Noninvasive Functional and Anatomical Imaging of the …Noninvasive Functional and Anatomical Imaging of the Human Medial Temporal Lobe Thackery I. Brown 1, Bernhard P. Staresina ,

Noninvasive Functional and Anatomical Imagingof the Human Medial Temporal Lobe

Thackery I. Brown1, Bernhard P. Staresina1, and Anthony D. Wagner1,2

1Department of Psychology, Stanford University, Stanford, California 94305-21302Neurosciences Program, Stanford University, Stanford, California 94305-2130

Correspondence: [email protected]

The ability to remember life’s events, and to leverage memory to guide behavior, defines whowe are and is critical for everyday functioning. The neural mechanisms supporting suchmnemonic experiences are multiprocess and multinetwork in nature, which creates chal-lenges for studying them in humans and animals. Advances in noninvasive neuroimagingtechniques have enabled the investigation of how specific neural structures and networkscontribute to human memory at its many cognitive and mechanistic levels. In this review, wediscuss how functional and anatomical imaging has provided novel insights into the types ofinformation represented in, and the computations performed by, specific medial temporallobe (MTL) regions, and we consider how interactions between the MTL and other corticaland subcortical structures influence what we learn and remember. By leveraging imaging,researchers have markedly advanced understanding of how the MTL subserves declarativememory and enables navigation of our physical and mental worlds.

One of the central aims of cognitive neuro-science research is to understand how hu-

man brain function relates to the mnemonic ex-periences that define much of who we are asindividuals. Recent advances in noninvasive hu-man neuroimaging have given rise to novel in-sights about the neural foundations of humanmemory, and have allowed neuroscientists todraw important connections between humanand nonhuman animal research. Although non-invasive imaging techniques remain limited intheir spatial resolution (we cannot yet describethe behavior of individual neurons in the hu-man brain without implanting electrodes in pa-tients undergoing brain surgery), they also have

important strengths that have allowed research-ers to significantly advance mechanistic ac-counts of learning and memory. In this review,we begin with a historical perspective on nonin-vasive neuroimaging techniques and their appli-cation to the study of memory encoding. Wethen introduce more recent cutting-edge meth-odological approaches, as we discuss specific do-mains of memory theory that they have helpedadvance. Drawing on research examining themedial temporal lobe (MTL), we emphasizethe power of such imaging techniques to allowscientists to make inferences about the types ofmnemonic information represented by distinctbrain areas, and to understand how the func-

Editors: Eric R. Kandel, Yadin Dudai, and Mark R. Mayford

Additional Perspectives on Learning and Memory available at www.cshperspectives.org

Copyright # 2015 Cold Spring Harbor Laboratory Press; all rights reserved; doi: 10.1101/cshperspect.a021840

Cite this article as Cold Spring Harb Perspect Biol 2015;7:a021840

1

Harbor Laboratory Press at Stanford University Libraries on April 10, 2015 - Published by Cold Springhttp://cshperspectives.cshlp.org/Downloaded from

mailto:[email protected]

mailto:[email protected]

http://www.cshperspectives.org



http://cshperspectives.cshlp.org/

tions of different regions and neural networksunderlie our ability to learn and remember thedetails of our lives.

Although it can be challenging to assay whatnonverbal animals are thinking and why, work-ing directly with humans allows researchers todirectly probe the subjective experience of re-membering in people, and to relate their cogni-tion to underlying neural processes. For exam-ple, it is easy to ask a human participant aboutthe contents of their memories, or to assay howconfident they are about having previously en-countered a stimulus. In this review, we docu-ment how the ability to relate behavioral indicesof memory to measures of neural activity allowsresearch to advance our thinking about the bi-ological underpinnings of human cognition.We focus primarily on recent functional mag-netic resonance imaging (fMRI), and anatomi-cal magnetic resonance imaging (MRI) researchaddressing the role of MTL subregions in hu-man declarative memory (principally, episodicmemory for individual life events), and alsohighlight findings addressing how declarativememory mechanisms interact with those ofprocedural memory in service of efficient mem-ory-guided behavior.

EPISODIC MEMORY ENCODING

In the course of a day, humans encounter asteady stream of sensations, emotions, thoughts,and actions elicited by the external world andinternal states. Encoding mechanismstransformthis stream of information into long-term neuralrepresentations of co-occurring event features—that is, into episodic memory traces—and,thus, establish lasting footprints in our mindsof life’s events.

Imaging and the Subsequent MemoryParadigm

How to noninvasively study memory-encodingprocesses in humans is not a trivial question. Inhealthy individuals, we cannot directly measurethe synaptic and cellular dynamics that allow anexperience to be encoded such that it can be laterremembered. Instead, cognitive neuroscientists

have leveraged noninvasive imaging methods toidentify neural correlates of memory formationin humans, evidenced as neural predictors ofsubsequent memory expression (Paller et al.1987; Wagner et al. 1999; Paller and Wagner2002; Spaniol et al. 2009; Uncapher and Wagner2009; Kim 2011). Subsequent memory para-digms involve recording neural activity, usingeither electromagnetic or hemodynamic mea-sures, while participants encounter and processstimuli (e.g., a series of words) during an “en-coding period.” Subsequently, memory for eachstimulus is tested after a delay, and encodingperiod activity is examined conditioned on thebehavioral expressions of memory, such aswhether the stimulus was subsequently remem-bered or forgotten, or remembered with highor low confidence. Differences in encoding pe-riod activity as a function of later memoryoutcomes, such as subsequently rememberedand forgotten stimuli (termed “subsequentmemory effects” or difference as a result ofmemory—“Dm”—effects), can then be inter-preted in terms of their potential contributionsto the formation of memories.

Early evidence for subsequent memory ef-fects came from electroencephalography (EEG)research in humans (e.g., Sanquist et al. 1980;reviewed in Wagner et al. 1999). EEG is a non-invasive technique in which electrodes, placedon the scalp of a participant, passively recordvoltages at the surface of the scalp induced byion currents in the underlying cortex (Nieder-meyer and Lopes da Silva 2005). Because EEGsignals reflect electrical current from neural ac-tivity, the temporal resolution of EEG is on theorder of milliseconds, allowing researchers toaddress hypotheses about the time course ofneural events. One of the most common waysto analyze EEG data is to study the profile ofevent-related potentials (ERPs), which are elec-trical responses time-locked to or evoked bythe onset of a stimulus or response. By analyzingthe averaged ERP time-course associated withstimulus encoding, Sanquist and colleagues(1980) provided initial evidence of a subsequentmemory effect in humans, finding that itemsthat were subsequently remembered had amore positive deflection in the ERP time-course

T.I. Brown et al.

2 Cite this article as Cold Spring Harb Perspect Biol 2015;7:a021840



�450 to 750 msec after stimulus presentation(relative to items later forgotten).

Following this early finding, extensive EEGresearch corroborated the presence of multiplesubsequent memory effects underlying success-ful encoding (e.g., Long et al. 2014; reviewed inRugg 1995; Wagner et al. 1999; Friedman andJohnson 2000; Paller and Wagner 2002; Nyhusand Curran 2010). Among their discoveries,EEG studies showed that the encoding of mem-ories is a multifaceted process characterized byno single neural response. Instead, this richliterature revealed that there are distinct electro-physiological signatures associated with encod-ing, which manifest both in different poststimu-lus time periods and different spatial locations onthe scalp. Furthermore, distinct ERPs are associ-ated with different types of subsequent memory,such as memory for specific stimuli versus thecontextual details surrounding an encoding ex-perience (Johnson et al. 1997; Bridger and Wild-ing 2010; Angel et al. 2013). Such data suggestthat multiple neural structures and mechanismsunderlie successful episodic encoding.

