REVIEWS

A CORTICAL–HIPPOCAMPAL SYSTEM
FOR DECLARATIVE MEMORY
Howard Eichenbaum
Recent neurobiological studies have begun to reveal the cognitive and neural coding
mechanisms that underlie declarative memory — our ability to recollect everyday events and
factual knowledge. These studies indicate that the critical circuitry involves bidirectional
connections between the neocortex, the parahippocampal region and the hippocampus.
Each of these areas makes a unique contribution to memory processing. Widespread high-
order neocortical areas provide dedicated processors for perceptual, motor or cognitive
information that is influenced by other components of the system. The parahippocampal
region mediates convergence of this information and extends the persistence of neocortical
memory representations. The hippocampus encodes the sequences of places and events that
compose episodic memories, and links them together through their common elements. Here I
describe how these mechanisms work together to create and re-create fully networked
representations of previous experiences and knowledge about the world.

EPISODIC REPRESENTATIONS Our everyday, conscious memories are not a passive or this area is dedicated to memory independent of other
Neural firing patterns, which disconnected collection of images and associations. cognitive functions. In addition, even within memory,
encode the sequence of events Rather, we create an intricate network of images and the role of the hippocampal region is selective both to a
that compose a unique, associations, constituting a record of our personal particular time window and a particular domain of
personal experience.
experiences that is continuously updated through an memory processing.
SEMANTIC KNOWLEDGE active organization of new information within the Immediate memory, the ability to repeat or recog-
An organization of factual context of previous experience1. Recollection is simi- nize items just brought into consciousness, is intact in
information independent of the larly re-creative, and the course of remembering is patients with damage to the hippocampal region3,4.
specific episodes in which that
information was acquired.
therefore determined by the nature of our memory Also, remote childhood memories and general world
organization. This type of memory is called ‘declara- knowledge acquired early in life are not affected. These
tive memory’, and is therefore a multifaceted process findings indicate that the hippocampal region plays a
involving a synthesis of EPISODIC REPRESENTATIONS with critical role between the initial formation of memories
our framework of general SEMANTIC KNOWLEDGE that and their final repository elsewhere in the brain. It has
mediates our capacity for recollection. In this review, I been suggested that the hippocampus is always critical
will summarize recent progress in characterizing a for the expression of autobiographical and spatial
functional circuit diagram for the brain system that memories5. However, it was recently shown that the
mediates declarative memory. role of the hippocampal region is time-limited even for
Laboratory of Cognitive memory of spatial environments learned in child-
Neurobiology, Department Hippocampal region and declarative memory hood6. The duration of critical hippocampal involve-
of Psychology, Boston The hippocampal region has been identified as central ment may depend on mediation by cortical areas
University, 64 Cummington
Street, Boston,
to our capacity for declarative memory. Scoville and adjacent to the hippocampus, and damage outside the
Massachusetts 02215, USA. Milner’s2 initial report of memory loss in humans fol- hippocampal region can result in temporally extensive
e-mail: hbe@[Link] lowing removal of the hippocampal region showed that memory loss7,8.

NATURE REVIEWS | NEUROSCIENCE VOLUME 1 | OCTOBER 2000 | 4 1

© 2000 Macmillan Magazines Ltd
REVIEWS

In contrast to the observations of temporal specifici- We are beginning to characterize the neural circuitry
ty, the early evidence indicated that the domain of mem- and information processing mechanisms that mediate
ory dependent on the hippocampal region was ‘global’. these aspects of memory through the use of animal
However, it is now clear that there are several memory models. Recent studies have shown that the general pat-
systems in the brain, of which the hippocampal system is tern of memory deficits and spared capacities, following
only one9. As Cohen and Squire10 first recognized, the damage to the hippocampal region in monkeys and
hippocampal region functions selectively in declarative rats, parallels the phenomenology of amnesia in
memory. Although the terminology used to characterize humans22,23. Sensory, motor, motivational and cognitive
this kind of memory has varied, there is consensus that processes are intact following hippocampal damage,
the phenomenology of declarative memory is composed confirming that this structure functions selectively in
of our capacity for episodic and semantic memory, and memory in animals as in humans. The role of the
our ability for conscious recollection and ‘flexible’ mem- hippocampal region in animals is limited both to the
ory expression11. By contrast, the hippocampal region is transition from immediate to permanent memory, and
not required for the acquisition of many skills and biases to a particular domain of memory. So as in human
that can be expressed unconsciously through alterations amnesics, animals with damage to the hippocampal
in performance on a broad variety of tasks (for example, region can have an intact immediate memory and sub-
REFS 12,13,17). Instead, systems that include the neostria- sequent loss of memory after interpolated material or
tum and cerebellum mediate PROCEDURAL MEMORY, the delay24,25. This pattern of intact immediate memory and
acquisition of motor skills and habits13–18. A system that abnormally rapid forgetting has also been observed after
includes the amygdala mediates EMOTIONAL MEMORY17 and damage limited to the hippocampus itself in some cases
modulates the strength and consolidation of memories of nonspatial memory26 and spatial memory27.
in other memory systems19. Cortical regions are critical However, in other spatial tasks an impairment is
in short-term or WORKING MEMORY20, and in the PRIMING of observed at the briefest possible delays28,29. In addition,
recently experienced stimuli21, as well as in long-term several studies on animals have shown that memories
declarative memory (see below). acquired shortly before hippocampal damage are lost

Primate Rodent

Neocortical
association
areas

PROCEDURAL MEMORY
The representation of a series of
actions or perceptual processing Parahippocampal
functions that occur region
unconsciously, and typically
result in increased speed or
accuracy with repetitions.

EMOTIONAL MEMORY
The representation of a positive
or negative affect associated
with specific stimuli. Typically Hippocampus
not subject to conscious
recollection but reflected in
attraction, avoidance or
autonomic nervous system
activation.

WORKING MEMORY
Figure 1 | The anatomy of the hippocampal memory system. In both monkeys and rats the origins of specific information
The representation of items
for the hippocampus include virtually every neocortical association area102,103. Each of these neocortical areas (blue) project to
held in consciousness during
one or more subdivisions of the parahippocampal region, which includes the perirhinal cortex (purple), the parahippocampal
experiences or after retrieval of
(or postrhinal) cortex (dark purple) and the entorhinal cortex (light purple)42,43. The subdivisions of the parahippocampal region are
memories. Short-lasting and
associated with active rehearsal interconnected and send principal efferents to many subdivisions of the hippocampus itself (green), the dentate gyrus, the CA3
or manipulation of and CA1 areas, and the subiculum. So the parahippocampal region serves as a convergence site for cortical input and mediates
information. the distribution of cortical afferents to the hippocampus. Within the hippocampus, there are broadly divergent and convergent
connections that could mediate a large network of associations104, and these connections support plasticity mechanisms that
PRIMING could participate in the rapid coding of new conjunctions of information105. The outcome of hippocampal processing is directed
The facilitation of recognition, back to the parahippocampal region, and the output of that region is directed in turn back to the same areas of the cerebral
reproduction or biases in cortex that were the source of input to this region42,43. Further structures have been included in this system, including the medial
selection of stimuli that have diencephalic structures that connect with the hippocampus along with other subcortical areas, through a major fibre bundle
recently been perceived. called the fornix106.

