neuron review: point/counterpoint - york university09_percepmem.pdf · neuron review:...
TRANSCRIPT
Neuron
Review: Point/Counterpoint
Involvement of Medial Temporal LobeStructures in Memory and Perception
Mark G. Baxter1
1Department of Experimental Psychology, Oxford University, South Parks Road, Oxford OX1 3UD, UKDOI 10.1016/j.neuron.2009.02.007
Beginning approximately a decade and a half ago, it was suggested that some structures that are consideredto be part of the ‘‘medial temporal lobe memory system’’ could play a role in perception as well. The implica-tions of this view, interpreted broadly, are that medial temporal lobe structures may be understood as anextension of the ventral visual stream and that their functions cannot be described exclusively in terms ofmemory. Considerable evidence now supports the view that medial temporal lobe structures are involvedin nonmnemonic aspects of cognition, such as perception. This discovery allows for a fuller understandingof the involvement of these structures in mental phenomena than does a purely mnemonic account of theirfunction. See the related review by Suzuki, ‘‘Perception and the Medial Temporal Lobe: Evaluating theCurrent Evidence,’’ in this issue of Neuron.
tant role in online perceptual processing would falsify the stan-
dard view of MTL function.
This article will review, briefly, the evidence that has been
interpreted as demonstrating a role for MTL structures in percep-
tual functions as opposed to strictly mnemonic functions. It will
draw from studies in animals and humans, and from both exper-
iments, such as investigations of effects of lesions or temporary
inactivations, and correlational approaches, such as functional
neuroimaging and behavioral neurophysiology. The differences
in interpretation of these findings will also be discussed. As the
goal of these paired articles is to synthesize and integrate dispa-
rate views of a single literature, some possible avenues for future
work will be outlined, but these are also discussed more exten-
sively in the accompanying commentary (Suzuki and Baxter,
2009). A central theme running through this work is the extent
to which perceptual and mnemonic functions can be differenti-
ated from one another, both psychologically and behaviorally,
a topic to which I will return at the end of the paper.
Some Effects of Perirhinal Cortex LesionsAre Consistent with an Impairment in PerceptionAs far as I am aware, Eacott, Gaffan, and Murray (Eacott et al.,
1994) were the first to propose that the rhinal cortex (comprising
entorhinal and perirhinal cortex) may be involved in visual identi-
fication and perceptual function, similar to the laterally adjacent
cortical area TE. This study is of interest, therefore, for historical
reasons, but also because it serves to highlight some key issues
in the interpretation of findings from lesion studies that attempt
to address the role of defined cortical structures in cognition.
These authors trained macaque monkeys preoperatively on
several visual recognition memory tasks (delayed matching-to-
sample, DMS) in an automated apparatus. In this paradigm,
one or more visual objects (images on a computer screen) are
presented as ‘‘sample’’ objects, and then choice tests are
offered in which the monkey is allowed to select either the
sample object or a foil that was not presented during the sample
phase. In DMS, the monkey is rewarded for selecting the object
that appeared as a sample during the sample phase of that
Structures in the medial temporal lobe (MTL), including the
hippocampus and connected areas (entorhinal, perirhinal, para-
hippocampalcortex), havebeenproposed toconstitutea ‘‘medial
temporal lobe memory system’’ that is specialized for memory
functions in the mammalian brain. Although it is acknowledged
that these are not the only structures in the brain that are involved
in memory—for example, the medial diencephalon is critical for
memory (Aggleton and Mishkin, 1983; Mitchell et al., 2007; Victor
et al., 1989; Zola-Morgan and Squire, 1985)—the primary func-
tions of these areas are thought to be in storage and recall of facts
and events (Squire and Knowlton, 2000; Squire et al., 2004). In
particular, the MTL ‘‘(a) is principally concerned with memory,
(b) operates with neocortex to establish and maintain long-term
memory, and (c) ultimately, through a process of consolidation,
becomes independent of long-term memory’’ (Squire et al.,
2004, p. 279). The precise extent to which particular elements
of this system are differentially involved in specific aspects of
memory remains a topic of discussion: for example, are struc-
tures within this system further specialized for particular types
or modalities of memory, or do they work together as a unitary
memory system (Murray and Wise, 2004)?
Notwithstanding these considerations, a central tenet of this
characterization of the MTL is that its primary function is in
memory and not in other aspects of cognition (Squire et al.,
2004). However, beginning �15 years ago, experimental data
began to suggest that MTL structures may be involved in
perceptual functions as well. Thus, an alternative hypothesis
about the involvement of the MTL in cognition has emerged
which suggests that perceptual functions of these cortical areas
are just as important as their involvement in memory, or indeed
that apparently disparate mnemonic and perceptual functions
may arise from common computational mechanisms (Bussey
and Saksida, 2002; Murray et al., 2007). According to this view,
‘‘the perirhinal cortex’’ (a component of the MTL) ‘‘represents
information about objects for both mnemonic and perceptual
purposes’’ (Murray et al., 2007, p. 99). This differs from the stan-
dard view of MTL function, which holds that its main function is in
memory. Thus, establishing that MTL structures play an impor-
Neuron 61, March 12, 2009 ª2009 Elsevier Inc. 667
Neuron
Review: Point/Counterpoint
particular test trial. In the more common delayed nonmatching-
to-sample, or DNMS, the monkey is rewarded for choosing the
other object that was not presented as a sample; see Figure 1
of Suzuki (2009). Eacott et al. (1994) found that monkeys with
rhinal cortex lesions were impaired in delayed matching-to-
sample performance, but this impairment was mitigated by
reducing the set size of objects from which the recognition
memory test was drawn and was eliminated entirely by reducing
the set of objects to two. Thus, monkeys with rhinal cortex
lesions could perform delayed matching-to-sample, a canonical
test of visual recognition memory, normally as long as the pool
from which the objects to be remembered were drawn was
very small (two). Moreover, with large sets of objects, deficits
could be seen even when the stimuli to be matched were pre-
sented simultaneously, or when a 0 s delay separated sample
and choice.
Although tasks such as DMS and DNMS are designed to study
memory rather than perceptual function, these findings were in-
terpreted as possibly suggesting a visual perceptual deficit,
rather than a mnemonic one, following lesions of the rhinal
cortex. Thus, some features of impaired and spared perfor-
mance may shed light on whether rhinal cortex damage impairs
perceptual function in addition to memory. Impaired perfor-
mance in the absence of memory demands (simultaneous and
0 s matching conditions) suggested an impairment in some
aspect of perceptual processing. The significance of this finding
has been criticized because a subsequent study reported that
monkeys with selective perirhinal cortex lesions were unim-
paired at learning a DNMS task with a 0.5 s delay between
sample and choice (Buffalo et al., 2000; see also Suzuki, 2009),
but because performance of these monkeys was not statistically
significantly different from that of controls at any delay (including
10 min delays between sample and choice), this result is difficult
to interpret. The statistical significance of the original finding of
Eacott et al. (1994) has also been questioned because it depends
on pooling data from several test conditions (Buffalo et al., 2000;
Suzuki, 2009).
