You are on page 1of 15

THE PREMACK PRINCIPLE IN HUMAN

EXPERIMENTAL AND APPLIED SETTINGS

TERRY J. KNAPP*

Behavror Therapy and Research Center, Nevada Mental Health Institute.


P.O. Box 2460. Reno. NV 89505. U.S.A.

(Rewired 9 Drcemher 1974)

Summary-Studies claiming support for the Premack Principle in human experimental and
applied settings are revtewed in terms of the standard test conditions associated with the Prin-
ciple and in terms of the Principle’s conceptual constrains. Little definitive evidence was found
to support the claim that a high probability response will reinforce a lower probability response.
nor was the converse contention supported. that a low probability response will act as a punisher
for a high probability response. More importantly. among the reports reviewed there was no
evidence for the reversibility of the reinforcement relationship.

The Premack Principle (PP) or differential-probability hypothesis (Premack, 1959, 1965.


1971) predicts outcomes of operant response-stimulus consequence relationships in terms
of the relative probability of the two behaviors entailed in any given contingency. If
RLp and RHp represent respectively low and high probability (relative frequency or
duration of responding) responses for an organism, then the PP predicts that RHP contin-
gent upon R,, will produce reinforcement results, and conversely, that R,, contingent
upon R,, will yield punishment results. Reinforcement and punishment results here
refer to the respective increase and decrease in response rate as a function of a stimulus
consequence. In the instance of an organism not initiating a stimulus, as with a food
reinforcer, but rather terminating one. as in shock avoidance, the PP can be extended
to the remaining conditions of negative reinforcement and negative punishment (Ter-
hune and Premack. 1971; Weisman and Premack, 1966). Table 1 presents the four
possible configurations. It necessarily follows from the PP, that by reversing the relative
probability relationship between the operant response and the stimulus consequence
(both of which may be regarded as responses by the organism) the contingency itself
can be reversed. Thus, reinforcement and punishment effects are claimed to be both
,relative to response probability and predictable from response probability.
The standard procedure for demonstrating the PP requires a haselir~e period during
which the independent probability of two (or more) responses is assessed by allowing
the organism unrestricted responding. Responding opportunities may be offered singly
or in pairs, the latter method usually serving to reduce the response frequency compared
to its independent rate. but continuing to maintain the ordinal relationship among
the events (Premack, 1963). The baseline period is followed by the eontirigrrrt phclsr.
during which the R,, is made contingent upon the R,,, and reinforcement results
demonstrated through response rate (RLP) increase. Finally, in the reversing period, the

Table I. Relationship between Response Probability. Response On-set and


Off-set. and Experimental Outcome

REINFORCEMENT PUNISHMENT

Positive Rr.+Gi,,,, RHP-RLP.,,

* Request for reprints should he sent to the author at the Department of Psychology. West Virginia Univer-
sity, Morgantown. West Virginia. 26506.
133
134 T. J. KZAPP

independent response probabilities are reversed, and consequently the response-rein-


forcer relationship reversed. thereby demonstrating the predicted reinforcer and the
reversibility of the reinforcement relationship. A recent animal study has elegantly exem-
plified the utilization of these procedures on a within-session basis (Bauermeister and
Schaeffer, 1974).
Several conceptual constraints, in addition to the standard test procedure, are typically
placed on the assessment of the PP. These require (1) that the subject apply the stimulus
to itself, as opposed to an environmental or experimenter determined stimulus pre-
sentation, (2) that the responding be measured under the conditions of free-operant
baseline-that is, not maintained by externally imposed contingencies of reinforcement,
and (3) that average response probabilities not be employed except in such circumstances
as where the experimenter can demonstrate that the respective response duration-time
curves are approximately comparable (Premack. 1971). Finally, two empirical limitations
have found some support. Demonstration of reinforcement effects may (1) require a
reduction in the total amount of responding to RHP (Premack, 1965) and (2) require
a necessary increase in instrumental responding (RLH), if R,, is to be maintained at
the free-operant baseline level (Eisenberger, Karpman and Trattner, 1967).
Frequent attention is directed toward the PP as an adjunct axiom to the general
principles of reinforcement. Prior 9 the Principle’s inception. applied behavior analysts
relied upon the supposed trans-&iational nature of reinforcers, or other post-hoc pro-
cedures, for the recognition of reinforcing events. The PP is alleged to alleviate this
difficulty by supplying the clinician and educator with a simple and effective technique
with which to identify reinforcers for any given individual: Simply observe the behaviors
that the organism engages in at a high response rate, and such behaviors will serve
as reinforcers for any behavior occurring at a lower rate of response. Advantages which
would flow from the empirical verification of the PP are often cited (Rachlin, 1970;
White, 1971; and Danaher, 1974), and they may be summarized as (1) giving a partial
answer to the often asked theoretical question, “why is a reinforcer reinforcing?” (2)
providing an independent method of identification and measurement of a stimulus rein-
forcer’s potential, (3) allowing an easy means with which to identify and utilize reinforc-
ing events in the natural and applied setting, (4) illustrating the conceptualization of
reinforcers as behaviors (eating, drinking, running, copulating, etc.) as opposed to static
objects or events (food, water, exercise, sex, etc.), and (5) permitting the use of reinforcing
events which may be more natural to the organism.
Despite the frequent citing of the PP in introductory operant and behavior modifica-
tion textbooks (Bandura, 1969: Kanfer and Phillips, 1970; Liberman, 1972; Sherman,
1973; Sulzer and Mayer, 1972; Tharp and Wetzel, 1969; Whaley and Mallot, 1971)
and extensive experimental animal studies (Bauermeister and Schaeffer, 1974; Holstein
and Hundt, 1965; Harrison. 1970; Hundt and Premack, 1963; Premack, 1959, 1962,
1963.1972 : Premack. Schaeffer and Hundt. 1964; Schaeffer, 1963. 1967; Weisman and Pre-
mack, 1966) relatively few published studies evidence the investigation of the Principle
with human subjects or the application of the Principle in natural settings.
This paper serves to review studies of the PP which employed human subjects in
an experimental setting or which assessed the PP in an applied or natural environment.
Each such study is critically evaluated in terms of the previously outlined standard
test procedure and conceptual criteria. The investigations reviewed may be partitioned
into three broad categories: those which intend to assess the PP directly, those which
assess the Principle within the context of Homme’s Coverant procedure, and those which
report a R,, reinforcing a R,,. but with neither an independent assessment of response
probabilities. nor a reversal of the response-reinforcer relationship.