42 | OCTOBER 2000 | VOLUME 1 [Link]/reviews/neuroscience

© 2000 Macmillan Magazines Ltd
 REVIEWS

in both rats and monkeys using a simple recognition

Box 1 | The ‘delayed nonmatch to sample’ task
memory task, called ‘delayed nonmatch to sample’
In this test subjects are initially presented with a single stimulus, called the sample, and (DNMS), where subjects must remember a single
must indicate that they have perceived it by an appropriate behavioural response. stimulus across a variable memory delay (BOX 1)23,44.
Subsequently the sample stimulus is removed and must be remembered across a There is emerging evidence that neocortical associa-
variable delay. In the memory test phase, the subject is presented with the sample tion areas and the parahippocampal region have distinct
concurrently or sequentially along with an alternative stimulus. The subject is required and complementary functions in DNMS performance.
to select against the sample in favour of the alternative, that is, to non-match to the In rats performing an odour-guided version of the
sample. In a version of the task commonly used in monkey studies, the sample and DNMS task (BOX 1), damage to the orbitofrontal cortex
alternative stimuli are three-dimensional ‘junk’ objects that are used on only one trial
resulted in a deficit in the acquisition of the task when
and then never presented again95. Typically the animals are initially trained with a
the memory delay was minimal, suggesting that it is
minimal memory delay, and the delay is subsequently elongated to increase the
important in perceptual processing or in learning the
memory demand. Several variants of the task are used in other behavioural and
nonmatching rule45. By contrast, rats with damage to
physiological studies. In physiological studies on monkeys, video-pictures are the
stimuli, the test stimuli are presented sequentially, and a match-to-sample is the parahippocampal region acquired the DNMS task
required63,64. In a ‘continuous nonmatch to sample’ variant of the task, used in at the normal rate and did well at brief memory delays.
behavioural and physiological studies on rats, the stimuli are a continuous series of However, their memories declined abnormally rapidly
odours for which each stimulus acts as both the test of memory for the previous as the memory delay was extended beyond a few sec-
stimulus and as the sample to be remembered on the next trial45. onds, indicating a selective role in maintaining a persis-
tent memory of the sample stimulus. Little if any deficit
whereas memories acquired much earlier are spared, in nonspatial DNMS is observed following damage to
similar to the temporally graded retrograde memory the hippocampus or its connections through the
loss observed in human amnesic patients30–35. fornix28,45–47, indicating that the parahippocampal region
In addition, in animals as in humans, the domain of itself mediates the persistence of memories for single
memory dependent on the hippocampal region is selec- items required to perform DNMS.
tive to a particular type of memory processing. It is im- Parallel results have been obtained in monkeys per-
possible to assess in animals some aspects of declarative forming visually guided versions of the DNMS task
memory, such as conscious recollection. Nevertheless, (BOX 1). Similar to rats, monkeys with damage to the
several studies have succeeded in showing a selective parahippocampal region do well when the memory
role for the hippocampal region in mediating other cen- delay is brief. But when the memory demand is
tral features of declarative memory. These include the increased by extending the delay period, severe deficits
linking of memories within a network of semantic in DNMS are observed48,49, and these impairments are
knowledge and flexible, inferential expression of memo- more severe than that following damage to the hippo-
ries, as outlined below (see section on ‘memory process- campus50 or its connections through the fornix51.
ing within the hippocampus’)36. Conversely, there is Examination of performance on the DNMS task with
abundant evidence that other brain systems in animals brief delays has been difficult because the standard pro-
mediate procedural learning37, emotional memory38–40 tocol used for monkeys is manual. However, using
and memory modulation41. These findings validate the another recognition task that allowed testing at very
application of animal models to the study of memory, brief delays, it has recently been shown that the infero-
and set the stage for a detailed neurobiological analysis temporal area of the cortex is critical for visual recogni-
aimed at identifying the relevant pathways and func- tion even for a one second delay, indicating a possible
tional mechanisms of the declarative memory system. function in perceptual processing as opposed to memo-
ry. In contrast, the parahippocampal region was critical
Dissecting the hippocampal memory system for memory in the same task only when recognition
The hippocampal memory system is composed of three was delayed52. The parahippocampal region may also
principal components: cerebral cortical areas, the para- act at the intersection of perception and memory in sit-
hippocampal region and the hippocampus itself42,43. uations where perceptual processes depend on learned
The main pathways are similar in rodents and primates associations among complex stimulus elements53,54.
(FIG. 1). This anatomical organization complements the Parallel electrophysiological studies that involve re-
findings from studies of amnesia, leading to the working cording from single cells in these same brain areas have
hypothesis that the parahippocampal region and hippo- provided a preliminary understanding of the neural
campus contribute to memory by altering the nature, coding mechanisms that underlie DNMS performance.
persistence and organization of memory representa- In both monkeys and rats, three general responses have
tions within the cerebral cortex. been observed (FIG. 2)55,59. First, many cells showed selec-
tive tuning to sample stimuli during the initial percep-
Different roles of cortical areas tion of the stimulus, indicating that these areas encode
To fully understand the contribution of the hippo- specific stimuli. Second, some cells continued firing in a
campus to memory, it is essential to characterize the stimulus-specific fashion during a memory period when
nature of the information processing performed by the the cue was no longer present, indicating the persistence
neocortical association areas and the parahippocampal of a representation of the sample. Lastly, many cells
region — areas that project to and are influenced by the showed enhanced or suppressed responses to the famil-
hippocampus. The role of these areas has been studied iar stimuli when they reappeared in the memory test

NATURE REVIEWS | NEUROSCIENCE VOLUME 1 | OCTOBER 2000 | 4 3

© 2000 Macmillan Magazines Ltd
REVIEWS

a addition, neurons in perirhinal and inferotemporal cor-

tex areas show long-lasting decrements in responsiveness

Firing rate
to highly familiar stimuli, which could provide signals
about familiarity for extended periods63,64.
It is difficult at this time to directly compare the data
Time across species from studies that use different experimen-
tal strategies, focus on different components of the pre-
b frontal and temporal cortex, and use different variants of
recognition memory tests. However, the evidence is gen-
Firing rate
erally consistent with the idea that several neocortical
and parahippocampal areas serve distinct functions in
recognition memory. Neocortical areas have specific
Time functions in the perceptual or cognitive processing
required to complete the task, and are able to mediate
c some aspects of working or short-term memory. The
parahippocampal region makes a different contribution.
Firing rate

