Some effects of punishment upon unpunished responding.

Dunham (1972) worked with rats in a small lab space. The rats could drink sweet water or run in a wheel. Each sip earned food. Each wheel turn earned food. The team then shocked the rats every time they drank. They watched what happened to running.

The goal was simple. See if punishing one action changes an unpunished action. The team also tracked time. They wanted numbers, not just yes or no.

Shock cut drinking fast. At the same time, running jumped up. The rise in running matched the drop in drinking almost exactly. The rats simply swapped how they spent each second.

The data fit a clean rule. Time stayed constant. Less time drinking meant more time running. The rule held across every rat.

SIDMAELLIOTT et al. (1962) saw the same swap in pigeons earlier. Pecking dropped while wing flaps rose. Both papers show punishment can boost an unpunished act. Dunham (1972) adds a math model to the story.

Sailor (1971) looks like a clash. He saw an upside-down U. Medium shock raised unpunished pecking a little, then high shock crushed it. The gap is method. W punished during extinction, when food was off. J punished during reinforcement, when food stayed on. Context drives the shape of the curve.

Sadowsky (1973) also seems to disagree. He found that shocking any link in a chain slowed the whole chain. Again, method matters. S timed how long rats waited to start. J counted how many times rats ran. Same shock, different measure, different tale.

You now have a ruler. If you punish one behavior, watch the clock. Any rise in an alternate act is likely time reallocation, not new motivation. Use this when planning punishment programs. Pair the punisher with a ready incompatible response that earns the same reinforcer. Check that the client simply trades seconds, not creates a fresh problem.