Concurrent schedules: Effects of time- and response-allocation constraints.

The team set up two levers side-by-side. Each lever paid off on its own variable-interval schedule. They then added rules about how long or how often the animal could use each lever. Some blocks capped time per lever. Other blocks capped the number of presses per lever. They watched which rule actually controlled where the animal worked.

Time caps barely moved response splits. The animals still pressed in the same ratio even when a timer cut time short. Response caps, however, instantly reset both time and presses. Once the animal hit the press limit on one lever, it switched and stayed on the other. Response allocation drove the show; time followed along.

Marcucella et al. (1978) had already shown that flashing a light before payoff can crash responding on that lever. Their payoff signal shifted time away, but the 1991 study shows the signal works only if it does not cap the count. Jarrold et al. (1994) later moved the setup to a classroom. Kids under-matched math tasks unless the teacher added timers or delays. The lab finding that 'press rules beat time rules' helps explain why those classroom fixes were needed. Leigland (2000) pooled many interval schedules and still saw matching. The 1991 result says matching holds because organisms manage their counts, not the clock.

When you run two programs at once, set response limits first. A daily '10-token max' on each center will push learners to sample both. A daily '10-minute max' may not. If you use a visual timer or signal, pair it with a response cap so the signal does not empty one option. Let the count, not the clock, steer choice.