Although providing some leverage on thetemporal dynamics of encoding activity, EEGdata present an important challenge for under-standing the neural bases of memory formationin humans. As a measure of electrical signal atthe scalp, EEG is most sensitive to postsynapticpotentials generated in superficial layers of thecortex. Furthermore, the signals from millionsof neurons in adjacent brain areas converge andare diffuse at the level of the scalp. As a result, al-though EEG provides excellent resolution about“when” a neural response happens, it is chal-lenging to discern “where” this processing occurs.This limitation is less pronounced in fMRI,which suffers from low temporal resolutionbut affords spatial resolution at the millimeterscale. fMRI inherits these strengths and limita-tions from its use of oxygenated blood flow asan indirect measure of neural activity. Themechanistic link between the blood oxygen lev-el–dependent (BOLD) fMRI signal and neuralactivity remains an active area of research (Lo-gothetis and Wandell 2004). Most fMRI studiesobtain data on the order of 3 to 4 mm3 resolu-tion, and the technology is continually improv-

ing, with recent advances in MRI hardware andfMRI pulse sequences allowing for whole-brainimaging at substantially higher spatial and tem-poral resolution (Feinberg et al. 2010; Moelleret al. 2010). For example, acquisition and anal-ysis techniques now allow for individual subjectand group-level analyses of human brain activ-ity at spatial resolutions approaching 1 mm(Zeineh et al. 2000; Grill-Spector et al. 2006;Ekstrom et al. 2009; Yassa and Stark 2009; Yassaet al. 2010), and in some cases even higher (Ya-coub et al. 2008; Heidemann et al. 2012), posi-tioning fMRI as a powerful tool for researchersto noninvasively pinpoint where activity relatedto memory, perception, and cognition occurs inthe brain.

For memory researchers, this increased spa-tial resolution is particularly beneficial as it al-lows fine-grained investigation of functionaldistinctions within the human MTL (e.g., Kir-wan et al. 2007; Bakker et al. 2008; Ekstrom et al.2009; reviewed in Carr et al. 2010; Chadwick etal. 2011; Bonnici et al. 2012, 2013; Libby et al.2012; LaRocque et al. 2013; Brown et al. 2014a).The MTL has long been understood to be crit-ical for episodic memory, based on decades ofresearch that was spurred by the landmark caseof Henry Molaison (HM), whose ability to re-member new experiences from his daily life wasdramatically impaired following bilateral resec-tion of the hippocampus, and portions of theneighboring MTL cortical structures (perirhi-nal, parahippocampal, and entorhinal cortex)(Scoville and Milner 1957; Corkin 2013; Squireand Dede 2015). Although studies of HM andother patients suffering MTL lesions have yield-ed a multitude of novel insights about the roleof the MTL in declarative memory (Eichen-baum and Cohen 2001; Squire et al. 2004; Mos-covitch et al. 2006; Squire and Bayley 2007; Gra-ham et al. 2010; Greenberg and Verfaellie 2010;Montaldi and Mayes 2010; Rosenbaum et al.2014), a challenge for neuropsychological stud-ies is to identify whether the observed memorydeficits reflect impairments at encoding, re-trieval, or both. Noninvasive human imagingtechniques complement lesion studies, as theyprovide a critical set of tools for studyinghealthy brain function by measuring neural re-

Noninvasive Functional Anatomical Imaging of MTL

Cite this article as Cold Spring Harb Perspect Biol 2015;7:a021840 3



sponses at the time memories are encoded and,as we will later discuss, at the time that they areretrieved. Indeed, the first fMRI studies usingthe subsequent memory paradigm (Brewer et al.1998; Wagner et al. 1998) provided early evi-dence that event-related levels of encoding pe-riod activation in the human MTL, as well as inthe lateral prefrontal cortex, predict whether astimulus will be later remembered or forgotten.We next consider how fMRI subsequent mem-ory data have informed theory about MTLmnemonic function.

Testing Theories of MTL FunctionalDifferentiation

Episodic memories often contain a remarkablewealth of detail from an object that caught ourattention to the environment in which we en-countered it. At retrieval, sometimes we onlyhave a vague sense of recognition when viewinga previously encountered stimulus, whereas, atother times, we vividly recollect many details ofthe prior experience. A number of theoreticalframeworks have been formulated in an effortto capture the link between this diversity in mne-monic experience and the functions of differentsubregions of the MTL, most notably the hippo-campus, perirhinal cortex, and parahippocam-pal cortex (Cohen and Eichenbaum 1993; Mur-ray and Bussey 1999; Brown and Aggleton 2001;Davachi 2006; Diana et al. 2007; Eichenbaum etal. 2007; Mayes et al. 2007; Graham et al. 2010).

Leveraging the subsequent memory para-digm combined with fMRI’s ability to distin-guish activity from distinct, but spatially prox-imal MTL cortical regions (Carr et al. 2010),researchers have begun to test competing hy-potheses about how specific MTL subregionscontribute to episodic memory. For example,fMRI studies have shown content specializationwithin the MTL cortex that is predictive of sub-sequent memory. Specifically, consistent withtheir differing connectivity with other corticalareas (Suzuki 2009), activity in the parahippo-campal cortex at encoding has been shown topredict later memory for scenes, while activityin the perirhinal cortex at encoding predictslater memory for faces, objects (Litman et al.

2009; Preston et al. 2010; Staresina and Davachi2010; Staresina et al. 2011), and their associa-tions (Staresina and Davachi 2008; Watson et al.2012). Convergent evidence suggests that suchfunctional differentiation along the anterior–posterior axis of the MTL cortex is best under-stood as a continuous gradient; for example,Liang and colleagues (2013) showed that face/object and scene representations are coded todiffering degrees across the MTL cortex, withthe greatest specialization for scene memoryin the posterior parahippocampal region andgreatest specialization for face/object process-ing in the perirhinal cortex (see also Lee et al.2008; Barense et al. 2010).

Although specific types of event content ap-pear to be differentially represented along theanterior–posterior axis of the MTL cortex, theconvergence of perirhinal and parahippocam-pal inputs on the hippocampus, via the entorhi-nal cortex, is thought to enable the binding ofthe distinct facets (i.e., “items” or “items andcontext”) of an event into a conjunctive memorytrace. This theoretical perspective has garneredsome support from fMRI subsequent memorystudies, which have revealed that MTL corticalactivity at encoding (principally in the perirhi-nal cortex) differentially predicts later itemrecognition memory, whereas hippocampal ac-tivity at encoding differentially predicts latermemory for item–context and item–item asso-ciations (e.g., Davachi et al. 2003; Kirwan andStark 2004; Ranganath et al. 2004).

Multivariate fMRI Analyses and MemoryTheory

The high spatial resolution of MRI has recentlybeen combined with multivariate analysis tech-niques (Norman et al. 2006; Kriegeskorte etal. 2008), providing a powerful new means toaddress the mechanisms giving rise to, and therepresentational contents of, episodic memo-ries. One such technique—representationalsimilarity analysis (RSA)—has been particular-ly informative for testing hypotheses about theneural mechanisms that support successful en-coding. Briefly, RSA measures the similarity(correlation) between event- or stimulus-spe-

T.I. Brown et al.




cific patterns of activation, which can be used toquery how the similarity (or dissimilarity) ofmultivoxel patterns measured during encodingrelate to subsequent memory. As we next illus-trate, such measures provide important tests oftheoretical predictions about MTL functionaldifferentiation underlying memory.

One dominant theoretical perspective onthe functional contributions of different MTLsubregions to memory, the “complementarylearning systems” model, holds that a hippo-campal “pattern separation” mechanism sup-ports memory for events by orthogonalizing(making more distinct) neural representationsduring encoding, allowing later retrieval cues totrigger memory for a unique event (O’Reillyand McClelland 1994; McClelland et al. 1995;O’Reilly and Rudy 2001; Norman and O’Reilly2003; Norman 2010). Conversely, this perspec-tive holds that the neighboring MTL cortex sup-

ports memory by gradually encoding neuralrepresentations that capture the commonalitiesacross similar stimuli, permitting later item rec-ognition on the basis of global similarity be-tween the present and past. Using RSA, coupledwith high-resolution fMRI of the MTL, LaRocque and colleagues (2013) computed thesimilarity of the multivoxel fMRI pattern elicit-ed by a stimulus at encoding to the patternselicited by other encoded stimuli. By computingthese across-item pattern similarities for eachMTL subregion—perirhinal cortex, parahippo-campal cortex, and these researchers obtainedstrong support for the “complementary learn-ing systems” framework: namely, greater across-item similarity in the MTL cortex, but reducedacross-item similarity in the hippocampus, waspredictive of later successful memory (Fig. 1).

