This study assessed the performance of 20 persons with autism in an identity matching-to-sample task, using two procedures with different stimuli arrangements and different response requirements. The typical matching procedure presented one sample stimulus and three comparison stimuli; the adapted matching displayed three sample stimuli and three comparison stimuli simultaneously. Three consecutive sets of trials were conducted: a 10-trial set of adapted matching, a 10-trial set of typical matching, and a 20-trial set intermixing both trial types in an unsystematic order. The average score of correct performance was significantly higher under the adapted matching than under the typical matching, but the inter-individual variability was large and the adapted arrangement favored mainly the performance of participants with lower scores. The participants' previous history in learning may have played a role in the way they solved the present task, but further investigation, with experimental controls additional will be necessary to evaluate this possibility.
Conditional relations; identity matching-to-sample; concept formation; autism
