Density and delay of punishment of free-operant avoidance.

Scientists taught rats to press a lever to avoid mild electric shocks. Once the rats were good at it, the team added extra shocks right after some lever presses. They changed two things: how many presses got shocked (density) and how soon the extra shock came (delay).

The goal was to see if these extra shocks would make the rats press less and maybe get fewer shocks overall.

More shocks and faster shocks did stop the rats from pressing. But here is the twist: total shocks usually went up, not down. When the rats froze, they missed chances to avoid the regular shocks, so they got zapped anyway.

The local rule—press equals quick shock—beat the global goal of fewer shocks.

Koegel et al. (1992) also used extra shocks, but they timed them to hit only certain response patterns. Their rats cut just those patterns, proving punishment works best when it is precise. Sachs et al. (1969) show that sloppy timing can backfire.

Mosk et al. (1984) asked how strong the regular shock must be to keep avoidance going. They found a minimum level; more intensity did not help. Sachs et al. (1969) add that how often and how soon you punish matters just as much as how strong.

Wilkie (1973) ran a similar lab and saw that switching from free shocks to response-linked shocks first raised, then lowered response rates. Together these papers warn: changing any shock parameter—intensity, density, delay, or contingency—can flip behavior from more to less.

If you use punishment or response-cost to cut a problem behavior, think about the whole chain, not just the one response. A quick consequence may stop the act today but lock the client into a bigger problem tomorrow. Check that the suppression actually lowers the total aversive events, not just the target response. If data show total negatives rising, fade density, add delay, or switch to differential reinforcement instead.