A Unified Framework for the Functional Organization of the MedialTemporal Lobes and the Phenomenology of Episodic Memory

Charan Ranganath*

ABSTRACT: There is currently an intense debate about the nature ofrecognition memory and about the roles of medial temporal lobe subre-gions in recognition memory processes. At a larger level, this debate hasbeen about whether it is appropriate to propose unified theories toexplain memory at neural, functional, and phenomenological levels ofanalysis. Here, I review findings from physiology, functional imaging,and lesion studies in humans, monkeys, and rodents relevant to the rolesof medial temporal lobe subregions in recognition memory, as well as inshort-term memory and perception. The results from these studies areconsistent with the idea that there is functional heterogeneity in themedial temporal lobes, although the differences among medial temporallobe subregions do not precisely correspond to different types of mem-ory tasks, cognitive processes, or states of awareness. Instead, the evi-dence is consistent with the idea that medial temporal lobe subregionsdiffer in terms of the kind of information they process and represent,and that these regions collectively support episodic memory by bindingitem and context information. VVC 2010 Wiley-Liss, Inc.

KEY WORDS: fmri; recollection; hippocampus; recognition; perirhinal

INTRODUCTION

This special issue focuses on a specific and particularly controversialtopic in episodic memory research, which is how best to characterize theneural substrates of recognition memory. This is an important, but rela-tively narrow question, which might leave many readers wondering whatthe controversy is all about. Make no mistake, however, this debate isnot just about recognition memory, but rather about whether we as afield are ready to propose unified models to explain the neural and func-tional organization of memory processes in humans and other mammals.At one point, it would have been difficult to conceive that neuroscienceor psychology has progressed to the point where such a ‘‘grand unifiedtheory’’ could be developed, but now many of us believe that such anachievement is not only possible, but likely.

Here, I will review research pertinent to our understanding of thefunctional organization of the medial temporal lobes (MTL), and how itrelates to the nature of memory processes and the phenomenology of

mnemonic experience. In this review, I will touch onresearch from several domains examining the contribu-tions of different MTL subregions to different types ofmemory tasks (recognition, recall, short-term memory,perceptual discrimination), retrieval processes (thresh-old/pattern completion vs. signal detection/globalmatching), and conscious mnemonic experiences(remembering vs. knowing). Many theories of MTLfunction emphasize one of these levels of description,and such approaches can explain a large amount of theextant evidence. Here, I propose a unified framework,the Binding of Items and Contexts (BIC) model, toexplain MTL function across multiple domains andlevels of analysis. The central tenet to this frameworkis that MTL function may be explained in terms of thekinds of information that typically support differentprocesses, mnemonic experiences, and performance ondifferent tasks.

WHY RECOGNITION MEMORY? AHISTORICAL PERSPECTIVE

To understand the controversy about the MTL andrecognition memory, it is useful to consider the recenthistory of work on this topic. Interest in the role ofthe MTL in memory was stimulated by Brenda Mil-ner and her colleagues (Milner et al., 1968; Milner,1972), who demonstrated that extensive damage tothe MTL region in humans severely impairs the for-mation of new memories for events, while sparingmany other processes, such as motor skill learning.Milner’s findings stimulated work on animal modelsto further pinpoint the roles of different brain regions,particularly in the MTL, in memory for events. Muchof the work in nonhuman primates focused on thedelayed matching (DMS) or nonmatching to sample(DNMS) task (Gaffan, 1974; Mishkin and Delacour,1975), which is a test of object recognition memory.In DNMS, one or more ‘‘sample’’ objects is shownand after a retention delay, the subject is shown thesample object, along with a novel object, and isrewarded for choosing the novel object. This task issimilar to paradigms used to study short-term memory(Konorski, 1959), but was adapted by using a largeset of complex stimuli, so that within a given testingsession, each stimulus to be remembered and the cor-

Center for Neuroscience and Department of Psychology, University ofCalifornia at Davis, Davis, CaliforniaGrant sponsor: National Institute of Mental Health; Grant numbers:R01MH068721, R01MH083734.*Correspondence to: Charan Ranganath, Ph.D., Center for Neuroscience,University of California at Davis, 1544 Newton Court, Davis, CA 95618.E-mail: cranganath@ucdavis.eduAccepted for publication 25 June 2010DOI 10.1002/hipo.20852Published online 6 October 2010 in Wiley Online Library(wileyonlinelibrary.com).

HIPPOCAMPUS 20:1263–1290 (2010)

VVC 2010 WILEY-LISS, INC.

responding foil object are trial-unique (Gaffan, 1974; Mishkinand Delacour, 1975). The usage of trial-unique stimuli mini-mizes the accumulation of proactive interference, making itpossible to perform the task by remembering the sampleobject during the memory decision, or by evaluating the dif-ference in familiarity between the sample and foil objects dur-ing the memory decision. Monkeys with large MTL lesionsshow severe impairment on this task, especially when thedelay length is long or if the list length (i.e., Number of sam-ple objects) is large (Mishkin, 1978; Mahut et al., 1982; Mur-ray and Mishkin, 1984; Zola-Morgan and Squire, 1985; Mur-ray and Mishkin, 1986; Zola-Morgan and Squire, 1986; Zola-Morgan et al., 1993; Meunier et al., 1996). On the basis ofthe findings described above, it was argued that MTL lesionsin monkeys provide an animal model of the kinds of severeepisodic memory impairments seen in amnesia (Mishkin, 1982;Squire et al., 1988). This was a significant breakthrough, in thatit paved the way for careful lesion studies to examine the roles ofspecific MTL subregions in recognition memory.

Building on emerging knowledge of the anatomical organiza-tion of the MTL, researchers outlined a ‘‘system’’ of regionsthat were likely to contribute to memory performance, includ-ing the perirhinal cortex (PRc), parahippocampal cortex(PHc),i entorhinal cortex (EC), hippocampal formation, amyg-dala, and diencephalon, which later came to be described asthe ‘‘medial temporal lobe memory system’’ (Squire and Zola-Morgan, 1991).ii Much of the subsequent research on theMTL memory system has focused on the hippocampus and theEC, PRc, and PHc, which are the primary sources of neocorti-cal inputs to the hippocampus. This research demonstrated aclear role for the PRc in object recognition memory, as assessedby the DNMS task (Murray and Mishkin, 1986; Zola-Morganet al., 1989a; Meunier et al., 1993; Suzuki et al., 1993; Zola-Morgan et al., 1993; Meunier et al., 1996; Nemanic et al.,2004), with one study revealing deficits even with no delay(Eacott et al., 1994). In contrast, hippocampal lesions (Murrayand Mishkin, 1998; Zola et al., 2000; Nemanic et al., 2004)or disruption of the hippocampus through fornix lesions(Gaffan, 1974; Mahut et al., 1982; Zola-Morgan et al., 1989b)elicited comparatively mild deficits, leading to a debate aboutwhether the hippocampus makes any contribution to perform-ance on the DNMS task (Baxter and Murray, 2001b; Zola andSquire, 2001). Consistent with the results in monkeys, numer-ous studies have shown that object recognition memory is rela-tively spared in rats with selective hippocampal damage(Mumby, 2001). The results also converged with a review ofhuman neuropsychology studies indicating that patients withrelatively selective hippocampal damage show relatively mildrecognition deficits, despite significant impairments in recall

(Aggleton and Shaw, 1996). That said, there has been sufficientdisagreement amongst studies to fuel a debate about whetherthe hippocampus makes any contribution to performance onitem recognition memory tasks (Aggleton and Shaw, 1996;Manns and Squire, 1999; Baxter and Murray, 2001a; Zola andSquire, 2001).

TWO COMPONENT MODELS OF THE MTL

Although the role of the hippocampus in object recognitionmemory was and is still controversial, most of the availablelesion evidence converges to suggest that, at least in rodents andmonkeys, the PRc makes a more essential contribution thandoes the hippocampus. Accordingly, a number of ‘‘two-compo-nent’’ models have been proposed to capture the functional dif-ferences between the hippocampus and surrounding corticalregions (particularly the PRc). For instance, the ‘‘relationalmemory theory’’ proposed by Cohen et al. (Eichenbaum et al.,1992; Cohen and Eichenbaum, 1993; Eichenbaum and Cohen,2001), proposed that: (1) the PRc and PHc (collectively termedthe ‘‘parahippocampal region’’) encode specific constituent ele-ments of an event, or ‘‘items’’ and (2) the hippocampus encodesrepresentations of the relationships of the items. Thus, the rela-tional memory theory emphasizes the different kinds of infor-mation incorporated in representations formed in the hippo-campus versus the parahippocampal region.

Another approach has been to distinguish between theprocesses used by the hippocampus and parahippocampalregion. For example, the ‘‘complementary learning systems’’(CLS) approach (McClelland et al., 1995; O’Reilly andRudy, 2001; Norman and O’Reilly, 2003), which builds onthe computational approaches of Marr (Marr, 1971) andRolls and Treves (Treves and Rolls, 1994; Rolls, 1996; Rollsand Kesner, 2006), emphasizes that the hippocampus differsfrom neocortical regions in terms of the computational mech-anisms of encoding and retrieval. The CLS family of modelsessentially proposes that the hippocampus is specialized torapidly encode new information, with sparse, minimally over-lapping representations that are well suited to encode specificepisodes. In contrast, the parahippocampal region (and neo-cortex in general) is proposed to represent stimuli and eventsin terms of their constituent features, such that stimuliwith overlapping representations have overlapping neuralrepresentations.

A third idea, proposed by Aggleton and Brown (1999; seealso contributions by Brown et al. and Yonelinas et al. in thisissue) is that the PRc is sufficient to support item recognitionbased on familiarity, and that the hippocampus supports recallor recognition based on conscious recollection. Thus, theAggleton and Brown model emphasized that the hippocampusand PRc differentially support different subjective experiences.In addition, when viewed in the context of dual process modelsof recognition memory (Yonelinas, 2002), their model empha-sizes the roles of different roles of the hippocampus and PRc in

iIt is believed that the postrhinal cortex is the rodent homologue ofthe PHc.iiOther regions not typically included in the MTL memory system,such as the retrosplenial cortex, septal nuclei, and basal forebrain arealso likely to play a role in episodic memory.

1264 RANGANATH

Hippocampus

terms of the processes that contribute to performance on differ-ent memory tasks.

The three models described above differ in terms of theirareas of emphasis. Whereas the relational memory theoryemphasizes that the hippocampus and parahippocampal regiondiffer in terms of the kinds of information that they receiveand associate, CLS emphasizes that they differ in terms of theirneurocomputational properties (see Moses and Ryan, 2006, fora more detailed comparison of the two theories), and Aggletonand Brown’s model emphasizes both processes and consciousexperience. That said, the three models are compatible witheach other in terms of their predictions about the roles of dif-ferent MTL subregions in recognition memory. The relationaltheory (Konkel and Cohen, 2009) would propose that the re-covery of relational information (e.g., recollection of the placeand time that an item was last encountered), should depend onthe hippocampus, but that the PRc should be sufficient to sup-port item recognition without relational information (i.e., fa-miliarity-based recognition). Aggleton and Brown’s (1999)model makes essentially the same prediction. The CLS modelmakes similar predictions, but emphasizes that the process dis-tinction between familiarity and recollection may be better con-ceived in terms of the degree of overlap in recognition memorysignals elicited by studied items and unstudied lures (O’Reillyand Norman, 2002; Norman and O’Reilly, 2003; Elfmanet al., 2008). Under typical circumstances, the cortical compo-nent of the model produces a signal that indexes the globalmatch between a cue and all previously learned information, asa consequence of broad, overlapping stimulus representations.Thus, the distributions of cortical recognition strength for oldand new stimuli are separate but overlapping, much like twoGaussian curves. In contrast, the hippocampal component ofthe model can produce a somewhat bimodal signal distribution,such that some proportion of old items are indistinguishablefrom new items and others have very strong responses that areeasily distinguishable. This response characteristic is due to thesparse, highly differentiated representations formed in the hip-pocampus. Viewed from the perspective of dual process models(Yonelinas, 2001, 2002; Yonelinas and Parks, 2007, see alsoYonelinas et al., this issue) which assume that recollectionoccurs primarily for previously encountered items and that fa-miliarity is more of a signal detection process (i.e., evenunstudied items can elicit a great deal of familiarity), then theCLS model would suggest that the hippocampus will typicallybe more involved in recollection and the parahippocampalregion should be more involved in familiarity (see Norman,this issue, for a fuller discussion of this topic).

THE BINDING OF ITEMS ANDCONTEXTS MODEL

Based on an extensive review of behavioral, neuropsychology,neurophysiology, and neuroimaging studies in rats, monkeys,and humans, Howard Eichenbaum, Andy Yonelinas, and I

(Eichenbaum et al., 2007) proposed the Binding of Items andContexts (BIC) model to explain the roles of different MTLsubregions in episodic memory (see also Diana et al., 2007;Ranganath, 2010). The BIC model builds upon the MTLmemory system framework originally outlined by Squire(1991), in that it proposes differing roles for the PRc and PHcin memory based on their differential connectivity (Fig. 1).Specifically, the PRc primarily receives input from neocorticalareas that process information about the qualities of objects(‘‘what’’), such as area TE or their reinforcement history (suchas the amygdala). The PHc not only receives input from areasinvolved in visual object processing, but also from areas thatprocess spatial (‘‘where’’) information, such as the posterior pa-rietal cortex. The PHc sends a strong projection to the PRc,whereas projections from the PRc to the PHc are comparablyweaker (Suzuki and Amaral, 1994). In the rodent brain, thePRc and PHc preferentially project to the lateral and medialentorhinal areas (Witter et al., 2000a; Kerr et al., 2007),respectively, and these inputs converge in the hippocampus.

FIGURE 1. Anatomy of the MTL ‘‘memory system.’’ A: Rela-tive locations of the perirhinal cortex (shown in blue), parahippo-campal cortex (shown in green), and hippocampus (red) are shownon a rendering of a brain with a cutaway to reveal the medial tem-poral lobes. B: A diagram illustrating our current model of howthe perirhinal and parahippocampal cortex, entorhinal cortex(with separate pathways from perirhinal and parahippocampal cor-tex indicated in blue and green, respectively), and hippocampusmay contribute to episodic memory. The anatomical connectionsbetween each region are illustrated with black lines and proposedroles of each region are shown in italic letters. For simplicity, thediagram presents only the most significant anatomical connectionsbetween these regions and omits other anatomically connectedregions that also may play a role in episodic memory formation orretrieval.

MEDIAL TEMPORAL LOBES AND EPISODIC MEMORY 1265

Hippocampus

Extrapolating from these aspects of MTL anatomy, we (Dianaet al., 2007; Eichenbaum et al., 2007; Ranganath, 2010) pro-posed that the PRc may represent information about specificitems (e.g., who and what), the PHc may represent informationabout the context (e.g., where and when) in which these itemswere encountered, and the hippocampus may represent bind-ings between items and context.

The BIC model builds upon Cohen and Eichenbaum’s rela-tional memory theory (Eichenbaum et al., 1992; Cohen andEichenbaum, 1993) in that both propose similar roles for thehippocampus and PRc. Both models also propose that MTLsubregions differ in terms of information content, and thatthese differences will often (but not always) lead to dissocia-tions at the level of tasks, processes, and conscious experiences.Although not directly stated in our original article, BIC is alsocompatible with the CLS model, which emphasizes the role ofthe hippocampus in pattern separation and completion(Norman and O’Reilly, 2003). Where BIC differs from thetwo-component models is that it proposes a role for the PHcin representing context information. BIC therefore predictsfunctional dissociations between the PHc and PRc. Althoughthe rest of this article is focused primarily on the BIC model,it is important to note that other researchers have recentlyadvanced three-component models of the MTL that are similarto BIC in many ways (Montaldi and Mayes, this issue; Eacottand Gaffan, 2005; Davachi, 2006; Bird and Burgess, 2008).iii

LESION EVIDENCE SUGGESTS ADISPROPORTIONATE ROLE FOR THEHIPPOCAMPUS IN RECOLLECTION

As noted earlier, the two-component models described above(Eichenbaum et al., 1992; Aggleton and Brown, 1999; Normanand O’Reilly, 2003; Eichenbaum et al., 2007) suggest that therecollection/familiarity distinction is useful for understandingthe relative roles of the hippocampus and PRc in item recogni-tion, although this distinction is more central to some modelsthan others. This is also the case for the BIC model, whichsuggests that familiarity-based recognition is supported by itemrepresentations in PRc, and that recollection should addition-ally depend on the hippocampus and PHc to recover informa-tion about the corresponding study context (i.e., the experienceof recollection is usually associated with the recovery of contex-tual information). Consistent with this prediction, a number oflabs have reported that patients with hippocampal damage ordysfunction have disproportionate deficits on measures of recol-lection as compared with familiarity, which is relatively spared(Yonelinas et al., 2002, Bastin et al., 2004; Quamme et al.,2004; Turriziani et al., 2004; Aggleton et al., 2005; Holdstocket al., 2005; Brandt et al., 2008; Tsivilis et al., 2008; Peters

et al., 2009; Vann et al., 2009b; Bowles et al., 2010; seeFig. 2), although different results have been reported for agroup of patients that has been studied by researchers from UCSan Diego (Manns et al., 2003; Wais et al., 2006; Jenesonet al., 2010; Kirwan et al., 2010). It is important to note thatone can expect a good deal of variability across studies of am-nesia patients because of necessarily small sample sizes ofpatient groups and correspondingly small sizes of controlgroups. The latter issue is particularly important given the highdegree of anatomical, demographic, and functional variabilityamong ‘‘healthy’’ controls. Studies often claim to study patientswith damage restricted to the hippocampus, but usually thesestudies do not include direct comparisons of MRI data frompatients and the control participants in the study. Given thetypical age of matched controls in amnesia studies, these partic-ipants might also have substantial hippocampal, cortical, orwhite matter atrophy. Another concern is that structural imag-ing methods can significantly underestimate the severity andextent of MTL damage (Rempel-Clower et al., 1996), and fewstudies investigate the integrity of gray and white matter acrossthe whole brain (i.e., areas outside of the MTL) in amnesiapatients. Finally, some sites select patients on functional criteriasuch as amnesic severity (e.g., Squire and Shimamura, 1986)rather than etiology, thereby leading to a bias towards patients

iiiTo my knowledge, these models make similar predictions to thosedescribed here for BIC, although I will not discuss them as such toavoid misrepresenting their viewpoints.

