Concept learning with one relevant dimension in the squirrel monkey

Abstract

Sixteen squirrel monkeys learned four discriminations in a procedure patterned on human concept learning (CL) , 1948) have used procedures such as sequences of 6-trial object discriminations. But in the typical human CL study, Sis given a different stimulus on each trial, and the whole array is specified by known relevant and irrelevant dimensions. The present study is an initial demonstration of CL in a convenient primate using procedures similar to those in human studies. MethodSs were .16 female squirrel monkeys maintained at 90% ad lib weight.The apparatus was a manually-operated WGT A with two Kluver-type pull-in carts replacing the single tray. In action, the carts were placed 7·in.apart, 14 in. from the animals, and the cord from each cart was tied to a ring just outside the bars ofS's compartment. A movable stimulus block covered the reward cup on each cart.Stimuli were mounted on the face of a 2 x 2 in. wood block, and represented all 81 combinations of size (large, medium, small); shape (triangle, circle, square); color of background (green, yellow, blue); and pattern within the figure (a cross, vertical stripes, crosshatching).In each discrimination problem, Ss were presented with a choice between two stimuli on each of 36 trials per day, in a non-correction procedure. Positive and negative stimuli were randomly paired, and correct side was randomized with a restriction of three consecutive placements on the right or left cart. A criterion of 15 out of 18 consecutive correct responses was used.Ss learned four discriminations in a series of nonreversal shifts. In each case there was one relevant and three irrelevant dimensions. Thefirstproblemalso served as a shaping task, in which the choices were made successively more complex by the addition of irrelevant dimensions. The positive cues in the four problems were Small, Circle, Stripes, and Green. A Latin square ensured that problems were encountered in different orders.