|  Memory is a broad topic that
has its roots in both biology and psychology. Many of the important questions
about memory concern structure, organization,
and function and must be addressed at a relatively global, systems level
of analysis. Two long-standing and interrelated questions of this kind
have been of central interest. The first is whether there is more than
one kind of memory. The second concerns what brain structures and pathways
are important for memory. Scientific work directed at these questions has
introduced the terms declarative memory and nondeclarative
memory to neuroscience
and identified a number of parallels between the neural organization of
memory in humans and other animals.
The modern era of memory research began in 1957, when the profound effects
of bilateral medial temporal lobe resection on memory were described in
a patient who became known as H.M. This case became a landmark in the history
of memory research for two reasons. First, the medial aspect of the temporal
lobe was identified as an important region for memory function as his severe
memory impairment could be linked directly to the brain tissue that had
been removed. Second, comprehensive testing of this patient indicated that
memory impairment could occur on a background of otherwise normal cognitive
function. This observation showed that memory is to some extent an isolatable
function, largely separable from perception and general intellectual functions.
These discoveries led ultimately to the development of an animal model
of amnesia in the monkey and to the identification of the anatomical structures
of what is now known as the medial temporal lobe memory system. The important
structures are the hippocampus and the adjacent entorhinal, perirhinal,
and parahippocampal cortices. The success of this effort in monkeys led
to similar studies in rodents aimed at understanding the contribution of
the hippocampus and related structures to memory. At the same time, continuing
studies of H.M. and other memory-impaired patients made fundamental discoveries
about how memory functions are organized.
A key discovery from the work with patients, monkeys, and rodents was that
medial temporal lobe structures are essential for just one kind of memory,
which has come to be termed declarative memory. Other kinds of memory,
collectively termed nondeclarative memory, have been linked to other brain
systems.
The important insight was that memory is not a single entity but is composed
of several separate and parallel systems (Fig.
1). The major distinction
is between the capacity for conscious knowledge of facts and events (declarative
memory) and other nonconscious (nondeclarative) knowledge systems that
support the capacity for skill learning, habit formation, the phenomenon
of priming in which an earlier exposure to words or other material facilitates
a latter performance, and certain other ways of interacting with the world
where memory is experienced through performance rather than recollection.
Declarative memory is dependent on the integrity of the hippocampus and
anatomically related structures in the medial temporal lobe and diencephalon.
Declarative memory provides the ability to associate the various aspects
of a context that are present at a particular time and place, thereby creating
a memory of an episode. Declarative memory is also well suited for connecting
the pieces of information needed to acquire a new fact (e.g., the capitol
of North Dakota is Bismarck). It is sometimes pointed out that declarative
memory allows one to model the external world, and in that sense it is
either true or false.
Declarative memory is the kind of memory impaired in medial temporal lobe
amnesia. While work with amnesic patients has emphasized the notion of
conscious recollection, the concept of declarative memory is not defined
solely in terms of what amnesic patients can and cannot learn. Other characteristics
have been identified that have made it possible to extend the concept to
experimental animals. Declarative memory is fast, it is not always reliable
(i.e., forgetting and retrieval failure can occur), and it is flexible
in the sense that it is accessible to multiple response systems. It is
especially suited for one-trial learning and for forming and maintaining
an association between two arbitrarily different pieces of material, e.g.,
as in the case of conventional paired-associate learning in which a person
is asked to remember pairs of unrelated words.
A particularly good example of declarative memory involves the capacity
to identify a recently encountered item as familiar, a capacity termed
recognition memory. Recognition memory is impaired in amnesic patients,
and it is impaired in monkeys and rats following damage to the hippocampal
region. The finding that the hippocampal region is essential for normal
recognition memory is consistent with current ideas about the role of the
hippocampus in declarative memory and with the view that the hippocampus
is essential for acquiring information about relationships or connections
between stimuli. The recognition test asks whether an item that had been
presented recently now appears familiar. For recognition to be successful,
a link must be made at the time of learning between the to-be-remembered
stimulus and its context or between the stimulus and an organism’s
interaction with it. It is this associating and the ability to retain relational
information across time that many have supposed is at the heart of declarative
memory and in turn is the function of the hippocampal region in both humans
and nonhuman animals.
Recently, there has been interest in the possibility that some aspect of
memory function might be associated specifically and uniquely with the
hippocampus itself and, correspondingly, that some aspect of declarative
memory might be independent of the hippocampus (and be supported instead
by adjacent medial temporal cortex). These ideas are currently active topics
of experimental work.
Whereas declarative memory is a brain-systems construct, tied to the brain
structures and connections damaged in amnesia, nondeclarative memory refers
to a heterogeneous collection of several kinds of memory that in turn depend
on distinct brain systems (Fig.
1). Thus classical conditioning of skeletal
musculature depends on the cerebellum, conditioning of emotional responses
depends on the amygdala, and habit learning (win-stay, lose-shift responding)
depends on the neostriatum. The amygdala can also modulate the strength
of both declarative and nondeclarative forms of memory. Finally, perceptual
priming likely depends on changes in early-stage cortical areas involved
in processing the stimuli that are primed.
