The learning of response patterns in choice situations

This study presents a theory by which to understand how pigeons learn respo nse patterns in simple choice situations. The theory assumes that, in a cho ice situation, patterns of responses compete for the final common path; tha t the competition is governed by two variables, the overall reinforcement p robability obtained by emitting the patterns, T, and the differences in rei nforcement probabilities among the patterns, D; and that the ratio D/T dete rmines the final strength of specific response patterns. To test these pred ictions, three experiments were run in which pigeons were more likely to re ceive food when they pecked the momentarily least-preferred of three respon se keys. On the basis of previous research, it was predicted that the birds would be indifferent among the keys (molar aspect) and would also acquire a response pattern that consisted of pecking each key once during three con secutive trials (molecular aspect). The present theory went further and pre dicted that the strength of that pattern would increase with the ratio D/T. In the first two experiments, D was manipulated while T remained constant, and in the third, T was manipulated while D remained constant. The results agreed with the theory, for the strength of the response pattern increased with D and decreased with T, whereas overall choice proportions were alway s close to the matching equilibrium.