Indifference between punishment and free shock: evidence for the negative law of effect.
Punishment only works well when the target response truly controls the aversive outcome and the reinforcement schedule is not too lean.
01Research in Context
What this study did
Thomas et al. (1968) worked with pigeons in a small lab chamber. The birds pecked a key for food on a variable-interval schedule. Sometimes electric shocks came no matter what. Other times shocks only followed key pecks. The team kept the total number of shocks the same in both setups. They wanted to know if the birds cared about who caused the shock.
What they found
The pigeons avoided the side that gave more total shocks. They did not care if their own pecks caused the shock or if shocks were free. The key finding: punishment worked best when the bird's own response controlled the shock. Stronger control by the bird meant stronger behavior drop. The authors called this a 'negative law of effect'.
How this fits with other research
Fontes et al. (2025) later showed the simple 'direct-suppression' idea breaks down. When reinforcement rates shift quickly, birds choose the richer side even if both sides give equal shocks. Their 2025 data update the 1968 rule: relative shock rate, not absolute, now drives choice.
Santi (1978) extended the idea. Shock hurt responding more when that same response earned little food. Lean reinforcement lets punishment bite harder. The negative law still holds, but only after you check how rich the payoff schedule is.
Gibbon (1967) ran a similar test one year earlier. He also saw bigger drops when birds could avoid shock. Both papers agree: contingency matters. The 1968 study just made the rule symmetrical and cleaner.
Why it matters
Before you add a punisher, ask: does the client's behavior really control the consequence? If the room already has lots of reprimands 'for free,' extra contingent ones may add little. Also check the reinforcement side. Thin schedules make any punisher more powerful. Use rich schedules or strong non-contingent reinforcers to buffer the bite when you must punish.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Track how often the problem behavior contacts the punisher versus how often aversives occur for free; adjust the environment or the contingency before raising intensity.
02At a glance
03Original abstract
Pigeons were trained to respond under two conditions with two identical variable-interval schedules of positive reinforcement. While the schedules operated for separate response keys, they were not available concurrently. During one condition, each response was punished with electric shock. During the other condition, shocks were delivered independently of responding. The punishment suppressed responding but the free shocks did not. However, when allowed to choose, the pigeons preferred the condition associated with the lowest rate of shock regardless of whether or not the shock was dependent on responding. In general, shocks exerted their greatest effect on whichever response had the greatest influence on shocks. In this respect, punishment is instrumental in suppressing behavior and the properties of punishment are symmetrical to those of reinforcement. This empirical symmetry dictates a corresponding conceptual symmetry in terms of a positive law of effect accounting for response increments and a negative law accounting for response decrements.
Journal of the experimental analysis of behavior, 1968 · doi:10.1901/jeab.1968.11-777