FIGURE 2. Recollection is disproportionately impaired in hy-poxia patients and rats with focal hippocampal damage. A: Resultsfrom Yonelinas et al. (2002). At left, receiver operating characteris-tic (ROC) curves are shown for human patients with likely hippo-campal damage due to mild hypoxia and for matched controls. Atright, mean estimates of familiarity and recollection are shown foreach subject group based on a dual-process model fit to the recog-nition data. B: Results from Fortin et al. (2004). At left, ROCcurves are shown for rats with hippocampal lesions and controlanimals. At right, mean estimates of recollection and familiarityare shown. Note that ROCs are lower and more symmetric foramnesic humans and rats with hippocampal damage than for thecorresponding control subjects.

1266 RANGANATH

Hippocampus

with more severe and global memory deficits than would nor-mally be expected given the etiology and extent of brain damage.

The problems described above, for the most part, do notapply to two recent studies of patients with incidental damageto the fornix during surgical removal of colloid cysts within thethird ventricle (Tsivilis et al., 2008; Vann et al., 2009b). Thestudies capitalized on the fact that fornix lesions in animalseffectively mimic the effects of focal hippocampal lesions, andon the fact that fornix lesions lead to atrophy of the mammi-lary bodies. Using a large sample of patients and several con-verging methods, these researchers found that patients withlikely fornix damage showed significant recollection impair-ments, whereas familiarity was relatively intact. Critically, thepatients with fornix damage and their corresponding controlswere matched for surgical procedure as well as age, IQ, andtime since surgery. These findings are consistent with the ideathat hippocampal damage affects recollection, while sparing fa-miliarity-based item recognition.

Further evidence for a disproportionate role for the hippo-campus in recollection has come from a study of recognitionmemory receiver operating characteristics (ROCs) in rats withhippocampal lesions. ROC curves are constructed by measuringhit and false alarm rates at different levels of response bias.Human recognition memory ROC curves typically are notsymmetric, and this has been assumed by some to reflect the factthat some items are recollected and, therefore, can be confidentlyrecognized even when one adopts a strict response bias (Yonelinas,1994). In a clever experiment, Fortin et al. (2004) calculatedROCs for rats by manipulating response bias in DNMS testswith odor stimuli. As shown in Figure 2, intact rats, like intacthumans, exhibited asymmetric ROCs, whereas ROCs were sym-metrical in rats with hippocampal lesions. The striking conver-gence between these data (Fortin et al., 2004)iv and studies ofROCs in patients with likely hippocampal damage or dysfunctionstrongly support the view that hippocampal lesions impair recol-lection while generally sparing familiarity-based recognition.

FUNCTIONAL IMAGING EVIDENCE FORDISSOCIABLE ROLES OF MTL SUBREGIONS

IN ITEM RECOGNITION

The BIC model not only makes predictions about lesioneffects but it also provides a framework for understanding theactivation of different MTL subregions in relation to the kindsof information that may support recollection- and familiarity-based recognition. Specifically, we assume that familiarity is

related to the ‘‘strength’’ of an item’s representation, and thatfamiliarity should emerge as a byproduct of experience-depend-ent tuning of the representation of an item during encoding(see contribution by Brown et al., this issue). Thus, PRc activ-ity should differ during encoding of items that will be recog-nized primarily on the basis of familiarity relative to items thatwill be missed, and PRc activity during retrieval should be sen-sitive to gradations in familiarity. We also assume that recollec-tion depends on the successful encoding of not only the item,but also of the association between the item and the context(i.e., spatial, temporal, semantic, social, affective, etc.) in whichthe item was encountered, which should depend on the PHc(context) and hippocampus (item-context bindings). During re-trieval, input from PRc to the hippocampus may trigger com-pletion of the activity pattern that occurred during the learningevent and lead to activation of the associated contextual repre-sentations in PHc networks. Finally, output from PHc to neo-cortical regions would elicit the reinstantiation of neocorticalrepresentations of the various aspects of the contextual state atthe time of the original encoding event, thereby leading to rec-ollection (Wheeler et al., 2000; Wheeler and Buckner, 2004;Johnson et al., 2009). Thus, hippocampal and PHc activityduring encoding and retrieval should be higher for items thatare subsequently recollected relative to items that are subse-quently recognized primarily on the basis of familiarity.

Because the BIC model makes predictions about the rela-tionship between MTL activity and recollection and familiarity,I need to digress for a moment to summarize how these contri-butions are typically estimated in event-related functional mag-netic resonance imaging (fMRI) studies of item recognition. Anumber of different measures can be used to disentangle recol-lection and familiarity, including remember/know, recognitionconfidence, or item and source recognition paradigms (seeYonelinas, 2002 for review). Recollection is typically expectedto differ between remember and know responses, source correctand source incorrect responses, and the highest confidencecompared to lower confidence responses. Familiarity is typicallyexpected to differ between recognized, non-recollected items,and non-recognized items (misses) and it is expected to increasemonotonically with increasing recognition confidence.

Wixted (Wixted, 2007, 2009; Wixted and Squire, in press,see also Wixted and Squire, this issue) has incorrectly assertedthat the interpretations of such imaging results hinge on strongassumptions unique to Yonelinas’ (2001 and 2002) dual-processmodel of recognition memory. It should be emphasized that‘‘recollection’’ or ‘‘familiarity’’ contrasts are not expected to beprocess-pure and they do not depend upon strong assumptionsfrom a particular behavioral model of recognition memory. Forinstance, consider FMRI studies using the remember-knowmethod.v In a remember-know recognition test, the participantis instructed that to give a ‘‘remember’’ (‘‘R’’) response if s/heivThese data were also analyzed with an unequal variance signal detec-

tion (UVSD) model of recognition memory. Both the dual processand UVSD models revealed similar values for d 0� 5 1.01 and bothmodels revealed no contribution of the other parameter (R 5 0; Vold�5 1) (Norbert Fortin, personal communication). However, the UVSDmodel does not provide a principled account of why hippocampallesions should eliminate the asymmetry in ROC curves.

vThe same point can be made about studies examining item and sourcememory. That is, substitute ‘‘R’’ for recognized items with correct sourcejudgments and ‘‘K’’ for recognized items with incorrect source judgments,and the same logic holds true.

MEDIAL TEMPORAL LOBES AND EPISODIC MEMORY 1267

Hippocampus

believes an item was studied based on the recollection of con-textual details about the study episode, and to give a Know(‘‘K’’) response if s/he believes that the item was studied basedon strong familiarity (with little or no recollection). It is com-mon to contrast trials associated with ‘‘R’’ responses and thoseassociated with ‘‘K’’ responses, to identify neural correlates ofrecollection. This interpretation does not hinge on assumingthat recollection is a threshold or all-or-none process, nor thatthese measures are process pure. Instead, one need only makethe reasonable assumptions that both recollection and familiar-ity can contribute to recognition, and that participants can tellthe difference between the two experiences. If so, then whenone recollects a reasonable amount of information, s/he willprobably report it as an ‘‘R,’’ and if an item is highly familiarand little or nothing is recollected, s/he will use the ‘‘K’’ response(Dudukovic and Knowlton, 2006; Slotnick, 2010b). Likewise,‘‘K’’ responses and Misses (or correct rejections) are comparedto identify neural correlates of familiarity. This comparisonwould be based on the reasonable assumption that familiarityshould be higher for ‘‘K’’s than for misses. These contrasts areof course relative, not absolute. On average, the R–K differ-ence should be more sensitive to recollection-related activitythan the K-Miss difference should be, and the K-Miss differ-ence should be more sensitive to differences in familiaritythan the R–K difference should be. To the extent that theseassumptions are incorrect (e.g., when familiarity is also muchhigher for R responses than for K responses), then one wouldnot expect to see reliable dissociations between the responsesof different MTL subregions in contrasts targeting activityrelated to recollection and familiarity.vi

A similar point can be made regarding analyses linking brainactivity to item recognition confidence ratings (Ranganathet al., 2003; Yonelinas et al., 2005; Montaldi et al., 2006; Kir-wan et al., 2008). In a 6-point confidence scale, a judgment of‘‘6’’ indicates the highest degree of confidence that an item wasstudied, whereas a judgment of ‘‘5’’ indicates a slightly lowerlevel of confidence. Typically, memory for contextual detailsmuch higher is much higher for items that receive a ‘‘6’’ thanfor items that receive a ‘‘5’’ responses (Slotnick et al., 2000;Yonelinas, 2001; Ranganath et al., 2003; Slotnick and Dodson,2005). Accordingly, it is reasonable to contrast ‘‘6’’ and ‘‘5’’responses to identify a neural correlate of recollection. One canalso look for monotonic changes in activity across confidencelevels 1–5 to identify a correlate of familiarity, under theassumption that familiarity increases across this range and thatchanges in recollection are fairly small. This assumption is rea-sonable because typically there is little subjective or objectiveevidence to suggest that any information is recollected for judg-ments in the 1–4 range (Slotnick et al., 2000; Yonelinas, 2001;Ranganath et al., 2003; Slotnick and Dodson, 2005).

Even though the assumptions described above are reasonable,it is essential to note that every method of measuring memoryhas limits that must be taken into account. For instance,although source memory is sometimes treated as a process-pureand complete measure of recollection (Wais et al., 2008;Mickes et al., 2009; Wais et al., 2010), familiarity can supportaccurate source memory (Yonelinas, 1999; Diana et al., 2008a;Elfman et al., 2008; Diana et al., 2010; Diana et al., submitted)and recollection can occur for items that are associated withincorrect source decisions (Vilberg and Rugg, 2007, 2009).Likewise, results from the remember-know procedure can behighly variable and highly sensitive to minor changes in theinstructions for making ‘‘R’’ versus ‘‘K’’ judgments (Eldridgeet al., 2002; Rotello et al., 2005). Studies that combine sourcememory and remember-know or recognition confidence ratingscan also produce aberrant results if recognition performance is atceiling and subjects must nonetheless distribute their responsesacross the range of confidence levels (e.g., Gold et al., 2006;Wais et al., 2010). In this event, participants can be expected torecollect a great deal of information on the majority of trials,including trials with relatively low confidence ratings. Because ofthese methodological limitations, one can expect results to varysomewhat across studies. Accordingly, it is essential to not focusexcessively on any one imaging study, but rather to determinewhether generally similar results have been observed across fMRIstudies using different measurement techniques, materials, etc.

To address this question, we recently reviewed results fromfMRI studies which contrasted trials as a function of memoryoutcome (Diana et al., 2007). As shown in Figure 3A, imag-ing studies have consistently shown that activity in the hippo-campus and PHc during encoding or retrieval is generallyincreased during processing of items that are recollected (oritems for which recollective strength is high), as compared withrecognized items for which little or nothing is recollected. Fur-thermore, activity in these regions is generally insensitive to gra-dations in the familiarity of an item. In contrast, activity in thePRc is not usually observed in contrasts examining recollectionof items but is often related to familiarity. Thus, imaging resultssuggest different roles for the PRc versus the hippocampus andPHc in item recognition. This pattern is not only evident acrossstudies (see also Davachi, 2006; Wais, 2008; Mitchell and John-son, 2009), but also within studies that found qualitative differ-ences in the response properties of these regions during itemencoding or recognition (Davachi et al., 2003; Ranganath et al.,2003; Daselaar et al., 2006; Kensinger and Schacter, 2006;Montaldi et al., 2006; Kirwan et al., 2008; Staresina and Dava-chi, 2009, see Figs. 3B–D).

ASSOCIATIVE AND SOURCE RECOGNITION:SOME EXCEPTIONS THAT PROVE THE RULE

Earlier, I noted that the BIC model generally suggests thatdifferences in information content captured by the PRc, PHc,and hippocampus may drive dissociations in the roles of these

viWixted and Squire (this issue) have argued that studies might findartifactual differences amongst MTL subregions due to differences inneurovascular coupling amongst these regions. In subsequent sections,I will review evidence which is inconsistent with this argument.

1268 RANGANATH

Hippocampus

regions with respect to particular memory tasks or processes.However, there is no one-to-one mapping between information,processes, and tasks. Thus, under the right circumstances, infor-mation about items that are represented in PRc can supportperformance on associative and source recognition tasks,whether based on recollection or familiarity. For instance, the

model suggests that, when one recollects associations amongitems, activation in both the hippocampus and PRc should beincreased. Consistent with this prediction, PRc activation is of-ten correlated with successful encoding and retrieval of item–item associations (Jackson and Schacter, 2004; Kirwan andStark, 2004; Eldridge et al., 2005; Fenker et al., 2005).

FIGURE 3. Functional MRI evidence that MTL subregionsdifferentially contribute to recognition memory. A: Results from areview of over 20 FMRI studies that linked trial-by-trial variabilityin MTL activity with recognition memory performance (Diana,et al., 2007). Of the studies that reported analyses targeting recol-lection, most reported significant activation in the hippocampus(red) and parahippocampal cortex (‘‘PHc,’’ green). Of the studiesthat reported analyses targeting familiarity-based recognition, themajority reported significant activation in the perirhinal cortex(‘‘PRc,’’ blue). B: Results from Davachi et al. (2003) showing thatactivation in PRc during encoding of words was predictive of sub-sequent item recognition, but PRc activation did not differentiatebetween recognized items that were associated with correct orincorrect source memory decisions. In contrast, hippocampal acti-vation was selectively enhanced for recognized items that wereassociated with correct source decisions, relative to incorrectsource decisions, but was not sensitive to item recognition accu-racy. C: Results from Ranganath et al. (2003) showing that activ-

ity in the hippocampus during encoding was predictive of subse-quent source memory accuracy for recognized words. Activity inthis region and other hippocampal regions was not significantlycorrelated with subsequent item recognition confidence. In con-trast, encoding activity in a region situated in left peri/entorhinalcortex was monotonically related to subsequent familiarity-basedrecognition, as indexed by confidence ratings. Activity in PRc andEC was not, however, correlated with subsequent source memoryaccuracy. D: Results from Montaldi et al. (2006). Activation inbilateral hippocampal regions was specifically increased during re-trieval for items that were subjectively recollected (R), as com-pared with non-recollected items, and was insensitive to gradationsin familiarity strength. In contrast, activation in bilateral PRc dur-ing retrieval monotonically decreased with increasing confidence[note differences across misses and items that elicited ‘‘familiar’’responses (F1–F3)], but it did not differ between highly familiaritems (F3) and items that were subjectively recollected.

MEDIAL TEMPORAL LOBES AND EPISODIC MEMORY 1269

Hippocampus

Findings from the animal lesion literature also suggest animportant role for the PRc in associative memory. For instance,rats (Bunsey and Eichenbaum, 1993) and monkeys (Murrayet al., 1993) with hippocampal lesions can show intact learningof novel associations between items, whereas performance onthese tasks is significantly impaired by PRc lesions. Importantly,recent evidence suggests that associative recognition may bequalitatively different in intact animals and animals with hippo-campal lesions. For instance, Sauvage et al. (2008b) examinedROC curves in rats during recognition of associations betweenodors and a digging media (e.g., wood chips, beads, and sand).Although rats with hippocampal lesions did not show overallrecognition impairments, the basis for their performance wasqualitatively different than that for intact rats. Analysis of rec-ognition ROC curves revealed that rats with intact hippocampirelied on recollection to support associative recognition,whereas rats with hippocampal lesions relied primarily on fa-miliarity. The BIC model can explain this data by assumingthat rats with an intact hippocampus encoded the odor andmedium as two separate items (e.g., cinnamon and wood) thatwere arbitrarily associated, and that when the relationshipbetween the two items was retrieved, they exhibited recollec-tion-like responses. In contrast, rats with hippocampal lesionsmight have been forced to rely on item representations in PRcby encoding each pairing as a single item (e.g., cinnamon-scented wood), such that they made familiarity-based decisionsbased on the match between each test pairing and the previ-ously acquired item representations.

In studies of human verbal memory, it is easy to manipulatethe extent to which two items can be perceived or encoded as asingle item (Graf and Schacter, 1985; Yonelinas et al., 1999;Rhodes and Donaldson, 2007). For instance, an associationbetween ‘‘house’’ and ‘‘boat’’ could be remembered as a single

compound word, ‘‘houseboat’’ (which is a unitized form of thetwo individual words). Familiarity can support recognition ofword pairs that comprise a compound word, and amnesicpatients with hippocampal dysfunction show significantimprovements in memory for compound word associations(Giovanello et al., 2006). Interestingly, even novel, unrelatedpairings (‘‘motor’’ and ‘‘bear’’) can be processed as componentsof a single compound word (‘‘motorbear’’) by providing a noveldefinition (‘‘a mechanized stuffed animal’’). Processing of wordpairs as compound words (Quamme et al., 2007; Rhodes andDonaldson, 2007; Haskins et al., 2008; Parks and Yonelinas,2009) has been shown to increase the ability to subsequentlyrecognize them based on familiarity (i.e., because the pair hasbeen processed as a single item, it is more familiar than pairsthat had not been treated as a single item), and significantlyimproves associative recognition in human patients with recol-lective impairments and likely hippocampal damage (Quammeet al., 2007).

According to BIC (Cohen and Eichenbaum, 1993), if onewere to process an unrelated word pair as a single novel com-pound word, then associative recognition could be supportedthe strength of item representations in PRc. Consistent withthis idea, we (Haskins et al., 2008) have found that PRc activa-tion was increased during encoding of word pairs in the con-text of a definition (thereby encouraging their treatment as asingle novel item or concept), as compared with encoding ofword pairs in the context of a sentence frame (thereby encour-aging them to be treated as separate items). Furthermore, themagnitude of PRc activation during word pair encoding wasmonotonically related to subsequent familiarity at test (Fig. 4).In the same study, we also found that memory for word pairsin the compound condition was disrupted by changing theword order between study and test, as is the case for real com-

FIGURE 4. PRc can support associative recognition based onfamiliarity (Haskins et al., 2008). Activation in the vicinity of theleft PRc was increased when encoding word pairs in a unitizedmanner, as compared to encoding them in the context of a sen-

tence frame (left graph). In addition, PRc activation duringencoding was monotonically related to subsequent associative rec-ognition confidence, suggesting a role for this region infamiliarity.

1270 RANGANATH

Hippocampus

pound words (e.g., ‘‘houseboat’’ vs. ‘‘boathouse’’). Thus, encour-aging participants to process each word pair as a compoundword led them to subsequently recognize the pair as though itwere a single item. These findings, along with the lesion resultsdescribed earlier suggest that the PRc can support associative rec-ognition based on familiarity, particularly if the items to belinked can be treated as a single configuration. The results fromthese studies are important because they support the counterin-tuitive point that item representations formed by the PRc cansupport performance on associative recognition tasks.

