NIH Public Access

Author Manuscript
Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.
Published in final edited form as:
NIH-PA Author Manuscript

Annu Rev Neurosci. 2011 ; 34: 259–288. doi:10.1146/annurev-neuro-061010-113720.

The Cognitive Neuroscience of Human Memory Since H.M

Larry R. Squire1,2,3,4 and John T. Wixted4
Larry R. Squire: lsquire@ucsd.edu; John T. Wixted: jwixted@ucsd.edu
1Veterans Affairs Healthcare System, San Diego, California 92161
2Department of Psychiatry, University of California, San Diego, La Jolla, California 92093
3Department of Neurosciences, University of California, San Diego, La Jolla, California 92093
4Department of Psychology, University of California, San Diego, La Jolla, California 92093

Abstract
Work with patient H.M., beginning in the 1950s, established key principles about the organization
of memory that inspired decades of experimental work. Since H.M., the study of human memory
NIH-PA Author Manuscript

and its disorders has continued to yield new insights and to improve understanding of the structure
and organization of memory. Here we review this work with emphasis on the neuroanatomy of
medial temporal lobe and diencephalic structures important for memory, multiple memory
systems, visual perception, immediate memory, memory consolidation, the locus of long-term
memory storage, the concepts of recollection and familiarity, and the question of how different
medial temporal lobe structures may contribute differently to memory functions.

Keywords
medial temporal lobe; hippocampus; neocortex; anterograde amnesia; retrograde amnesia

INTRODUCTION
In the earliest systematic writings about human memory, it was already appreciated that the
study of memory impairment can provide valuable insights into the structure and
organization of normal function (Ribot 1881, Winslow 1861). This tradition of research has
continued to prove fruitful and has yielded a broad range of fundamental information about
NIH-PA Author Manuscript

the structure and organization of memory. What is memory? Is it one thing or many? What
are the concepts and categories that guide our current understanding of how memory works
and that underlie the classification of its disorders? It is sometimes not appreciated that the
concepts and categories used in current discussions of memory are not fixed and were not
easily established. Even the question of which cognitive operations reflect memory and
which depend on other faculties has a long history of empirical work and discussion.

One needs only to sample nineteenth-century writings to recognize how differently memory
was viewed then and now. For example, in his classic treatment of memory disorders, Ribot
(1881) considered amnesias due to neurological injury together with amnesias due to
psychological trauma. And he viewed aphasia and agnosia as disorders of memory, wherein

Copyright © 2011 by Annual Reviews. All rights reserved

DISCLOSURE STATEMENT
The authors are not aware of any affiliations, memberships, funding, or financial holdings that might be perceived as affecting the
objectivity of this review.
 Squire and Wixted Page 2

(in aphasia, for example) patients have lost their memory for words or memory for the
movements needed to produce words. Today, aphasia is considered a deficit of language,
and agnosia a deficit of visual perception. Memory is affected but only as part of a more
NIH-PA Author Manuscript

fundamental defect in a specific kind of information processing.

The notion that the study of brain injury can elucidate the organization of memory was itself
a matter for empirical inquiry. If brain regions were highly interconnected, and the brain’s
functions distributed and integrated one with another, then damage to any one area would
produce a global impairment, blurred across multiple faculties and affecting all of mental
life. But the fact of the matter is different. The brain is highly specialized and modular, with
different regions dedicated to specific operations. As a result, localized damage can produce
strikingly specific effects, including a selective and circumscribed impairment of memory.

The idea that functions of the nervous system can be localized was already well accepted by
the end of the nineteenth century. This localizationist view had its roots in the writings of
Gall (1825) and was supported by the experimental work of Broca (1861), Ferrier (1876),
Fritsch & Hitzig (1870), and others (see Finger 1994). Yet, these ideas centered mainly
around sensory functions, motor control, and language and did not usefully address the topic
of memory. Then, in the early twentieth century, an influential program of experimental
work in rodents investigated directly the localization of memory with the conclusion that
memory is distributed throughout the cortex and that the contribution to memory is
NIH-PA Author Manuscript

equivalent across regions (Lashley 1929). This idea was strongly challenged (Hebb 1949,
Hunter 1930) by the alternative, and more modern, interpretation that memory storage is
indeed distributed but that different areas store different features of the whole. Still, as the
midpoint of the twentieth century approached, memory functions, while distributed, were
thought to be well integrated with perceptual and intellectual functions, and no region of the
brain was believed to be disproportionately dedicated to memory. All that was about to
change.

In 1957, Brenda Milner reported the profound effect on memory of bilateral medial temporal
lobe resection, carried out to relieve epilepsy in a patient who became known as H.M.
(1926–2008) (Scoville & Milner 1957, Squire 2009) (Figure 1). Remarkably, H.M.
exhibited profound forgetfulness but in the absence of any general intellectual loss or
perceptual disorders. He could not form new memories (anterograde amnesia) and also
could not access some memories acquired before his surgery (retrograde amnesia). His
impairment extended to both verbal and non-verbal material, and it involved information
acquired through all sensory modalities. These findings established the fundamental
principle that memory is a distinct cerebral function, separable from other perceptual and
cognitive abilities, and also identified the medial aspect of the temporal lobe as important for
NIH-PA Author Manuscript

memory. The early descriptions of H.M. can be said to have inaugurated the modern era of
memory research, and the findings from H.M. enormously influenced the direction of
subsequent work.

ANATOMY OF MEMORY
The work with H.M. is sometimes cited incorrectly as evidence of the importance of the
hippocampus for memory, but this particular point could not of course be established by a
large lesion that included not only the hippocampus but also the amygdala together with the
adjacent parahippocampal gyrus. Which structures within H.M.’s lesion are important for
memory became understood only gradually during the 1980s following the successful
development of an animal model of human amnesia in the non-human primate (Mishkin
1978). Cumulative studies in the monkey (Murray 1992, Squire & Zola-Morgan 1991, Zola-
Morgan et al. 1994) considerably clarified this issue. The important structures proved to be

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 3

the hippocampus and the adjacent entorhinal, perirhinal, and parahippocampal cortices,
which make up much of the parahippocampal gyrus (Figure 2).
NIH-PA Author Manuscript

One particularly instructive case of human memory impairment became available during this
same time period (Zola-Morgan et al. 1986). R.B. developed a moderately severe, enduring
impairment following an ischemic episode in 1978. During the five years until his death, his
memory deficit was well documented with formal tests. Detailed histological examination of
his brain revealed a circumscribed bilateral lesion involving the entire CA1 field of the
hippocampus. Note that a lesion confined to the CA1 field must substantially disrupt
hippocampal function because the CA1 field is a bottleneck in the unidirectional chain of
processing that begins at the dentate gyrus and ends in the subiculum and entorhinal cortex.
R.B. was the first case of memory impairment following a lesion limited to the hippocampus
that was supported by extensive neuropsychological testing as well as neuropathological
analysis.

The findings from R.B., considered together with the much more severe impairment in
H.M., made two useful points. First, damage to the hippocampus itself is sufficient to
produce a clinically significant and readily detectable memory impairment. Second,
additional damage to the adjacent cortical regions along the parahippocampal gyrus (as in
H.M.) greatly exacerbates the memory impairment. These same conclusions about the
neuroanatomy of modest and severe memory impairment were also established in the
NIH-PA Author Manuscript

monkey (Zola-Morgan et al. 1994).

Another case was subsequently described (patient G.D.) with a histologically confirmed
bilateral lesion confined to the CA1 field and with a memory impairment very similar to
R.B. (Rempel-Clower et al. 1996). Two other patients were also of interest. L.M. and W.H.
had somewhat more severe memory impairment than did R.B. and G.D., but the impairment
was still moderate in comparison to H.M. (Rempel-Clower et al. 1996). Histological
examination revealed extensive bilateral lesions of the hippocampal region, involving all the
CA fields and the dentate gyrus. There was also some cell loss in entorhinal cortex and, for
W.H., cell loss in the subiculum, as well. The more severe memory impairment in these two
cases, in comparison to R.B. and G.D., could be due to the additional damage within the
hippocampus or to the cell loss in entorhinal cortex.

There are only a small number of cases where detailed neuropsychological testing and
thorough neurohistological analysis have combined to demonstrate memory impairment
after limited hippocampal damage or larger medial temporal lobe lesions (see also Victor &
Agamanolis 1990). Yet, neuroanatomical information is essential because it lays the
groundwork for classifying memory disorders, for understanding qualitative and quantitative
NIH-PA Author Manuscript

differences between patients, and for addressing questions about how specific structures
may contribute differently to memory functions. Nonetheless, in the absence of histological
data, valuable information can be obtained from structural imaging. Methods for high-
resolution imaging of hippocampal damage were developed some time ago (Press et al.
1989), and quantitative data can now be obtained that provide reliable estimates of tissue
volume (Gold & Squire 2005). These estimates are based on guidelines defined
histologically and use landmarks in the medial temporal lobe that are visible on MRI
(Insausti et al. 1998a,b).

An interesting observation has emerged from calculations of hippocampal volume in

memory-impaired patients, usually patients who have sustained an anoxic episode. Across a
number of reports, hippocampal volume (or area in the coronal plane) is typically reduced
by ~40% [41%, n = 10 (Isaacs et al. 2003); 44%, n = 5 (Shrager et al. 2008); 43%, n = 4
(Squire et al. 1990); 45%, n = 1 (Cipolotti et al. 2001); 46%, n = 1 (Mayes et al. 2002)].

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 4

Neurohistological data from two of these patients (L.M. and W.H.) suggest an explanation
for this striking consistency. As described above, these two patients had extensive cell loss
in the hippocampus as well as in the dentate gyrus. Accordingly, a reduction in hippocampal
NIH-PA Author Manuscript

volume of 40%, as estimated by MRI, may indicate a nearly complete loss of hippocampal
neurons. The tissue collapses, but it does not disappear entirely. A volume loss in the
hippocampus of ~40% may represent a maximum value for some etiologies of memory
impairment.

While medial temporal lobe structures have received the most attention in studies of
memory and memory impairment, it is notable that damage to the diencephalic midline also
impairs memory. The deficit has essentially the same features as in medial temporal lobe
amnesia. The best-known cause of diencephalic amnesia is alcoholic Korsakoff’s syndrome.
Here, damage to the medial dorsal thalamic nucleus (alone or perhaps in combination with
damage to the mammillary nuclei) has been associated with memory impairment (Victor et
al. 1989). Another survey of Korsakoff’s syndrome documented damage to these two
structures and, in addition, identified a role for the anterior thalamic nuclei (Harding et al.
2000). Six cases that were studied both neuropsychologically and neurohistologically (Gold
& Squire 2006, Mair et al. 1979, Mayes et al. 1988) consistently identified damage in the
medial thalamus (as well as in the mammillary nuclei for the five cases with Korsakoff’s
syndrome). Two regions of thalamus were implicated by these cases and by two
neuroimaging studies of diencephalic amnesia (Squire et al. 1989, von Cramon et al. 1985):
NIH-PA Author Manuscript

first, the medial dorsal nucleus and the adjacent internal medullary lamina; and second, the
mammillothalamic tract and its target, the anterior thalamic nuclei. Damage to either of
these regions can cause memory impairment. These diencephalic nuclei and tracts are
anatomically related to the medial temporal lobe. The perirhinal cortex originates
projections to the medial dorsal nucleus that enter through the internal medullary lamina,
and the hippocampal formation projects both to the rostrally adjacent anterior nuclei and to
the mammillary nuclei. These anatomical connections likely explain why patients with
medial temporal or diencephalic lesions exhibit the same core deficit.

PRINCIPLES OF ORGANIZATION SUGGESTED BY H.M.’S FINDINGS

The early descriptions of H.M suggested four principles about how memory is organized in
the brain. First, despite his debilitating and pervasive memory impairment, H.M.
successfully acquired a motor skill. This finding raised the possibility that memory is not a
single thing. Second, because his memory impairment appeared to be well circumscribed,
the structures damaged in memory-impaired patients were thought not to be involved in
intellectual and perceptual functions. Third, H.M. had a considerable capacity for sustained
attention, including the ability to retain information for a period of time after it was first
NIH-PA Author Manuscript

encountered. This finding suggested that medial temporal lobe structures are not needed for
immediate memory or for the rehearsal and maintenance of material in what would now be
termed working memory. Fourth, H.M. appeared to have good access to facts and events
from time periods remote to his surgery. This observation suggested that the medial
temporal lobe cannot be the ultimate storage site for long-term memory. Permanent memory
must be stored elsewhere, presumably in neocortex. In the years since H.M. was described,
each of these ideas has been the topic of extensive experimental work.

