| |
Abstract:
Restricted or token-specific semantic fields, expressed as
stimulus overselectivity, are often believed to be typical of
low-functioning autism. We present evidence that semantic fields in
these individuals may be more abstract, and the impression
otherwise may have been an artifact of the methods typically used
to test conceptual and categorical knowledge. A 12 year old, low
functioning and nonverbal autistic male was given an auditory
word-to-picture matching task. The distracter photographs were
either both visually and semantically related, semantically but not
visually related, visually but not semantically related, or
unrelated to the correct target. 300 target items were assessed
across 1500 trials. He made errors on 35% of the trials.
Distracters that shared both visual and semantic features were the
most frequent errors (45% of errors), while distracters with
semantic features were 29% of the total errors. Errors were rarely
made for items that had no features in common with the correct
answer (7%). He generalized to 56 items that had never been
explicitly trained. He was significantly better with natural kind
categories, but his comparable performance on specific non-natural
categories suggests that motivation and training influenced
performance. His errors indicate that his semantic fields are
perhaps not as narrowly focused as thought. This methodology may be
helpful for further investigations of semantic representations in
autism.
|