This region seems to be critical in extending the persis-

tence of memory for single stimuli over brief periods in
the absence of interference, and maintains information
Time about stimulus familiarity for prolonged periods even
with interference.
It seems that memory mediated by the hippocampus
Sample Delay Test itself is not critical for performance in standard DNMS
tasks, in that the deficits observed are, at most, modest
Figure 2 | Firing patterns associated with memory
performance in ideal cortical neurons. Responses are
compared with the effects of damage to the parahippo-
shown for a sample period when a stimulus is presented, a campal region. However, the hippocampus seems to be
delay period during which it must be remembered, and finally essential in other types of simple recognition memory
a match period when the memory of the sample must be tests26,65 and in memory for configurations of items
matched to choice stimuli (BOX 1). a | Selective activation for within scenes or places66–69.
some stimuli (green) compared with others (red), consistent
with an encoding of stimulus properties. b | Stimulus-selective
Memory processing within the hippocampus
activation (green) during a delay period when the memory of
a stimulus must be maintained, consistent with a role in The findings from studies using animal models point to
maintaining a representation of the sample. c | Enhanced a critical role for the hippocampus itself in central
(dotted line) or suppressed (solid line) responses to stimulus aspects of declarative memory. To understand this role
repetition compared with initial presentations, consistent with it is important to reconsider the fundamental properties
processing comparisons between the sample and choice of declarative memory introduced earlier. We acquire
stimuli.
our declarative memories through everyday personal
experiences, and the ability to retain and recall these
phase of the task, indicating involvement in the match/ ‘episodic’ memories is highly dependent on the hippo-
nonmatch judgment. campus in humans70. But the full scope of hippocampal
All three types of representations have been found in involvement also extends to semantic memory71. For
several neocortical areas and in the parahippocampal example, a typical episodic memory might involve
region, indicating that information about all aspects of recalling the specific events and places surrounding the
the task may be shared among these areas. However, it is meeting of a long-lost cousin. Your general knowledge
likely that each area makes a distinct contribution to the about your family tree, and other facts about the history
performance of the task. For example, in rats more cells of your family, comes in great part from a synthesis of
in the parahippocampal region showed sustained stimu- the representations of many meetings with relatives and
lus-specific activity during the delay, whereas more cells other episodes in which family personalities or events
in the orbitofrontal area showed stimulus-selective are observed or discussed. Similarly, our episodic mem-
match enhancement or suppression60. In monkeys, a ory mediates the capacity to remember a sequence of
greater proportion of cells in the lateral prefrontal region events, places passed, and turns taken while walking
showed sustained responses during the delay, and con- across a city, and a synthesis of many such representa-
veyed more information about the match–nonmatch tions provides general knowledge about the spatial
status of the test stimuli compared with the perirhinal layout of the city.
cortex in a task where the memory delay was filled with In addition, declarative memory for both the episod-
SPATIAL LEARNING interpolated material61. By contrast, more neurons in the ic and semantic information is special in that the con-
Acquisition of information perirhinal cortex and inferotemporal cortex show tents of these memories are accessible through various
about spatial relations among greater stimulus selectivity. Furthermore, in a recogni- routes. Most commonly in humans, declarative memo-
objects in the environment,
typically reflected in the ability
tion task where the memory delay is not filled with inter- ry is expressed through conscious, effortful recollection.
to navigate through the polated material, a large fraction of temporal neurons This means that you can access and express declarative
environment using new routes. show sustained stimulus-specific delay activity62. In memories to solve new problems by making inferences

44 | OCTOBER 2000 | VOLUME 1 [Link]/reviews/neuroscience

© 2000 Macmillan Magazines Ltd
 REVIEWS

a from memory. For example, even without ever explic-

itly studying your family tree, you can infer indirect
relationships or the sequence of central events in the
family history, from the set of episodic memories
about your family. Similarly, without ever studying the
map of a city, you can make navigational inferences
from the synthesis of many episodic memories of pre-
vious routes taken. Large-scale networks for family
trees and city layouts are but two examples of the kind
of ‘memory space’ proposed to be mediated by the
hippocampal system72. Within this view, a broad range
of such networks can be created, with their central
organizing principle the linkage of episodic memories
through their common events and places, and a conse-
quent capacity to move among related memories with-
in the network.
These properties of declarative memory suggest an
approach for the development of animal models. So a
b Variable start positions Constant start position
way to study the creation of a memory space from over-
lapping experiences, and to make inferences from the
network knowledge, is to train subjects on several dis-
tinct experiences that share common elements and then
test whether these experiences have been linked in
memory to solve new problems. One can conceive of
this approach as applied to various domains relevant to
the lives of animals, from knowledge about spatial rela-
c 70 120 tions among stimuli in an environment, to categoriza-
60 100 tions of foods, learned organizations of odour or visual
stimuli, or social relationships. Progress is being made
Mean latency (s)

Mean latency (s)

50
80
in investigating these domains.
40
60 In some experimental protocols, the requirement to
30 synthesize several overlapping experiences is enough to
40
20 require hippocampal function. One case involves SPATIAL
10 20 LEARNING, similar to the example of the learning of routes

0 0
through a city given above, but involving rats and the
0 2–6 7–12 13–18 0 2 4 6 8 10 12 Morris water maze task. In this test, rats or mice learn to
Trials Trials escape from submersion in a pool by swimming
d 20 e Control towards a platform located just underneath the surface.
Importantly, training in the conventional version of the
task involves an intermixing of four kinds of trial
Escape latency (s)

episodes that differ in the starting point of the swim.

Under this condition, animals with hippocampal dam-
10
age typically fail to acquire the task73. However, if the
Lesion
demand for synthesizing a solution from four types of
episodes is eliminated by allowing the animal to repeat-
edly start from the same start position, animals with
0 hippocampal damage acquire the task almost as readily
Instruction trials Probe trials
as normal rats and use the same distant spatial cues in
Figure 3 | Performance of rats with hippocampal damage in the Morris water maze. identifying the escape site74 (FIG. 3).
a | An illustration of the Morris water maze and typical environmental cues74. The escape Other experiments indicate that the hippocampus
platform, submerged just below the surface of the water, cannot be seen by the rat. b | In the
conventional version of the task (left), the rat begins each trial from one of four starting locations,
may be required for new problem solving in familiar
and the time required for it to locate the escape platform is measured. In the constant start environments. So when rats with hippocampal damage
position version of the task (right), one start location is used consistently. c | In the conventional that have successfully learned to locate the escape plat-
version of the task (left), normal rats (blue) rapidly improve their swim latencies to find the platform form from a single start position are tested from new
across trials, whereas rats with hippocampal damage (red) do not. In the constant start position start positions, they fail to readily locate the platform.
version of the task (right), rats with hippocampal damage are slightly impaired in acquisition In contrast, normal animals swim directly to the escape
rate, but successfully learn to locate the platform. d | During probe testing, normal rats (blue)
locus on each new probe trial (FIG. 3)74. In another
rapidly locate the escape platform both on repetitions of the original instruction trials and on
probe trials that begin at new start positions. Rats with hippocampal damage (red) also do well example, hippocampal damage results in failure to
on repetitions of the instruction trials, but poorly on the probe trials. e | Example swim paths in express memory for a single experience in social learn-
new probe trials by normal rats (blue) and rats with hippocampal damage (red). Normal rats ing of food odours. Training in this task involves a
swim directly to the platform, but rats with hippocampal damage are severely impaired. social encounter during which the subject interacts

NATURE REVIEWS | NEUROSCIENCE VOLUME 1 | OCTOBER 2000 | 4 5

© 2000 Macmillan Magazines Ltd
REVIEWS

a b 85 choices by the subject. Rats with selective hippocampal

Demonstrator damage show intact memory when tested immediately

Trained odour selection (%)