Eacott et al. also showed that if the monkey was very familiar
with the objects to be remembered in the DMS task, because
they were used repeatedly across trials instead of being unique
from trial to trial, the rhinal cortex was apparently not required for
remembering which one had been seen in the sample trial. This
also supports a perceptual view of MTL function, because when
there was no demand on discrimination between the stimuli, the
rhinal cortex was also not necessary to remember them during
the delay between sample and choice. It might be argued that
this version of DMS no longer taxes visual recognition memory
but instead taxes working or recency memory. That is, rather
than simply recognizing whether a stimulus has been seen
before, the monkey must remember which of the two stimuli he
saw on the sample trial immediately preceding the current
choice trial in order to respond correctly. However, it is not
obvious why a similar strategy could not be applied to choice
trials with trial-unique stimuli.
At the time of the Eacott et al. (1994) study, it had been
demonstrated that lesions of entorhinal cortex, perirhinal cortex,
or both together impaired DNMS with trial-unique objects
(Meunier et al., 1993) and that lesions of rhinal cortex impaired
668 Neuron 61, March 12, 2009 ª2009 Elsevier Inc.
DMS but left object discrimination learning intact (Gaffan and
Murray, 1992). Furthermore, lesions of perirhinal and parahippo-
campal cortex together impaired DNMS in both visual and
tactual modalities, indicating that the deficit in DNMS perfor-
mance extended beyond the visual modality (Suzuki et al.,
1993). (For a review of the literature on stimulus recognition
leading up to the focus on the rhinal cortex, see Murray,
1996.) The finding that monkeys with rhinal cortex lesions could
succeed at DMS when tested with familiar stimuli and that
deficits in DMS could be seen with no or minimal memory
demands challenged the notion that the rhinal cortex was
responsible for all aspects of stimulus memory, although it
certainly was not an unambiguous demonstration of a perceptual
function of rhinal cortex.
A series of studies from Buckley and Gaffan extended this
initial observation. These authors reported that perirhinal cortex
lesions impaired concurrent discrimination learning with large,
but not small, sets of objects, as well as transfer of discrimination
learning between 2D and 3D representations of the same objects
and generalization of discrimination learning to new views of
familiar objects (Buckley and Gaffan, 1997, 1998a, 1998b).
These findings support a role for perirhinal cortex in being able
to perceptually identify objects, not just remember them. Again,
these data are not unambiguous. For example, it has been
argued that the set-size effects on discrimination learning simply
represent a magnification of a unitary impairment per problem
(Hampton, 2005) such that this impairment does not reach signif-
icance with a small number of discriminations but the additive
effect across a large number of discriminations creates a signifi-
cant difference from controls. Another criticism of these experi-
ments is that the impairment in transfer between 3D objects and
2D photographs of the same objects could just reflect an impair-
ment in discrimination learning, but this criticism can be rejected
on the basis of reanalysis of data from this experiment, showing
impairment in monkeys with perirhinal cortex lesions even on the
first trial of each problem (Buckley, 2005).
A subsequent experiment provided a more explicit test of
perceptual function by examining performance of monkeys
with perirhinal cortex lesions on oddity discriminations with
a number of different types of stimulus material (Buckley et al.,
2001). Monkeys with perirhinal cortex lesions were impaired on
oddity problems based on different viewpoints of objects,
degraded images of objects, human and monkey faces pre-
sented from different viewpoints, and visual scenes, but not on
problems based on color, shape, or size, irrespective of the diffi-
culty of particular problems, or same-viewpoint presentations of
human or monkey faces. Because the impairments were
observed on discriminations taxing complex object discrimina-
tion and not on difficult discriminations per se, this argues in
favor of a perceptual deficit rather than a problem with memory.
On the other hand, although associations with particular stimuli
and reward could not be used to solve the oddity problems
(because individual stimuli appeared equally often as the odd
one out and the foils), it could be argued that the impairment in
these tasks is expressed through learning as a consequence of
repeated exposure to particular objects, or association of
different views of the same stimuli with one another, and there-
fore has more to do with memory function than with perception
Neuron
Review: Point/Counterpoint
(Suzuki, 2009). These limitations can be addressed, to some
extent, through the use of an oddity task with trial-unique stimuli
(discussed later in this article).
Formal Models of Perirhinal Cortex Involvementin Perceptual FunctionOne problem with all of these experiments is that the definition of
what constitutes ‘‘perception’’ is nebulous. Bussey and Saksida
(2002) devised a computational model in order to make explicit
predictions of how damage to perirhinal cortex would affect
performance in visual learning tasks, based on the idea that
the perirhinal cortex is a rostral extension of the ventral visual
stream that is specialized to process conjunctions of simple
perceptual features. In this model, the perirhinal cortex is at
the apex of a hierarchically organized system of visual cortical
areas. By virtue of its position in the hierarchy, the perirhinal
cortex is required to resolve discriminations with a high degree
of ‘‘feature ambiguity,’’ which occurs when the stimuli to be
discriminated have a large number of features in common.
However, this does not solve the problem of deciding what
a ‘‘feature’’ is. Some neuropsychological data bear on this point,
for example that ablations of the middle temporal gyrus
(including much of area TE) impair color discrimination but not
DNMS, whereas the reverse pattern is obtained with lesions of
perirhinal cortex (Buckley et al., 1997). Nevertheless, the model
does make explicit predictions about performance in a number
of tasks in which ‘‘features’’ are operationally defined. For
example, discrimination problems that can only be solved based
on a combination of features should require the perirhinal cortex
for their solution. One case of this is the biconditional discrimina-
tion problem, of the form AB+, CD+, BC�, AD�, where each
capital letter represents an individual feature and each pair of
letters represents an individual stimulus that is associated with
reward (+) or nonreward (�). When confronted with a stimulus
(for example AB), it is only possible to know whether that stim-
ulus is rewarded based on the combination of features A and
B, because A and B individually are each also part of nonre-
warded stimuli. Thus, a biconditional visual discrimination
problem requires processing of feature conjunctions—because
each individual feature is ambiguous—and should be impaired
by perirhinal cortex damage. In contrast, a discrimination
problem of the form AB+, CD+, EF�, GH� could be solved on
the basis of individual features: A only appears as part of the re-
warded stimulus AB. Because each feature is unambiguous, this
discrimination problem—which also requires learning about four
stimuli, each composed of two features—should not require
perirhinal cortex. An important property of these discrimination
problems is that they should place equal demands on memory,
because the number of stimuli to be associated with reward (or
nonreward) is the same in each case.
Bussey, Saksida, and Murray (Bussey et al., 2002) tested this
prediction directly in monkeys with lesions of perirhinal cortex,
by training them on visual discrimination problems in conditions
of minimal, intermediate, or maximal feature ambiguity (Figure 1).