Some evidence suggests that the PRc might spontaneouslyform associations between different elements even when this isnot explicitly demanded by a task. For instance, single-unit re-cording studies have reported the existence of object-selective‘‘pair coding’’ neurons in the PRc that preferentially respondedto objects that had previously been paired together (Miyashita,1988; Sakai and Miyashita, 1991; Erickson and Desimone,1999). The responses of pair coding neurons might reflect rep-resentations that are developed over the course of trainingwhich treat the two objects as a single, fused item. Related tothis idea, some findings indicate that the PRc might spontane-ously associate co-occurring items that are from the same do-main (e.g., an association between two faces). For instance,some patients with damage restricted to the hippocampus havebeen shown to perform well at recognition of associationsbetween stimuli from the same domain, despite severe impair-ments in cross-domain associative recognition (Vargha-Khademet al., 1997; Mayes et al., 2004). On the basis of these find-ings, Mayes et al. (2007, see also Montaldi and Mayes, thisissue) recently proposed that the PRc might generally associateitem pairs from the same domain, such that the pairing couldsubsequently recognized based on familiarity.

The studies described above illustrate how the distinctionbetween item and associative recognition tasks can be mislead-ing, as the key determinant of MTL involvement is really thetype of information that is used to support task performance.Another example of this principle comes from studies in whichthe distinction between item and source information has beenmanipulated. In source recognition studies, participants areshown an item and asked to make a decision (usually forcedchoice) about contextual information associated with the item,such as the study task, spatial position, or background colorassociated with an item. As with associative recognition, it isgenerally agreed that source recognition depends more on rec-ollection of contextual information than does item recognition.However, research has also shown that item information cancontribute to source recognition (Diana et al., 2008a, 2010,submitted; Zimmer and Ecker, 2010), and that the involve-ment of different MTL regions in source memory depends on theextent to which item or context information supports performance(Staresina and Davachi, 2006, 2008; Diana et al., 2010).

For example, in one fMRI study, participants were scannedwhile encoding concrete words against a colored background(Staresina and Davachi, 2006). To encourage participants to as-sociate each word with the background color, they wereinstructed to decide whether it would be plausible to see the

item in that color. Following scanning, participants were askedto recall the words that were previously studied. Next, theywere tested on item recognition, and for items judged old, theywere asked to make a source judgment about the associatedbackground color. Consistent with previous results (Davachiet al., 2003; Ranganath et al., 2003), hippocampal activity dur-ing encoding was predictive of subsequent recall performanceand source recognition. A different pattern of activity wasobserved in PRc, where activity was correlated with subsequentsource recognition, but not with recall of the items. Staresinaand Davachi (2006) speculated that, unlike previous studies,participants in this study were encouraged to think of sourceinformation (i.e., the screen color) as an item feature (e.g., ared balloon), rather than an arbitrary contextual association.On the basis of these results, Staresina and Davachi proposedthat PRc can support the encoding of intra-item associations,which reflect ‘‘binding of a concrete item feature (color) withthe item (word/object representation) itself, as opposed to bind-ing of more contextual/abstract information such as the experi-mental task in which the item was encountered.’’ To test thisproposal, Staresina and Davachi (2008) ran a replication studyin which participants completed source memory tests for boththe item-color associations and for the orienting task that wasused to study the item. Consistent with their prediction, theyreplicated the finding that PRc activity was predictive of subse-quent memory for item-color associations, but activation wasnot predictive of subsequent source memory for the orientingtask (unless color information was also successfully retrieved).In contrast, hippocampal activity was predictive of subsequentmemory for all types of associations, and activation increasedwith the amount of information that was retrieved.

The results from Staresina and Davachi (2006 and 2008) areconsistent with the idea that the hippocampus and PRc playdifferent roles in encoding of item and context information,but are limited by the fact that the dissociations they observedwere based on different retrieval tests. It would be preferable toshow that, even while holding the retrieval task constant, onecan see dissociations between different MTL subregions accord-ing to the type of information one is using to make the deci-sion. Rachel Diana, Andy Yonelinas, and I therefore conductedan experiment which examined the neural correlates of sourcememory retrieval, while manipulating the extent to which par-ticipants made their decisions based on item or context infor-mation (Diana et al., 2010). In this experiment, participantswere asked to learn associations between a word and a back-ground color, either by encoding color as a feature of the itemthat they are encoding (e.g., ‘‘The elephant is red because it issunburned’’) or as a contextual association (e.g., ‘‘The elephantstopped at the red light’’). Our behavioral and event-relatedpotential data suggest that (as in most studies) color sourcememory was supported primarily by recollection if color wasencoded as a contextual association, but source memory was alsosupported by familiarity if color was encoded as an item feature(Diana et al., 2008a, 2010, submitted; see Fig. 5A). In an fMRIstudy using the same paradigm, we observed a qualitative dif-ference in the involvement of different MTL subregions during

MEDIAL TEMPORAL LOBES AND EPISODIC MEMORY 1271

Hippocampus

FIGURE 5. PRc can support source memory for item details.Counter-clockwise from right. A: Results from Diana et al. (2008a).Source (color) memory ROCs were more curvilinear when color wasencoded as an item detail than when it was encoded as a contextualdetail. Note that the ROCs cross over, indicating that this cannot beexplained as an overall change inmemory strength. Instead, the resultssuggest that encoding background color as an item feature increasedthe contribution of familiarity to source recognition. B and C: Resultsfrom Diana et al. (2010). B: Regions showing MTL activation in eachcontrast are plotted simultaneously to reveal areas of overlap across:familiarity/weak memory in the item detail condition (red), recollec-tion/strong memory in the item detail condition (yellow), and recol-lection/strong memory in the context detail condition (blue). Com-mon areas of activation for item detail familiarity/weak memory and

recollection/strong memory occurred in PRc (plotted in orange).Common areas of activation for recollection/strong memory in boththe item detail and context detail conditions occurred in the hippo-campus (plotted in green). C: Parameter estimates extracted from thecommon areas of activation in A (orange and green areas) reveal sig-nificant PRc activation for the item detail conditions only and signifi-cant hippocampal activation only for responses that were associatedwith subjective recollection. D: PRc activation is not related to mem-ory strength. A plot of PRc activation (y-axis) against discriminability(x-axis) for each retrieval condition reveals that the relationshipbetween the two is nonmonotonic. Thus, PRc activation was relatedto the kind of information that supported source decisions, and not tooverall memory strength.

the retrieval test (Diana et al., 2010). Consistent with resultsfrom previous imaging studies, we found that hippocampal andparahippocampal activity was enhanced during color memorydecisions when participants indicated that their decisions werebased on recollected contextual details (Fig. 5C). In the PRc,however, activity was only correlated with successful colormemory if color was encoded as an item feature, and this wastrue for recollection-based responses or familiarity-basedresponses (Figs. 5B,C).

It should be mentioned that the results of Diana et al.(2010) also address a concern raised by Squire et al. about theinterpretation of dissociations between MTL subregions inmemory studies (Squire et al., 2007). Squire et al. speculatedthat dissociations amongst MTL subregions in FMRI studiesmight be due to differences in neurovascular coupling. Specifi-cally, their concern is that differences among MTL areas inresponses to recognition confidence, remember versus knowresponses, etc., could really reflect differences in the couplingbetween BOLD responses and neural responses that signal overallmemory strength (based on both recollection and familiarity). Ifthis is the case, one could make a clear prediction about theresults from the study of Diana et al. (2010), which is that MTLactivation should be monotonically related to discriminability, asopposed to the way items were encoded. As shown in Figure 5D,results were not consistent with this prediction, as BOLDresponses in PRc did not monotonically vary with discriminabil-ity. Instead, PRc activity was related to recollection of item details(i.e., a condition of high memory strength), and to familiarity ofitem details (i.e., a condition of low memory strength), and it wasnot related to recollection of contextual details (i.e., a conditionof high memory strength). Thus, the relationship between PRcactivation and memory strength is nonmonotonic (see also Stare-sina and Davachi, 2008). Indeed, informal inspection of resultsfrom many studies indicates that activity in the hippocampus isalso not monotonically related to memory strength (e.g., see Figs.3B–D, see also Cohn et al., 2009).

The consistent message from imaging studies of associativememory is that MTL subregions do not merely differ in therelationship between overall memory strength and BOLD sig-nal, nor do they not differ solely in terms of their contributionsto item versus associative memory tasks or familiarity versusrecollection processes. Instead, MTL subregions appear to differin a more fundamental manner that is related to the kind ofinformation that is processed by these regions.

THE MTL IS NOT THE SITE OF CONSCIOUSRECOLLECTION

As noted earlier, the BIC model makes predictions aboutrecollection and familiarity based on the assumption that MTLsubregions differ in terms of the types of information theyreceive and process (Diana et al., 2007; Eichenbaum et al.,2007; Ranganath, 2010). This does not mean that the MTL isthe site where conscious mnemonic experiences arise. NealCohen and colleagues (Cohen et al., 1997; Konkel and Cohen,

2009) have made this point more emphatically, proposing thatthe MTL can support recovery of information about the pasteven in the absence of awareness. Although few studies have beenable to successfully disentangle conscious/explicit retrieval fromaccess to item, context, or relational information, results fromthese studies indicate that the MTL may support performanceeven in some tasks that do not require explicit memory. Forexample, the MTL has been implicated in contextual cueing tasks(Chun and Phelps, 2000; Manns and Squire, 2001; Preston andGabrieli, 2008), implicit detection of changes in scenes througheye-movements (Ryan et al., 2000), and conceptual priming(Voss et al., 2009; Wang et al., in press). Of course, there arealso many kinds of implicit memory that are spared followingMTL lesions (see Squire et al., 1993 for review), so it is probablymost appropriate to conclude that the MTL can support recoveryof some kinds of information about the past, and this informa-tion does not always correspond to conscious experience.

If this is the case, how do we make sense of the role of thehippocampus in conscious experiences like recollection?Although the availability of contextual information is generallya prerequisite for recollection, recollective experience ultimatelyis the outcome of a constructive process by which recovered in-formation is used to make an attribution about the past (Bart-lett, 1932; Johnson and Raye, 1981; Johnson et al., 1993;Schacter et al., 1998). Thus, recollection should involve notonly the MTL regions that support access to item and contextinformation but also neocortical areas, such as prefrontal andparietal regions, which may be required to construct consciousattributions about this information (Moscovitch, 1992, 1995,2008). It follows that attributions about the past will notalways correspond to the information that is output from theMTL (Cabeza et al., 2001; Preston and Gabrieli, 2008; Han-nula and Ranganath, 2009; Slotnick, 2010a).

Debbie Hannula and I found evidence for this idea in arecent fMRI study in which we used eye movements to indi-rectly measure the expression of relational memory, separatefrom participants’ explicit attributions (Hannula and Ranga-nath, 2009). Participants were asked to learn a series of scene–face pairs and were tested on these pairings while their eyemovements were monitored (Fig. 6A). On each test trial, astudied scene was shown, and after a delay, a test display con-sisting of the scene and three previously studied faces was pre-sented. Consistent with previous studies using a similar para-digm (Hannula et al., 2007), participants tended to dispropor-tionately look at the face that was previously paired with thescene, suggesting that eye movements were influenced by thepreviously learned face–scene association. Activity in the hippo-campus and PRc during initial presentation of the scene waspredictive of the extent to which participants subsequentlyviewed the correct face, even when they failed to explicitly rec-ognize it as the associate (Fig. 6B). Interestingly, activity in thelateral prefrontal cortex was more closely correlated withresponse accuracy than it was with eye movement measures ofrelational memory, and functional connectivity between theprefrontal cortex and hippocampus was higher on correct trialsthan on incorrect trials.

MEDIAL TEMPORAL LOBES AND EPISODIC MEMORY 1273

Hippocampus

FIGURE 6. Hippocampal activity predicts expression of rela-tional memory through eye movements, even when recollectionfails (Hannula and Ranganath, 2009). A: Participants studied a se-ries of faces, each superimposed on a scene context. On each testtrial, a previously studied scene cue was shown, and after a delay,three studied faces were superimposed, and participants wereinstructed to select the associated face. B: Hippocampal activationin response to the scene cue was significantly higher on trials forwhich participants disproportionately fixated on the associatedface (‘‘Disproportionate Match’’) than on trials for which partici-pants spent more time viewing another face (‘‘DisproportionateMismatch’’). Furthermore, even on trials for which explicit mem-

ory decisions were incorrect, hippocampal activity was correlatedwith the amount of time spent viewing the correct face. C: Activa-tion in dorsolateral (‘‘DLPFC’’) and ventrolateral (‘‘VLPFC’’) pre-frontal regions during the scene cue was increased on correct rela-tive to incorrect trials (left). Furthermore, during presentation ofthe test display, the correlation between activity in the DLPFCand hippocampus was increased during correct trials, relative toincorrect trials (right). The results suggest that hippocampal activ-ity is related to the recovery of relational memory even when con-scious recollection fails. Furthermore, recollection may involve abroader cortical network which includes regions in the PFC.

The findings are consistent with the BIC model—when par-ticipants successfully recovered items in response to a contextcue (i.e., the studied scene), activation was increased in the hip-pocampus (corresponding to the activated item-context associa-tion) and PRc (corresponding to the reactivated item representa-tion). The data go even further, however, in that they demon-strate that output from the hippocampus does not alwayscorrespond to conscious experience. The results suggest that hip-pocampal activity only reflects the recovery of relational informa-tion, and that accurate conscious attributions about this informa-tion (i.e., recollection) may require interactions between the hip-pocampus and other brain areas such as the PFC (see below).

The idea that conscious recollection depends on corticalregions that receive hippocampal input is not just an academicissue. There is now a growing appreciation of the fact that dys-function of areas in the frontal and parietal cortices might play arole in memory dysfunction that occurs in ‘‘normal’’ aging (Nor-dahl et al., 2006; Andrews-Hanna et al., 2007), dementing dis-orders (Nordahl et al., 2005; Buckner et al., 2008), and psychi-atric conditions like schizophrenia (Ranganath et al., 2008,2009). For instance, we have shown that, even among cogni-tively typical elderly individuals, age-related changes in whitematter integrity are associated with reduced ability to recruit pre-frontal regions during episodic and working memory tasks (Nor-dahl et al., 2006). Furthermore, white matter deterioration caneventually lead to memory impairments that are comparable tothat seen in preclinical stages of Alzheimer’s disease, even in theabsence of gross hippocampal atrophy (Nordahl et al., 2005). Itis possible that these memory deficits might be addressedthrough development of pharmaceutical (McDowell et al., 1998;Minzenberg and Carter, 2008) and/or behavioral (Nyberg et al.,2003; Olesen et al., 2004; Davidson et al., 2006; McNab et al.,2009) interventions to improve frontal and parietal functioning.

BEYOND LONG-TERM MEMORY: MTLCONTRIBUTIONS TO SHORT-TERM MEMORY

AND PERCEPTION

The work reviewed in the previous sections focused on therole of the MTL in long-term memory. However, a representa-tional view of MTL organization implies that the involvementof the MTL in a memory task is not dependent on the lengthof the retention interval. Instead, the issue is whether a particu-lar task requires access to information that is disproportionatelysupported by MTL subregions (for a similar perspective, seeCowell et al., this issue). If so, one should be able to find evi-dence of MTL involvement in retention of information acrossbrief delays of seconds, or possibly even in perception.Although this idea would have seemed ridiculous to neuro-scientists 15 years ago, there is now substantial evidence to sug-gest that the MTL may play an important and previouslyunappreciated role in short-term memory (Ranganath and Blu-menfeld, 2005; Hasselmo and Stern, 2006; Jonides et al.,2008; Ranganath, 2009) and perception (Murray et al., 2007;Baxter, 2009; Graham et al., 2010).

Initial hints of MTL involvement in short-term memorycame from studies in monkeys showing MTL activation duringshort-term memory tasks (Sybirska et al., 2000; Davachi andGoldman-Rakic, 2001) and that cooling of the MTL (Horeland Pytko, 1982; Horel et al., 1984, 1987; Horel, 1994) orextensive MTL lesions (Murray and Mishkin, 1986; Suzukiet al., 1993; Zola-Morgan et al., 1993; Meunier et al., 1996)could impair object recognition memory even across delays of afew seconds (Ringo, 1988, 1991). Some findings from humanamnesics also indicated potential involvement of the MTL inshort-term memory (Aggleton et al., 1992; Holdstock et al.,1995; Owen et al., 1995, 1996; Buffalo et al., 1998; Holdstocket al., 2000; Kesner and Hopkins, 2001), but these findingswere largely ignored, possibly because of the widespread knowl-edge that many aspects of short-term memory are spared fol-lowing MTL damage (Cave and Squire, 1992).

Interest in the potential role of the human MTL in short-term memory was renewed, however, following the publicationof two fMRI studies in 2001. The first, by Stern and colleagues(2001), investigated the neural correlates of WM for novelscenes. Activity was examined while participants performed a‘‘2-back’’ task that required them to decide whether each scenematched the scene that was presented two trials ago. Activity inthe hippocampus was increased during blocks with novelscenes, as compared to blocks with scenes that were highly fa-miliar. Stern et al. also examined activation in the hippocampusduring performance of a target detection task that placed mini-mal demands on WM maintenance. Hippocampal activationdid not differentiate between blocks of target detection withnovel stimuli and blocks with highly familiar stimuli. Thisfinding suggests that hippocampal activation during the WMtask was not driven by passive processing of novel stimuli, butrather by the demand to actively maintain these stimuli.

In a separate study, I (Ranganath and D’Esposito, 2001)used event-related FMRI to investigate the neural correlates ofWM for novel faces. In the first experiment, participants per-formed a delayed recognition task, which required activemaintenace of a face across a 7s delay. The faces presented oneach trial were novel and not repeated on subsequent trials.We found that hippocampal activation was significantlyincreased during the memory delay, as might be expected ifthis region was involved in maintenance of each novel face.vii

In a second experiment, we replicated our previous finding ofhippocampal activation during maintenance of novel faces in a

viiThe finding of hippocampal activation during maintenance of facesmight seem to be inconsistent with the idea that the hippocampus rep-resents items-in-context. However, our study showed that hippocampalactivity was elevated during the delay, but not during recognition deci-sions. It is possible that, on each working memory trial, participantsrepeatedly recalled the previously seen face during the delay. Thus, hip-pocampal activation in this working memory task might have reflectedpersistent reactivation of the recent encoding episode, in a similar man-ner to what might be seen during ‘‘long-term memory’’ tasks such asfree recall. Admittedly, this is a post hoc explanation; however, later inthis section I summarize findings from studies which more directly testthe predictions of the BIC model.

MEDIAL TEMPORAL LOBES AND EPISODIC MEMORY 1275

Hippocampus

new sample of participants, and extended it by demonstrating thatdelay period activation in the same region was increased duringmaintenance of novel faces, as compared with familiar faces.