75
after the social encounter, but no memory when the
test is delayed by 24 hours (FIG. 4)33,76. The observation
Exchange of information
of intact short-term memory is similar to the sparing of
65
immediate memory in humans with amnesia, and
indicates that the hippocampus is not required for the
55
Chance perceptual or motivational components of learning, for
the critical social interactions, or for the ability to
45
Preference test express a learned food selection. The loss of differential
choice behaviour within a day indicates that the hippo-
35 campus is required for expressing the memory acquired
Immediate 24 hr delay
during a single social encounter in a new situation
Figure 4 | The social transmission of food preferences task. a | Initially a ‘demonstrator’ involving food selection.
rat eats food containing a new odour. Then, during a social encounter, the demonstrator In several other experimental protocols, animals
exchanges information about the food odour with the subject rat76. Subsequently the subject with hippocampal damage successfully acquire a set of
is given a preference test for the new food odour versus another food odour. b | Preference overlapping experiences, often at a rate not substantial-
test results. Normal rats (blue) show a strong preference for the demonstrated food odour both
immediately and one day following the social encounter. Rats with hippocampal lesions (red)
ly different from that of normal subjects. But they fail
shown intact performance on the immediate test but forget within one day. to express their memories of the experience in new sit-
uations that require an inference on the basis of linking
the distinct experiences in memory (BOX 2). In one of
with a ‘demonstrator’ rat that has recently eaten a par- these studies, rats were trained on sets of odour ‘paired
ticular food (FIG. 4). During this exposure the subject associates’ with shared elements and were then tested to
sniffs the breath of the demonstrator and acquires an see if they could infer an association between elements
association between the odour of the recently eaten that were only indirectly related77. In another study, rats
food and an odorous constituent of rat’s breath, carbon were trained on a series of four odour discriminations,
disulphide75. The subsequent memory test involves with shared items such that the odour set could be con-
presentation of a choice of foods, one of which is the strued as a hierarchy, and then were tested to see if they
same kind of food eaten by the demonstrator, in the could infer transitive relations according to the hierar-
absence of the social context. Memory for the learned chical organization78. The results of these studies
association is reflected in an alteration of food selection showed that some forms of stimulus–stimulus repre-
sentations can be acquired independently of the hippo-
campus itself. However, these representations are
Box 2 | Tests of transitive inference ‘hyperspecific’, that is, they can only be expressed with-
In one experiment rats learned overlapping sets of associations between odour in the confined context of the reproduction of each of a
stimuli77. On each trial one of two odours was initially presented, followed by a choice set of distinct learning events79. Only a hippocampally
between two odours, one of which was baited as the assigned ‘associate’ for a particular mediated representation can support the inferential
initial odour (A goes with B, not Y; X goes with Y, not B). Following training on two expression of associations that must be linked across
sets of overlapping odour–odour associations (A–B and X–Y, then B–C and Y–Z), separated experiences.
subsequent probe tests were used to characterize the extent to which learned
representations could be linked to support inferential memory expression. Control Linking episodic memories in the hippocampus
rats learned paired associates rapidly and hippocampal damage did not affect How are these memory capacities mediated within the
acquisition rate on either of the two training sets. Intact rats also showed that they circuitry of the hippocampus? Recent observations
could link the information from overlapping experiences, and use this information to from extracellular recordings in behaving animals indi-
make inferential judgments in two ways. First, normal rats showed strong transitivity cate that hippocampal neuronal networks may repre-
across odour pairings that contained a shared item. For example, having learned that sent sequences of events and places that compose
odour A goes with odour B, and B goes with C, they could infer that A goes with C. episodic memories. The content of information encod-
Second, control rats could infer symmetry in paired associate learning. For example, ed by the firing patterns of these neurons includes both
having learned that B goes with C, they could infer that C goes with B. By contrast, rats
specific conjunctions of events and places unique to
with selective hippocampal lesions were severely impaired, showing no evidence of
particular experiences and features that are common to
transitivity or symmetry. A subsequent study examined the ability of rats to solve the
overlapping experiences. Indeed, there is now evidence
classic transitive inference task78. Animals initially learned a series of overlapping
pairwise discrimination problems in which they were rewarded for selecting one
that the hippocampus creates separate and linked
odour over another (A > B, B > C, C > D and D > E, where the item before ‘>’ is to be episodic-like representations even when the overt
selected over the other item). In later probe tests, all the initial pairs were presented in behaviours, and places where they occur, are the same
random order, together with occasional probe trials with the pair B versus D as the but the events are parts of distinct experiences.
critical test of transitive inference. Control rats acquired each of the premise pairs Hippocampal principal cells show firing patterns
rapidly, and showed a robust capacity for transitive inference, indicating that rats are that are readily related to a broad range of events, which
capable of linking information about the odours acquired across distinct experiences, occur during sequences of behaviour in all tasks exam-
and of making inferential judgments based on knowledge about the orderly series. ined (BOX 3)72. For example, as rats complete spatial tasks
Animals with different kinds of hippocampal damage acquired the premise pairs at the where they are required to shuttle between a common
normal rate but showed no capacity for the transitive inference. starting location and one or more reward locations,

46 | OCTOBER 2000 | VOLUME 1 [Link]/reviews/neuroscience

© 2000 Macmillan Magazines Ltd
 REVIEWS

cells fired as the rat approached the odour stimulus, or

Box 3 | Hippocampal ‘place’ cells, and more
as it sniffed a particular odour, regardless of where the
In 1971, O’Keefe and Dostrovsky96 reported the observation of principal cells in the trial occurred, and cells fired as the rat performed the
hippocampus that fired when a rat was in a particular location in its environment. trial at a particular location, regardless of what odour
Since then, there have been many characterizations of hippocampal ‘place’ cells. The was presented (FIG. 5).
spatial coding properties of these cells are most readily observed in rats moving Finally, there is emerging evidence of coding for
randomly in an open field while foraging for food97. In this task, the animal’s information specific to particular types of episodes
movements and behaviours are homogeneous throughout the environment, but many even in situations where the overt behavioural events
hippocampal cells are active only when the animal traverses a particular area within and the locations in which they occur are identical
the environment98. These findings have been interpreted as evidence that the between several types of experience. For example, in a
hippocampus is dedicated to mapping spatial layouts of the environment99. However,
spatial DNMS task, some hippocampal cells were acti-
information about places reflects only part of the domain of hippocampal information
vated when the rat was pressing one of two levers dur-
coding. In many situations where rats move towards or away from important locations
ing the sample, or during the test phase of the task86.
in the environment, the spatial firing patterns of these cells are strongly affected by the
direction and speed of movement81,82, by the targets of movement within the
These cells can be characterized as elements encoding
environment80, and by demands of the behavioural test (FIG. 6)83,100. Furthermore, one temporally, spatially and behaviourally defined
hippocampal cell firing is also associated with many nonspatial events, including event in the network representation of a particular trial
conditioned behavioural responses84, olfactory cues85 and, in humans, categories of type. The firing of other cells was associated with com-
visual stimuli101. mon events (a particular lever position) regardless of
trial phase, or during the sample or test phase regardless
hippocampal PLACE CELLS fire during each moment as the of location. These cells could be used to link the sepa-
animal traverses its path, with each neuron activated rate representations of different trial phases or episodes,
when the animal is in a particular place and moving and these codes were topographically segregated within
toward the goal. A largely different set of cells fires simi- the hippocampus. More direct evidence of episodic-like
larly in sequence as the rat returns to the starting point, coding was found in a recent study where rats per-
such that each cell can be characterized as an element of formed a spatial alternation task on a T-maze (FIG. 6).
a network representing an outbound or inbound part of Each trial commenced when the rat traversed the stem
the episode80–83. One can imagine the network activity as of the ‘T’ and then selected either the left- or the right-
similar to a videoclip of each trial episode, with each cell choice arm87. To alternate successfully, the rats were
capturing the information about where the rat is and required to distinguish between their left-turn and
what it is doing in each sequential ‘frame’ of the clip. right-turn experiences and to use their memory of the
Similarly, in both simple and complex learning tasks, most recent experience to guide the current choice.
hippocampal cells fire at virtually every moment associ- Different hippocampal cells fired as the rats passed
ated with specific relevant events83,84.. For example, through the sequence of locations within the maze dur-
when rats complete an odour discrimination task, ing each trial. Most important, the firing patterns of
hippocampal cells fire during each sequential event, many of the cells depended on whether the rat was in
with different neurons firing during the approach to the the midst of a left- or right-turn episode, even when the
odour stimuli, sampling of odours, execution of a rat was on the stem of the T and running similarly on
behavioural response and reward consumption83. Again, both types of trials — minor variations in the animal’s
it is as if each hippocampal cell encodes one of the speed, direction of movement or position within areas
sequential trial events with its activity reflecting both on the stem did not account for the different firing pat-
aspects of the continuing behaviour and the place where terns on left-turn and right-turn trials. Other cells fired
that behaviour occurred. In all of these situations, some when the rat was at the same point in the stem on either
cells fire during common events or places that occur on trial type. Therefore, the hippocampus encoded both
every trial, whereas the firing of other cells was associated the left-turn and right-turn experiences using distinct
with events that occurred only during a particular type representations, and included elements that could link
of episode, such as sampling a particular configuration them by their common features. In each of these experi-
of two odours presented on that trial. ments, the representations of event sequences, linked by
In an extension of these studies, we were recently codings of their common events and places, could con-
able to distinguish hippocampal neurons that encoded stitute the substrate of a network of episodic memories.
specific combinations of both events and places, which
were unique to particular experiences as well as partic- Organization and consolidation
ular features that were common across many related The studies described here indicate that each principal
experiences85. In this experiment, rats performed a component of the memory system contributes differen-
variant of the DNMS task at several locations in an tially to declarative memory, although interactions
open field. Again, different cells fired during each between these areas are also essential. Initially, perceptu-
sequential trial event. Some cells were activated only in al information as well as information about behaviour
association with a specific event, for example, when the is processed in many dedicated neocortical areas. This
PLACE CELLS
rat sniffed a particular odour at a particular place when processing includes complex cognitive rules and con-
Hippocampal principal cells
that fire selectively when an
it was a nonmatch with the odour presented on the cepts, such as those likely to be processed in the pre-
animal is in a particular previous trial. Other cells fired in association with fea- frontal cortex or other association areas58,88 (see Miller in
location in its environment. tures of the task that were common across many trials: this issue). However, the capacity of these components