On the view that perirhinal cortex is involved in visual learning
and memory irrespective of the degree of feature ambiguity
within the visual discrimination problem, one would predict that
perirhinal cortex lesions would produce equivalent impairments
in all three conditions. Alternatively, if perirhinal cortex is required
for solving visual discrimination problems only when resolution
of feature ambiguity is required, then the degree of impairment
should interact with the perceptual demands of the visual
discrimination problem. This latter pattern of performance is
exactly what was found (Figure 1). Importantly, the specification
of ‘‘feature ambiguity’’ as a critical factor in determining whether
perirhinal cortex is necessary for solving a visual task is consis-
tent with the observation that rotating or otherwise degrading
previously learned visual stimuli affects the performance of
control monkeys and monkeys with perirhinal cortex lesions
similarly (Hampton and Murray, 2002). Because these manipula-
tions do not affect the degree of feature ambiguity, they would
not be expected to produce any impairment after damage to
perirhinal cortex (and indeed they do not).
It may still be argued that this result tells us nothing about visual
perception, because the impairment is expressed in the context
of a learning problem. Alternatively, if the demands on learning
and memory are consistent between the various task conditions
(which they were in this study), then it is difficult to attribute any
observed deficits to impaired learning or memory. Instead, it
seems that the feature ambiguity manipulation is critical in deter-
mining whether the perirhinal cortex is involved in the task, rather
than any demands on learning and memory per se. Furthermore,
it is difficult to attribute the impairment to damage to cortical
areas outside of perirhinal cortex, both because the perirhinal
cortex lesions in Bussey et al. (2002) were extremely discrete
and because such damage (which would most likely be located
in laterally adjacent area TE) would produce impairments in visual
discrimination learning regardless of the degree of perceptual
ambiguity (Buffalo et al., 1998, 1999; Iwai and Mishkin, 1968).
Remarkably, the same pattern of impairment in discrimination
learning is seen in humans with temporal cortex lesions that
include the perirhinal cortex (Barense et al., 2005). Subjects in
this study with lesions limited to the hippocampus, who were
amnesic, were not impaired in any aspect of visual discrimination
learning, consistent with the effect of selective hippocampal
lesions in monkeys (Saksida et al., 2006). However, subjects in
the MTL lesion group were impaired as a function of the degree
of feature ambiguity within problems (Figure 2), just as monkeys
with perirhinal cortex lesions are. Related experiments have
used a ‘‘morphing’’ approach to create feature ambiguity
between visual stimuli and have examined impairments in
learning ‘‘morphed’’ discriminations in monkeys with perirhinal
cortex lesions, or in performance on probe trials of ‘‘morphed’’
stimuli created from ‘‘unmorphed’’ photographs that monkeys
have learned to discriminate. Again perirhinal cortex lesions are
without effect on single-pair discrimination learning of ‘‘un-
morphed’’ photographs, but impairments emerge as the degree
of feature ambiguity, created by ‘‘morphing,’’ increases (Bussey
et al., 2003, 2006). Some studies have reported similar findings
in humans with MTL damage (Lee et al., 2005b, 2007; cf. Levy
et al., 2005; Shrager et al., 2006), although there are concerns
about the anatomical specificity of MTL damage in patients that
are impaired on these tasks (discussed in the next section and
by Suzuki, 2009).
These investigations have focused on the involvement of peri-
rhinal cortex in perceptual function. It has been suggested that
Neuron 61, March 12, 2009 ª2009 Elsevier Inc. 669
Neuron
Review: Point/Counterpoint
Figure 1. Effects of Manipulation of Feature Ambiguity on Discrimination Learning in Monkeys with Perirhinal Cortex (PRh) Lesions(Bussey et al., 2002)Example minimum and maximum feature ambiguity problems are illustrated. Each column shows a set of four problems, represented as a screen that would bepresented to the monkey. Each problem contains one rewarded stimulus (in the + column) and one unrewarded stimulus (in the� column); the left/right position ofthe stimuli was random for each trial. Each stimulus is composed of two grayscale photographs apposed to one another; the monkey’s task was to choose onepair of photographs by touching it. Each individual photograph composes a ‘‘feature’’ of the stimulus, so each stimulus is composed nominally of two features. Inthe minimum feature ambiguity problem, each feature appears only once as part of a stimulus, so each feature is either always rewarded or never rewarded. Forexample, the photograph of bandages is always part of a rewarded stimulus, and the photograph of the surgeon is always part of an unrewarded stimulus. In themaximum feature ambiguity problem, each feature appears as part of a rewarded stimulus and as part of an unrewarded stimulus, so it is impossible to solve thetask without relying on conjunctions of features because each individual feature is ambiguous. For example, the image of the ‘‘duck crossing’’ sign is part ofa rewarded stimulus in combination with the ‘‘Lot 6’’ sign, but part of an unrewarded stimulus when in combination with the picture of the child. Intermediatefeature ambiguity problems (not illustrated) have some features that are ambiguous and some that are not. Monkeys with PRh lesions are unimpaired at learningsets of discrimination problems with minimum feature ambiguity but are impaired at learning sets of problems with intermediate or maximum feature ambiguity.Thus, the perceptual qualities of the visual stimuli, rather than the memory demands of the task, determine whether an impairment is seen. Error bars indicate± SEM. Adapted from Figures 4, 6, and 7 of Bussey et al. (2002), by permission of Wiley-Blackwell.
670 Neuron 61, March 12, 2009 ª2009 Elsevier Inc.
and choice (Hartley et al., 2007). Although this latter result
suggests a more reliable impairment in very short-term memory
for spatial relationships rather than perception per se following
hippocampal damage, neither finding is consistent with the
viewpoint that the hippocampus is only required to maintain
memory beyond the capacity of short-term/working memory
(cf. Shrager et al., 2008). These findings, taken together, may
suggest a role for the hippocampus in the perception of scenes,
in the same manner as the perirhinal cortex is involved in the
other components of the MTL may also have perceptual func-
tions, such as the hippocampus. Humans with apparently selec-
tive hippocampal lesions were impaired at oddity discriminations
among virtual-reality ‘‘scenes’’ (Lee et al., 2005a; Figure 3). A
separate study of a different group of humans with hippocampal
lesions examined performance in a simultaneous matching-to-
sample task with topographical relations in visual scenes and
found impairments in two of the four patients, although all four
were impaired when a 2 s delay was introduced between sample
Neuron
Review: Point/Counterpoint
Figure 2. Effects of Manipulation of Feature Ambiguity on Discrimination Learning in Humans with Hippocampal or MTL Damage(Barense et al., 2005)Two kinds of stimuli (‘‘barcodes’’ and ‘‘bugs’’) are illustrated, with three different problem sets in each. The problems are constructed in the same way as thoseused in testing monkeys with perirhinal cortex lesions (Figure 1). For the barcodes, the ‘‘features’’ are the left and right halves of each barcode; for the bugs, thefeatures are the body type and the legs. In the example illustrated with the bugs, in the minimum feature ambiguity condition, both body type and legs are unam-biguous; there are four different body types and four different leg shapes, and each body type or leg shape is part of an always-rewarded bug or a never-rewardedbug. Thus, it is possible to solve this discrimination by focusing on the presence or absence of individual features. In the intermediate feature ambiguity conditionillustrated, body type is ambiguous: there are only two different body types, and each body type is part of an always-rewarded bug or a never-rewarded bugequally often. In the maximum feature ambiguity condition, there are only two body types and two leg shapes, and all are ambiguous: each body type andleg shape is part of an always-rewarded bug and a never-rewarded bug. Thus, the conjunction of leg shape and body type must be used in order to solvethe problems, because each individual leg shape and body type is ambiguous. Performance of amnesic humans with either restricted hippocampal damageor larger MTL lesions including perirhinal cortex is shown in comparison with age-matched, neurologically intact control participants. Humans with hippocampallesions learn all the discrimination problems normally, regardless of feature ambiguity, but humans with MTL lesions are impaired as a function of feature ambi-guity. Error bars indicate ± SEM. Adapted from Figures 2 and 3 of Barense et al. (2005). Reprinted by permission of The Journal of Neuroscience.
