The effect of foreperiod duration on reaction time and its relation to interval schedules of reinforcement.
The way you space the cue matters as much as how often you reinforce.
01Research in Context
What this study did
Keller et al. (1978) tested how the wait time before a cue changes the speed of the bird's response. They used pigeons in a lab box. The cue could appear after short, medium, or long waits, and the waits were mixed up like a variable-interval schedule.
They recorded how fast the bird pecked each time the wait changed. The goal was to see if the wait pattern alone could shape reaction time, just like reinforcement rate does.
What they found
Reaction time moved up and down with the wait pattern. Short waits made the bird peck faster. Long waits made it slower. The pattern looked like the curves you see under variable-interval food schedules.
The authors say the same clock-like process may control both pause length and reaction time.
How this fits with other research
STEBBINROSS et al. (1962) showed that thicker reinforcement schedules give shorter, steadier latencies. Keller et al. (1978) now add that the wait before the cue can mimic those same latency shifts, even when food rate is fixed. The two studies stack together: one proves rate matters, the other shows timing structure alone can do the same.
Blough (1992) found that only reinforcement probability, not stimulus frequency, changed pigeon reaction time. Keller et al. (1978) agree that timing rules matter more than cue features, but they push the idea further by showing the wait distribution itself is the active variable.
Glover et al. (1976) split pause control from running-rate control under VR schedules. Keller et al. (1978) echo this split by showing latency can be pulled away from overall rate when foreperiod rules change. Both warn us to watch time-based rules, not just food count.
Why it matters
If you want faster client responses, do not just add more tokens. Tighten the wait window. Use short, mixed foreperiods like a VI 10-s schedule. You can rehearsal this in match-to-sample drills, DTT trials, or even between-instruction pauses. Start with 2-4 s waits, then mix in occasional 6-8 s probes. Track latency for ten sessions. You should see the same speed-up V saw, without touching your reinforcement rate.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Cut your DTT wait time to 3 s average and randomize the interval; graph first-response latency for one week.
02At a glance
03Original abstract
Two groups of pigeons were exposed to a simple reaction-time procedure in which mean foreperiod duration was 5, 10, or 20 seconds. For one group, the foreperiods had an arithmetic, or rectangular, distribution; for the second group, they had a constant-probability, or Bernoulli, distribution. Under both distributions, mean response latency was an increasing, negatively accelerated function of mean foreperiod duration. On a given trial, response latency was a function of its associated foreperiod duration: latency was a decreasing function of foreperiod duration in the arithmetic distribution, and an increasing function of foreperiod duration in the constant-probability distribution. Examination of the distribution of latencies revealed a harmonic structure reminiscent of distributions of interresponse times under variable-interval schedules of reinforcement. Taken together, the results confirm and extend previous findings with human subjects, and also suggest numerous similarities to behavior maintained by variable-interval schedules.
Journal of the experimental analysis of behavior, 1978 · doi:10.1901/jeab.1978.30-19