record of the experience to the storage bin; consequently the number of records
expands over time. Similarly, the number of books stored in a library increases
over time. The records actually stored may be more accurately described as
interpretations of experiences. (2) Remembering involves searching through a
network of memory locations for some particular record, as one might search
for a particular book in a library. Once found, the target memory record is
‘‘read’’ or in some sense reexperienced. The search need not be done haphaz-
ardly, since the memory records may be connected or organized in such a way
as to improve the efficiency of the search. Libraries, for example, organize
books by subject matter in order to make finding the books easier. (3) Forget-
ting is primarily due to search failure caused by the interfering effect of the
presence of lots of memory records, just as in a library the huge number of
books stored there makes finding any one book difficult. Some versions of
the record-keeping theory claim that no memory record is ever really lost. All
records of past experiences are potentially recoverable.
The metaphor of record keeping is compelling for several reasons. The word
memoryimplies a preserving of the past; we sometimes have vivid and accurate
recollections of the past, and nearly all of the artificial memory systems we
know about, such as libraries, videotapes, and computers, are record-keeping
systems designed to preserve information. Indeed, it is difficult to imagine any
other basis for memory. Nevertheless, I will argue in this and other chapters
that the record-keeping approach to human memory is a misleading one
(Schacter, 1996). Human memory works according to a different set of principles.
An alternative to the record-keeping approach may be called aconstructionist
approach to memory. We know that knowledge from sources outside of the
stimulus stream affects the perception of the stimulus. A similar notion plays a
role in a constructionist account of memory.
The constructionist account begins with the important insight that human
memory is not designed primarily to preserve the past, but to anticipate the
future (Morris, 1988). Most constructionist theories are characterized by these
principles: (1) Each new experience causes changes in the various cognitive
systems that perceive, interpret, respond emotionally, and act on the environ-
ment, but no record-by-record account of the experiences that gave rise to those
changes is stored anywhere. That is, memory reflects how the cognitive sys-
tems have adapted to the environment. Usually this adaptation takes the form
of noting regularities in experiences and basing future responses on these reg-
ularities. The cognitive systems are also sensitive to unexpected exceptions
to the regularities ordinarily observed. (2) Recollection of the past involves a
reconstruction of past experiences based on information in the current environ-
ment and on the way cognitive processing is currently accomplished. Remem-
bering is a process more akin to fantasizing or planning for the future than
searching for and then ‘‘reading’’ memory records, or in any sense reexperi-
encing the past. The past does not force itself on a passive individual; instead,
the individual actively creates some plausible account of her or his past. (3)
Forgetting is not due to the presence of other memory records but to the con-
tinuous adaptive changes made to the various cognitive systems in response to
events.
312 R. Kim Guenther