehlen, W. (1995)Fornix damage and memory. A case report. Cortex 31:555–64. [arJPA]

Cameron, A. S. & Archibald, Y. M. (1981) Verbal memory deficit after left fornixremoval: A case report. International Journal of Neuroscience 12:201. [aJPA]

Carmel, P. W. (1985) Tumours of the third ventricle. Acta Neurochirurgica75:136–46. [aJPA]

Cassaday, H. J. & Rawlins, J. N. P. (1995) Fornix-fimbria section and workingmemory deficits in rats: Stimulus complexity and stimulus size. BehavioralNeuroscience 109:594–606. [aJPA]

(1997) The hippocampus, objects and their contexts. Behavioral Neuroscience111:1228–44. [aJPA, JKE]

Castaigne, P., Lhermitte, F., Buge, A., Escourolle, R., Hauw, J. J. & Lyon-Caen, O.(1981) Paramedian thalamic and midbrain infarcts: Clinical andneuropathological study. Annals of Neurology 10:127–48. [aJPA]

Cave, C. B. & Squire, L. R. (1992) Intact verbal and nonverbal short-term memoryfollowing damage to the human hippocampus. Hippocampus 2:151–64.[aJPA]

Cermak, L. S., Talbot, N., Chandler, K. & Reale Wolbarst, L. (1985) Theperceptual priming phenomenon in amnesia. Neuropsychologia 23:615–22.[aJPA]

Chen, H.-C. & Burwell, R. D. (1996) An anterograde tract-tracing study of theperirhinal and postrhinal cortical projections to the thalamus in the rat brain.Journal of Undergraduate Research 3:47–68. [MPW]

Clark, R. E., Teng, E., Squire, L. R. & Zola, S. (1996) The visual paired-comparison task and the medial temporal lobe memory system. Society forNeuroscience Abstract 22:116.10. [aJPA, SMZ]

(1997) Perirhinal damage impairs memory on the visual paired-comparison task.Society for Neuroscience Abstract 23:12.8. [aJPA]

Clarke, S., Assal, G., Bogousslavsky, J., Regli, F., Townsend, D. W., Leenders, K. L.& Blecic, S. (1994) Journal of Neurology, Neurosurgery and Psychiatry57:27–34. [aJPA]

Clayton, N. S. & Dickinson, A. (1998) Episodic-like memory during cache recoveryby scrub jays. Nature 395:272–74. [rJPA]

Cohen, N. J. & Eichenbaum, H. (1993) Memory, amnesia and the hippocampalsystem. MIT Press. [DJM]

Colmant, H. J. (1965) Enzephalopien bei chronischem Alkoholismus, insbesondereThalamusbefunde bei Wernickescher Enzephalopathie. F. Enke. [aJPA]

Colombo, M. & Gross, C. G. (1996) Hippocampus, delay neurons, and sensoryheterogeneity. Behavioral and Brain Sciences 19:766–67. [JCD-A]

Corkin, S. (1984) Lasting consequences of bilateral medial temporal lobectomy:Clinical course and experimental findings in HM. Seminars in Neurology4:249–59. [NK]

Corkin, S., Amaral, D. G., González, R. G., Johnson, K. A. & Hyman, B. T. (1997)H. M.’s medial temporal lobe lesion: Findings from magnetic resonanceimaging. Journal of Neuroscience 17:3964–79. [rJPA, KSG]

Cramon, D. Y., von Hebel, N. & Schuri, U. (1985) A contribution to the anatomicalbasis of thalamic amnesia. Brain 108:993–1008. [aJPA]

Crick, F. & Jones, E. (1993) The backwardness of human neuroanatomy. Nature361:109–10. [NK]

Cruce, J. A. F. (1975) An autoradiographic study of the projections of themammillothalamic tract in the rat. Brain Research 85:211–19. [aJPA]

Cummings, J. L., Tomiyasu, S., Read, S. & Benson, D. F. (1984) Amnesia withhippocampal lesions after cardiopulmonary arrest. Neurology 34:679–81.[aJPA]

Curran, T. & Hintzman, D. L. (1995) Violations of the independence assumptionin process dissociation. Journal of Experimental Psychology: Learning,Memory, and Cognition 21:531–47. [ADP]

Daum, I. & Ackermann, H. (1994) Frontal-type memory impairment associatedwith thalamic damage. International Journal of Neuroscience 77:187–98.[aJPA]

Day, L. B. & Schallert, T. (1996) Anticholinergic effects on acquisition of placelearning in the Morris water task: Spatial mapping deficit or inability in inhibitnonplace strategies? Behavioral Neuroscience 110:998–1005. [IQW]

Dehn, D. M. & Engelkamp, J. (1997) Process dissociation procedure: Doubledissociations following divided attention and speeded responding. QuarterlyJournal of Experimental Psychology 50A:318–36. [aJPA]

DeJong, R. N., Itabashi, H. H. & Olson, J. R. (1969) Memory loss due tohippocampal lesion. Archives of Neurology 20:339–48. [aJPA]

Delay, J. & Brion, S. (1969) Le syndrome de Korsakoff. Masson. [aJPA, JKF, AP]Delbecq-Derousne, J., Beauvois, M. & Shallice, T. (1990) Preserved recall versus

impaired recognition. Brain 113:1054–74. [NK]DeLuca, J. & Diamond, B. J. (1995) Aneurysm of the anterior communicating

artery: A review of neuroanatomical and neuropsychological sequelae. Journalof Clinical and Experimental Neuropsychology 17:100–121. [HJM]

Diamond, B. J., DeLuca, J. & Kelly, S. M. (1997) Memory and executive functionsin amnesic and non-amnesic patients with aneurysms of the anteriorcommunicating artery. Brain 120:1015–25. [HJM]

Diesfeldt, H. F. A. (1992) Impaired and preserved semantic memory functions indementia. In: Memory functioning in dementia, ed. L. Backman. Elsevier.[KSG]

Dolan, R. J. & Fletcher, P. C. (1997) Dissociating prefrontal and hippocampalfunction in episodic memory encoding. Nature 388:582–85. [aJPA]

Donaldson, W. (1996) The role of decision processes in remembering andknowing. Memory and Cognition 24:523–33. [aJPA]

Dum, R. P. & Strick, P. L. (1993) Cingulate motor areas. In: Neurobiology ofcingulate cortex and limbic thalamus, ed. B. A. Vogt & M. Gabriel. Birkhauser.[EEK]

References/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (57)

Dusek, J. A. & Eichenbaum, H. (1997) The hippocampus and memory for orderlystimulus relations. Proceedings of the National Academy of Sciences USA94:7109–14. [RDB]

Dusoir, H., Kapur, N., Byrnes, D. P., McKinstry, S. & Hoare, R. D. (1990) The roleof diencephalic pathology in human memory disorder. Brain 113:1695–706.[aJPA]

Duva, C. A., Floresco, S. B., Wunderlich, G. P., Lao, T. L., Pinel, J. P. J. & Phillips,A. G. (1997) Disruption of spatial but not object-recognition memory byneurotoxic lesions of the dorsal hippocampus in rats. Behavioral Neuroscience111:1184–96. [aJPA]

Düzel, E., Yonelinas, A. P., Mangun, G. R., Heinze, H. & Tulving, E. (1997) Event-related brain potential correlates of two states of conscious awareness inmemory. Proceedings of the National Academy of Sciences, USA 94:5973–78.[APY]

Eacott, M. J. (1998) Acquisition and retention of visual discrimination learningafter ablation of perirhinal cortex in the rat. Psychobiology 26:36–41. [MJE]

Eacott, M. J., Gaffan, D. & Murray, E. A. (1994) Preserved recognition memoryfor small sets, and impaired stimulus identification for large sets, followingrhinal cortex ablation in monkeys. European Journal of Neuroscience 6:1466–78. [rJPA, MJE]

Eacott, M. J. & Heywood, C. A. (1995) Perception and memory: Action andinteraction. Critical Reviews in Neurobiology 9:311–20. [MJE]

Eichenbaum, H. (1997) Declarative memory: Insights from cognitive neurobiology.Annual Review of Psychology 48:547–72. [SMZ]

Eichenbaum, H., Otto, T. & Cohen, N. J. (1994) Two functional components of thehippocampal memory system. Behavioral and Brain Sciences 17:449–518.[aJPA, RDB, JCD-A, DLH, MPW, SMZ]

Eichenbaum, H., Stewart, C. & Morris, R. G. M. (1990) Hippocampalrepresentation in spatial learning. Journal of Neuroscience 10:331–39.[RDB]

Eif*cku, S., Nishijo, H., Kita, T. & Ono, T. (1995) Neuronal activity in the primatehippocampal formation during a conditional association task based on thesubject’s location. Journal of Neuroscience 15:4952–69. [aJPA, AT]

Ennaceur, A. & Aggleton, J. P. (1994) Spontaneous recognition of objectconfigurations in rats: Effects of fornix lesions. Experimental Brain Research100:85–92. [aJPA]

(1997) The effects of neurotoxic lesions of the perirhinal cortex combined tofornix transection on object recognition memory in the rat. Behavioural BrainResearch 88:181–93. [arJPA, DKB]

Ennaceur, A., Neave, N. & Aggleton, J. P. (1996) Neurotoxic lesions of theperirhinal cortex do not mimic the behavioural effects of fornix transection inthe rat. Behavioural Brain Research 80:9 –25. [aJPA, DKB]

(1997) Spontaneous object recogntion and object location memory in rats: Theeffects of lesions in the cingulate cortices, the medial prefrontal cortex, thecingulum bundle and the fornix. Experimental Brain Research 113:509–519.[aJPA]

Eskandar, E. N., Richmond, B. J. & Optican, L. M. (1992) Role of inferiortemporal neurons in visual memory: I. Temporal encoding of informationabout visual images, recalled images, and behavioral context. Journal ofNeurophysiology 68:1277–95. [aJPA]

Fahy, F. L., Riches, I. P. & Brown, M. W. (1993a) Neuronal signals of importanceto the performance of visual recognition memory tasks: Evidence fromrecordings of single neurones in the medial thalamus of primates. Progress inBrain Research 95:401–16. [arJPA]

(1993b) Neuronal activity related to visual recognition memory: Long-termmemory and the encoding of recency and familiarity information in theprimate anterior and medial inferior temporal and rhinal cortex. ExperimentalBrain Research 96:457–72. [arJPA, MJE]

Fazio, F., Perani, D., Gilardi, M. C., Colombo, F., Cappa, S. F., Vallar, G.,Bettinardi, V., Paulesu, E., Alberoni, M., Bressi, S., Franceshci, M. & Lenzi,G. L. (1992) Metabolic impairment in human amnesia: A PET study ofmemory networks. Journal of Cerebral Blood Flow and Metabolism 12:353–58. [aJPA]

Foster, J. K., ed. (in press) Special edition of Memory on the theme of“Neuroimaging and memory.” To be published in 1999. This will also bepublished in book form by Psychology Press (Taylor & Francis). [JKF]

Foster, J. K., Ainsworth, J., Faratin, P. & Shapiro, J. (1997) Investigating themnemonic functions of the hippocampus: A computational neuroscienceperspective. In: Cognitive models of memory, ed. M. Conway & S. Gathercole.Psychology Press (Taylor & Francis). [JKF]

Foster, J. K., Black, S. E., Buck, B. H. & Bronskill, M. J. (1997) Executive andassociated functions in normal and pathological aging: A neuroimagingperspective. In: Methodology of frontal and executive function, ed. P. Rabbitt.Psychology Press (Taylor & Francis). [JKF]

Foster, J. K., Lidder, P. & Suenram, S. I. (1998) Glucose facilitation of memoryperformance in normal young participants: Fractionation of enhancementeffects? Psychopharmacology 137:259–70. [JKF]

Foster, J. K. & Rawlins, J. N. P. (1992a) The role of the hippocampus in the

performance of a continuous non-matching task. Psychobiology 20:189–97.[JKF]

(1992b) The effects of a hippocampal aspiration lesion upon the performance ofa delayed alternation Go/No-Go task. Behavioural Brain Research 47:35–48.[JKF]

Foster, J. K., Rossell, S. L. & Meudell, P. R. (in preparation) Temporal judgement:Judging what has been seen is preserved in schizophrenia, but not when.[JKF]

Freed, D. M. & Corkin, S. (1988) Rate of forgetting in H.M.: 6-month recognition.Behavioral Neuroscience 102:823–27. [aJPA]

Freed, D. M., Corkin, S. & Cohen, N. J. (1987) Forgetting in H.M. – A secondlook. Neuropsychologia 25:461–72. [aJPA]

Freeman, J. H., Jr., Cuppernell, C., Flannery, K. & Gabriel, M. (1996) Context-specific multi-site cingulate cortical, limbic thalamic and hippocampalneuronal activity during concurrent discriminative approach and avoidancetraining in rabbits. Journal of Neuroscience 16:1538–49. [MG]

Freeman, J. H., Jr., Weible, A., Rossi, J. & Gabriel, M. (1997) Entorhinal cortexlesions disrupt processing and alter amygdalar and cingulate cortical neuronalactivity during extinction of discriminative avoidance learning. ExperimentalBrain Research 115:445–57. [MG]