RSA has also been used to gain leverage on alongstanding debate about whether the stabil-

1.0

0.8

0.6

0.4

0.2

0.0–5 0

Similarity score (z)

Hipp *** PRc *

5

1.0

0.8

0.6

0.4

0.2

0.0–5 0

Similarity score (z)

Parahippocampal cortexPerirhinal cortex

rwrw

ra

ra

Hippocampus

5

Hipp *** PHc **

A C

B D

P (

rem

embe

red)

rw

P (

rem

embe

red)

A

B

Figure 1. Pattern similarity and separation in different medial temporal lobe (MTL) subregions supportsencoding. (A) Anatomically defined hippocampus, perirhinal cortex, and parahippocampal cortex regions ofinterest, overlaid on a structural magnetic resonance imaging (MRI). (B) Across-stimuli representational sim-ilarity (within category, rw; across category, ra) was computed using Pearson correlations between the patternsof neural activity, across voxels, for pairs of stimuli. (C,D) Critically, logistic regression revealed that thesimilarity of an item’s encoding pattern to those of other items in the perirhinal cortex (PRc) and parahippo-campal cortex (PHc) positively predicted subsequent memory. Conversely, hippocampal (Hipp) pattern sim-ilarity negatively predicted subsequent memory. �p , 0.05; ��p , 0.01; ��p , 0.005. (From LaRocque et al.2013; adapted from several in the source, with permission, from the Society for Neuroscience # 2013.)





ity of a neural representation across encodingexperiences is beneficial for subsequent memo-ry, or whether variability in the neural repre-sentations of a stimulus across experiences isbeneficial (the “encoding-variability” hypothe-sis) (Martin 1968). In particular, RSA analysesof fMRI encoding data suggest that, in somecontexts, representational stability may be ben-eficial to later remembering (Xue et al. 2010;Ward et al. 2013; but see Wagner et al. 2000),demonstrating that greater pattern similarity ofan item’s neural representations across multi-ple encoding trials predicts better subsequentmemory for the item. Although some questionsremain (Xue et al. 2013; Davis et al. 2014), re-searchers are now positioned to measure, at theindividual trial level and within an individualhuman brain, the large-scale distributed neuralrepresentations that underlie important aspectsof memory behavior.

A related multivariate technique—multi-voxel pattern analysis (MVPA)—has also beeneffectively leveraged to advance understand-ing of episodic memory, including testing theprediction that the “strength” of a neural repre-sentation at encoding predicts later memory ex-pression. In MVPA, multivoxel activation pat-terns for two or more classes of stimuli (e.g.,faces vs. scenes) are used to train a “classifier”to identify the characteristic activity patternsthat maximally discriminate between the stim-ulus classes. As such, from training data, a clas-sifier learns to partition neural patterns intoclass-labeled decision regions (e.g., patterns rep-resentative of faces and patterns representative ofscenes), and then can be used to estimate where anovel test pattern falls with respect to the bound-aries between these decision regions. Thus,when presented with a new test pattern, the clas-sifier is used to predict to which of the learnedclasses the new event belongs (Norman et al.2006; Rissman and Wagner 2012). Importantly,pattern classifiers can output probabilistic pre-dictions about the new event’s likely class, whichprovides a trial-specific quantitative measure ofthe strength of neural evidence. Recently, re-searchers have shown that the “strength” of con-tent-specific (i.e., face vs. scene) neural evidencein the visual cortex correlates with the magni-

tude of hippocampal univariate fMRI activity atencoding and, critically, predicts whether thatinformation will later be remembered or forgot-ten (Kuhl et al. 2012; Gordon et al. 2013). Thesedata suggest that the success of hippocampallymediated encoding of event details is influencedby, or at least covaries with, the strength or fidel-ityof the corresponding cortical representations.We expect that these, and other recent MVPAand RSA observations (Johnson et al. 2009;McDuff et al. 2009; Rissman et al. 2010; Wardet al. 2013), will be the first of many instances,in which the application of multivariate analyt-ic techniques, combined with the spatial resolu-tion of fMRI, allows researchers to make criticalprogress on open questions about the neurobi-ological mechanisms governing memory.

EPISODIC MEMORY RETRIEVAL

At its core, retrieval can be considered as thereinstatement or reconstruction of informationthat was encoded in memory (Dudai 2015).With the ability to measure large-scale neuralnetworks involved in retrieval, researchersstudying memory in humans are able to consid-er not only where features of encoded memoriesare “stored” in the brain, but which regions andprocesses contribute to what is successfully re-membered (and in how much detail). In thissection, we discuss noninvasive functional im-aging findings that provide critical insights into(1) the role of MTL subregions at retrieval, and(2) how distributed neural networks can inter-act to guide goal-directed memory retrieval andmemory-guided behavior.

Multivariate fMRI and Connectivity Analysesof Episodic Retrieval

As with encoding, multivariate pattern analysesprovide a powerful approach to examine theexpression of memory content at retrieval. Forinstance, early MVPA work from Polyn et al.(2005) showed that a classifier, trained to dis-tinguish between neural patterns elicited by theencoding of face, object, and scene classes ofstimuli, can predict which category of stimuliwill be imminently freely recalled based on

T.I. Brown et al.




shifts in these categorical neural patterns dur-ing retrieval. Other data indicate that the repre-sentational strength of face and scene patternsin the visual cortex during encoding is not onlypredictive of subsequent memory behavior, butalso of the strength of pattern reinstatement atretrieval (Gordon et al. 2013), providing an im-portant demonstration of the link between rep-resentations of experiences as they unfold and ofmemory for those experiences as they are re-trieved.

Episodic memories are characterized by therepresentation of unique experiences in ourlives. One strength of RSA for assessing episod-ic reinstatement lies in the ability to examinestimulus- or event-specific representations atretrieval. For instance, in a recent fMRI study,participants were presented with word–scenecombinations during encoding and then, at re-trieval, they were provided with only the word asa retrieval cue. It was found that stimulus-spe-cific scene patterns were reactivated in the para-

hippocampal cortex when participants indicat-ed they had recollected the target scene, but notwhen they indicated they had no recollection(Fig. 2) (Staresina et al. 2012; see also, Ritcheyet al. 2008; Bosch et al. 2014). This is an impor-tant step toward demonstrating that episodicreinstatement in the MTL can be observed atthe level of individual memory representations.

Although multivariate pattern analyses fa-cilitate the study of representational contentin specific brain regions, functional connectiv-ity measures provide a means of noninvasivelystudying how different regions and networksinteract in support of memory and cognition.Connectivity measures leverage variability inthe magnitude and timing of fMRI responsesacross trials to index functional relationships be-tween distinct regions (e.g., Friston et al. 1993,2003; McIntosh and Gonzalez-Lima 1994; Cor-des et al. 2000; Lowe et al. 2000; Greicius et al.2003; Rissman et al. 2004; Sun et al. 2004; re-viewed in Stephan and Friston 2010; Friston

0.08*

Encoding/retrieval similarity (ERS)Anatomically defined hippocampus (HIPP)and parahippocampal cortex (PhC)

A B

0.07

0.06

0.05

0.04

0.03

0.02

Pea

rson

cor

rela

tion

coef

ficie

nt

0.01

0.00

Word andscene

Word andscene

PhC HIPP

Word only

Word only

ns

Figure 2. Encoding period patterns in the medial temporal lobe (MTL) are reinstated during successful retrieval.(A) Anatomically defined hippocampus and parahippocampal cortex regions of interest, overlaid on a structuralmagnetic resonance imaging (MRI). (B) Representational similarity (Pearson correlation) in the parahippo-campal cortex (PhC) showed that when word–scene stimulus pairs are successfully recollected, neural patternsat retrieval are significantly more similar to those elicited for the unique word–scene combinations duringencoding than when only the word is remembered. These reinstatement measures were not present in thehippocampus (HIPP), but trial-by-trial hippocampal activity predicted reinstatement, consistent with patterncompletion mechanisms supporting retrieval. ns, not significant; �p , 0.05. (From Staresina et al. 2012;reprinted, with permission, from the Society for Neuroscience 2012.)





2011; Friston et al. 2013) and, within the con-text of memory, their relation to memory be-havior. Such functional relationships can be ei-ther direct (i.e., activity in one region directlyinfluences activity in the other) or indirect (i.e.,mediated by the function of another region).The ability to study large-scale networks acrossthe brain is a particular strength of noninvasivefunctional imaging measures in humans.

