We demonstrate the usefulness of Bayesian methods in developing, evaluating, and using psychological models in the experimental analysis of behavior. We do this through a case study, involving new experimental data that measure the response count and time allocation behavior in pigeons under concurrent random‐ratio random‐interval schedules of reinforcement. To analyze these data, we implement a series of behavioral models, based on the generalized matching law, as graphical models, and use computational methods to perform fully Bayesian inference. We demonstrate how Bayesian methods, implemented in this way, make inferences about parameters representing psychological variables, how they test the descriptive adequacy of models as accounts of behavior, and how they compare multiple competing models. We also demonstrate how the Bayesian graphical modeling approach allows for more complicated modeling structures, including hierarchical, common cause, and latent mixture structures, to formalize more complicated behavioral models. As part of the case study, we demonstrate how the statistical properties of Bayesian methods allow them to provide more direct and intuitive tests of theories and hypotheses, and how they support the creative and exploratory development of new theories and models.

Reduced response-inhibition capacity is a defining feature of attention-deficit hyperactivity disorder. The fixed minimum interval (FMI) schedule has been systematically validated to assess such capacity in rats. On each FMI trial, the first lever press initiates an inter-response time (IRT); a potentially consummatory response terminates the IRT; only IRTs longer than a target interval result in access to food. Despite task validity, steady-state FMI performance in the most common animal model of attention-deficit hyperactivity disorder, the spontaneously hypertensive rat (SHR), is similar to normotensive control performance, even though SHR performs at lower levels, especially during acquisition, in similar response-withholding tasks. To determine whether such limitations of the model are specific to stable-state performance, this experiment compared FMI 6-s performance in SHR and Wistar rats during acquisition and in steady state, and assessed the effect of acute D-amphetamine (AMP) administration (0.1, 0.5, and 1.0 mg/kg) on steady-state performance. Median latencies to first lever press were consistently shorter in SHR than in Wistar rats; IRTs were shorter for SHR than for Wistar rats during acquisition, but substantially less so during asymptotic performance. AMP dose-dependently reduced latencies, shortened IRTs, and, at the highest dose, increased the proportion of IRTs under schedule control. These results suggest that, relative to Wistar rats, SHR have a reduced capacity to learn to withhold a reinforced response; once the FMI is acquired, high doses of D-AMP disrupt withholding performance in both strains, but they also enhance the responsiveness of both strains to reinforcement contingencies.

There is a growing interest in studying individual differences in choices that involve trading off reward amount and delay to delivery because such choices have been linked to involvement in risky behaviors, such as substance abuse. The most ubiquitous proposal in psychology is to model these choices assuming delayed rewards lose value following a hyperbolic function, which has one free parameter, named discounting rate. Consequently, a fundamental issue is the estimation of this parameter. The traditional approach estimates each individual’s discounting rate separately, which discards individual differences during modeling and ignores the statistical structure of the population. The present work adopted a different approximation to parameter estimation: each individual’s discounting rate is estimated considering the information provided by all subjects, using state-of-the-art Bayesian inference techniques. Our goal was to evaluate whether individual discounting rates come from one or more subpopulations, using Mazur’s (1987) hyperbolic function. Twelve hundred eighty-four subjects answered the Intertemporal Choice Task developed by Kirby, Petry and Bickel (1999). The modeling techniques employed permitted the identification of subjects who produced random, careless responses, and who were discarded from further analysis. Results showed that one-mixture hierarchical distribution that uses the information provided by all subjects suffices to model individual differences in delay discounting, suggesting psychological variability resides along a continuum rather than in discrete clusters. This different approach to parameter estimation has the potential to contribute to the understanding and prediction of decision making in various real-world situations where immediacy is constantly in conflict with magnitude.