The role of timing in reinforcement schedule performance

Abstract

Reinforcement schedules produce distinctive and reliable temporal patterns of behavior. Temporal discrimination is known to be an important ingredient in these patterns. We sketch a simple dynamic model for the discrimination of short time intervals, linear waiting, in terms of four assumptions: that the time before the onset of the reinforced response, waiting time, is determined by the food delay in the just-preceding interfood interval; that waiting is triggered by the most recent time marker; that different delays signaled by the same time marker are averaged; and that the effectiveness of a given stimulus as a time marker is limited by memory constraints. We show that recursive application of linear waiting can generate contiguity learning and many molecular patterns of behavior on interval schedules, such as the fixed-interval “scallop” and interresponse times on variable-interval schedules, as well as molar properties such as the hyperbolic relation between response and reinforcement rates on variable-interval schedules. Linear waiting also implies the high response rates on ratio schedules and their instability. Linear waiting does not seem able to account for differences in response rates between ratio and interval schedules equated for reinforcement rate, molar response functions on ratio schedules, and some features of responding on cyclic schedules. These failures highlight our limited understanding of the role of memory in timing and hint at additional mechanisms.