- The Premack principle also called the relativity theory of reinforcement, and the differential-probability hypothesis, states that a more desirable activity (e.g. such as eating chocolate) can be used to reinforce a less desirable one (such as writing an essay).
- David Premack established his theory through experiments on rats, children, and later chimpanzees.
- Although the Premack principle has been criticized for a number of reasons, its derivative, response deprivation theory, still holds great influence in the psychology of learning.
In This Article
The Premack principle states that more probable behaviors will reinforce less probable behaviors. Behavior in itself can reinforce behavior, and the presence of a high-probability behavior can make a low-probability behavior more likely.
For example, an unstudious young child may be incentivized to do their homework (a normally low-probability behavior) if their parent tells them that they can go to the park afterward (something that the child is likely to want to do on their own).
Conversely, the C predicts that the presence of a high-frequency behavior above baseline will discourage the low-frequency behavior. For example, to use the homework example, a child allowed to go to the park excessively is less likely to do their homework (Knapp, 1974).
Prior to the Premack principle, behaviorists believed that reinforcers had a so-called trans-situational nature, and something that is a reinforcer in one context will always be a reinforcer. For example, a pigeon conditioned to seek pellets will always see pellets as a reinforcer, regardless of their abundance.
In contrast, Premack’s view establishes that animals have a preference for ordering of responses, and stronger responses will reinforce weaker responses (Knapp, 1974).
Premack’s research into the Premack principle emerged from his 1959 research on the “rate-differential” or “probability-differential” effect.
The probability-differential effect states that “any response A will reinforce any response B, if and only if, the independent rate of A is greater than that of B.”
That is to say, a response that happens at a high rate can reinforce a response that happens at a low rate. For example, according to this probability-differential effect, a rat will continue to run in a wheel where they are rewarded with water if and only if the probability of the rat drinking water is greater than that of the wheel running (Premack, 1961).
The probability-differential effect has found applications such as making the probability that those with developmental disabilities will engage in exercise greater by making access to games (a high-rate response) contingent upon exercising (Allen and Iwata, 1980).
Premack (1965), following his original research into the probability-differential effect, found that this was accompanied by another: when opportunities to engage in a high-probability behavior reinforced low-probability behavior, the rate of high-probability behavior in itself reduced (because of the barrier of the low-probability activity), and thus incentivized the opportunity to engage in the high-probability behavior.
This is called response reduction. Any response A can reinforce a response B if the rate of A is suppressed below its baseline, regardless of which response happens more often (Mazur, 1975; Timberlake and Allison, 1974).
For instance, in one experiment by Eisenberger et al., people tasked with turning a wheel are more likely to press a bar if the rate of the wheel turning was suppressed below baseline, regardless of how much more frequent wheel-turning was than bar-pushing (1967).
This idea of response suppression then evolved into response-deprivation (Timberlake and Allison, 1974), which has found a number of applications in changing human behavior.
For example, Dougher (1983) found that the response-deprivation of drinking a preferred beverage could increase socially-appropriate behavior in adults diagnosed with schizophrenia when access to drinking depended on this appropriate social behavior (Klatt and Morris, 2001).
Language Training in Chimpanzees
David Premack asked the question in his seminal 1971 article, Language in Chimpanzees, of whether apes could be taught language. Premack believed this to be an important question not just biologically, but for the fundamental question of what language is.
Premack approaches this question in two parallel ways: through a list of examples, and through creating a corresponding list of instructions for training organisms so that they can be taught these examples.
Premack (1971) defined exemplars as aspects of words, sentences, questions, and using language to teach language; concepts of class described through language such as color, shape, and size; the copula, which in linguistics is the word or phrase that links subjects to subject complements (such as is in “The cat is fluffy””); and the logical connector if-then.
Although this list is far from comprehensive, Premack argues
Premack attempted to teach Cebus chimpanzees language through pieces of metal-backed plastic adhering to magnetized slates.
Trainers fed the chimpanzees and introduced the language system pieces gradually, until the chimp placed the plastic piece on a “language board.”
The trainers gave the chimpanzees different types of fruits depending on the part of speech the word represented. Premack then attempted to test whether or not the chimpanzees had formed an association between the plastic objects and the parts of speech that they represented through two different tests.
The researchers found that the chimpanzee was aware of which word went with each fruit.
The researchers changed the fruit donors such that each change in donor was associated with a change in the second language element given to the chimpanzee.