Neuron 61, March 12, 2009 ª2009 Elsevier Inc. 671
Neuron
Review: Point/Counterpoint
perception of objects (Murray et al., 2007). However, this hypoth-
esis remains speculative pending accumulation of further data,
in particular from studies in animals in which discrete hippo-
campal lesions can be produced experimentally. This is an
important avenue for future investigations.
What Are the Criticisms of a Role for MTL Structuresin Perception?Some criticisms of these experiments have already been alluded
to. They fall primarily into two categories, related to anatomical
and cognitive specificity of impairments. First, it has been argued
that the extent of brain damage in amnesic patients who are
impaired on ‘‘perceptual’’ tasks has not been adequately delin-
eated, and damage outside the MTL may be able to account
for their perceptual impairments without needing to hypothesize
a role for MTL structures in perceptual function. For example,
Shrager et al. (2006) have suggested that patients tested by
Lee, Barense, and colleagues (Barense et al., 2005; Lee et al.,
2005b) may have damage in temporal cortical structures outside
the MTL that can account for their perceptual impairments (see
also Suzuki, 2009). Recent volumetric analyses of the extent of
Figure 3. Performance of OddityDiscrimination Problems, with Same andDifferent Views of Faces and Scenes, inHumans with Hippocampal or MTL DamageParticipants view displays of four objects, threeviews of the same face or scene and one ofa different face or scene, and are asked to identifythe ‘‘odd one out.’’ In ‘‘same-view’’ problems, thethree images of the same face or scene are pre-sented from the same viewpoint; in ‘‘different-view’’ problems, the three images of the sameface or scene are presented from different view-points. All participants find it easy to identify theodd stimulus in ‘‘same-view’’ face and sceneoddity problems. Humans with hippocampallesions are unimpaired relative to their age-matched controls (‘‘young controls’’) on different-view face oddity, but are impaired in different-view scene oddity. Notably, controls find thesekinds of problems equally difficult, so this is notthe result of a task-difficulty effect. Humans withMTL damage that extends beyond the hippo-campus into the adjacent cortex (including perirhi-nal cortex) are impaired in both face and sceneoddity problems with different views, relative totheir age-matched controls (‘‘elderly controls’’).Error bars indicate ± SEM. Adapted from Figures3 and 6 of Lee et al. (2005a) by permission.
lesions in this patient population support
the conclusions drawn from visual ratings
of MRI scans (A.C. Lee et al., 2008,
Society for Neuroscience, abstract;
A.C.H. Lee, personal communication,
July 14, 2008), so it is unlikely that there
is undetected damage in these patients
that has not been described.
In any case, this is an issue with any
lesion analysis in human patient popula-
tions, where the lesion extent is variable
and, in the case of neurodegenerative
diseases, progressive. The striking similarity between the perfor-
mance of monkeys with discrete perirhinal cortex lesions and
that of humans with broader MTL damage on nearly identical
tasks is compelling in terms of localization of the perceptual
deficits to the MTL. Converging information from fMRI in neuro-
logically intact humans reveals perirhinal cortex activation corre-
lated with oddity judgments (Devlin and Price, 2007; Lee et al.,
2008) as well as for other aspects of object processing associ-
ated with feature ambiguity (Tyler et al., 2004; Moss et al.,
2005) that are impaired by MTL lesions (Moss et al., 2005).
This correlational evidence provides confirmation of the neuro-
anatomical locus of these functions in intact brains, because it
is not possible to localize function with this degree of specificity
based on lesions in human patients. Although neurophysiolog-
ical data have indicated similarity in visual properties of neurons
in perirhinal cortex and laterally adjacent area TE (Suzuki, 2009),
these firing correlates may reflect information passing between
these areas. Disruption of activity in perirhinal cortex or TE while
recording in the complementary area could help resolve this
question (see, for example, Higuchi and Miyashita, 1996). Two
other points are noteworthy in this context. First, to the extent
672 Neuron 61, March 12, 2009 ª2009 Elsevier Inc.
Neuron
Review: Point/Counterpoint
that damage outside of core MTL structures accounts for
perceptual impairments in visual discrimination and oddity tasks
in humans with MTL damage, it is not clear why this damage
would cause impairments that are related to the degree of
feature ambiguity in the stimuli that are presented. Second, there
is little variability in the performance of the MTL patients whose
lesion localization has been questioned (those tested by Lee
et al., 2005a, 2005b), despite differences in the extent of damage
outside of core MTL structures (Suzuki, 2009). For example, if
lateral temporal damage is responsible for the discrepant perfor-
mance of MTL patients in Lee et al. (2005b) and Shrager et al.
(2006), then the subject with the most lateral temporal damage
(MTL1) should be dramatically worse than the other two MTL
patients on the perceptual tasks, but this is not the case.
The second criticism is that cognitive impairments that are
apparently perceptual are instead the result of impaired memory.
There are at least two ways in which impaired learning and
memory could account for effects ascribed to impaired percep-
tion. First, it could be the case that feature-ambiguous discrimi-
nation problems are simply more difficult and therefore unmask
a subtle impairment in visual discrimination learning caused by
perirhinal cortex lesions. But Buckley et al. (2001) show that
there is no relationship between discrimination difficulty, per
se, and impairment (or lack thereof) in their monkeys with perirhi-
nal cortex lesions, so it cannot be that oddity problems that
require ‘‘object identification’’ or discrimination of feature ambi-
guity are simply more difficult. Similarly, color and size discrimi-
nations that are equally difficult to feature-ambiguous discrimi-
nations were learned normally by monkeys with perirhinal
cortex lesions (Bussey et al., 2003). Second, because the impair-
ments in oddity discrimination in monkeys with perirhinal cortex
lesions occur over the course of repeated presentations of
oddity problems, they could simply reflect an impairment in
visual discrimination learning or long-term familiarity with the
stimuli. But learning is not apparent across blocks of oddity
problems in humans (Lee et al., 2005a). Similarly, when monkeys
learn to discriminate a pair of target images and perception is
probed with probe trials with morphed stimuli, learning is not
seen across probe trials (Bussey et al., 2003, 2006). It seems
imparsimonious to hypothesize a memory deficit which is not
strong enough to impair discrimination learning but results in
impairment in probe trials as a function of feature ambiguity (Su-
zuki, 2009). When the memory demands are constant across the
various task conditions and the only variable that is manipulated
is feature ambiguity, it is difficult to explain oddity discrimination
impairments and impairments in discriminating high feature-
ambiguity stimuli in terms of anything other than a perceptual
impairment (Murray et al., 2005). The argument that higher
feature-ambiguity discriminations place a greater demand on
memory because the degree of detail required to discriminate
the stimuli is greater (Suzuki, 2009) is simply an appeal to
a task difficulty explanation, and there is ample evidence that
monkeys (and humans) with MTL damage are not impaired on
difficult discrimination problems. Nevertheless, these are prob-
lems with any procedure in which trial-unique stimuli are not pre-
sented. Recent studies in human patients with MTL damage
(Barense et al., 2007) and rats (Bartko et al., 2007), which also
find impairments in oddity discrimination in high feature-ambi-
guity conditions using trial-unique stimulus presentation, can
address this limitation.
