Oddity-from-Sample Abstract-Concept Learning by Pigeons

Abstract

The ability to determine whether or not objects are the same has been established as an ability essential to human cognition (James, 1890). Concept formation, such as that of sameness, has been shown to rely on the number of training exemplars for humans and nonhumans. For example, by systematically increasing the number of training items pigeons acquired, performance on completely novel items increased. This shift indicates an early reliance on stimulus features to the relationship between the stimuli. This function of training set-size has been previously shown in same/different (Katz & Wright, 2006) and matching-to-sample (Bodily, Katz, & Wright, 2008) conditional discrimination tasks, but it has yet to be shown in an oddity-from-sample task. Pigeons trained with a set size of 3 cartoon items were required to respond to the sample item (FR 10) and select which of the two comparison items differed from the sample item. Once the pigeons demonstrated consistent performance with this task (85% accuracy), novel items were presented systematically in a series of transfer sessions to test for abstract-concept learning. During these transfer sessions, 12 of the 96 trials presented trial-unique pairings. The set-size was then systematically doubled (8 times), increasing the set-size from 3 to 768 items. With an increase in set-size, performance on novel items became equivalent with that of trained items, suggesting that pigeons use the same strategy across these sessions. This transfer performance suggests the first evidence of the oddity concept formation in the pigeon. The function relationship of set size was different than that of the identity concept learned during matching-to-sample, indicating different factors that influence how pigeons learn these two abstract concepts.