NATURE REVIEWS | NEUROSCIENCE VOLUME 1 | OCTOBER 2000 | 4 7

© 2000 Macmillan Magazines Ltd
REVIEWS

Trial n Trial n + 1 Trial n + 2

b c

Firing rate (Hz)

6 10
Firing rate (Hz)

8
4
6
2 4
2
0 0
1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9

Position Position

10
6

Firing rate (Hz)

Firing rate (Hz)

8
4 6
4
2
2
0 0
1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9
Odour Odour

Figure 5 | Hippocampal neuronal firing patterns in rats during an odour DNMS task. a | On each trial the rat is presented
with one of nine odours at any of nine randomly selected locations. To obtain a buried reward, the rat must identify whether the
odour is the same as (matches, trial n + 1) or differs from (non-matches, trial n + 2) the odour presented on the previous trial85.
Panels b and c show the average firing rates of two cells associated with all the places and all the odours. b | This cell fires
selectively when the rat samples odour 5, but does not encode where the trial was performed. c | This cell fires selectively when
the rat performs the trial at adjacent positions 2 and 3, but does not encode odour.

is limited in time, such that their representations may these areas, processing within the cortex can take advan-
not outlast conscious processing by more than a few tage of lasting parahippocampal representations, and so
seconds64. The parahippocampal region, which receives come to reflect complex associations between events
convergent inputs from the neocortical association that are processed separately in different cortical regions
areas and sends return projections to all of these areas, or occur sequentially in the same or different areas53,54.
seems to mediate the extended persistence of these cor- However, these individual contributions and their
tical representations. Through interactions between interactions are not conceived as sufficient to link repre-
sentations of events that are separated by long time
periods or to form generalizations. Such an organiza-
tion requires the capacity to rapidly encode a sequence
of events that make up an episodic memory, to retrieve
that memory by re-experiencing one facet of the event,
and to link the continuing experience to stored episodic
representations. It appears that the neuronal elements of
the hippocampus contain the fundamental coding
properties that can support this kind of organization.
However, interactions among the components of the
system are again undoubtedly critical. It is unlikely that
the hippocampus has the storage capacity to contain all
episodic memories. Indeed, sparing of remote memo-
Figure 6 | Place cell firing patterns associated with ries obtained before hippocampal damage indicate that
performance in a spatial memory task on a T-maze87. the hippocampus is not the final storage site2. Therefore,
On left-turn trials, individual hippocampal cells fired as the it seems likely that the hippocampal neurons are
animal passed through each of a series of locations (red involved in mediating the re-establishment of detailed
circled areas and arrows) running up the stem of the T-maze cortical representations, rather than storing the details
and turning onto the left choice arm where it received a reward
themselves. Furthermore, one can imagine that repeti-
(black well). On right-turn trials, a different set of cells fired as
the animal passed through the same set of locations on the
tive interactions between the cortex and hippocampus
stem as well as when it turned onto the right choice arm (with the parahippocampal region as intermediary)
(green circles and arrows). These findings indicate that the serve to co-activate widespread cortical areas so that
hippocampus represents each type of trial separately. they eventually develop linkages between detailed mem-

48 | OCTOBER 2000 | VOLUME 1 [Link]/reviews/neuroscience

© 2000 Macmillan Magazines Ltd
 REVIEWS

ories without hippocampal mediation. In this way, the must elaborate and extend these observations to show
networking provided by the hippocampus may also how rapid episodic coding in the hippocampus might
underlie its temporary role in the consolidation of corti- mediate the protracted process of creating permanent
cal memories72,89,90. links between representations within the cortex. This
question may be pursued by relating the development
Future directions of cortical representations to the development of
The model presented above combines many of the episodic representations in the hippocampus, or by
recent findings about hippocampal function in humans determining whether the development of such cortical
and animals. However, future studies are required to test representations depends on intact hippocampal func-
and elaborate the model. In particular, we have only tion. A recent study on monkeys showed that the devel-
begun to conceptualize how information acquired in opment of representations of visual paired associates in
single learning episodes is encoded and preserved for the inferotemporal cortex depends on an intact para-
long periods within this system91,92. The problem of cre- hippocampal region94. This approach can be extended
ating behavioural protocols for testing episodic memory to an examination of the role of the hippocampus in
in animals is formidable. However, there is evidence both the acquistion and consolidation of cortical net-
that, for example, birds can remember a particular food work representations.
cached at a particular time and in a particular place dur-
ing a single episode93. Other recent studies have offered Links
insights into hippocampal representation of single FURTHER INFORMATION Howard Eichenbaum lab page |
learning experiences in rodents and, in particular, about John O’Keefe lab page | Richard Morris lab page
the role of NMDA-receptor-dependent plasticity in ENCYCLOPEDIA OF LIFE SCIENCES Learning and
episodic-like memory29,65. memory | Amnesia | Neural activity and the development
These findings are consistent with the proposal that of brain circuits | Neural networks and behaviour | Long-
the hippocampus is critical for rapid encoding of events term potentiation | Protein phosphorylation and long-
that compose episodic representations. Future studies term synaptic plasticity

