Concurrent schedules of reinforcement: effects of gradual and abrupt increases in changeover delay.
Raise changeover delays 0.5 s at a time to keep client response splits matched to your programmed reinforcement rates.
01Research in Context
What this study did
Blue et al. (1971) worked with six pigeons in a chamber with two keys. Each key paid off on its own variable-interval schedule. A peck on the other key too soon triggered a changeover delay (COD).
The team raised the COD in two ways. One group got sudden 3-second jumps. The other group got tiny 0.5-second steps every few sessions.
What they found
Birds in the gradual group kept their response split close to the programmed payoff split. Birds hit with sudden jumps drifted off the mark.
Tiny steps kept obtained reinforcement rates almost identical to scheduled rates. Big jumps did not.
How this fits with other research
Fahmie et al. (2013) later showed that a long-COD history itself becomes a reinforcer. Pigeons picked the side that once had the longer COD even when payoffs were equal. That finding extends the 1971 work: the way you build the COD history shapes later choice.
Rutland et al. (1996) used the same two-key VI setup but asked a different question. They tracked how response patterns shift within a single session. Both studies confirm that summed reinforcer rate drives responding, yet only S et al. tested how the COD is introduced.
Rider (1983) mixed short and long delays in a discrete trial. Rats liked the mix once short delays topped a large share. That paper is topically related but uses rats and a different procedure, so the results complement rather than clash with the pigeon COD data.
Why it matters
When you fade in a delay or a reinforcement schedule, take tiny steps. Sudden jumps let the client’s behavior drift away from the plan. Next time you thin a DRL or introduce a COD, move in 0.5-second bites and check the obtained rate stays on target.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →If you use a COD, drop it by 0.5 s each session instead of 3 s chunks and graph obtained vs scheduled reinforcer rates to stay on track.
02At a glance
03Original abstract
Pigeons' pecks were maintained on concurrent variable-interval 1-min variable-interval 3-min schedules of reinforcement, with a changeover delay of 2 sec. When changeover delay was increased successively to 5.0, 7.5, and 12.5 sec (Exp. I) the actual relative rate of reinforcement for the variable-interval 3-min key decreased progressively for two birds, abruptly for two other birds, and the subjects devoted proportionately less of their time and responding to that key. However, the relative performance measures (relative time and relative responding) approximated the actual relative rate of reinforcement, with a maximum discrepancy of 11%, over all changeover delay values investigated. Experiment II attempted to lengthen response-run durations on the variable-interval 3-min key so that they were long enough to meet the changeover delay requirement at each new changeover delay value, by progressively increasing the changeover delay by 0.5-sec increments. With this procedure the actual relative rate of reinforcement approximated more closely the scheduled relative rate as changeover delay increased. As in Exp. I, relative performance measures approximated the actual relative reinforcement rate (maximum discrepancy 17%).
Journal of the experimental analysis of behavior, 1971 · doi:10.1901/jeab.1971.16-67