Connectivity measures provide an impor-tant means for studying how information isshared within the MTL system, and betweenthe MTL and other networks in the brain. Inthe preceding section on encoding, we dis-cussed how different features of memories, spe-cifically object/item information and scene/spatial information, are differentially represent-ed by the perirhinal and parahippocampal cor-tex, respectively. The integration of these signalsvia the hippocampus is theorized to be a keyneurobiological step in episodic memory en-coding, and provides a basis for stimuli to cueassociated memories at retrieval (Marr 1971;Teyler and DiScenna 1986; Treves and Rolls1994). For example, how is it that when welook at our desk lamp it can trigger a memoryfor the location where the lamp was purchasedlast week? Researchers have posited that thismight occur via a systematic flow of informa-tion, whereby information about the lampmight elicit a spread of activity through associ-ations in the hippocampus (known as “patterncompletion”) that leads to reactivation of theother representations associated with the lamp.A recent development in functional connectivi-ty approaches, known as effective connectivity,allows researchers to test evidence for such direc-tional predictions about the flow of informationin brain networks (McIntosh and Gonzalez-Lima 1994; Stephan and Friston 2010; Smithet al. 2011; Friston et al. 2013). Effective connec-tivity measures seek to support inferences aboutcausality in connectivity, by modeling the fit ofpredicted directional relationships with an fMRIsignal from a target network of brain regions.Using one form of effective connectivity, knownas dynamic causal modeling (DCM) (Fristonet al. 2003), researchers have provided evidencethat retrieval of a memory from a cue arises from

information transfer within the MTL. Specifi-cally, activity caused by processing a scene cuein the parahippocampal cortex can drive activityfor an associated object in the perirhinal cortex,with this interaction being mediated by the hip-pocampus (Staresina et al. 2013). These dataprovide novel support for a fundamental predic-tion about the role of the hippocampus in me-diating the link between segregated item or fea-ture representations (here, scene and objectinformation) in support of rich episodic re-membering.

SPATIAL MEMORY AND THECONTRIBUTIONS OF MULTIPLEMEMORY SYSTEMS

The functional interaction between separatespatial and nonspatial representations, mediat-ed via the hippocampus, has important impli-cations for the direction of future memory re-search. Although the early findings in patientHM provided a framework for decades of re-search on the role of the hippocampus in epi-sodic memory, the existence of “place cells” inthe rodent hippocampus (O’Keefe and Dostrov-sky 1971; O’Keefe 1976; Moser et al. 2008, 2015)led to a separate hypothesis that the hippocam-pus creates internal “cognitive maps” of envi-ronments (Tolman 1948; O’Keefe and Nadel1978). Although research examining the roles ofthe hippocampus in episodic and spatial mem-ories increasingly crosses paths, these have his-torically remained two distinct areas of studyin human cognitive neuroscience. One meansof bridging the gap between these areas of re-search is to consider hippocampal representa-tions of location as one mechanism underlyingits broader role in associating stimuli and expe-riences across space and time (Eichenbaum andCohen 2014).

In real-world scenarios, episodic memoriesencompass the “who, what, when, and where” ofan experience and, thus, require the ability toembed nonspatial information (e.g., faces andobjects) in memory for environments (e.g., Bur-gess et al. 2001; reviewed in Burgess et al. 2002;Bird and Burgess 2008). The effective connec-tivity data described in the preceding section

T.I. Brown et al.




(Staresina et al. 2013) are an important step to-ward understanding how spatial and nonspatialmemory signals combine within the hippocam-pus to support the expression (in addition to theencoding and construction) of such integratedknowledge. Moreover, recent high-resolutionfMRI data show that, when scene informationis presented as a cue for memory of a specificnavigational episode, trial-by-trial responses inthe parahippocampal cortex and the hippo-campal CA1 subfield during cue processing cor-relate with prospective retrieval of the desirednavigational event (Brown et al. 2014a). CA1 rep-resents a final stage of processing in the hippo-campal circuit, and theories of hippocampalfunction propose that the convergence in CA1of representations from the CA3 subfield and theMTL cortex facilitates sequential retrieval, andgates hippocampal output to memories that arecongruent with current context (Hasselmo andWyble 1997; Hasselmo and Eichenbaum 2005;Kesner 2007). Brown and colleagues’ high-reso-lution fMRI data, among other recent navigationwork (Brown et al. 2010; Brown and Stern 2013),suggest that parahippocampal scene representa-tions could underlie hippocampal reactivationof navigational episodes, and support key pre-dictions about CA1’s role in the flexible retrievalof goal-relevant memories. Moreover, comple-mentary fMRI data (Suthana et al. 2009) haveassociated CA1 activity with the learning of al-locentric (map level) representations of envi-ronments; such flexible spatial representationsmay be critical as a scaffold for rememberingthe locations of, and relationships between, spe-cific events in our lives.

Building from the hypothesis that the hip-pocampus represents cognitive maps of envi-ronments, spatial navigation research has oftencontrasted hippocampal-dependent memoryfor complex spatial relationships between loca-tions with navigation based on striatal-depen-dent motor associations for specific cues andlandmarks (Hartley et al. 2003; Iaria et al.2003; Doeller et al. 2008). Noninvasive measuresof anatomical morphology have recently linkedthese concepts. Specifically, using one class ofanatomical MRI analysis known as voxel-basedmorphometry (VBM) (Mechelli et al. 2005),

researchers have shown that volume estimatesin the hippocampus and caudate nucleus dif-ferentially correlate with the predisposition of aperson to rely on spatial knowledge or response-based strategies to solve navigational problems(Bohbot et al. 2007; Konishi and Bohbot 2013),as well as with the level of an individual’s “exper-tise” as a spatial navigator (Maguire et al. 2006).Briefly, VBM leverages regional volumetric dif-ferences between anatomical MRI images foreach participant and a standardized templatebrain. By examining the degree to which a brainregion (e.g., hippocampus) must be enlarged orcompressed to fit the template brain, researcherscan infer volumetric differences between partic-ipants in their dataset. VBM analyses have beencombined with automated methods for seg-menting MRI images into gray matter structures(Fischl et al. 2002; Patenaude et al. 2011) to showthat hippocampal volume predicts an individu-al’s ability to learn and remember map-level in-formation (Hartley and Harlow 2012; Schinaziet al. 2013) (such techniques have also been usedwith nonspatial memory paradigms to showfunctional specialization within the hippocam-pus—linking posterior hippocampal volume,specifically, with contextual memory perfor-mance [Poppenk and Moscovitch 2011]). Simi-larly, researchers have used diffusion tensor im-aging (a method for tracking water moleculediffusion along white matter tracts in the brain)to show that greater directionality in water mol-ecule diffusion in the hippocampus, putativelyindicative of greater white matter integrity andorganization, correlates with improved ability ofparticipants to learn and retrieve cognitive mapinformation (Iaria et al. 2008).

Critically, in the real world, we need a mech-anism for flexibly translating both spatial andnonspatial forms of episodic memory into goal-directed actions. The hippocampus is not ana-tomically positioned to directly control motorbehavior, and early evidence from rodent navi-gation studies (Devan and White 1999) led tothe prediction that the hippocampus may directbehavior via engagement with striatal circuitry.Therefore, although hippocampal and striatalforms of memory may differ in fundamentalways (see Graybiel 2015), given that navigation-





al memories are often complex and can incor-porate both spatial and behavioral information,our ability to navigate in real-world settingsmay draw on both regions; more generally, in-tegration of MTL and frontostriatal computa-tions may be important for memory and mem-ory-guided behavior in many scenarios as afunction of their combined relevance to currenttask demands. The interplay between these sys-tems can be compensatory; for example, fMRIresearch in Huntington’s disease patients, whosuffer from progressive pathology affectingstriatal circuitry, has shown that route naviga-tion—potentially supported by response asso-ciations in the striatum (Hartley et al. 2003)—can fall back on hippocampal computations asthe striatal system fails (Voermans et al. 2004).Importantly, navigation of familiar routes inhealthy populations can also rely on the hippo-campus when navigational responses dependon explicit knowledge of the current naviga-tional context, that is, when navigation drawsmore strongly on features of episodic memory.

