Reinforcement · Grandma’s rule
The Premack Principle
"Eat your vegetables, then you can have dessert." Every grandmother knows the rule; David Premack showed why it works, discovered that it runs backward under the right conditions, and changed what a reinforcer is.
Definition
The Premack principle states that the opportunity to engage in a more probable behavior will reinforce a less probable behavior when access to the first is made contingent on performing the second. Reinforcers, on this view, are not stimuli but behaviors: it is not the dessert that reinforces vegetable-eating but the eating of the dessert.[1]
Informally it is called Grandma's rule, or in classrooms the first–then rule: first the work, then the play.
In brief
- The opportunity to perform a more probable behavior reinforces a less probable one when access to the first is made contingent on the second.
- Reinforcers are behaviors, not objects, and the relation reverses with circumstances: water-deprived rats ran to drink, running-deprived rats drank to run.
- The response deprivation hypothesis is the better-supported statement: a contingency reinforces whenever it restricts the contingent behavior below its free baseline.
Premack's experiments
David Premack's 1959 study gave first-grade children free access to a candy dispenser and a pinball machine and measured which each child used more. He then made the two contingent on each other. For children who preferred pinball, making the machine available only after they ate a candy increased candy eating; making candy available only after a game did nothing much. For children who preferred candy, the result reversed: candy eating reinforced pinball, and pinball did not reinforce candy eating.[1] Whether an activity worked as a reinforcer depended entirely on whether it was the more probable of the pair for that child.
The decisive test came in 1962 with rats, running wheels, and water. Rats deprived of water would run in a wheel to earn access to drinking — the standard result, drinking reinforcing running. But rats given free water and deprived of running learned to drink in order to unlock the wheel: running reinforced drinking.[2] The same two behaviors reversed roles with the animal's circumstances. Drinking was not a reinforcer in itself; it was a reinforcer only when drinking was the more probable behavior.
Why this was radical
Thorndike and Skinner had treated reinforcers as things — food, water, a pellet — that were reinforcing more or less everywhere. Premack argued that reinforcement is a relation between behaviors: any behavior can reinforce any less probable one and be reinforced by any more probable one. The list of reinforcers is not a list of objects; it is a ranking of what the organism would do right now if it could.[3]
The punishment side
Premack later extended the principle to punishment. If being made to perform a less probable behavior is made contingent on a more probable one, the more probable one declines. Forcing a rat that would rather drink to run a wheel after each drink suppresses drinking.[4] Reinforcement and punishment become two sides of a single relation: the direction of the probability difference determines which you get.
The response deprivation hypothesis
The principle has a known failure. Eisenberger, Karpman, and Trattner found in 1967 that a less probable behavior can reinforce a more probable one, provided the schedule restricts the less probable behavior below the level the organism would normally choose.[5] William Timberlake and James Allison built that into a more general rule in 1974, the response deprivation hypothesis. Measure how much of each behavior the organism performs when both are freely available — the baseline. A contingency will make one behavior reinforce another if, under the contingency, performing the instrumental behavior at its baseline level would leave the organism with less of the contingent behavior than its baseline. Formally, with I the required instrumental responses, C the contingent responses earned, and O the baseline levels:
When the inequality holds, the organism is "response deprived," and it will increase the instrumental behavior to recover access to the contingent one — regardless of which was more probable to begin with.[6] Premack's principle turns out to be the special case in which the contingent behavior is the more probable one, which almost guarantees deprivation. Timberlake and Allison's version explains the reversals, and it explains why a small requirement often fails: "read one page, then play for an hour" deprives the child of nothing.
Applied tests followed. Konarski and colleagues showed in classrooms that schedules meeting the response-deprivation condition increased children's academic work even when the contingent activity was the less preferred one, exactly as the hypothesis and not the original principle predicted.[7] A later review concluded that the response deprivation hypothesis is the better-supported statement, and that both are best understood through the lens of motivating operations: restricting a behavior below baseline is an establishing operation for it.[8]
How to use the Premack principle
- Observe before you arrange. Watch what the person (or animal, or you) actually does when free to choose. The most probable behaviors are the reinforcers available to you, whatever anyone says they enjoy.