During the 1960s and 1970s, when human memory impairment began to be systematically
studied, there was considerable debate about whether medial temporal and diencephalic
structures were concerned more with storage or with retrieval. The findings from H.M. led
to the view that these structures are needed for memory storage, that is, for the establishment
of new representations in long-term memory. If these structures are unable to participate in
forming long-term memory, then representations established in immediate memory are

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 5

presumably lost or perhaps achieve some disorganized state. Consider the case of transient
amnesic episodes (transient global amnesia or the memory impairment associated with
electroconvulsive therapy). Here, the events that occur during the period of anterograde
NIH-PA Author Manuscript

amnesia are not subsequently remembered after recovery from the amnesic condition. New
learning again becomes possible, but events from the amnesic episode do not return to
memory. Thus, if medial temporal lobe or diencephalic structures are not functional at the
time of learning, memory is not established in a usable way and does not become available
at a later time. More direct investigations of this issue using single-cell recording in
monkeys have reached similar conclusions (Higuchi & Miyashita 1996; see Squire 2006).
The idea is that the synaptic changes that would ordinarily represent acquired information in
long-term memory either are lost altogether or fail to develop into a stable, coherent
ensemble.

MULTIPLE MEMORY SYSTEMS

The memory impairment in H.M. and other patients is narrower than once thought in that
not all kinds of learning and memory are affected. The first hint of this idea came when
H.M. was found capable of learning a hand-eye coordination skill (mirror drawing) over a
period of days, despite having no recollection of practicing the task before (Milner 1962).
Although this finding showed that memory was not unitary, for some time it was thought
that motor skill learning was a special case and that all the rest of memory is of one piece
NIH-PA Author Manuscript

and is impaired in amnesia. Subsequently, it was discovered that motor-skill learning is but
one example of a large domain of learning and memory abilities, all of which are intact in
H.M. and other patients. H.M.’s motor skill learning marked the beginning of a body of
experimental work that would eventually establish the biological reality of two major forms
of memory.

An early insight was that perceptual skills and cognitive skills, not just motor skills, are
preserved in amnesia. Specifically, amnesic patients acquired at a normal rate the perceptual
skill of reading mirror-reversed words, despite poor memory for the task itself and for the
words that were read (Cohen & Squire 1980). This finding was the basis for the formulation
of a brain-based distinction between two major forms of memory, which afford either
declarative or procedural knowledge. Declarative knowledge referred to knowledge
available as conscious recollections about facts and events. Procedural knowledge referred
primarily to skill-based information, where what has been learned is embedded in acquired
procedures.

Subsequently, memory-impaired patients were found to exhibit intact priming effects (see
Tulving & Schacter 1990). For example, patients (like healthy volunteers) could name
NIH-PA Author Manuscript

pictures of objects 100 ms faster when the pictures had been presented previously than when
they were presented for the first time and independently of whether patients could recognize
the pictures as familiar (Cave & Squire 1992).

Another important insight was the idea that the neostriatum (not the medial temporal lobe) is
important for the sort of gradual, feedback-guided learning that results in habit memory
(Mishkin et al. 1984). Thus, memory-impaired patients learned at a normal rate when
explicit memorization was not useful (for example, when the outcome of each trial was
determined probabilistically and performance needed to be based on a gut feeling)
(Knowlton et al. 1996). Furthermore, tasks that healthy volunteers could learn rapidly by
memorization (such as the concurrent learning of eight different, two-choice object
discriminations) could also be learned successfully by profoundly amnesic patients, albeit
very gradually (healthy volunteers required fewer than 80 trials; patients required more than
1000 trials). Although memory became robust in the patients after extended training (>90%

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 6

accuracy), it differed from the memory acquired by healthy volunteers in that what was
learned was outside of awareness and was rigidly organized (performance collapsed when
the task format was modified) (Bayley et al. 2005a).
NIH-PA Author Manuscript

Given the wide variety of learning and memory phenomena that could be demonstrated in
patients (for example, priming and habit learning), the perspective eventually shifted to a
framework that accommodated multiple memory systems, not just two kinds of memory.
Indeed, one could ask what the various kinds of memory that were preserved in patients had
in common aside from the fact that they were not declarative. Accordingly, the term non-
declarative was introduced with the idea that declarative memory refers to one kind of
memory system and that nondeclarative memory is an umbrella term referring to several
additional memory systems (Squire & Zola-Morgan 1988). Nondeclarative memory includes
skills and habits, simple forms of conditioning, emotional learning, priming, and perceptual
learning, as well as phylogenetically early forms of behavioral plasticity such as habituation
and sensitization.

Declarative memory is the kind of memory that is referred to when the term memory is used
in everyday language. Declarative memory allows remembered material to be compared and
contrasted. The stored representations are flexible, accessible to awareness, and can guide
performance in a variety of contexts. Declarative memory is representational. It provides a
way of modeling the external world, and it is either true or false. Nondeclarative memory is
NIH-PA Author Manuscript

neither true nor false. It is dispositional and is expressed through performance rather than
recollection. These forms of memory provide for myriad unconscious ways of responding to
the world. In no small part, by virtue of the unconscious status of the nondeclarative forms
of memory, they create some of the mystery of human experience. Here arise the
dispositions, habits, and preferences that are inaccessible to conscious recollection but that
nevertheless are shaped by past events, influence our behavior and mental life, and are an
important part of who we are.

VISUAL PERCEPTION
Formal testing of patient H.M. over the years documented his good performance on
intelligence tests and on other tests of perceptual function and lexical knowledge (Kensinger
et al. 2001, Milner et al. 1968). He could detect the anomalous features of cartoon drawings,
and he performed above the control mean on the Mooney “Closure” task, which requires
participants to find a face in a chaotic black and white pattern with incomplete contour
(Milner et al. 1968). This perspective, that visual perception is intact after large medial
temporal lobe lesions, was eventually challenged, first by work in monkeys (Eacott et al.
1994) and later by studies in humans (Lee et al. 2005a,b). These studies proposed that the
NIH-PA Author Manuscript

perirhinal cortex, one of the structures damaged in H.M., is important for complex visual
perceptual tasks involving stimuli with substantial feature overlap. It was also proposed that
the hippocampus is needed when spatial processing is required, as in visual discriminations
involving scenes.

Although some subsequent studies appeared to provide additional support for this
perspective (Barense et al. 2007, Lee & Rudebeck 2010), attempts to replicate some of the
key early work and to find impairments with new tests were unsuccessful (Shrager et al.
2006). Comprehensive reviews of this topic (Suzuki 2009, 2010) raised three important
issues. First, a consideration of neuroanatomic and neurophysiological data emphasizes that
the perirhinal cortex has unique characteristics that distinguish it from the laterally adjacent,
unimodal visual area TE. The perirhinal cortex is a poly-modal association area with strong
connections to the hippocampus and entorhinal cortex, and it is difficult to view the

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 7

perirhinal cortex as a visual area and as a continuation of the ventral visual pathway (Suzuki
2010).
NIH-PA Author Manuscript

Second, many of the studies designed to test visual perception, particularly studies in
monkeys, involve a significant memory requirement. Thus, impaired associative learning or
impaired long-term memory for the stimulus material could have contributed to many of the
deficits reported after perirhinal lesions in monkeys. Even in studies of humans, impaired
associative learning could result in deficient performance when different test stimuli need to
be judged against the same two comparison stimuli on every trial (Graham et al. 2006).
Indeed, in a new study that explored this issue, patients with hippocampal lesions were
impaired when the same comparison stimuli were used on every trial but were fully intact
when the stimuli were unique to every trial (Kim et al. 2011). Using fixed comparison
stimuli gives an advantage to those who can remember because one can learn what to look
for in the test stimuli to decide which comparison stimulus it most closely resembles.

Third, patients who exhibit impaired performance on tasks of visual perception may have
significant damage to lateral temporal cortex in addition to medial temporal lobe damage.
This idea merits consideration, given that two of the three patients with medial temporal
lobe damage who were impaired were reported to have damage lateral to the medial lobe
(Barense et al. 2007; Lee et al. 2005a,b; Lee & Rudebeck 2010). Also, estimates of damage
in most of the patients who were impaired were based on ratings of single sections through
NIH-PA Author Manuscript

the lateral temporal cortex, not on quantitative measures of the entire region, thus leaving
large amounts of tissue unexamined.

The importance of thorough neuroanatomical measurement in neuropsychological studies of

memory cannot be overstated. Many current disagreements about the facts and ideas
emerging from neuropsychological research on human memory can be traced to concerns
about the locus and extent of lesions. If a deficit is expected but not found, perhaps the
damage is less extensive than believed. If a deficit is not expected but is found, perhaps the
damage is more extensive than has been detected. There is no substitute for thorough,
quantitative descriptions of damage based on magnetic resonance imaging, as well as (where
possible) detailed neurohistological description of the postmortem brain.

The possible role of perirhinal cortex in certain kinds of visual perception remains a topic of
discussion and will benefit from detailed analysis of the lesions in the cases under study. At
the present time, the weight of evidence from experimental lesion studies in monkeys,
neurophysiological studies, and human neuropsychological studies continues to support the
view that medial temporal lobe structures are important for declarative memory and not for
perceptual functions (see also Clark et al. 2011).
NIH-PA Author Manuscript

IMMEDIATE MEMORY AND WORKING MEMORY

The early descriptions of H.M. emphasized how capable he was at focusing his attention and
at retaining information for short periods of time (Milner et al. 1968). For example, he could
retain a three-digit number for 15 minutes by continuous rehearsal, using what would now
be termed working memory (Baddeley 2003). Yet when his attention was diverted, he forgot
the whole event. In one dramatic demonstration, participants heard digit strings of increasing
length (Drachman & Arbit 1966) (Figure 3a). Each string was presented as many times as
needed until it was reported back correctly. Then, a new digit string was presented that was
one digit longer than the previous one. Controls made their first errors with strings of eight
digits and were eventually able to repeat strings as long as 20 digits (with no more than 25
repetitions at any one string length). In contrast, H.M. exhibited a marked discontinuity in
performance as the string length increased. He repeated up to six digits correctly on his first
try (six was his preoperative digit span), but he never succeeded at seven digits, even though

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 8

he was given 25 repetitions of the same string. The interpretation was that at short string
lengths H.M. could rely on his intact immediate memory and that he failed when the
material to be remembered was more than could be held in mind. That is, he failed when the
NIH-PA Author Manuscript

material exceeded his immediate memory capacity.

Time is not the key factor that determines how long information can be retained by patients
like H.M. The relevant factors are immediate memory capacity and how successfully
material can be maintained in working memory through rehearsal. Maintenance of
information is difficult when material is difficult to rehearse (e.g., faces and designs).
Moreover, working memory capacity can be quite limited, and typically only three or four
simple visual objects can be maintained (Cowan 2001, Fukuda et al. 2010). With these
considerations in mind, it is perhaps not surprising that impaired performance after medial
temporal lobe lesions has sometimes been reported at short retention intervals, usually when
the task requires learning complex material or learning the relations between items (e.g.,
object-location associations) (Finke et al. 2008, Hannula et al. 2006, Kan et al. 2007, Olson
et al. 2006). In these cases, the important question is whether working memory capacity has
been exceeded and performance must rely on long-term memory, or whether working
memory sometimes depends on the medial temporal lobe. Methods that are independent of
the particular task that is used are needed to decide this question.

One approach to this issue seems promising in cases where the retention interval is long
NIH-PA Author Manuscript

enough (about 8 seconds) to allow a manipulation to be introduced during the interval

(Shrager et al. 2008). Controls (but not patients) were given either distraction or no
distraction between study and test. Across experiments involving names, faces, or object-
location associations, patient performance was related to how distraction affected controls.
The patients were impaired when distraction had no effect on control performance, and the
patients were intact when distraction disrupted control performance. These results suggested
that the patients were impaired when the task depended minimally on working memory (as
indicated by the ineffectiveness of distraction on control performance), and they performed
well when the task depended substantially on working memory (as indicated by the
disruptive effect of distraction on controls). Thus, for the kinds of material studied here,
including relational information for objects and locations, working memory appears to be
intact after medial temporal lobe damage.

