ABSTRACT
Because primate researchers cannot afford the luxury of using a new and “naive” group of subjects for each new experiment, it is not surprising that investigators of cognitive skills in primates have had a continuing interest in transfer of learning. One striking transfer phenomenon is learning set formation. In a learning set experiment, the subject is trained on a series of many different discriminations, each presented for only a few trials. The stimuli typically are pairs of objects that differ from each other in a variety of dimensions (e.g., color, form, size, texture). These objects usually are presented on the form board of a Wisconsin General Test Apparatus. For simultaneous object discriminations, the displacement of one object of each pair is always rewarded and the displacement of the other is consistently not rewarded (see Figure 2.1). All other cues, including positional cues, are irrelevant. After a few trials on one discrimination problem have been given, a new pair of objects is presented as a new discrimination problem. A special feature of learning set experiments is that the particular stimulus properties that are associated with reward on one problem (e.g., blue, large) are as likely to be associated with nonreward as with reward on a new problem. Indeed, performance on Trial 1 of each new problem stays at chance as long as the experiment is properly controlled.