A real-world example that most of us arefamiliar with is the experience of traversing anintersection between two familiar navigationalroutes. In this scenario, we need to choose be-tween two possible directions based on memoryfor which path is most relevant to the goal of thecurrent navigational episode. Consistent withthe episodic memory demands of this scenario,recent VBM data show that hippocampal graymatter volume in young adults correlates withthe ability to perform such context-dependentroute navigation (Fig. 3A) (Brown et al. 2014b).Moreover, fMRI research has shown that peoplefaced with alternative route memories in a vir-tual navigation task draw on both hippocampaland striatal processes to identify and selectwhich path to take (Brown et al. 2010). In fact,not only do the hippocampus and the medialcaudate (a striatal subregion implicated in be-havioral flexibility) support learning of naviga-tional episodes (Brown and Stern 2013), buttheir recruitment during navigation of familiarroutes increases as memories for new alternative

0.2 0.4

VB

M g

ray

mat

ter

volu

me

0.1

0.2

0.3

0.4

0.6 0.8 1.0

0

0.5

1

1.5

2

2.5

3

3.5

OLold

OLnew

Start

Start

End

End

Y

Y YY

Y

Y

Y

t

0.70.60.50.40.30.20.1

0–0.1–0.2–0.3

Early Late

ROI: Left hippocampal tail

Par

amet

er e

stim

ates

ROI: Right medial caudate

0.80.60.40.2

0

–1–0.8–0.6–0.4–0.2

Early Late

Par

amet

er e

stim

ates *

*

ROI: Left medial caudate

0.80.60.40.2

0

–1–0.8–0.6–0.4–0.2

Early Late

Par

amet

er e

stim

ates

*

0.2 0.4

VB

M g

ray

mat

ter

volu

me

0.4

0.5

0.7

0.8

0.6 0.8 1.0

0.6

Right hippocampusLeft hippocampus

A BR R

OLoldNOLold

Figure 3. Hippocampus and striatum support flexible navigational learning and memory. (A) Statistical map ofbilateralhippocampalregion, inwhichvolumepositivelycorrelatedwithcontext-dependentrouteretrievalability inhealthy young adults (upper frames). Volume estimates extracted from the hippocampus tightly predict individualdifferences inperformance(lower frames).(B)Navigationoffamiliarroutes(overlapping[OL]old;bluepathonmap)becomes increasingly reliant on hippocampal and striatal mechanisms as a novel interfering route memory isintroduced (OLnew; green path on map)—fMRI activity in the left hippocampus and bilateral medial caudateincreased from early to late navigation trials (blue line in graphs) as participants became more familiar with thenovel competing route memory. In contrast, navigation of familiar nonoverlapping routes (nonoverlapping[NOL]old; red line in graphs), for which contextual retrieval demands were limited, relied minimallyon these medialtemporal lobe (MTL) and striatal subregions from early to late trials (even decreasing for the medial caudate withcontinued practice). ROI, region of interest; VBM, voxel-based morphometry; R, right hemisphere; �p , 0.05.(From Brown et al. 2014b and Brown and Stern 2013; adapted, with permission, from the authors.)

T.I. Brown et al.




paths are introduced to the environment, in-creasing the need to draw on episodic memoryto guide selection of behavior (Fig. 3B). Impor-tantly, functional connectivity research suggeststhat the hippocampus and striatal circuitry in-teract cooperatively at retrieval when peopleneed to use episodic memory to guide naviga-tion (Brown et al. 2012). (Recent evidence alsosuggests that corecruitment of, and functionalinteractions between, the hippocampus and stri-atum may also be important in non-navigationsettings for forming episodic memories [Ben-Yakov and Dudai 2011; Sadeh et al. 2011] andmaking episodic and relational memory judg-ments [Moses et al. 2010; Ross et al. 2011]). Fur-thermore, Brown and colleagues (2012) showedthat the hippocampus and striatum functionallyinteract with regions of the prefrontal cortexduring contextual navigation, suggesting thatthe prefrontal cortex may mediate goal-directedmemory and the interaction between these sys-tems. Ultimately, this line of research illustrateshow episodic memory supported by the hippo-campus plays a critical part in spatial navigation,and demonstrates that the distinct functions ofthe MTL and striatal systems can combine tosupport navigation in real-world scenarios inhumans; more broadly, the imaging data de-scribed in this section have laid importantgroundwork for understanding (1) how we le-verage multiple memory systems to achievelong-term goals, and (2) the importance of net-work interactions for constructing and navigat-ing mnemonic representations of our lives.

CONCLUSIONS

The ability to remember an event from last weekor to plan which route to take to the grocerystore are fundamentally multiprocess andmultinetwork acts, integrating declarativememory with systems of attention, and cogni-tive and behavioral control. Noninvasive func-tional imaging techniques are essential for ad-vancing memory neuroscience as a field,enabling the study of human memory at itsmany cognitive and mechanistic levels. Thesetechniques have yielded novel insights into thetypes of information represented in distinct

brain regions, the mnemonic computationsthat specific regions perform, and how the func-tions of different regions interact to influencewhat we learn and remember about our world.Moreover, data addressing functional organiza-tion and neural network interactions from hu-man research can serve as a crucial guide fordirecting neural recordings, invasive high-reso-lution imaging, and genetic and pharmacolog-ical manipulations in nonhuman animals. Byleveraging these techniques to study how repre-sentational features, functional interactions,and anatomical morphology of brain areas sup-port mnemonic experience, researchers will beable to better understand the neural basis ofdeclarative memory and how changes associat-ed with both development and disease affectthis core element of the human condition.

ACKNOWLEDGMENTS

Supported by grants from the Wallenberg Foun-dation’s Network Initiative on Culture, Brain,and Learning (A.D.W.) and the Wellcome Trust(B.P.S.).

REFERENCES�Reference is also in this collection.

Angel L, Isingrini M, Bouazzaoui B, Fay S. 2013. Neuralcorrelates of encoding processes predicting subsequentcued recall and source memory. Neuroreport 24: 176–180.

Bakker A, Kirwan CB, Miller M, Stark CE. 2008. Patternseparation in the human hippocampal CA3 and dentategyrus. Science 319: 1640–1642.

Barense MD, Henson RNA, Lee ACH, Graham KS. 2010.Medial temporal lobe activity during complex discrim-ination of faces, objects, and scenes: Effects of viewpoint.Hippocampus 20: 389–401.

Ben-Yakov A, Dudai Y. 2011. Constructing realistic engrams:Poststimulus activity of hippocampus and dorsal stria-tum predicts subsequent episodic memory. J Neurosci 31:9032–9042.

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

Bohbot VD, Lerch J, Thorndycraft B, Iaria G, Zijdenbos AP.2007. Gray matter differences correlate with the sponta-neous strategies in a human virtual navigation task. JNeurosci 27: 10078–10083.

Bonnici HM, Kumaran D, Chadwick MJ, Weiskopf N, Has-sabis D, Maguire EA. 2012. Decoding representations of





scenes in the medial temporal lobes. Hippocampus 22:1143–1153.

Bonnici HM, Chadwick MJ, Maguire EA. 2013. Represen-tations of recent and remote autobiographical memoriesin hippocampal subfields. Hippocampus 23: 849–854.

Bosch SE, Jehee JFM, Fernandez G, Doeller CF. 2014. Rein-statement of associative memories in early visual cortex issignaled by the hippocampus. J Neurosci 34: 7493–7500.

Brewer JB, Zhao Z, Desmond JE, Glover GH, Gabrieli JD.1998. Making memories: Brain activity that predicts howwell visual experience will be remembered. Science 281:1185–1187.

Bridger EK, Wilding EL. 2010. Requirements at retrievalmodulate subsequent memory effects: An event-relatedpotential study. Cogn Neurosci 1: 254–260.

Brown MW, Aggleton JP. 2001. Recognition memory: Whatare the roles of the perirhinal cortex and hippocampus?Nat Rev Neurosci 2: 51–61.

Brown TI, Stern CE. 2013. Contributions of medial tempo-ral lobe and striatal memory systems to learning andretrieving overlapping spatial memories. Cereb Cortex24: 1906–1922.

Brown TI, Ross RS, Keller JB, Hasselmo ME, Stern CE. 2010.Which way was I going? Contextual retrieval supportsthe disambiguation of well learned overlapping naviga-tional routes. J Neurosci 30: 7414–7422.

Brown TI, Ross RS, Tobyne SM, Stern CE. 2012. Cooperativeinteractions between hippocampal and striatal systemssupport flexible navigation. Neuroimage 60: 1316–1330.

Brown TI, Hasselmo ME, Stern CE. 2014a. A high resolutionstudy of hippocampal and medial temporal lobe corre-lates of spatial context and prospective overlapping routememory. Hippocampus 24: 819–839.

Brown TI, Whiteman AS, Aselcioglu I, Stern CE. 2014b.Structural differences in hippocampal and prefrontalgray matter volume support flexible context-dependentnavigation ability. J Neurosci 34: 2314–2320.

Burgess N, Maguire EA, Spiers HJ, O’Keefe J. 2001. A tem-poroparietal and prefrontal network for retrieving thespatial context of lifelike events. Neuroimage 14: 439–453.

