Choice between two-component chained and tandem schedules.
Middle stimuli in a chain do not work as mini-reinforcers—cut extra steps to keep value high.
01Research in Context
What this study did
Pigeons pecked two keys in a chamber. Each key led to a different two-part schedule. One schedule showed a brief light between parts (chained). The other schedule ran straight through with no light change (tandem).
The birds chose freely between the two keys. Researchers counted pecks to see if the light itself worked like a tiny reward.
What they found
The pigeons split their pecks fifty-fifty. They did not favor the key with the extra light. The light did not act as a conditioned reinforcer.
Instead, the birds seemed to track only the final food delay. The middle light made no difference to their choice.
How this fits with other research
Duncan et al. (1972) ran a similar test the same year. They also saw that pigeons picked simple schedules over chained ones. Together, the two papers show the same point: extra links in the chain hurt preference.
Wacker et al. (1985) and Warren et al. (1986) stretched the idea further. They proved the effect grows when the final delay gets longer. The later work turns the null finding into a clear rule: simpler is better.
GOLLUMIGLER (1964) saw the first hint. Pigeons paused more in early chain links. The 1972 paper explains why: the early links carry no reward value.
Why it matters
When you build token boards or sticker charts, keep the path short. Each extra step dilutes value, especially for kids who already wait poorly. Try dropping middle tokens or merge steps. One clear cue to the final reinforcer beats a long chain every time.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Audit your token board: if a child needs three tokens before a break, test two instead and watch response rate rise.
02At a glance
03Original abstract
Pigeons were trained on a two-key choice procedure in which a pair of equal and concurrently available variable-interval schedules (initial links) arranged entry into one or the other of two mutually exclusive schedules (terminal links) that ended in primary reinforcement. The terminal links were two-component chained or tandem schedules. Responses during the initial links were distributed equally on the two keys whenever the terminal links were associated with the same sets of interreinforcement intervals. Whether or not the terminal-link interreinforcement intervals were the same on the two keys, initial-link responding was affected by neither the presence nor relative durations of differentially signalled components within a terminal-link schedule. The simplest interpretation of these results is that initial-link responding is maintained directly by delayed primary reinforcement, rather than conditioned reinforcement afforded by the stimuli correlated with the terminal-link schedule components. This finding suggests that aspects of chained schedule performance usually attributed to conditioned reinforcement might best be reinterpreted in terms of delayed primary reinforcement and various discriminative functions served by the component-correlated stimuli.
Journal of the experimental analysis of behavior, 1972 · doi:10.1901/jeab.1972.18-45