The findings from the studies by Ranganath and D’Esposito(2001) and Stern et al. (2001) prompted several groups to re-examine the question of whether short-term memory is intactfollowing MTL damage. For example, one group (Nicholset al., 2006) ran an FMRI study aimed at replicating our previ-ous findings (Ranganath and D’Esposito, 2001) and then testedamnesic patients using essentially the same paradigm. Consist-ent with our study, they also found that activity in the hippo-campus was increased during maintenance of novel faces. Inaddition, amnesic participants with medial temporal lobe dam-age due to anoxia or encephalitis were impaired at performingthe task at a 7 s delay. Similar studies have also reportedimpaired short-term memory for novel faces in patients withMTL damage (Olson et al., 2006a; Shrager et al., 2008). Morerecently, an intracranial electroencephalography study showedthat the amplitude of slow potentials and gamma oscillations inMTL regions during maintenance of novel faces is modulatedby memory load (Axmacher et al., 2008).

The results described above provide strong evidence forMTL involvement in short-term memory, but they do not indi-cate the conditions under which the MTL will make a criticalcontribution. It is well known that many kinds of information,such as sequences of letters or digits, can be maintained acrossshort delays even with severe MTL damage, so the MTL is notalways necessary for short-term memory. Overlearned or simplestimuli such as digits and letters may be well represented inother neocortical areas, so this information can be temporarilyaccessed without necessitating MTL involvement. Short-termretention of complex, novel stimuli such as scenes or faces, how-ever, might necessitate activation of item representations in thePRc, context representations in PHc, or recall of the entire studyevent via activation of hippocampal representations.

The BIC framework predicts that the PRc should be involvedin short-term memory for objects, and particularly for novelobjects that are not well-represented in earlier visual areas (e.g.,inferior temporal cortex). Consistent with that idea, results fromcontrolled lesion studies in monkeys suggest that MTL lesionsthat include the PRc can cause severe STM deficits for novelobjects (Mahut et al., 1982; Murray and Mishkin, 1984; Zola-Morgan and Squire, 1985; Murray and Mishkin, 1986; Zola-Morgan and Squire, 1986; Zola-Morgan et al., 1989a; Meunieret al., 1993; Zola-Morgan et al., 1993; Eacott et al., 1994), evenwhen damage to adjacent areas is minimal. Interestingly, one ofthese studies even reported impaired performance with a zerosecond retention delay (Eacott et al., 1994). Results from single-unit recording studies of monkeys (Miyashita and Chang, 1988;Miller et al., 1993; Nakamura and Kubota, 1995; Suzuki et al.,1997) complement the lesion evidence by demonstrating thatperirhinal and entorhinal neurons exhibit persistent, object-selec-tive activity during retention delays in STM tasks—a putativeneural mechanism for active maintenance.

The role of the hippocampus in short-term memory may bemore restricted to situations that require representations of rela-

tionships between items and/or their context.viii Consistentwith this idea, patients with hippocampal amnesia have beenshown to exhibit impaired short-term memory for object-loca-tion associations (Ryan and Cohen, 2004; Hannula et al.,2006; Olson et al., 2006b; Shrager et al., 2008), spatial layouts(Hartley et al., 2007), and arbitrary face–scene relationships(Hannula et al., 2006). Converging evidence has come fromfMRI studies showing hippocampal involvement in short-termmemory for object–location associations (Mitchell et al., 2000;Piekema et al., 2006; Hannula and Ranganath, 2008). In onesuch study, participants were scanned while performing a testthat required short-term memory for the locations of objectson a 3 3 3 grid. On each trial, participants saw a 3d renderingof four objects on the grid (see Fig. 7A). After an 11 seconddelay, they were shown a test grid (‘‘probe’’) and asked todecide whether the objects were in the same locations, if one ofthe objects changed position (1 object–location bindingchanged), or if the positions of two of the objects wereswitched (2 object–location bindings changed). Critically, oneach trial, the test grid was rotated 908 from the original view-point, so participants could not base their decisions on overallvisual similarity. Results (Fig. 7B) showed that activity in thehippocampus was predictive of accuracy on this working mem-ory task. Furthermore, hippocampal activity during correct tri-als was parametrically related to the number of object–locationbindings that were preserved in the test probe. Hippocampalactivity was highest for matching test probes, less so for probesin which 1 object–location binding was changed, and smallestfor probes in which 2 object–location bindings were changed.Because both matching and nonmatching probe displays werevisually dissimilar to the display that had been encoded, thefindings suggest that hippocampal activity must have been sen-sitive to the relational match between the probe display andwhat had previously been encoded. Thus, hippocampal activityduring short-term memory retrieval,ix as in long-term memory,appears to reflect the formation and reactivation of item-con-text bindings.

viiiMuch as the hippocampus can support long-term item recognitionbased on recollection, the hippocampus could also support short-termmemory for items based on recollection and also by active maintenanceof a representation of the encoding episode.ixIt may be worth noting that these effects happened during the retrievalphase of each trial, rather than during maintenance. This brings up somethorny semantic issues. Some people feel that ‘‘short-term’’ or ‘‘workingmemory’’ should only refer to ‘‘activity-based’’ memory that persists acrossmemory delays rather than to ‘‘weight-based’’ changes in synaptic strengththat would produce phasic activity changes during retrieval. This view iscompletely arbitrary. There are many forms of weight-based changes insynaptic strength that have a very short time scale, and would be well-suited to support short-term memory under conditions of high interference(i.e., during ‘‘working memory tasks’’). In any case, such debates aboutsemantics miss the larger point. Persistent activity and synaptic plasticityare phenomena that can be seen in any cortical area, including the areas inthe MTL (Ranganath and Blumenfeld, 2005; Postle, 2006). Thus, there isno reason to believe that the MTL solely supports weight-based changes inmemory.

1276 RANGANATH

Hippocampus

Relevant to the findings described above, recent work indi-cates a role for the hippocampus in online construction of rep-resentations of scenes or episodes (Burgess et al., 2001; Bucknerand Carroll, 2007; Hassabis et al., 2007a; Schacter et al.,2007). For instance, patients with hippocampal amnesia werefound to be impaired at constructing imagined experiences inresponses to verbal cues (Hassabis et al., 2007b). Furthermore,hippocampal activation is increased during tasks that requireimagination of hypothetical future events (Addis et al., 2007,2009). Summarizing this work, Schacter and Addis (2009) pro-posed that the hippocampus plays a critical role in constructingsimulations of hypothetical episodes as well as in reconstructingelements of past episodes. They speculated that the involvementof the hippocampus in episodic simulation may emerge fromits more basic role in the representation of arbitrary associationsbetween the elements of an event. Considered from the per-spective of the BIC framework, episodic simulation involves

maintaining a representation of the elements of an event andalso placing them within a hypothetical context. Accordingly,the hippocampus, through its role in binding item and contextinformation, may be involved in the active maintenance of hy-pothetical events, a process which is extensively and routinelyused to plan future behavior.

If short-term memory is dependent on the MTL, it is reasona-ble to suspect that the MTL also supports perception. Forinstance, when viewing a complex scene, people typically movetheir eyes to sample different regions of the scene, and theacquired information must be integrated into a coherent on-linerepresentation. Thus, every time one makes an eye movement tosample a different part of a scene, representations of objects andtheir spatial context must be accessed and updated to provide astable perceptual experience (e.g., Hollingworth et al., 2001), andperception in these situations probably relies on the MTL. Evenin situations where trans-saccadic integration is not required, the

FIGURE 7. Hippocampal activation predicts accurate short-term memory for spatial relations (Hannula and Ranganath,2008). A: Example stimuli and relative timing of events on eachtrial. In this task, participants maintained the locations of fourobjects in a grid, and were tested with a rendering of the grid thatwas rotated 908 clockwise. A ‘‘match’’ probe is shown in this exam-ple. B: Three examples of probe displays are shown. C: Hippocam-

pal activation during retrieval decisions was graded as a functionof display type; activity was greatest for correctly identified matchdisplays, followed by correctly rejected mismatch-position (inwhich one item-location binding changed), correctly rejected mis-match-swap (in which two item-location bindings changed), andfinally incorrect trials.

MEDIAL TEMPORAL LOBES AND EPISODIC MEMORY 1277

Hippocampus

MTL might still contribute. This is because, anatomically speak-ing, regions in the MTL are in a position to encode more inte-grative representations of the world, as compared with unimodalcortical regions. For instance, whereas extrastriate regions mayrepresent visual objects in a viewpoint dependent manner (Grill-Spector and Malach, 2001), the PRc might form viewpoint-inde-pendent object representations which also incorporate semanticknowledge (Taylor et al., 2006) and reward value (Buckley andGaffan, 1998a; Liu and Richmond, 2000; Liu et al., 2004).Thus, representations in the PRc could be more abstract thanthose in extrastriate cortex, while at the same time incorporatinginformation which could differentiate stimuli whose low-level vis-ual features are highly similar (Bussey and Saksida, 2002). Con-sistent with this idea, PRc damage has generally been shown toimpair performance on perceptual tasks that require highly differ-entiated object representations (Buckley and Gaffan, 1997; Buck-ley et al., 1997; Buckley and Gaffan, 1998b,c; Bussey et al.,2002; Barense et al., 2005; Lee et al., 2005a,b,c, 2006a; Barenseet al., 2007) (but see also Buffalo et al., 2000; Stark and Squire,2000), and PRc activation is correlated with performance onsuch tasks (Lee et al., 2006b; O’Neil et al., 2009). Less is knownabout the role of the hippocampus and PHc in perception, butsome evidence from human studies implicates these regions inscene perception (Aguirre et al., 1998; Epstein and Kanwisher,1998; Epstein et al., 1999; Lee et al., 2005b,c), which is consist-ent with the roles of these regions in representing context anditems-in-context respectively. Further work (Bar and Aminoff,2003) indicates that the PHc might also contribute to the percep-tion of objects that have a well-defined semantic context (i.e.,objects that tend to be seen only in certain kinds of situations,like a microscope). Thus, the PHc might play a role in percep-tion and memory that extends beyond scenes or spatial layouts(Diana et al., 2008b; Preston et al., 2010). I will return to thisissue later in the article.

THINKING ABOUT THE MTL IN THE CONTEXTOF EPISODIC MEMORY

The most significant innovation of the BIC framework is itsemphasis on context as a key component of episodic memory,and on the difference between items and contexts. The distinc-tion between items and contexts cannot be easily distilled into asimple dichotomy,x but it can be operationalized. Unfortunately,

neither our previous descriptions of the BIC model nor compet-ing models of MTL organization have made it clear how to doso. In human behavioral studies, context has been operational-ized in a number of ways, including not only spatial locationbut also less tangible variables such as cognitive, emotional, andphysical state. Reinstantiating any of these contextual variablescan facilitate memory for events that occurred in a similar con-text.xi For instance, recall is facilitated if one is in the sameroom (Smith, 1979), in the same emotional state (Bower,1981), or in the same intoxicated state (Goodwin et al., 1969)during study and test. These findings suggest that when learninga list of items, people naturally tend to associate each item witha particular contextual state, which explains why reinstantiatingthe study context facilitates recall of the associated items. Thus,context, both spatial and nonspatial, is incidentally encoded asan integral part of any episodic memory (Tulving, 1985).

Based on the behavioral research described above (and somespeculation), we can arrive at three principles to define contex-tual variables (see Table 1). One principle is that contextualvariables correspond to states or experiences that extend acrosstime and/or space. More precisely, context may disproportion-ately consist of information acquired at low spatial and tempo-ral frequencies. For instance, when viewing a complex scene,

TABLE 1.

Distinctions Between Items and Contexts in Practice and in Principle

Item Context

Typical examples

used in

experiments

Objects, faces,

words

Spatial locations, scenes,xii

orienting tasks, background

colors, voices, emotional/

physical state,

semantic ‘‘gist’’

Spatial/Temporal

Frequencies

High Low

Differentiation High Low

Attentional status Foreground, focus

of processing

Background

xThe same could be said about any theoretically meaningful distinctionin psychology or neuroscience. For instance, even the distinctionbetween objects and spatial relations can be difficult to define. Con-sider a swarm of bees. Intuitively, the bees are different objects residingin different spatial locations. But if the bees are at a distance andmove in perfect coherence, a viewer might treat the swarm as a singleobject. In other words, it is not trivial to distinguish between an object(which is a spatial configuration of features) and the spatial relationsbetween a series of objects. Of course, one can come up with princi-ples to make such a distinction, just as I am proposing principles todistinguish between items and contexts.

xiContext-dependent retrieval effects are most pronounced in recalltasks, in which one must reconstruct the past context in order to gen-erate previously studied items. In contrast, because the items aredirectly presented in recognition tests, reinstating a past context doesnot help recognition as much as recall.xiiIn many studies, it is implicitly assumed that a scene is a spatial con-text. Scenes usually have a strong contextual element that is difficult toignore, but the information that is encoded can sometimes dependconsiderably on the content of the scene and how it is processed. Forinstance, a picture of an empty room is likely to be viewed as a con-text, whereas a picture of four patrons sitting at a bar might be experi-enced as four items in a context, and a picture of the Statue of Libertyset against the New York skyline might be viewed as an item (a pictureof a statue).

1278 RANGANATH

Hippocampus

context can be apprehended from the information carried inlow spatial frequencies (Bar, 2004). When exploring a novelenvironment, context is acquired by gradually integratingknowledge of one’s current movements and visual cues with in-formation in the recent past. When reading, a semantic context(or ‘‘gist’’) is gradually built up over the course of reading sev-eral sentences. A second principle is that items are differenti-ated from one another, whereas context is an integrated frame-work in which items may appear. Thus, one can simultaneouslyprocess two items in the same context, but one cannot processsimultaneously process one item in multiple contexts.xiii Thethird principle is that contextual information serves as thebackground for current processing. That is, during the experi-ence of an event, some information will be brought to the fore-ground for processing while other information will be relegatedto the background. Attentional selection probably serves todynamically bind foregrounded information into discrete itemrepresentations (Treisman and Kanwisher, 1998), whereas back-grounded information may contribute to what is experienced asthe context. For example, in the Diana et al. (2010) experi-ment, the color of the background screen that was associatedwith each word might normally be processed as a contextualbackdrop, but the color could be encoded as an item feature ifparticipants were oriented to process the color and item in abound manner.

The BIC model suggests a role for the PHc in context repre-sentation based on research suggesting that the PHc is criticalfor remembering spatial contexts (Aguirre and D’Esposito,1999; Bohbot et al., 2000; Nemanic et al., 2004), and thatPHc activation is also often correlated with successful recoveryof nonspatial contextual information (such as source memoryfor the task that was used to encode a word) (Davachi et al.,2003; Ranganath et al., 2003). Involvement of the PHc insource memory (such as the task that was used to encode aword) is consistent with a role in memory for ‘‘cognitivecontext,’’ including task set and goal states (Cohen andServan-Schreiber, 1992). This information could be conveyedto PHc via substantial inputs from the dorsolateral prefrontalcortex (Petrides and Pandya, 1999). PHc activation is alsosensitive to certain kinds of semantic context, such as informa-tion about the kinds of situations in which a particular kindof object might be encountered (Bar and Aminoff, 2003;Aminoff et al., 2007, 2008; Bar et al., 2008). In addition, PHcactivation is higher for objects that have a history of repeatedco-occurrence with other objects than for objects that do not(Aminoff et al., 2007).

Although there is evidence that the PHc may support encod-ing and retrieval of both spatial and nonspatial context varia-bles, this hypothesis needs to be more rigorously tested. Theprinciples described in Table 1 provide a starting point for test-ing a number of hypotheses. For instance, one could expectthat different patterns of MTL activation might be seen duringrecollection, depending on whether an item cue elicits success-ful activation of the corresponding context or whether a con-text cue is used to elicit successful activation of an item.Another testable prediction is that, during processing of a newitem, PHc activity could be modulated as a function of the fa-miliarity of the context. One could also use decoding or classi-fication techniques on FMRI or intracranial electroencephalog-raphy data to examine the kinds of information incorporatedin PHc representations (e.g., Diana et al., 2008b). I wouldexpect that PHc activity should carry information about spatialand nonspatial context, including the semantic gist of recentlyprocessed items (Sederberg et al., 2008) and the currentlyactive task context (Polyn et al., 2009). I also expect that infor-mation about attentionally backgrounded information wouldbe carried to a greater extent in PHc than in PRc activity.

Substantial evidence also suggests that the hippocampus maybe critical for remembering contexts (O’keefe and Nadel, 1978;Smith and Mizumori, 2006; Polyn and Kahana, 2008). Thisidea is not at odds with the BIC framework, which proposesthat the hippocampus receives and integrates information aboutcontext from the PHc, but it begs the question of how contex-tual representation differs between the two regions. To extendthe framework I have presented so far, I propose that the PHcmay be encoding different contextual dimensions separately(e.g., space, task set, semantic gist), and that the EC and hippo-campus serve to integrate these different dimensions to arrive ata more precise, multi-dimensional representation of current con-text with respect to space, time, and other nonspatial contextualvariables (along with item information conveyed by PRc). Thisidea is consistent with evidence from a number of studies of spa-tial navigation. For instance, the PHc has been shown to encodespatial context in a relatively global sense, such as the room thatsomeone is in (Hassabis et al., 2009) or general position relativeto a salient landmark (Aguirre et al., 1998; Ekstrom et al.,2003; Janzen and van Turennout, 2004; Ekstrom and Book-heimer, 2007; Janzen et al., 2007), and ‘‘grid cells’’ in the medialEC (which integrates inputs from the PHcxiv) provide a system-atic map of spatial contexts (Fyhn et al., 2004). In contrast, thehippocampus appears to specify position within a particular spa-tial context (O’Keefe and Dostrovsky, 1971; Ekstrom et al.,2003; Hassabis et al., 2009) or temporal sequence (Ekstrom andBookheimer, 2007; Manns et al., 2007; Pastalkova et al., 2008;

xiiiOne can experience different contextual dimensions at the sametime, but one cannot experience different points along a single contextualdimension at the same time. For instance one can be happy (emotionalcontext) while sitting in the backyard (spatial context). However, mostpeople do not experience multiple emotional states or feel as if they areresiding in multiple locations at the same time (quantum mechanicsnotwithstanding).

xivTechnically, because this is in reference to the rodent brain, it is moreaccurate to say that the dorsomedial EC receives substantial input fromthe postrhinal cortex. In general, researchers treat the postrhinal cortexas comparable to PHc (Amaral, 1999; Burwell, 2000). I therefore choseto stick with PHc to minimize confusion for nonspecialists.

