Delay and amount of reward in a concurrent chain.

Pigeons pecked two keys in a lab. Each key led to a different food path. One path gave more food later. The other gave less food sooner.

The team held the food ratio steady but stretched the wait times. They tested four delay ratios: 6:1, 3:1, 3:2, and 1:1. They wanted to see how longer waits changed the birds’ picks.

At 6:1, 3:1, and 3:2, birds switched to the small-quick snack as delays grew. At 1:1, the same birds switched the other way and waited for the big snack.

In short, longer waits hurt the big reward’s pull—unless both sides kept the same wait ratio.

Sanford et al. (1980) ran the same ratios and saw the same flip. Their 1980 paper is the direct forerunner; Haemmerlie (1983) adds a fourth ratio and firmer data.

Lloyd (2002) kept the delay gap fixed at 10 s while stretching both sides. Preference still fell, showing that absolute length—not just the ratio—steers choice.

Hinson (1988) took the same numbers and plugged them into a hyperbolic-decay formula. The model fit well, turning the mixed pigeon data into a clean curve.

For BCBAs, the lesson is timing beats size. A big reinforcer loses value if the wait feels long. Keep delays short, or keep both sides equal, to protect the power of your larger reward. Check your token boards, DRO schedules, or response chains—if one path drags, the learner may quit before they reach the payoff.