1. Bartlett, F. C. Remembering (Cambridge Univ. Press, 14. Woodruff-Pak, D. S., Papka, M. & Ivry, R. B. Cerebellar 29. Steele, R. J. & Morris, R. G. M. Delay dependent
Cambridge,1932). involvement in eyeblink classical conditioning in humans. impairment in matching-to-place task with chronic and
2. Scoville, W. B. & Milner, B. Loss of recent memory after Neuropsychology 10, 443–458 (1996). intrahippocampal infusion of the NMDA-antagonist D-AP5.
bilateral hippocampal lesions. J. Neurol. Neurosurg. 15. Gabrieli J. in Models of information processing in the basal Hippocampus 9, 118–136 (1999).
Psychiat. 20, 11–21 (1957) ganglia (eds Houk, J. C., Davis, J. L. & Beiser, D. G.) This study suggests a specific role for NMDA-
3. Corkin, S. Lasting consequences of bilateral medial 227–294 ( MIT, Cambridge, Massachusetts, 1995). receptor dependent plasticity as critical to the role
temporal lobectomy: Clinical course and experimental 16. Hallet, M., Pascual-leone, A. & Topka, H. in The of the hippocampus in memory for single episodes.
findings in H.M. Sem. Neurol. 4, 249–259 (1984). Acquisition of Motor Behavior in Vertebrates (eds Bloedel, 30. Cho, Y. H., Beracochea, D. & Jaffard, R. Extended
4. Squire, L. R., Knowlton, B. & Musen, G. The structure and J. R., Ebner, T. J. & Wise, S. P.) 289–302 (MIT, Cambridge, temporal gradient for the retrograde and anterograde
organization of memory. Ann. Rev. Psychol. 44, 453–495 Massachusetts,1996). amnesia produced by ibotenate entorhinal cortex lesions
(1993) 17. Salmon, D. P. & Butters, N. Neurobiology of skill and habit in mice. J. Neurosci. 13, 1759–1766 (1993).
5. Nadel, L. & Moscovitch, M. Memory consolidation, learning. Curr. Opin Neurobiol. 5, 184–190 (1995). 31. Kim, J. J. & Fanselow, M. S. Modality-specific retrograde
retrograde amnesia and the hippocampal complex. Curr. 18. Bechera, A. et al. Double dissociation of conditioning and amnesia of fear. Science 256, 675–677 (1992).
Opin Neurobiol. 7, 217–227 (1997). declarative knowledge relative to the amygdala and 32. Kim, J. J., Clark, R. E. & Thompson, R. F.
6. Teng, E. & Squire, L. R. Memory for places learned long hippocmapus in humans. Science 269, 1115–1118 Hippocamectomy impairs the memory of recently, but not
ago is intact after hippocampal damage. Nature 400, (1995). remotely, aquired trace eyeblink conditioned responses.
675–677 (1999). A set of case studies show the separation of Behav. Neurosci. 109, 195–203 (1995).
An elegant case study of an amnesic patient who memory systems in humans, one involving mediation 33. Winocur, G. Anterograde and retrograde amnesia in rats
has not been able to learn about the geography of of fear conditioning by the amygdala and another with dorsal hippocampal or dorsomedial thalamic lesions.
the city in which he now lives, but has normal recall involving mediation of declarative memory by the Behav. Brain Res. 38, 145–154 (1990).
and navigational capacities about the city in which hippocampus. 34. Zola-Morgan, S. & Squire, L. R. The primate hippocampal
he grew up long before he became amnesic. 19. Cahill, L. F., Babinksy, R., Markowitsch, H. J. & McGaugh, formation: evidence for a time-limited role in memory
7. Reed, J. M. & Squire, L. R. Retrograde amnesia for facts J. L. The amygdala and emotional memory. Nature 377, storage. Science 250, 288–290 (1990).
and events: Recent findings from four new cases. 6547 (1995). 35. Milner, B., Squire, L. R. & Kandel, E. R. Cognitive
J. Neurosci. 18, 3943–3954 (1998). 20. Vallar, G. & Shallice, T. Neuropsychological Impairments of neuroscience and the study of memory. Neuron 20,
8. Warrington, E. K. Studies of retrograde memory: A long Short-Term Memory (Cambridge Univ. Press, Cambridge, 445–468 (1999).
term view. Proc. Natl Acad. Sci. USA 93, 13523–13526 1990). A recent review of neurobiological studies of
(1996). 21. Tulving, E. & Schacter, D. L. Priming and human memory memory, outlining progress in understanding both
9. Eichenbaum, H. & Cohen, N. J. From Conditioning to systems. Science 247, 301–306 (1990). the molecular and neuropsychological aspects of
Conscious Recollection: Memory Systems of the Brain 22. Squire, L. Memory and the hippocampus: A synthesis hippocampal function since the discovery of the
(Oxford Univ. Press, in the press). from findings with rats, monkeys, and humans. Psychol. patient H.M.
10. Cohen, N. J. & Squire, L. R. Preserved learning and Rev. 99, 195–231 (1992). 36. Eichenbaum, H. Declarative memory: insights from
retention of pattern-analyzing skill in amnesia: dissociation 23. Eichenbaum, H., Alvarez, P. & Ramus, S. in Handbook of cognitive neurobiology. Annu. Rev. Psychol. 48, 547–572
of knowing how and knowing that. Science 210, 207–210 Neuropsychology 2nd edn Vol. 4 (ed. Cermak, L.) (Elsevier (1997).
(1980). Sciences, in the press). 37. White, N. M. Mnemonic functions of the basal ganglia.
11. Schacter, D. L. & Tulving, E. in Memory Systems (eds 24. Gaffan, D. Recognition impaired and association intact in Curr. Opin Neurobiol. 7, 164–172 (1997).
Schacter, D. L. & Tulving, E.) 1–38 (MIT, Cambridge, the memory of monkeys after transection of the fornix. 38. LeDoux, J. E. Brain mechanisms of emotion and
Massachusetts,1994). J. Comp. Physiol. Psychol. 86, 1100–1109 (1974). emotional learning. Curr. Opin Neurobiol. 2, 191–197
12. Chun, M. M & Phelps, E. A. Memory deficits for implicit 25. Mishkin, M. Memory in monkeys severely impaired by (1992).
contextual information in amnesic subjects with combined but not separate removal of the amygdala and 39. Davis, M. The role of the amygdala in emotional learning.
hippocampal damage. Nature Neurosci. 2, 844–847 hippocampus. Nature 273, 297–298 (1978). Int. Rev. Neurobiol. 36, 225–266 (1994).
(1999). 26. Zola, S. M. et al. Impaired recognition memory in monkeys 40. McDonald, R. J. & White, N. M. A triple dissociation of
13. Knowlton, B. J., Mangels, J. A. & Squire, L. R. A after damage limited to the hippocampal region. J. memory systems: Hippocampus, amygdala, and dorsal
neostriatal habit learning system in humans. Science 273, Neurosci. 20, 451–463 (2000). striatum. Behav. Neurosci. 107, 3–22 (1993).
1399–1401 (1996). 27. Kesner, R. P. & Novak, J. M. Serial position curve in rats: 41. McGaugh, J. L., Cahill, L. & Roozendaal, B. Involvement of
This study elegantly dissociates the role of the Role of the dorsal hippocampus. Science 218, 173–175 the amygdala in memory storage: Interactions with other
neostriatal system in mediating the acquisition of (1982). brain systems. Proc. Natl Acad. Sci. USA 93,
habitual responses to complex stimuli and that of 28. Aggleton, J. P. Hunt, P. R. & Rawlins, J. N. P. The effects of 13508–13514 (1996).
the hippocampal system in mediating declarative hippocampal lesions upon spatial and non-spatial tests of 42. Burwell, R. D., Witter, M. P. & Amaral, D. G. Perirhinal and
memory. working memory. Behav. Brain Res. 19, 133–146 (1986). postrhinal cortices in the rat: A review of the