MEDIAL TEMPORAL LOBES AND EPISODIC MEMORY 1279

Hippocampus

Tubridy and Davachi, in press; Jenkins and Ranganath, in press).Hippocampal activity also appears to reflect information aboutitems (Wood et al., 1999; Kreiman et al., 2000; Viskontas et al.,2006) and conjunctions of item, spatial, and nonspatial contextinformation (Wood et al., 2000; Komorowski et al., 2009;Manns and Eichenbaum, 2009). These findings are consistentwith the idea that contextual representation may be more precisein the hippocampus than in the PHc, but more research isneeded to directly test this conjecture.

COMPARISON BETWEEN BIC AND OTHERMODELS IN THIS SPECIAL ISSUE

Because BIC is a framework that attempts to explain MTLfunction across different levels of analysis and researchdomains, the model shares many features with other modelsthat have been proposed. For instance, as noted earlier, BIC,like the two-component models described in this special issue(e.g., Brown et al., this issue; Norman, this issue; Yonelinaset al., this issue), can explain dissociations between the roles ofthe hippocampus and PRc in contextual recollection and famili-arity-based item recognition, and the explanations are largelycompatible with one another. The differences amongst thesetwo-component models lie primarily in the questions they areoriented towards (physiology/lesion studies in animal models vs.computational considerations vs. behavioral dissociations andhuman amnesia research). BIC differs from these models in thatit emphasizes separate roles for the PHc and PRc in episodicmemory. BIC also emphasizes information content as the sourceof differences between MTL subregions, whereas the other mod-els place greater emphasis on process-related differences and/orsubjective mnemonic experiences. In this regard, the BICmodelxv has more in common with the models proposed byMontaldi and Mayes (this issue) and Henson and Gagnepain(this issue). Shimamura (this issue) and Cowell et al. (this issue),propose that different MTL subregions differ in terms of sup-porting relatively weak versus strong memories. This view wouldseem to suggest that PHc and PRc should support weaker mem-

ories and that the strongest memories would recruit the hippo-campus, which is at odds with the BIC model. As noted earlier,PHc activity is often associated with contextual recollection (Di-ana et al., 2007; Eichenbaum et al., 2007), and in general, ac-tivity patterns of different MTL subregions differ in ways thatcannot easily be explained in terms of memory strength (Stare-sina and Davachi, 2006, 2008; Cohn et al., 2009; Diana et al.,2010).

BIC is neutral about some issues that are at the heart ofsome other models. For instance, Cowell et al. and Normanfundamentally disagree about whether the hippocampus andPRc differ more in terms of computational process (patterncompletion vs. global matching) or in terms of informationcontent (objects vs. spatial). Norman’s CLS account is moreconsistent with the dearth of evidence linking the hippocampusto item familiarity (Brown et al., this issue; Yonelinas et al.,this issue) and with the fact that hippocampal activity usuallydoes not covary in a graded manner with recognition confi-dence, as has been shown for PRc (Ranganath et al., 2003;Montaldi et al., 2006; Kirwan et al., 2008). In addition, theCLS model can explain a range of behavioral data (Normanand O’Reilly, 2003; Norman et al., 2005; Elfman et al., 2008;Norman et al., 2008; Migo et al., 2009). Cowell et al.’s modelcan better explain evidence implicating the PRc in perceptualdiscriminations (Bussey et al., 2002) and its emphasis on infor-mation content as a key factor in the functional organization ofthe MTL is more consistent with the perspective presentedhere.

Other papers in the special issue also presented novel ideasthat are potentially compatible with the BIC framework. Forinstance, Henson and Gagnepain emphasize the importance ofpredictive coding in explaining MTL activation across a vari-ety of tasks. Montaldi and Mayes emphasize that the PRc cansupport memory for associations between items from the samedomain, and they distinguish between familiarity for itemsand familiarity for inter-item associations. These ideas areinnovative, and more research will be needed to assess theirviability.

Finally, it is difficult to compare the BIC framework with thecontribution of Wixted and Squire (this issue), because they donot propose a model of MTL organization. Instead, they arguethat: (1) recollection is a continuous process, (2) most fMRIand lesion experiments of recognition memory rely on theassumption that recollection is categorical, and (3) results fromthese studies are uninterpretable because they confound memorystrength with the recollection/familiarity distinction. As notedearlier, dissociations in imaging studies between different MTLsubregions can be interpreted even if one does not believe thatrecollection is a categorical process (e.g., Mitchell and Johnson,2009). Furthermore, activity patterns in MTL subregions donot reliably covary with memory strength, so one cannot explainpreviously reported dissociations in terms of a memory strengthconfound (Staresina and Davachi, 2006, 2008; Cohn et al.,2009; Staresina and Davachi, 2009; Diana et al., 2010). Finally,the measurement approaches that they advocate rely on a num-ber of assumptions, the most problematic of which is that

xvCowell et al. eschew the recollection/familiarity process distinctionand claim that their model is more consistent with the idea that differ-ent MTL subregions support strong vs. weak memories. However,Cowell et al. appear to share my view that recollection and familiarityare typically associated with the availability of different kinds of infor-mation. Cowell et al. imply that familiarity is usually related to activa-tion of item/object representations and that recollection is related toactivation of item/context bindings. If this is to be taken seriously,then they should agree that one cannot make predictions about therelative contributions of different MTL subregions to item recognitiontasks unless one can estimate the relative influences of recollection andfamiliarity processes. As noted earlier, dissociations in MTL contribu-tions to recollection vs. familiarity have turned out to be one of themost consistent sources of evidence for functional heterogeneity in thehuman MTL.

1280 RANGANATH

Hippocampus

source recognition is a process pure and complete measure ofrecollection.xvi

The more significant problem with Wixted and Squire’s con-tribution is that, although they clearly eschew the idea that therecollection/familiarity distinction can explain functional differ-ences between MTL subregions, they fail to articulate a testabletheoretical account to explain the functional organization of theMTL. Although they cite Squire et al. (2007) as a source ofpromising ‘‘alternative ideas about the functional organizationof the MTL,’’ the cited paper does not propose a theory. Thisis a significant problem because even if one adopts a relativelyneutral stance on the recollection/familiarity distinction, thereis still clear evidence for functional distinctions between MTLsubregions that necessitates some sort of theoretical account.Squire et al. (2007) acknowledged such evidence but only con-cluded that, ‘‘What these differences imply about the functionalorganization of the medial temporal lobe is not yet clear’’(p.881). From a scientific perspective, few would argue that itis preferable to proceed with no theory than one that is imper-fect or incomplete. Accordingly, until Wixted and Squire pro-pose a testable theory that can provide a better account of theevidence than the theories described in this special issue, theirarguments are of limited use in advancing our understandingof the functional organization of the MTL.

LIMITATIONS OF THE MOST EXISTINGMODELS

There are several important issues that have not beenaddressed by BIC or by most other models of the functional or-ganization of the MTL. Perhaps, the most significant of theseissues is the role of the MTL in memory ‘‘systems consolida-tion.’’ One of the most influential ideas to come from earlystudies of amnesic patients was that the MTL only plays a tem-porary role in memory storage (Squire, 2009). A logical refine-ment of this view is that, through cortico-hippocampal interac-tion, associations among neocortical representations become suf-ficiently strengthened so that the hippocampus is no longerrequired to support recovery of the elements of a remote event(Paller, 1997). An alternative account has been that the hippo-

campus is always necessary for spatial or episodic memory,although cortico-hippocampal interactions play an importantrole in consolidation of semantic information in neocortex(Nadel et al., 2000). Although this is an important, and cur-rently unresolved debate, systems consolidation has generallybeen treated separately from theoretical accounts of the func-tional organization of MTL subregions. For instance, whenresearchers refer to the ‘‘neocortex’’ in systems consolidation the-ories (McClelland et al., 1995), it is not clear whether they arereferring to the parahippocampal region or to neocorticalregions outside of the ‘‘MTL memory system.’’ This is a poten-tially important issue because if the parahippocampal region isthe site of ‘‘consolidated’’ episodic memories (e.g., direct con-nections between PRc and PHc developed through interactionswith the hippocampus), one might expect unusual patterns ofretrograde amnesia following damage to these regions (Kapur,1993) that would differ from what would occur following hip-pocampal damage. A comprehensive account of MTL functionwill therefore have to incorporate both anatomical specificityand statements about the relative roles of these subregions insystems consolidation.

Another issue that requires further development is the roleof the EC in human episodic memory. A central challenge inaddressing this issue is that, in most fMRI studies, substantialmagnetic field inhomogeneity artifacts complicate localizationand interpretation of activity in the EC. There are also basicquestions about the anatomical organization of the EC thatneed to be addressed. For instance, it is unclear whether thehuman EC can be subdivided into regions that preferentiallyreceive input from the PRc or PHc, as has been shown inrodents (Witter et al., 2000b; Hargreaves et al., 2005; Kerret al., 2007). Even if this is the case, it is not clear whetherwhat goes on in medial and lateral EC should be viewed as anextension of what is processed by the PHc and PRc, or whetherthere is an added level of information processing and integra-tion in the EC. The latter seems likely (Witter et al., 2000c),but there has been little attention given to this question.

The BIC model, like most models of the functional organiza-tion of the MTL, treats the hippocampus as a functional unit.It will be important to incorporate more detail about the rolesof different hippocampal subfields (dentate gyrus, CA1, CA3,and subiculum). Some insights into this issue have come fromprecise lesion studies (Kesner, 2003), single-unit recording stud-ies (Leutgeb et al., 2004; Knierim et al., 2006; Leutgeb et al.,2007), and immediate early gene imaging studies (Vann et al.,2000) in rats. High-resolution FMRI studies have providedinteresting insights into the functions of different hippocampalsubfields in humans (e.g., Zeineh et al., 2000, 2003; Kirwanet al., 2007; Bakker et al., 2008; Olsen et al., 2009; Prestonet al., 2010), although a consistent picture has not yet emergedfrom this research (Carr et al., 2010).

Another important issue that has not been addressed by mostmodels (but see Aggleton and Brown, 2006) is the role of sub-cortical connections to the hippocampus. It is clear that the ante-rior and mediodorsal thalamic nuclei (Aggleton and Mishkin,

xviAlthough they have claimed to demonstrate that hippocampal activ-ity is sensitive to familiarity, Wixted et al. have not shown that hippo-campal activation is sensitive to experimental manipulations of famili-arity. Instead, their evidence (e.g., Wais et al., 2010) relies on theassumption that if one fails to accurately remember a particular detailabout a recognized item, then the item must have been recognizedsolely on the basis of familiarity. This assumption hinges on theunstated premise that recollection is an all-or-none process that is per-fectly measured by source memory accuracy, which seems to contradicttheir assertion that recollection is continuous. In fact, paradigms likethe one used by Wais et al. (2010) are only feasible if item recognitionaccuracy is near ceiling, and in this event it is likely that one will rec-ollect something about a successfully recognized item even if the corre-sponding source judgment is incorrect.

MEDIAL TEMPORAL LOBES AND EPISODIC MEMORY 1281

Hippocampus

1983) and mamillary bodies (Aggleton and Mishkin, 1985; Tsi-vilis et al., 2008) play a critical role in memory processes, butthe contributions of these areas, along with other subcorticalregions is unclear. In addition, dopaminergic (Lisman and Grace,2005) and cholinergic (Hasselmo, 1999; Ranganath and Rainer,2003) inputs probably play critical roles in modulating plasticityin the MTL, but further work needs to be done in this area.

Finally, to pick up a point made earlier in this article, recol-lective experience depends on interactions between the MTLand other neocortical areas (Hannula and Ranganath, 2009).Conscious recollection probably involves the synchronous reacti-vation of networks that are distributed across the disparate neo-cortical areas that participated in the original memory experi-ence (Johnson et al., 2009). Virtually nothing is known abouthow this occurs, but at least there is some knowledge about thedifferent cortical areas that may be involved in coordinating theactivation of these networks. One such area is the prefrontalcortex, which is widely believed to play a critical role in cogni-tive control processes that regulate the processing of informationduring encoding and retrieval. During encoding, lateral prefron-tal areas probably contribute to the selection of task-relevant in-formation and the implementation of strategies to find associa-tions amongst items that are currently being processed (Wagner,1999; Blumenfeld and Ranganath, 2007; Ranganath and Blu-menfeld, 2008). During retrieval, prefrontal areas play a role inorienting and constraining the retrieval process (Rugg andWilding, 2000; Ranganath and Blumenfeld, 2008) to maximizethe likelihood of recovering goal-relevant information that canguide decisions and actions. Furthermore, prefrontal areas playa critical role in resolving interference and initiating strategies tofacilitate retrieval (Stuss and Benson, 1984; Buckner et al.,1999; Wagner, 1999; Shimamura, 2000; Ranganath and Blu-menfeld, 2008). For all of these reasons, one would expect theprefrontal cortex to contribute to episodic memory functioning.

In addition to the prefrontal cortex, results from functionalimaging studies have shown regions in the retrosplenial cortexand medial and lateral parietal cortex show large increases in ac-tivity during recollection of past events. The retrosplenial cortexis heavily interconnected with regions in the MTL (Kobayashiand Amaral, 2003; Kerr et al., 2007), and damage to this regioncauses amnesia (Valenstein et al., 1987), but little has been doneto functionally characterize the role of this region in memory(Vann et al., 2009a). In contrast, the role of parietal cortex inmemory processes has received considerable attention recently(Wagner et al., 2005; Cabeza et al., 2008; Vilberg and Rugg,2008; Hutchinson et al., 2009; Uncapher and Wagner, 2009).Activity in left ventrolateral parietal cortex, for instance, hasbeen reliably associated with recollection in numerous FMRIstudies (Vilberg and Rugg, 2008). Although patients with pari-etal lobe lesions do not seem to have significant deficits inmemory for past events (Ally et al., 2008; Simons et al.,2008), some evidence suggests that they lack the experience ofconscious recollection (Davidson et al., 2008; Berryhill et al.,2010; Simons et al., 2010). It is possible that, during retrieval,lateral parietal areas play a role in dynamic binding of reacti-vated neocortical representations, which in turn is related to

the experience of recollection (Vilberg and Rugg, 2008). Arelated idea is that lateral parietal areas drive shifts of attentionto recovered information about past episodes in a manner thatis analogous to how other parietal areas contribute to atten-tional capture by external events (Cabeza et al., 2008). Both ofthese accounts suggest a role for the parietal cortex in coordi-nating the activation of neocortical representations during re-trieval, and this process might be critical for the experience ofconscious recollection.

LOOKING TO THE FUTURE

In science, as in politics, fierce debates usually attract a greatdeal of attention, but at the expense of polarizing the varioussides and shifting focus away from the most significant issuesthat are at stake. In this debate, the important question iswhether it is feasible or appropriate to pose unified models toexplain memory at neural, functional, and phenomenologicallevels of analysis. Over the past 4 decades, we have accumu-lated a critical mass of data about the neurocognitive processesthat support episodic memory, only a fraction of which hasbeen reviewed here. Given our current knowledge about themind and brain of episodic memory, we can no longer be satis-fied with descriptive behavioral measurement models (Dunn,2004; Wixted, 2007) and taxonomies of brain systems (Squireand Zola, 1996; Squire, 2004). It is appropriate at this pointto develop and test theories about episodic memory, and fur-thermore, it would be scientifically irresponsible to proceedwithout reference to such theories.

Although there is not universal agreement on the best model,theory, or computational framework, I suspect that readers willfind a good deal of common ground among many of the theo-ries presented in this issue. This common ground can be seen asan indication of the progress we have made in understandingthe neural basis of memory processes. The differences amongthe models are just as important, if not more so. Some differen-ces correspond to a preferred level of description or a particularbody of research that is to be explained, and others correspondto issues that cut across domains. Reconciling the differencesamongst these models and testing their competing predictionsmay eventually lead to the development of a unified frameworkto explain memory at the neural, functional, and phenomeno-logical levels. In that spirit, even if future studies falsify some ofthe predictions made here, the results will reveal new and im-portant insights into the mechanisms of episodic memory.

Acknowledgements

The author gratefully acknowledges the intellectual contribu-tions of Howard Eichenbaum, Andy Yonelinas, Rachel Diana,Debbie Hannula, and the other members of the UC DavisMemory Group. This article also greatly benefitted from dis-cussions with Joe Manns, Scott Slotnick, and Peter Wais, andfrom helpful suggestions and constructive criticism from Joel

1282 RANGANATH

Hippocampus

Voss, Ken Norman, and one anonymous reviewer. The contentis solely the responsibility of the author and does not necessar-ily represent the official views of the National Institute of Men-tal Health or the National Institutes of Health.

REFERENCES

Addis DR, Wong AT, Schacter DL. 2007. Remembering the past andimagining the future: Common and distinct neural substrates dur-ing event construction and elaboration. Neuropsychologia 45:1363–1377.

Addis DR, Pan L, Vu MA, Laiser N, Schacter DL. 2009. Constructiveepisodic simulation of the future and the past: Distinct subsystemsof a core brain network mediate imagining and remembering. Neu-ropsychologia 47:2222–2238.

Aggleton JP, Brown MW. 1999. Episodic memory, amnesia, andthe hippocampal-anterior thalamic axis. Behav Brain Sci 22:425–444.

Aggleton JP, Brown MW. 2006. Interleaving brain systems for episodicand recognition memory. Trends Cogn Sci 10:455–463.

Aggleton JP, Mishkin M. 1983. Memory impairments following re-stricted medial thalamic lesions in monkeys. Experimental BrainResearch 52:199–209.

Aggleton JP, Mishkin M. 1985. Mamillary-body lesions and visual rec-ognition in monkeys. Experimental Brain Research 58:190–197.

Aggleton JP, Shaw C. 1996. Amnesia and recognition memory: A re-analysis of psychometric data. Neuropsychologia 34:51–62.

Aggleton JP, Shaw C, Gaffan EA. 1992. The performance of postence-phalitic amnesic subjects on two behavioural tests of memory:Concurrent discrimination learning and delayed matching-to-sam-ple. Cortex 28:359–372.

Aggleton JP, Vann SD, Denby C, Dix S, Mayes AR, Roberts N, Yone-linas AP. 2005. Sparing of the familiarity component of recognitionmemory in a patient with hippocampal pathology. Neuropsycholo-gia 43:1810–1823.

Aguirre GK, D’Esposito M. 1999. Topographical disorientation: Asynthesis and taxonomy. Brain 122 (Part 9):1613–1628.

Aguirre GK, Zarahn E, D’Esposito M. 1998. Neural componentsof topographical representation. Proc Natl Acad Sci USA 95:839–846.

Ally BA, Simons JS, McKeever JD, Peers PV, Budson AE. 2008. Parie-tal contributions to recollection: Electrophysiological evidence fromaging and patients with parietal lesions. Neuropsychologia 46:1800–1812.

Amaral DG. 1999. Introduction: What is where in the medial tempo-ral lobe? Hippocampus 9:1–6.

