Abstract # 2269 Poster # 147:

Scheduled for Friday, June 22, 2007 05:00 PM-07:00 PM: Session 14 (South Main Hall) Poster Presentation

A Capuchin Monkey Masters Mazes

T. Pickering, J. Pan, B. von Ammon, E. H. Kennedy and D. Fragaszy
University of Georgia, Athens, GA 30602, USA
     We examined how a capuchin monkey (Cebus apella) improved in navigating two-dimensional digital mazes. The 192 mazes contained one to five binary choices including "non-obvious" choices (where the incorrect path led toward the goal). On his first encounter with the mazes in a random order of difficulty, the monkey made errors (wrong turns) at 60% of choice points. Subsequently, he completed sets of all 192 mazes in replicates. Each replicate was followed by a probe series of twenty-four mazes composed of mazes drawn from the experimental set of 192. Errors on the probe series declined over the first four replicates from 42% to 26%, and then fluctuated from 19% through 23% to the last, tenth, replicate [Kolmogorov-Smirnov One-Sample test, a=0.05]. At this point, three new sets of twenty-four mazes were given. Each set contained eight familiar mazes, mirror images of eight familiar mazes, and eight novel mazes. The monkey performed as well on these final probe series (averaging 20% errors) as on the previous probes with the familiar series. Moreover, he performed equally well on the novel mazes as on the mirror image and familiar mazes. The capuchin learned to attend to the most relevant spatial property of each choice point (continuation), and to alter behavior flexibly in accord with continuation. Learning to manage behavior in this way supports flexible problem-solving.