Matching with a key-peck response in concurrent negative reinforcement schedules.

Wheatley et al. (1978) worked with six pigeons in a small chamber. The birds could peck left or right keys.

Each key gave breaks from mild electric shock. The shocks came on different schedules. The team watched which side each bird chose.

The birds matched. If the left key gave 70 percent of the shock-free time, the birds spent about 70 percent of their pecks there.

The fit was almost perfect. R squared topped .95. Matching law works for negative reinforcement just like for food rewards.

Ward-Horner et al. (2017) later showed that learners often pick bigger, rarer payoffs. Their review includes concurrent schedules, so the 1978 shock data sits inside their bigger picture.

Shahan et al. (2021) found that thinning alternative reinforcement too fast brings back problem behavior. Their work rests on the same matching idea: less payoff on one side shifts responses toward the other.

Yeh et al. (2025) used a discounting model to predict human choices between risky and delayed rewards. Both studies use lab tasks to quantify choice, but Yeh’s math predicts shifts where L et al. simply describe the steady split.

You now know that clients will allocate their responses to whatever option gives the most relief. If you thin reinforcement too quickly on the replacement behavior, the problem behavior gains share, just like the pigeons shifted back to the richer shock-free side. Keep alternative reinforcement dense during early thinning and make steps small.