INDIVIDUAL DIFFERENCES IN RELATION LEARNING IN CAPUCHINS: EVIDENCE FOR HYPOTHESIS TESTING*

relations among objects is fundamental to analogical reasoning. The extent to which non-human primates can learn such relations has been investigated primarily in Matching-To-Sample studies. These studies suggest that members of a number of primate species can learn abstract relations. How does this learning take place? In this paper we ask how members of one species, tufted capuchins, learn the abstract relation ‘identity ’ in an Matching-To-Sample study carried out with computer touchscreens. We use a novel statistical analysis of runs of correct and incorrect responses to distinguish between explicit learning (based on hypothesis-testing and falsification) and implicit learning (based on the formation of positive associations between stimuli and rewards). We find that individual capuchins with the best learning performance in this task reliably respond in a way which is consistent with explicit hypothesis testing. One common form of learning occurs whenever participants are asked to make repeated choices between alternative stimuli which, when chosen, give different levels of reward. In such situations, participants will learn to distinguish between the presented stimuli and will learn to identify and select stimuli which return a higher level of reward, and to avoid selecting stimuli which return a lower level of reward. Researchers studying this form of learnin