- Make the probable behavior contingent on the improbable one. First homework, then screen; first the walk, then the coffee; first three sits, then the game of tug. Access to the reinforcing activity is granted only after the target behavior.
- Keep the ratio deprivational. The requirement has to leave the person with less of the preferred activity than they would otherwise have taken, or there is no reason to work for it. Short requirements, delivered often, are usually better than a huge requirement for a huge reward.
- Deliver promptly. The preferred activity should begin within seconds of the target behavior ending, or be bridged with a conditioned reinforcer such as a token or a check-off.
- Watch for satiation. Probabilities change. After an hour of screen time, screen time is no longer the most probable behavior, and the contingency stops working until it is restored.
Examples
| Setting | Less probable behavior (first) | More probable behavior (then) |
|---|---|---|
| Parenting | Clearing the table | Playing outside |
| Classroom | Ten minutes of math problems | Five minutes of free choice |
| Special education | Completing a task strip | Time with a preferred toy, shown on a "first–then" board |
| Dog training | Sitting at the door | Going through it for the walk |
| Self-management | Writing 200 words | Checking messages |
| Exercise | The workout | The podcast you only listen to at the gym |
| Workplace | Filing the expense report | Starting the interesting design task |
The dog example is the one most people already use without a name. A dog that wants to go outside will sit, wait, and make eye contact for the privilege, and the door opening is a more reliable reinforcer than any treat because, at that moment, going through the door is the dog's most probable behavior.[9] The classroom "first–then" board, standard in applied behavior analysis, is Premack made visible.
Common mistakes
- Reversing the order. "You can play now if you promise to do homework after" delivers the reinforcer before the behavior. It reinforces promising.
- Guessing the probabilities. Parents and managers routinely pick a "reward" nobody would choose. The only test is observation: what does the person do when free?
- Requirements that deprive nobody. If the person could get as much of the preferred activity as they want anyway, the contingency has no bite. Access must actually be restricted.
- Making the target behavior aversive. A punishing requirement — an hour of tedium for five minutes of play — teaches avoidance of the whole arrangement. Small, frequent contingencies work better.
- Forgetting that the reinforcer is an activity. Ending the preferred activity abruptly to start the next requirement can function as negative punishment and provoke resistance. Signal transitions in advance.
Where it fits in the theory
The Premack principle and its successor sit alongside the matching law as the two great "relativity" results of operant research. Matching says a behavior's strength depends on what the alternatives pay; Premack says whether something reinforces at all depends on what the organism would otherwise be doing. Both replaced the picture of reinforcers as fixed objects with a picture of organisms distributing their time among activities, and both feed directly into motivating operations, the modern term for the deprivation and satiation that make an activity more or less probable.[8]
Key takeaways
- A more probable behavior reinforces a less probable one when access to it is granted only after the less probable one. Grandma's rule and the classroom first–then board are the same idea.
- Reinforcement is a relation between behaviors, not a property of objects. Which behavior reinforces which depends on what the organism would otherwise be doing, deprivation can reverse the roles, and running the relation the other way produces punishment.
- The response deprivation hypothesis is the more accurate statement: a contingency reinforces the instrumental behavior whenever it leaves the organism with less of the contingent behavior than its free baseline, regardless of which behavior was more probable to begin with.
- To use it, observe what the person actually does when free, make the probable behavior contingent on the improbable one, keep the requirement deprivational but small and frequent, deliver promptly, and watch for satiation.
- The common mistakes are delivering the preferred activity first, which reinforces promising; guessing the probabilities instead of observing them; requirements that deprive nobody; and requirements so large that the whole arrangement becomes aversive.
Check yourself
A manager announces a team lunch as a reward for finishing reports on time. Reports get no faster. Why might the lunch have failed?
A reinforcer is not whatever the manager thinks people enjoy; it is what people would actually be doing if free to choose, and the only test is observation. If the lunch is not a more probable activity than what it displaces, or if people would get it anyway, the contingency has no bite.
A child spends more free time on math worksheets than on coloring. The teacher makes coloring available only after math and restricts it below the amount the child would normally do. Can coloring reinforce math, the more probable behavior?
