Computational model of selection by consequences: patterns of preference change on concurrent schedules.

The computational model of selection by consequences is an ontogenetic dynamic account of adaptive behavior based on the Darwinian principle of selection by consequences. The model is a virtual organism based on a genetic algorithm, a class of computational algorithms that instantiate the principles of selection, fitness, reproduction and mutation. The computational model has been thoroughly tested in experiments with a variety of single alternative and concurrent schedules. A number of published reports demonstrate that the model generates patterns of behavior that are quantitatively equivalent to the findings from live organisms. The experiments and analyses in this study assess the behavior of the computational model for evidence of preference change phenomena in environments with rapidly changing reinforcement rate ratios. Molar and molecular effects of behavioral adjustment were consistent with those observed in live organisms. The results of this study provide strong evidence supporting the selectionist account of adaptive behavior.