Armando Machado - US grants

DESCRIPTION (adapted from applicant's abstract): How animals learn to choose among alternatives, and how they learn about temporal cues in their environment, are enduring problems for any theory of learning. This project suggests an answer to each question in the form of a dynamic model, and proposes a set of experiments to test the models. The model for choice assumes that in recurrent choice situations, three factors determine how preference for the better alternative is acquired: the overall reward rate, the discriminability of the local reward rates, and the degree of response/reward feedback. In addition, the model predicts that when animals develop partial preferences and show local response patterns, the asymptotic strength of these patterns is also a function of the preceding factors. Five experiments will use pigeons to test the model's predictions concerning the acquisition of molar preference and the molecular organization of behavior. The timing model assumes a serial activation of behavioral states during the interreward interval, a learning process that couples the states with the instrumental response, and a rule mapping the states onto response rate or probability. Four experiments with pigeons will test the model's description of the learning process in three temporal tasks, fixed -- interval schedules, the bisection of temporal intervals, and a psychometric procedure. Taken together, the experiments will help to answer a set of fundamental questions about the processes of choice and timing: How are preferences acquired? How are the molar and the molecular structures of choice related? What do animals learn when rewards are time based? How do learning processes affect temporal judgements? A common feature of the work proposed here is the continual emphasis on the dynamics of behavior and learning.