Burgess N, Maguire EA, O’Keefe J. 2002. The human hip-pocampus and spatial and episodic memory. Neuron 35:625–641.

Carr VA, Rissman J, Wagner AD. 2010. Imaging the humanmedial temporal lobe with high-resolution fMRI. Neuron65: 298–308.

Chadwick MJ, Hassabis D, Maguire EA. 2011. Decodingoverlapping memories in the medial temporal lobes us-ing high-resolution fMRI. Learn Mem 18: 742–746.

Cohen NJ, Eichenbaum HB. 1993. Memory, amnesia, andthe hippocampal system. MIT Press, Cambridge, MA.

Cordes D, Haughton VM, Arfanakis K, Wendt GJ, TurskiPA, Moritz CH, Quigley MA, Meyerand ME. 2000. Map-ping functionally related regions of brain with functionalconnectivity MR imaging. AJNR Am J Neuroradiol 21:1636–1644.

Corkin S. 2013. Permanent present tense: The unforgettablelife of the amnesic patient. Basic Books, New York.

Davachi L. 2006. Item, context and relational episodic en-coding in humans. Curr Opin Neurobiol 16: 693–700.

Davachi L, Mitchell JP, Wagner AD. 2003. Multiple routes tomemory: Distinct medial temporal lobe processes builditem and source memories. Proc Natl Acad Sci 100: 2157–2162.

Davis T, LaRocque KF, Mumford J, Norman KA, WagnerAD, Poldrack RA. 2014. What do differences betweenmulti-voxel and univariate analysis mean? How subject-,voxel-, and trial-level variance impact fMRI analysis.Neuroimage 97: 271–283.

Devan BD, White NM. 1999. Parallel information in pro-cessing in the dorsal striatum: Relation to hippocampalfunction. J Neurosci 19: 2789–2798.

Diana RA, Yonelinas AP, Ranganath C. 2007. Imaging rec-ollection and familiarity in the medial temporal lobe: Athree-component model. Trends Cogn Sci 11: 379–386.

Doeller CF, King JA, Burgess N. 2008. Parallel striatal andhippocampal systems for landmarks and boundaries inspatial memory. Proc Natl Acad Sci 105: 5915–5920.

� Dudai Y. 2015. Retrieval. Cold Spring Harb Perspect Biol doi:10.1101/cshperspect.a021790.

Eichenbaum HB, Cohen NJ. 2001. From conditioning to con-scious recollection: Memory systems of the brain. OxfordUniversity Press, New York.

Eichenbaum H, Cohen NJ. 2014. Can we reconcile the de-clarative memory and spatial navigation views on hippo-campal function? Neuron 83: 764–770.

Eichenbaum H, Yonelinas AP, Ranganath C. 2007. The me-dial temporal lobe and recognition memory. Annu RevNeurosci 30: 123–152.

Ekstrom AD, Bazih AJ, Suthana NA, Al-Hakim R, Ogura K,Zeineh M, Burggren AC, Bookheimer SY. 2009. Advancesin high-resolution imaging and computational unfold-ing of the human hippocampus. Neuroimage 47: 42–49.

Feinberg DA, Moeller S, Smith SM, Auerbach E, Ramanna S,Gunther M, Glasser MF, Miller KL, Ugurbil K, Yacoub E.2010. Multiplexed echo planar imaging for sub-secondwhole brain FMRI and fast diffusion imaging. PLoS ONE5: e15710.

Fischl B, Salat DH, Busa E, Albert M, Dieterich M, Hasel-grove C, van der Kouwe A, Killiany R, Kennedy D, Klave-ness S, et al. 2002. Whole brain segmentation: Automatedlabeling of neuroanatomical structures in the humanbrain. Neuron 33: 341–355.

Friedman D, Johnson R. 2000. Event-related potential(ERP) studies of memory encoding and retrieval: A se-lective review. Microsc Res Tech 51: 6–28.

Friston KJ. 2011. Functional and effective connectivity: Areview. Brain Connect 1: 13–36.

Friston KJ, Frith CD, Liddle PF, Frackowiak RS. 1993. Func-tional connectivity: The principal-component analysis oflarge (PET) data sets. J Cereb Blood Flow Metab 13: 5–14.

Friston KJ, Harrison L, Penny W. 2003. Dynamic causalmodelling. Neuroimage 19: 1273–1302.

Friston K, Moran R, Seth AK. 2013. Analysing connectivitywith Granger causality and dynamic causal modelling.Curr Opin Neurobiol 23: 172–178.

Gordon AM, Rissman J, Kiani R, Wagner AD. 2013. Corticalreinstatement mediates the relationship between con-tent-specific encoding activity and subsequent recollec-tion decisions. Cereb Cortex 24: 3350–3364.

T.I. Brown et al.




Graham KS, Barense MD, Lee ACH. 2010. Going beyondLTM in the MTL: A synthesis of neuropsychological andneuroimaging findings on the role of the medial temporallobe in memory and perception. Neuropsychologia 48:831–853.

� Graybiel A. 2015. The striatum: Where skills and habitsmeet. Cold Spring Harb Perspect Biol doi: 10.1101/cshperspect.a021691.

Greenberg DL, Verfaellie M. 2010. Interdependence of epi-sodic and semantic memory: Evidence from neuropsy-chology. J Int Neuropsychol Soc 16: 748–753.

Greicius MD, Krasnow B, Reiss AL, Menon V. 2003. Func-tional connectivity in the resting brain: A network anal-ysis of the default mode hypothesis. Proc Natl Acad Sci100: 253–258.

Grill-Spector K, Sayres R, Ress D. 2006. High-resolutionimaging reveals highly selective nonface clusters in thefusiform face area. Nat Neurosci 9: 1177–1185.

Hartley T, Harlow R. 2012. An association between humanhippocampal volume and topographical memory inhealthy young adults. Front Hum Neurosci 6: 1–11.

Hartley T, Maguire EA, Spiers HJ, Burgess N. 2003. The well-worn route and the path less traveled: Distinct neuralbases of route following and wayfinding in humans. Neu-ron 37: 877–888.

Hasselmo ME, Eichenbaum H. 2005. Hippocampal mech-anisms for the context-dependent retrieval of episodes.Neural Networks 18: 1172–1190.

Hasselmo ME, Wyble BP. 1997. Free recall and recognitionin a network model of the hippocampus: Simulating ef-fects of scopolamine on human memory function. BehavBrain Res 89: 1–34.

Heidemann RM, Ivanov D, Trampel R, Fasano F, Meyer H,Pfeuffer J, Turner R. 2012. Isotropic submillimeter fMRIin the human brain at 7 T: Combining reduced field-of-view imaging and partially parallel acquisitions. MagnReson Med 68: 1506–1516.

Iaria G, Petrides M, Dagher A, Pike B, Bohbot VD. 2003.Cognitive strategies dependent on the hippocampus andcaudate nucleus in human navigation: Variability andchange with practice. J Neurosci 23: 5945–5952.

Iaria G, Lanyon LJ, Fox CJ, Giaschi D, Barton JJS. 2008.Navigational skills correlate with hippocampal fractionalanisotropy in humans. Hippocampus 18: 335–339.

Johnson MK, Kounios J, Nolde SF. 1997. Electrophysiolog-ical brain activity and memory source monitoring. Neu-roreport 8: 1317–1320.

Johnson JD, McDuff SGR, Rugg MD, Norman KA. 2009.Recollection, familiarity, and cortical reinstatement: Amultivoxel pattern analysis. Neuron 63: 697–708.

Kesner RP. 2007. Behavioral functions of the CA3 subregionof the hippocampus. Learn Mem 14: 771–781.

Kim H. 2011. Neural activity that predicts subsequent mem-ory and forgetting: A meta-analysis of 74 fMRI studies.Neuroimage 54: 2446–2461.

Kirwan CB, Stark CEL. 2004. Medial temporal lobe activa-tion during encoding and retrieval of novel face-namepairs. Hippocampus 14: 919–930.

Kirwan CB, Jones CK, Miller MI, Stark CEL. 2007. High-resolution fMRI investigation of the medial temporallobe. Hum Brain Mapp 28: 959–966.

Konishi K, Bohbot VD. 2013. Spatial navigational strategiescorrelate with gray matter in the hippocampus of healthyolder adults tested in a virtual maze. Front Aging Neurosci5: 1.

