Learning variable and stereotypical sequences of responses: Some data and a new model

Abstract

In a left-right (L-R) choice situation, pigeons were exposed to a frequency-dependent schedule that selectively reinforced either the momentarily least-frequent pairs of responses (Experiment 1), or the least-frequent triplets of responses (Experiment 2). In Experiment 1, the birds developed various stereotypical response patterns including double alternation (i.e. RRLLRRLL…), the stable pattern that satisfied the schedule constraint. In Experiment 2, three out of four birds developed random-like behavior. A linear operator reinforcement model, in the tradition of Bush and Mosteller, accounted for the qualitative features of pigeons' behavior in frequency-dependent schedules: strict alternation (i.e. RLRL…) when the schedule selectively reinforces the least-frequent individual response, double alternation when the least-frequent pairs are selectively reinforced, random-like behavior when the least-frequent triplets are selectively reinforced, and various stereotypical response patterns apparently not related to the target of selection.