NATURE REVIEWS | NEUROSCIENCE VOLUME 1 | OCTOBER 2000 | 4 9

© 2000 Macmillan Magazines Ltd
REVIEWS

neuroanatomical literature and comparison with findings 64. Miller, E. K., Li, L. & Desimone, R. A neural mechanism for 87. Wood, E. R., Dudchenko, P. A., Robitsek, J. R. &
from the monkey brain. Hippocampus 5, 390–408 (1995). working and recognition memory in inferior temporal Eichenbaum, H. Hippocampal neurons encode
This paper reviews the anatomy of the hippocampal cortex. Science 254, 1377–1379 (1991). information about different types of memory episodes
system in rodents and primates, showing striking 65. Rampon, C. et al. Enrichment induces structural changes occuring in the same location. Neuron (in the press).
similarities in the basic flow of information, as well and recovery from non-spatial memory deficits in CA1 88. Miller, E. K. The prefrontal cortex: Complex properties for
as species differences in the cortical inputs to the NMDAR1-knockout mice. Nature Neurosci. 3, 238–244 complex behavior. Neuron 22, 15–17 (1999).
hippocampal region. (2000). A review of recent successes in understanding the
43. Suzuki, W. A. Neuroanatomy of the monkey entorhinal, 66. Gaffan, D. & Harrison, S. Place memory and scene nature of information coding in the prefrontal cortex,
perirhinal, and parahippocampal cortices: Organization of memory: Effects of fornix transection in the monkey. showing an integrated role in memory and cognition.
cortical inputs and interconnections with amygdala and Exp. Brain Res. 74, 202–212 (1989). 89. McClelland, J. L., McNaughton, B. L. & O’Reilly, R. C. Why
striatum. Semin. Neurosci. 8, 3–12 (1996). 67. Gaffan, D. Scene-specific memory for objects: A model of there are complimentary learning systems in the
44. Mishkin, M. A memory system in the monkey. Phil. Trans episodic memory impairment in monkeys with fornix hippocampus and neocortex: Insights from the successes
R. Soc. Lond. B 298, 85–95 (1982). transection. J. Cogn. Neurosci. 6, 305–320 (1994). and failures of connectionist models of learning and
45. Otto, T. & Eichenbaum, H. Complementary roles of orbital 68. Wan, H., Aggleton, J. P. & Malcolm, W. B. Different memory. Psychol. Rev. 102, 419–457 (1995).
prefrontal cortex and the perirhinal-entorhinal cortices in contributions of the hippocampus and perirhinal cortex to This paper reviews efforts to create computational
an odor-guided delayed non-matching to sample task. recognition memory. J. Neurosci. 19, 1142–1148. (1999). models of hippocampal function in memory, and
Behav. Neurosci. 106, 763–776 (1992). 69. Casaday, H. J. & Rawlins, J. N. P. Fornix-fimbria section outlines a potential role for the hippocampal
46. Mumby, D. G., Wood, E. R. & Pinel, J. P. Object and working memory deficits in rats: Stimulus complexity network in organizing memories stored in the
recognition memory is only mildly impaired in rats with and stimulus size. Behav. Neurosci. 109, 594–606 (1995). cerebral cortex.
lesions of the hipocampus and amygdala. Psychobiology 70. Vargha-Khadem, F. et al. Differential Effects of Early 90. Squire, L. R. & Alvarez, P. Retrograde amnesia and
20, 18–27 (1992). Hippocampal Pathology on Episodic and Semantic memory consolidation: a neurobiological perspective. Curr.
47. Rothblat, L. A. & Kromer, L. F. Object recognition memory Memory. Science 277, 376–380 (1997). Opin Neurobiol. 5, 169–177 (1995).
in the rat: The role of the hippocampus. Behav. Brain Res. 71. Squire, L. R. & Zola, S. M. Episodic memory, semantic 91. Lisman, J. E. Relating hippocampal circuitry to function:
42, 25–32 (1991). memory and amnesia. Hippocampus 8, 205–211 (1998). Recall of memory sequences by recperocal dentate-CA3
48. Meunier, M., Bachevalier, J., Mishkin, M. & Murray, E. A. 72. Eichenbaum, H., Dudchenko, P., Wood, E., Shapiro, M. & interactions. Neuron 22, 233–242 (1999).
Effects on visual recognition of combined and separate Tanila, H. The hippocampus, memory, and place cells: Is it This theoretical review suggests a mechanism by
ablations of the entorhinal and perirhinal cortex in rhesus spatial memory or memory space? Neuron 23, 1–20 which hippocampal circuitry, plasticity and neuronal
monkeys. J. Neurosci. 13, 5418–5432 (1993). (1999). firing properties can be combined to understand
49. Zola-Morgan, S., Squire, L. R., Amaral, D. G. & Suzuki, W. A recent review of the phenomenon of place cells how the hippocampus mediates its critical role in
Lesions of perirhinal and parahippocampal cortex that that interprets the findings on location-specific episodic memory.
spare the amygdala and the hippocampal formation activity of hippocampal neurons in rodents within the 92. Wallenstein, G. V., Eichenbaum, H. & Hasselmo, M. E. The
produce severe memory impairment. J. Neurosci. 9, broader perspective of memory processing by the hippocampus as an associator of discontiguous events.
4355–4370 (1989). hippocampus in animals and humans. Trends Neurosci. 21, 317–323 (1998).
50. Murray, E. A. & Mishkin, M. Object recognition and 73. Morris, R. G. M., Garrud, P., Rawlins, J. P. & O’Keefe, J. 93. Clayton, N. S. & Dickinson, A. Episodic-like memory during
location memory in monkeys with excitotoxic lesions of the Place navigation impaired in rats with hippocampal cache recovery by scrub jays. Nature 395, 272–274 (1998).
amygdala and hippocampus. J. Neurosci. 18, 6568–6582 lesions. Nature 297, 681–683 (1982). This study offers a way to study episodic-like
(1998). 74. Eichenbaum, H., Stewart, C. & Morris, R. G. M. memory in non-human species, suggesting that
51. Gaffan, D. Dissociated effects of perirhinal cortex ablation, Hippocampal representation in spatial learning. episodic memory can be defined across species as a
fornix transection and amygdalectomy: Evidence for J. Neurosci. 10, 331–339 (1990). representation of the ‘what’, ‘where’ and ‘when’ of a
multiple memory systems in the primate temporal lobe. 75. Galef, B. G., Mason, J. R., Preti, G. & Bean, N. J. Carbon single event.
Exp. Brain Res. 99, 411–422 (1994). disulfide: A semiochemical mediating socially-induced diet 94. Higuchi, S. & Miyashita, Y. Formation of mnemonic neural
