Effects of training different levels of operant variability in rats

If high levels of variability are initially trained, responding remains variable even if the criterion for variability is eliminated. This effect, reported in college students, could have resulted from rule-governed behavior. To extend the generality of this finding, this experiment explored the effects of increasing or decreasing the criteria for reinforcing variability in rat subjects. The variability of a sequence of 3 responses distributed on five levers was reinforced using a Lag n procedure. For one group of rats (0-50) the criterion in successive conditions was Lag 0, 2, 10, 25, and 50. For the other group (50-0) the criteria were presented in decreasing order. For both groups, variability, measured as U values was higher with higher Lag values. For Group 50 - 0, U values with Lags 10, 25, and 50 were higher than for Group 0-50. The results suggest that in rats and humans, high levels of variability established early in training produced more variable behavior even if the requirement for variability decreases. Different from human participants in a previous study, rats' behavior, stop varying when the criterion was markedly reduced (Lag 2) or eliminated (Lag 0).