It is also notable that different investigators have reported
different results from patients with MTL lesions on apparently
similar tasks. For example, Stark and Squire (2000) report intact
oddity-discrimination performance in two amnesic patients with
large MTL lesions, whereas Lee et al. (2005a) report oddity-
discrimination impairments in their amnesic patients. These
differences have been attributed by some, as mentioned, to
differences in the location and specificity of MTL lesions. It is
also worth comparing the performance of control subjects
across different investigations on various tasks. For example,
the control subjects in Stark and Squire (2000) find some of the
perceptual tasks as difficult as the amnesic patients do in Lee
et al. (2005a). Both studies tested oddity discriminations with
displays of six faces, five orientations of one face and a sixth
different face (experiment 1 of Lee et al., 2005a, and task 7 of
Stark and Squire, 2000) using apparently nearly identical proce-
dures. The control subjects in Stark and Squire (2000) perform at
�68.2% correct on face oddity and the MTL subjects at 57.3%
correct (estimated from their Figure 4). In Lee et al. (2005a), the
control subjects perform at �88.3% correct on face oddity and
the MTL subjects at 62.3% correct (estimated from their
Figure 4). Thus, some apparent differences in results between
laboratories may relate to the control data that are used for
comparison, rather than the effects of MTL damage per se,
although in other cases performance of control subjects is well
matched across studies (Lee et al., 2005b; Shrager et al.,
2006; discussion in Suzuki, 2009). This is an issue that is difficult
to address given geographic, demographic, and other differ-
ences between patient populations that are tested by different
groups of investigators. Ideally, investigators would attempt to
make their patients and control subjects accessible to testing
by other research groups, in an attempt to exclude possible
methodological or other confounds that might complicate
comparison of results across different laboratories. (For
example, the surgical method used to produce hippocampal
damage in monkeys was excluded as a factor for interlaboratory
differences in effects of hippocampal damage on recognition
memory by testing monkeys operated by one group of investiga-
tors in the laboratory of another group of investigators [Zola-
Morgan and Squire, 1986].) Other differences are more difficult
to resolve: Shrager et al. (2006) used trial-unique presentations
of ‘‘morphed’’ discriminations, using similar stimuli to Lee et al.
(2005b), and failed to find impairments in their population of
amnesic patients. This may be unique to the ‘‘morphing’’
discriminations, which may be solvable based on identification
of single features when simultaneously presented pictures are
compared, and thus do not adequately tax feature ambiguity
(M.D. Barense, personal communication, May 9, 2008; Figure 4).
I have discussed arguments for and against the interpretation
of behavioral impairments after MTL damage as a consequence
of impaired perception rather than memory. Another argument
against a role for the MTL in perception is that preserved perfor-
mance in some tasks is indicative of intact perception after MTL
damage. For example, preserved performance at short delays in
recognition memory tasks is often taken as evidence that
subjects have intact perceptual function, attention, motivation,
Neuron 61, March 12, 2009 ª2009 Elsevier Inc. 673
Neuron
Review: Point/Counterpoint
Figure 4. Even Difficult ‘‘Morphed’’Discrimination Problems, Designed to TaxFeature Ambiguity, May Be Solvable Basedon Individual FeaturesA subject asked to judge which of two blendedimages (on the left and right) is most similar tothe middle image may be able to do so by focusingon individual features of the images, potentiallylimiting the utility of these types of problems for as-sessing the impact of MTL damage on perception.In the top row, the baseball player’s belt is clearlymore apparent in one blended image than theother (red arrowheads), rendering this discrimina-tion a shading judgment rather than a test offeature ambiguity. Similarly, in the bottom row,the darkness of the animal’s nose (arrowheads)provides a single feature that can be focused on.Top row adapted from Figure 5 of Levy et al.(2005) by permission of Cold Spring Harbor Labo-ratory Press (ª 2005). Images in bottom rowprovided by Andy C.H. Lee.
674 Neuron 61, March 12, 2009 ª2009 Elsevier Inc.
controversies here. However, a role for MTL in perception may
allow for an explanation of some experimental observations
that a purely mnemonic account of MTL function does not.
The ‘‘perceptual-mnemonic’’ (Bussey and Saksida, 2002;
Murray et al., 2007) hypothesis may provide a framework to
understand effects of MTL damage on tasks like visual discrim-
ination learning that cannot readily be accommodated in terms
of an exclusive role for the MTL in memory. According to this
view, neurons in the MTL encode representations of objects or
scenes that are necessary for both perception and memory.
The prevailing view of MTL involvement in memory is that the
MTL is specialized for declarative memory, that is, memory for
facts and events that in humans is consciously accessible
(Squire and Knowlton, 2000; Squire and Zola-Morgan, 1991). It
has been difficult, a priori, to decide whether a learning and
memory task is ‘‘declarative’’ or not without making reference
to the performance of amnesic patients on the task (or the perfor-
mance of animals with lesions of structures within the MTL).
Of course, this leads to problems with circularity of definitions,
which have been previously discussed (Morris, 1984; Nadel,
1992). For example, visual discrimination learning has been sug-
gested to be ‘‘declarative’’ on the basis that it is rapidly acquired
(Teng et al., 2000). However, discrimination problems that are
learned rapidly do not necessarily require the hippocampus
(Broadbent et al., 2007; Jonasson et al., 2004). Indeed, the
maximum feature ambiguity discrimination problems tend to
be learned more slowly, yet these are the ones that are impaired
by MTL lesions. Thus, there is nothing ‘‘declarative’’ or not about
visual discrimination learning, although such problems may be
acquired by different learning mechanisms, some slow and
some fast, perhaps in parallel in neurologically intact humans
(Bayley et al., 2005; Hood et al., 1999). Although it is difficult to
account for these data with a purely mnemonic hypothesis of
MTL function, they are readily explained by the ‘‘perceptual-
mnemonic’’ view.