A possible approach in cases where the retention interval is very short (1–3 seconds) is
based on the early study of digit span, described above. Participants saw different numbers
of objects (1 to 7) arranged on a tabletop and then immediately tried to reproduce the array
on an adjacent table ( Jeneson et al. 2010) (Figure 3b). The same study-test sequence was
repeated (up to a maximum of ten times) until participants correctly placed each object
NIH-PA Author Manuscript

within a specified distance of its original location. The finding was that performance was
intact when only a few object locations needed to be remembered. However, just as was
found for digit strings, there was an abrupt discontinuity in performance with larger numbers
of object locations. For example, patient G.P. (who has large medial temporal lobe lesions
similar to H.M.’s lesions) learned 1, 2, or 3 object locations as quickly as did controls,
needing no more than one or two tries at each stage. However, when four object locations
needed to be remembered, he could not succeed even in 10 attempts with the same array.
These findings suggest that the maintenance of relational information (in this case, object-
location associations) can proceed normally, even in patients with large medial temporal
lobe lesions. An impairment is evident only when a capacity limit is reached, at which point
performance must depend, at least in part, on long-term memory.

These observations support the view that patients with medial temporal lobe lesions can
succeed at remembering whatever they have encountered, so long as the material to be

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 9

remembered can be supported by a limited-capacity, short-term memory system (see also

Jeneson et al. 2011). This formulation touches on a large and fundamental issue: whether
there is any ability at all that depends on the hippocampus and related structures, even when
NIH-PA Author Manuscript

a task can be managed within working memory. That is, do these structures perform any
online computations for which the distinction between working memory and long-term
memory is irrelevant?

This is a question of considerable current interest. It runs through discussions of perceptual

functions and discussions of relational memory (as considered in this section and the
preceding section). The issue is especially prominent in discussions of spatial cognition. For
example, the ability to path integrate (i.e., the ability to use self-motion cues to keep track of
a reference location as one moves though space) has been proposed to have a fundamental
dependency on the hippocampus and entorhinal cortex. That is, these structures are proposed
to carry out computations essential for path integration, regardless of the memory load or the
retention interval (Whitlock et al. 2008). Furthermore, the hippocampus is proposed to be
necessary for constructing a spatially correct mental image of either a remembered scene or
an imagined scene (Bird et al. 2010, Bird & Burgess 2008), a task that need not involve
recollection at all.

In the case of path integration, humans can succeed at simple paths in the absence of
hippocampus and entorhinal cortex so long as the task can be managed within 30–40
NIH-PA Author Manuscript

seconds (presumably supported by working memory) (Shrager et al. 2008). In the case of
spatial imagining, patients with severe memory impairment can describe routes around their
childhood neighborhoods, including when main routes are blocked and alternative routes
must be found (Rosenbaum et al. 2000, Teng & Squire 1999). Furthermore, in one study,
patients with hippocampal damage successfully imagined future events and provided a
normal number of spatial referents (Squire et al. 2010; see Hassabis et al. 2007 for a deficit
on a similar task). These demonstrations appear straightforward and would seem to raise
doubts about the idea that the hippocampus performs online computations. Yet there is an
alternate perspective. Specifically, it has been suggested that spatial representations can be
established outside the hippocampus, and in parallel with hippocampal representations, but
using somewhat different computations (Bird & Burgess 2008, Whitlock et al. 2008). By
this account, some spatial tasks that are accomplished successfully after hippocampal
damage are in fact being accomplished using different structures and different computations
than are used by healthy individuals.

The idea is that, despite intact performance in patients, some tasks are still hippocampus-
dependent and could be shown to be so if one could devise tasks that can only be done with
computations unique to the hippocampus. This is an interesting perspective and one that, in
NIH-PA Author Manuscript

principle, could be applied to any example of intact performance in patients. It will be

difficult to resolve issues like these without understanding which strategies are used in any
particular case and without gaining experimental control over them. In addition, tasks that
can be solved by different structures and using different strategies may be associated with
inconsistent deficits after hippocampal lesions. In contrast, there are some tasks that depend
on the medial temporal lobe, where performance deficits are invariably pronounced, and
where performance cannot be made to appear normal by recruiting other brain structures or
by using different strategies. These are tasks that assess the ability to form conscious long-
term memory of facts and events, and the inability to carry out this function appears to be
the central deficit in H.M. and other patients with medial temporal lobe lesions.

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 10

REMOTE MEMORY AND MEMORY CONSOLIDATION

A key insight about the organization of memory came with early observations of H.M.’s
NIH-PA Author Manuscript

capacity to remember information that he acquired before his surgery in 1953. Initially, he
was described as having a loss of memory (retrograde amnesia) covering the three years
immediately preceding surgery and with earlier memories “seemingly normal” (Scoville &
Milner 1957, p. 17). About ten years later, the impression was similar as there did not appear
to have been any change in H.M.’s capacity to recall remote events antedating his operation,
such as incidents from his early school years, a high school attachment, or jobs he had held
in his late teens and early twenties (Milner et al. 1968, p. 216).

The first study of this issue with formal tests asked H.M. to recognize faces of persons who
had become famous in the decades 1920–1970 (Marslen-Wilson & Teuber 1975). As
expected, he performed poorly in the postmorbid period (the 1950s and 1960s) but did as
well as or better than age-matched controls at recognizing faces from the premorbid period
(the 1920s–1940s). This important finding implied that medial temporal lobe structures are
not the ultimate storage sites for acquired memories. Memories that initially require the
integrity of medial temporal lobe structures must be reorganized as time passes after
learning so as to gradually become independent of these structures. The extent of retrograde
amnesia provides an indication of how long this process takes.
NIH-PA Author Manuscript

Retrograde amnesia can be either temporally limited, covering a few years, or prolonged,
depending on the locus and extent of the damage. Patients with damage thought to be
restricted to the hippocampus had retrograde amnesia for past news events that extended
only a few years into the premorbid period (Manns et al. 2003b). By contrast, patients with
large medial temporal lobe lesions (damage to hippocampus plus parahippocampal gyrus)
exhibited extended retrograde amnesia that covered several decades, albeit sparing
memories acquired in early life (patients E.P. and G.P.; Bayley et al. 2006, Bright et al.
2006). The possibility that some amount of more lateral damage (e.g., in the fusiform gyrus)
contributed to the extended retrograde impairment in E.P. and G.P. cannot be excluded.

There has been particular interest in the status of autobiographical memories for unique
events following medial temporal lobe damage, and in recent years methods have been
developed to assess the detail with which such recollections can be reproduced. In the
earliest formal assessments of H.M. (Sagar et al. 1985), he produced well-formed
autobiographical memories from age 16 and younger (his surgery occurred at age 27).
However, the situation seemed to change as H.M. aged. In a later update (Corkin 2002),
H.M. (now 76 years old) was reported to have memories of childhood, but the memories
appeared fact-like and lacked detail. It was stated that he could not reproduce a single event
NIH-PA Author Manuscript

that was specific to time and place. In a formal study reported a few years later (Steinvorth
et al. 2005), he was also impaired in recollecting events from his early life. It was concluded
that autobiographical memories remain dependent on the medial temporal lobe so long as
the memories persist.

This conclusion about H.M. is complicated by the findings from MRI scans obtained in
2002 and 2003 (Salat et al. 2006). These scans documented a number of significant changes
since his first MRI scans from 1992–1993 (Corkin et al. 1997) (Figure 1). Specifically, the
scans showed cortical thinning, subcortical atrophy, large amounts of abnormal white
matter, and subcortical infarcts. All these features were thought to have developed during
the past decade, and they complicate the interpretation of neuropsychological data collected
during and after this period. Considering the earlier reports that he could successfully
retrieve past autobiographical memories (Milner et al. 1968, Scoville & Milner 1957), it is
possible that remote autobiographical memories were in fact intact during the early years

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 11

after surgery but were later compromised by neurological change. It is also possible that the
available memories faded with time because they could not be strengthened through
rehearsal and relearning.
NIH-PA Author Manuscript

Other work has supported the earlier descriptions of H.M. For example, methods similar to
those used to assess H.M. have also been used to evaluate autobiographical memory in other
patients with hippocampal damage or larger medial temporal lobe lesions (Bright et al. 2006,
Kirwan et al. 2008). These patients had intact autobiographical memory from their early
lives. The following example illustrates a well-formed autobiographical memory produced
by E.P. about his early life, one of 18 that he produced. In this case, he was asked for a
specific recollection in response to the cue word “fire.” Like most recollections, his narrative
contains both fact-based and event-specific information. Note the several repetitions in the
narrative, which reflect his severe anterograde amnesia.
Dad had 31/2 acres of property in Castro Valley and the back property would just
grow and would be dry and for some reason, I didn’t do it, but somehow or other
the next thing we knew is that it was starting to burn. I told dad and he called the
Castro Valley fire department. They came up and they got it out real quick.
However it started I don’t know. He had 31/2 acres of property and he just let it
grow. It would be grass or whatever. Who knows how it started, but it started to
burn. Dad called the Castro Valley fire department and they came up and all the
volunteers came in and they got it out in a matter of 10–15 minutes. They stamped
NIH-PA Author Manuscript

it out. They don’t know how it started. I was 16–17, in that bracket. Dad had 31/2
acres of property. It was summer time, 1938. Those sort of things I think you
remember. (Bayley et al. 2003, p. 139)
The same finding of intact early memories was reported in 10 patients with medial temporal
lobe lesions in a study of emotional (and remote) autobiographical memories (Buchanan et
al. 2005), and in two other patients (M.R. and P.D.), using a simpler assessment device
(Eslinger 1998). In another study of four patients with medial temporal lobe damage and
variable damage to anterior and posterior temporal neocortex (Rosenbaum et al. 2008), one
patient (S.J.) was reported to have extended retrograde amnesia for autobiographical
memory. The other three patients were less impaired, performing poorly in time periods
closer to the onset of their amnesia. The impairment in S.J. was attributed to hippocampal
damage. Alternatively, it is difficult to rule out a substantial contribution from the damage
that was identified in neocortex.

It is noteworthy that, not infrequently, patients have been described as having extensive and
ungraded retrograde amnesia (i.e., unrelated to how long ago the memory was formed) (for
examples, see Bright et al. 2006, Cipolotti et al. 2001, Noulhiane et al. 2007, Rosenbaum et
NIH-PA Author Manuscript

al. 2008, Sanders & Warrington 1971). This pattern of impairment has sometimes been
taken to mean that the hippocampus (or related structures) is required as long as a memory
persists. Yet, in many cases testing did not cover early adulthood and adolescence, so it is
possible that the amnesia was not as ungraded as it appeared to be. In other cases, the
damage was known to extend substantially into lateral temporal neocortex (see Bright et al.
2006 and Squire & Bayley 2007 for consideration of several cases). In one report of patients
with unilateral temporal lobe resections, autobiographical memory was impaired across all
past time periods (Noulhiane et al. 2007). In these patients, damage was recorded in the
medial temporal lobe as well as in the temporal pole and in the anterior aspect of the
superior, middle, and inferior temporal gyri. It is difficult to know to what extent this
damage outside the medial temporal lobe might have contributed to the impairment.
Significant damage to lateral temporal or frontal cortex can severely impair performance on
tests of remote memory, including tests of autobiographical memory about early life [7
cases, Bright et al. (2006); patients H.C., P.H., and G.T., Bayley et al. (2005b); patient E.K.,

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 12

Eslinger (1998)]. If lateral temporal cortex, for example, is a site of long-term memory
storage (Mishkin 1982, Miyashita 1993), then lateral temporal damage would be expected to
cause severe and extended retrograde amnesia. The difficulty is knowing in any particular
NIH-PA Author Manuscript

case to what extent such damage is responsible for impaired remote memory.