Nondeclarative memory is expressed through performance. Unlike declarative
memory, it is neither true nor false. Non-declarative memory refers to
the variety of ways in which experience can lead to altered dispositions,
preferences, and judgments without affording any necessary conscious memory
content. Performance changes as the result of experience and in this sense
deserves the term memory, but performance changes without an accompanying
sense that memory is being consulted. The organism simply behaves differently
than it did previously. In many cases, performance changes slowly, as when
one learns gradually about the causal structure of the environment and
acquires new procedures for interacting with the world (in the case of
conditioning, skill learning, or habit learning). Sometimes performance
can change rapidly (in the case of fear conditioning or conditioned taste
aversion). In the latter cases, the possibility of rapid change may be
built into evolutionarily important systems that are specialized to process
or associate particular kinds of information.
Eyeblink classical conditioning has provided a useful paradigm for exploring
the distinction between declarative and nondeclarative forms of memory
in humans and other animals. In eyeblink classical conditioning, a conditioned
stimulus (CS; typically a tone) is presented just prior to an unconditioned
stimulus (US; typically a puff of air to the eye). After repeated pairings
of the CS and US, subjects begin to blink in response to the CS. The eyeblink
response is a learned or conditioned response. The two most commonly studied
forms of eyeblink classical conditioning are delay conditioning and trace
conditioning. In delay conditioning, the CS is presented and remains on
until the US is presented. The two stimuli then overlap and coterminate.
In trace conditioning, an empty or “trace” interval separates
the CS and the US.
Work with rabbits first demonstrated a clear distinction between delay
and trace conditioning. The acquisition and retention of delay conditioning
require the cerebellum and associated brainstem structures. No tissue above
the level of the midbrain, including the hippocampus, is required. Thus
delay conditioning appears to be an example of nondeclarative memory. Trace
conditioning is fundamentally different. Like delay conditioning, successful
trace conditioning requires the cerebellum but trace conditioning differs
from delay conditioning in that it also requires the hippocampus and specific
regions of neocortex. Trace conditioning appears to require the hippocampus
because declarative knowledge of the CS–US relationship must build
up and be maintained across many trials.
This link between trace conditioning and declarative knowledge was first
demonstrated by showing that awareness of the stimulus contingencies is
critical for differential trace conditioning. In differential conditioning,
the CS+ (e.g., a tone) is followed by the US, and the CS– (e.g.,
a static noise) is presented alone. Successful differential conditioning
occurs when more conditioned responses are elicited by the CS+ than by
the CS–. Because there are several relationships among the stimuli
about which a participant can become aware, a variety of questions can
be asked about the stimulus contingencies, and a participant’s answers
to these questions can be related to conditioning performance. The finding
of interest in the case of trace conditioning was that only individuals
who developed awareness of the CS–US relationship conditioned successfully.
Individuals who did not develop knowledge of the CS–US relationship
did not acquire trace conditioning. Studies of amnesic patients with damage
that included the hippocampus have also been informative. Amnesic patients
failed to acquire differential trace conditioning and also failed to become
aware of the stimulus contingencies. The same patients were subsequently
able to acquire differential delay conditioning as readily as intact subjects.
These results indicate that trace conditioning requires an additional level
of processing that is not required for delay conditioning. Specifically,
trace conditioning (but not delay conditioning) requires the participation
of the hippocampus and presumably its interaction with neocortex. Awareness
may emerge during trace conditioning because awareness is a typical feature
of hippocampus-dependent learning. In this sense, awareness is a reliable
indicator of a brain state (a state of interaction between the hippocampus
and neocortex) that is essential for forming and storing declarative memory.
Finally, the notion of multiple memory systems provides a way to think
about the phenomenon of infantile amnesia (i.e., the relative unavailability
of memories for events that occur before the third year of life). There
is good evidence that the declarative memory system is functional, to at
least some degree, in early life. Accordingly, the absence or slow development
of this memory system cannot account for the phenomenon of infantile amnesia.
If declarative memory is available to infants, then what accounts for infantile
amnesia? One clue comes from the finding in monkeys that suggests that
the capacity for forming and maintaining declarative memories may be limited,
not by the maturation of the structures essential for declarative memory
but rather by gradual maturation of the neocortical areas that are served
by these structures and that are believed to be the repositories of long-term,
permanent memory. This perspective is an appealing one because it provides
a point of contact between a neurological account of infantile amnesia
and accounts grounded in cognitive psychology that emphasize the gradual
maturation of cognition, the emergence of skills and strategies for organizing
information into knowledge systems, the development of language, and the
growth of individual identity.
The cognitive and neuroanatomical work described here is a first step in
analyzing how the brain has organized its memory functions. With respect
to declarative memory, neuroscience is approaching a time when it will
be possible to study representations directly in neocortex with single-cell
recording, to observe directly the development of neural plasticity, and
to determine how the medial temporal lobe interacts with neocortex during
learning, the consolidation of memories, and their retrieval. With respect
to nondeclarative memory, it has been possible to identify particular brain
systems that are essential for particular kinds of memory. The next step
will be to determine whether these systems are essential for the acquisition,
storage, or expression of memory and to identify exactly where the synaptic
changes occur that support each kind of memory. Cellular and molecular
studies of experimental animals will be particularly useful in this work;
some of the work in this area will be examined over the next several months
in this column.
top of page
|