DIRECT EVIDENCE
Direct evidence for the PP naturally sub-divides into three categories: studies which
involved an actual manipulation of the experimental variables in the laboratory, or
The Premack Prmciple 135

Table 2. Manipulation studies directly assessing the Premack Principle

Stud, SUbJeCtS M ‘CS* GDlSSt Rcsponx I Response ?

Premack. 1959 Children M GD Pm hall machme Chocolate cand\


Mclntrc. 1963 Students M GD Hlph frequency words
Bcrgei. 1965 Children M GD Micro-sultch-wsual
Schaeffcr cf ~1. 1964 Students M ss Lever press CRF Lever press FRS
Elsenberger CI al, 1967 Students M GD Wheel turnmg Bar prcssmg
Was,k. 1967. 196X. 1969 Students M SS Lever press: CRF. FR5. 10. 20 Lever press: CRF. FR5

Ayllon & Arrm IY6h PXl,ClIl\ v GD Tokens


Fratr. IYhX Rcurd.lla M ss Play acLl”ltles Task duration. mappropnatc
bchawor
Wa,,k 1969 Student, M GD Lever press’ CRF Lever press F-R.‘. IO.
20. 40. xu
Johansson <‘I of. IY69 PatlCIltS M-CS SS Depresswc verbal behavEor Lou raw verbal heha\,“rs
8r BC1I”I1ICS
Incc. 1969 PatlenIS cs ss Attendance. physxal therapy Attcndancc. speech thcrap)
Lewnsohn cl al.. 1969 Patlcflt cs Depresswe verbal hehavlor Vocational goals
Roberts. 1969 Pat,c!lt M ss Posmre verbal hchawor C,garette put%
Walk. 1970 Children M GD Acadcmx hehavtors PIa) actlrltles
Hoit. 1971 Retardates M ss Acadcmac behawors Play actl”ltlcS
MacDonald. 1972 Chddren M ss Prclerred 1”) Non-prclcrrcd to)
Robmson & Lcwnsohn 1’)‘: Patients M GD Depreswe verbal behawor Low raw verbal hehawor
Har~Jc. 1973 Chddren M GD Academic behawors Academx behariors
Mltchcll & StofTclma!r. lY7.2 Pd1IC”1S M SS Slttl”g COll strlpplng

* Manipulation/case study.
t Group design&ingle subject.

in an applied setting, and a set of investigations reported only as case studies or anecdo-
tal applications of the principle (see Table 2).

Manipulation studies-laboratory settings


Some of the experimental studies employ mechanical manipulandum reminiscent of
investigations in the animal literature. The earliest such report was conducted by Pre-
mack (1959). Thirty-three children. average age 6.7 yr, were first given unrestricted access
to freely delivered standard size chocolate candies (E) and to a pin-ball machine manipu-
landum (M). Under the baseline phase. the R,, was pin-ball manipulation for 61:;
of the subjects, while eating candy was the R,, for the remaining 30% of the children.
&., and &, classifications were made simply in terms of which response had the
greatest occurrence. During the experimental condition the R,,, whether eating or mani-
pulating, was made contingent upon the R,, under a CRF schedule. The control condi-
tion provided subjects with free access to the R,, to determine whether the R,, rate
might increase independent of the contingency. The mean increase in the number of
R,,‘s from the baseline to experimental treatments was statistically significant at the
0.01 level for both experimental groups in comparison with their respective control
conditions. Thus, it is alleged that R,, contingent upon R,, increased the frequency
of the latter response. hence. R,, acted as a reinforcer for R,,.
While this study seemingly lends credence to the PP, a clear interpretation is not
possible due to one major defect: lack of the appropriate control group. For example.
in the experimental M contingent upon E condition, an appropriate control would
entail locking M, while allowing access to E-that is, a control allowing for potential
increases in E’s (RJ rate as a function of the non-availability of M (RHP), as opposed
to the contingency of M.* The control condition that was provided did assure that
R,, did not increase independently of the contingency, however, the contingency, as
Premack later discusses (1965) is a complex relationship, one component of which is
the dependency between R,, and R,,, and another of which is the restricted access
to R,,. The lack of the appropriate control condition (namely, locking the R,,) is
a common fault among tests of the PP. although there are two exceptions (Eisenberger.
Karpman and Trattner. 1967; Robinson and Lewinsohn, 1973) discussed later. The em-
pirical. aside from design. importance of such a control procedure can be gleaned from

