The effects of the stimulus-reinforcer correlation in a discrete-trials IRT>t procedure.
Extra free reinforcers cut response rates only when they outshine earned ones and carry no unique cue.
01Research in Context
What this study did
Carr (1979) ran a lab study with rats. Each trial gave the rat six seconds to press a lever.
If the rat waited longer than six seconds, the trial ended and nothing happened.
The twist: between trials the rat sometimes got free food. The study asked, does extra free food change how often the rat presses?
What they found
When free food matched or beat the food earned by pressing, the rats pressed less.
Surprise: giving free food did not make the rats better at pressing. Their hit rate stayed flat.
The drop in pressing only showed up when the food truly depended on the lever press.
How this fits with other research
Van Hemel (1973) saw the same pattern. When a light told the rat food was coming, the rat looked at the light more. Both studies say the same thing: cues that predict food control behavior.
Cullinan et al. (2001) later showed that free food can also make pressing stronger during extinction. Their free food came with a special sound. The sound acted like a safety signal, so the rat kept pressing. Carr (1979) gave no special signal, so pressing fell.
Corrigan et al. (1998) used the rule in kids. In FCT, a card told the child which reinforcer was coming. The card kept treatment working even when the favorite toy was absent. The lab rule holds in the clinic.
Why it matters
Check the signals you give between trials. If free attention, tangibles, or edibles come too often, the learner may work less during teaching trials. Pair any free reinforcer with its own clear cue, or save it for after the session. This keeps the response-reinforcer link strong and your teaching efficient.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Put a unique bracelet on your wrist before you deliver any free edible between trials; remove it during trial time.
02At a glance
03Original abstract
The correlation between a keylight and food in a discrete-trials, interresponse-time-greater-than 6-sec (IRT>6-sec) procedure was varied by manipulating the rate of response-independent food presentation in the intertrial interval. When the correlation was positive, the rates of pecking in the IRT>6-sec condition were high and food was obtained on only about 5% of the trials. Likewise, responding was maintained at a high rate in yoked birds that received the same presentations of the light and food as the birds in the IRT>6-sec condition. When the rate of reinforcement between trials was equated to or made greater than the rate of reinforcement within trials, the response rate decreased for all birds, and those decreases were considerably larger for the yoked birds. However, the percentage of trials in which reinforced responses occurred under the IRT>6-sec procedure did not increase substantially when the light and food were either uncorrelated or negatively correlated. The percentage of trials in which a reinforcer was obtained increased when the keylight was left on continuously and the discriminative stimulus was not presented on the key. The results show that the stimulus-reinforcer correlation affects responding in the discrete-trials IRT>6-sec procedure, but that the effects of the stimulus-reinforcer correlation vary as a function of whether reinforcement is response-dependent or response-independent. The differences between the effects of response-independent and response-dependent pairings and nonpairings of the light and food are best accounted for in terms of differences in the control of responding by background stimuli.
Journal of the experimental analysis of behavior, 1979 · doi:10.1901/jeab.1979.31-307