Some experiments on the organization of a class of imitative behaviors.
Mix unreinforced imitations among reinforced ones to keep them alive, but don’t withhold rewards for too long or they’ll vanish.
01Research in Context
What this study did
Peterson (1968) worked with children who had intellectual disabilities.
The team wanted to know how reinforcement keeps untrained imitations alive.
They mixed reinforced imitations with unreinforced ones and later stopped all rewards.
What they found
Imitations that never earned their own rewards kept going if they sat among reinforced ones.
When the experimenters removed all rewards in one long block, the unreinforced copies vanished.
Bringing reinforcement back made the lost imitations return right away.
How this fits with other research
Lowe et al. (1974) later showed the same idea works for greetings.
They used two trainers and saw the wave generalize to over twenty new staff without extra work.
Both studies used prompting and fading with kids who had ID, but the 1974 paper moved from lab to real-life staff.
Bland et al. (2018) saw pigeons chase local reinforcer rates even when it hurt overall payoff.
Their data warn us that learners may copy what was last rewarded, so spacing out payoffs matters.
Why it matters
You can keep untrained skills in a learner’s toolbox by sprinkling them between reinforced responses.
Avoid long stretches of zero rewards; instead, give occasional praise or tokens so the behavior survives.
Next time you run imitation drills, reward every third or fourth response and watch the rest stay strong for free.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →During imitation drills, praise or token every third response and track if the unreinforced copies keep going.
02At a glance
03Original abstract
A retarded child was taught to imitate diverse demonstrations made by an experimenter, until new demonstrations were imitated correctly upon first presentation without direct training. These imitations could be maintained without reinforcement, when they were distributed among other reinforced imitations. Factors responsible for the continued performance of these unreinforced imitations were examined. When subjected to massed extinction trials, unreinforced imitations eventually disappeared; they reappeared when again interspersed among reinforced imitations. In addition, the stimulus function of "similarity of response between subject and experimenter" was examined. The subject was taught a set of non-imitative responses, through discriminative stimuli controlled by the experimenter, and a comparable imitative set. Unreinforced non-imitations, like reinforced imitations, were maintained only when interspersed among reinforced imitations. When all reinforcement was discontinued, all responses extinguished similarly, indicating that reinforcement was necessary to maintain the response-class organization, but not confirming an essential role for "similarity" as such.
Journal of applied behavior analysis, 1968 · doi:10.1901/jaba.1968.1-225