Temporal Constraint on Choice

Abstract

This work proposes a concurrent analysis of multiple schedule performance which emphasizes concurrent interactions during multiple schedule components as determinants of absolute-rate behavioral contrast and response-ratio sensitivity and, bias. Specifically, this analysis assumes no direct interaction of multiple schedule components in determining component performances. Indeed, it assumes that the allocation of behavior between responding on the schedule and engaging in other activities during a component depends only on the reinforcers for those behaviors in that component, and is independent of reinforcers obtained in alternated components. In a series of experiments using multiple-concurrent schedules of reinforcement, an experimental analog of ‘extraneous’ reinforcement was arranged on one key, and varied multiple-schedule components were arranged on another key. In Experiment l, multiple schedule sensitivity increased as a function of the absolute rates of analog extraneous reinforcers in components, and in Experiment 2,multiple schedule bias was an inverse function of the ratio of rates of extraneous reinforcers in components. Thus multiple schedule response ratios were sensitive to obtained rates of extraneous reinforcers, a relation which received additional confirmation in changes in ratios of response rates on the second (analog extraneous reinforcement) key, where reinforcer ratios were constant, which were an inverse function of multiple schedule reinforcer ratios. In Experiments 3 and 4, conditions were arranged where the total reinforcers (multiple schedule plus analog extraneous reinforcers) in both components were constant across changes in multiple schedule reinforcer ratios. Multiple schedule sensitivity equalled the concurrent sensitivity observed during components, with bias in Experiment 4 which favored a component which consistently had. reinforcer totals which were low relative to those in the other component. Quantitative predictions of multiple schedule performances, made from the present analysis, closely matched obtained performances in each experiment. In Experiment 5, conditions were similar to those in Experiment 1 and the results provided a partial replication of the dependence of sensitivity on absolute rates of extraneous reinforcers in components. All of these experiments provided evidence supporting the assumption that response allocation in a component does not depend on reinforcement in the alternated component -- that successive components do not interact. Experiments 3, 4 and 5 examined this question directly by maintaining a constant relative reinforcer rate across keys in one component. The allocation of responses and time between keys in the constant component was invariant with respect to changes in reinforcer ratios inthe alternated component in all three experiments. In Experiment 5, changes in reinforcer totals in the varied component were correlated with.changes in varied-component reinforcer ratios. Response and time allocation between keys in the constant component was invariant with respect to changes in varied-component reinforcer ratios (and totals) but total. responses in the constant component were an inverse function of varied-component. reinforcement. These results are consistent with the assumption that successive components do not interact, and provide indirect support for an alternative account of the apparent successive dependencies (behavioral contrast) observed in multiple schedules.