52. Buffalo, E. A., Ramus, S. J., Clark, R. E., Teng, E., Squire, choice in rats. Physiol. Behav. 42, 119–124 (1988). responses to visual paired associates in inferotemporal
L. R. & Zola, S. M. Dissociation between the effects of 76. Bunsey, M. & Eichenbaum, H. Selective damage to the cortex is impaired by perirhinal and entorhinal lesions.
damage to the perirhinal cortex and area TE. Learning hippocampal region blocks long term retention of a natural Proc. Natl Acad. Sci. USA 93, 739–743 (1996).
Memory 6, 572–599 (1999). and nonspatial stimulus-stimulus association. 95. Mishkin, M. & Delacour, J. An analysis of short-term visual
53. Eichenbaum, H. & Bunsey, M. On the binding of Hippocampus 5, 546–556 (1995). memory in the monkey. J. Exp. Psychol. Anim. Behav.
associations in memory: Clues from studies on the role of 77. Bunsey, M. & Eichenbaum, H. Conservation of Process 1, 326–334 (1975).
the hippocampal region in paired-associate learning. Curr. hippocampal memory function in rats and humans. Nature 96. O’Keefe, J. & Dostrovsky, J. The hippocampus as a spatial
Direct. Psychol. Sci. 4, 19–23 (1995). 379, 255–257 (1996). map. Preliminary evidence from unit activity in the freely-
54. Murray, E. A. & Bussey, T. J. Perceptual-mnemonic Insights into the fundamental role of the moving rat. Brain Res. 34, 171–175 (1971).
functions of the perirhinal cortex. Trends Cogn. Sci. hippocampus in memory in animals, showing a 97. Muller, R. U., Kubie, J. L. & Ranck, J. B. Jr Spatial firing
3, 142–151 (1999). common role in organizing and flexibly expressing patterns of hippocampal complex spike cells in a fixed
A review of our understanding of the role of one part memories about distinct learning experiences. environment. J. Neurosci. 7, 1935–1950 (1987).
of the parahippocampal region in memory, as an 78. Dusek, J. A. & Eichenbaum, H. The hippocampus and 98. Muller, R. U. A quarter of a century of place cells. Neuron
intermediary between cortical perceptual memory for orderly stimulus relations. Proc. Natl Acad. 17, 813–822 (1996).
processing and hippocampal memory processing. Sci. USA 94, 7109–7114 (1997). 99. O’Keefe, J. A. & Nadel, L. The Hippocampus as a
55. Young, B. J., Otto, T., Fox, G. D. & Eichenbaum, H. 79. Schacter, D. L. in Memory Systems of the Brain (eds N. M. Cognitive Map (Oxford Univ. Press, Oxford 1978).
Memory Representation within the Parahippocampal Weinberger, J. L. McGaugh, G. Lynch) 351–380. (Guilford, 100. Markus, E. J. et al. Interactions between location and task
Region. J. Neurosci. 17, 5183–5195 (1997). 1985). affect the spatial and directional firing of hippocampal
56. Brown, M. W. & Xiang, J.-Z. Recognition memory: 80. Gothard, K. M., Skaggs, W. E. & McNaughton, B. L. neurons. J. Neurosci. 15, 7079–7094 (1995).
Neuronal substrates of the judgement of prior occurrence. Dynamics of mismatch correction in the hippocampal 101. Fried, I., MacDonald, K. A. & Wilson, C. L. Single neuron
Prog. Neurobiol. 55, 149–189 (1998). ensemble code for space: interaction between path activity in human hippocampus and amygdala during
57. Desimone, R., Miller, E. K., Chelazzi, L. & Lueschow, A. in integration and environmental cues. J. Neurosci. 16, recognition of faces and objects. Neuron 18, 753–765
The Cognitive Neurosciences (ed. Gazzaniga, M. S.) 8027–40 (1996). (1997).
475–486 (MIT, Cambridge, Massachusetts, 1995). 81. McNaughton, B. L., Barnes, C. A. & O’Keefe, J. The 102. Burwell, R. D. & Amaral, D. G. The cortical afferents of the
58. Fuster, J. M. Memory in the cerebral cortex (MIT, contributions of position, direction, and velocity to single perirhinal, postrhinal, and entorhinal corices of the rat.
Cambridge, Massachusetts 1995). unit activity in the hippocampus of freely-moving rats. J. Comp. Neurol. 398, 179–205 (1998).
59. Suzuki, W. & Eichenbaum, H. The neurophysiology of Exp. Brain Res. 52, 41–49 (1983). 103. Suzuki, W. A. & Amaral, D. G. Perirhinal and
memory. Ann. New York Acad. Sci. 911, 175–191 (2000). 82. Muller, R. U., Bostock, E., Taube, J. S. & Kubie, J. L. On parahippocampal cortices of the macaque monkey:
60. Ramus, S. J. & Eichenbaum, H. Neural correlates of the directional firing properties of hippocampal place cells. Cortical afferents. J. Comp. Neurol. 350, 497–533 (1994).
olfactory recognition memory in the rodent orbitofrontal J. Neurosci. 14, 7235–7251 (1994). 104. Amaral, D. G. & Witter, M. P. The three-dimensional
cortex. J. Neurosci. (in the press). 83. Wiener, S. I., Paul, C. A. & Eichenbaum, H. Spatial and organization of the hippocampal formation: A review of
61. Miller, E. K., Erickson, C. A. & Desimone, R. Neural behavioral correlates of hippocampal neuronal activity. anatomical data. J. Neurosci. 31, 571–591(1989).
mechanisms of visual working memory in prefrontal cortex J. Neurosci. 9, 2737–2763 (1989). 105. Bliss, T. V. P. & Collingridge, G. L. A synaptic model of
of the macaque. J. Neurosci. 16, 5154–5167 (1996). 84. Berger, T. W., Alger, B. & Thompson, R. F. Neuronal memory: Long-term potentiation in the hippocampus.
62. Miyashita, Y. & Chang, H. S. Neuronal correlate of pictorial substrate of classical conditioning in the hippocampus. Nature 361, 31–39 (1993).
short-term memory in the primate temporal cortex. Nature Science 192, 483–485 (1976). 106. Aggleton, J. P. & Brown, M. W. Episodic memory,
331, 68–70 (1988). 85. Wood, E. R., Dudchenko, P. A. & Eichenbaum, H. The amnesia, and the hippocampal-anterior thalamic axis.
63. Brown, M. W., Wilson, F. A. W. & Riches, I. P. Neuronal global record of memory in hippocampal neuronal activity. Behav. Brain Sci. 22, 425–489 (1999).
evidence that inferomedial temporal cortex is more Nature 397, 613–616.(1999).
important than hippocampus in certain processes 86. Hampson, R. E., Simeral, J. D. & Deadwyler, A.
underlying recognition memory. Brain Res. 409, 158–162 Distribution of spatial and nonspatial information in dorsal Acknowledgements
(1987). hippocampus. Nature 402, 610–614 (1999). This work was funded by grants from the NIMH and the NIA.

50 | OCTOBER 2000 | VOLUME 1 [Link]/reviews/neuroscience

© 2000 Macmillan Magazines Ltd