Natural language categories are considered as concepts
which can be described in terms of a prototype, wit h some
category i terns more "typical" of their categories than others .
One hypothesis tested was that typicality effects on free
recall performance were due to atypical items not being
encoded by their category labels. An alternative hypothesis
argues that the structural base behind typicality effects
is "family resemblances" (feature overlap). In the task,
subjects were given either a list of typical or atypical
items for three study/recall trials. Further, one half of
the subjects in each materials condition received category
cues at input to test the differential encoding hypothesis.
The results showed large typicality effects in the no cue
condition. However, there were small typicality effects
in the cue condition. These results were interpreted as
indicating that typicality effects on a free recall performance
task are largely eliminated when category coding is relatively
complete. Thus, the feature overlap hypothesis cannot be
accepted.