An implementation of punishment in the evolutionary theory of behavior dynamics

Abstract

An implementation of punishment in the evolutionary theory of behavior dynamics is proposed, and is applied to responding on concurrent schedules of reinforcement with superimposed punishment. In this implementation, punishment causes behaviors to mutate, and to do so with a higher probability in a lean reinforcement context than in a rich one. Computational experiments were conducted in an attempt to replicate three findings from experiments with live organisms. These are (1) when punishment is superimposed on one component of a concurrent schedule, response rate decreases in the punished component and increases in the unpunished component, (2) when punishment is superimposed on both components at equal scheduled rates, preference increases over its no‐punishment baseline, and (3) when punishment is superimposed on both components at rates that are proportional to the scheduled rates of reinforcement, preference remains unchanged from the baseline preference. Artificial organisms animated by the theory, and working on concurrent schedules with superimposed punishment, reproduced all of these findings. Given this outcome, it may be possible to discover a steady‐state mathematical description of punished choice in live organisms by studying the punished choice behavior of artificial organisms animated by the evolutionary theory.

All Behavior is choice: Revisiting an evolutionary theory's account of behavior on single schedules

“…One is to study the effect of punishment superimposed on single schedules of reinforcement. In McDowell and Klapes's (2019) implementation of punishment in the ETBD, the punishment of a target behavior causes phenotypes in the target class to mutate with a specific probability. This mutation may result in a phenotype mutating out of the class altogether, thereby decreasing the probability that a behavior from that class will be emitted again.…”

“…The wide range of phenomena for which the theory accounts, and various untested predictions of the theory, are also discussed in these articles. Not included in these articles, however, is McDowell and Klapes' (2019) recent implementation of punishment in the evolutionary theory. They studied the behavior of AOs working on concurrent schedules of reinforcement with punishment superimposed on one or both alternatives, and found that the behavior of the AOs was consistent with the behavior of live organisms in similar experiments.…”

“…This produces a continuous stream of behavior that can be studied in the same way that the behavior of a live organism is studied. The behavior of AOs animated by the theory has been shown to reproduce the behavior of live organisms in qualitative and quantitative detail in a wide variety of experimental environments (McDowell, 2013b; McDowell, 2019; McDowell & Klapes, 2019).…”

“…This defect is more than just an abstract theoretical failing, for it calls into question the single‐schedule results the theory has produced (McDowell, 2004; McDowell & Caron, 2007). It also bears importantly on future applications of the theory to single schedules, such as the superimposition of punishment on these schedules (McDowell & Klapes, 2019), and the development of stimulus control in chained and multiple schedules (McDowell et al, 2006). Moreover, this flaw means that the theory cannot even in principle account for empirical findings that may be produced by manipulating background (i.e., nontarget) behaviors or reinforcers in single‐schedule experiments (e.g., Belke & Heyman, 1994; Dougan & McSweeney, 1985; Soto et al, 2005).…”

A NetLogo Implementation of the Norms and Metanorms Game: Behavior Analysis Meets Agent-Based Modeling

“…Computational models, on the other hand, are extremely rare. McDowell and colleagues have developed an intriguing and fruitful research project simulating behavior selection in a computer environment (McDowell, 2013, McDowell & Klapes, 2019. Their implementations have to make explicit the implicit assumptions adopted by verbal theories and in the process they are able to not only specify these parameters, but also show how the parameters interact when the programs run, and to determine the limits of the model, such as how much variation is needed for a selection-by-reinforcement to be adaptive (McDowell, 2017).…”