Effects of delayed reinforcement in a concurrent situation.

Pigeons pecked two keys. Food followed pecks on each key, but one key paid off after a short wait and the other after a longer wait.

The researchers slowly stretched the wait time for one key. They counted how often birds switched keys as the delay grew.

When the wait for food got longer, pecking on that key dropped fast. The drop looked like a sliding curve, not a straight line.

Birds moved most of their pecks to the key that still paid quickly.

Hinson (1988) kept the curve but swapped the math. The 1965 paper said the drop is exponential; Mazur showed a hyperbola fits better. Same birds, same keys, tighter equation.

Davidson et al. (1992) added a twist: a brief light told the pigeon "wait here." With the signal, birds kept pecking even when the wait reached 20 s. The 1965 study had no signal, so pecking crashed sooner. The papers seem to clash until you spot the light cue.

Cudré-Mauroux (2010) review bundles both results. It reminds us: delay hurts only if the learner cannot see or hear what bridges the gap.

When you shape two behaviors, the one that pays faster will win. If you must delay reinforcement, fill the gap with a signal—a token, a click, a brief praise—so the learner knows the deal is still on. Keep the wait short relative to the task rhythm, or the response you want may disappear before your eyes.