Friedman, H. R., Janas, J. D. & Goldman-Rakic, P. S. (1990) Enhancement ofmetabolic activity in the diencephalon of monkeys performing workingmemory tasks: A 2-deoxyglucose study in behaving rhesus monkeys. Journal ofCognitive Neuroscience 2:18–31. [aJPA]

Fulvi-Mari, C. & Treves, A. (1998) Modeling neocortical areas with a modularneural network. Biosystems 48:47–55. [AT]

Fuster, J. M. (1989) The prefrontal cortex: Anatomy, physiology andneuropsychology of the frontal lobe. Raven Press. [JKF]

Fuster, J. M. & Jervey, J. P. (1981) Inferotemporal neurons distinguish and retainbehaviorally relevant features of visual stimuli. Science 212:952–55. [aJPA]

Gabriel, M. (1993) Discriminative avoidance learning: A model system. In:Neurobiology of cingulate cortex and limbic thalamus: A comprehensivehandbook, ed. B. A. Vogt & M. Gabriel. Birkhauser. [aJPA, JCD-A, MG]

Gabriel, M., Sparenborg, S. & Stolar, N. (1987) Hippocampal control of cingulatecortical and anterior thalamic information processing during learning inrabbits. Experimental Brain Research 67:131–52. [MG]

Gabriel, M., Vogt, B. A., Kubota, Y., Poremba, A. & Kang, E. (1991) Training-stage related neuronal plasticity in limbic thalamus and cingulate cortexduring learning: A possible key to mnemonic retrieval. Behavioural BrainResearch 46:175–85. [MG]

Gabrieli, J., Brewer, J. B., Desmond, J. E. & Glover, G. H. (1997) Separate neuralbases of two fundamental memory processes in the human medial temporallobe. Science 276:264–66. [aJPA]

Gaffan, D. (1974) Recognition impaired and association intact in the memory ofmonkeys after transection of the fornix. Journal of Comparative andPhysiological Psychology 86:1100–1109. [aJPA]

(1991) Spatial organization of episodic memory. Hippocampus 1:262–64. [arJPA](1992a) The role of the hippocampus-fornix-mammillary system in episodic

memory. In: Neuropsychology of memory (2nd edition), ed. L. R. Squire & N.Butters. Guilford Press. [arJPA, MJE]

(1992b) Amnesia for complex naturalistic scenes and for objects following fornixtransection in the rhesus monkey. European Journal of Neuroscience 4:381–88. [arJPA]

(1994a) Dissociated effects of perirhinal cortex ablation, fornix transection andamygdalectomy: Evidence for multiple memory systems in the primatetemporal lobe. Experimental Brain Research 99:411–22. [aJPA, DKB, MJE,SMZ]

(1994b) Scene-specific memory for objects: A model of episodic memoryimpairment in monkeys with fornix transection. Journal of CognitiveNeuroscience 6:305–20. [aJPA, AP]

(1998) Ideothetic input into object-place configuration as the contribution tomemory of the monkey and human hippocampus. Experimental BrainResearch 123:201–209. [AP]

Gaffan, D. & Eacott, M. J. (1995) Uncinate fascicle section leaves delayedmatching-to-sample intact, with both large and small sets. Experimental BrainResearch 105:175–80. [arJPA]

Gaffan, D. & Gaffan, E. A. (1991) Amnesia in man following transection of thefornix. Brain 114:2611–18. [arJPA]

Gaffan, D. & Harrison, S. (1989) Place memory and scene memory: Effects offornix transection in the monkey. Experimental Brain Research 74:202–12.[arJPA]

Gaffan, D. & Lim, C. (1991) Hippocampus and the blood supply to TE:Parahippocampal pial section impairs visual discrimination learning inmonkeys. Experimental Brain Research 87:227–31. [arJPA]

Gaffan, D. & Murray, E. A. (1992) Monkeys with rhinal cortex lesions succeed inobject discrimination learning despite 24-hour intertrial intervals and fail atmatch to sample despite double sample presentations. BehavioralNeuroscience 106:30–38. [MJE]

References/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (58)

Gaffan, D. & Parker, A. (1996) Interaction of perirhinal cortex with the fornix-fimbria: Memory for objects and “object-in-place” memory. Journal ofNeuroscience 16:5864–69. [aJPA]

Gaffan, D., Parker, A. & Easton, A. (1998) Dense amnesia in macaques aftersection of anterior temporal stem, amygdala and fornix. Society forNeuroscience Abstracts 24:18. [AP]

Gaffan, D., Shields, C. & Harrison, S. (1984) Delayed matching by fornix-transected monkeys: The sample, the push and the bait. Quarterly Journal ofExperimental Psychology 36B:305–17. [arJPA]

Gaffan, E. A., Gaffan, D. & Hodges, J. R. (1991) Amnesia following damage to theleft fornix and to other sites. Brain 114:1297–313. [aJPA]

Garcia-Bengochea, F. & Friedman, W. A. (1987) Persistent memory loss followingsection of the anterior fornix in humans. Surgical Neurology 27:361–64.[arJPA]

Gardiner, J. M. (1988) Functional aspects of recollective experience. Memory andCognition 16:309–13. [aJPA]

Gardiner, J. M. & Java, R. I. (1990) Recollective experience in word and nonwordrecognition. Memory and Cognition 19:617–23. [aJPA]

Gardiner J. M. & Parkin, A. J. (1990) Attention and recollective experience inrecognition memory. Memory and Cognition 18:579–83. [aJPA]

Geffen, G., Walsh, A., Simpson, D. & Jeeves, M. (1980) Comparison of the effectsof transcortical and transcallosal removal of intraventricular tumours. Brain103:773–88. [aJPA]

Gentilini, M., De Renzi, E., Crisi, G. (1987) Bilateral paramedian thalamic arteryinfarcts: Report of eight cases. Journal of Neurology, Neurosurgery andPsychiatry 50:900–909. [aJPA]

Glanzer, M. & Bowles, N. (1976) Analysis of the word frequency effect inrecognition memory. Journal of Experimental Psychology: Human Learningand Memory 2:21–31. [LN]

Glees, P. & Griffiths, H. B. (1952) Bilateral destruction of the hippocampus(Cornu ammonis) in a case of dementia. Psychiatric Neurological and MedicalPsychology 123:193–204. [aJPA]

Glenn, M. J. & Mumby, D. G. (1996) Place- and object-recognition deficitsfollowing lesions of the hippocampus or perirhinal cortex in rats: A doubledissociation. Society for Neuroscience Abstracts 22:1120. [DGM]

(1998) Place-memory is intact in rats with perirhinal cortex lesions. BehavioralNeuroscience 112:1353–65. [DGM]

Godden, D. R. & Baddeley, A. D. (1975) Context-dependent memory in twonatural environments: On land and underwater. British Journal of Psychology66:325–31. [aJPA]

Goldenberg, G., Wimmer, A. & Maly, J. (1983) Amnesic syndrome with a unilateralthalamic lesion: A case report. Journal of Neurology 229:79–86. [aJPA]

Gothard, K. M., Skaggs, W. E. & McNaughton, B. L. (1996) Dynamics ofmismatch correction in the hippocampal ensemble code for space: Interactionbetween path integration and environmental cues. Journal of Neuroscience16:8027–640. [aJPA, AT]

Goulet, S., Douré, F. Y. & Murray, E. A. (1998) Aspiration lesions of the amygdaladisrupt the rhinal corticothalamic projection system in rhesus monkeys.Experimental Brain Research 119:131–40. [rJPA, AP]

Graf, P. & Ryan, L. (1990) Transfer appropriate processing for implicit and explicitmemory. Journal of Experimental Psychology: Learning, Memory, andCognition 16:978–92. [LN]

Graf, P. & Schacter, D. L. (1985) Implicit and explicit memory for new associationsin normal and amnesic subjects. Journal of Experimental Psychology:Learning, Memory and Cognition 11:501–18. [DGM]

Graff-Radford, N. R., Tranel, D., van Hoesen, G. W. & Brandt, J. P. (1990)Diencephalic amnesia. Brain 113:1–25. [aJPA, MPW]

Graham, K. S., Becker, J. T. & Hodges, J. R. (1997) On the relationship betweenknowledge and memory for pictures: Evidence from the study of patients withsemantic dementia and Alzheimer’s disease. Journal of the InternationalNeuropsychological Society 3:534–44. [KSG]

Graham, K. S. & Hodges, J. R. (1997) Differentiating the roles of the hippocampalcomplex and the neocortex in long term memory storage: Evidence from thestudy of semantic dementia and Alzheimer’s disease. Neuropsychology 11:77–89. [aJPA, KSG, BJK]

Graham, K. S., Pratt, K. H. & Hodges, J. R. (in press) A reverse temporalgradient for public events in a single case of semantic dementia. Neurocase.[KSG]

Graham, K. S., Simons, J. S., Pratt, K. H., Patterson, K. & Hodges, J. R.(submitted) Insights from semantic dementia on the relationship betweenepisodic and semantic memory. Neuropsychologia. [KSG]

Granon, S., Vidal, C., Thinus-Blanc, C., Changeus, J-P. & Poucet, B. (1994)Working memory, response selection, and effortful processing in rats withmedial prefrontal lesions. Behavioral Neuroscience 108:883–91. [aJPA]

Gray, J. A. (1982) The neuropsychology of anxiety: An enquiry into the functions ofthe septohippocampal system. Oxford University Press. [JKF, ADP]

Greene, E. & Naranjo, J. N. (1986) Thalamic role in spatial memory. BehavioralBrain Research 19:123–31. [aJPA, JCD-A]

Gregg, V. H. & Gardiner, J. M. (1994) Recognition memory and awareness: A largeeffect of study-test modalities on “Know” responses following a highlyperceptual orienting task. European Journal of Cognitive Psychology 6:131–47. [BJK]

Groenewegen, H. J. (1988) Organization of the afferent connections of themediodorsal thalamic nucleus in the rat, related to the mediodorsal-prefrontal topography. Neuroscience 24:379–431. [RGM]

Gronlund, S. D., Edwards, M. B. & Ohrt, D. D. (1997) Comparison of the retrievalof item versus spatial position information. Journal of ExperimentalPsychology: Learning, Memory, and Cognition 23:1261–74. [DLH]

Gudden, H. (1896) Klinische und anatommische Beitrage zur Kenntnis dermultiplen Alkoholneuritis nebst Bemerzungen uber dieRegenerationsvorgange im peripheren Nervensystem. Archiv Psychiatrie28:643–741. [aJPA]

Habib, M. & Sirigu, A. (1987) Pure topographical disorientation: A definition andanatomical basis. Cortex 23:73–85. [aJPA]

Haist, F., Shimamura, A. P. & Squire, L. R. (1992) On the relationship betweenrecall and recognition memory. Journal of Experimental Psychology 18:691–702. [aJPA]

Hamann, S. B. & Squire, L. R. (1997) Intact perceptual memory in the absence ofconscious memory. Behavioral Neuroscience 111(4):850–54. [aJPA]

Hanley, J. R., Davies, A. D. M., Downes, J. J. & Mayes, A. R. (1994) Impairedrecall of verbal material following rupture and repair of an anteriorcommunicating artery aneurysm. Cognitive Neuropsychology 11:543–78.[aJPA]

Harasty, J. A., Halliday, G. M., Code, C. & Brooks, W. S. (1996) Quantification ofcortical atrophy in a case of progressive fluent aphasia. Brain 119:181–90.[KSG]

Hart, M., Poremba, A. & Gabriel, M. (1997) The nomadic engram: Overtrainingeliminates the impairment of discriminative avoidance behavior produced bylimbic thalamic lesions. Behavioural Brain Research 82:169–77. [MG]

Hasselmo, M. E. & Wyble, B. P. (1997) Free recall and recognition in a networkmodel of the hippocampus: Simulating effects of scopolamine on humanmemory functions. Behavioural Brain Research 89:1–34. [ARM]

Hassler, F. (1962) New aspects of brain functions revealed in brain diseases. In:Frontiers in brain research, ed. J. D. French. Columbia University Press.[aJPA]

Heilman, K. M., Bowers, D., Watson, R. T., Day, A., Valenstein, E., Hammond, E.& Duara, R. (1990) Frontal hypermetabolism and thalamic hypometabolismin a patient with abnormal orienting and retrosplenial amnesia.Neuropsychologia 28:161–69. [aJPA]

Heilman, K. M. & Sypert, G. W. (1977) Korsakoff ’s syndrome resulting frombilateral fornix lesions. Neurology 27:490–93. [aJPA]

Herdegen, T. (1996) Jun, Fos, and CREB/ATF transcription factors in the brain:Control of gene expression under normal and pathophysiological conditions.Neuroscientist 2:153–61. [aJPA]

Higuchi, S. I. & Miyash*ta, Y. (1996) Formation of mnemonic neuronal responsesto visual paired associates in inferotemporal cortex is impaired by perirhinaland entorhinal lesions. Proceedings of the National Academy of Sciences, USA93:739–43. [aJPA]