Yes. The original Premack principle says only a more probable behavior can reinforce, but the response deprivation hypothesis says a contingency reinforces whenever it restricts the contingent behavior below its baseline, whichever behavior was more probable. Konarski's classroom studies found exactly this, which is why the response deprivation hypothesis is the better-supported statement.
A parent says, "You can play video games now if you promise to do your homework afterward." Is this the Premack principle?
No. The preferred activity is delivered before the target behavior, so the arrangement reinforces promising, not homework. Access to the probable behavior has to come only after the less probable one.
"Read one page, then you can play for an hour" produces almost no reading. What went wrong?
The requirement deprives the child of nothing: one page for an hour of play leaves the child with as much play as ever, so there is no reason to work for it. The ratio has to be deprivational, and short requirements delivered often work better than a huge requirement for a huge reward.
Explain it to a friend. Explain why "first the work, then the play" works and when it stops working, in three sentences that use two activities from your own routine.
Frequently asked questions
What is the Premack principle in simple terms?
A behavior you are likely to do can be used to reinforce a behavior you are unlikely to do, if the likely one is only allowed after the unlikely one. Vegetables before dessert; homework before video games; the walk before the coffee. It is often called Grandma's rule.
What is an example of the Premack principle?
A child who would rather play than tidy up is told that play begins once the toys are put away. Tidying increases. In the laboratory, water-deprived rats ran in a wheel to earn drinking, while rats deprived of running drank to earn access to the wheel — the same two behaviors, reversed by circumstances.
What is the response deprivation hypothesis?
Timberlake and Allison's 1974 refinement: a contingency reinforces the instrumental behavior whenever it restricts the contingent behavior below the amount the organism would freely perform. It explains cases where a less-preferred activity reinforces a more-preferred one, which the original Premack principle cannot, and it is generally considered the more accurate statement.
Is the Premack principle positive reinforcement?
Yes. Access to the preferred activity is added after the target behavior, and the target behavior increases. The novelty is in what counts as the reinforcer — an opportunity to behave, rather than a stimulus — not in the quadrant.
Does the Premack principle work on yourself?
Yes, with the caveat that you are both the person setting the rule and the person tempted to break it. It works best when the preferred activity is physically gated — the podcast that only exists at the gym, the café you only visit after the writing — so that the contingency does not depend on willpower.
Who was David Premack?
An American psychologist (1925–2015) who, besides the principle that bears his name, was a pioneer of primate cognition research: he taught the chimpanzee Sarah to communicate with plastic symbols and, with Guy Woodruff, coined the term "theory of mind" in 1978.
References
- Premack, D. (1959). Toward empirical behavior laws: I. Positive reinforcement. Psychological Review, 66(4), 219–233.
- Premack, D. (1962). Reversibility of the reinforcement relation. Science, 136(3512), 255–257.
- Premack, D. (1965). Reinforcement theory. In D. Levine (Ed.), Nebraska Symposium on Motivation (Vol. 13, pp. 123–180). University of Nebraska Press.
- Premack, D. (1971). Catching up with common sense or two sides of a generalization: Reinforcement and punishment. In R. Glaser (Ed.), The Nature of Reinforcement (pp. 121–150). Academic Press.
- Eisenberger, R., Karpman, M., & Trattner, J. (1967). What is the necessary and sufficient condition for reinforcement in the contingency situation? Journal of Experimental Psychology, 74(3), 342–350.
- Timberlake, W., & Allison, J. (1974). Response deprivation: An empirical approach to instrumental performance. Psychological Review, 81(2), 146–164.
- Konarski, E. A., Johnson, M. R., Crowell, C. R., & Whitman, T. L. (1980). Response deprivation and reinforcement in applied settings: A preliminary analysis. Journal of Applied Behavior Analysis, 13(4), 595–609.
- Klatt, K. P., & Morris, E. K. (2001). The Premack principle, response deprivation, and establishing operations. The Behavior Analyst, 24(2), 173–180.
- Pryor, K. (1984). Don't Shoot the Dog! The New Art of Teaching and Training. Simon & Schuster.