Kriegeskorte N, Mur M, Bandettini P. 2008. Representation-al similarity analysis—Connecting the branches of sys-tems neuroscience. Front Syst Neurosci 2: 4.

Kuhl BA, Rissman J, Wagner AD. 2012. Multi-voxel patternsof visual category representation during episodic encod-ing are predictive of subsequent memory. Neuropsycho-logia 50: 458–469.

LaRocque KF, Smith ME, Carr VA, Witthoft N, Grill-SpectorK, Wagner AD. 2013. Global similarity and pattern sep-aration in the human medial temporal lobe predict sub-sequent memory. J Neurosci 33: 5466–5474.

Lee ACH, Scahill VL, Graham KS. 2008. Activating the me-dial temporal lobe during oddity judgment for faces andscenes. Cereb Cortex 18: 683–696.

Liang JC, Wagner AD, Preston AR. 2013. Content represen-tation in the human medial temporal lobe. Cereb Cortex23: 80–96.

Libby LA, Ekstrom AD, Ragland JD, Ranganath C. 2012.Differential connectivity of perirhinal and parahippo-campal cortices within human hippocampal subregionsrevealed by high-resolution functional imaging. J Neuro-sci 32: 6550–6560.

Litman L, Awipi T, Davachi L. 2009. Category-specificity inthe human medial temporal lobe cortex. Hippocampus19: 308–319.

Logothetis NK, Wandell BA. 2004. Interpreting the BOLDsignal. Annu Rev Physiol 66: 735–769.

Long NM, Burke JF, Kahana MJ. 2014. Subsequent memoryeffect in intracranial and scalp EEG. Neuroimage 84:488–494.

Lowe MJ, Dzemidzic M, Lurito JT, Mathews VP, PhillipsMD. 2000. Correlations in low-frequency BOLD fluctu-ations reflect cortico–cortical connections. Neuroimage12: 582–587.

Maguire EA, Woollett K, Spiers HJ. 2006. London taxi driv-ers and bus drivers: A structural MRI and neuropsycho-logical analysis. Hippocampus 16: 1091–1101.

Marr D. 1971. Simple memory: A theory for archicortex.Philos Trans R Soc Lond B Biol Sci 262: 23–81.

Martin E. 1968. Stimulus meaningfulness and paired-asso-ciate transfer: An encoding variability hypothesis. PsycholRev 75: 421–441.

Mayes A, Montaldi D, Migo E. 2007. Associative memoryand the medial temporal lobes. Trends Cogn Sci 11: 126–135.

McClelland JL, McNaughton BL, O’Reilly RC. 1995. Whythere are complementary learning systems in the hippo-campus and neocortex: Insights from the successes andfailures of connectionist models of learning and memory.Psychol Rev 102: 419–457.

McDuff SGR, Frankel HC, Norman KA. 2009. Multivoxelpattern analysis reveals increased memory targeting andreduced use of retrieved details during single-agendasource monitoring. J Neurosci 29: 508–516.

McIntosh AR, Gonzalez-Lima F. 1994. Structural equationmodeling and its application to network analysis in func-tional brain imaging. Hum Brain Mapp 2: 2–22.





Mechelli A, Price CJ, Friston KJ, Ashburner J. 2005. Voxel-based morphometry of the human brain: Methods andapplications. Curr Med Imag Rev 1: 105–113.

Moeller S, Yacoub E, Olman CA, Auerbach E, Strupp J, HarelN, Ugurbil K. 2010. Multiband multislice GE-EPI at 7tesla, with 16-fold acceleration using partial parallel im-aging with application to high spatial and temporalwhole-brain fMRI. Magn Reson Med 63: 1144–1153.

Montaldi D, Mayes AR. 2010. The role of recollection andfamiliarity in the functional differentiation of the medialtemporal lobes. Hippocampus 20: 1291–1314.

Moscovitch M, Nadel L, Winocur G, Gilboa A, RosenbaumRS. 2006. The cognitive neuroscience of remote episodic,semantic and spatial memory. Curr Opin Neurobiol 16:179–190.

Moser EI, Kropff E, Moser M-B. 2008. Place cells, grid cells,and the brain’s spatial representation system. Annu RevNeurosci 31: 69–89.

� Moser MB, Rowland DC, Moser EI. 2015. Place cells, gridcells, and memory. Cold Spring Harb Perspect Biol 7:a021808.

Moses SN, Brown TM, Ryan JD. 2010. Neural system inter-actions underlying human transitive inference. Hippo-campus 20: 894–901.

Murray E, Bussey T. 1999. Perceptual–mnemonic functionsof the perirhinal cortex. Trends Cogn Sci 3: 142–151.

Niedermeyer E, Lopes da Silva F (eds.). 2005. Electroenceph-alography: Basic principles, clinical applications, and relat-ed fields, 5th ed. Lippincott Williams and Wilkins, Phil-adelphia.

Norman KA. 2010. How hippocampus and cortex contrib-ute to recognition memory: Revisiting the complemen-tary learning systems model. Hippocampus 20: 1217–1227.

Norman KA, O’Reilly RC. 2003. Modeling hippocampaland neocortical contributions to recognition memory:A complementary-learning-systems approach. PsycholRev 110: 611–646.

Norman KA, Polyn SM, Detre GJ, Haxby JV. 2006. Beyondmind-reading: Multi-voxel pattern analysis of fMRI data.Trends Cogn Sci 10: 424–430.

Nyhus E, Curran T. 2010. Functional role of g and u oscil-lations in episodic memory. Neurosci Biobehav Rev 34:1023–1035.

O’Keefe J. 1976. Place units in the hippocampus of the freelymoving rat. Exp Neurol 51: 78–109.

O’Keefe J, Dostrovsky J. 1971. The hippocampus as a spatialmap. Preliminary evidence from unit activity in the freelymoving rat. Brain Res 34: 171–175.

O’Keefe J, Nadel L. 1978. The hippocampus as a cognitivemap. Oxford University Press, Oxford.

O’Reilly RC, McClelland JL. 1994. Hippocampal conjunc-tive encoding, storage, and recall: Avoiding a trade-off.Hippocampus 4: 661–682.

O’Reilly RC, Rudy JW. 2001. Conjunctive representations inlearning and memory: Principles of cortical and hippo-campal function. Psychol Rev 108: 311–345.

Paller KA, Wagner AD. 2002. Observing the transformationof experience into memory. Trends Cogn Sci 6: 93–102.

Paller KA, Kutas M, Mayes AR. 1987. Neural correlates ofencoding in an incidental learning paradigm. Electroen-cephalogr Clin Neurophysiol 67: 360–371.

Patenaude B, Smith SM, Kennedy DN, Jenkinson M. 2011. ABayesian model of shape and appearance for subcorticalbrain segmentation. Neuroimage 56: 907–922.

Polyn SM, Natu VS, Cohen JD, Norman KA. 2005. Catego-ry-specific cortical activity precedes retrieval duringmemory search. Science 310: 1963–1966.

Poppenk J, Moscovitch M. 2011. A hippocampal marker ofrecollection memory ability among healthy young adults:Contributions of posterior and anterior segments. Neu-ron 72: 931–937.

Preston AR, Bornstein AM, Hutchinson JB, Gaare ME,Glover GH, Wagner AD. 2010. High-resolution fMRI ofcontent-sensitive subsequent memory responses in hu-man medial temporal lobe. J Cogn Neurosci 22: 156–173.

Ranganath C, Yonelinas AP, Cohen MX, Dy CJ, Tom SM,D’Esposito M. 2004. Dissociable correlates of recollectionand familiarity within the medial temporal lobes. Neuro-psychologia 42: 2–13.

Rissman J, Wagner AD. 2012. Distributed representations inmemory: Insights from functional brain imaging. AnnuRev Psychol 63: 101–128.

Rissman J, Gazzaley A, D’Esposito M. 2004. Measuringfunctional connectivity during distinct stages of a cogni-tive task. Neuroimage 23: 752–763.

Rissman J, Greely HT, Wagner AD. 2010. Detecting individ-ual memories through the neural decoding of memorystates and past experience. Proc Natl Acad Sci 107: 9849–9854.

Ritchey M, Dolcos F, Cabeza R. 2008. Role of amygdalaconnectivity in the persistence of emotional memoriesover time: An event-related FMRI investigation. CerebCortex 18: 2494–2504.

Rosenbaum RS, Gilboa A, Moscovitch M. 2014. Case studiescontinue to illuminate the cognitive neuroscience ofmemory. Ann NY Acad Sci 1316: 105–133.