Hintzmann, D. L. & Caulton, D. A. (1997) Recognition memory and modalityjudgements: A comparison of retrieval dynamics. Journal of Memory andLanguage 37:1–23. [DLH]

Hintzmann, D. L., Caulton, D. A. & Levitin, D. J. (1998) Retrieval dynamics inrecognition and list discrimination: Further evidence of separate processes offamiliarity and recall. Memory and Cognition 26:449–62. [DLH]

Hintzmann, D. L. & Curran, T. (1994) Retrieval dynamics of recognition andfrequency judgments: Evidence for separate processes of familiarity andrecall. Journal of Memory and Language 33:1–18. [DLH]

(1997) Comparing retrieval dynamics in recognition memory and lexicaldecision. Journal of Experimental Psychology: General 126:228–47. [DLH]

Hirshman, E. & Master, S. (1997) Modeling the conscious correlates of recognitionmemory: Reflections on the remember-know paradigm. Memory andCognition 25:345–51. [aJPA]

Hirst, W., Johnson, M. K., Kim, J. K., Phelps, E. A., Risse, G. & Volpe, B. T. (1986)Recognition and recall in amnesics. Journal of Experimental Psychology12:445–51. [aJPA]

Hirst, W., Johnson, M. K., Phelps, E. A. & Volpe, B. T. (1988) More on recognitionand recall in amnesics. Journal of Experimental Psychology 14:758–62.[aJPA]

Hodges, J. R. & Carpenter, K. (1991) Anterograde amnesia with fornix damagefollowing removal of IIIrd ventricle colloid cyst. Journal of Neurology,Neurosurgery and Psychiatry 54:633–38. [aJPA]

Hodges, J. R. & Graham, K. S. (1998) A reversal of the temporal gradient forfamous person knowledge in semantic dementia: Implications for the neuralorganisation of long-term memory. Neuropsychologia 36:803–25. [KSG]

Hodges, J. R. & McCarthy, R. A. (1993) Autobographical amnesia resulting frombilateral paramedian thalamic infarction. Brain 116:921–40. [aJPA]

References/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (59)

Hodges, J. R. & Patterson, K. (1996) Nonfluent progressive aphasia and semanticdementia: A comparative neuropsychological study. Journal of theInternational Neuropsychological Society 2:511–24. [KSG]

Hodges, J. R., Patterson, K., Oxbury, S. & Funnell, E. (1992) Semantic dementia:Progressive fluent aphasia with temporal lobe atrophy. Brain 115:1783–806.[aJPA, KSG]

Holdstock, J. S., Shaw, C. & Aggleton, J. P. (1995) The performance of amnesicsubjects on tests of delayed matching-to-sample and delayed matching-to-position. Neuropsychologia 33:1583–96. [arJPA]

Holmes, E. J. & Butters, M. S. (1983) An examination of the effects ofmammillary-body lesions on reversal learning sets in monkeys. PhysiologicalPsychology 3:159–65. [aJPA]

Honey, R. C., Watt, A. & Good, M. (1998) Hippocampal lesions disrupt anassociative-mismatch process. Journal of Neuroscience 18:2226–30. [aJPA,LN, ADP]

Horel, J. A., Pytko-Joiner, D.Voytko, M. L. & Salsbury, K. (1987) The performanceof visual tasks while segments of the inferotemporal cortex are suppressed bycold. Behavioural Brain Research 23:29–42. [MJE]

Horel, J. A., Voytko, M. L. & Salsbury, K. G. (1984) Visual learning suppressed bycooling the temporal pole. Behavioral Neuroscience 98:310–24. [MJE]

Hull, C. L. (1943) Principles of behavior. Appleton-Century-Crofts. [EEK]Hunt, P. R. & Aggleton, J. P. (1991) Medial dorsal thalamic lesions and working

memory in the rat. Behavioural and Neural Biology 55:227–46. [aJPA, JCD-A]

(1998) Neurotoxic lesions of the dorsomedial thalamus impair the acquisitionbut not the performance of delayed matching to place by rats: A deficit inshifting response rules. Journal of Neuroscience 97:129–41. [rJPA]

Huppert, F. A. & Piercy, M. (1978) The role of trace strength in recency andfrequency judgments by amnesic and control subjects. Quarterly Journal ofExperimental Psychology 30:347–54. [aJPA]

(1979) Normal and abnormal forgetting in organic amnesia: Effect of locus oflesion. Cortex 15:385–90. [aJPA]

Insausti, R., Amaral, D. G. & Cowan, W. M. (1987) The entorhinal cortex of themonkey: II. Cortical afferents. Journal of Comparative Neurology 264:356–95. [aJPA]

Jackson-Smith, P., Kesner, R. P. & Chiba, A. A. (1993) Continuous recognition ofspatial and nonspatial stimuli in hippocampal-lesioned rats. Behavioral andNeural Biology 59:107–19. [RPK]

Jacoby, L. L. (1983) Remembering the data: Analyzing interactive process inreading. Journal of Verbal Learning and Verbal Behavior 22:485–508. [LN]

(1991) A process dissociation framework: Separating automatic from intentionaluses of memory. Journal of Memory and Language 30:513–41. [aJPA, ADP,APY]

Jacoby, L. L. & Dallas, M. (1981) On the relationship between autobiographicalmemory and perceptual learning. Journal of Experimental Psychology:General 3:306–40. [aJPA]

Jarrard, L. E. (1995) What does the hippocampus really do? Behavioural BrainResearch 71:1–10. [DKB]

Jones, B. & Mishkin, M. (1972) Limbic lesions and the problem of stimulus-reinforcement associations. Experimental Neurology 36:362–77. [aJPA]

Jones, E. G. (1985) The thalamus. Plenum Press. [RGM]Jones, G. V. (1987) Independence and exclusivity among psychological processes:

Implications for the structure of recall. Psychological Review 94:229–35.[aJPA]

Jones, G. V. & Gardiner, J. M. (1990) Recognition failure when recognition targetsand recall cues are identical. Bulletin of the Psychonomic Society 28:105–108.[aJPA]

Kang, E. & Gabriel, M. (1998) Hippocampal modulation of cingulo-thalamicneuronal activity and discriminative avoidance learning rabbits. Hippocampus8:491–510. [MG]

Kapur, N., Crewes, H., Wise, R., Abbott, P., Carter, M., Millar, J. & Lang, D.(1999) Mammillary body damage results in memory impairment but notamnesia. Neurocase (in press). [NK]

Kapur, N., Ellison, D., Parkin, A. J., Hunkin, N., Burrows, E., Sampson, S. A. &Morrison, E. A. (1994) Bilateral temporal lobe pathology with sparing ofmedial temporal lobe structures: Lesion profile and pattern of memorydisorder. Neuropsychologia 32:1–16. [aJPA, KSG]

Kapur, N., Ellison, D., Smith, M. P., McLellan, D. L. & Burrows, E. H. (1992)Focal retrograde amnesia following bilateral temporal lobe pathology: Aneuropsychological and magnetic resonance study. Brain 115:73–85. [aJPA,KSG]

Kapur, N., Scholey, K., Moore, E., Barker, S., Brice, J., Mayes, A. & Fleming, J.(1995) The mammillary bodies revisited: Their role in human memoryfunctioning. In: Neuropsychological explorations in memory and cognition:Essays in honor of Nelson Butters, ed. L. Cermak. Plenum Press. [aJPA,NK]

Kartsounis, L. D., Rudge, P. & Stevens, J. M. (1995) Bilateral lesions of CA1 andCA2 fields of the hippocampus are sufficient to cause a severe amnesic

syndrome in humans. Journal of Neurology, Neurosurgery and Psychiatry59:95–98. [aJPA]

Kesner, R. P. (1998) Neurobiological views of memory. In: Neurobiology oflearning and memory, ed. J. Martinez & R. Kesner. Academic Press. [RPK]

Kesner, R. P., Ravindranath, A., Jackson-Smith, P., Giles, R. & Chiba, A. A.(submitted) A neural circuit analysis of visual object working memory: Role ofperirhinal, entorhinal and TE2 cortex. [RPK]

Kievit, J. & Kuypers, H. G. J. M. (1977) Organization of the thalamo-corticalconnexions to the frontal lobe in the rhesus monkey. Experimental BrainResearch 29:299–322. [aJPA]

Kim, J. J. & Fanselow, M. S. (1992) Modality-specific retrograde amnesia of fear.Science 256:675–77. [aJPA]

Knight, R.T. (1996) Contribution of human hippocampal region to noveltydetection. Nature 383:256–59. [aJPA]

Knoth, R. L. & Mair, R. G. (1991) Response latency and accuracy on a pretrainednonmatch to sample task in rats recovered from pyrithiamine-induced thiamindeficiency. Behavioral Neuroscience 105:375–85. [RGM]

Knowlton, B. J. (1998) The relationship between remembering and knowing: Acognitive neuroscience perspective. Acta Psychologica 98:253–65. [BJK]

Knowlton, B. J. & Squire, L. R. (1995) Remembering and knowing: Two differentexpressions of declarative memory. Journal of Experimental Psychology21(3):699–710. [arJPA, BJK, APY, SMZ]

Koehler, S., Black, S. E., Sinden, M., Kidron, D., Szekely, C., Parker, J., Foster, J.K., Moscovitch, M., Winocur, G., Szalai, J. P. & Bronskill, M. (1998) Memoryimpairments in relation to hippocampal and parahippocampal atrophy inAlzheimer’s disease: An MR volumetry study. Neuropsychologia 36:101–14.[JKF]

Koger, S. M. & Mair, R. G. (1994) Comparison of the effects of frontal cortical andthalamic lesions on measures of olfactory learning and memory in the rat.Behavioral Neuroscience 108:1088–100. [RGM]

Kolb, B., Pittman, K., Sutherland, R. J. & Whishaw, I. Q. (1982) Dissociation of thecontributions of the prefrontal cortex and dorsomedial thalamic nucleus tospatially guided behavior in the rat. Behavioural Brain Research 6:365–78.[aJPA]

Kopelman, M. D. (1995) The Korsakoff ’s syndrome. British Journal of Psychiatry166:154–73. [JCD-A]

Kosel, K. C., Van Hoesen, G. W. & Rosene, D. (1983) A direct projection from theperirhinal cortex (area 35) to the subiculum in the rat. Brain Research269:347–51. [RDB]

Krieckhaus, E. E. (1964) Decrements in avoidance behavior followingmammillothalamic tractotomy in cats. Journal of Neurophysiology 27:753–67.[EEK]

(1998) Papez circuit inhibits stress and alarm and initiates voluntary action.Behavioral and Brain Sciences (submitted). [EEK]

Krieckhaus, E. E., Donahoe, J. W. & Morgan, M. A. (1992) Paranoid schizophreniamay be caused by dopamine hyperactivity of CA1 hippocampus. Journal ofBiological Psychiatry 31:560–70. [EEK]

Kritchevsky, M., Graff-Radford, N. R. & Damasio, A. R. (1987) Normal memoryafter damage to medial thalamus. Archives of Neurology 44:959–64. [arJPA,NK, MPW]

Kroll, N. E. A., Knight, R. T., Metcalfe, J., Wolf, E. S. & Tulving, E. (1996)Cohesion failure as a source of memory illusions. Journal of Memory andLanguage 35:176–96. [APY]

Langlais, P. J. & Savage, L. M. (1995) Thiamine deficiency in rats producescognitive and memory deficits on spatial tasks that correlate with tissue loss indiencephalon, cortex and white matter. Behavioural Brain Research 68:75–89. [aJPA, JCD-A, RGM]

Langlais, P. J., Zhang, S.-X. & Savage, L. M. (1996) Neuropathology of thiaminedeficiency: An update on the comparative analysis of human disorders andexperimental models. Metabolic Brain Disease 11:19–37. [RGM]

Lauro-Grotto, R., Piccini, C., Borgo, F. & Treves, A. (1997) What remains ofmemories lost in Alzheimer and herpetic encephalitis. Society forNeuroscience Abstracts 23:1889. [AT]

Leng, N. R. C. & Parkin., A. J. (1989) Aetiological variation in the amnesicsyndrome: Comparisons using the Brown-Peterson task. Cortex 25:251–59.[aJPA]

Leonard, B. W., Amaral, D. G., Squire, L. R. & Zola-Morgan, S. (1995) Transientmemory impairment in monkeys with bilateral lesions of the entorhinal cortex.Journal of Neuroscience 15:5637–59. [aJPA]

Levasseur, M., Baron, J. C., Sette, G., LeGault-Demare, F., Pappata, S.,Mauguiere, F., Benoit, N., Tran Dinh, S., Degos, J. D., Laplane, D. &Mazoyer, B. (1992) Brain energy metabolism in bilateral paramedian thalamicinfarcts. Brain 115:795–807. [aJPA]

Li, L., Miller, E. K. & Desimone, R. (1993) The representation of stimulusfamiliarity in anterior inferior temporal cortex. Journal of Neurophysiology69:1918–29. [aJPA, MJE]

Lindeboom, J. (1989) The assessment of anterograde amnesia. Journal of Clinicaland Experimental Neuropsychology 11:345. [MPW]