Aminoff E, Gronau N, Bar M. 2007. The parahippocampal cortexmediates spatial and nonspatial associations. Cereb Cortex17:1493–1503.

Aminoff E, Schacter DL, Bar M. 2008. The cortical underpinnings ofcontext-based memory distortion. J Cogn Neurosci 20:2226–2237.

Andrews-Hanna JR, Snyder AZ, Vincent JL, Lustig C, Head D,Raichle ME, Buckner RL. 2007. Disruption of large-scale brainsystems in advanced aging. Neuron 56:924–935.

Axmacher N, Helmstaedter C, Elger CE, Fell J. 2008. Enhancementof neocortical-medial temporal EEG correlations during non-REMsleep. Neural Plast 2008:563028.

Bakker A, Kirwan CB, Miller M, Stark CE. 2008. Pattern separationin the human hippocampal CA3 and dentate gyrus. Science319:1640–1642.

Bar M. 2004. Visual objects in context. Nat Rev Neurosci 5:617–629.Bar M, Aminoff E. 2003. Cortical analysis of visual context. Neuron

38:347–358.

Bar M, Aminoff E, Schacter DL. 2008. Scenes unseen: The parahip-pocampal cortex intrinsically subserves contextual associations, notscenes or places per se. J Neurosci 28:8539–8544.

Barense MD, Bussey TJ, Lee AC, Rogers TT, Davies RR, SaksidaLM, Murray EA, Graham KS. 2005. Functional specialization inthe human medial temporal lobe. J Neurosci 25:10239–10246.

Barense MD, Gaffan D, Graham KS. 2007. The human medial tem-poral lobe processes online representations of complex objects.Neuropsychologia 45:2963–2974.

Bartlett FC. 1932. Remembering: A Study in Experimental and SocialPsychology. New York: MacMillan.

Bastin C, Linden M, Charnallet A, Denby C, Montaldi D, RobertsN, Andrew M. 2004. Dissociation between recall and recognitionmemory performance in an amnesic patient with hippocam-pal damage following carbon monoxide poisoning. Neurocase10:330–344.

Baxter MG. 2009. Involvement of medial temporal lobe structures inmemory and perception. Neuron 61:667–677.

Baxter MG, Murray EA. 2001a. Effects of hippocampal lesions ondelayed nonmatching-to-sample in monkeys: A reply to Zola andSquire. Hippocampus 11:201–203.

Baxter MG, Murray EA. 2001b. Opposite relationship of hippocampaland rhinal cortex damage to delayed nonmatching-to-sample defi-cits in monkeys. Hippocampus 11:61–71.

Berryhill ME, Picasso L, Arnold R, Drowos D, Olson IR. 2010. Simi-larities and differences between parietal and frontal patients inautobiographical and constructed experience tasks. Neuropsycholo-gia 48:1385–1393.

Bird CM, Burgess N. 2008. The hippocampus and memory: Insightsfrom spatial processing. Nat Rev Neurosci 9:182–194.

Blumenfeld RS, Ranganath C. 2007. Prefrontal cortex and long-termmemory encoding: An integrative review of findings from neuro-psychology and neuroimaging. Neuroscientist 13:280–291.

Bohbot VD, Allen JJ, Nadel L. 2000. Memory deficits characterizedby patterns of lesions to the hippocampus and parahippocampalcortex. Ann NY Acad Sci 911:355–368.

Bower GH. 1981. Mood and memory. Am Psychol 36:129–148.Bowles B, Crupi C, Pigott S, Parrent A, Wiebe S, Janzen L, Kohler S.

2010. Double dissociation of selective recollection and familiarityimpairments following two different surgical treatments fortemporal-lobe epilepsy. Neuropsychologia 48:2640–2647.

Brandt KR, Gardiner JM, Vargha-Khadem F, Baddeley AD, MishkinM. 2008. Impairment of recollection but not familiarity in a caseof developmental amnesia. Neurocase 15:60–65.

Buckley MJ, Gaffan D. 1997. Impairment of visual object-discrimina-tion learning after perirhinal cortex ablation. Behav Neurosci111:467–475.

Buckley MJ, Gaffan D. 1998a. Learning and transfer of object-reward associations and the role of the perirhinal cortex. BehavNeurosci 112:15–23.

Buckley MJ, Gaffan D. 1998b. Perirhinal cortex ablation impairs con-figural learning and paired-associate learning equally. Neuropsycho-logia 36:535–546.

Buckley MJ, Gaffan D. 1998c. Perirhinal cortex ablation impairs vis-ual object identification. J Neurosci 18:2268–2275.

Buckley MJ, Gaffan D, Murray EA. 1997. Functional double dissocia-tion between two inferior temporal cortical areas: Perirhinal cortexversus middle temporal gyrus. J Neurophysiol 77:587–598.

Buckner RL, Andrews-Hanna JR, Schacter DL. 2008. The brain’sdefault network: Anatomy, function, and relevance to disease. AnnNY Acad Sci 1124:1–38.

Buckner RL, Carroll DC. 2007. Self-projection and the brain. TrendsCogn Sci 11:49–57.

Buckner RL, Kelley WM, Petersen SE. 1999. Frontal cortex contrib-utes to human memory formation. Nat Neurosci 2:311–314.

Buffalo EA, Reber PJ, Squire LR. 1998. The human perirhinal cortexand recognition memory. Hippocampus 8:330–339.

MEDIAL TEMPORAL LOBES AND EPISODIC MEMORY 1283

Hippocampus

Buffalo EA, Ramus SJ, Squire LR, Zola SM. 2000. Perception andrecognition memory in monkeys following lesions of area TE andperirhinal cortex. Learn Mem 7:375–382.

Bunsey M, Eichenbaum H. 1993. Critical role of the parahippocampalregion for paired-associate learning in rats. Behav Neurosci107:740–747.

Burgess N, Becker S, King JA, O’Keefe J. 2001. Memory for eventsand their spatial context: Models and experiments. Philos Trans RSoc Lond B Biol Sci 356:1493–1503.

Burwell RD. 2000. The parahippocampal region: Corticocortical con-nectivity. Ann NY Acad Sci 911:25–42.

Bussey TJ, Saksida LM. 2002. The organization of visual object repre-sentations: A connectionist model of effects of lesions in perirhinalcortex. Eur J Neurosci 15:355–364.

Bussey TJ, Saksida LM, Murray EA. 2002. Perirhinal cortex resolvesfeature ambiguity in complex visual discriminations. Eur J Neuro-sci 15:365–374.

Cabeza R, Ciaramelli E, Olson IR, Moscovitch M. 2008. The parietalcortex and episodic memory: An attentional account. Nat RevNeurosci 9:613–625.

Cabeza R, Rao SM, Wagner AD, Mayer AR, Schacter DL. 2001. Canmedial temporal lobe regions distinguish true from false? An event-related functional MRI study of veridical and illusory recognitionmemory. Proc Natl Acad Sci USA 98:4805–4810.

Carr VA, Rissman J, Wagner AD. 2010. Imaging the human medialtemporal lobe with high-resolution fMRI. Neuron 65:298–308.

Cave CB, Squire LR. 1992. Intact verbal and nonverbal short-termmemory following damage to the human hippocampus. Hippo-campus 2:151–163.

Chun MM, Phelps EA. 2000. Memory deficits for implicit contextualinformation in amnesic subjects with hippocampal damage. NatNeurosci 2:844–847.

Cohen JD, Servan-Schreiber D. 1992. Context, cortex, and dopamine:A connectionist approach to behavior and biology in schizophrenia.Psychol Rev 99:45–77.

Cohen NJ, Eichenbaum H. 1993. Memory, Amnesia, and the Hippo-campal System. Cambridge, MA: MIT Press.

Cohen NJ, Poldrack RA, Eichenbaum H. 1997. Memory for itemsand memory for relations in the procedural/declarative memoryframework. Memory 5:131–178.

Cohn M, Moscovitch M, Lahat A, McAndrews MP. 2009. Re-collection versus strength as the primary determinant of hippo-campal engagement at retrieval. Proc Natl Acad Sci USA 106:22451–22455.

Daselaar SM, Fleck MS, Cabeza R. 2006. Triple dissociation in themedial temporal lobes: Recollection, familiarity, and novelty.J Neurophysiol 96:1902–1911.

Davachi L. 2006. Item, context and relational episodic encoding inhumans. Curr Opin Neurobiol 16:693–700.

Davachi L, Goldman-Rakic PS. 2001. Primate rhinal cortex partici-pates in both visual recognition and working memory tasks: Func-tional mapping with 2-DG. J Neurophysiology 2590–2601.

Davachi L, Mitchell JP, Wagner AD. 2003. Multiple routes to mem-ory: Distinct medial temporal lobe processes build item and sourcememories. Proc Natl Acad Sci USA 100:2157–2162.

Davidson PS, Troyer AK, Moscovitch M. 2006. Frontal lobe contribu-tions to recognition and recall: Linking basic research with clinicalevaluation and remediation. J Int Neuropsychol Soc 12:210–223.

Davidson PS, Anaki D, Ciaramelli E, Cohn M, Kim AS, Murphy KJ,Troyer AK, Moscovitch M, Levine B. 2008. Does lateral parietalcortex support episodic memory? Evidence from focal lesionpatients. Neuropsychologia 46:1743–1755.

Diana RA, Yonelinas AP, Ranganath C. 2007. Imaging recollectionand familiarity in the medial temporal lobe: A three-componentmodel. Trends Cogn Sci 11:379–386.

Diana RA, Yonelinas AP, Ranganath C. 2008a. The effects of unitiza-tion on familiarity-based source memory: Testing a behavioral pre-

diction derived from neuroimaging data. J Exp Psychol LearnMem Cogn 34:730–740.

Diana RA, Yonelinas AP, Ranganath C. 2008b. High-resolution multi-voxel pattern analysis of category selectivity in the medial temporallobes. Hippocampus 18:536–541.

Diana RA, Van Den Boom W, Yonelinas AP, Ranganath C. (sub-mitted). ERP Correlates of Source Memory: Unitized Source Infor-mation Increases Familiarity-Based Retrieval.

Diana RA, Yonelinas AP, Ranganath C. 2010. Medial temporal lobeactivity during source retrieval reflects information type, not mem-ory strength. J Cogn Neurosci 22:1808–1818.

Dudukovic NM, Knowlton BJ. 2006. Remember-Know judgments andretrieval of contextual details. Acta Psychol (Amst) 122:160–173.

Dunn JC. 2004. Remember-know: A matter of confidence. PsycholRev 111:524–542.

Eacott MJ, Gaffan D, Murray EA. 1994. Preserved recognition mem-ory for small sets, and impaired stimulus identification for largesets following rhinal cortex ablations in monkeys. Eur J Neurosci6:1466–1478.

Eacott MJ, Gaffan EA. 2005. The roles of perirhinal cortex, postrhinalcortex, and the fornix in memory for objects, contexts, and eventsin the rat. Q J Exp Psychol B 58:202–217.

Eichenbaum H, Cohen NJ. 2001. From Conditioning to ConsciousRecollection: Memory Systems of the Brain. New York: OxfordUniversity Press.

Eichenbaum H, Otto T, Cohen NJ. 1992. The hippocampus: Whatdoes it do? Behav Neural Biol 57:2–36.

Eichenbaum H, Yonelinas AP, Ranganath C. 2007. The medial tempo-ral lobe and recognition memory. Annu Rev Neurosci 30:123–152.

Ekstrom AD, Bookheimer SY. 2007. Spatial and temporal episodicmemory retrieval recruit dissociable functional networks in thehuman brain. Learn Mem 14:645–654.

Ekstrom AD, Kahana MJ, Caplan JB, Fields TA, Isham EA, NewmanEL, Fried I. 2003. Cellular networks underlying human spatialnavigation. Nature 425:184–188.

Eldridge LL, Sarfatti S, Knowlton BJ. 2002. The effect of testing proce-dure on remember-know judgments. Psychon Bull Rev 9:139–145.

Eldridge LL, Engel SA, Zeineh MM, Bookheimer SY, Knowlton BJ.2005. A dissociation of encoding and retrieval processes in thehuman hippocampus. J Neurosci 25:3280–3286.

Elfman KW, Parks CM, Yonelinas AP. 2008. Testing a neurocomputa-tional model of recollection, familiarity, and source recognition. JExp Psychol Learn Mem Cogn 34:752–768.

Epstein R, Harris A, Stanley D, Kanwisher N. 1999. The parahippo-campal place area: Recognition, navigation, or encoding? Neuron23:115–125.

Epstein R, Kanwisher N. 1998. A cortical representation of the localvisual environment. Nature 392:598–601.

Erickson CA, Desimone R. 1999. Responses of macaque perirhinalneurons during and after visual stimulus association learning. JNeurosci 19:10404–10416.

Fenker DB, Schott BH, Richardson-Klavehn A, Heinze HJ, Duzel E.2005. Recapitulating emotional context: Activity of amygdala, hip-pocampus and fusiform cortex during recollection and familiarity.Eur J Neurosci 21:1993–1999.

Fortin NJ, Wright SP, Eichenbaum H. 2004. Recollection-like mem-ory retrieval in rats is dependent on the hippocampus. Nature431:188–191.

Fyhn M, Molden S, Witter MP, Moser EI, Moser MB. 2004. Spatialrepresentation in the entorhinal cortex. Science 305:1258–1264.

Gaffan D. 1974. Recognition impaired and association intact in thememory of monkeys after transection of the fornix. J Comp Phys-iol Psychol 86:1100–1109.

Giovanello KS, Keane MM, Verfaellie M. 2006. The contribution offamiliarity to associative memory in amnesia. Neuropsychologia44:1859–1865.

1284 RANGANATH

Hippocampus

Gold JJ, Smith CN, Bayley PJ, Shrager Y, Brewer JB, Stark CE, Hop-kins RO, Squire LR. 2006. Item memory, source memory, and themedial temporal lobe: Concordant findings from fMRI and mem-ory-impaired patients. Proc Natl Acad Sci USA 103:9351–9356.

Goodwin DW, Powell B, Bremer D, Hoine H, Stern J. 1969.Alcohol and recall: State-dependent effects in man. Science 163:1358–1360.

Graf P, Schacter DL. 1985. Implicit and explicit memory for newassociations in normal and amnesic subjects. J Exp Psychol LearnMem Cogn 11:501–518.

Graham KS, Barense MD, Lee AC. 2010. Going beyond LTM in theMTL: A synthesis of neuropsychological and neuroimaging findingson the role of the medial temporal lobe in memory and percep-tion. Neuropsychologia 48:831–853.

Grill-Spector K, Malach R. 2001. fMR-adaptation: A tool for studyingthe functional properties of human cortical neurons. Acta Psychol(Amst) 107:293–321.

Hannula DE, Ranganath C. 2008. Medial temporal lobe activitypredicts successful relational memory binding. J Neurosci 28:116–124.

Hannula DE, Ranganath C. 2009. The eyes have it: Hippocampal ac-tivity predicts expression of memory in eye movements. Neuron63:592–599.

Hannula DE, Tranel D, Cohen NJ. 2006. The long and the short ofit: Relational memory impairments in amnesia, even at short lags. JNeurosci 26:8352–8359.

Hannula DE, Ryan JD, Tranel D, Cohen NJ. 2007. Rapid onset rela-tional memory effects are evident in eye movement behavior, butnot in hippocampal amnesia. J Cogn Neurosci 19:1690–1705.

Hargreaves EL, Rao G, Lee I, Knierim JJ. 2005. Major dissociationbetween medial and lateral entorhinal input to dorsal hippocam-pus. Science 308:1792–1794.

Hartley T, Bird CM, Chan D, Cipolotti L, Husain M, Vargha- KhademF, Burgess N. 2007. The hippocampus is required for short-term to-pographical memory in humans. Hippocampus 17:34–48.

Haskins AL, Yonelinas AP, Quamme JR, Ranganath C. 2008. Perirhi-nal cortex supports encoding and familiarity-based recognition ofnovel associations. Neuron 59:554–560.

Hassabis D, Kumaran D, Maguire EA. 2007a. Using imagination tounderstand the neural basis of episodic memory. J Neurosci27:14365–14374.

Hassabis D, Kumaran D, Vann SD, Maguire EA. 2007b. Patientswith hippocampal amnesia cannot imagine new experiences. ProcNatl Acad Sci USA 104:1726–1731.

Hassabis D, Chu C, Rees G, Weiskopf N, Molyneux PD, MaguireEA. 2009. Decoding neuronal ensembles in the human hippocam-pus. Curr Biol 19:546–554.

Hasselmo ME. 1999. Neuromodulation: Acetylcholine and memoryconsolidation. Trends Cogn Sci 3:351–359.

Hasselmo ME, Stern CE. 2006. Mechanisms underlyingworking memory for novel information. Trends Cogn Sci 10:487–493.

Holdstock JS, Shaw C, Aggleton JP. 1995. The performance of amne-sic subjects on tests of delayed matching-to-sample and delayedmatching-to-position. Neuropsychologia 33:1583–1596.

Holdstock JS, Gutnikov SA, Gaffan D, Mayes AR. 2000. Perceptualand mnemonic matching-to-sample in humans: Contributions ofthe hippocampus, perirhinal and other medial temporal lobe corti-ces. Cortex 36:301–322.

Holdstock JS, Mayes AR, Gong QY, Roberts N, Kapur N. 2005. Itemrecognition is less impaired than recall and associative recognitionin a patient with selective hippocampal damage. Hippocampus15:203–215.

Hollingworth A, Williams CC, Henderson JM. 2001. To see andremember: Visually specific information is retained in memoryfrom previously attended objects in natural scenes. Psychon BullRev 8:761–768.

Horel JA. 1994. Some comments on the special cognitive functionsclaimed for the hippocampus. Cortex 30:269–280.

Horel JA, Pytko DE. 1982. Behavioral effect of local cooling in tem-poral lobe of monkeys. J Neurophysiol 47:11–22.

Horel JA, Voytko ML, Salsbury KG. 1984. Visual learning suppressedby cooling the temporal pole. Behav Neurosci 98:310–324.

Horel JA, Pytko-Joiner DE, Voytko ML, Salsbury K. 1987. The per-formance of visual tasks while segments of the inferotemporal cor-tex are suppressed by cold. Behav Brain Res 23:29–42.

Hutchinson JB, Uncapher MR, Wagner AD. 2009. Posterior parietalcortex and episodic retrieval: Convergent and divergent effects ofattention and memory. Learn Mem 16:343–356.

Jackson O III, Schacter DL. 2004. Encoding activity in anteriormedial temporal lobe supports subsequent associative recognition.Neuroimage 21:456–462.