Another advantage of this point of view is that it may explain
the dependence of memory impairments in amnesia on stimulus
modality that are difficult to accommodate in other ways.
and understanding of the task rules, in addition to intact memory
at short delays between sample and choice. There are several
difficulties with this argument. First, extensive training at short
delays is often required to achieve good performance at short
delays between sample and choice in monkeys with MTL
lesions, whether they were trained preoperatively in recognition
memory or not (e.g., Meunier et al., 1993; Zola-Morgan et al.,
1989). This may introduce a bias in performance favoring the
short delay that is trained extensively (Ringo, 1993). Second,
monkeys with MTL damage may fail to learn recognition memory
tasks even after extensive training (e.g., Suzuki et al., 1993), chal-
lenging the conclusion that the only effect of these lesions is on
intermediate- or long-term memory. Third, even when there are
no differences in performance at short delays, typically the
objects used are not designed to tax the hypothesized percep-
tual functions of the MTL (that is, they are not explicitly feature
ambiguous). Finally, the delay dependence of many MTL lesion
effects on recognition memory performance may be related to
the use of a percent correct scale of measurement, because
these effects are not delay dependent when analyzed in a d’
metric (Baxter and Murray, 2001; Ringo, 1988, 1991). Thus, the
effect of MTL lesions on performance of tasks like DNMS may
not necessarily reflect memory impairment. It may be possible
to explain these lesion effects in terms of a similar computational
mechanism to that which is involved in resolution of feature
ambiguity in discrimination learning (Cowell et al., 2006).
How Does a Perceptual/‘‘Perceptual-Mnemonic’’View of MTL Function Advance the Understandingof the Organization of Brain Systems for Cognition?Why might it be useful to consider that MTL structures, including
perirhinal cortex and hippocampus, may play a role in perception
as well as in memory? Considerable effort has been devoted to
identifying which structures in the MTL are essential for memory
function, as well as the potential for different MTL structures to be
involved in different aspects of memory (for instance, semantic
versus, episodic, or recognition versus recall). This remains
extremely controversial, and I will not attempt to address these
Neuron
Review: Point/Counterpoint
Patients with hippocampal damage show intact perceptual
learning and categorization of faces, but not of virtual-reality
scenes (Graham et al., 2006). Not only are these tasks thought
to be ‘‘nondeclarative’’ and therefore independent of the MTL,
but there is no reason to expect the hippocampus to be involved
in memory for scenes but not faces if it has a general role in
declarative memory. Similarly, recognition memory for scenes
is impaired in patients with amnesia consequent to hippocampal
damage, but recognition memory for faces is not (Bird et al.,
2008; Taylor et al., 2007). A dissociation between impaired
recognition memory for words, but intact recognition memory
for faces, in patients with hippocampal damage has been attrib-
uted to hippocampal involvement in providing contextual
support for recognition of familiar items (Bird and Burgess,
2008). A role for the hippocampus in representation of spatial
or contextual information, used for both perception and memory,
is able to account for all of these observations.
It is clear that many questions remain to be resolved in trying to
understand the contributions of MTL structures to memory and
perception. However, progress in this area depends on at least
acknowledging the possibility that the involvement of MTL struc-
tures in cognition extends beyond the domain of memory. It is
common to argue that ‘‘perception’’ is intact after MTL damage
because humans or animals with MTL damage perform normally
on some tasks that are thought to tax perceptual functions. For
example, because monkeys with MTL damage can perform
the DNMS task at short delays with minimal memory demand
(Suzuki, 2009), or because rotating or masking visual stimuli in
monkeys with perirhinal cortex lesions is no more deleterious
to their performance than it is in controls, then their perceptual
functions must be intact (Buffalo et al., 1999; Hampton and Mur-
ray, 2002; Squire et al., 2004). But, as indicated earlier, these
tasks do not adequately tax the perceptual functions that MTL
structures are thought to mediate. The use of feature ambiguity
to tax perceptual discriminations among visual stimuli may tap
into the role of perirhinal cortex in mediating object identification
function more generally (Murray et al., 2000). The finding that
some aspects of perceptual function are intact after perirhinal
cortex damage no more excludes a role for perirhinal cortex in
perception, than the finding that some aspects of memory func-
tion are intact after hippocampal damage excludes a role for the
hippocampus in memory.
The strong version of the ‘‘perceptual-mnemonic’’ view of MTL
function argues that it is not useful to distinguish between
‘‘perception’’ and ‘‘memory’’ as psychological functions, to the
extent that they may arise from a single mechanism at the neural
level. Most would agree that a difference in forgetting rate, when
it has been established that information has been acquired to the
same degree, reflects an impairment in memory rather than
perception. Similarly, most would agree that tasks in which
performance depends on simultaneous comparison of trial-
unique stimuli tax perceptual function rather than memory.
However, these ‘‘folk psychological’’ views of cognition may
impede progress toward understanding how cognitive functions
arise from the physiological activity of the brain (e.g., Bussey,
2004). It is challenging to move beyond these views, especially
given the compelling phenomenology of densely amnesic
patients with MTL lesions, who present with amnesia rather
than agnosia. Nevertheless, the accumulation of evidence that
patients with MTL damage have cognitive impairments that
extend beyond memory forces a reconsideration of the view
that the function of the MTL is exclusively concerned with
memory. Rather than trying to determine the location in the brain
of a sharp line between perception and memory, it may be worth
viewing MTL function in terms of representational capacities that
are used for both cognitive processes (Murray et al., 2007).
Indeed, it is likely that the influence between apparently separate
processes of ‘‘perception’’ and ‘‘memory’’ is bidirectional, with
stored representations affecting online sensory processing
(e.g., Hochstein and Ahissar, 2002). The recent demonstration
that patients with selective hippocampal damage are impaired,
not just in recalling past events, but in imagining future ones
(Hassabis et al., 2007) suggests that even the hippocampus,
for decades the single structure in the brain most closely identi-
fied with memory function, may also be necessary for the
perception and representation of episodes, not merely their
storage and recall. These investigations promise a fuller under-
standing of how brain function gives rise to the phenomena of
mind.
ACKNOWLEDGMENTS
I thank Morgan Barense, Tim Bussey, David Gaffan, Andy Lee, and ElisabethMurray for helpful comments and discussion. My research is supported by theWellcome Trust.
REFERENCES
Aggleton, J.P., and Mishkin, M. (1983). Visual recognition impairment followingmedial thalamic lesions in monkeys. Neuropsychologia 21, 189–197.
Barense, M.D., Bussey, T.J., Lee, A.C., Rogers, T.T., Davies, R.R.,Saksida, L.M., Murray, E.A., and Graham, K.S. (2005). Functional speciali-zation in the human medial temporal lobe. J. Neurosci. 25, 10239–10246.
Barense, M.D., Gaffan, D., and Graham, K.S. (2007). The human medialtemporal lobe processes online representations of complex objects.Neuropsychologia 45, 2963–2974.
Bartko, S.J., Winters, B.D., Cowell, R.A., Saksida, L.M., and Bussey, T.J.(2007). Perirhinal cortex resolves feature ambiguity in configural object recog-nition and perceptual oddity tasks. Learn. Mem. 14, 821–832.
Baxter, M.G., and Murray, E.A. (2001). Opposite relationship of hippocampaland rhinal cortex damage to delayed nonmatching-to-sample deficits inmonkeys. Hippocampus 11, 61–71.
Bayley, P.J., Frascino, J.C., and Squire, L.R. (2005). Robust habit learning inthe absence of awareness and independent of the medial temporal lobe.Nature 436, 550–553.
Bird, C.M., and Burgess, N. (2008). The hippocampus supports recognitionmemory for familiar words but not unfamiliar faces. Curr. Biol. 18, 1932–1936.
