110
informed about free recall clusters and were asked to circle
“he recall clusters in their own recall lists. Clusters of
only one member were allowed. Subjects were then asked to
identity both the best (i.e. the most representative or
typical) and the worst exemplars in each recall list.
Finally, subjects were asked to describe, if they could, the
strategies they used to recall exemplars of what they owned
and what they did not own.
Results: First, all subjects were able to list
exemplars of things they did not own. The mean number of
items listed was 11.46 (SD = 4.99), as compared to 18.93 (SD
= 7.8) for things owned. Since the pilot study did not
balance the order of the categories across the subjects, it
was not possible to determine whe ther the smaller output for
not owned things was due to it being more difficult to
generate exemplars in that category or due to a general
fatigue effect of recalling fewer items on each successive
category. Second, the mean number of recall clusters in the
first 10 items was 5.2 (SD = 1.76). Therefore, it appears
that 10-item recall lists are not unrepresentative because
they include only one recall cluster. Ten-item recall lists
also should not be unrepresentative because their exemplars
are atypical, since the best exemplars tended to be in the
top half of the recall lists, in the 4th or 5th positions,
and the worst exemplars tended to be in the bottom half. of