For example, in order to receive an apple with Mary present, the chimpanzee would have to signal “Mary apple,” and in order to receive an apple with Randy present, the chimpanzee had to signal “Randy apple” and not “Mary apple” or “apple Randy.”
Eventually, the chimpanzee in Premack’s 1971 paper was able to construct the sentence, “Mary give apple Sarah” (which would result in the chimpanzee getting the apple) and “Mary give apple Gussie,” (resulting in another chimpanzee getting the apple).
Premack (1971) then used his study of the chimpanzee to test the hypothesis that the acquisition of language is the mapping of existing knowledge. The researchers placed two ups in front of the chimpanzee and attempted to teach them the difference between the words “same” and “different.”
Again, the chimpanzee was largely able to differentiate between objects that were the same and different. The researchers then asked the chimp yes-no questions such as, “is X different from X?: to which the chimpanzee was conditioned to answer no.
Additionally, Premack (1971) was able to introduce such concepts as color, shape, size, and so on through this reinforcement.
Premack’s 1971 paper reinforced the Premack Principle in that the researchers used a high-probability behavior (a chimpanzee wanting to eat various types of fruit) to reinforce a low-probability one (the chimpanzee learning the fundamentals of language).
Researchers have tested the Premack Principle in a number of clinical applications.
In one such study, Horan and Johnson (1971) recruited 96 female undergraduate volunteers and assigned them randomly to one of four treatment groups: a control, and three experimental groups with three half-hour counseling interviews where participants were asked to identify a specific highly probable behavior (such as “sitting down on a particular chair”) and think about a specific negative-positive pair of associations associated with being overweight (such as “a shortened life span” and “clothes fitting better”).
The researchers found that those asked to use the Premack Principle lost significantly more weight than the control or those asked to simply think of the negative-positive association pair throughout the day (Horan and Johnson, 1971).
Premack’s early rat studies
In an earlier study, Premack (1963) controlled rats’ baseline rates of drinking (how often they lick a drinkometer) and running (number of wheel rotations).
Premack found that longer periods of time spent drinking reinforced longer periods of time spent running and vice-versa. Premack used three different concentrations of sugar in the drinking solution (64%, 32%, and 16%) and two different levels of difficulty in running in the activity wheel (weighted either with an 18 or 80 gram weight).
In his observation of rats in a “free-choice” situation, he observed the following preference ordering (Allison, 2019):
16% drink > 32% drink > run with 18 gram weight > 64% drink > run with 80g weight.
In all, Premack ran an experiment using five groups of rats where one of the previous responses was a reinforcer for lever pressing.
In-line with the Premack principle, the results revealed that the average number of lever responses per 10 minute session by the rats depended on which response was used as a reinforcer.
As the second response increased in preference, the performance of a learned response also did. While those given a brief run in the heavily-weighted wheel pressed the lever about 20 times per session, those given a drink of 16% sucrose pressed the lever about 37 times per session (Allison, 2019).
This theory also applies to punishment. Premack claims that punishment involves a situation where a more preferred response is followed by the requirement that someone engage in a less preferred response.
For example, a rat drinking the 16% sucrose solution would be punished if they were then forced to run on a heavily weighted wheel. Premack established in other studies that a punishment system decreases the amount of time that the animal does a preferred response. Thus, any response can reinforce a weaker response but punish a stronger one (Allison, 2019).
However, researchers have since criticized Prremack’s methodology. For example, the experimental design either measured whether the running response was evoked independently from drinking or if this was an effect of the other behavior nor if there were related responses, such as approaching the drinkometer.
This leads to a recurring issue with Premack’s studies where, while deprivation seems to be necessary for the less-likely behavior to happen, this effect is often unmeasured (Klatt and Morris, 2001).
Premack’s study of first-graders
Premack observed first-grade students in a situation where they had to choose between candy-dispensing machines and pinball machines.
The researchers established each of the children’s preferences and they were subsequently identified as players or eaters. The researchers then separated the children into four groups which established a contingency between the candy-dispensing machine and the pinball response, in an attempt to determine whether the children’s behavior could be modified.
The researchers predicting, in-line with the Premack principle, that eating should be a reinforcer for the eaters and playing a reinforcer for the players but not vice-versa (Allison, 2019).
|1a||player||candy-dispensing machine followed by chance to use the pinball machine|
|1b||player||pinball machine followed by chance to use the candy-dispensing machine|
|2a||Eater||candy-dispensing machine followed by chance to use the pinball machine|
|2b||Eater||pinball machine followed by chance to use the candy-dispensing machine|
Aligning with these predictions of the Premack principle, researchers found that there was a congruency effect where the learning success in the two types of children differed.
