Choice, rate of reinforcement, and the changeover delay.

Stubbs et al. (1970) watched pigeons choose between two keys. Each key gave grain on its own schedule. A change-over delay (COD) made the bird wait a few seconds after it switched keys before food could pop.

The team counted every peck and timed how long the birds stayed on each side. They wanted to see if the birds matched their responses and their time to the grain rates.

Matching was not one smooth process. Quick bursts of pecks right after switching, plus the forced wait of the COD, created the numbers we call matching. When the researchers erased those first post-switch pecks, response matching vanished but time matching stayed.

In plain words: the birds’ schedule of staying and switching, not just the grain odds, wrote the final scoreboard.

MacDonall (2003) later paid birds for staying and for switching. The COD still lengthened visits, proving the delay itself is a local reinforcer that can be strengthened or weakened.

Pliskoff et al. (1981) tightened or loosened the number of pecks needed to switch. Bigger requirements slid the data from under-matching to over-matching, showing the same lever A et al. spotted can be dialed up or down.

Oliver et al. (2002) moved the matching law out of the lab. Children with severe problem behavior matched their aggression to the attention it earned at home, even without any COD. The core relation holds, but the delay mechanism is optional for humans.

When you see a client split their time or responses across two tasks, check the local rules, not just the pay-off odds. A brief ‘cool-off’ period, a fixed number of responses before switching, or extra attention for staying put can each tilt the numbers without changing the overall reinforcement rate. Try adjusting these micro-rules first; they often fix allocation problems faster than re-weighting the big reinforcers.