* I am indebted to Prof. R. Allen Gardner for first calling to my attention the appropriateness of such
a control procedure.
136 T. J. E;\APP

an animal study by Premack and Premack (1963) in which rats dcprivcd of the oppor-
tunity to run in an activity wheel increased their daily food intake. The authors suggest
that “when the rat is deprived of a behavior that recurrently comprises a large part
of its total daily activity.. an increase may occur in some other behavior.” (p. 209).
Presumably. the same thing might be said of a sub_ject deprived of a large portion
of available responding within the confines of an experimental time constraint.
What followed Premack’s first report was a series of investigations conducted within
the controlled environments of experimental laboratories. Most such studies employed
sub-human species as previously cited. McIntire (1963) was the first to attempt a replica-
tion of the PP employing human subjects. He utilized a paired-associate learning task.
Forty-five college students were assigned to one of three conditions: Group H. in which
a frequently occurring word. as measured by the Thorndike-Lorge word list. followed
each paired-associate; Group L. in which a low frequency word followed the paircd-
associate. and Group C. the control. in which no word followed the pair. McIntire
reasoned that the frequently occurring words should reinforce the paired-associate. hence
Group H should learn the task faster than Group L. and Group L faster than the
non-reinforced control. Group C. Group H did learn significantly faster (I, < 0.01) than
Group L. but did no better than the non-reinforced Group C. McIntire attributed
the discrepancy to possible intertrial interval practice by Group C subjects. He alleged
general support for the PP. However, the estimation of response frequency on the bases
of a word list does not meet the conceptual criteria outlined by Premack. particularly
with respect to non-contingent (free-operant baseline) and momentary response probabi-
lities.
Berger (1965) in an unpublished study provided 30 fourth grade students with access
to two micro-switches. Responding on the first switch produced an S-set exposure to
the visual component of a film strip. similar responding on the second switch produced
the corresponding audio component for 8 sec. The students were assigned to one of
two groups: a free response group in which both responses were available non-con-
tingently during two 20-min sessions. or a contingent response group which was exposed
to a series of conditions, free responding, R HP (as determined by free responding probabi-
lities) contingent upon R,,. and a return to the baseline conditions of free responding.
The experimental group produced a mean increase in R,, from baseline to contingent
condition of 29.9 responses which is significant at the p < 0.01 level. The control group
remained relatively stable from session 1 to session 2, decreasing during session 3 by
a mean R,, of 7.4 responses. Unfortunately. as in the earlier Premack study. the appro-
priate R,, locked control group was not included in this experiment.
Employing what was later to become a standard procedure, Schaeffer. Hanna and
Russo (1966) had two subjects press a telephone key for points under a CRF and
FR5 schedule of reinforcement. As would be expected, during baseline conditions the
FR.5 was the R,, and the CRF the R,,. When either was put on extinction, their
respective rates dropped. and when the contingency R,, produces R,, was arranged,
RLP increased in rate of responding and later returned to baseline levels in the absence
of the contingency. While this experiment also lacks the RHP locked control. it does
demonstrate. as the authors suggest. that the PP generalizes to responses which both
depend upon the same reinforcers. They also suggest that the rate of response in the
contingent period is a function of both the independent rate of R,, and R,,.
A year later. Eisenberger. Karpman and Trattner (1967) conducted one of the few
studies which claim to offer evidence for a modification of the PP. During the first
experiment 36 of 39 college students spent more time turning a wheel than pressing
a bar. when both were presented concurrently for 5-min periods. An R,, contingent
upon R,, was then arranged for periods of 5 min. Only seven of the 18 subjects who
were above the group median for wheel turning increased their bar press rate. while
14 of the I8 subjects who were below the median for wheel turning increased their
bar pressing rate Thus. the R,,, acted as a reinforcer only for those subjects having
a relatively low rate baseline for the instrumental response. The authors conclude that
The Premack Prlnclplc 137

