Punishment

A consequence that follows a behaviour and decreases the likelihood of that behaviour happening again.

Punishment is one of the two primary categories of consequence within operant conditioning (the other being reinforcement). It comes in two forms: positive punishment, in which something unpleasant is added (a verbal correction, a tap, a startle), and negative punishment, in which something pleasant is removed (the handler walks away, the play stops, the food is withdrawn).

The term is often confused with negative reinforcement in lay discussions, but the two are opposites. Reinforcement increases the likelihood of a behaviour; punishment decreases it. The two have very different implications for welfare and for what the animal is actually learning.

The behavioural science literature has consistently found that training systems built around punishment, particularly positive punishment, produce more side effects than reinforcement-based alternatives. Documented side effects include increased fear, increased aggression toward the punisher and toward unrelated targets, suppression of behaviour without learning what to do instead, and in chronic cases, learned helplessness. Modern applied animal behaviour practice across species has moved firmly away from punishment-based methods, with the consensus drawing on both the welfare evidence and the practical observation that reinforcement-based methods tend to produce more durable and flexible learning.

This does not mean punishment is never used. In particular, negative punishment (removing something pleasant) is widely accepted as the most welfare-positive of the four operant quadrants when applied carefully. The concerns are mostly with positive punishment, especially when escalated, repeated, or paired with poor timing.