Abstract # 5940 Event # 40:

Scheduled for Saturday, September 13, 2014 04:45 PM-05:00 PM: Session 10 (Henry Oliver) Oral Presentation


L. A. Heimbauer1, T. Qian2, R. N. Aslin2 and D. J. Weiss1
1The Pennsylvania State University, The Comparative Communication Laboratory, 107 Moore Building, University Park, PA 16802, USA, 2University of Rochester
     To effectively track the statistical regularities of the environment, learners must distinguish when a single underlying causal structure rather than multiple underlying structures generates the observed statistics. That is, when encountering variance, learners must decide whether to expand the currently assumed model or infer that the input emanates from a new causal structure. In a series of implicit learning experiments modeled after research with humans (Qian, Jaeger, & Aslin, in review), we examined whether nonhuman primates are capable of inferring and retaining new structures to the same extent as human learners. Using a symmetrical-response serial reaction time task, we tested four rhesus macaques for their response to change in underlying statistical probabilistic structures. Structures were produced using four stimuli locations—one with a 70% occurrence rate and the others with a 10% rate. Results of Experiment 1 demonstrated four monkeys learned the underlying probability schedule of one structure (t-tests, p < 0.01). In Experiment 2, there were multiple alternating structures (i.e., the probabilities in individual locations changed). Two of the four monkeys succeeded in this task while the other two monkeys needed a correlated contextual cue (screen background color change) to learn. Experiments with four variable structures, without and with contextual cues, are currently being conducted. Findings shed light on how nonhuman primates track statistics in complex environments when variance more closely approximates real-world input.