A two-state analysis of fixed-interval responding in the pigeon.

Pigeons pecked a key for food on a fixed-interval schedule. Every 3 minutes the first peck paid off.

The team recorded each peck to see how the birds spaced their responses through the interval.

The birds produced two clear states. They waited a long time after food, then suddenly switched to rapid pecking about two-thirds of the way through the interval.

The switch point was sharp, not gradual, for every bird.

Winett et al. (1972) later broke the same 3-min FI into short 10-s trials. The same two-state jump appeared, but the data were cleaner because birds could not peck randomly between trials.

Fay (1979) added a twist: if you let only one peck per interval, the next pause shrank. This shows the two-state rule can be nudged by response limits, even though the basic pattern stays intact.

Gibbon (1967) had already shown that brief flashes of light alone can keep the FI pattern running. The two-state model therefore holds even when food itself is delayed and only conditioned reinforcers mark the cycle.

When you shape behavior with FI-like schedules — think of a 5-minute DRO or a token board that opens only after a set time — expect two phases: silence, then rush. You can plan for the rush by prompting early or by reinforcing spaced responses just before the usual switch point. If data look messy, try breaking the interval into visible trials or adding brief conditioned reinforcers, tricks that produced cleaner patterns in later pigeon work.