References/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (60)

Liu, P. (1998) Perirhinal cortex contributions to spatial memory. Ph. D. thesis,University of Otago, Dunedin, New Zealand. [JCD-A]

Liu, P. & Bilkey, D. K. (1998a) Excitotoxic lesions centered on perirhinal cortexproduce delay-dependent deficits in a test of spatial memory. BehavioralNeuroscience 112:512–24. [DKB]

(1998b) Lesions of perirhinal cortex produce spatial memory deficits in theradial maze. Hippocampus 8:114–21. [DKB, JCD-A, DGM]

(1998c) Perirhinal cortex contributions to performance in the Morris watermaze. Behavioral Neuroscience 112:304–15. [DKB, DGM]

Lorente de No, R. (1934) Studies on the structure of the cerebral cortex. II.Continuation of the study of the ammonic system. Journal für Psychologie undNeurologie 46:113–77. [aJPA]

Lueschow, A., Miller, E. K. & Desimone, R. (1994) Inferior temporal mechanismsfor invariant object recognition. Cerebral Cortex 4:523–31. [aJPA]

Maguire, E. A., Frackowiak, R. S. J. & Frith, C. D.(1996) Learning to find yourway: A role for the human hippocampal formation. Proceedings of the RoyalSociety of London B 263:1745–50. [aJPA]

Mahut, H. (1971) Spatial and object reversal learning in monkeys with partialtemporal lobe ablations. Neuropsychologia 9:409–24. [aJPA]

(1972) A selective spatial deficit in monkeys after transection of the fornix.Neuropsychologia 10:65–74. [aJPA]

Mair, R. G. (1994) On the role of thalamic pathology in diencephalic amnesia.Reviews in the Neurosciences 5:105–40. [RGM]

Mair, R. G., Anderson, C. D., Langlais, P. J. & McEntee, W. J. (1988) Behavioralimpairments, brain lesions and monoaminergic activity in the rat followingrecovery from a bout of thiamin deficiency. Behavioural Brain Research27:223–39. [RGM]

Mair, R. G., Burk, J. A. & Porter, M. C. (1998) Lesions of the frontal cortex,hippocampus, and intralaminar thalamic nuclei have distinct effects onremembering. Behavioral Neuroscience 112:772–92. [RGM]

Mair, W. G. P., Warrington, E. K. & Weiskrantz, L. (1979) Memory disorders inKorsakoff”s psychosis. A neuropathological and neuropsychologicalinvestigation of two cases. Brain 102:749–83. [aJPA, RGM]

Malamud, N. & Skillicorn, S. A. (1956) Relationship between the Wernicke andthe Korsakoff syndrome. American Medical Association Archives of Neurologyand Psychiatry 76:585–96. [RGM]

Malamut, B. L., Graff-Radford, N., Chawluk, J., Grossman, R. I. & Gur, R. C.(1992) Memory in a case of bilateral thalamic infarction. Neurology 42:163–69. [aJPA]

Malkova, L. & Mishkin, M. (1997) Memory for the location of objects afterseparate lesions of the hippocampus and parahippocampal cortex in rhesusmonkeys. Society for Neuroscience Abstracts 23:12. [SMZ]

Mandler, G. (1980) Recognizing: The judgment of previous occurrence.Psychological Review 87:252–71. [aJPA, LN, APY]

Mark, V. H., Barry, H., Mclardy, T. & Ervin, F. R. (1970) The destruction of bothanterior thalamic nuclei in a patient with intractable agitated depression.Journal of Nervous and Mental Disorders 150:266–72. [aJPA]

Markowitsch, H. J. (1982) Thalamic mediodorsal nucleus and memory: A criticalevaluation of studies in animals and man. Neuroscience and BiobehavioralReviews 6:351–80. [aJPA, RGM, HJM]

(1988) Diencephalic amnesia: A reorientation towards tracts? Brain ResearchReviews 13:351–70. [aJPA, HJM]

(1992) Diencephalic amnesia. In: Trastornos de la memoria, ed. D. BarciaSalorio. Editorial MCR. [HJM]

(1994) Effects of emotion and arousal on memory processing by the brain. In:Memory, learning and the brain, ed. J. Delacour. World Scientific. [HJM]

(1999) The limbic system. In: The MIT encyclopedia of cognitive science, ed. R.Wilson & F. Keil. MIT Press. [HJM]

Markowitsch, H. J., Calabrese, P., Haupts, M., Durwen, H. F., Liess, J. & Gehlen,W. (1993) Searching for the anatomical basis of retrograde amnesia. Journal ofClinical and Experimental Neuropsychology 15:947–76. [aJPA]

Markowitsch, H. J., Emmans, D., Irle, E., Streicher, M. & Preilowski, B. (1985)Cortical and subcortical afferent connections of the primate’s temporal pole: Astudy of rhesus monkeys, squirrel monkeys, and marmosets. Journal ofComparative Neurology 242:425–58. [arJPA]

Markowitsch, H. J., von Cramon, D. Y. & Schuri, U. (1993) Mnestic performanceprofile of a bilateral diencephalic infarct patient with preserved intelligenceand severe amnesic disturbances. Journal of Clinical and ExperimentalNeuropsychology 15:627–52. [HJM]

Markowitsch, H. J., Weber-Luxemburger, G., Ewald, K., Kessler, J. & Heiss, W.-D. (1997) Patients with heart attacks are not valid models for medial temporallobe amnesia. A neuropsychological and FDG-PET study with consequencesfor memory research. European Journal of Neurology 4:178–84. [aJPA, HJM]

Markowska, A. L., Olton, D. S., Murray, E. A. & Gaffan, D. (1989) A comparativeanalysis of the role of the fornix and cingulate cortex in memory: Rats.Experimental Brain Research 74:187–201. [aJPA]

Martin, K. M. & Aggleton, J. P. (1993) Contextual effects on the ability of divers touse decompression tables. Applied Cognitive Psychology 7:311–16. [aJPA]

Mayes, A. R. & Downes, J. J. (1997) What do theories of the functional deficits(s)underlying amnesia have to explain? Memory 5:3–36. [aJPA, JKF, ADP]Also published as Chapter 1 of Theories of organic amnesia, ed. A. R. Mayes& J. J. Downes. Psychology Press (Taylor & Francis). [JKF]

Mayes, A. R., Meudell, P. R., Mann, D. & Pickering, A. (1988) Location of lesionsin Korsakoff ’s syndrome: Neuropsycholgical and neuropathological data ontwo patients. Cortex 24:367–88. [aJPA]

McClelland, J. L., McNaughton, B. L. & O’Reilly, R. C. (1995) Why there are twocomplementary learning systems in the hippocampus and neocortex: Insightsfrom the successes and failures of connectionist models of learning andmemory. Psychological Review 102:419–57. [LN]

McKee, R. D. & Squire, L. R. (1992) Equivalent forgetting rates in long-termmemory for diencephalic and medial temporal lobe amnesia. Journal ofNeuroscience 12:3765–72. [aJPA]

McMackin, D., co*ckburn, J., Anslow, P. & Gaffan, D. (1995) Correlation of fornixdamage with memory impairment in six cases of colloid cyst removal. ActaNeurochirugica 135:12–18. [arJPA]

McNaughton, B. L., Barnes, C. A., Gerrard, J. L., Gothard, K., Jung, M. W.,Knierin, J. J., Kudrimoiti, Q. Y., Skaggs, W. E., Suster, M. & Weaver, K. L.(1996) Deciphering the hippocampal polyglot: The hippocampus as a pathintegration system. Journal of Experimental Biology 199:173–85. [AP]

Mennemeier, M., Fennell, E., Valenstein, E. & Heilman, K. M. (1992)Contributions of the left intralaminar and medial thalamic nuclei to memory:Comparisons and a case report. Archives of Neurology 49:1050–58. [RGM]

Meunier, M., Bachevalier, J., Mishkin, M. & Murray, E. A. (1993) Effects on visualrecognition of combined and separate ablations of the entorhinal andperirhinal cortex in rhesus monkeys. Journal of Neuroscience 12:5418–32.[aJPA, MJE]

Meunier, M., Hadfield, J., Bachevalier, J. & Murray, E. A. (1996) Effects of rhinalcortex lesions combined with hippocampectomy on visual recognition memoryin rhesus monkeys. Journal of Neurophysiology 75:1190–205. [aJPA]

Miller, E. K. & Desimone, R. (1994) Parallel neuronal mechanisms for short-termmemory. Science 263:520–22. [aJPA]

Miller, E. K., Li, L. & Desimone, R. (1993) Activity of neurons in anterior inferiortemporal cortex during a short-term memory task. Journal of Neuroscience13:1460–78. [aJPA, DLH]

Miller, L. A., Munoz, D. G. & Finmore, M. (1993) Hippocampal sclerosis andhuman memory. Archives of Neurology 50:391–94. [aJPA]

Mirsky, R., Adolphs, R., Volkov, I., Bechara, A., Damasio, H. & Howard, M. A., III.(1997) Single-unit neuronal activity in human amygdala and ventral frontalcortex recorded during emotional experience. Society for NeuroscienceAbstracts 23:1318. [AT]

Mishkin, M. (1978) Memory in monkeys severely impaired by combined, but notby separate removal of amygdala and hippocampus. Nature 273:297–98.[aJPA]

Mishkin, M. & Delacour, J. D. (1975) An analysis of short-term visual memory inthe monkey. Journal of Experimental Psychology: Animal Behavior Processes1:326–34. [rJPA]

Mishkin, M., Malamut, B. & Bachevalier, J. (1984) Memories and habits: Twoneural systems. In: Neurobiology of learning and memory, ed. G. Lynch, J. L.McGaugh & N. M. Weinberger. Guilford Press. [RDB]

Mishkin, M. & Murray, E. M. (1994) Stimulus recognition. Current Opinion inNeurobiology 4:200–206. [aJPA, SMZ]

Miyash*ta, Y., Okuno, H., Tokuyama, W., Ihara, T. & Nakajima, K. (1996) Feedbacksignal from medial temporal lobe mediates visual associative mnemonic codesof inferotemporal neurons. Cognitive Brain Research 5:81–86. [aJPA]

Miyash*ta, Y., Rolls, E. T., Cahusac, P. M. B., Niki, H. & Feigenbaum, J. D. (1989)Activity of hippocampal formation neurons in the monkey related to aconditional spatial response task. Journal of Neurophysiology 61:669–77.[aJPA, AT]

Mizumori, S. J. Y., Miya, D. Y. & Ward, K. E. (1994) Reversible inactivation of thelateral dorsal thalamus disrupts hippocampal place representation and impairsspatial learning. Brain Research 644:168–74. [aJPA, JCD-A]

Mizumori, S. J. Y. & Williams, J. D. (1993) Directionally sensitive mnemonicproperties of neurons in the lateral dorsal nucleus of the thalamus in rats.Journal of Neuroscience 13:4015–28. [aJPA]

Mori, E., Yamadori, A. and Mitani, Y (1986) Left thalamic infarction anddisturbance of verbal memory: A clinicoanatomical study with a new methodof computed tomographic stereotaxic lesion localization. Annals of Neurology20:671–76. [aJPA]

Morris, R. G. M., Garrud, P., Rawlins, J. N. P. & O’Keefe, J. (1982) Placenavigation impaired in rats with hippocampal lesions. Nature 297:681–83.[aJPA, RDB]

Moscovitch, M. & Nadel, L. (1998) Consolidation revisited: In defense of themultiple-trace model. Current Opinion in Neurobiology 8:297–300. [LN]

Moser, E., Moser, M. & Anderson, P. (1993) Spatial learning impairment parallelsthe magnitude of dorsal hippocampal lesions, but is hardly present followingventral lesions. Journal of Neuroscience 13:3916–25. [SMZ]

References/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (61)

Moser, M.-B., Moser, E. I., Forrest, E., Andersen, P. & Morris, R. G. M. (1995)Spatial learning with a minislab in the dorsal hippocampus. Proceedings of theNational Academy of Sciences USA 92:9697–701. [rJPA, SMZ]

Muller, R. U., Ranck, J. B., Jr. & Taube, J. S. (1996) Head direction cells:Properties and functional significance. Current Opinions in Neurobiology6:196–206. [JB]

Mulligan, N. & Hirshman, E. (1995) Speed-accuracy trade-offs and the dualprocess model of recognition memory. Journal of Memory and Language34:1–18. [DLH]

Mumby, D. G. & Pinel, J. P. J. (1994) Rhinal cortex lesions and object recognitionin rats. Behavioral Neuroscience 108:11–18. [aJPA, MJE, DGM]

Mumby, D. G., Pinel, J. P. J. & Dastur, F. N. (1993) Mediodorsal thalamic lesionsand object recognition in rats. Psychobiology 21:27–36. [aJPA]

Mumby, D. G., Pinel, J. P. J., Kornecook, T. J., Shen, M. J. & Redila, V. A. (1995)Memory deficits following lesions of hippocampus or amygdala in rat:Assessment by an object-memory test battery. Psychobiology 23:26–36.[rJPA, SMZ]