Among several single-case studies reporting impaired memory for early-life events (see
Squire & Bayley 2007 for discussion), patient V.C. has been the most carefully documented.
The volume of his lateral temporal lobes was reported as normal. Yet, it is striking that
V.C.’s 1/9 score on the childhood portion of the autobiographical memory interview differs
sharply from the good scores (and sometimes maximum scores of 9) obtained on the same
test by as many as 12 patients with MRI documentation of limited medial temporal lobe
damage [n = 8, Bayley et al. (2006); n = 2, Eslinger (1998); n = 1, Kapur & Brooks (1999);
n = 1, Schnider et al. (1995). With the possible exception of V.C., we are unaware of
memory-impaired patients who have damage limited to the medial temporal lobe (as
documented by neurohistology or thorough MRI) and who do so poorly at recollecting
remote autobiographical memories (Figure 4).

The finding that retrograde amnesia is temporally limited after damage to the medial
temporal lobe implies a process of reorganization whereby over time memories become less
dependent on medial temporal lobe structures. As time passes after learning, the role of
medial temporal lobe structures diminishes and a more permanent memory gradually
NIH-PA Author Manuscript

develops, presumably in neocortex. According to a different perspective, only fact-based

memories (not autobiographical memories) make this transition (Winocur et al. 2010). This
view discounts the possible importance of neocortical damage in patients with impaired
autobiographical remembering of remote events and attributes the impairment specifically to
hippocampal damage.

Some studies in experimental animals have directly tracked neural activity and structural
changes in the hippocampus and neocortex after learning. Expression patterns of c-Fos
described gradually decreasing activity in the mouse hippocampus after learning and parallel
increases in a number of cortical regions (Frankland & Bontempi 2005). These findings and
others (Restivo et al. 2009) reflect the increasing importance of distributed cortical regions
for the representation of memory as time passes. The idea is not that memory is literally
transferred from the hippocampus to neocortex but that gradual changes in the neocortex
increase the complexity, distribution, and connectivity among multiple cortical regions. The
next section considers what the study of patients has contributed to understanding the
organization and storage of long-term memory.

MEMORY IN THE NEOCORTEX

NIH-PA Author Manuscript

The view that emerged from the study of H.M. and other patients is that medial temporal
lobe structures are uniquely specialized to establish and maintain declarative memories.
Other structures support the initial perception and processing of an experience, and these
other structures are also critical for the long-term storage of the experience. A long-standing
view is that the cortical processing of a multisensory experience leaves a distributed record
in the same multiple regions that initially performed the processing. For example, neurons in
visual areas store the visual aspect of a multisensory experience, neurons in auditory areas
store the auditory aspect of the experience, other areas store the spatial aspects, and so on.
According to this view, any act of remembering consists of the coordinated reactivation of
the distributed neocortical regions that were engaged at the time of encoding (Damasio
1989, De Renzi 1982, Mishkin 1982, Squire 1987). When a memory is first formed, this
reactivation depends on the hippocampus and related structures, but once memory is fully
consolidated, reactivation can occur independently in neocortex.

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 13

A considerable body of evidence supporting the reactivation view has come from studies
using fMRI (see Buckner & Wheeler 2001, Danker & Anderson 2010 for reviews). For
example, several studies have found that the modality-specific or category-specific
NIH-PA Author Manuscript

processes engaged at encoding tend to be re-engaged at retrieval (e.g., Polyn et al. 2005,
Wheeler et al. 2000, Woodruff et al. 2005). This perspective of remembering implies that
the dedicated processing areas of the neocortex can also be viewed as memory areas.
However, rather than broadly encoding and consolidating memories, like the structures of
the medial temporal lobe, each neocortical region operates within a very specific domain,
and each region stores only specific features of an experience. It follows then that the same
neocortical lesions that selectively impair processing in one particular domain should also
cause correspondingly specific anterograde and retrograde memory impairments within the
same domain. Although an extensive literature documents the selective information-
processing deficits that are associated with different cortical lesions, the effects of those
lesions on new learning and past remembering are only rarely considered. Here, we consider
the cognitive effects of selective processing deficits with a view toward also identifying the
effects on memory.

Achromatopsia
Finding selective anterograde memory impairment in association with a selective perceptual
processing deficit would not be surprising. That is, if a perceptual deficit is present in one
modality (e.g., visual perception), it should also be difficult to learn new material presented
NIH-PA Author Manuscript

in the same modality. In addition, there should be consequences for remembering the past.
Specifically, a selective deficit in processing particular features of visual material should
selectively compromise the ability to recollect the same features in a previous memory,
while leaving other aspects of the memory intact. This idea is illustrated by “The Case of the
Color-blind Painter” (Sacks 1995). An accomplished painter was involved in an automobile
accident at the age of 65, which rendered him completely color blind. Although the
anatomical basis of his disability was not identified, it was thought to have been caused by
damage to regions dedicated to the perception of color (possibly including area V4). The
disability itself was striking. The patient could discriminate between wavelengths of light,
even though the different wavelengths no longer gave rise to the perception of different
colors. Instead, different wavelengths gave rise to the perception of different shades of gray.
Because this was a case of acquired cerebral achromatopsia (i.e., cortical color blindness), it
was possible to ask about the status of previously established memories that had once
included the subjective experience of color. If color in early memories depends on the same
cortical structures that support the perception of color, then previously intact memories that
were once retrieved in color should now be retrieved in black and white. Indeed, the case
description leaves little doubt that the patient’s experience—both going forward and looking
NIH-PA Author Manuscript

back—was now completely (and selectively) devoid of color. Although he retained abstract
semantic knowledge of color, he could neither perceive nor later remember the color of
objects presented to him (anterograde impairment). In addition, he could not subjectively
experience color in his earlier (and once chromatic) memories (retrograde impairment). For
example, he knew that his lawn was green, but he reported that he could no longer visualize
it in green when he tried to remember what it once looked like.

Prosopagnosia
Similar effects have been documented by formal testing in cases of acquired prosopagnosia
(impaired recognition of faces, or face blindness). The cardinal complaint of patients
diagnosed with prosopagnosia is that they have a selective retrograde memory deficit. That
is, once-recognizable faces no longer yield a memory signal, even though other aspects of
one’s memory for the same individuals are preserved. For example, a patient who could not

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 14

recognize his mother’s face might continue to recognize the sound of her voice and still be
able to recall his prior experiences with her.
NIH-PA Author Manuscript

Patient L.H., a 37-year-old man, sustained a severe closed-head injury in an automobile

accident at the age of 18 (Farah et al. 1995a,b). His brain damage involved bilateral inferior
temporo-occipital regions, as well as the right inferior frontal lobe and right anterior
temporal lobe. Although general intellectual and elementary visual capabilities were
preserved following the accident, L.H. became profoundly impaired at recognizing
previously familiar faces. Along with this retrograde memory deficit, L.H. also exhibited a
perceptual processing deficit that was selective for upright faces. For example, on a same/
different face discrimination task, L.H. performed worse than controls at discriminating
upright faces (consistent with a face perception deficit), but he performed unexpectedly
better than controls at discriminating inverted faces (indicating that general perceptual
abilities were preserved). Patient L.H. also exhibited anterograde amnesia for new faces. For
example, L.H. and controls were presented with black and white photographs of both faces
and common objects and asked to memorize them (Farah et al. 1995a). On a later
recognition test, control subjects performed at the same level for faces and nonface objects.
L.H’s ability to remember faces was selectively impaired.

The retrograde memory deficit associated with acquired prosopagnosia is not confined to
recognition memory but applies as well to recalling and imaging the past. In one study,
NIH-PA Author Manuscript

(Barton & Cherkasova 2003), seven patients with adult-onset prosopagnosia performed
comparative judgments about the configuration of famous faces that they tried to retrieve
from memory (e.g., “Who has the more angular face: George Washington or Abraham
Lincoln?”). The famous faces used in this test were presumably familiar before the onset of
prosopagnosia. Even so, the patients were severely impaired on the face imagery task.
Together, the findings from acquired prosopagnosia—a modular perceptual processing
deficit associated with selective anterograde and retrograde amnesia—suggest that the same
areas that support the perception of faces also support the long-term memory of faces.

Amusia
This same set of findings, whereby an acquired and relatively modular processing deficit is
associated with corresponding memory deficits (both anterograde and retrograde), has also
been reported in a patient who lost the ability to recognize familiar music while retaining
other perceptual and intellectual functions (amusia). Patient I.R. suffered bilateral brain
damage at the age of 28 after undergoing a series of operations to clip aneurysms on the left
and right middle cerebral arteries (Peretz et al. 1998, Peretz & Gagnon 1999). At the time
she was tested (in her early 40s), CT scans indicated that the superior temporal gyrus was
severely damaged bilaterally, and the lesion also extended to involve structures in the frontal
NIH-PA Author Manuscript

cortex and anterior inferior parietal lobule.

I.R. was of normal intelligence, and her overall memory ability was normal as well. In
addition, she exhibited no evidence of a hearing impairment according to standard
audiometric tests, and except for music she had no difficulty recognizing familiar
environmental sounds. However, tunes that were once familiar to her were now
unrecognizable, and she could no longer sing music from memory (which she had
previously been able to do). Her selective retrograde amnesia for previously familiar music
was also accompanied by a selective perceptual deficit for music. Musical perception was
tested using a same/different format in which two short excerpts were presented in
succession (e.g., Mozart’s piano concerto #27 followed by Mozart’s piano concerto #23).
Controls found this task so easy that they made no errors even when the interstimulus
interval was long (20 s) and filled with conversation, but I.R.’s performance was no better
than 80% correct even when the interstimulus interval was short (4 s).

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 15

She also exhibited anterograde amnesia for new music. A list of 15 briefly presented
melodies was presented for study. On a subsequent old/new recognition test involving the 15
old melodies intermixed with 15 new ones, her memory performance was no better than
NIH-PA Author Manuscript

chance (whereas control performance exceeded 85% correct). Thus, as with the cases of
acquired achromatopsia and acquired prosopagnosia discussed earlier, impairments
associated with acquired amusia imply a close connection between information processing
and storage. The specificity of her anterograde and retrograde memory deficits corresponded
directly to the specificity of her perceptual deficit.

Knowledge Systems
The findings considered here are consistent with the idea that memory storage in the
neocortex reflects the outcome of the perceptual processing and analysis that occurred at the
time of learning. A related literature concerns the status of stored semantic knowledge and
its relation to information processing. These studies do not document a deficit in specific
perceptual processing modules. Instead, they document the effects of cortical lesions (e.g., to
posterior temporal cortex) on previously acquired knowledge within specific semantic
categories, and they relate these deficits to the kinds of processing involved when the
knowledge was first acquired.

The idea that knowledge systems may be organized by semantic categories was discussed by
Warrington & Shallice (1984). They described four patients with widespread bilateral
NIH-PA Author Manuscript

lesions (following herpes simplex encephalitis) that included the medial and lateral temporal
lobes. In addition to having global amnesia, all four patients exhibited an asymmetry in their
ability to identify animate and inanimate objects. They had a selective impairment in the
ability to name or describe pictures of animate objects (e.g., animals and plants). By
contrast, their ability to name or describe pictures of inanimate objects (e.g., broom, pencil,
umbrella) appeared to be preserved. Assuming that all the objects were previously familiar
to the patients, the findings describe a category-specific retrograde memory impairment.

Other patients exhibited the opposite impairment. For example, patient Y.O.T., who had
damage to the left temporoparietal region (thought to have resulted from a
thromboembolism), showed relatively preserved knowledge of living things and poor
knowledge of inanimate objects (Warrington & McCarthy 1987). However, her
comprehension of body parts and fabrics was anomalous in that she exhibited knowledge
about fabric names (non-living things) and poor knowledge about body parts (living things).
In addition, Warrington & McCarthy (1987) noted that patient J.B.R. [one of the four
patients previously described by Warrington & Shallice (1984)], who had exhibited a
selective loss of knowledge about living things, nevertheless had preserved knowledge about
body parts (living things) and poor knowledge about fabrics (nonliving things). These
NIH-PA Author Manuscript

findings suggested that the principle by which knowledge is organized in the brain concerns
whether objects are identified mainly by their physical features (form, color, texture, etc.) or
by their function and how they are used. Generally, the animate/inanimate distinction fits
this principle, but the exceptions are telling. Most animals are identified by their physical
attributes, not by what can be done with them. By contrast, small inanimate objects are
usually identified by their functions and how they are used (e.g., sweep with a broom, write
with a pencil). However, some living things (such as body parts) are identified largely by
their function, and some nonliving things (such as fabrics) are identified largely by their
texture and shape. A recent comprehensive review of neuroimaging evidence strongly
supports this account of stored semantic knowledge (Martin 2007).

