Multiple schedules: the effects of temporal factors on behaviour

Abstract

Domestic hens served in three experiments examining the effects of varying the temporal proximity of component schedules on behaviour on multiple variable-interval variable-interval schedules. In all experiments response rates on each component schedule and in successive subintervals of the components were determined. Data were analysed to obtain estimates of the parameters of the Generalised Matching Law for total responding and for responding in successive subintervals of components. Four hens served in the first experiment which examined the effects of reducing component duration while reinforcement rates in the component schedules were frequently reversed. The results indicated that values of the parameter α of the Generalised Matching Law generally increased as component duration was reduced. Changes in response rates contributing to the decrease in α values varied across hens. Local positive and local negative contrast were present in all hens’ data in some conditions but not in others. The conditions in which these phenomena occurred varied across hens. For three hens values of α declined across successive subintervals of components at all component durations. This effect was also present at the longer component durations in the fourth hen’s data but, for this hen’s data, α values increased over successive subintervals when component duration was reduced. The second experiment examined the effect of increasing and then decreasing the duration of an intercomponent blackout. Six hens served in this experiment. Blackout durations examined were 1 s, 10 s, 30 s, 60 s, 120 s, 180 s and 297 s. At each blackout duration the reinforcement rates in component schedules were varied. Response rates on both components of the multiple schedules tended to increase as the blackout duration was increased. Values of the parameter α decreased as blackout duration was increased and for five birds became negative. Changes in α values and response rates across successive subintervals similar to those seen in the first experiment were present when the blackout duration was 1 s. As blackout duration was increased these changes were attenuated. Values of α and response rates for overall responding observed when blackout duration was 1 s and 30 s were not replicated when blackout duration was then decreased. The third experiment examined the effects of blackout duration using multiple-multiple schedules. In the first phase of this experiment the behaviour of six hens on the multiple schedules of the multiple-multiple schedules was compared with behaviour of the same hens on each multiple schedule when one schedule was presented in each condition. The results of this phase suggested that behaviour on each multiple schedule of the multiple-multiple was qualitatively similar to behaviour on the same multiple schedules when they were presented singly. In the second phase of this experiment three of the hens employed in the first phase were exposed to a series of conditions where the duration of an intercomponent blackout was varied. The blackout durations examined in the order they were presented were 1 s, 30 s, 60 s, 0 s, 120 s. For two of the hens’ data values of α for overall responding decreased as a function of blackout duration. For one of these hen’s data α values became negative. For the third hen’s data α values decreased when the blackout duration was increased to 120 s. Absolute response rates contributing to these α values varied across hens. Within component changes in response rates also varied across hens. Values of α decreased across successive subintervals for all hens’ data when no blackout separated components but this effect was present in only one hen’s data in other conditions. In general changes in response rates and α values as a function of component or blackout duration were not explicable in terms of the changes in behaviour across successive subintervals of components. These results were examined in terms of several theories previously proposed to explain multiple-schedule performance. It was concluded that none of these could easily accommodate the results obtained in the experiments presented here.