Bird, C.M., Vargha-Khadem, F., and Burgess, N. (2008). Impaired memory forscenes but not faces in developmental hippocampal amnesia: a case study.Neuropsychologia 46, 1050–1059.
Broadbent, N.J., Squire, L.R., and Clark, R.E. (2007). Rats depend on habitmemory for discrimination learning and retention. Learn. Mem. 14, 145–151.
Buckley, M.J. (2005). The role of the perirhinal cortex and hippocampus inlearning, memory, and perception. Q. J. Exp. Psychol. B 58, 246–268.
Buckley, M.J., and Gaffan, D. (1997). Impairment of visual object-discrimina-tion learning after perirhinal cortex ablation. Behav. Neurosci. 111, 467–475.
Buckley, M.J., and Gaffan, D. (1998a). Learning and transfer of object-rewardassociations and the role of the perirhinal cortex. Behav. Neurosci. 112, 15–23.
Neuron 61, March 12, 2009 ª2009 Elsevier Inc. 675
Neuron
Review: Point/Counterpoint
Buckley, M.J., and Gaffan, D. (1998b). Perirhinal cortex ablation impairs visualobject identification. J. Neurosci. 18, 2268–2275.
Buckley, M.J., Gaffan, D., and Murray, E.A. (1997). Functional double dissoci-ation between two inferior temporal cortical areas: perirhinal cortex versusmiddle temporal gyrus. J. Neurophysiol. 77, 587–598.
Buckley, M.J., Booth, M.C., Rolls, E.T., and Gaffan, D. (2001). Selectiveperceptual impairments after perirhinal cortex ablation. J. Neurosci. 21,9824–9836.
Buffalo, E.A., Stefanacci, L., Squire, L.R., and Zola, S.M. (1998). A reexamina-tion of the concurrent discrimination learning task: the importance of anteriorinferotemporal cortex, area TE. Behav. Neurosci. 112, 3–14.
Buffalo, E.A., Ramus, S.J., Clark, R.E., Teng, E., Squire, L.R., and Zola, S.M.(1999). Dissociation between the effects of damage to perirhinal cortex andarea TE. Learn. Mem. 6, 572–599.
Buffalo, E.A., Ramus, S.J., Squire, L.R., and Zola, S.M. (2000). Perception andrecognition memory in monkeys following lesions of area TE and perirhinalcortex. Learn. Mem. 7, 375–382.
Bussey, T.J. (2004). Mutiple memory systems: Fact or fiction? Q. J. Exp. Psy-chol. 57B, 89–94.
Bussey, T.J., and Saksida, L.M. (2002). The organization of visual object repre-sentations: A connectionist model of effects of lesions in perirhinal cortex.Eur. J. Neurosci. 15, 355–364.
Bussey, T.J., Saksida, L.M., and Murray, E.A. (2002). Perirhinal cortex resolvesfeature ambiguity in complex visual discriminations. Eur. J. Neurosci. 15,365–374.
Bussey, T.J., Saksida, L.M., and Murray, E.A. (2003). Impairments in visualdiscrimination after perirhinal cortex lesions: testing ’declarative’ vs. ’percep-tual-mnemonic’ views of perirhinal cortex function. Eur. J. Neurosci. 17,649–660.
Bussey, T.J., Saksida, L.M., and Murray, E.A. (2006). Perirhinal cortex andfeature-ambiguous discriminations. Learn. Mem. 13, 103–105.
Cowell, R.A., Bussey, T.J., and Saksida, L.M. (2006). Why does brain damageimpair memory? A connectionist model of object recognition memory inperirhinal cortex. J. Neurosci. 26, 12186–12197.
Devlin, J.T., and Price, C.J. (2007). Perirhinal contributions to human visualperception. Curr. Biol. 17, 1484–1488.
Eacott, M.J., Gaffan, D., and Murray, E.A. (1994). Preserved recognitionmemory for small sets, and impaired stimulus identification for large sets,following rhinal cortex ablations in monkeys. Eur. J. Neurosci. 6, 1466–1478.
Gaffan, D., and Murray, E.A. (1992). Monkeys (Macaca fascicularis) with rhinalcortex ablations succeed in object discrimination learning despite 24-hr inter-trial intervals and fail at matching to sample despite double sample presenta-tions. Behav. Neurosci. 106, 30–38.
Graham, K.S., Scahill, V.L., Hornberger, M., Barense, M.D., Lee, A.C.,Bussey, T.J., and Saksida, L.M. (2006). Abnormal categorization and percep-tual learning in patients with hippocampal damage. J. Neurosci. 26,7547–7554.
Hampton, R.R. (2005). Monkey perirhinal cortex is critical for visual memory,but not for visual perception: reexamination of the behavioural evidencefrom monkeys. Q. J. Exp. Psychol. B 58, 283–299.
Hampton,R.R., and Murray,E.A. (2002). Learningof discriminations is impaired,but generalization to altered views is intact, in monkeys (Macaca mulatta) withperirhinal cortex removal. Behav. Neurosci. 116, 363–377.
Hartley, T., Bird, C.M., Chan, D., Cipolotti, L., Husain, M., Vargha-Khadem, F.,and Burgess, N. (2007). The hippocampus is required for short-term topo-graphical memory in humans. Hippocampus 17, 34–48.
Hassabis, D., Kumaran, D., Vann, S.D., and Maguire, E.A. (2007). Patients withhippocampal amnesia cannot imagine new experiences. Proc. Natl. Acad. Sci.USA 104, 1726–1731.
Higuchi, S., and Miyashita, Y. (1996). Formation of mnemonic neuronalresponses to visual paired associates in inferotemporal cortex is impaired byperirhinal and entorhinal lesions. Proc. Natl. Acad. Sci. USA 93, 739–743.
676 Neuron 61, March 12, 2009 ª2009 Elsevier Inc.
Hochstein, S., and Ahissar, M. (2002). View from the top: Hierarchies andreverse hierarchies in the visual system. Neuron 36, 791–804.
Hood, K.L., Postle, B.R., and Corkin, S. (1999). An evaluation of the concurrentdiscrimination task as a measure of habit learning: performance of amnesicsubjects. Neuropsychologia 37, 1375–1386.
Iwai, E., and Mishkin, M. (1968). Two visual foci in the temporal lobe ofmonkeys. In Neurophysiological Basis of Learning and Behavior, N. Yoshiiand N.A. Buchwald, eds. (Osaka, Japan: Osaka University Press), pp. 1–11.
Jonasson, Z., Ballantyne, J.K., and Baxter, M.G. (2004). Preserved antero-grade and retrograde memory of rapidly acquired olfactory discrminationsafter neurotoxic hippocampal lesions. Hippocampus 14, 28–39.
Lee, A.C., Buckley, M.J., Pegman, S.J., Spiers, H., Scahill, V.L., Gaffan, D.,Bussey, T.J., Davies, R.R., Kapur, N., Hodges, J.R., and Graham, K.S.(2005a). Specialization in the medial temporal lobe for processing of objectsand scenes. Hippocampus 15, 782–797.