Mumby, D. G., Wood, E. R., Duva, C. A., Kornecook, T. J., Pinel, P. J. & Phillips,A. G. (1996) Ischemia-induced object-recognition deficits in rats areattenuated by hippocampal ablation before or soon after ischemia. BehavioralNeuroscience 110:266–81. [aJPA, DGM]

Mumby, D. G., Wood, E. R. & Pinel, J. P. J. (1992) Object recognition memory inrats is only mildly impaired by lesions of the hippocampus and amygdala.Psychobiology 20:18–27. [DGM]

Mummery, C. J., Patterson, K., Wise, R. J. S., Price, C. J. & Hodges, J. R. (in press)Disrupted temporal lobe connections in semantic dementia. Brain. [KSG]

Murray, D. J. (in press) Language and psychology: 19th-century developmentsoutside Germany: A survey. In: History of the language sciences: Aninternational handbook on the evolution of the study of language, ed. S.Auroux, K. Koerner, H.-J. Niederehe & K. Versteegh. De Gruyter. [DJM]

Murray, E. A. (1996) What have ablation studies told us about the neural substratesof stimulus memory? Seminars in the Neurosciences 8:13–22. [aJPA, DBK,MJE]

Murray, E. A., Davidson, M., Gaffan, D., Olton, D. S. & Suomi, S. (1989) Effectsof fornix transection and cingulate coritical ablation on spatial memory inrhesus monkeys. Experimental Brain Research 74:173–86. [aJPA]

Murray, E. A., Gaffan, D. & Mishkin, M. (1993) Neural substrates of visualstimulus-stimulus association in rhesus monkeys. Journal of Neuroscience13:4549–61. [aJPA, MJE]

Murray, E. A. & Mishkin, M. (1986) Visual recognition in monkeys following rhinalcortical ablations combined with either amygdalectomy or hippocampectomy.Journal of Neuroscience 7:1991–2003. [aJPA]

(1996) Forty-minute visual recognition memory in rhesus monkeys withhippocampal lesions. Society for Neuroscience Abstracts 22:281, 116.9.[aJPA, DJM]

(1998) Object recognition and location memory in monkeys with excitotoxiclesions of the amygdala and hippocampus. Journal of Neuroscience 18:6568–82. [rJPA, SMZ]

Murray, E. A. & Wise, S. P. (1996) Role of the hippocampus plus subjacent cortexbut not amygdala in visuomotor conditional learning in rhesus monkeys.Behavioral Neuroscience 110:1261–70. [AP]

(1997) Role of orbitoventral prefrontal cortex in conditional motor learning.Society for Neuroscience Abstracts 23:11. [AP]

Naber, P. A., Cabellero-Bleda, M., Jorritsma-Byham, B. & Witter, M. (1997)Parallel inputs to the hippocampal memory system through the peri- andpostrhinal cortices. Neuroreport 8(11):2617–21. [aJPA, RDB]

Naber, P. A., Witter, M. P. & Lopes da Silva, F. H. (submitted) Perirhinal cortexinput to the hippocampal formation in the rat: Evidence for parallel pathways,both direct and indirect. A combined physiological and anatomical study.European Journal of Neuroscience. [MPW]

Nadel, L. & Moscovitch, M. (1997) Consolidation, retrograde amnesia and thehippocampal formation. Current Opinion in Neurobiology 7:217–27. [JKF,LN]

(1998) Hippocampal contributions to cortical plasticity. Neuropharmacology37:431–39. [LN]

Neave, N., Lloyd, S., Sahgal, A. & Aggleton, J. P. (1994) Lack of effect of lesions inthe anterior cingulate cortex and retrosplenial cortex on certain tests of spatialmemory in the rat. Behavioural Brain Research 65:89–101. [aJPA]

Neave, N., Nagle, S. & Aggleton, J. P. (1997) Evidence for the involvement of themamillary bodies and cingulum bundle in allocentric spatial processing byrats. European Journal Neuroscience 9:101–15. [aJPA]

Neave, N., Nagle, S., Sahgal, A. & Aggleton, J. P. (1996) The effects of discretecingulum bundle lesions in the rat on the acquisition and performance of twotests of spatial working memory. Behavioural Brain Research 80:75–85.[aJPA, SMZ]

Nunn, J. & Hodges, H. (1994) Cognitive deficits induced by global cerebralischaemia: Relationship to brain damage and reversal by transplants.Behavioural Brain Research 65:1–31. [arJPA]

O’Boyle, V. J., Murray, E. A. & Mishkin, M. (1993) Effects of excitotoxic amygdalo-hippocampal lesions on visual recognition in rhesus monkeys. Society forNeuroscience Abstracts 19:438, 186.4. [aJPA]

O’Keefe, J. (1993) Hippocampus, theta rhythms and spatial memory. CurrentOpinion in Neurobiology 3:917–24. [aJPA]

O’Keefe, J. & Burgess, N. (1996) Geometric determinants of the place fields ofhippocampal neurons. Nature 381:425–28. [aJPA, AT]

O’Keefe, J. & Nadel, L. (1978) The hippocampus as a cognitive map. OxfordUniversity Press. [aJPA, JB, DJM, IQW]

Okuno, H. & Miyash*ta, Y. (1996) Expression of the transcription factor Zif268 inthe temporal cortex of monkeys during visual paired associate learning.European Journal of Neuroscience 8:2118–28. [aJPA]

Olton, D. S., Becker, J. T. & Handelmann, G. E. (1979) Hippocampus, space andmemory. Behavioral and Brain Sciences 2:313–65. [JKF]

Olton, D. S., Walker, J. A. & Woolf, W. A. (1982) A disconnection analysis ofhippocampal function. Brain Research 233:241–53. [arJPA]

Owen, M. J. & Butler, S. R. (1984) Does amnesia after transection of the fornix inmonkeys reflect abnormal sensitivity to proactive interference? BehaviouralBrain Research 14:183–92. [rJPA]

Otto, T. & Eichenbaum, H. (1992a) Complementary roles of the orbital prefrontalcortex and the perirhinal-entorhinal cortices in an odor-guided delayed-nonmatching to sample task. Behavioral Neuroscience 106:762–75. [MJE]

(1992b) Neuronal activity in the hippocampus during delayed non-match tosample performance in rats: Evidence for hippocampal processing inrecognition memory. Hippocampus 2:323–34. [aJPA]

Packard, M. G., Cahill, L. & McGaugh, J. L. (1994) Amygdala modulation ofhippocampal-dependent and caudate nucleus-dependent memory processes.Proceedings of the National Academy of Sciences USA 91:8477–81. [rJPA]

Paller, K. A., Acharya, A., Richardson, B. C., Plaisant, O., Shimamura, A. P., Reed,B. R. & Jagust, W. J. (1997) Functional neuroimaging of cortical dysfunctionin alcoholic Korsakoff ’s syndrome. Journal of Cognitive Neuroscience 9:277–93. [arJPA]

Papez, J. W. (1937) A proposed mechanism of emotion. Archives of Neurology andPsychology 38:725–43. [RDB, HJM]

Parker, A., Eacott, M. J. & Gaffan, D. (1997) The recognition memory deficitcaused by mediodorsal thalamic lesion in nonhuman primates: A comparisonwith rhinal cortex lesion. European Journal of Neuroscience 9:2423–31.[arJPA, JCD-A, AP]

Parker, A. & Gaffan, D. (1997a) Mamillary body lesions in monkeys impair object-in-place memory: Functional unity of the fornix-mamillary system. Journal ofCognitive Neuroscience 9:512–21. [aJPA, AP]

(1997b) The effects of anterior thalamic and cingulate cortex lesions on “object-in -place” memory in monkeys. Neuropsychologia 35:1093–102. [aJPA, AP]

(1997c) Frontal/temporal disconnection in monkeys: Memory for strategies andmemory for visual objects. Society for Neuroscience Abstracts 23:11. [AP]

(1998a) Interaction of frontal and perirhinal cortices in visual object recognitionmemory in monkeys. European Journal of Neuroscience 10:3044–57. [rJPA,AP]

(1998b) Memory after frontal/temporal disconnection in monkeys: Conditionaland non-conditional tasks, unilateral and bilateral frontal lesions.Neuropsychologia 36:259–71. [AP]

(1998c) Memory systems in primates: Episodic, semantic and perceptuallearning. In: Comparative neuropsychology, ed. D. Milner. Oxford UniversityPress. [MJE]

(in press) Magnocellular mediodorsal thalamic ablation does not disruptpreoperatively learned visual object associations, but impairs both newacquisition of object associations and scene learning. Behavioral Neuroscience.[AP]

Parker, A., Wilding, E. & Akerman, C. (1998) Novelty and memory encoding: VonRestorff effects in normal human and non-human primate recognitionmemory, and the role of frontal/perirhinal interaction. Journal of CognitiveNeuroscience 10:691–703. [AP]

Parkin, A. J. (1984) Memory and amnesia. Blackwell. [aJPA]Parkin, A. J., ed. (1997) Case studies in the neuropsychology of memory.

Psychology Press (Taylor & Francis). [JKF]Parkin, A. J., Dunn, J. C., Lee, C., O’Hara, P. F. & Nussbaum, L. (1993)

Neuropsychological sequelae of Wernicke’s encephalopathy in a 20-year- oldwoman: Selective impairment of a frontal memory system. Brain andCognition 21:1–19. [aJPA]

Parkin, A. J. & Hunkin, N. M. (1993) Impaired temporal context memory onanterograde but not retrograde tests in the absence of frontal pathology.Cortex 29:267–80. [aJPA]

Parkin, A. J. & Leng, N. R. C. (1993) Neuropsychology of the amnesic syndrome.Erlbaum. [aJPA]

Parkin, A. J., Leng, N. R. C. & Hunkin, N. M. (1990) Differential sensitivity tocontext in diencephalic and temporal lobe amnesia. Cortex 26:373–80.[aJPA]

Parkin, A. J., Yeomans, J. & Bindschaedler, C. (1994) Further characterization of

References/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (62)

the executive memory impairment following frontal lobe lesions. Brain andCognition 26:23–42. [aJPA]

Parkinson, J. A. & Aggleton, J. P. (1994) The failure of context shifts to alter therecognition of faces: Implications for contextual deficits in amnesia. Cortex30:351–54. [aJPA]

Parkinson, J. K., Murray, E. A. & Mishkin, M. (1988) A selective mnemonic rolefor the hippocampus in monkeys: Memory for the location of objects. Journalof Neuroscience 8:4159–67. [aJPA, SMZ]

Pearce, J. M., Roberts, A. D. L. & Good, M. (1998) Hippocampal lesions disruptnavigation based on cognitive maps but not heading vectors. Nature 396:75–77. [rJPA]

Penfield, W. & Mathieson, G. (1974) Memory: Autopsy findings and comments onthe role of hippocampus in experiential recall. Archives of Neurology 31:145–54. [aJPA]

Perfect, T. J., Mayes, A. R., Downes, J. J. & Van Eijk, R. (1996) Does contextdiscriminate recollection from familiarity in recognition memory? QuarterlyJournal of Experimental Psychology 49A:797–813. [aJPA]

Pickering, A. D. (1997) New approaches to the study of amnesic patients: Whatcan a neurofunctional philosophy and neural network methods offer? Memory5:255–300. [ADP]

Porter, M. C. & Mair, R. G. (1997) The effects of frontal cortical lesions onremembering depend on the procedural demand of tasks performed in theradial arm maze. Behavioural Brain Research 87:115–25. [RGM]

Poucet, B. (1993) Spatial cognitive maps in animals: New hypotheses on theirstructure and neural mechanisms. Psychological Review 100:163–82. [DJM]

Prepscius, C. & Levy, W. B. (1994) Sequence prediction and cognitive mapping bya biologically plausible neural network. Proceedings of the INNS WorldCongress on Neural Networks 5:164–69. [ADP]

Rajaram, S. (1993) Remembering and knowing: Two means of access to thepersonal past. Memory and Cognition 21:89–102. [aJPA, BJK]

Ramus, S. J., Zola-Morgan, S. & Squire, L. R. (1994) Effects of lesions ofperirhinal cortex or parahippocampal cortex on memory in monkeys. Societyfor Neuroscience Abstracts 20:1074. [aJPA, SMZ]

Ratcliff, R., Van Zandt, T. & McKoon, G. (1995) Process dissociation, single-process theories, and recognition memory. Journal of ExperimentalPsychology: General 124:353–74. [aJPA]

Rawlins, J. N. P. (1985) Associations across time: The hippocampus as a temporarymemory store. Behavioural and Brain Sciences 8:479–528. [JKF]

Reed, J. M. & Squire, L. R. (1997) Impaired recognition memory in patients withlesions limited to the hippocampal formation. Behavioural Neuroscience111:667–75. [aJPA, SMZ]

Rempel-Clower, N. L., Zola, S., Squire, L. R. & Amaral, D. G. (1996) Three casesof enduring memory impairment after bilateral damage limited to thehippocampal formation. Journal of Neuroscience 16:5233–55. [aJPA, SMZ]

Remy, M. (1942) Contribution a l’etude de la maladie de Korsakow. Monatsschriftfür Psychiatrie und Neurologie 106:128–44. [aJPA]