the PP predicted correctly only when the contingent behavior was suppressed beneath
the free-operant baselevel.* Reinforcement is attributed to the necessity of an increase
in the instrumental response rate if the contingent responding rate is to be maintained
at or near the free-performance level. Subsequent experiments in the study offer support
for such an interpretation. assess the effect of instructions. and finally present the appro-
priate. locking RHP, control procedure.
In this last experiment subjects were given the same five minute concurrent baseline
period. followed by a 5-min period in which the R,, was available. but during which
the subjects were told not to manipulate the R,,. Under such conditions all subjects
decreased in their R,, rate. Thus, preventing R,,, rather than making it contingent.
did not result in reinforcement. apparently the contingency is necessary. However. the
subjects were instructed not to emit. rather than prevented from emitting. the R,,.
the latter might be a preferable procedure.
In ,a similar, but more recent set of studies by Wasik (1967. 1968. 1969) the PP
was supported. The first investigation involved 34 college students participating in four
different experiments. The first two employed levers. one on CRF. the other on FR5.
which produced points on counters. Concurrent baselines were determined, then the
R,, (CRF) made contingent upon the R,, (FR.5). Later the contingency was reversed,
and still later returned to baseline conditions. The data supported the PP. The last
two experiments of this study employed multiple response levers on CRF. FR5, FRlO.
FR20. After determining baseline rates. all possible contingencies were arranged. Results
showed that the R,, (CRF) reinforced all R,, responses, while the intermediately prob-
able responses (FR5 and FRlO) reinforced LP responses, and hence were both reinforce-
able and reinforcing. The final experiment replicated the earlier animal multiple response
chain study of Schaeffer (1967). and concluded that the PP needs revision when applied
to response chains. The revision notes that for any response within a chain the R,,
will reinforce all chained combinations of R,,.
A second set of investigations (Wasik, 1969) extended the earlier work on human
lever pressing to the interaction effects of FR (5. 10. 20, 40, and 80) schedules and
contingent time (2. 5. 10. 20). the duration during which R,, was available. College
students were again employed as subjects. and the same procedures were utilized. As
CT increased, the rate of R,, respondmg (CRF) did also, while the R,, (FR) responding
decreased. If the FR (R,,) requirement were increased. R,, (CRF) responding lessened
and FR responding increased. When the R,, and R,, responses were combined. total
responding was relatively constant across increases in FR and CT values, indicating
that the effect of such increases is primarily upon the distribution of responding between
the R,, and R,, and that the rate of responding is high and constant regardless of
changes in FR and CT values. Thus. this second set of investigations represents a para-
metric study rather than a direct assessment of the PP. and again, the R,, locked
control was not included.
Thus far the laboratory investigations have largely utilized arbitrary manipulandum.
experimental work on verbal behavior was introduced with a paper by Johansson. Lew-
insohn and Flippo (1969). The verbal behavior of three paid subjects scoring high on
the depressive scale of the MMPI, and having elevated ratings on the Feelings and
Concerns Checklist was assessed during three 30-min free-talk sessions. Subjects were
subsequently seen in seven to ten 30-min interviews in which they were instructed that
high probable verbal behavior (RHP) could be emitted only in the presence of a green
light which would be turned on only at certain times during the last 15 min of the
* Allison and Timberlake (19751 in an experimental analysis and in a later review paper (Timberlake and
Allison. 1974) present additional pcrsuasix evidence and arguments to the effect that. “the probabiiity-
differential condition is neither necessary nor sutiicirnt for instrumental performance”. (1974, p. 154). Rather.
they argue. following Eiscnbcrger. Karpman. and Trattner (1967). that instrumental responding is B function
of response deprivation. The latter concept is “defined to occur if the animal. by performing its baseline
amount of the instrumental response. IS unable to obtain access to its baseline amount of the contingent
response.” (1974. p. 152). Thus. remforccment effects occur not because of a probability disparity between
the two responses. but because of the contingency imposed requirement of increasing instrumental responding
in order to gain access to the same level (baseline) of contingent responding.
138 T. J. KNAPP

interview. Green light time was then made proportional to the R,, verbal behavior
emitted during the first 15 min of the interview. In a final phase the subjects were
placed on a DRO schedule in which verbal behavior other than the R,, and R,,
categories was reinforced. The control subject was yoked for non-contingent green light
time. The experimental subjects increased low frequency categories. the control did not.
During the DRO phase, the R, remained higher for one experimental subject, and
returned to a low rate for another. It remained low for the control.
Unfortunately, while this study pioneered in the examination of verbal behavior. a
definitive interpretation of the results is not possible for two reasons: (1) The increase
in R,, could result from restricting R,, rather than making it contingent. that is. the
R,, locked control procedure was not utilized. (2) The green light is a stimulus change
confounded with the access to R,,.
The most recent laboratory investigation (Robinson and Lowinsohn, 1973) extended
the work of Johansson er al. (1969). Twenty depressed females were assigned to one
of four treatment conditions, each involving R,, of depressive talk and R,, of low
frequency verbal behavior. Group 1 was a standard PP treatment. R,, contingent upon
R . Group II was a contingent group in which LP was consequated bv another R,,.
Ggup III, termed deprivation group. was a random restriction of Rip. the second
time the appropriate control has been employed. The results showed that the PP group
significantly different from the no treatment control (p < 0.05) and showed significant
gains over its own baseline. The contingent group did not differ significantly from the
control. The deprivation group produced results paralleling those of the PP treatment.
Thus, when the appropriate control condition, preventing RHP, was included. that group
did not differ from the PP treatment group.
In summarizing the results of manipulation studies in laboratory settings. it is evident
that the investigations generally claim systematic replication of the PP across a variety
of responses and subjects. however, when the evidence is restricted to those studies
employing the appropriate control procedure, no definitive conclusions are possible.
One set of investigators (Eisenberger. Karlman and Trattner, 1967) employing mechani-
cal manipulation found the contingency between R,, and R,, crucial. while Robinson
and Lowinsohn (1973) examining verbal behavior found no difference between the R,,
locked control and the PP treatment group.