Lee, A.C., Bussey, T.J., Murray, E.A., Saksida, L.M., Epstein, R.A., Kapur, N.,Hodges, J.R., and Graham, K.S. (2005b). Perceptual deficits in amnesia:challenging the medial temporal lobe ’mnemonic’ view. Neuropsychologia43, 1–11.
Lee, A.C., Levi, N., Davies, R.R., Hodges, J.R., and Graham, K.S. (2007).Differing profiles of face and scene discrimination deficits in semanticdementia and Alzheimer’s disease. Neuropsychologia 45, 2135–2146.
Lee, A.C., Scahill, V.L., and Graham, K.S. (2008). Activating the medialtemporal lobe during oddity judgment for faces and scenes. Cereb. Cortex18, 683–696.
Levy, D.A., Shrager, Y., and Squire, L.R. (2005). Intact visual discrimination ofcomplex and feature-ambiguous stimuli in the absence of perirhinal cortex.Learn. Mem. 12, 61–66.
Meunier, M., Bachevalier, J., Mishkin, M., and Murray, E.A. (1993). Effects onvisual recognition of combined and separate ablations of the entorhinal andperirhinal cortex in rhesus monkeys. J. Neurosci. 13, 5418–5432.
Mitchell, A.S., Baxter, M.G., and Gaffan, D. (2007). Dissociable performanceon scene learning and strategy implementation after lesions to magnocellularmediodorsal thalamic nucleus. J. Neurosci. 27, 11888–11895.
Morris, R.G.M. (1984). Is the distinction between procedural and declarativememory useful with respect to animal models of amnesia? In Neurobiologyof Learning and Memory, G. Lynch, J. McGaugh, and N. Weinberger, eds.(New York: Guilford Press), pp. 119–124.
Moss, H.E., Rodd, J.M., Stamatakis, E.A., Bright, P., and Tyler, L.K. (2005).Anteromedial temporal cortex supports fine-grained differentiation amongobjects. Cereb. Cortex 15, 616–627.
Murray, E.A. (1996). What have ablation studies told us about the neuralsubstrates of stimulus memory? Semin. Neurosci. 8, 13–22.
Murray, E.A., and Wise, S.P. (2004). What, if anything, is the medial temporallobe, and how can the amygdala be part of it if there is no such thing?Neurobiol. Learn. Mem. 82, 178–198.
Murray, E.A., Bussey, T.J., Hampton, R.R., and Saksida, L.M. (2000). Theparahippocampal region and object identification. Ann. N Y Acad. Sci. 911,166–174.
Murray, E.A., Graham, K.S., and Gaffan, D. (2005). Perirhinal cortex and itsneighbours in the medial temporal lobe: contributions to memory and percep-tion. Q. J. Exp. Psychol. B 58, 378–396.
Murray, E.A., Bussey, T.J., and Saksida, L.M. (2007). Visual perception andmemory: a new view of medial temporal lobe function in primates and rodents.Annu. Rev. Neurosci. 30, 99–122.
Nadel, L. (1992). Multiple memory systems: What and why. J. Cogn. Neurosci.4, 179–188.
Ringo, J.L. (1988). Seemingly discrepant data from hippocampectomizedmacaques are reconciled by detectability analysis. Behav. Neurosci. 102,173–177.
Neuron
Review: Point/Counterpoint
Ringo, J.L. (1991). Memory decays at the same rate in macaques with andwithout brain lesions when expressed in d’ or arcsine terms. Behav. BrainRes. 42, 123–134.
Ringo, J.L. (1993). Spared short-term memory in monkeys following medialtemporal lobe lesions is not yet established: A reply to Alvarez-Royo, Zola-Morgan and Squire. Behav. Brain Res. 59, 65–72.
Saksida, L.M., Bussey, T.J., Buckmaster, C.A., and Murray, E.A. (2006).No effect of hippocampal lesions on perirhinal cortex-dependent feature-ambiguous visual discriminations. Hippocampus 16, 421–430.
Shrager, Y., Gold, J.J., Hopkins, R.O., and Squire, L.R. (2006). Intact visualperception in memory-impaired patients with medial temporal lobe lesions.J. Neurosci. 26, 2235–2240.
Shrager, Y., Levy, D.A., Hopkins, R.O., and Squire, L.R. (2008). Workingmemory and the organization of brain systems. J. Neurosci. 28, 4818–4822.
Squire, L.R., and Knowlton, B.J. (2000). The medial temporal lobe, the hippo-campus, and the memory systems of the brain. In The New Cognitive Neuro-sciences, M.S. Gazzaniga, ed. (Cambridge, MA: MIT Press), pp. 765–779.
Squire, L.R., and Zola-Morgan, S. (1991). The medial temporal lobe memorysystem. Science 253, 1380–1386.
Squire, L.R., Stark, C.E., and Clark, R.E. (2004). The medial temporal lobe.Annu. Rev. Neurosci. 27, 279–306.
Stark, C.E., and Squire, L.R. (2000). Intact visual perceptual discrimination inhumans in the absence of perirhinal cortex. Learn. Mem. 7, 273–278.
Suzuki, W.A. (2009). Perception and the medial temporal lobe: evaluating thecurrent evidence. Neuron 61, this issue, 657–666.
Suzuki, W.A., and Baxter, M.G. (2009). Memory, perception and the medialtemporal lobe: a synthesis of opinions. Neuron 61, this issue, 678–679.
Suzuki, W.A., Zola-Morgan, S., Squire, L.R., and Amaral, D.G. (1993). Lesionsof the perirhinal and parahippocampal cortices in the monkey produce long-lasting memory impairment in the visual and tactual modalities. J. Neurosci.13, 2430–2451.
Taylor, K.J., Henson, R.N., and Graham, K.S. (2007). Recognition memory forfaces and scenes in amnesia: dissociable roles of medial temporal lobe struc-tures. Neuropsychologia 45, 2428–2438.
Teng, E., Stefanacci, L., Squire, L.R., and Zola, S.M. (2000). Contrastingeffects on discrimination learning after hippocampal lesions and conjointhippocampal-caudate lesions in monkeys. J. Neurosci. 20, 3853–3863.
Tyler, L.K., Stamatakis, E.A., Acres, K., Bright, P., Abdallah, S., Rodd, J.M.,and Moss, H.E. (2004). Processing objects at different levels of specificity.J. Cogn. Neurosci. 16, 351–362.
Victor, M., Adams, R.D., and Collins, G.H. (1989). The Wernicke-KorsakoffSyndrome, Second Edition (Philadelphia, PA: F.A. Davis).
Zola-Morgan, S., and Squire, L.R. (1985). Amnesia in monkeys after lesions ofthe mediodorsal nucleus of the thalamus. Ann. Neurol. 17, 558–564.
Zola-Morgan, S., and Squire, L.R. (1986). Memory impairment in monkeysfollowing lesions limited to the hippocampus. Behav. Neurosci. 100, 155–160.
Zola-Morgan, S., Squire, L.R., Amaral, D.G., and Suzuki, W.A. (1989). Lesionsof perirhinal and parahippocampal cortex that spare the amygdala andhippocampal formation produce severe memory impairment. J. Neurosci. 9,4355–4370.
Neuron 61, March 12, 2009 ª2009 Elsevier Inc. 677