Rescorla, R. A. (1967) Pavlovian conditioning and its proper control procedures.Psychological Review 74:71–80. [DJM]

Riches, I. P., Wilson, F. A. W. & Brown, M. W. (1991) The effects of visualstimulation and memory on neurons of the hippocampal formation and theneighboring parahippocampal gyrus and inferior temporal cortex of theprimate. Journal of Neuroscience 11:1763–79. [aJPA]

Rigges, H. E. & Boles, R. S. (1944) Wernicke’s encephalopathy: Clinical andpathological studies of 42 cases. Quarterly Journal Studies of Alcohol 5:361–70. [aJPA]

Ringo, J. L. (1996) Stimulus specific adaptation in inferior temporal and medialtemporal cortex of the monkey. Behavioural Brain Research 76:191–97.[aJPA]

Robinson, J. K. & Mair, R. G. (1992) MK-801 prevents brain lesions anddelayed non-matching to sample deficits produced by pyrithiamine-induced encephalopathy in rats. Behavioural Neuroscience 106:623–33.[RGM]

Roediger, H. L. & McDermott, K. B. (1994) The problem of differing false-alarmrates for the process dissociation procedure: Comment on Verfaellie andTreadwell (1993) Neuropsychology 8:284–88. [aJPA]

Rolls, E. T. (1996) A theory of hippocampal function in memory. Hippocampus6:601–20. [JKF]

Rolls, E. T., Cahusac, P. M. B., Feigenbaum, J. D. & Miyash*ta, Y. (1993)Responses of single neurons in the hippocampus of the macaque related torecognition memory. Experimental Brain Research 93:299–306. [aJPA,DLH, AT

Rolls, E. T. & O’Mara, S. M. (1995) View-responsive neurons in the primatehippocampal complex. Hippocampus 5:409–24. [aJPA, AT]

Rolls, E. T., Treves, A., Robertson, R. G., Georges-Francois, P. & Panzeri, S. (1998)Information about spatial view in an ensemble of primate hippocampal cells.Journal of Neurophysiology 79:1797–813. [AT]

Romanes, G. J. (1889) Mental evolution in man. D. Appleton & Co. [DJM]Rombouts, S. A. R. B., Machielson, W. C. M., Witter, M. P., Barkhof, F.,

Lindeboom, J. & Scheltens, P. (1997) Visual association encoding showsactivation in the medial temporal lobe. A functional magnetic resonanceimaging study. Hippocampus 7:594–601. [MPW]

Room, P. & Groenewegen, H. J. (1986) Connections of the parahippocampalcortex in the cat. II. Subcortical afferents. Journal of Comparative Neurology251:451–73. [MPW]

Rothblat, L. A. & Kromer, L. F. (1991) Object recognition memory in the rat: Therole of the hippocampus. Behavioural Brain Research 42:25–32. [aJPA]

Rugg, M. D., Fletcher, P. C., Frith, C. D., Frackowiak, R. S. J. & Dolan, R. J.(1996) Differential activation of the prefrontal cortex in successful andunsuccessful memory retrieval. Brain 119:2073–83. [aJPA]

Rupniak, N. M. J. & Gaffan, D. (1987) Monkey hippocampus and learning aboutspatially directed movements. Journal of Neuroscience 7:2331–37. [rJPA]

Russchen, F. T., Amaral, D. G. & Price, J. L. (1987) The afferent input to themagnocellular division of the mediodorsal thalamic nucleus in the monkeyMacaca fascicularis. Journal of Comparative Neurology 256:175–210.[arJPA, MPW]

Sakai, K. & Miyash*ta, Y. (1991) Neuronal organisation for the long-term memoryof paired associates. Nature 354:152–55. [MJE]

Saunders, R. C. & Rosene, D. L. (1988) A comparison of the efferents of theamygdala and the hippocampal formation in the rhesus monkey: I.Convergence in the entorhinal, prorhinal, and perirhinal cortices. Journal ofComparative Neurology 271:153–84. [aJPA]

Saunders, R. C. & Weiskrantz, L. (1989) The effects of fornix transection andcombined fornix transection, mammillary body lesions and hippocampalablations on object-pair association memory in the rhesus monkey.Behavioural Brain Research 35:85–94. [aJPA]

Savage, L. M., Sweet, A. J., Castillo, R. & Langlais, P. J. (1997) The effects oflesions to thalamic lateral internal medullary lamina and posterior nuclei onlearning, memory and habituation in the rat. Behavioural Brain Research82:133–47. [aJPA]

Schacter, D. L. (1987) Memory, amnesia, and frontal lobe dysfunction.Psychobiology 15:21–36. [aJPA]

Schacter, D. L., Chiu, C.-Y, P. & Ochsner, K. N. (1993) Implicit memory: Aselective review. Annual Review of Neuroscience 16:159–82. [aJPA]

Schacter, D. L., Tharan, M., Cooprt, L. A. & Rubens, A. B. (1991) Preservedpriming of novel objects in patients with memory disorders. Journal ofCognitive Neuroscience 3:117–30. [aJPA]

Schacter, D. L., Verfaellie, M. & Anes, M. D. (1997) Illusory memories in amnesicpatients: Conceptual and perceptual false recognition. Neuropsychology11:331–42. [APY]

Schacter, D. L., Verfaellie, M. & Pradere, D. (1996) Neuropsychology of memoryillusions: False recall and recognition in amnesic patients. Journal of Memoryand Language 35(2):319–34. [APY]

Schnider, A., Gutbrod, K., Hess, C. W. & Schroth, G. (1996) Memory withoutcontext: Amnesia with confabulation after infarction of the right capsulargenu. Journal of Neurology, Neurosurgery, and Psychiatry 61:186–93.[aJPA]

Shaw, C. & Aggleton, J. P. (1993) The effects of fornix and medial prefrontallesions on delayed non-matching-to-sample by rats. Behavioural BrainResearch 54:91–102. [aJPA]

(1995) Evidence for the independence of recognition and recency memory inamnesic subjects. Cortex 31:57–71. [aJPA]

Shibata, H. (1992) Topographic organization of subcortical projections to theanterior thalamic nuclei in the rat. Journal of Comparative Neurology323:117–27.

(1993a) Direct projections from the anterior thalamic nuclei to theretrohippocampal region in the rat. Journal of Comparative Neurology337:431–45. [aJPA, JCD-A]

(1993b) Efferent projections from the anterior thalamic nuclei to the cingulatecortex in the rat. Journal of Comparative Neurology 330:533–42.

(1996) Direct projections from the entorhinal area to the anteroventral andlaterodorsal thalamic nuclei in the rat. Neuroscience Research 26:83–87.[aJPA]

Shimamura, A. P. (1995) Memory and frontal lobe function. In: The cognitiveneurosciences, ed. M. S. Gazzaniga. MIT Press. [BJK]

Shimamura, A. P., Janowsky, J. S. & Squire, L. R. (1990) Memory for the temporalorder of events in patients with frontal lobe lesions and amnesic patients.Neuropsychologia 28:803–13. [aJPA]

Shuren, J. E., Jacobs, D. H. & Heilman, K. M. (1997) Diencephallic temporalorder amnesia. Journal of Neurological and Neurosurgical Psychiatry 62:163–68. [MPW]

Sobotka, S. & Ringo, J. L. (1993) Investigations of long-term recognition andassociation memory in unit responses from inferotemporal cortex.Experimental Brain Research 96:28–38. [aJPA]

Sobotka, S. & Ringo, J. L. (1994) Stimulus specific adaptation in excited but not ininhibited cells in inferotemporal cortex of macaque. Brain Research 646:95–99. [aJPA]

References/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (63)

Snowden, J. S., Goulding, P. J. & Neary, D. (1989) Semantic dementia: A form ofcirc*mscribed cerebral atrophy. Behavioural Neurology 2:167–82. [KSG]

Snowden, J. S., Griffiths, H. L. & Neary, D. (1996) Semantic-episodic memoryinteractions in semantic dementia: Implications for retrograde memoryfunction. Cognitive Neuropsychology 13:1101–37. [KSG]

Squire, L. R. (1981) Two forms of human amnesia: An analysis of forgetting.Journal of Neuroscience 1:635–40. [aJPA]

(1992) Memory and the hippocampus: A synthesis from findings with rats,monkeys, and humans. Psychological Review 99:195–231. [JKF, ADP, SMZ]

Squire, L. R. & Alvarez, P. (1995) Retrograde amnesia and memory consolidation:A neurobiological perspective. Current Opinion in Neurobiology 5:169–77.[aJPA, LN]

Squire, L. R., Amaral, D. G., Zola-Morgan, S., Kritchevsky, M. & Press, G. (1989)Description of brain injury in the amnesic patient N.A. based on magneticresonance imaging. Experimental Neurology 105:23–35. [aJPA]

Squire, L. R. & Knowlton, B. J. (1995) Memory, hippocampus, and brain systems.In: The cognitive neurosciences, ed. M. Gazzinaga. MIT Press. [aJPA]

Squire, L. R., Knowlton, B. J. & Musen, G. (1993) The structure and organizationof memory. Annual Reviews in Psychology 44:453–95. [RDB]

Squire, L. R. & Shimamura, A. P. (1986) Characterizing amnesic patients forneurobehavioural study. Behavioural Neuroscience 100:866–77. [aJPA]

Squire, L. R. & Zola, S. (1996) Ischemic brain damage and memory impairment: Acommentary. Hippocampus 6:546–52. [SMZ]

Squire, L. R. & Zola-Morgan, S. (1991) The medial temporal lobe memory system.Science 253:1380–86. [aJPA, SMZ]

(1996) Ischemic brain damage and memory impairment: A commentary.Hippocampus 6:546–52. [aJPA]

Squire, L. R., Zola-Morgan, S. & Chen, K. S. (1988) Human amnesia and animalmodels of amnesia: Performance of amnesic patients on tests designed for themonkey. Behavioural Neuroscience 102:210–21. [aJPA]

Steckler, T., Drinkenburg, W. H. I. M., Sahgal, A. & Aggleton, J. P. (1998a)Recognition memory in rats - I. Concepts and classification.. Progress inNeurobiology 54:289–311. [RPK]

(1998b) Recognition memory in rats - II. Neuroanatomical substrates. Progressin Neurobiology 54:313–32. [JCD-A]

Steele, K. & Rawlins, J. N. P. (1993) The effects of hippocampectomy onperformance by rats of a running recognition task using long lists of non-spatial items. Behavioural Brain Research 54:1–10. [aJPA]

Stern, C. E., Corkin, S., Gonzalez, G., Guimaraes, A. R., Baker, J. R., Jennings,P. J., Carr, C. A., Sugiura, R. M., Vedantham, V. & Rosen, B. R. (1996) Thehippocampal formation participates in novel picture encoding: Evidence fromfunctional magnetic resonance imaging. Proceedings of the National Academyof Sciences, USA 93:8660–65. [LN]

Stuss, D. T., Eskes, G. A. & Foster, J. K. (1994) Experimental neuropsychologicalstudies of frontal lobe functions. In: Handbook of neuropsychology, vol. 9, ed.H. Spinnler & F. Boller. Elsevier. [JKF]

Stuss, D. T., Guberman, A., Nelson, R. & Larochelle, S. (1988) Theneuropsychology of paramedian thalamic infarction. Brain and Cognition8:348–78. [aJPA]

Sugita, K., Doi T., Mutsuga, N. & Takaoka, Y. (1971) Clinical study of fornicotomyfor psychom*otor epilepsy and behavior disorder. In: Special topics instereotaxis: Epilepsy, disorders of behaviour, and autonomic balance, ed. W.Umbach. Hippokrates. [aJPA]

Sutherland, R. J. & Hoesing, J. M. (1993) Posterior cingulate cortex and spatialmemory: A microlimnology analysis. In: Neurobiology of cingulate cortex andlimbic thalamus, ed. B. A. Vogt & M. Gabriel. Birkhauser. [aJPA]

Sutherland, R. J. & Rodriguez, A. J. (1989) The role of the fornix/fimbria and somerelated subcortical structures in place learning and memory. BehaviouralBrain Research 32:265–77. [aJPA]

Sutherland, R. J. & Rudy, J. W. (1989) Configural association theory: The role ofthe hippocampal formation in learning, memory, and amnesia. Psychobiology17:129–44. [SMZ]

Sutherland, R. J., Whishaw, I. Q. & Kolb, B. (1988) Contributions of cingulatecortex to two forms of spatial learning and memory. Journal of Neuroscience8:1863–72. [aJPA]

Suzuki, W. A. (1996a) The anatomy, physiology and functions of the perirhinalcortex. Current Opinion in Neurobiology 6:179–86. [aJPA, RGM]

(1996b) Neuroanatomy of the monkey entorhinal, perirhinal andparahippocampal cortices: Organization of cortical inputs andinterconnections with amygdala and striatum. Seminars in the Neurosciences8:3–12. [aJPA]

Suzuki, W. A. & Amaral, D. G. (1990) Cortical inputs to the CA1 field of themonkey hippocampus originate from the perirhinal and parahippocampalcortex but not from area TE. Neuroscience Letters 115:43–48. [aJPA, RDB]

