entree for any given dinner. Similarly, people might not be able to remember
very well such things as what color shirt they wore on any given night out on
the town or exactly where in the lot they parked their car on any given excur-
sion to the beach. But it should be easy for people to remember theinvariantsor
enduring patterns of events, such as always drinking a beverage with dinner,
always wearing a casual shirt to the night club, or always parking in the cheaper
lotatthebeach.
Record-keeping theories, like constructionist theories, would also predict that
accurate memory for any one event is likely to decline as more records are
stored (see, for example, Anderson, 1976). But without embellishment, record-
keeping theories have no ready way to explain why memory should be strong
for the enduring patterns of experience. At the very least, a record-keeping
theory would have to postulate the existence of another cognitive mechanism
designed only to extract patterns from experiences. That is, it is not a natural
consequence of keeping records that enduring patterns are extracted from those
records. The advantage of constructionist theory is that it postulates that the
creation of memories and the extraction of patterns from experience are accom-
plished by the same mechanism; namely, the altering of connection strengths
among the concepts and ideas that constitute knowledge.
Evidence for the Constructionist Account of Retention
Empirical Evidence That Memory Preserves Patterns but Not Details of Experiences
A nice example of the principle that memory preserves the enduring patterns
and themes but not the changing elements in events comes from the testimony
of John Dean, a key figure in the Watergate scandal of the early 1970s (Neisser,
1981). John Dean had been President Nixon’s attorney and testified against him
in a highly publicized Senate hearing on the Watergate break-in. Dean tried to
recollect the details of meetings, including who participated, what was said,
and when the meeting took place. Dean’s memory seemed quite remarkable
(and damaging to Nixon); he was able to supply many details that other mem-
bers of Nixon’s administration claimed to be unable to recall.
It was discovered later that all meetings in the Oval Office had been tape
recorded, so that many of Dean’s recollections could be compared with the
actualtranscriptsofthosemeetings.ItturnsoutthatDeanwasofteninaccurate
about details of the meetings but was accurate in his recollection of the general
tenor of a number of the meetings; namely, that Nixon and other high-ranking
members of his administration knew about the Watergate break-in and tried to
cover it up. What distinguished Dean’s testimony from that of the others was
that Dean decided to tell the truth about the coverup. Dean’s memory was not
especially accurate about those elements that were always changing, like the
details of conversations or which participants were at particular meetings, but
hismemorywasquiteaccurateaboutthesortsoftopicsandissuesthat
endured across many meetings.
Many memory experiments also make the point that our memories permit
easier recall of enduring patterns than of details of specific experiences (e.g.,
Bartlett, 1932; Bransford & Franks, 1971; Thorndyke & Hayes-Roth, 1979). For
example, participants in a weekly seminar on math were asked to recall the
320 R. Kim Guenther