Operant Conditioning

Definition

The four contingencies, which are named confusingly and are worth stating precisely:

• Positive reinforcement: adding something, behaviour increases.
• Negative reinforcement: removing something, behaviour increases.
• Positive punishment: adding something, behaviour decreases.
• Negative punishment: removing something, behaviour decreases.

Positive and negative here mean added and removed, not pleasant and unpleasant. Negative reinforcement is not punishment, and this is probably the single most common error in applied psychology writing.

Schedules of reinforcement — fixed or variable, ratio or interval — produce characteristically different response patterns, with variable ratio producing the highest and most extinction-resistant rates.

History

Thorndike's law of effect (1898) came first. Skinner formalised the analysis from the 1930s onward and built the experimental apparatus and the vocabulary the field still uses.

Radical behaviourism's claim that internal states were not needed for explanation was rejected by mainstream psychology during the cognitive revolution. The experimental findings about contingencies were not rejected, and remain robust; the philosophical position and the empirical results are separable and are often confused with each other.

Theory

The core claim is that behaviour is selected by its consequences, an analogy to selection in evolution that Skinner drew deliberately.

The explanatory unit is the three-term contingency: antecedent, behaviour, consequence. Applied work usually intervenes on antecedents and consequences rather than on the behaviour directly.

Avoidance learning is the theoretically awkward case: a behaviour maintained by something that does not happen. Two-factor theory explains it with classical conditioning of fear plus negative reinforcement of escape, and the account is still debated.

Research

The basic contingency findings replicate reliably across species and settings, which is more than can be said for much of psychology.

Punishment works but has poor side effects: suppression that does not generalise, avoidance of the punisher, and no teaching of an alternative behaviour. The applied consensus favours reinforcing an incompatible behaviour over punishing the target one.

The overjustification effect — external reward reducing later intrinsic motivation — is real but narrower than popularly claimed. Meta-analytic work locates it mainly with expected, tangible rewards given for merely engaging in a task, and it does not appear for unexpected rewards or for rewards tied to performance quality.

Applications

Behaviour support in education and disability services, habit formation, contingency management in addiction treatment — which has strong trial evidence and is used far less than that evidence justifies — and the design of feedback loops in software.

Variable-ratio scheduling is also the mechanism behind the most criticised patterns in gambling and in engagement-optimised product design.

Criticism

• Negative reinforcement is misused constantly, including in professional writing.
• Skinner's theoretical claims are frequently attributed to the empirical findings, so criticising radical behaviourism is mistaken for refuting reinforcement.
• Extending contingency accounts to language was Skinner's weakest move and Chomsky's 1959 review of Verbal Behavior, whatever its fairness, ended it as a mainstream position.
• Ethical concerns in applied settings are real where the goal is compliance rather than a goal the person holds.

References

• Skinner, B. F. (1938). The Behavior of Organisms. Appleton-Century.
• Ferster, C. B., & Skinner, B. F. (1957). Schedules of Reinforcement. Appleton-Century-Crofts.
• Deci, E. L., Koestner, R., & Ryan, R. M. (1999). A meta-analytic review of experiments examining the effects of extrinsic rewards. Psychological Bulletin, 125(6), 627-668.
• Chomsky, N. (1959). Review of Skinner's Verbal Behavior. Language, 35(1), 26-58.