Choice and segmented interreinforcement intervals.

Pigeons pecked two keys in a chamber. One key gave food on a simple VI schedule. The other key started a two-link chain: first link VI, second link another VI.

The overall rate of food was the same on both keys. The only difference was how the wait time was split.

Researchers wanted to know if birds cared about the chopped-up wait inside the chain.

Birds usually picked the simple VI key. But preference flipped when the first chain link was very short.

Choice tracked the local segment, not the total reinforcement rate. A tiny first link felt like a faster payoff, even if total delay was equal.

Najdowski et al. (2003) later showed pigeons can flip their choice almost overnight when terminal-link delays change daily. That extends Moore (1982): the local segment still rules, but the system updates fast.

Bensemann et al. (2015) moved from two keys to four and found the constant-ratio rule only once the schedule sat still for many sessions. Again the lesson holds: steady state smooths out the local bumps J first spotted.

Matthews et al. (1987) swapped VI for fixed-ratio chains and saw the same pattern—short first links won. The two studies echo: whatever the response type, the opening segment carries extra weight.

When you build token boards or chained schedules for learners, the first step should feel easy or quick. A short, sweet initial link can boost engagement even if later steps stay long. Watch local delays, not just the overall rate—splitting wait time can shift preference more than you think.