Mar~ipulatior~ studies--natural settings


Ayllon and Azrin in the Token Economy (1968) have presented the most comprehen-
sive program of applied use of the PP. The Principle serves, according to the authors.
both as a mechanism with which to identify potential reinforcers, and as a means with
which to specify contingencies. The authors state the Principle in this fashion. “Probabi-
lity of Behavior Rule: Observe what the individual does when the opportunity exists.
Those activities that are very probable at a given time will serve as reinforcers.” (1968.
p. 60). This is. of course. not a restatement of the PP, but a logical implication given
the PP and the empirical law of effect.
The experiments conducted by Ayllon and Azrin at best serve as “applied examples
or demonstrations” rather than as applied support for the PP. The research design
is similar in each case. The first experiment involves attendance at religious services.
Here. an R,,. based on duration of responding, can only be engaged in by presenting
a token which is earned through engaging in an RLP. The design is ABA: baseline.
contingent on token. return to baseline. With 29 patients as subjects mean attendance
at religious services under baseline was 10.2, under contingency X.6, and under return
to baseline 10.0. Thus. Ayllon and Azrin concluded that religious services were a power-
ful reinforcer for some patients since only a slight difference in attendance occurred
when it was contingent on tokens. The removal of the token requirement, and sub-
sequent return to baseline, demonstrated. say the authors, the relationship between the
token requirement and reduction in base rate.
The Premack Prmclple 139

In a later set of experiments, the design was changed only slightly. A became the
token contingency. Thirty-seven patients participated in a token requirement for attend-
ance at a commissary. Baseline mean attendance was 16.5, free access rate 21.0. and
contingency reimposed 17.3. The same number of patients served in a going out of
doors study. A one token contingency produced a mean rate of 5.5, free rate was 5.5,
and contingency reimposed 5.8. Forty-five patients took part in a musical activities
study. Mean attendance was ten under the contingency, 14 under free access, and 12
under return to baseline. A final study involved watching motion pictures. Contingency
requirements of one token resulted in a mean of eight of 45 patients attending movies.
free access 13, and return to token conditions ten. The authors conclude that these
studies show that the activities examined could be used as effective reinforcers for the
patients involved because they did not lose their effectiveness, that is, decreased only
slightly, when a one token contingency was imposed. And finally, the authors noted
that’the ABA design showed the slight decreases that did occur were related to the
token requirement.
The Ayllon and Azrin study fails to meet the standard procedures on several counts.
(1) There is no independent measurement of response probability, except for R,, in
experiment 1. (2) There is no test of the R,, as a reinforcer for any behavior. (3)
There is no reversal of a R,, and R, relationship. (4) In only one of the experiments
was the behavior a reinforcer for over SO’/, of the patients. (5) Finally, maintenance
of responding at a near free-access level during a response cost contingency does not
require a PP interpretation.
Roberts (1969) applied the PP to a 40-yr-old male paranoid schizophrenic who had
a long history of institutionalization and a phobia of confinement. The experimenter
observed that during confinement training, sessions in which the subject was to remain
in a stairwell in an attempt to extinguish the phobic behavior, a high rate of cigarette
smoking occurred. The experimenter decided to make cigarette puffs contingent upon
positive verbal statements emitted by the subject with respect to the benefits of confine-
ment training. Baseline produced about 0.7 positive verbal statements per min. 0.5 nega-
tive statements per min. and 2.7 cigarette puffs per min. The contingency increased
positive statements to a 1.6 per min, decreased negative to 0.2 per min. The results
generalized to a locked office and locked utility room. However, the experimenter
recorded all of his own data, conducted no reliability checks, and failed to return to
baseline conditions. although the generalization tests might be viewed as a multiple
baseline across settings.
The PP was employed by Frair (1968) in the contingency management of five non-
institutionalized trainable retardates. R,, were determined by making observations for
1 week, these were then subsequently depicted in cartoon drawings comprising a re-
inforcement menu (Addison and Homme, 1966; Daley, 1969). The R,,‘s were tasks
derived from the Peabody Language Development Kit and in a second experiment
desired increases in appropriate behaviors and decreases in inappropriate behaviors.
When R,, was made contingent upon completion of R,,, the time devoted to task
behavior increased. as did appropriate behaviors. Inappropriate behaviors when conse-
quated decreased. Unfortunately, the experimenter reports only informal observations
in estimating R,, probabilities, no independent estimates of R,,, no reliability data.
no return to baseline conditions, nor a multiple baseline across subjects.
Another applied study is best placed within the group administered contingency litera-
ture although it bears Premack in its title. Wasik (1970) coded (Coping Analysis Schedule
for Educational Settings) appropriate and inappropriate behavior under two conditions
(individual and group), of 19 second-grade culturally deprived children under a 3-day
baseline condition. The study employed an ABAB design. Condition B made appropriate
behaviors reinforced by free-time activities (toys, games, and crafts). Appropriate beha-
viors did increase during the contingent (B) condition and returned to baseline during
reinstatement of the A, and increased again in returning to B. The suggestion was
made that R,, can be determined by (1) asking the children to name those things
140 T. J. KNAI’P

