Relative frequency of reinforcement and rate of punished behavior.
Reinforcement richness in alternate schedule components can override punishment and swing punished rates up or down.
01Research in Context
What this study did
The team set up a two-part schedule. In one part, responses earned food on a variable-interval (VI) schedule. In the other part, responses earned food on a second VI schedule but also produced a brief punisher.
They changed how much food each part delivered. Then they watched whether the punished behavior sped up or slowed down.
What they found
When the punishment part started paying more food, the punished responses increased. When that part paid less, the responses dropped.
The same swing happened in the unpunished part. The changes matched classic behavioral contrast: richer alternate reinforcement can accidentally strengthen punished behavior.
How this fits with other research
Kaufman (1965) saw the same swing earlier, but with shock avoidance instead of punishment. Both studies show that changing payoff in one schedule component pulls response rates in the other.
Szatmari (1992) and Sturmey (1995) later showed the swing is driven by extra reinforcers quietly moving between components. The 1970 data fit that story: punishment did not override the pull of relative payoff.
Jensen et al. (1973) replaced punishment with simple omission of food and still got contrast. Together these papers say contrast is a core schedule effect, not a special-case punishment glitch.
Why it matters
If you run multiple schedules, DRA, or mixed reinforcement programs, remember that boosting reinforcement in one alternative can unintentionally revive problem behavior you are trying to punish. Check response rates across all components, not just the one where punishment is present. When you see a bounce-back, first look at the reinforcement ratio, not the punisher intensity.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Graph response rates in every component; if punished behavior rises, check whether reinforcement rose elsewhere first.
02At a glance
03Original abstract
After training on a multiple variable-interval variable-interval schedule of reinforcement, each response in one component of the schedule was followed by a brief electric shock. When the rate of punished responses stabilized, the frequency of reinforcement in the other component was first decreased and then increased from the baseline frequency. The effects of these manipulations were consistent with reports of interactions in multiple schedules involving only unpunished behavior, i.e., the rate of punished responses increased with a higher relative frequency of reinforcement in the punishment component and decreased with a lower relative frequency of reinforcement in that component. The relevance of such findings to a further generality of behavioral contrast effects is discussed.
Journal of the experimental analysis of behavior, 1970 · doi:10.1901/jeab.1970.13-319