Persistence and relapse of reinforced behavioral variability

Abstract

The present study examined persistence and relapse of reinforced behavioral variability in pigeons. Pigeons emitted four‐response sequences across two keys. Sequences produced food according to a lag schedule, in which a response sequence was followed by food if it differed from a certain number of previous sequences. In Experiment 1, food was delivered for sequences that satisfied a lag schedule in both components of a multiple schedule. When reinforcement was removed for one component (i.e., extinction), levels of behavioral variability decreased for only that component. In Experiment 2, food was delivered for sequences satisfying a lag schedule in one component of a multiple schedule. In the other component, food was delivered at the same rate, but without the lag variability requirement (i.e., yoked). Following extinction, levels of behavioral variability returned to baseline for both components after response‐independent food delivery (i.e., reinstatement). In Experiment 3, one group of pigeons responded on a lag variability schedule, and the other group responded on a lag repetition schedule. For both groups, levels of behavioral variability increased when alternative reinforcement was suspended (i.e., resurgence). In each experiment, we observed some evidence for extinction‐induced response variability and for variability as an operant dimension of behavior.

An alternative approach to relapse analysis: Using Monte Carlo methods and proportional rates of response

“…The results of the Monte Carlo are generally consistent with the original findings of Galizio et al, . In Experiment 1, pigeons responded on a two‐component multiple schedule (Vary / Vary Ext).…”

“…Figure displays the histograms of the simulated means for each comparison. The histograms are similar to those described in the preceding section on the reanalysis of Odum and Shahan () and Galizio et al (). The distributions of simulated means appear normal as they match the expected normal distribution that was imposed on the figure.…”

“…** p < .01, *** p < .001 for the Monte Carlo results. Effect size is D. Bolded and italicized values do not match the results as reported in Galizio et al, .…”

“…Galizio et al () used several different procedures to investigate relapse of operant variability. Variability can be maintained by reinforcement and may be an operant dimension of behavior (see Neuringer, , for a review).…”

“…Monte Carlo methods may represent an improved analytic technique, because these methods require no prior assumptions, permit smaller sample sizes, and can be tailored to account for all of the data from an experiment instead of some limited set. In the present study, we conducted reanalyses of three studies of relapse (Berry, Sweeney, & Odum, ; Galizio et al, ; Odum & Shahan, ) using Monte Carlo techniques to determine if relapse occurred and if there were differences in rate of response based on relevant independent variables (such as group membership or schedule of reinforcement). These reanalyses supported the previous findings.…”