If these category-specific retrograde memory deficits reflect the loss of knowledge that was
initially acquired through category-specific processing, then a corresponding anterograde
memory deficit would be expected, as well. Thus, for example, a patient who exhibits a

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 16

selective deficit in naming or describing objects that are defined by how they are used
should also exhibit a selective deficit in learning novel objects that are defined by how they
are used. To our knowledge, this prediction has not been tested.
NIH-PA Author Manuscript

RECOLLECTION AND FAMILIARITY

In recent years, there has been extended investigation of the idea that the different medial
temporal lobe structures (hippocampus, entorhinal cortex, perirhinal cortex, and
parahippocampal cortex) may support different memory functions. The study of H.M. could
not address this issue because his bilateral lesions included most of these structures.
However, other patients, especially patients with limited hippocampal lesions, have been
useful in this regard.

One issue that has commanded considerable attention concerns the roles played by the
hippocampus and perirhinal cortex in recognition memory. Recognition memory is thought
to be supported by two processes, recollection and familiarity (Atkinson & Juola 1974,
Mandler 1980). Recollection involves remembering specific contextual details about a prior
learning episode; familiarity involves simply knowing that an item was presented without
having available any additional information about the learning episode. According to one
view, both the hippocampus and the perirhinal cortex contribute to recollection and
familiarity (Squire et al. 2007, Wixted & Squire 2010). According to a different view, the
NIH-PA Author Manuscript

hippocampus and perirhinal cortex selectively support recollection and familiarity,

respectively (Brown & Aggleton 2001, Eichenbaum et al. 2007).

Recall versus Recognition

One approach to investigating this issue has been to compare performance on an old/new
recognition task, which is widely thought to be supported by both recollection and
familiarity, with performance on a task of free recall, which is thought to depend mainly on
recollection. (In a free recall task, subjects are presented with a list of items to memorize and
are later asked to recall those items in any order they wish.) Because old/new recognition
can be partially supported by familiarity, the question of interest is whether the performance
of patients with hippocampal lesions is disproportionately better on an old/new recognition
task in comparison to free recall.

Several case studies and group studies have asked this question of patients with adult-onset
bilateral lesions that, according to quantitative MRI, are limited to the hippocampus. The
case studies differ in their findings about the status of old/new recognition memory
(Aggleton et al. 2005, Cipolotti et al. 2006, Mayes et al. 2002). Because the differing results
may reflect individual differences, group studies are more informative. Two group studies
NIH-PA Author Manuscript

have shown that the degree of impairment is similar when old/new recognition and free
recall are compared (Kopelman et al. 2007, Manns et al. 2003a) (Figure 5). Another group
study involved 56 hypoxic patients with damage believed to be limited to the hippocampus
(no radiological information was available) (Yonelinas et al. 2002). The patients were less
impaired on old/new recognition than on free recall. However, this conclusion was later
shown to result from the remarkably aberrant recognition performance of a single 1 of the 55
control subjects (Wixted & Squire 2004). With that one outlier removed from the analysis,
the patients and controls exhibited similar levels of impairment on recall and recognition.
The recognition z-score for the patients was −0.59 (before removal of the outlier, z =
−0.39), and the recall z-score was a statistically indistinguishable −0.68. Thus, the available
group studies are consistent in showing that the degree of memory impairment in patients
with lesions limited to the hippocampus is similar for old/new recognition (which is
substantially supported by familiarity) and for free recall (which is fully dependent on

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 17

recollection). These findings suggest that the hippocampus is important for both recollection
and familiarity.
NIH-PA Author Manuscript

Remember/Know Procedure
Another method that has been used to investigate the role of the medial temporal lobe in
recollection and familiarity is the Remember/ Know procedure, which is based on subjective
reports of whether recollection is available when an item is declared old. Participants report
Remember when they can recollect something about the original encounter with the item
(e.g., its context, what thoughts they had), and they report Know when they judge the item to
be familiar but cannot recollect anything about its presentation. The Remember/Know
judgments made by patients and controls are often converted into quantitative estimates of
recollection and familiarity based on a widely used but controversial model of recognition
memory (Yonelinas 1994). Using this method, some studies have reported that recollection
is selectively impaired in patients with hippocampal lesions (Yonelinas et al. 2002), whereas
other studies have found impairments in both recollection and familiarity (Manns et al.
2003a). A difficulty with deriving quantitative estimates of recollection and familiarity from
Remember/Know judgments is that the assumptions of the model that is used to derive
estimates have generally not been supported by empirical test (e.g., Heathcote 2003, Rotello
et al. 2005, Slotnick 2010, Slotnick & Dodson 2005). In particular, Know judgments reflect
weaker memory than do Remember judgments, as measured by both confidence and
accuracy (e.g., Dunn 2004, Squire et al. 2007). Thus, a supposed impairment in recollection
NIH-PA Author Manuscript

(Remembering) after hippocampal lesions could simply mean that the patients have few
strong memories (and that what would have been strong memories are now weak
memories), not that recollection is selectively affected. The Remember/Know procedure
could be used to study recollection and familiarity effectively if Remember and Know
judgments were first equated for confidence and accuracy, but this approach has not been
used in patient studies to date.

Analysis of the Receiver Operating Characteristic

Still another method that has been used to estimate recollection and familiarity has been to
fit the Yonelinas (1994) dual-process model to receiver operating characteristic (ROC) data.
This is the same model that has often been used to estimate recollection and familiarity
using the Remember/Know procedure. An ROC is a plot of the hit rate versus the false
alarm rate across different decision criteria. Typically, multiple pairs of hit and false alarm
rates are obtained by asking subjects to provide confidence ratings for their old/new
recognition decisions. A pair of hit and false alarm rates is then computed for each level of
confidence, and the paired values are plotted across the confidence levels. The points of an
ROC typically trace out a curvilinear path that can be characterized in terms of its symmetry
NIH-PA Author Manuscript

relative to the negative diagonal (Figure 6). The dual-process model proposed by Yonelinas
(1994) holds that the degree of asymmetry in an ROC directly reflects the degree to which
the recollection process is involved in recognition decisions. Accordingly, a symmetrical
ROC indicates that recognition decisions were based solely on familiarity, and an
asymmetrical ROC indicates that recollection occurred for some of the items, as well.

The finding that memory-impaired patients produce symmetrically curvilinear ROCs,

whereas controls produce asymmetrical curvilinear ROCs, has been interpreted to mean that
the recollection process is selectively impaired by hippocampal lesions (Yonelinas et al.
1998, 2000). However, once again, this is a model-dependent interpretation, and much
evidence that has accumulated against this model in recent years instead supports an
alternative signal-detection model (e.g., Dunn 2004, 2008; Wixted 2007; Wixted & Mickes
2010). According to the signal-detection model, a symmetrical ROC does not indicate
familiarity-based responding but simply reflects weaker memory. Because patients have

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 18

weaker memory than do controls, the fact that patients tend to exhibit symmetrical ROCs is
not surprising.
NIH-PA Author Manuscript

The question is whether patients can exhibit asymmetrical ROCs (like controls) once the
strength of memory is equated. In one study, patients with lesions limited to the
hippocampus were studied under two conditions (weak and strong memory) (Wais et al.
2006). In the weak condition, patients studied 50-item word lists, as did matched controls.
As expected, the controls performed better than the patients did. In addition (again as
expected), the control ROC was asymmetrical, and the patient ROC was symmetrical. To
equate for overall memory strength, patients also studied lists of 10 items, which improved
their memory performance to a level similar to that of the controls who had studied 50-item
lists. In this condition, the patient ROC and the control ROC were similarly asymmetrical.
These results show that patients can exhibit asymmetric ROCs, which have been taken to
denote performance based on recollection. The results further suggest that the typical finding
of asymmetrical ROCs for controls and symmetrical ROCs for patients does not necessarily
indicate a selective deficit in recollection but can reflect a difference in overall memory
strength.

Newer (Model-Free) Methods

The effects of hippocampal lesions on recollection and familiarity can also be studied in a
way that does not depend on the assumptions of any specific psychological model. If
NIH-PA Author Manuscript

hippocampal lesions selectively impair recollection, and preserve familiarity, then patients
with hippocampal lesions should commonly experience strong, familiarity-based recognition
that is unaccompanied by recollection. Furthermore, this experience should occur even more
frequently in patients than controls because ordinarily when strong, familiarity-based
recognition occurs (e.g., seeing a familiar face), details about prior encounters are
remembered, as well.

In a formal test of this prediction, five patients with circumscribed hippocampal damage
studied 25 words in one of two contexts (source A or source B) (Kirwan et al. 2010). Old/
new recognition memory for the words was then tested using a six-point confidence scale (1
= sure new, 6 = sure old). For items endorsed as old, participants were also asked to make a
source recollection decision (was the item learned in context A or B?). Old decisions made
with high confidence but in the absence of successful source recollection would thus
correspond to strong, familiarity-based recognition without recollection. The results were
that there was no increased tendency for this experience to occur in patients relative to
controls. If anything, the experience was less frequent in the patients. The simplest
explanation for this result is that hippocampal damage impairs familiarity as well as
recollection.
NIH-PA Author Manuscript

In summary, a large body of evidence based on the Remember/Know procedure and ROC
analysis has been interpreted to mean that the hippocampus subserves recollection and plays
no role in familiarity. It is often not appreciated that this interpretation is based on a specific
model that equates weak memory with familiarity and strong memory with recollection [the
model proposed by Yonelinas (1994)]. However, familiarity can sometimes be strong, and
recollection can sometimes be weak (Wixted & Mickes 2010). In studies that do not depend
on this model, the results suggest that hippocampal lesions impair both recollection and
familiarity (Kirwan et al. 2010, Wais et al. 2006).

The fact that a memory strength confound can explain why earlier studies have failed to
detect impaired familiarity in hippocampal patients should not be taken to mean that
“memory strength” is a concept that usefully informs the functional organization of medial
temporal lobe structures. Consideration of how these structures contribute differently to

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 19

memory properly begins with neuroanatomy. Information from neocortex enters the medial
temporal lobe at different points (Suzuki & Amaral 1994). Perirhinal cortex receives strong
input from unimodal visual areas, and the parahippocampal cortex receives prominent
NIH-PA Author Manuscript

projections from areas important for spatial cognition, including posterior parietal cortex.
This anatomical specialization suggests that perirhinal cortex may be especially important
for visual memory (regardless whether a task requires recollection), and the
parahippocampal cortex may be important for spatial memory. The finding of severe
impairment in monkeys in visual associative tasks after perirhinal lesions (Murray et al.
1993) and in spatial tasks after parahippocampal cortex lesions (Malkova & Mishkin 2003)
conforms to this suggestion. The hippocampus itself receives input from the adjacent cortex
and is thus in a position to combine the operations of memory formation that are carried out
by the more specialized structures that project to it. As expected, hippocampal lesions impair
both visual memories and spatial memories. The impairment in memory formation is only
modestly severe because many memory functions can be carried out by the adjacent cortex
[for additional discussion of differences in the function of medial temporal lobe structures,
see Squire et al. (2007), Wixted & Squire (2011)].

GROUP STUDIES AND MULTIPLE METHODS

The study of patients with medial temporal lobe lesions (especially the severely impaired
patient H.M.) has led to dramatic advances in understanding the structure and organization
NIH-PA Author Manuscript

of memory. As work progressed, many studies came to focus on smaller lesions and less
severe impairments. Furthermore, many of these studies were based on single cases and
investigated specific questions about particular aspects of the impairment (for example,
recall versus recognition or recollection versus familiarity). Such questions are difficult to
settle from individual case studies because the expected effects are relatively small. Under
such conditions, a deficit may reflect (unmeasured) premorbid individual differences rather
than the effect of a focal brain lesion. Accordingly, for many questions single case studies
are more suggestive than conclusive, and group studies are needed to answer experimental
questions in a compelling way.