Ross RS, Sherrill KR, Stern CE. 2011. The hippocampus isfunctionally connected to the striatum and orbitofrontalcortex during context dependent decision making. BrainRes 1423: 53–66.

Rugg MD. 1995. Event-related potential studies of humanmemory. In The cognitive neurosciences (ed. GazzanigaMS), pp. 789–802. MIT Press, Cambridge.

Sadeh T, Shohamy D, Levy DR, Reggev N, Maril A. 2011.Cooperation between the hippocampus and the striatumduring episodic encoding. J Cogn Neurosci 23: 1597–1608.

Sanquist TF, Rohrbaugh JW, Syndulko K, Lindsley DB. 1980.Electrocortical signs of levels of processing: Perceptualanalysis and recognition memory. Psychophysiology 17:568–576.

Schinazi VR, Nardi D, Newcombe NS, Shipley TF, EpsteinRA. 2013. Hippocampal size predicts rapid learning of acognitive map in humans. Hippocampus 23: 515–528.

Scoville WB, Milner B. 1957. Loss of recent memory afterbilateral hippocampal lesions. J Neurol Neurosurg Psychi-atry 20: 11–21.

Smith SM, Miller KL, Salimi-Khorshidi G, Webster M,Beckmann CF, Nichols TE, Ramsey JD, Woolrich MW.

T.I. Brown et al.




2011. Network modelling methods for FMRI. Neuro-image 54: 875–891.

Spaniol J, Davidson PSR, Kim ASN, Han H, Moscovitch M,Grady CL. 2009. Event-related fMRI studies of episodicencoding and retrieval: Meta-analyses using activationlikelihood estimation. Neuropsychologia 47: 1765–1779.

Squire LR, Bayley PJ. 2007. The neuroscience of remotememory. Curr Opin Neurobiol 17: 185–196.

� Squire LR, Dede AJO. 2015. Memory systems. Cold SpringHarb Perspect Biol 7: a021667.

Squire LR, Stark CEL, Clark RE. 2004. The medial temporallobe. Annu Rev Neurosci 27: 279–306.

Staresina BP, Davachi L. 2008. Selective and shared contri-butions of the hippocampus and perirhinal cortex toepisodic item and associative encoding. J Cogn Neurosci20: 1478–1489.

Staresina BP, Davachi L. 2010. Object unitization and asso-ciative memory formation are supported by distinctbrain regions. J Neurosci 30: 9890–9897.

Staresina BP, Duncan KD, Davachi L. 2011. Perirhinal andparahippocampal cortices differentially contribute to lat-er recollection of object- and scene-related event details. JNeurosci 31: 8739–8747.

Staresina BP, Henson RNA, Kriegeskorte N, Alink A. 2012.Episodic reinstatement in the medial temporal lobe. JNeurosci 32: 18150–18156.

Staresina BP, Cooper E, Henson RN. 2013. Reversible infor-mation flow across the medial temporal lobe: The hippo-campus links cortical modules during memory retrieval.J Neurosci 33: 14184–14192.

Stephan KE, Friston KJ. 2010. Analyzing effective connec-tivity with fMRI. Wiley Interdiscip Rev Cogn Sci 1: 446–459.

Sun FT, Miller LM, D’Esposito M. 2004. Measuring interre-gional functional connectivity using coherence and par-tial coherence analyses of fMRI data. Neuroimage 21:647–658.

Suthana NA, Ekstrom AD, Moshirvaziri S, Knowlton B,Bookheimer SY. 2009. Human hippocampal CA1 in-volvement during allocentric encoding of spatial infor-mation. J Neurosci 29: 10512–10519.

Suzuki WA. 2009. Comparative analysis of the cortical affer-ents, intrinsic projections and interconnections of theparahippocampal region in monkeys and rats. In Thecognitive neurosciences, 4th ed. (ed. Gazzaniga SM), pp.659–674. MIT Press, Cambridge.

Teyler TJ, DiScenna P. 1986. The hippocampal memory in-dexing theory. Behav Neurosci 100: 147–154.

Tolman EC. 1948. Cognitive maps in rats and men. PsycholRev 55: 189–208.

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

Uncapher MR, Wagner AD. 2009. Posterior parietal cortexand episodic encoding: Insights from fMRI subsequentmemory effects and dual-attention theory. NeurobiolLearn Mem 91: 139–154.

Voermans NC, Petersson KM, Daudey L, Weber B, vanSpaendonck KP, Kremer HPH, Fernandez G. 2004. Inter-action between the human hippocampus and the caudatenucleus during route recognition. Neuron 43: 427–435.

Wagner AD, Schacter DL, Rotte M, Koutstaal W, Maril A,Dale AM, Rosen BR, Buckner RL. 1998. Building mem-ories: Remembering and forgetting of verbal experiencesas predicted by brain activity. Science 281: 1188–1191.

Wagner AD, Koutstaal W, Schacter DL. 1999. When encod-ing yields remembering: Insights from event-related neu-roimaging. Philos Trans R Soc Lond B Biol Sci 354: 1307–1324.

Wagner AD, Maril A, Schacter DL. 2000. Interactions be-tween forms of memory: When priming hinders newepisodic learning. J Cogn Neurosci 12: 52–60.

Ward EJ, Chun MM, Kuhl BA. 2013. Repetition suppressionand multi-voxel pattern similarity differentially track im-plicit and explicit visual memory. J Neurosci 33: 14749–14757.

Watson HC, Wilding EL, Graham KS. 2012. A role for peri-rhinal cortex in memory for novel object-context associ-ations. J Neurosci 32: 4473–4481.

Xue G, Dong Q, Chen C, Lu Z, Mumford JA, Poldrack RA.2010. Greater neural pattern similarity across repetitionsis associated with better memory. Science 330: 97–101.

Xue G, Dong Q, Chen C, Lu Z-L, Mumford JA, Poldrack RA.2013. Complementary role of frontoparietal activity andcortical pattern similarity in successful episodic memoryencoding. Cereb Cortex 23: 1562–1571.

Yacoub E, Harel N, Ugurbil K. 2008. High-field fMRI unveilsorientation columns in humans. Proc Natl Acad Sci 105:10607–10612.

Yassa MA, Stark CEL. 2009. A quantitative evaluation ofcross-participant registration techniques for MRI studiesof the medial temporal lobe. Neuroimage 44: 319–327.

Yassa MA, Stark SM, Bakker A, Albert MS, Gallagher M,Stark CEL. 2010. High-resolution structural and func-tional MRI of hippocampal CA3 and dentate gyrus inpatients with amnestic mild cognitive impairment. Neu-roimage 51: 1242–1252.

Zeineh MM, Engel SA, Bookheimer SY. 2000. Application ofcortical unfolding techniques to functional MRI of thehuman hippocampal region. Neuroimage 11: 668–683.





March 16, 20152015; doi: 10.1101/cshperspect.a021840 originally published onlineCold Spring Harb Perspect Biol

Thackery I. Brown, Bernhard P. Staresina and Anthony D. Wagner Temporal LobeNoninvasive Functional and Anatomical Imaging of the Human Medial

Subject Collection Learning and Memory

ConsolidationThe Regulation of Transcription in Memory

Cristina M. Alberini and Eric R. Kandelof the Human Medial Temporal LobeNoninvasive Functional and Anatomical Imaging

Anthony D. WagnerThackery I. Brown, Bernhard P. Staresina and

Conscious and Unconscious Memory SystemsLarry R. Squire and Adam J.O. Dede

Nonassociative Learning in InvertebratesJohn H. Byrne and Robert D. Hawkins

Place Cells, Grid Cells, and Memory

MoserMay-Britt Moser, David C. Rowland and Edvard I.

http://cshperspectives.cshlp.org/cgi/collection/ For additional articles in this collection, see

Copyright © 2015 Cold Spring Harbor Laboratory Press; all rights reserved


http://cshperspectives.cshlp.org/cgi/collection/

http://cshperspectives.cshlp.org/cgi/collection/

http://cshperspectives.cshlp.org/cgi/adclick/?ad=42791&adclick=true&url=http%3A%2F%2Fwww.genelink.com%2Fnewsite%2Fproducts%2FmodoligosINFO.asp


Documents

Noninvasive Functional and Anatomical Imaging of the …Noninvasive Functional and Anatomical Imaging of the Human Medial Temporal Lobe Thackery I. Brown 1, Bernhard P. Staresina ,