Only those in groups 1a and 2b were more likely to engage in their non-preferred behavior than they were before, while the other two groups exhibited the same amount of responding on the first machine as they had during the pre-learning session (Allison, 2019).
Although the Premack Principle has had wide appeal in fields such as education (Allison, 2019), academics have supplanted it with alternate theories because of several apparently inconsistent predictions.
Often, the Premack Principle can predict learning in a way contradicts experimental data. For example, Timberlake and Allison (1974) have noted that weaker responses can occasionally seem to reinforce stronger responses, and that stronger responses will occasionally not reinforce a weaker response.
Additionally, there can be additional features of the situation not covered by the Premack principle, where the type of learning that occurs depends on the relative rate at which each response must be performed in order to be able to do the other.
Lastly, some events that do not qualify as responses per say can also moderate the behaviors that follow. For example, shocks can punish behavior, but they are not instrumental responses that animals engage in (Staddon and Ettinger, 1989).
Response Deprivation Theory
In response to these objections, Timberlake and Allison (1974) developed the response deprivation hypothesis. Response deprivation is an element of establishing operations.
According to the response deprivation hypothesis, a weak response can become a reinforcer of a stronger response through deprivation, and that many to most of the studies of instrumental learning conducted by Premack had an element of deprivation.
For example, in the experiment with the pinball and candy-dispensing machines, players who preferred to eat candy in one group were deprived of eating by having to play first.
To consider another example of the response deprivation hypothesis, researchers can deprive a rat of the ability to run. Even if running is not the rat’s preferred response, the rat will then increase its rate of running above baseline when it is able to run again.
Modifying the Premack principle, the response deprivation hypothesis generalizes that an animal will perform a non-deprived or less-deprived activity in order to be able to do a more-deprived activity.
The idea of Allison and Timberlake’s (1974) approach is that animals in a free-choice situation are in a state of equilibrium, able to engage in activities at the preferred levels, what scholars call bliss points.
Being above or below a bliss point can trigger either appetitive or aversive responses. In one of Allison and Timberlake’s rat experiments, rats gravitated toward drinking the sweet solution 60% of the time and the dry solution 40% of the time when in a free-choice situation.
However, when the researchers put the rats in a situation where they had to drink the sweet solution at 10 times the rate of the dry solution, they effectively forced the rats into a level below its bliss point, giving the rat motivation to do things that would allow it to drink more dry solution, even if this is not preferred in a free-choice situation (Allison, 2019)
Allen, L. D., & Iwata, B. A. (1980). Reinforcing exercise maintenance: Using existing high-rate activities. Behavior Modification, 4(3), 337-354.
Allison, J. (2019). The nature of reinforcement. Contemporary learning theories, 13-40.
Cooper, J. O., Heron, T. E., & Heward, W. L. (2007). Applied behavior analysis.
Dougher, M. J. (1983). Clinical effects of response deprivation and response satiation procedures. Behavior Therapy, 14(2), 286-298.
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.
Horan, J. J., & Johnson, R. G. (1971). Coverant conditioning through a self-management application of the Premack principle: Its effect on weight reduction. Journal of Behavior Therapy and Experimental Psychiatry, 2(4), 243-249.
Klatt, K. P., & Morris, E. K. (2001). The Premack principle, response deprivation, and establishing operations. The Behavior Analyst, 24(2), 173-180.
Knapp, T. J. (1976). The Premack principle in human experimental and applied settings. Behaviour Research and Therapy, 14(2), 133-147.
Mazur, J. E. (1975). The matching law and quantifications related to Premack’s principle. Journal of Experimental Psychology: Animal Behavior Processes, 1(4), 374.
Premack, D. (1959). Toward empirical behavior laws: I. Positive reinforcement. Psychological review, 66(4), 219.
Premack, D. (1961). Predicting instrumental performance from the independent rate of the contingent response. Journal of Experimental Psychology, 61(2), 163.
Premack, D. (1965). Reinforcement theory. Paper presented at the Nebraska symposium on motivation.
Premack, D. (1971). Language in chimpanzee? Science, 172(3985), 808-822.
Staddon, J. E., & Ettinger, R. H. (1989). Learning: An introduction to the principles of adaptive behavior: Harcourt Brace Jovanovich.
Timberlake, W., & Allison, J. (1974). Response deprivation: An empirical approach to instrumental performance.