The advantage of group studies is that individual variability tends to be averaged out.
However, group studies are useful only to the extent that the lesions can be documented and
quantified with MRI. Some group studies have studied patients with assumed lesions, such
as patients with modest memory impairments due to hypoxia who are studied on the
untested assumption that their lesions are limited to the hippocampus. Given techniques
currently available for quantifying the locus and extent of lesions, the use of such techniques
in both single-case and group studies should become standard practice.
NIH-PA Author Manuscript

It is important to emphasize that studies of patients with lesions provide only one of many
experimental approaches to investigating the organization of memory. The same issues have
been usefully investigated in experimental animals with lesions (e.g., rats and monkeys), in
single-unit recording studies of animals and humans, in studies using functional
neuroimaging or transcranial magnetic stimulation (TMS), and in studies of genetically
modified mice. Each approach has its own advantages and disadvantages, such that one can
expect that their combined application will provide the best opportunity for further
discovery.

CONCLUSIONS
The early descriptions of H.M. changed how human memory was understood. What became
clear as a result of work with H.M.—and what remains clear today—is that the structures of
the medial temporal lobe are essential for normal memory function. Specifically, these

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 20

structures are thought to be important for the formation of memory and for the maintenance
of memory for a period of time after learning. Although active lines of research investigate
the possibility that these structures also contribute to other domains of cognitive function
NIH-PA Author Manuscript

(e.g., visual perception, working memory, and online computations supporting spatial
cognition), the half-century of research that began with H.M. has shown that profound
impairment after medial temporal lobe damage occurs in only one domain, specifically, in
what is now termed declarative memory.

The elements of long-term memory are stored in the neocortex (not in the medial temporal
lobe) as products of the distributed, domain-specific processing that occurred in different
regions of neocortex at the time of learning. Thus, long-term memory for whole events is
widely represented, but the multiple areas that are involved each store distinct components
of information. In addition, acts of remembering involve the reactivation of the same
neocortical regions that initially processed and stored what was learned. The role of the
medial temporal lobe is to consolidate the distributed elements of memory into a coherent
and stable ensemble (a process that can take years). Many questions remain about how
consolidation occurs, as well as about memory storage, memory retrieval, and the specific
functions of the different medial temporal lobe structures and the different areas of
neocortex. These topics encompass what has become a substantial and fruitful tradition of
research within systems and cognitive neuroscience—a tradition that began with the study of
H.M.
NIH-PA Author Manuscript

Acknowledgments
This work was supported by the Medical Research Service of the Department of Veterans Affairs, NIMH grant
MH24600 to L.R.S., and NIMH grant MH082892 to J.T.W.

LITERATURE CITED
Aggleton JP, Vann SD, Denby C, Dix S, Mayes AR, et al. Sparing of the familiarity component of
recognition memory in a patient with hippocampal pathology. Neuropsychologia. 2005; 43:810–23.
Atkinson, RC.; Juola, JF. Search and decision processes in recognition memory. In: Krantz, DH.;
Atkinson, RC.; Suppes, P., editors. Contemporary Developments in Mathematical Psychology. San
Francisco: Freeman; 1974. p. 243-90.
Baddeley A. Working memory: looking back and looking forward. Nat Rev Neurosci. 2003; 4:829–39.
[PubMed: 14523382]
Barense MD, Gaffan D, Graham KS. The human medial temporal lobe processes online
representations of complex objects. Neuropsychologia. 2007; 45:2963–74. [PubMed: 17658561]
Barton JJS, Cherkasova M. Face imagery and its relation to perception and covert recognition in
prosopagnosia. Neurology. 2003; 61:220–25. [PubMed: 12874402]
NIH-PA Author Manuscript

Bayley PJ, Frascino JC, Squire LR. Robust habit learning in the absence of awareness and independent
of the medial temporal lobe. Nature. 2005a; 436:550–53. [PubMed: 16049487]
Bayley PJ, Gold JJ, Hopkins RO, Squire LR. The neuroanatomy of remote memory. Neuron. 2005b;
46:799–810. [PubMed: 15924865]
Bayley PJ, Hopkins RO, Squire LR. Successful recollection of remote autobiographical memories by
amnesic patients with medial temporal lobe lesions. Neuron. 2003; 37:135–44. [PubMed:
12526779]
Bayley PJ, Hopkins RO, Squire LR. The fate of old memories after medial temporal lobe damage. J
Neurosci. 2006; 26:13311–17. [PubMed: 17182781]
Bird CM, Burgess N. The hippocampus and memory: insights from spatial processing. Nat Rev
Neurosci. 2008; 9:182–94. [PubMed: 18270514]
Bird CM, Capponi C, King JA, Doeller CF, Burgess N. Establishing the boundaries: the hippocampal
contribution to imagining scenes. J Neurosci. 2010; 30:11688–95. [PubMed: 20810889]

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 21

Bright P, Buckman JR, Fradera A, Yoshimasu H, Colchester ACF, Kopelman MD. Retrograde
amnesia in patients with hippocampal, medial temporal, temporal lobe, or frontal pathology. Learn
Mem. 2006; 13:545–57. [PubMed: 17015852]
NIH-PA Author Manuscript

Broca, PP. Remarks on the seat of the faculty of articulate language, followed by an observation of
aphemia. In: von Bonin, G., translator. Some Papers on the Cerebral Cortex. Springfield: IL: 1861.
p. 199-220.
Thomas Brown MW, Aggleton JP. Recognition memory: What are the roles of the perirhinal cortex
and hippocampus? Nat Rev Neurosci. 2001; 2:51–61. [PubMed: 11253359]
Buchanan TW, Tranel D, Adolphs R. Emotional autobiographical memories in amnesic patients with
medial temporal lobe damage. J Neurosci. 2005; 25:3151–60. [PubMed: 15788772]
Buckner RL, Wheeler ME. The cognitive neuroscience of remembering. Nat Rev Neurosci. 2001;
2:624–34. [PubMed: 11533730]
Cave C, Squire LR. Intact and long-lasting repetition priming in amnesia. J Exp Psychol Learn Mem
Cog. 1992; 18:509–20.
Cipolotti L, Bird C, Good T, Macmanus D, Rudge P, Shallice T. Recollection and familiarity in dense
hippocampal amnesia: a case study. Neuropsychologia. 2006; 44:489–506. [PubMed: 16023686]
Cipolotti L, Shallice T, Chan D, Fox N, Scahill R, et al. Long-term retrograde amnesia…the crucial
role of the hippocampus. Neuropsychologia. 2001; 2:151–72. [PubMed: 11163373]
Clark RE, Reinagel P, Broadbent N, Flister E, Squire LR. Intact performance on feature ambiguous
discriminations in rats with lesions of the perirhinal cortex. Neuron. 2011 In press.
Cohen NJ, Squire LR. Preserved learning and retention of pattern analyzing skill in amnesia:
NIH-PA Author Manuscript

dissociation of knowing how and knowing that. Science. 1980; 210:207–9. [PubMed: 7414331]
Corkin S. What’s new with the amnesic patient H.M.? Nat Rev Neurosci. 2002; 3:153–60. [PubMed:
11836523]
Corkin S, Amaral DG, Gonzalez RG, Johnson KA, Hyman BT. H.M.’s medial temporal lobe lesion:
findings from magnetic resonance imaging. J Neurosci. 1997; 17:3964–80. [PubMed: 9133414]
Cowan N. The magical number 4 in short-term memory: a reconsideration of mental storage capacity.
Behav Brain Sci. 2001; 24:87–185. [PubMed: 11515286]
Damasio AR. Time-locked multiregional retroactivation: a systems-level proposal for the neural
substrates of recall and recognition. Cognition. 1989; 33:25–62. [PubMed: 2691184]
Danker JF, Anderson JR. The ghosts of brain states past: remembering reactivates the brain regions
engaged during encoding. Psychol Bull. 2010; 136:87–102. [PubMed: 20063927]
De Renzi E. Memory disorders following focal neocortical damage. Philos R Soc London [Biol].
1982; 298:73–83.
Drachman DA, Arbit J. Memory and the hippocampal complex. II Is memory a multiple process? Arch
Neurol. 1966; 15:52–61. [PubMed: 5937496]
Dunn JC. Remember-Know: a matter of confidence. Psychol Rev. 2004; 111:524–42. [PubMed:
15065921]
Dunn JC. The dimensionality of the remember–know task: a state-trace analysis. Psychol Rev. 2008;
NIH-PA Author Manuscript

115:426–46t. [PubMed: 18426296]

Eacott MJ, Gaffan D, Murray EA. Preserved recognition memory for small sets, and impaired stimulus
identification for large sets, following rhinal cortex ablations in monkeys. Euro J Neurosci. 1994;
6:1466–78.
Eichenbaum H, Yonelinas AR, Ranganath C. The medial temporal lobe and recognition memory.
Annu Rev Neurosci. 2007; 30:123–52. [PubMed: 17417939]
Eslinger PJ. Autobiographical memory after temporal lobe lesions. Neurocase. 1998; 4:481–95.
Farah MJ, Levinson KL, Klein KL. Face perception and within-category discrimination in
prosopagnosia. Neuropsychologia. 1995a; 33:661–74. [PubMed: 7675159]
Farah MJ, Wilson KD, Drain HM, Tanaka J. The inverted face inversion effect in prosopagnosia:
evidence for mandatory, face-specific perceptual mechanisms. Vision Res. 1995b; 35:2089–93.
[PubMed: 7660612]
Ferrier, D. The Functions of the Brain. London: Smith, Elder and Company; 1876.

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 22

Finger, S. Origins of Neuroscience: A History of Explorations into Brain Function. New York: Oxford
Univ. Press; 1994.
Finke C, Braun M, Ostendorf F, Lehmann TN, Hoffmann KT, et al. The human hippocampal
NIH-PA Author Manuscript

formation mediates short-term memory of colour-location associations. Neuropsychologia. 2008;

46:614–23. [PubMed: 18023459]
Frankland PW, Bontempi B. The organization of recent and remote memories. Nat Rev Neurosci.
2005; 6:119–30. [PubMed: 15685217]
Fritsch, G.; Hitzig, E. On the electrical excitability of the cerebrum. In: von Bonin, G., translator.
Some Papers on the Cerebral Cortex. Springfield, IL: Thomas; 1870. p. 73-86.
Fukuda K, Edward A, Vogel EK. Discrete capacity limits in visual working memory. Curr Opin
Neurobiol. 2010; 20:177–82. [PubMed: 20362427]
Gall, FJ. Sur les Fonctions du Cerveau et sur Celles de Chacune des Ses Parties. Vol. 6. Paris:
Bailliere; 1825.
Gold JJ, Squire LR. Quantifying medial temporal lobe damage in memory-impaired patients.
Hippocampus. 2005; 15:79–85. [PubMed: 15390163]
Gold JJ, Squire LR. The anatomy of amnesia: neurohistological analysis of three new cases. Learn
Mem. 2006; 13:699–710. [PubMed: 17101872]
Graham KS, Scahill VL, Hornberger M, Barense MD, Lee AC, et al. Abnormal categorization and
perceptual learning in patients with hippocampal damage. J Neurosci. 2006; 26:7547–54.
[PubMed: 16855082]
Hannula DE, Tranel D, Cohen NJ. The long and short of it: relational memory impairments in
NIH-PA Author Manuscript

amnesia, even at short lags. J Neurosci. 2006; 26:8352–59. [PubMed: 16899730]

Harding A, Halliday G, Caine D, Kril J. Degeneraton of anterior thalamic nuclei differentiates
alcoholics with amnesia. Brain. 2000; 123:141–54. [PubMed: 10611128]
Hassabis D, Kumaran D, Vann SD, Maguire EA. Patients with hippocampal amnesia cannot imagine
new experiences. Proc Natl Acad Sci USA. 2007; 104:1726–31. [PubMed: 17229836]
Heathcote A. Item recognition memory and the ROC. J Exp Psych Learn Mem Cogn. 2003; 29:1210–
30.
Hebb, DO. The Organization of Behavior. New York: Wiley; 1949.
Higuchi S, Miyashita Y. Formation of mnemonic neuronal responses to visual paired associates in
inferotemporal cortex is impaired by perirhinal and entorhinal lesions. Proc Natl Acad Sci USA.
1996; 93:739–43. [PubMed: 8570626]
Hunter WS. A consideration of Lashley’s theory of equipotentiality of cerebral action. J Gen Psych.
1930; 3:444–68.
Insausti R, Insausti AM, Sobreviela M, Salinas A, Martinez-Penuela J. Human medial temporal lobe in
aging: anatomical basis of memory preservation. Microsc Res Techn. 1998a; 43:8–15.
Insausti R, Juottonen K, Soininen H, Insausti AM, Partanen K, et al. MR volumetric analysis of the
human entorhinal, perirhinal, and temporopolar cortices. Am J Neuroradiol. 1998b; 19:659–71.
[PubMed: 9576651]
NIH-PA Author Manuscript