Janzen G, van Turennout M. 2004. Selective neural representation ofobjects relevant for navigation. Nat Neurosci 7:673–677.

Janzen G, Wagensveld B, van Turennout M. 2007. Neural representa-tion of navigational relevance is rapidly induced and long lasting.Cereb Cortex 17:975–981.

Jeneson A, Kirwan CB, Hopkins RO, Wixted JT, Squire LR. 2010.Recognition memory and the hippocampus: A test of the hippo-campal contribution to recollection and familiarity. Learn Mem17:63–70.

Jenkins LJ, Ranganath C. Prefrontal and medial temporal lobe activity atencoding predicts temporal context memory. J Neurosci (in press).

Johnson JD, McDuff SG, Rugg MD, Norman KA. 2009. Recollec-tion, familiarity, and cortical reinstatement: A multivoxel patternanalysis. Neuron 63:697–708.

Johnson MK, Hashtroudi S, Lindsay DS. 1993. Source monitoring.Psychol Bull 114:3–28.

Johnson MK, Raye CL. 1981. Reality monitoring. Psychol Rev88:67–85.

Jonides J, Lewis RL, Nee DE, Lustig CA, Berman MG, Moore KS.2008. The mind and brain of short-term memory. Annu Rev Psy-chol 59:193–224.

Kapur N. 1993. Focal retrograde amnesia in neurological disease: Acritical review. Cortex 29:217–234.

Kensinger EA, Schacter DL. 2006. Amygdala activity is associatedwith the successful encoding of item, but not source, informationfor positive and negative stimuli. J Neurosci 26:2564–2570.

Kerr KM, Agster KL, Furtak SC, Burwell RD. 2007. Functional neu-roanatomy of the parahippocampal region: The lateral and medialentorhinal areas. Hippocampus 17:697–708.

Kesner RP. 2003. Functional specificity of memory function associatedwith different subregions of the medial temporal lobe. Curr NeurolNeurosci Rep 3:449–451.

Kesner RP, Hopkins RO. 2001. Short-term memory for duration anddistance in humans: Role of the hippocampus. Neuropsychology15:58–68.

Kirwan CB, Jones CK, Miller MI, Stark CE. 2007. High-resolutionfMRI investigation of the medial temporal lobe. Hum Brain Mapp28:959–966.

Kirwan CB, Stark CE. 2004. Medial temporal lobe activation duringencoding and retrieval of novel face-name pairs. Hippocampus14:919–930.

Kirwan CB, Wixted JT, Squire LR. 2008. Activity in the medial tem-poral lobe predicts memory strength, whereas activity in the pre-frontal cortex predicts recollection. J Neurosci 28:10541–10548.

Kirwan CB, Wixted JT, Squire LR. 2010. A demonstration that thehippocampus supports both recollection and familiarity. Proc NatlAcad Sci USA 107:344–348.

Knierim JJ, Lee I, Hargreaves EL. 2006. Hippocampal place cells: Par-allel input streams, subregional processing, and implications for ep-isodic memory. Hippocampus 16:755–764.

Kobayashi Y, Amaral DG. 2003. Macaque monkey retrosplenial cor-tex: II. Cortical afferents. J Comp Neurol 466:48–79.

MEDIAL TEMPORAL LOBES AND EPISODIC MEMORY 1285

Hippocampus

Komorowski RW, Manns JR, Eichenbaum H. 2009. Robust conjunc-tive item-place coding by hippocampal neurons parallels learningwhat happens where. J Neurosci 29:9918–9929.

Konkel A, Cohen NJ. 2009. Relational memory and the hippocampus:Representations and methods. Front Neurosci 3:166–174.

Konorski J. 1959. A new method of physiological investigation ofrecent memory in animals. Bull Acad Polym Sci 7:115–117.

Kreiman G, Koch C, Fried I. 2000. Imagery neurons in the humanbrain. Nature 408:357–361.

Lee AC, Barense MD, Graham KS. 2005a. The contribution of thehuman medial temporal lobe to perception: Bridging the gapbetween animal and human studies. Q J Exp Psychol B 58:300–325.

Lee AC, Buckley MJ, Pegman SJ, Spiers H, Scahill VL, Gaffan D,Bussey TJ, Davies RR, Kapur N, Hodges JR. 2005b. Specializationin the medial temporal lobe for processing of objects and scenes.Hippocampus 15:782–797.

Lee AC, Bussey TJ, Murray EA, Saksida LM, Epstein RA, Kapur N,Hodges JR, Graham KS. 2005c. Perceptual deficits in amnesia:Challenging the medial temporal lobe ‘‘mnemonic’’ view. Neuro-psychologia 43:1–11.

Lee AC, Bandelow S, Schwarzbauer C, Henson RN, Graham KS.2006a. Perirhinal cortex activity during visual object discrimina-tion: An event-related fMRI study. Neuroimage 33:362–373.

Lee AC, Buckley MJ, Gaffan D, Emery T, Hodges JR, Graham KS.2006b. Differentiating the roles of the hippocampus and perirhinalcortex in processes beyond long-term declarative memory: A doubledissociation in dementia. J Neurosci 26:5198–5203.

Leutgeb JK, Leutgeb S, Moser MB, Moser EI. 2007. Pattern separa-tion in the dentate gyrus and CA3 of the hippocampus. Science315:961–966.

Leutgeb S, Leutgeb JK, Treves A, Moser MB, Moser EI. 2004. Dis-tinct ensemble codes in hippocampal areas CA3 and CA1. Science305:1295–1298.

Lisman JE, Grace AA. 2005. The hippocampal-VTA loop: Controllingthe entry of information into long-term memory. Neuron 46:703–713.

Liu Z, Richmond BJ. 2000. Response differences in monkey TE andperirhinal cortex: Stimulus association related to reward schedules.J Neurophysiol 83:1677–1692.

Liu Z, Richmond BJ, Murray EA, Saunders RC, Steenrod S, Stubble-field BK, Montague DM, Ginns EI. 2004. DNA targeting of rhi-nal cortex D2 receptor protein reversibly blocks learning of cuesthat predict reward. Proc Natl Acad Sci USA 101:12336–12341.

Mahut H, Zola-Morgan S, Moss M. 1982. Hippocampal resectionsimpair associative learning and recognition memory in the monkey.J Neurosci 2:1214–1229.

Manns JR, Eichenbaum H. 2009. A cognitive map for object memoryin the hippocampus. Learn Mem 16:616–624.

Manns JR, Hopkins RO, Reed JM, Kitchener EG, Squire LR. 2003.Recognition memory and the human hippocampus. Neuron37:171–180.

Manns JR, Howard MW, Eichenbaum H. 2007. Gradual changes inhippocampal activity support remembering the order of events.Neuron 56:530–540.

Manns JR, Squire LR. 1999. Impaired recognition memory on thedoors and people test after damage limited to the hippocampalregion. Hippocampus 9:495–499.

Manns JR, Squire LR. 2001. Perceptual learning, awareness, and thehippocampus. Hippocampus 11:776–782.

Marr D. 1971. Simple memory: A theory for archicortex. Philos TransR Soc London 262:23–81.

Mayes A, Montaldi D, Migo E. 2007. Associative memory and themedial temporal lobes. Trends Cogn Sci 11:126–135.

Mayes AR, Holdstock JS, Isaac CL, Montaldi D, Grigor J, GummerA, Cariga P, Downes JJ, Tsivilis D, Gaffan D. 2004. Associativerecognition in a patient with selective hippocampal lesions and rel-atively normal item recognition. Hippocampus 14:763–784.

McClelland JL, McNaughton BL, O’Reilly RC. 1995. Why there arecomplementary learning systems in the hippocampus and neocor-tex: Insights from the successes and failures of connectionist modelsof learning and memory. Psychol Rev 102:419–457.

McDowell S, Whyte J, D’Esposito M. 1998. Differential effect of adopaminergic agonist on prefrontal function in traumatic braininjury patients. Brain 121 (Part 6):1155–1164.

McNab F, Varrone A, Farde L, Jucaite A, Bystritsky P, Forssberg H,Klingberg T. 2009. Changes in cortical dopamine D1 receptorbinding associated with cognitive training. Science 323:800–802.

Meunier M, Bachevalier J, Mishkin M, Murray EA. 1993. Effects onvisual recognition of combined and separate ablations of the ento-rhinal and perirhinal cortex in rhesus monkeys. J Neurosci13:5418–5432.

Meunier M, Hadfield W, Bachevalier J, Murray EA. 1996. Effectsof rhinal cortex lesions combined with hippocampectomy onvisual recognition memory in rhesus monkeys. J Neurophysiol 75:1190–1205.

Mickes L, Wais PE, Wixted JT. 2009. Recollection is a continuousprocess: Implications for dual-process theories of recognition mem-ory. Psychol Sci 20:509–515.

Migo E, Montaldi D, Norman KA, Quamme J, Mayes A. 2009. Thecontribution of familiarity to recognition memory is a function oftest format when using similar foils. Q J Exp Psychol (Colchester)62:1198–1215.

Miller EK, Li L, Desimone R. 1993. Activity of neurons in anteriorinferior temporal cortex during a short-term memory task. JNeurosci 13:1460–1478.

Milner B. 1972. Disorders of learning and memory after temporallobe lesions in man. Clin Neurosurg 19:421–446.

Milner B, Corkin S, Teuber HL. 1968. Further analysis of the hippo-campal amnesic syndrome: 14-year follow-up study of H.M. Neu-ropsychologia 6:215–234.

Minzenberg MJ, Carter CS. 2008. Modafinil: A review of neurochem-ical actions and effects on cognition. Neuropsychopharmacology33:1477–1502.

Mishkin M. 1978. Memory in monkeys severely impaired by com-bined but not separate removal of the amygdala and hippocampus.Nature 273:297–298.

Mishkin M. 1982. A memory system in the monkey. Phil R Soc Lon-don [Biol] 298:85–92.

Mishkin M, Delacour J. 1975. An analysis of short-term visual mem-ory in the monkey. J Exp Psychol: Anim Behav Proc 1:326–334.

Mitchell KJ, Johnson MK. 2009. Source monitoring 15 years later:What have we learned from fMRI about the neural mechanisms ofsource memory? Psychol Bull 135:638–677.

Mitchell KJ, Johnson MK, Raye CL, D’Esposito M. 2000. fMRI evi-dence of age-related hippocampal dysfunction in feature binding inworking memory. Brain Res Cogn Brain Res 10:197–206.

Miyashita Y. 1988. Neuronal correlate of visual associative long-termmemory in the primate temporal cortex. Nature 335:817–820.

Miyashita Y, Chang HS. 1988. Neuronal correlate of pictorial short-term memory in the primate temporal cortex. Nature 331:68–70.

Montaldi D, Spencer TJ, Roberts N, Mayes AR. 2006. The neural sys-tem that mediates familiarity memory. Hippocampus 16:504–520.

Moscovitch M. 1992. Memory and working-with-memory: A compo-nent process model based on modules and central systems. J CognNeurosci 4:257–267.

Moscovitch M. 1995. Recovered consciousness: A hypothesis concern-ing modularity and episodic memory. J Clin Exp Neuropsychol17:276–290.

Moscovitch M. 2008. The hippocampus as a ‘‘Stupid,’’ domain-specificmodule: Implications for theories of recent and remote memory, and ofimagination. Can J Exp Psychol Rev Can De Psychol Exp 62:62–79.

Moses SN, Ryan JD. 2006. A comparison and evaluation of the pre-dictions of relational and conjunctive accounts of hippocampalfunction. Hippocampus 16:43–65.

1286 RANGANATH

Hippocampus

Mumby DG. 2001. Perspectives on object-recognition memory follow-ing hippocampal damage: Lessons from studies in rats. Behav BrainRes 127:159–181.

Murray EA, Mishkin M. 1984. Severe tactual as well as visual memorydeficits follow combined removal of the amygdala and hippocam-pus in monkeys. J Neurosci 4:2565–2580.

Murray EA, Mishkin M. 1986. Visual recognition in monkeys follow-ing rhinal cortical ablations combined with either amygdalectomyor hippocampectomy. J Neurosci 6:1991–2003.

Murray EA, Mishkin M. 1998. Object recognition and location mem-ory in monkeys with excitotoxic lesions of the amygdala and hip-pocampus. J Neurosci 18:6568–6582.

Murray EA, Gaffan D, Mishkin M. 1993. Neural substrates of visualstimulus-stimulus association in rhesus monkeys. J Neurosci 13:4549–4561.

Murray EA, Bussey TJ, Saksida LM. 2007. Visual perception andmemory: A new view of medial temporal lobe function in primatesand rodents. Annu Rev Neurosci 30:99–122.

Nadel L, Samsonovich A, Ryan L, Moscovitch M. 2000. Multipletrace theory of human memory: Computational, neuroimaging,and neuropsychological results. Hippocampus 10:352–368.

Nakamura K, Kubota K. 1995. Mnemonic firing of neurons in themonkey temporal pole during a visual recognition memory task. JNeurophysiol 74:162–178.

Nemanic S, Alvarado MC, Bachevalier J. 2004. The hippocampal/par-ahippocampal regions and recognition memory: Insights from vis-ual paired comparison versus object-delayed nonmatching in mon-keys. J Neurosci 24:2013–2026.

Nichols EA, Kao YC, Verfaellie M, Gabrieli JD. 2006. Working mem-ory and long-term memory for faces: Evidence from fMRI andglobal amnesia for involvement of the medial temporal lobes. Hip-pocampus 16:604–616.

Nordahl CW, Ranganath C, Yonelinas AP, DeCarli C, Reed BR, JagustWJ. 2005. Different mechanisms of episodic memory failure inmild cognitive impairment. Neuropsychologia 43:1688–1697.

Nordahl CW, Ranganath C, Yonelinas AP, Decarli C, Fletcher E,Jagust WJ. 2006. White matter changes compromise prefrontal cor-tex function in healthy elderly individuals. J Cogn Neurosci 18:418–429.

Norman KA, Newman EL, Perotte AJ. 2005. Methods for reducinginterference in the complementary learning systems model: Oscil-lating inhibition and autonomous memory rehearsal. Neural Netw18:1212–1228.

Norman KA, O’Reilly RC. 2003. Modeling hippocampal and neocort-ical contributions to recognition memory: A complementary-learn-ing-systems approach. Psychol Rev 110:611–646.

Norman KA, Tepe K, Nyhus E, Curran T. 2008. Event-related poten-tial correlates of interference effects on recognition memory. Psy-chon Bull Rev 15:36–43.

Nyberg L, Sandblom J, Jones S, Neely AS, Petersson KM, Ingvar M,Backman L. 2003. Neural correlates of training-related memoryimprovement in adulthood and aging. Proc Natl Acad Sci USA100:13728–13733.

O’Keefe J, Dostrovsky J. 1971. The hippocampus as a spatial map.Preliminary evidence from unit activity in the freely-moving rat.Brain Res 34:171–175.

O’keefe J, Nadel L. 1978. The Hippocampus as a Cognitive Map.London: Oxford University Press.

O’Neil EB, Cate AD, Kohler S. 2009. Perirhinal cortex contributes toaccuracy in recognition memory and perceptual discriminations.J Neurosci 29:8329–8334.

O’Reilly RC, Norman KA. 2002. Hippocampal and neocortical con-tributions to memory: Advances in the complementary learningsystems framework. Trends Cogn Sci 6:505–510.

O’Reilly RC, Rudy JW. 2001. Conjunctive representations in learningand memory: Principles of cortical and hippocampal function. Psy-chol Rev 108:311–345.

Olesen PJ, Westerberg H, Klingberg T. 2004. Increased prefrontal andparietal activity after training of working memory. Nat Neurosci7:75–79.

Olsen RK, Nichols EA, Chen J, Hunt JF, Glover GH, Gabrieli JD,Wagner AD. 2009. Performance-related sustained and anticipatoryactivity in human medial temporal lobe during delayed match-to-sample. J Neurosci 29:11880–11890.

Olson IR, Moore KS, Stark M, Chatterjee A. 2006a. Visual workingmemory is impaired when the medial temporal lobe is damaged.J Cogn Neurosci 18:1087–1097.

Olson IR, Page K, Moore KS, Chatterjee A, Verfaellie M. 2006b.Working memory for conjunctions relies on the medial temporallobe. J Neurosci 26:4596–4601.

Owen AM, Sahakian BJ, Semple J, Polkey CE, Robbins TW. 1995.Visuo-spatial short-term recognition memory and learning aftertemporal lobe excisions, frontal lobe excisions or amygdalo-hippo-campectomy in man. Neuropsychologia 33:1–24.

Owen AM, Morris RG, Sahakian BJ, Polkey CE, Robbins TW. 1996.Double dissociations of memory and executive functions in work-ing memory tasks following frontal lobe excisions, temporal lobeexcisions or amygdalo-hippocampectomy in man. Brain 119 (Part5):1597–1615.

Paller KA. 1997. Consolidating dispersed neocortical memories: Themissing link in amnesia. Memory 5:73–88.

Parks CM, Yonelinas AP. 2009. Evidence for a memory threshold insecond-choice recognition memory responses. Proc Natl Acad SciUSA 106:11515–11519.

Pastalkova E, Itskov V, Amarasingham A, Buzsaki G. 2008. Internallygenerated cell assembly sequences in the rat hippocampus. Science321:1322–1327.

Peters J, Thoma P, Koch B, Schwarz M, Daum I. 2009. Impairmentof verbal recollection following ischemic damage to the right ante-rior hippocampus. Cortex 45:592–601.

Petrides M, Pandya DN. 1999. Dorsolateral prefrontal cortex: Com-parative cytoarchitectonic analysis in the human and the macaquebrain and corticocortical connection patterns. Eur J Neurosci11:1011–1036.

Piekema C, Kessels RP, Mars RB, Petersson KM, Fernandez G.2006. The right hippocampus participates in short-term memorymaintenance of object-location associations. Neuroimage 33:374–382.

Polyn SM, Kahana MJ. 2008. Memory search and the neural represen-tation of context. Trends Cogn Sci 12:24–30.

Polyn SM, Norman KA, Kahana MJ. 2009. Task context and organi-zation in free recall. Neuropsychologia 47:2158–2163.

Postle BR. 2006. Working memory as an emergent property of themind and brain. Neuroscience 139:23–38.

Preston AR, Gabrieli JD. 2008. Dissociation between explicit memoryand configural memory in the human medial temporal lobe. CerebCortex 18:2192–2207.

Preston AR, Bornstein AM, Hutchinson JB, Gaare ME, Glover GH,Wagner AD. 2010. High-resolution fMRI of content-sensitive sub-sequent memory responses in human medial temporal lobe.J Cogn Neurosci 22:156–173.