they like. and (2) making available simultaneously many responses that are potential
reinforcers (reinforcing menu). While these are excellent suggestions in applying the
PP, they can not substitute for the actual test conditions necessary for a demonstration
of the PP.
In another utilization of the reinforcing event menu, Holt (1971) was able to decrease
the mean time to task completion of academic behaviors of two retarded children.
R,,‘s, for example, coloring, playing with dolls, were “empirically determined” in an
unreported manner. and then pictured schematically on 5 x 5 cards. One such menu
represented R,,’ s. and others (2-5) ranged in descending order to the picturing of R,,‘s,
e.g. academic tasks. Subjects were then required to complete a fixed unit of academic
behavior before gaining access to 5 min of R,,. Subjects developed about an 80”; reduc-
tion in time to task completion with a 90s; correction criterion. Then menus 2-5 were
faded in. thus reinforcing academic tasks with academic tasks. While the study claims
support for the PP. and does offer the innovations of fading in R,,, no clear support
can be found. Neither an ABA design nor multiple baseline across subjects was
employed. hence, the results could be attributed to the instructions, requirement of
907; success before being released from the task, or simple practice effects. Thus. it
is not possible to justify a claim of reinforcement effects. let alone a PP interpretation
of such effects.
MacDonald (1972) employed preschool children as subjects and toy preference as
responses in an attempt to demonstrate the PP. Toy preference was measured under
baseline conditions of concurrent availability of three toys, and subsequently, concurrent
availability of the R,, and R,, selected on the bases of baseline condition 1. The
per cent of the total time in contact with a particular toy became the estimate of
response probability. Most subjects were then placed under the first contingent condition
(IL + R HP), and later a verbalized contingency condition. Of the six subjects, one was
dropped due to instability in baselines, four failed to show the expected contingent
effect. a fifth showed the contingent effect only during the phase when the contingency
was explicitly verbalized (two other subjects failed to replicate this finding). and the
final subject demonstrated a reinforcement effect during the contingency condition. This
last subject also showed decreases in response probability under extinction. variability
under a non-contingent scheduled. and a return to baseline when it was reimposed.
The great variability across subjects speaks to limited experimental control.
Testing the PP in an academic setting. Hartje (1973) had 108 fifth-graders indicate
their preference for the subject matters of reading, social studies. and science. Then
students were assigned to one of three conditions: reinforcement, in which the subject
matters were scheduled in a R,, to R,, preferential sequence. punishment in which
the students began with the R,, then sequenced to the R,,, and a control condition
in which the subject matter sequencing was self-selected. The reinforcement group
showed a statistically significant increase in their preference for R,, over a pretest mea-
sure on a I-10 scale point scale, and the punishment group showed a statistically signifi-
cant decrement in preference for R,, over the pretest scores. However, the control
group also showed similar changes in preferences making any interpretation dubious.
The most recent publication examining the PP in an applied setting (Mitchell and
Stoffelmayr. 1973) comes from Aberdeen, Scotland. The subjects were two long-term
schizophrenics who engaged in little or no activity, and in addition consistently refused
such potential reinforcers as candies. cigarettes, fruit, etc. The R,, selected was a stan-
dard work-task engaged in by many patients on the ward; it consisted of unwinding
wire from a tightly wound copper coil. R,, was measured employing 30-set interval
recording for 30 min each day for 40 sessions. Behavior was recorded under conditions
of baseline (six sessions), instructions (three sessions), instructions plus reinforcement
(twelve sessions), return to instructions only (twelve sessions), and a return to instructions
with modified reinforcement (seven sessions). During baseline coil stripping occurred
in about 3.5)‘; of the intervals for each subject. and remained at about the same level
under the instructions condition. During the next phase, sitting (RHP) was made contin-
The Premack Principle 141

gent upon coil stripping by asking the patient to stand. strip a few inches, then allowing
him to sit down again for 90 sec. Stripping under these conditions increased and stabi-
lized at about SOo/,of the intervals across twelve sessions. The instructions only condition
was reinstituted and coil stripping dropped to near the baseline level. Finally. the sub-
jects were instructed to work, with work materials removed from them contingent upon
completion of a determined amount of stripping. This served to maintain the behavior
at about 89 and 67% of the intervals respectively. even though both patients were allowed
to remain seated throughout the period. This is a most interesting study because the
R,, is such a common one in an institutional setting. One is hard put to find a behavior
in institutions which exceeds the rate of sitting. Given the final condition. however.
it is not clear that sitting was acting as a reinforcer. It certainly appears to have served
as such during the shaping period. However. in the final condition it was sufficient
to remove the working materials, thus brief escape from the task may have maintained
the coil stripping.
In summarizing studies of the PP undertaken in the natural environment. there
appears even less justificatory evidence than in the case of laboratory investigations.
In general the applied studies employ inadequate group or single-subject designs, and
tend toward rather spurious and speciously derived applications. Clear support for the
PP will have to come from elsewhere.