Isaacs EB, Vargha-Khadem F, Watkins KE, Lucas A, Mishkin M, Gadian DG. Developmental
amnesia and its relationship to degree of hippocampal atrophy. Proc Natl Acad Sci USA. 2003;
100:13060–63. [PubMed: 14555756]
Jeneson A, Mauldin KN, Hopkins RO, Squire LR. The role of the hippocampus in retaining relational
information across short delays: the importance of memory load. Learn Mem. 2011 In press.
Jeneson A, Mauldin KN, Squire LR. Intact working memory for relational information after medial
temporal lobe damage. J Neurosci. 2010; 30:13624–29. [PubMed: 20943903]
Kan IP, Giovanello KS, Schnyer DM, Makris N, Verfaellie M. Role of the medial temporal lobes in
relational memory: neuropsychological evidence from a cued recognition paradigm.
Neuropsychologia. 2007; 45:2589–97. [PubMed: 17433382]
Kapur N, Brooks DJ. Temporally-specific retrograde amnesia in two cases of discrete bilateral
hippocampal pathology. Hippocampus. 1999; 9:247–54. [PubMed: 10401640]

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 23

Kensinger E, Ullman MT, Corkin S. Bilateral medial temporal lobe damage does not affect lexical or
grammatical processing: evidence from amnesic patient H.M. Hippocampus. 2001; 11:347–60.
[PubMed: 11530839]
NIH-PA Author Manuscript

Kim S, Jeneson A, van der Horst AS, Frascino JC, Hopkins RO, Squire LR. Memory, visual
discrimination performance, and the human hippocampus. J Neurosci. 2011; 31:2624–29.
[PubMed: 21325530]
Kirwan CB, Bayley PJ, Galvan VV, Squire LR. Detailed recollection of remote autobiographical
memory after damage to the medial temporal lobe. Proc Natl Acad Sci USA. 2008; 105:2676–80.
[PubMed: 18252820]
Kirwan CB, Wixted JT, Squire LR. A demonstration that the hippocampus supports both recollection
and familiarity. Proc Natl Acad Sci USA. 2010; 107:344–48. [PubMed: 19966306]
Knowlton BJ, Mangels JA, Squire LR. A neostriatal habit learning system in humans. Science. 1996;
273:1399–402. [PubMed: 8703077]
Kopelman MD, Bright P, Buckman J, Fradera A, Yoshimasu H, et al. Recall and recognition memory
in amnesia: patients with hippocampal, medial temporal, temporal lobe or frontal pathology.
Neuropsychologia. 2007; 45:1232–46. [PubMed: 17140609]
Kopelman MD, Wilson BA, Baddeley AD. The autobiographical memory interview: a new assessment
of autobiographical and personal semantic memory in amnesic patients. J Clin Exp Neuropsy.
1989; 5:724–44.
Lashley, KS. Brain Mechanisms and Intelligence: A Quantitative Study of Injuries to the Brain.
Chicago: Chicago Univ. Press; 1929.
NIH-PA Author Manuscript

Lee AC, Buckley MJ, Pegman SJ, Spiers H, Scahill VL, et al. Specialization in the medial temporal
lobe for processing of objects and scenes. Hippocampus. 2005a; 15:782–97. [PubMed: 16010661]
Lee AC, Bussey TJ, Murray EA, Saksida LM, Epstein RA, et al. Perceptual deficits in amnesia:
challenging the medial temporal lobe ‘mnemonic’ view. Neuropsychologia. 2005b; 43:1–11.
[PubMed: 15488899]
Lee AC, Rudebeck SR. Human medial temporal lobe damage can disrupt the perception of single
objects. J Neurosci. 2010; 30:6588–94. [PubMed: 20463221]
Mair WGP, Warrington EK, Weiskrantz L. Memory disorder in Korsakoff psychosis. A
neuropathological and neuropsychological investigation of two cases. Brain. 1979; 102:749–83.
[PubMed: 116710]
Malkova L, Mishkin M. One-trial memory for object-place associations after separate lesions of
hippocampus and posterior parahippocampal region in the monkey. J Neurosci. 2003; 23:1956–65.
[PubMed: 12629201]
Mandler G. Recognizing: the judgment of previous occurrence. Psychol Rev. 1980; 87:252–71.
Manns JR, Hopkins RO, Reed JM, Kitchener EG, Squire LR. Recognition memory and the human
hippocampus. Neuron. 2003a; 37:171–80. [PubMed: 12526782]
Manns JR, Hopkins RO, Squire LR. Semantic memory and the human hippocampus. Neuron. 2003b;
37:127–33.
NIH-PA Author Manuscript

Marslen-Wilson WD, Teuber HL. Memory for remote events in anterograde amnesia: recognition of
public figures from news photographs. Neuropsychologia. 1975; 13:353–64. [PubMed: 1161134]
Martin A. The representation of object concepts in the brain. Ann Review Psychol. 2007; 58:25–45.
Mayes AR, Holdstock JS, Isaac CL, Hunkin NM, Roberts N. Relative sparing of item recognition
memory in a patient with adult-onset damage limited to the hippocampus. Hippocampus. 2002;
12:325–40. [PubMed: 12099484]
Mayes AR, Meudell PR, Mann D, Pickering A. Location of lesions in Korsakoff ’s syndrome:
neuropsychological and neuropathological data on two patients. Cortex. 1988; 3:367–88.
[PubMed: 3191722]
Milner, B. Physiologie de l’hippocampe. Paris: Cent. Natl. Rech. Sci; 1962. Les troubles de la
memoire accompagnant des lesions hippocampiques bilaterales; p. 257-72.
Milner B, Corkin S, Teuber HL. Further analysis of the hippocampal amnesic syndrome: 14 year
follow-up study of H.M. Neuropsychologia. 1968; 6:215–34.

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 24

Mishkin M. Memory in monkeys severely impaired by combined but not by separate removal of
amygdala and hippocampus. Nature. 1978; 273:297–98. [PubMed: 418358]
Mishkin M. A memory system in the monkey. Philos Trans R Soc London Ser B. 1982; 1089:83–95.
NIH-PA Author Manuscript

[PubMed: 6125978]
Mishkin, M.; Malamut, B.; Bachevalier, J. Memories and habits: two neural systems. In: Lynch, G.;
McGaugh, JL.; Weinberger, NM., editors. Neurobiology of Learning and Memory. New York:
Guilford; 1984. p. 65-77.
Miyashita Y. Inferior temporal cortex: where visual perception meets memory. Annu Rev Neurosci.
1993; 16:245–63. [PubMed: 8460893]
Murray, EA. Medial temporal lobe structures contributing to recognition memory: the amygdaloid
complex versus rhinal cortex. In: Aggleton, JP., editor. The Amygdala: Neurobiological Aspects
of Emotion, Memory, and Mental Dysfunction. London: Wiley-Liss; 1992. p. 453-70.
Murray EA, Gaffan D, Mishkin M. Neural substrates of visual stimulus-stimulus association in rhesus
monkeys. J Neurosci. 1993; 13:4549–61. [PubMed: 8410203]
Noulhiane M, Piolino P, Hasboun D, Clemenceau S, Baulac M, Samson S. Autobiographical memory
after temporal lobe resection: neuropsychological and MRI volumetric findings. Brain. 2007;
130:3184–99. [PubMed: 17986479]
Olson IR, Page K, Moore KS, Chatterjee A, Verfaellie M. Working memory for conjunctions relies on
the medial temporal lobe. J Neurosci. 2006; 26:4596–601. [PubMed: 16641239]
Peretz I, Gagnon L. Dissociation between recognition and emotional judgments for melodies.
Neurocase. 1999; 5:21–30.
NIH-PA Author Manuscript

Peretz I, Gagnon L, Bouchard B. Music and emotion: perceptual determinants, immediacy, and
isolation after brain damage. Cognition. 1998; 68:111–41. [PubMed: 9818509]
Polyn SM, Natu VS, Cohen JD, Norman KA. Category-specific cortical activity precedes retrieval
during memory search. Science. 2005; 310:1963–66. [PubMed: 16373577]
Press GA, Amaral DG, Squire LR. Hippocampal abnormalities in amnesic patients revealed by high-
resolution magnetic resonance imaging. Nature. 1989; 341:54–57. [PubMed: 2770879]
Rempel-Clower N, Zola SM, Squire LR, Amaral DG. Three cases of enduring memory impairment
following bilateral damage limited to the hippocampal formation. J Neurosci. 1996; 16:5233–55.
[PubMed: 8756452]
Restivo L, Vetere G, Bontempi B, Ammassari-Teule M. The formation of recent and remote memory
is associated with time-dependent formation of dendritic spines in the hippocampus and anterior
cingulate cortex. J Neurosci. 2009; 29:8206–14. [PubMed: 19553460]
Ribot, T. Les Maladies de la Memoire [English translation: Diseases of Memory]. New York:
Appleton-Century-Crofts; 1881.
Rosenbaum RS, Moscovitch M, Foster JK, Schnyer DM, Gao F, et al. Patterns of autobiographical
memory loss in medial-temporal lobe amnesic patients. J Cogn Neurosci. 2008; 20:1490–506.
[PubMed: 18303977]
Rosenbaum RS, Priselac S, Kohler S, Black SE, Gao F, et al. Remote spatial memory in an amnesic
NIH-PA Author Manuscript

person with extensive bilateral hippocampal lesions. Nat Neurosci. 2000; 3:1044–48. [PubMed:
11017178]
Rotello CM, Macmillan NA, Reeder JA, Wong M. The remember response: subject to bias, graded,
and not a process-pure indicator of recollection. Psychon Bull Rev. 2005; 12:865–73. [PubMed:
16524003]
Sacks, O. An Anthropologist on Mars. New York: Random House; 1995. The case of the colorblind
painter; p. 3-41.
Sagar, HH.; Cohen, NJ.; Corkin, S.; Growdon, JM. Dissociations among processes in remote memory.
In: Olton, DS.; Gamzu, E.; Corkin, S., editors. Memory Dysfunctions. New York: Ann. NY
Acad. Sci; 1985. p. 533-35.
Salat DH, van der Kouwe AJW, Tuch DS, Quinn BT, Fischl B, et al. Neuroimaging H.M.: a 10-year
follow-up examination. Hippocampus. 2006; 16:936–45. [PubMed: 17016801]
Sanders HI, Warrington DK. Memory for remote events in amnesic patients. Brain. 1971; 94:661–68.
[PubMed: 5132964]

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 25

Schnider A, Bassetti C, Schnider A, Gutbrod K, Ozdoba C. Very severe amnesia with acute onset after
isolated hippocampal damage due to systemic lupus erythematosus. J Neurol Neurosurg
Psychiatry. 1995; 59:644–46. [PubMed: 7500113]
NIH-PA Author Manuscript

