Matching and maximizing with variable-time schedules.

The experimenter placed three pigeons in separate chambers. Each bird had two keys and a grain feeder.

Food dropped at random times no matter what the pigeon did. The only rule: stay on the key that last paid off and you kept the next grain. Switch keys and you lost it.

Sessions ran until birds had pecked 5 000 times. The question: would the birds maximize time on the richer key, or spread responses to match the payoff rates?

The birds did not pile onto the better key. Instead, their pecks lined up almost exactly with the grain rates.

If the left key paid 70 % of the time, about 70 % of pecks hit that key. This fits the matching law, not a maximizing rule.

Dube et al. (1998) also watched choice in a single-case lab set-up. They dropped the one-trial rule and kids with severe ID finally learned identity matching. Both papers show that small schedule tweaks change what the organism does.

Scull et al. (1973) used social rewards instead of grain. A tiny newspaper bounty made strangers hand over job leads. Like the pigeons, people allocated more help when the payoff rate was higher, again matching rather than maximizing.

Jenkins et al. (2016) asked how many practice swings staff need. One rehearsal plus feedback hit the fidelity ceiling, just like one grain schedule hit the response ceiling. All three studies say the same thing: extra input after the sweet spot is waste.

When you set up differential reinforcement, clients will likely match the payoff rates you give them, not hunt for the single richest source. If you want more eye contact than toy play, make sure eye contact pays off more often, not just bigger. Count the rates, not the magnitude, and you will see the behavior spread follow the numbers.