Clustering in the output of behavior.
Daily response rates under fixed-interval schedules move in predictable waves, not random scatter.
01Research in Context
What this study did
Spealman et al. (1978) watched pigeons peck a key every day on fixed-interval schedules.
They tracked daily response rates to see if the numbers bounced around at random.
The birds worked under several FI values; the team looked for wave-like patterns across sessions.
What they found
Response rates did not scatter randomly. They rose and fell in steady waves across days.
The clustering looked like feedback or a built-in rhythm, not noise.
How this fits with other research
Palya (1992) zoomed closer and found 3 Hz mini-bursts inside each FI scallop, extending the 1978 wave idea to the micro level.
Blough (1992) used math to show the same messy-looking scallop can be predicted with nonlinear equations, giving a deterministic reason for the clusters.
Kono (2017) moved from time to space: longer FI schedules made pigeons peck in more varied spots, showing FI control spreads beyond just rate.
Davison et al. (1968) had already said local reinforcement probability, not overall rate, drives responding—an earlier hint that moment-by-moment factors shape the patterns D et al. saw.
Why it matters
Your client’s behavior may also cluster across days, not just within sessions. Check for weekly waves before you tweak a program. If gains slip every few days, the schedule itself might be feeding a rhythm. Try shortening the interval or adding a stretch-to-reset prompt to break the cycle.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Graph your client’s daily response totals for the last two weeks—look for a wave and adjust the interval length if you see a repeating slump.
02At a glance
03Original abstract
Pigeons exposed to fixed-interval schedules of 3, 5, 15, 40, and 120 minutes all maintained considerable variability in daily response rates for as many as 104 sessions. However, variations did not occur at random. Instead, rate in a session appeared dependent on those occurring previously. The series displayed a wave-like form arising because a group of high rates was followed by a group of low rates and vice versa. These sequential relations produced a curve having irregular periodicity, sometimes superimposed on a declining or rising linear trend. Whether grouping of response rates stemmed from experimental or extra-experimental sources was not determined. If the phenomenon was either totally or partially produced by the schedule itself, it suggests that response rate is determined by a combination of positive and negative feedback. Control by factors extrinisic to the experiment itself implies that response rate may be influenced by some rhythmic physiological process.
Journal of the experimental analysis of behavior, 1978 · doi:10.1901/jeab.1978.29-363