Scoville WB, Milner B. Loss of recent memory after bilateral hippocampal lesions. J Neurol
Neurosurg Psychiatry. 1957; 20:11–21. [PubMed: 13406589]
Shrager Y, Gold JJ, Hopkins RO, Squire LR. Intact visual perception in memory-impaired patients
with medial temporal lobe lesions. J Neurosci. 2006; 26:2235–40. [PubMed: 16495450]
Shrager Y, Levy DA, Hopkins RO, Squire LR. Working memory and the organization of brain
systems. J Neurosci. 2008; 28:4818–22. [PubMed: 18448658]
Slotnick SD. “Remember” source memory ROCs indicate recollection is a continuous process.
Memory. 2010; 18:27–39. [PubMed: 19937493]
Slotnick SD, Dodson CS. Support for a continuous (single-process) model of recognition memory and
source memory. Mem Cogn. 2005; 33:151–70.
Squire, LR. Memory and Brain. New York: Oxford Univ. Press; 1987.
Squire LR. Lost forever or temporarily misplaced? The long debate about the nature of memory
impairment. Learn Mem. 2006; 13:522–29. [PubMed: 17015849]
Squire LR. The legacy of patient H.M. for neuroscience. Neuron. 2009; 61:6–9. [PubMed: 19146808]
Squire LR, Amaral DG, Press GA. Magnetic resonance imaging of the hippocampal formation and
mammillary nuclei distinguish medial temporal lobe and diencephalic amnesia. J Neurosci. 1990;
10:3106–17. [PubMed: 2118948]
Squire LR, Amaral DG, Zola-Morgan S, Kritchevsky M, Press GA. Description of brain injury in the
NIH-PA Author Manuscript

amnesic patient N.A. based on magnetic resonance imaging. Exp Neurol. 1989; 105:23–25.
[PubMed: 2744126]
Squire LR, Bayley PJ. The neuroscience of remote memory. Curr Opin Neurobiol. 2007; 17:185–96.
[PubMed: 17336513]
Squire LR, van der Horst AS, McDuff SGR, Frascino JC, Hopkins RO, Mauldin KN. Role of the
hippocampus in remembering the past and imagining the future. Proc Natl Acad Sci USA. 2010;
107:19044–48. [PubMed: 20956323]
Squire LR, Wixted JT, Clark RE. Recognition memory and the medial temporal lobe: a new
perspective. Nat Rev Neurosci. 2007; 8:872–83. [PubMed: 17948032]
Squire LR, Zola-Morgan S. Memory: brain systems and behavior. Trends Neurosci. 1988; 11:170–75.
[PubMed: 2469186]
Squire LR, Zola-Morgan S. The medial temporal lobe memory system. Science. 1991; 253:1380–86.
[PubMed: 1896849]
Steinvorth S, Levine B, Corkin S. Medial temporal lobe structures are needed to re-experience remote
autobiographical memories: evidence from H.M. and W.R. Neuropsychologia. 2005; 43:479–96.
[PubMed: 15716139]
Suzuki WA. Perception and the medial temporal lobe: evaluating the current evidence. Neuron. 2009;
61:657–66. [PubMed: 19285462]
NIH-PA Author Manuscript

Suzuki WA. Untangling memory from perception in the medial temporal lobe. Trends Cogn Sci. 2010;
1:195–200. [PubMed: 20227327]
Suzuki WA, Amaral DG. Topographic organization of the reciprocal connections between the monkey
entorhinal cortex and the perirhinal and parahippocampal cortices. J Neurosci. 1994; 14:1856–77.
[PubMed: 8126576]
Teng E, Squire LR. Memory for places learned long ago is intact after hippocampal damage. Nature.
1999; 400:675–77. [PubMed: 10458163]
Tulving E, Schacter DL. Priming and human memory systems. Science. 1990; 247:301–6. [PubMed:
2296719]
Victor, M.; Adams, RD.; Collins, GH., editors. The Wernicke-Korsakoff Syndrome and Related
Neurological Disorders due to Alcoholism and Malnutrition. Philadelphia: F.A. Davis; 1989.
Victor M, Agamanolis J. Amnesia due to lesions confined to the hippocampus: a clinical-pathological
study. J Cogn Neurosci. 1990; 2:246–57.

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 26

von Cramon DY, Hebel N, Schuri U. A contribution to the anatomical basis of thalamic amnesia.
Brain. 1985; 108:993–1008. [PubMed: 3935270]
Wais P, Wixted JT, Hopkins RO, Squire LR. The hippocampus supports both the recollection and the
NIH-PA Author Manuscript

familiarity components of recognition memory. Neuron. 2006; 49:459–68. [PubMed: 16446148]

Warrington EK, McCarthy RA. Categories of knowledge. Further fractionations and an attempted
integration. Brain. 1987; 110:1273–96. [PubMed: 3676701]
Warrington EK, Shallice T. Category specific semantic impairments. Brain. 1984; 107:829–53.
[PubMed: 6206910]
Wheeler ME, Petersen SE, Buckner RL. Memory’s echo: vivid remembering activates sensory-specific
cortex. Proc Natl Acad Sci USA. 2000; 97:11125–29. [PubMed: 11005879]
Whitlock JR, Sutherland RJ, Witter MP, Moser MB, Moser EI. Navigating from hippocampus to
parietal cortex. Proc Natl Acad Sci USA. 2008; 105:14755–62. [PubMed: 18812502]
Winocur G, Moscovitch M, Bontempi JB. Memory formation and long-term retention in humans and
animals: convergence towards a transformation account of hippocampal-neocortical interactions.
Neuropsychologia. 2010; 48:2339–56. [PubMed: 20430044]
Winslow, F. On Obscure Diseases of the Brain and Disorders of the Mind. London: John W. Davies;
1861.
Wixted JT. Dual-process theory and signal-detection theory of recognition memory. Psychol Rev.
2007; 114:152–76. [PubMed: 17227185]
Wixted JT, Mickes L. A continuous dual-process model of remember/know judgments. Psychol Rev.
2010; 117:1025–54. [PubMed: 20836613]
NIH-PA Author Manuscript

Wixted JT, Squire LR. Recall and recognition are equally impaired in patients with selective
hippocampal damage. Cogn Aff Behav Neurosci. 2004; 4:58–66.
Wixted JT, Squire LR. The role of the human hippocampus in familiarity-based recognition memory.
Behav Brain Res. 2010; 215:197–208. [PubMed: 20412819]
Wixted JT, Squire LR. The medial temporal lobe and the attributes of memory. Trends Cogn Sci. 2011
In press.
Woodruff CC, Johnson JD, Uncapher MR, Rugg MD. Content-specificity of the neural correlates of
recollection. Neuropsychologia. 2005; 43:1022–32. [PubMed: 15769488]
Yonelinas AP. Receiver-operating characteristics in recognition memory: evidence for a dual-process
model. J Exp Psychol Learn Mem Cogn. 1994; 20:1341–54. [PubMed: 7983467]
Yonelinas AP, Kroll NEA, Dobbins IG, Lazzara MM, Knight RT. Recollection and familiarity deficits
in amnesia: convergence of remember/know, process dissociation, and receiver operating
characteristic data. Neuropsychology. 1998; 12:1–17.
Yonelinas AP, Kroll NEA, Quamme JR, Lazzara MM, Sauve MJ, et al. Effects of extensive temporal
lobe damage or mild hypoxia on recollection and familiarity. Nat Neurosci. 2002; 5:1236–41.
[PubMed: 12379865]
Zola-Morgan S, Squire LR, Amaral DG. Human amnesia and the medial temporal region: enduring
memory impairment following a bilateral lesion limited to field CA1 of the hippocampus. J
NIH-PA Author Manuscript

Neurosci. 1986; 6:2950–67. [PubMed: 3760943]

Zola-Morgan S, Squire LR, Ramus SJ. Severity of memory impairment in monkeys as a function of
locus and extent of damage within the medial temporal lobe memory system. Hippocampus.
1994; 4:483–95. [PubMed: 7874239]

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 27
NIH-PA Author Manuscript
NIH-PA Author Manuscript

Figure 1.
Left column Magnetic resonance images arranged from rostral (a) to caudal (c) through the
temporal lobe of patient H.M. (in 1993 at age 67) and a 66-year-old healthy male (right).
The comparison brain illustrates the structures that appear to have been removed during
H.M.’s surgery in 1953. The lesion was bilaterally symmetrical, extending caudally 5.4 cm
on the left side and 5.1 cm on the right. The full caudal extent of abnormal tissue is not
NIH-PA Author Manuscript

illustrated. The damage included medial temporal polar cortex, most of the amygdaloid
complex, virtually all the entorhinal cortex, and approximately the rostral half of the
hippocampal region (dentate gyrus, hippocampus, and subicular complex). The perirhinal
cortex was substantially damaged except for its ventrocaudal aspect. The more posterior
parahippocampal cortex (areas TF and TH, not shown here) was largely intact. Adapted
from Corkin et al. (1997) with permission from the Society for Neuroscience.

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 28
NIH-PA Author Manuscript
NIH-PA Author Manuscript

Figure 2.
(a) Schematic view of the medial temporal lobe memory system for declarative memory,
which is composed of the hippocampus and the perirhinal, entorhinal, and parahippocampal
cortices. In addition to the connections shown here, there are also weak projections from the
perirhinal and parahippocampal cortices to the CA1-subiculum border. (b) Ventral view of a
human brain (upper left), monkey brain (upper right), and a lateral view of a rat brain (lower
center). The major cortical components of the medial temporal lobe are highlighted and
NIH-PA Author Manuscript

outlined. The hippocampus is not visible from the surface and in the human lies beneath the
cortex of the medial temporal lobe. Its anterior extent lies below the posterior entorhinal
(red ) and perirhinal ( purple) cortices, and the main body of the hippocampus lies beneath
the parahippocampal cortex. In the rat, the parahippocampal cortex is termed postrhinal
cortex. Abbreviations: EC, entorhinal cortex; PH, parahippocampal cortex (dark yellow);
Por, postrhinal cortex; PR, perirhinal cortex.

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 29
NIH-PA Author Manuscript

Figure 3.
Intact working memory and impaired long-term memory. (a) The number of trials needed to
succeed at each string length for patient H.M. and controls. H.M. could not succeed at
repeating back 7 digits even after 25 attempts with the same string. (b) The number of trials
needed to learn the locations of different numbers of objects for patient G.P. and controls.
G.P. could not reproduce the locations of four objects, even after 10 attempts with the same
display (panel a adapted from Drachman & Arbit 1966, with permission from the American
Medical Association, and panel b adapted from Jeneson et al. 2010).
NIH-PA Author Manuscript
NIH-PA Author Manuscript

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 30
NIH-PA Author Manuscript
NIH-PA Author Manuscript

Figure 4.
NIH-PA Author Manuscript

(a) Participants copied the Rey-Osterrieth figure illustrated in the small box in the upper left
and 10–15 min later, without forewarning, tried to reproduce it from memory. The
reproduction by a representative control is shown below the target figure. The left panel also
shows the reproduction by patient R.B., who had histologically identified lesions of the CA1
field of the hippocampus (Zola-Morgan et al. 1986). Patient E.P., who had large medial
temporal lobe lesions, did not recall copying a figure and declined to guess. The right
section shows reproductions by seven patients with circumscribed damage to the
hippocampus. Panels b and c show scores for the same seven patients (H) and 13 controls on
the autobiographical memory interview, childhood portion (Kopelman et al. 1989). These
findings suggest that patients who fail to produce any of the complex figure (like E.P.) or
who are deficient at producing either remote semantic memories (A, maximum score, 21) or
remote autobiographical events (B, maximum score, 9) will prove to have damage beyond
the hippocampus. Indeed, even E.P. with his large lesions limited mainly to the medial
temporal lobes, obtained maximal scores on these two tests (21/21 and 9/9).

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 31
NIH-PA Author Manuscript
NIH-PA Author Manuscript

Figure 5.
Individual recognition (a) and recall scores (b) for hippocampal patients (n = 7) and healthy
controls (n = 8) from Manns et al. (2003a). When the patient scores for recognition and
recall are converted to z-scores based on the mean and standard deviation of the
corresponding control scores, the recognition deficit (−1.59) is statistically indistinguishable
from the recall deficit (−1.81), p > 0.60. d′ = discriminability.
NIH-PA Author Manuscript

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.

 Squire and Wixted Page 32
NIH-PA Author Manuscript

Figure 6.
Symmetrical (a) and asymmetrical (b) receiver operating characteristic (ROC) plots with
hypothetical data shown as filled red circles. The axis of symmetry is the negative diagonal
(dashed gray line), and chance performance is indicated by the positive diagonal (solid blue
line). The symmetrical ROC (a) reflects relatively weak memory (the data fall close to the
positive diagonal), and the asymmetrical ROC (b) reflects stronger memory (the data fall
farther from the positive diagonal).
NIH-PA Author Manuscript
NIH-PA Author Manuscript

Annu Rev Neurosci. Author manuscript; available in PMC 2012 January 1.