Effect of punishment on human variable-interval performance.

Researchers asked adults to press a button for money. They paid on a variable-interval schedule. That means money came no matter how fast you pressed, just at random times.

While people pressed, the team also took away small amounts of money. The loss happened on a VR-34 schedule. Roughly every 34th press cost you cash.

The scientists ran an ABAB design. They turned the money-loss on, off, on, off. They wanted to see if the loss would slow pressing.

Money loss cut button pressing every time it was turned on. The drop happened no matter how rich or lean the payoff schedule was.

Even though people still earned cash, the tiny fines kept them from pressing. The effect showed up right away and stayed strong across all phases.

Goldman et al. (1979) ran almost the same setup and got the same drop. They added math to show the loss also bent the matching law. Together the two papers prove VR-34 money fines reliably slow human responding.

WEINELong (1962) did an earlier version with point loss instead of cash. He saw suppression too, but also noticed new FI-like pauses. The 1977 study keeps the finding cleaner by sticking to one schedule at a time.

Fontes et al. (2022) moved from money to electric shock. They showed a steady, moderate shock works better than slowly turning it up. Their result widens the 1977 claim: any well-timed aversive, not just cash fines, can cut behavior.

If you use response cost with clients, know that even small, random losses can cut behavior fast. The drop happens no matter how often you pay for good responses. Check that the lost item is truly worth losing and that skills you want to keep are reinforced elsewhere.