(1994a) Perirhinal and parahippocampal cortices of the Macaque monkey:Cortical afferents. Journal of Comparative Neurology 350:497–533. [aJPA,RDB, SMZ]

(1994b) Topographic organization of the reciprocal connections between the

monkey entorhinal cortex and the perirhinal and parahippocampal cortices.Journal of Neuroscience 14:1856–77. [aJPA, RDB]

Suzuki, W. A., Miller, E. K. & Desimone, R. (1995) Object and place memory inthe monkey entorhinal cortex. Society for Neuroscience Abstracts 21:19.[aJPA]

Suzuki, W. A., Zola-Morgan, S., Squire, L. R. & Amaral, D. G. (1993) Lesions ofthe perirhinal and parahippocampal cortices in the monkey produce long-lasting memory impairment in the visual and tactual modalities. Journal ofNeuroscience 13:2430–51. [aJPA]

Sweet, W. H., Talland, G. A. & Ervin, F. R. (1959) Loss of recent memoryfollowing section of fornix. Transactions of the American NeurologicalAssociation 84:76–82. [aJPA]

Taube, J. S. (1995) Head direction cells recorded in the anterior thalamic nuclei offreely moving rats. Journal of Neuroscience 15:70–86. [arJPA]

Teng, E., Squire, L. R. & Zola, S. (1997) Different memory roles for theparahippocampal and perirhinal cortices in spatial reversal. Society forNeuroscience Abstracts 23:12. [SMZ]

(1998) Effects of removal from the testing apparatus on delayed nonmatching tosample performance in monkeys with lesions of the hippocampal region.Society for Neuroscience Abstracts 24:17. [SMZ]

Thomas, G. J. & Brito, G. N. O. (1980) Recovery of delayed alternation in rats afterlesions in medial prefrontal cortex and septum. Journal of Comparative andPhysiological Psychology 94:808–18. [aJPA]

Thomas, G. J. & Spafford, P. S. (1984) Deficits for representational memoryinduced by septal and cortical lesions (singly and combined) in rats.Behavioural Neuroscience 98:394–404. [aJPA]

Thompson, R. F. (1990) Neural mechanisms of classical conditioning in mammals.Philosophical Transactions of the Royal Society London B 329:161–70.[aJPA]

Torvik, A. (1987) Topographic distribution and severity of brain lesions inWernicke’s encephalopathy. Clinical Neuropathology 6:25–29. [aJPA]

Treves, A., Georges-Francois, P., Panzeri, S., Robertson, R. G. & Rolls, E. T. (1998)The metric content of spatial views as represented in the primatehippocampus. In: Neuronal circuits and networks, ed. V. Torre & J.Nicholls.NATO ASI series, vol. 167. Springer. [AT]

Treves, A., Panzeri, S., Robertson, R. G., Georges-Francois, P. & Rolls, E. T.(1996a) The emergence of structure in neuronal representations. Society forNeuroscience Abstracts 22:281. [AT]

Treves, A., Skaggs, W. E. & Barnes, C. A. (1996b) How much of the hippocampuscan be explained by functional constraints? Hippocampus 6:666–74. [AT]

Tucker, D. M., Roeltgen, D. P., Tully, R., Hartmann, J. & Boxell, C. (1988)Memory dysfunction following unilateral transection of the fornix: Ahippocampal disconnection syndrome. Cortex 24:465–72. [aJPA]

Tulving, E. (1983) Elements of episodic memory. Clarendon Press. [aJPA](1985a) Memory and consciousness. Canadian Psychologist 26(1):1–12. [aJPA,

BJK, APY](1985b) How many memory systems are there? American Psychologist 40:385–

98. [SMZ](1991) Concepts in human memory. In: Memory: Organization and locus of

change, ed. L. R. Squire, N. M. Weinberger, G. Lynch & J. McGaugh. OxfordUniversity Press. [SMZ]

Tulving, E., Kapur, S., Markowitsch, H. J., Craik, F. I. M., Habib, R. & Houle, S.(1994) Neuroanatomical correlates of retrieval in episodic memory: Auditorysentence recognition. Proceedings of the National Academy of Sciences, USA91:2012–15. [LN]

Tulving, E. & Markowitsch, H. J. (1994) What do animal models of memorymodel? Behavioural and Brain Sciences 17:498–99. [aJPA]

(1998) Episodic and declarative memory: Role of the hippocampus.Hippocampus 8:198–204. [HJM]

Tulving, E., Markowitsch, H. J., Craik, F. I. M., Habib, R. & Houle, S. (1996)Novelty and familiarity activations in PET studies and memory encodingretrieval. Cerebral Cortex 6:71–79. [aJPA]

Ungerleider, L. G. & Mishkin, M. (1982) Two cortical visual systems. In: Analysisof visual behavior, ed. D. J. Ingle, M. A. Goodale & R. J. W. Mansfield. MITPress. [EEK]

Valenstein, E., Bowers, D., Verfaellie, M., Heilman, K. M., Day, A. & Watson, R. T.(1987) Retrosplenial amnesia. Brain 110:1631–46. [aJPA]

Vandenberghe, R., Dupont, P., Bormans, G., Mortelmans, L. & Orban, G. (1995)Rapid communication blood flow in human anterior temporal cortexdecreases with stimulus familiarity. Neuroimage 2:306–13. [aJPA]

Van der Werf, Y. D., Weerts, J. G. E., Jolles, J., Witter, M. P., Lindeboom, J. &Scheltens, P. (1998) Neuropsychological correlates of a right unilateral lacunarthalamic infarction. Journal of Neurological and Neurosurgical Psychiatry (inpress). [MPW]

Van Hoesen, G. W. (1997) Ventromedial temporal lobe anatomy, with commentson Alzheimer’s disease and temporal injury. Journal of Neuropsychiatry andClinical Neurosciences 9:331–41. [KSG]

Vargha-Khadem, F., Gadian, D. G., Watkins, K. E., Connelly, A., Van Paesschen,

References/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (64)

W. & Mishkin, M. (1997) Differential effects of early hippocampal pathologyon episodic and semantic memory. Science 277:376–80. [arJPA, KSG, SMZ]

Verfaellie, M. (1994) A re-examination of recognition memory in amnesia: Reply toRoediger and McDermott. Neuropsychology 8:289–92. [aJPA]

Verfaellie, M. & Treadwell, J. R. (1993) Status of recognition memory in amnesia.Neuropsychology 7(1):5–13. [arJPA, APY]

Victor, M., Adams, R. D. & Collins, G. H. (1971) The Wernicke-Korsakoffsyndrome. Blackwell. [aJPA]

(1989) The Wernicke-Korsakoff syndrome and related neurologic disorders dueto alcoholism and malnutrition, 2nd edition. F. A. Davis. [aJPA, RGM]

Victor, M. & Agamanolis, D. (1990) Amnesia due to lesions confined to thehippocampus: A clinical-pathologic study. Journal of Cognitive Neuroscience2:246–57. [aJPA]

Vogt, B. A., Gabriel, M. Vogt, L. J., Poremba, A., Jensen, E. L., Kubota, Y. & Kang,E. (1991) Muscarinic receptor binding increases in anterior thalamus andcingulate cortex during discriminative avoidance learning. Journal ofNeuroscience 11:1508–14. [aJPA]

Volpe, B. T., Holtzman, J. D. & Hist, W. (1986) Further characterization of patientswith amnesia after cardiac arrest: Preserved recognition memory. Neurology36:408–11. [aJPA]

Von Cramon, D. Y. (1992) Focal cerebral lesions damaging (subcortical) fibreprojections relating to memory and learning functions in man. In:Neuropsychological disorders associated with subcortical lesions, ed. G. Valler,S. F. Cappa & C. W. Wallesch. Oxford University Press. [aJPA]

Von Cramon, D. Y., Hebel, N. & Schuri, U. (1985) A contribution to theanatomical basis of thalamic amnesia. Brain 108:993–1008. [RGM, MPW]

Von Cramon, D. Y., Markowitsch, H. J. & Schuri, U. (1993) The possiblecontributions of the septal region to memory. Neuropsychologia 31:1159–80.[HJM]

Wagner, A. D., Stebbins, G. T., Masciari, F., Fleischman, D. A. & Gabriel, J. D. E.(1998) Neuropsychological dissociation between recognition familiarity andperceptual priming in visual long-term memory. Cortex 34:493–511. [rJPA]

Wan, H., Aggleton, J. P. & Brown, M. W. (1999) Different contributions of thehippocampus and perirhinal cortex to recognition memory. Journal ofNeuroscience 19:1142–48. [aJPA, JB]

Warburton, E. C. & Aggleton, J. P. (1999) Differential effects in the Morris watermaze following cytotoxic lesions of the anterior thalamus and fornixtransection. Behavioural Brain Research 98:27–38. [arJPA]

Warburton, E. C., Aggleton, J. P. & Muir, J. L. (1998) Comparing the effects ofselective cingulate cortex and cingulum bundle lesions on a spatial navigationtask. European Journal of Neuroscience 10:622–34. [arJPA]

Warburton, E .C., Baird, A. L. & Aggleton, J. P. (1997) Assessing the magnitude ofthe allocentric spatial deficit associated with complete loss of the anteriorthalamic nuclei in rats. Behavioural Brain Research 87:223–32. [aJPA,JCD-A]

Warrington, E. K. (1975) Selective impairment of semantic memory. QuarterlyJournal of Experimental Psychology 27:635–57. [KSG]

(1984) The recognition memory test. NFER-Nelson. [arJPA, KSG, MPW]Warrington, E. K. & Weiskrantz, L. (1974) The effect of prior learning on

subsequent retention in amnesia patients. Neuropsychologia 12:419–28.[aJPA]

Weiskrantz, L. (1985) On issues and theories of the human amnesic syndrome. In:Memory systems of the brain, ed. N. M Weinberger, J. L. McGaugh & G.Lynch. Guilford Press. [aJPA]

(1990) Problems of learning and memory: One or multiple memory systems?Philosophical Transactions of the Royal Society London B. 329:99–108.[aJPA]

Wheeler, M. A., Stuss, D. T. & Tulving, E. (1995) Frontal lobe damage producesepisodic memory impairment. Journal of the InternationalNeuropsychological Society 1:525–36. [aJPA]

Whishaw, I. Q. (1998) Place learning in hippocampal rats and the path integrationhypothesis. Neuroscience and Biobehavioural Reviews 22:209–20. [IQW]

Whishaw, I. Q., Cassel, J.-C. & Jarrard, L. E. (1995) Rats with fimbria-fornixlesions display a place response in a swimming pool: A dissociation betweengetting there and knowing where. Journal of Neuroscience 15:5779–88.[IQW]

Whishaw, I. Q. & Jarrard, L. E. (1995) Similarities vs. differences in place learningand circadian activity in rats after fimbria-fornix section or ibotenate removalof hippocampal cells. Hippocampus 5:595–604. [IQW]

(1996) Evidence for extrahippocampal involvement in place learning andhippocampal involvement in path integration. Hippocampus 6:513–24. [AP,IQW]

Whishaw, I. Q. & Maaswinkel, H. (1998) Rats with fimbria-fornix lesions areimpaired in path integration: A role of the hippocampus in “sense ofdirection.” Journal of Neuroscience 18:3050–58. [IQW]

Whishaw, I. Q., McKenna, J. E. & Maaswinkel, H. (1997) Hippocampal lesions andpath integration. Current Opinion in Neurobiology 7:228–34. [AP, IQW]

Whishaw, I. Q. & Tomie, J. (1997a) Piloting and dead reckoning dissociated by

fimbria-fornix lesions in a rat food carrying task. Behavioural Brain Research89:87–97. [IQW]

(1997b) Perseveration on place reversals in spatial swimming pool tasks: Furtherevidence for place learning in hippocampal rats. Hippocampus 7:361–70.[IQW]

Wiig, K. A. & Bilkey, D. K. (1994) The effects of perirhinal cortical lesions onspatial reference memory in the rat. Behavioural Brain Research 63:101–109.[DGM]

(1995) Lesions of rat perirhinal cortex exacerbate the memory deficit observedfollowing damage to the fimbria-fornix. Behavioural Neuroscience 109:620–30. [aJPA]

Wiig, K. A., Booth, S. J., Lui, P., Cooper, L. & Bilkey, D. K. (1995) The effects ofseparate and combined lesions of perirhinal cortex, fornix and septum onmemory in the rat. Society for Neuroscience Abstracts 2:1935. [MJE]

Wilding, E. L. & Rugg, M. D. (1996) An event-related potential study ofrecognition memory with and without retrieval of source. Brain 119:889–905.[aJPA]

Wilson, B., Baddeley, A. & Kapur, N. (1995) Dense amnesia in a professionalmusician following herpes simplex virus encephalitis. Journal of Clinical andExperimental Neuropsychology 17:668–81. [NK]

Witter, M. P. & Amaral, D. G. (1991) Entorhinal cortex of the monkey: V.Projections to the dentate gyrus, hippocampus, and subicular complex.Journal of Comparative Neurology 307:437–39. [aJPA]

