Control over response number by a targeted percentile schedule: reinforcement loss and the acute effects of d-amphetamine.
Tight percentile schedules amplify d-amphetamine’s power to destroy response control, while looser yoked schedules let the learner ride out the storm.
01Research in Context
What this study did
Dunlap et al. (1991) gave lab animals a percentile schedule. The animals had to hit a moving target of how many times to press a lever. If they pressed too much or too little, they lost some sugar pellets.
The team also ran a yoked group. These animals got the same pellets on the same times, but with no response rule. Both groups then got a shot of d-amphetamine. The researchers watched who kept steady responding and who fell apart.
What they found
The drug wrecked response control under the percentile schedule. Animals could not stay near the target number.
Oddly, the yoked group kept more stable. Losing pellets only helped the yoked animals, not the percentile ones. The schedule itself decided who could adjust and who could not.
How this fits with other research
McAuley et al. (1986) also saw d-amphetamine ruin timing on fixed-interval schedules. Both studies show the same drug hurting stimulus control, just on different schedule shapes.
Goldman et al. (1979) found the drug raised errors when pigeons had to learn new rules. That paper and this one agree: learning or adjusting is more fragile than steady performance.
Locurto et al. (1980) looks like a contradiction. They showed stimulus fading protected acquisition from the same drug. The difference is task type: fading gives extra cues, while percentile schedules make cues stricter. Extra help shields behavior; tighter rules break faster.
Why it matters
If you run precision teaching or percentile-based fluency goals, know that stimulant side effects can shatter self-control. Build in safety cues or brief yoked breaks so the learner has a fallback when the drug hits. Watch for sudden jumps or drops in response count; that is the first sign the schedule is too brittle.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Add a brief ‘free win’ trial every few minutes so the learner can reset if response count starts to swing wildly.
02At a glance
03Original abstract
Two fixed-consecutive-number-like procedures were used to examine effects of acute d-amphetamine administration on control over response number. In both procedures, rats were required to press the left lever at least once and then press the right lever to complete a trial. The consecutive left-lever presses on each trial comprised a "run." Under the targeted percentile schedule, reinforcement was provided if the current run length was closer to the target length (16) than half of the most recent 24 runs. This differentially reinforced run length while holding reinforcement probability constant at .5. A second group acquired the differentiation under the targeted percentile schedule, but were then shifted to a procedure that yoked reinforcement probability by subject and run length to that obtained under the targeted percentile schedule. The two procedures generated practically identical control run lengths, response rates, reinforcement probabilities, and reinforcement rates. Administration of d-amphetamine disrupted percentile responding to a greater degree than yoked control responding. This disruption decreased reinforcement frequency less in the former than the latter procedure. The similar baseline responding under these two procedures suggests that this difference in sensitivity was due to behavioral adjustments to drug prompted by reduction of reinforcement density in the yoked control but not the percentile schedule. These adjustments attenuate the drug's effects under the former, but not the latter, procedure.
Journal of the experimental analysis of behavior, 1991 · doi:10.1901/jeab.1991.56-205