Case studies and anecdotal accounts


Case studies and anecdotal reports of the PP’s employment in clinical and educational
settings are finding increasing acceptance in behaviorist and clinical journals. The fre-
quently cited Homme, deBaca. Devine. Skinhost and Rickert (1963) paper is the earliest
of these reports. No quantitative data are presented, but rather a descriptive account
of the PP as applied to the behavior of nursery school children. RHp’s. running around
the room, screaming, pushing chairs, or working on puzzles, were made contingent
on R,,‘s, sitting quietly and looking at the blackboard. The authors noted that they
were able to predict what behaviors would be reinforcing, since most such behaviors
had been previously punished. The authors concluded “even in this preliminary. un-
systematic application, the Premack hypothesis proved to be an exceptionally practical
principle for controlling the behavior of nursery school Ss.” (p. 544). The importance
and value of Homme’s report does not lie in providing support for the PP. since such
anecdotal descriptions fail to meet any of the standard test conditions, but rather in
providing illustrations of how the principle, if supported by appropriate experimental
work, might be employed in natural settings.
Similarly. another series of Homme reports are often cited (e.g., Wasik, 1970) as having
demonstrated the use of the PP in controlling the behavior of psychotics (Homme
and deBaca. 1965) and middle class children (Homme and Tosti, 1965). The reports
themselves are difficult to obtain; they were only recently published as a part of the
Behaaior Technolog_v series by Homme and Tosti. The Homme and deBaca (1965)
paper contains two descriptive accounts of PP employment. The first concerns
a 14-yr-old female patient hospitalized for 1Oyr. The R,,‘s were cranking her bed up
and down. brushing crumbs off it. and putting her shoes on. These were reinforced
by showing her a doll. pictures. and allowing her to visit another patient. In the
second example, a 16-yr-old blind female patient hospitalized since she was 8 yr was
reinforced for talking in increasingly longer phrases (RLP) by allowing her to touch
the experimenter’s hand (RHP). This is the extent of the descriptions, no data. no details,
no follow-up are provided in either case. The second study (Homme and Tosti, 1965)
contains no actual examples, but mere hypothetical illustrations of how the PP might
be used in certain situations. Despite the anecdotal nature of these reports, they are
frequently cited as evidence supporting the PP.
Lewinsohn, Weinstein and Shaw (1969) offered the first reported clinical case study.
The client was a 22-yr-old male divorcee whose high frequency verbal behavior (depres-
142 T. J. KNAPP

sive talk) was made contingent upon his completion of specific vocational goals estab-
lished by the therapist. For each hour of therapy. and hence. access to an audience
supportive of depressive talk. the patient had to complete tasks such as talking to
his vocational counselor, arranging for a loan. making employment inquiries. etc. Failure
to do so resulted in termination of therapy for that session. According to the authors
several important vocational decisions were made by the subject during the course
of therapy, and he himself reported “gaining many insights”. In a similar vein to Lewin-
sohn et al. (1969). Johansson et al. (1969) reported a case which also employed depressive
speech during therapy as the R,,. and the number of activities engaged in the week
previous as the R,,. Improvement is reported in terms of pre compared to post MMPI
profile.
Ince (1969) made attendance at physical therapy (RHP) contingent upon attending
speech therapy (RLP) for a 68-yr-old male, and in a similar arrangement, for a 68-yr-old
female, made attendance at physical therapy contingent on presence at both occupational
therapy and speech therapy. Within a few days both patient’s attendance at the RLP
therapy increased from zero to lOOo/
All such case studies and anecdotal accounts suffer from the same obvious weakness:
no behavioral data. The cases are often written in such an abbreviated fashion. with
only impressions employed to assess outcome, as to contribute little to the literature.
For example, in a case study employing coverant techniques (Seitz, 1971). the therapist
indicates that according to staff reports, psychological tests, and the patient’s own report.
eight hourly sessions produced “dramatic improvement.” It has even been claimed that
Ben Franklin employed the PP (Knapp and Shodahl, 1974). In such instances as these,
it may be more accurate to say that the existence of a principle served to control
the author’s verbal behavior. while the behavior of the subjects under examination may
have been controlled by a variety of variables other than those subsummed under the
rubric: Premack Principle.

COVERANT STUDIES
Lloyd Homme in a series of papers (1965, 1966) systematically extended the PP in
two directions. First, he offered a set of amendum (1966) derived in large part from
Guthrie’s contiguity theory (1952). Secondly, he presented an analysis of operants of
the mind. or coverants as he named them. others have termed such responses private
events, mental events. or simply thoughts and ideas. Homme suggested. for example.
that smoking behavior might be modified by following smoking stimuli with an anti-
smoking coverant. followed by a pro-non-smoking coverant, and finally followed by
an R,, which takes the subject out of the stimulus situation. The PP allegedly provides
“the easy availability of a large number of self-reinforcers.” (1966, p. 503). Following
Homme’s lead, a number of investigators lay claim to the utilization of the PP in
assisting with the modification of behavior through coverant control. These studies are
reviewed in this section (see Table 3).