Witter, M. P. & Groenewegen, H. J. (1986) Connections of the parahippocampalcortex in the cat. III. Cortical and thalamic efferents. Journal of ComparativeNeurology 252:1–31. [MPW]

Witter, M. P., Groenewegen, H. J., Lopes da Silva, F. H. & Lohman, A. H. M.(1989) Functional organization of the extrinsic and intrinsic circuitry of theparahippocampal region. Progress in Neurobiology 33:161–253. [aJPA]

Wood, E. R., Mumby, D. G., Pinel, J. P. J. & Phillips, A. G. (1993) Impaired objectrecognition memory in rats following ischemia-induced damage to thehippocampus. Behavioural Neuroscience 107:51–62. [aJPA]

Wood, E. R. & Phillips, A. G. (1991) Deficits on a one trial object recognition taskby rats with hippocampal CA1 lesions produced by cerebral ischemia.Neuroscience Research Communications 9:177–82. [aJPA]

Woods, B. T., Schoene, W. & Kneisley, L. (1982) Are hippocampal lesionssufficient to cause lasting amnesia? Journal of Neurology, Neurosurgery andPsychiatry 45:243–47. [aJPA]

Woolsey, R. M. & Nelson, J. S. (1975). Asymptomatic destruction of the fornix inman. Archives of Neurology 32:566–68. [aJPA]

Xiang, J.-Z. & Brown, M. W. (1997) Neuronal encoding of the prior occurrence ofvisual stimuli in rhinal cortex and area TE of the monkey. Brain ResearchAssociation Abstracts 14:42. [aJPA]

(1998) Differential neuronal encoding of novelty, familiarity and recency inregions of the anterior temporal lobe. Neuropharmacology (in press).[arJPA]

Yoneda, Y., Mori, E., Yamash*ta, H. & Yamadori, A. (1994) MRI volumetry ofmedial temporal lobe structures in amnesia following herpes simplexencephalitis. European Neurology 34:243–52. [rJPA, NK]

Yonelinas, A. P. (1994) Receiver-operating characteristics in recognition memory:Evidence for a dual process model. Journal of Experimental Psychology:Learning, Memory and Cognition 20:1341–54. [aJPA, ADP, APY]

(1997) Recognition memory ROCs for item and associative information: Thecontribution of recollection and familiarity. Memory and Cognition 25:747–63. [aJPA]

Yonelinas, A. P., Dobbins, I., Szymanski, M. D., Dhaliwal, H. S. & King, L. (1996)Signal-detection, threshold, and dual-process models of recognition memory:ROCs and conscious recollection. Consciousness and Cognition 5:418–41.[ADP]

Yonelinas, A. P. & Jacoby, L. L. (1995) The relation between remembering andknowing as bases for recognition: Effects of size congruency. Journal ofMemory and Language 34:622–43. [aJPA]

Yonelinas, A. P., Kroll, N. E. A., Dobbins, I. G., Lazzara, M. & Knight, R. T. (1998)Recollection and familiarity deficits in amnesia: Convergence of remember-know, process dissociation, and receiver operating characteristic data.Neuropsychology 12:323–39. [APY]

Zhang, Y. P., Burk, J. A., Glode, B. M. & Mair, R. G. (1998) The effects of thalamicand olfactory cortical lesions on continuous olfactory DNMTS and olfactorydiscrimination in the rat. Behavioural Neuroscience 112:39–53. [RGM]

Zhu, X. O., Brown, M. W. & Aggleton, J. P. (1995a) Neuronal signalling ofinformation important to visual recognition memory in rat rhinal andneighbouring cortices. European Journal of Neuroscience 7:753–65.[arJPA]

Zhu, X. O., Brown, M. W., McCabe, B. J. & Aggleton, J. P. (1995b) Effects of thenovelty or familiarity of visual stimuli on the expressions of the immediateearly gene c-fos in rat brain. Neuroscience 69:821–29. [arJPA]

Zhu, X. O., McCabe, B. J., Aggleton, J. P. & Brown, M. W. (1996) Mappingrecognition memory through the differential expression of the immediate

References/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (65)

early gene c-fos induced by novel or familiar visual stimulation. Neuroreport7:1871–75. [aJPA, JB]

(1997) Differential activation of the hippocampus and perirhinal cortex by novelvisual stimuli and a novel environment. Neuroscience Letters 229:141–43.[aJPA]

Zola, S. & Squire, L. R. (in press) Memory and brain systems. In: The Oxfordhandbook of memory, ed. E. Tulving & F. Craik. Oxford University Press.[SMZ]

Zola, S., Teng, E., Clark, R. E., Stefanacci, L., Buffalo, E. A. & Squire, L. R. (1998)Impaired recognition memory and simple discrimination learning in monkeysfollowing lesions limited to the hippocampal region made by radio frequency.,ischemia, or ibotenic acid. Society for Neuroscience Abstracts 24:17. [SMZ]

Zola-Morgan, S., Dabrowska, J., Moss, M. & Mahut, H. (1983) Enhancedpreference for perceptual novelty in the monkey after section of the fornix butnot after ablation of the hippocampus. Neuropsychologia 21:433–54. [aJPA]

Zola-Morgan, S. & Squire, L. R. (1985a) Amnesia in monkeys after lesions of themediodorsal nucleus of the thalamus. Annals of Neurology 17:558–64. [arJPA]

(1985b) Medial temporal lesions in monkeys impair memory on a variety of taskssensitive to human amnesia. Behavioural Neuroscience 99:22–34. [aJPA,DGM]

(1986) Memory impairment in monkeys following lesions limited to thehippocampus. Behavioural Neuroscience 100:155–60. [aJPA]

(1993) Neuroanatomy of memory. Annual Review of Neurocience 16:547–63.[aJPA]

Zola-Morgan, S., Squire, L. R. & Amaral, D. G. (1986) Human amnesia and themedial temporal region: Enduring memory impairment following bilaterallesions limited to field CA1 of the hippocampus. Journal of Neuroscience10:2950–67. [aJPA, NK, SMZ]

(1989a) Lesions of the hippocampal formation but not lesions of the fornix orthe mammillary nuclei produce long-lasting memory impairment in monkeys.Journal of Neuroscience 9:898–913. [aJPA]

Zola-Morgan, S., Squire, L. R., Amaral, D. G. & Suzuki, W. A. (1989b) Lesions ofperirhinal and parahippocampal cortex that spare the amygdala andhippocampal formation produce severe memory impairment. Journal ofNeuroscience 9:4355–70. [aJPA]

Zola-Morgan, S., Squire, L. R., Clower, R. P. & Rempel, N. L. (1993) Damage tothe perirhinal cortex exacerbates memory impairment following lesions tothe hippocampal formation. Journal of Neuroscience 13:251– 65. [aJPA,SMZ]

Zola-Morgan, S., Squire, L. R. & Mishkin, M. (1982) The neuroanatomy ofamnesia: Amygdala-hippocampal versus temporal stem. Science 218:1337–39.[aJPA]

Zola-Morgan, S., Squire, L. R. & Ramus, S. J. (1994) Severity of memoryimpairment in monkeys as a function of locus and extent of damage within themedial temporal lobe memory system. Hippocampus 4:483–94. [SMZ]

Zola-Morgan, S., Squire, L. R., Rempel, N. L., Clower, R. P. & Amaral, D. G.(1992) Enduring memory impairment in monkeys after ischemic damage tothe hippocampus. Journal of Neuroscience 12:2582–96. [arJPA, SMZ]

References/Aggleton & Brown: Episodic memory, amnesia, and hippocampus


Episodic memory, amnesia, and the hippocampal–anterior ... - [PDF Document] (2024)


What happens to the hippocampus in amnesia? ›

Temporally graded retrograde amnesia and memory consolidation. In addition to impairment in long-term storage of new memories, damage to the hippocampal region results in temporally graded retrograde amnesia.

What is the role of the hippocampus in episodic and procedural memory? ›

In episodic memory the hippocampus supports the recollection of contextual details, whereas regions such as the perirhinal cortex support familiarity. In contrast, in working memory the hippocampus supports a global familiarity matching signal, whereas a frontoparietal network supports recollection.

Does damage to the hippocampus affect episodic memory? ›

The hippocampus and surrounding structures in the temporal lobe are important in episodic memory and are part of an important network called the default mode network, which includes several brain areas including frontal and parietal regions and has been implicated in episodic memory functioning.

Which type of memory is most impaired by damage to the hippocampus? ›

If the hippocampus is damaged by disease or injury, it can influence a person's memories as well as their ability to form new memories. Hippocampus damage can particularly affect spatial memory, or the ability to remember directions, locations, and orientations.

How do you tell if your hippocampus is damaged? ›

What Are Signs That Something Could Be Wrong With Your Hippocampus?
  • Short-term memory loss.
  • Difficulty with directions and spatial skills.
  • Trouble forming new memories (even if your long-term memories are not affected)
  • Having a hard time with factual information like dates, times, names, and locations.
Sep 1, 2022

What type of memory is affected by hippocampal damage mainly? ›

While your hippocampus affects your memory, damage to it usually only targets your short-term memory. Your long-term memory isn't stored in your hippocampus. This is why someone with a condition like Alzheimer's disease might remember things that happened long ago but won't remember the names of people they just met.

What is the hippocampal function in episodic memory? ›

Studies in humans and animal models indicate that the hippocampus plays a key part in organizing memories in the context in which they were experienced — a defining feature of episodic memory — whereas the PFC controls the retrieval of context-appropriate memories by suppressing competing, context-inappropriate ...

What part of memory does the hippocampus control? ›

The hippocampus helps humans process and retrieve two types of memory, declarative memories and spatial relationships. Declarative memories are those related to facts and events. Examples can include learning how to memorize speeches or lines in a play. Spatial relationship memories involve pathways or routes.

How does amnesia affect episodic memory? ›

There is now a consensus that amnesia associated with hippocampal (and, in many cases, broader medial temporal lobe) damage results in deficits in episodic memory, delayed recall, and recollective experience.

What part of the brain controls episodic memory? ›

Hippocampus. The hippocampus, located in the brain's temporal lobe, is where episodic memories are formed and indexed for later access. Episodic memories are autobiographical memories from specific events in our lives, like the coffee we had with a friend last week.

What is an example of an episodic memory in everyday life? ›

Your first kiss, first day of school, a friend's birthday party, and your brother's graduation are all examples of episodic memories. In addition to your overall recall of the event itself, the episodic memory include the locations and times of the events.

What is the amnesia caused by hippocampal damage? ›

Damage to the hippocampal region and related medial temporal lobe structures (perirhinal, entorhinal, and parahippocampal cortices) impairs new learning (anterograde amnesia) as well as memory for information that was acquired before the damage occurred (retrograde amnesia).

Can hippocampus damage be reversed? ›

Regeneration and repair

There is good news on that front. Abstinence from alcohol will reverse hippocampal damage within 6 to 12 months. Antidepressants have been found to stimulate production of new brain cells (neurogenesis) and to gradually rebuild the structure of the hippocampus in depressed individuals.

How to improve episodic memory? ›

Write in a journal. Journaling is an excellent tool for supporting episodic memory because it requires you to reflect on specific events. Naidoo explains that when writing in a journal, you'll have to recall things like specific facts and what you felt during the event.

What happens to the brain during amnesia? ›

Bilateral damage to either the medial temporal lobe or the diencephalic midline causes an amnesic syndrome, i.e. a global impairment in the ability to acquire new memories regardless of sensory modality, and a loss of some memories, especially recent ones, from the period before amnesia began.

Does damage to the hippocampus cause anterograde or retrograde amnesia? ›

Studies in humans and laboratory animals show that damage to the hippocampus produces profound retrograde amnesia. The magnitude of the loss of previously learned information is related to the extent of the damage to the hippocampus (Epp et al., 2008; Lehmann et al., 2010; Mumby et al., 1999; Sutherland et al., 2001).

What function did the hippocampus serve and which type of amnesia did its removal cause? ›

These important findings show that even partial damage to the hippocampal formation, involving just CA1, can give rise to anterograde amnesia, thus indicating that this area of the hippocampus performs a vital role in memory.

What happens to the hippocampus? ›

Summary. The hippocampus plays a key role in learning and memory. Several conditions can affect the hippocampus' function, including Alzheimer's disease, depression, stress, and epilepsy. Making changes to a person's diet and exercise may help preserve brain function and protect the health of the hippocampus.


Top Articles
Latest Posts
Article information

Author: Fredrick Kertzmann

Last Updated:

Views: 5877

Rating: 4.6 / 5 (46 voted)

Reviews: 85% of readers found this page helpful

Author information

Name: Fredrick Kertzmann

Birthday: 2000-04-29

Address: Apt. 203 613 Huels Gateway, Ralphtown, LA 40204

Phone: +2135150832870

Job: Regional Design Producer

Hobby: Nordic skating, Lacemaking, Mountain biking, Rowing, Gardening, Water sports, role-playing games

Introduction: My name is Fredrick Kertzmann, I am a gleaming, encouraging, inexpensive, thankful, tender, quaint, precious person who loves writing and wants to share my knowledge and understanding with you.