| |
Abstract:
Computational models of concept formation have been derived
from critical but limited information generated by studies of
brain-damaged individuals. We examined the importance of different
factors-familiarity, category, item distinctiveness, age of
acquisition, concreteness, and word frequency- in the accuracy and
RTs of Alzheimer's disease (AD) subjects in semantic tasks. In a
picture naming task, the subjects named objects, matched for
familiarity, but encountered either in childhood (e.g., animals) or
in adulthood (i.e., appliances). The ADs' accuracy was strongly
affected by an item's familiarity and distinctiveness but not by
age of acquisition. In a task of semantic judgement the relative
effects of concreteness, word association, and word frequency were
measured. Subjects were presented with pairs of words (e.g.,
monarch-king, and monarch-host) and asked to indicate which words
were closest in meaning. ADs were disproportionately slower and
less accurate in responding to abstract concepts. Regression
analyses of the factors affecting RTs to items revealed that the
same factors partly predicted ithe RTs of normal and ADs, the
strongest factor being word association strength, with concreteness
making a lesser contribution. Concreteness, however, was the best
predictor of accuracy in the ADs. Word frequency contributed little
to predicting RT or accuracy. Different factors, then, influence
the retrieval and recognition of concepts and need to be accounted
for in comprehensive models of the semantic structure.
|