Quamme JR, Yonelinas AP, Widaman KF, Kroll NE, Sauve MJ. 2004.Recall and recognition in mild hypoxia: Using covariance structuralmodeling to test competing theories of explicit memory. Neuropsy-chologia 42:672–691.

Quamme JR, Yonelinas AP, Norman KA. 2007. Effect of unitizationon associative recognition in amnesia. Hippocampus 17:192–200.

Ragland JD, Laird AR, Ranganath C, Blumenfeld RS, Gonzales SM,Glahn DC. 2009. Prefrontal activation deficits during episodicmemory in schizophrenia. Am J Psychiatry 166:863–874.

Ranganath C. 2009. Interrelationships between working memory andlong-term memory. In: Rosler F, Ranganath C, Roder B, KluweRH, editors. Neuroimaging of Human Memory. New York:Oxford. pp 227–254.

MEDIAL TEMPORAL LOBES AND EPISODIC MEMORY 1287

Hippocampus

Ranganath C. 2010. Binding items and contexts: The cognitive neuro-science of episodic memory. Curr Direct Psychol Sci 19:131–137.

Ranganath C, Blumenfeld RS. 2005. Doubts about double dissocia-tions between short- and long-term memory. Trends Cogn Sci374–380.

Ranganath C, Blumenfeld RS. 2008. Prefrontal cortex and memory.In: Byrne JH, editor. Learning and Memory: A ComprehensiveReference. Oxford: Academic Press. pp 261–279.

Ranganath C, D’Esposito M. 2001. Medial temporal lobe activityassociated with active maintenance of novel information. Neuron31:865–873.

Ranganath C, Minzenberg MJ, Ragland JD. 2008. The cognitive neu-roscience of memory function and dysfunction in schizophrenia.Biol Psychiatry 64:18–25.

Ranganath C, Rainer G. 2003. Neural mechanisms for detecting andremembering novel events. Nat Rev Neurosci 4:193–202.

Ranganath C, Yonelinas AP, Cohen MX, Dy CJ, Tom SM, D’EspositoM. 2003. Dissociable correlates of recollection and familiaritywithin the medial temporal lobes. Neuropsychologia 42:2–13.

Rempel-Clower NL, Zola SM, Squire LR, Amaral DG. 1996. Threecases of enduring memory impairment after bilateral damage lim-ited to the hippocampal formation. J Neurosci 16:5233–5255.

Rhodes SM, Donaldson DI. 2007. Electrophysiological evidence forthe influence of unitization on the processes engaged during epi-sodic retrieval: Enhancing familiarity based remembering. Neuro-psychologia 45:412–424.

Ringo JL. 1988. Seemingly discrepant data from hippocampectomizedmacaques are reconciled by detectability analysis. Behav Neurosci102:173–177.

Ringo JL. 1991. Memory decays at the same rate in macaques withand without brain lesions when expressed in d’ or arcsine terms.Behav Brain Res 42:123–134.

Rolls ET. 1996. A theory of hippocampal function in memory. Hippo-campus 6:601–620.

Rolls ET, Kesner RP. 2006. A computational theory of hippocampalfunction, and empirical tests of the theory. Prog Neurobiol 79:1–48.

Rotello CM, Macmillan NA, Reeder JA, Wong M. 2005. The remem-ber response: Subject to bias, graded, and not a process-pure indi-cator of recollection. Psychon Bull Rev 12:865–873.

Rugg MD, Wilding EL. 2000. Retrieval processing and episodic mem-ory. Trends Cogn Sci 4:108–115.

Ryan JD, Althoff RR, Whitlow S, Cohen NJ. 2000. Amnesia is a defi-cit in relational memory. Psychol Sci 11:454–461.

Ryan JD, Cohen NJ. 2004. The nature of change detection and onlinerepresentations of scenes. J Exp Psychol Hum Percept Perform30:988–1015.

Sakai K, Miyashita Y. 1991. Neural organization for the long-termmemory of paired associates [see comments]. Nature 354:152–155.

Schacter DL, Addis DR. 2009. On the nature of medial temporallobe contributions to the constructive simulation of future events.Philos Trans R Soc Lond B Biol Sci 364:1245–1253.

Schacter DL, Norman KA, Koutstaal W. 1998. The cognitive neuro-science of constructive memory. Annu Rev Psychol 49:289–318.

Schacter DL, Addis DR, Buckner RL. 2007. Remembering the past toimagine the future: The prospective brain. Nat Rev Neurosci8:657–661.

Sederberg PB, Howard MW, Kahana MJ. 2008. A context-based theoryof recency and contiguity in free recall. Psychol Rev 115:893–912.

Shimamura AP. 2000. The role of the prefrontal cortex in dynamic fil-tering. Psychobiology 28:207–218.

Shrager Y, Levy DA, Hopkins RO, Squire LR. 2008. Workingmemory and the organization of brain systems. J Neurosci 28:4818–4822.

Simons JS, Peers PV, Hwang DY, Ally BA, Fletcher PC, Budson AE.2008. Is the parietal lobe necessary for recollection in humans?Neuropsychologia 46:1185–1191.

Simons JS, Peers PV, Mazuz YS, Berryhill ME, Olson IR. 2010. Dis-sociation between memory accuracy and memory confidence fol-lowing bilateral parietal lesions. Cereb Cortex 20:479–485.

Slotnick SD. 2010a. Does the hippocampus mediate objective bindingor subjective remembering? Neuroimage 49:1769–1776.

Slotnick SD. 2010b. ‘‘Remember’’ source memory ROCs indicate rec-ollection is a continuous process. Memory 18:27–39.

Slotnick SD, Dodson CS. 2005. Support for a continuous (single-pro-cess) model of recognition memory and source memory. MemCognit 33:151–170.

Slotnick SD, Klein SA, Dodson CS, Shimamura AP. 2000. An analysisof signal detection and threshold models of source memory. J ExpPsychol Learn Mem Cogn 26:1499–1517.

Smith DM, Mizumori SJ. 2006. Hippocampal place cells, context,and episodic memory. Hippocampus 16:716–729.

Smith SM. 1979. Remembering in and out of context. J Exp PsycholLearn Mem Cogn 5:460–471.

Squire LR. 2004. Memory systems of the brain: A brief history andcurrent perspective. Neurobiol Learn Mem 82:171–177.

Squire LR. 2009. The legacy of patient H.M. for neuroscience. Neu-ron 61:6–9.

Squire LR, Knowlton B, Musen G. 1993. The structure and organiza-tion of memory. Annu Rev Psychol 44:453–495.

Squire LR, Shimamura AP. 1986. Characterizing amnesic patients forneurobehavioral study. Behav Neurosci 100:866–877.

Squire LR, Wixted JT, Clark RE. 2007. Recognition memory and themedial temporal lobe: A new perspective. Nat Rev Neurosci8:872–883.

Squire LR, Zola SM. 1996. Structure and function of declarative andnondeclarative memory systems. Proc Natl Acad Sci USA 93:13515–13522.

Squire LR, Zola-Morgan S. 1991. The medial temporal lobe memorysystem. Science 253:1380–1386.

Squire LR, Zola-Morgan S, Chen KS. 1988. Human amnesia and ani-mal models of amnesia: Performance of amnesic patients on testsdesigned for the monkey. Behav Neurosci 102:210–221.

Staresina BP, Davachi L. 2006. Differential encoding mechanisms forsubsequent associative recognition and free recall. J Neurosci 26:9162–9172.

Staresina BP, Davachi L. 2008. Selective and shared contributions ofthe hippocampus and perirhinal cortex to episodic item and asso-ciative encoding. J Cogn Neurosci 20:1478–1489.

Staresina BP, Davachi L. 2009. Mind the gap: Binding experiencesacross space and time in the human hippocampus. Neuron63:267–276.

Stark CE, Squire LR. 2000. Intact visual perceptual discrimination inhumans in the absence of perirhinal cortex. Learn Mem 7:273–278.

Stern CE, Sherman SJ, Kirchhoff BA, Hasselmo ME. 2001. Medialtemporal and prefrontal contributions to working memory taskswith novel and familiar stimuli. Hippocampus 11:337–346.

Stuss DT, Benson DF. 1984. Neuropsychological studies of the frontallobes. Psychol Bull 95:3–28.

Suzuki WA, Amaral DG. 1994. Perirhinal and parahippocampal corti-ces of the macaque monkey: Cortical afferents. J Comp Neurol350:497–533.

Suzuki WA, Zola-Morgan S, Squire LR, Amaral DG. 1993. Lesions ofthe perirhinal and parahippocampal cortices in the monkey pro-duce long-lasting memory impairment in the visual and tactualmodalities. J Neurosci 13:2430–2451.

Suzuki WA, Miller EK, Desimone R. 1997. Object and placememory in the macaque entorhinal cortex. J Neurophysiol78:1062–1081.

Sybirska E, Davachi L, Goldman-Rakic PS. 2000. Prominence ofdirect entorhinal-CA1 pathway activation in sensorimotor and cog-nitive tasks revealed by 2-DG functional mapping in nonhumanprimate. J Neurosci 20:5827–5834.

1288 RANGANATH

Hippocampus

Taylor KI, Moss HE, Stamatakis EA, Tyler LK. 2006. Binding cross-modal object features in perirhinal cortex. Proc Natl Acad Sci USA103:8239–8244.

Treisman AM, Kanwisher NG. 1998. Perceiving visually presentedobjects: Recognition, awareness, and modularity. Curr Opin Neu-robiol 8:218–226.

Treves A, Rolls ET. 1994. Computational analysis of the role of thehippocampus in memory. Hippocampus 4:374–391.

Tsivilis D, Vann SD, Denby C, Roberts N, Mayes AR, Montaldi D,Aggleton JP. 2008. A disproportionate role for the fornix andmammillary bodies in recall versus recognition memory. Nat Neu-rosci 11:834–842.

Tubridy S, Davachi L. Medial temporal lobe contributions to episodicsequence encoding. Cereb Cortex (in press).

Tulving E. 1985. Elements of Episodic Memory. New York: OxfordUniversity Press.

Turriziani P, Fadda L, Caltagirone C, Carlesimo GA. 2004. Recogni-tion memory for single items and for associations in amnesicpatients. Neuropsychologia 42:426–433.

Uncapher MR, Wagner AD. 2009. Posterior parietal cortex and epi-sodic encoding: Insights from fMRI subsequent memory effectsand dual-attention theory. Neurobiol Learn Mem 91:139–154.

Valenstein E, Bowers D, Verfaellie M, Heilman K, Day A, Watson R.1987. Retrosplenial amnesia. Brain 110:1631–1646.

Vann SD, Brown MW, Erichsen JT, Aggleton JP. 2000. Using fosimaging in the rat to reveal the anatomical extent of the disruptiveeffects of fornix lesions. J Neurosci 20:8144–8152.

Vann SD, Aggleton JP, Maguire EA. 2009a. What does the retrosple-nial cortex do? Nat Rev Neurosci 10:792–802.

Vann SD, Tsivilis D, Denby CE, Quamme JR, Yonelinas AP, AggletonJP, Montaldi D, Mayes AR. 2009b. Impaired recollection butspared familiarity in patients with extended hippocampal systemdamage revealed by 3 convergent methods. Proc Natl Acad SciUSA 106:5442–5447.

Vargha-Khadem F, Gadian DG, Watkins KE, Connelly A, VanPaesschen W, Mishkin M. 1997. Differential effects of early hippo-campal pathology on episodic and semantic memory. Science277:376–380.

Vilberg KL, Rugg MD. 2007. Dissociation of the neural correlates ofrecognition memory according to familiarity, recollection, andamount of recollected information. Neuropsychologia 45:2216–2225.

Vilberg KL, Rugg MD. 2008. Memory retrieval and the parietal cor-tex: A review of evidence from a dual-process perspective. Neuro-psychologia 46:1787–1799.

Vilberg KL, Rugg MD. 2009. Left parietal cortex is modulatedby amount of recollected verbal information. Neuroreport 20:1295–1299.

Viskontas IV, Knowlton BJ, Steinmetz PN, Fried I. 2006. Differencesin mnemonic processing by neurons in the human hippocampusand parahippocampal regions. J Cogn Neurosci 18:1654–1662.

Voss JL, Hauner KK, Paller KA. 2009. Establishing a relationshipbetween activity reduction in human perirhinal cortex and priming.Hippocampus 19:773–778.

Wagner AD. 1999. Working memory contributions to human learningand remembering. Neuron 22:19–22.

Wagner AD, Shannon BJ, Kahn I, Buckner RL. 2005. Parietal lobecontributions to episodic memory retrieval. Trends Cogn Sci9:445–453.

Wais PE. 2008. FMRI signals associated with memory strength in themedial temporal lobes: A meta-analysis. Neuropsychologia46:3185–3196.

Wais PE, Wixted JT, Hopkins RO, Squire LR. 2006. The hippocam-pus supports both the recollection and the familiarity componentsof recognition memory. Neuron 49:459–466.

Wais PE, Mickes L, Wixted JT. 2008. Remember/know judgmentsprobe degrees of recollection. J Cogn Neurosci 20:400–405.

Wais PE, Squire LR, Wixted JT. 2010. In search of recollection and fa-miliarity signals in the hippocampus. J Cogn Neurosci 22:109–123.

Wang W, Lazzara MM, Ranganath C, Knight RT, Yonelinas AP. Themedial temporal lobe supports conceptual implicit memory. Neuron(in press).

Wheeler ME, Buckner RL. 2004. Functional-anatomic correlates ofremembering and knowing. Neuroimage 21:1337–1349.

Wheeler ME, Petersen SE, Buckner RL. 2000. Memory’s echo: Vividremembering reactivates sensory-specific cortex. Proc Natl Acad SciUSA 97:11125–11129.

Witter MP, Naber PA, van Haeften T, Machielsen WC, RomboutsSA, Barkhof F, Scheltens P, Lopes da Silva FH. 2000a. Cortico-hip-pocampal communication by way of parallel parahippocampal-sub-icular pathways. Hippocampus 10:398–410.

Witter MP, Naber PA, van Haeften T, Machielsen WC, RomboutsSA, Barkhof F, Scheltens P, Lopes da Silva FH. 2000b. Cortico-hippocampal communication by way of parallel parahippocampal-subicular pathways. Hippocampus 10:398–410.

Witter MP, Wouterlood FG, Naber PA, Van Haeften T. 2000c. Ana-tomical organization of the parahippocampal-hippocampal net-work. Ann NY Acad Sci 911:1–24.

Wixted JT. 2007. Dual-process theory and signal-detection theory ofrecognition memory. Psychol Rev 114:152–176.

Wixted JT. 2009. Remember/Know judgments in cognitive neuro-science: An illustration of the underrepresented point of view.Learn Mem 16:406–412.

Wixted JT, Squire LR. The role of the human hippocampus in famili-arity-based and recollection-based recognition memory. BehavBrain Res (in press).

Wood ER, Dudchenko PA, Eichenbaum H. 1999. The global recordof memory in hippocampal neuronal activity. Nature 397:613–616.

Wood ER, Dudchenko PA, Robitsek RJ, Eichenbaum H. 2000. Hippo-campal neurons encode information about different types of mem-ory episodes occurring in the same location. Neuron 27:623–633.

Yonelinas AP. 1994. Receiver-operating characteristics in recognitionmemory: Evidence for a dual-process model. J Exp Psychol LearnMem Cogn 20:1341–1354.

Yonelinas AP. 1999. The contribution of recollection and familiarity torecognition and source-memory judgments: A formal dual-processmodel and an analysis of receiver operating characterstics. J ExpPsychol: Learn Mem Cogn 25:1415–1434.

Yonelinas AP. 2001. Consciousness, control, and confidence: The 3 Csof recognition memory. J Exp Psychol: Gen 130:361–379.

Yonelinas AP. 2002. The nature of recollection and familiarity: Areview of 30 years of research. J Mem Lang 46:441–517.

Yonelinas AP, Kroll NEA, Dobbins IG, Soltani M. 1999. Recognitionmemory of faces: When familiarity supports associative recognitionjudgments. Psychonom Bull Rev 6:654–661.

Yonelinas AP, Kroll NE, Quamme JR, Lazzara MM, Sauve MJ, Wida-man KF, Knight RT. 2002. Effects of extensive temporal lobe dam-age or mild hypoxia on recollection and familiarity. Nat Neurosci5:1236–1241.

Yonelinas AP, Otten LJ, Shaw KN, Rugg MD. 2005. Separating thebrain regions involved in recollection and familiarity in recognitionmemory. J Neurosci 25:3002–3008.

Yonelinas AP, Parks CM. 2007. Receiver operating characteristics (ROCs)in recognition memory: A review. Psychol Bull 133:800–832.

Zeineh MM, Engel SA, Bookheimer SY. 2000. Application of corticalunfolding techniques to functional MRI of the human hippocam-pal region. Neuroimage 11:668–683.

Zeineh MM, Engel SA, Thompson PM, Bookheimer SY. 2003. Dy-namics of the hippocampus during encoding and retrieval of face-name pairs. Science 299:577–580.

Zimmer HD, Ecker UK. 2010. Remembering perceptual featuresunequally bound in object and episodic tokens: Neural mechanismsand their electrophysiological correlates. Neurosci Biobehav Rev34:1066–1079.

MEDIAL TEMPORAL LOBES AND EPISODIC MEMORY 1289

Hippocampus

Zola SM, Squire LR. 2001. Relationship between magnitude of dam-age to the hippocampus and impaired recognition memory inmonkeys. Hippocampus 11:92–98.

Zola SM, Squire LR, Teng E, Stefanacci L, Buffalo EA, Clark RE.2000. Impaired recognition memory in monkeys after damage lim-ited to the hippocampal region. J Neurosci 20:451–463.

Zola-Morgan S, Squire L. 1985. Medial temporal lesions on monkeysimpair memory in a variety of tasks sensitive to human amnesia.Behav Neurosci 99:22–34.

Zola-Morgan S, Squire LR. 1986. Memory impairment in monkeysfollowing lesions limited to the hippocampus. Behav Neurosci100:155–160.

Zola-Morgan S, Squire L, Amaral D, Suzuki W. 1989a. Lesions ofperirhinal and parahippocampal cortex that spare the amygdala andthe hippocampal formation produce severe memory impairment.J Neurosci 9:4355–4370.

Zola-Morgan S, Squire LR, Amaral D. 1989b. Lesions of the hippo-campal formation but not lesions of the fornix or the mammilarynuclei produce long-lasting memory impairment in monkeys.J Neurosci 9:898–913.

Zola-Morgan S, Squire LR, Clower RP, Rempel NL. 1993. Damage tothe perirhinal cortex exacerbates memory impairment followinglesions to the hippocampal formation. J Neurosci 13:251–265.

1290 RANGANATH

Hippocampus