Table 3. Studies employing Homme’s coverant technique

Study Suh,rcts M/CS* GD’SSt Response I

Tooley & Pratt. 1967 CS ss Anu-smokme thouehls


Horan & Johnson. ,970 (IL),,I
Lawson & May. 1970
Tyler & Straughan. 1970 GD Ne@~ve and po&
non-eatmg thoughts
Johnson, 1971 _. Readq statement.
pos,t,ve sell-thought
Positive self-thoughts

(‘S ss Posmvc sell-thoughts


CS Pos~twe sell-thoughts

* Manipulation/case study.
t Group design/single subject
The Premack Prmciplc 143

In the first of these studies. Tooley and Pratt (1967) assessed three procedures: covert
sensitization, Cautela’s technique (1966); contingency management, the Homme coverant
technique; and contractual management. These procedures were applied in succession
to the smoking behavior of two married professionals. In the Homme procedure the R,
was anti-smoking coverants, such as “smoking is a bad influence upon my children,”
these were followed by an RHP. drinking coffee in one case and drinking water in the
other. The number of cigarettes smoked per day dropped from a baseline of 50 for
subject 1 and 35 for subject 2 to zero for each subject. No attempt was made to sort
out the contribution of each procedure, however, nor was there a long term follow-up.
Lawson and May (1970) continued in the attempt to compare the Homme technique
to other procedures. Twelve student volunteers were distributed among four conditions
for a 5-week treatment period designed to decrease the rate of cigarette smoking. One
group received the Cautela (1966) covert sensitization procedure. another the Homme
coverant technique. Here the R,, was subject-determined thoughts which were “incom-
patible with smoking”. the R,, selected was entering a’building. A third group nego-
tiated contracts specifying times and places where they would go without smoking.
The fourth group simply counted the number of cigarettes they smoked and discussed
their progress at each treatment session. The outcome was assessed in terms of the
mean number consumed as a percentage of the pretreatment baseline. While all groups
showed a statistically significant reduction in consumption, no significant differences
were found between the groups. In addition, subjects were “permitted to supplement
the treatment with any secondary strategies for quitting they found useful”, hence, the
treatment conditions were not pure. Nor was any follow-up data available or an indepen-
dent assessment of R,, made.
Attention next turned to the use of Homme’s technique in the control of weight
(Horan, 1971). Ninety-six female coeds were randomly dispersed among four treatment
conditions. Group I was a delay treatment control, Group II a placebo control. Group
III paired negative and positive statements (e.g. a shortened life span. clothes fitting
better) seven times daily, and Group IV, emitted such pairings prior to the commission
of a selected R,, (e.g. sitting down in a particular chair). An analysis of variance showed
the only significant pair-wise comparison was between the delayed treatment group
and Group IV. The authors concluded that the applicability of the PP in self-manage-
ment was unconfirmed.
Tyler and Straughn (1970) investigated the Homme technique in contrast with two
as yet unutilized control procedures. Fifty-seven female volunteers from a self-help
weight reduction club were distributed across the three conditions. The Homme covert
strategy was one group. a breath-holding technique found useful in anti-smoking studies
the second group. Under this condition subjects would image a tasty food, and then
hold their breath for a period of time. The third group was taught a modification
of the Jacobson relaxation procedure. None of the groups produced a significant reduc-
tion in weight loss. The coverant group did best with a mean loss of 0.75 lb. followed
by the relaxation group with a mean loss of 0.53 lb, and the breath-holding group
with a mean loss of 0.43 lb. A no treatment control was unfortunately not employed,
thus it is difficult to draw any conclusions from this study.
The coverant literature contains a number of case studies. Johnson (1971) reports
the case of a 22-yr-old male college junior who was required to read statements and
rehearse potential interactions with his girl friend (RLP), and then follow them by R,,‘s
such as eating and driving after stopping. The procedure is reported as being very
effective, although no data are offered. In a similar case, a 17-yr-old male college fresh-
man made urination (RHP) contingent upon reading non-depressive thoughts. After 2
weeks the subject reported the depressive thoughts dropped to near zero.
Seitz (1971) reports a very similar case involving a 36-yr-old widowed male suffering
from depression following a suicide attempt. Simultaneously four treatment procedures
were employed: systematic desensitization, assertive training, sensory deprivation. and
finally Homme’s coverant procedure. Smoking (RHP) was made contingent upon thinking
144 T. J. KNAPP

one of six positive self-characteristics (R,,). Following 2 months of such treatment the
patient dropped 60 Tscores points on the depression scale of the MMPI. and improved
according to staff and self-reports. Unfortunately. medication was simultaneously being
administered throughout the 2-month treatment period. hence, it is impossible to draw
any definitive conclusions.
Mahoney (1971) reports a self-management technique employing a modified Homme
procedure used by a 22-yr-old male outpatient exhibiting compulsive and depressive
thoughts. Smoking was made contingent upon reading positive self-thoughts which were
attached to a cigarette package. Such thoughts increased from a baseline mean of
two per 2-hr time block to 56 per 2-hr time block. The patient was reported as doing
well in a 4-month follow-up.
Finally. Todd (1972) evaluated the coverant control of self-evaluative responses in
an attempt to treat depression. He presents two case studies. In the first a 49-yr-old
female depressive patient was instructed to read, from a list attached to her cigarette
package, one or two positive statements about herself prior to smoking. Smoking was
regarded as an R,, and positive self-evaluative statements as an R,,. The second case
concerned a 27-yr-old male college student suffering from test anxiety. He was told
to read one or two positive statements about himself before making a telephone call
(a RHP due to job requirements). Within 6 weeks the patient reported improvement.
It should be quite evident that the coverant studies employing Homme’s extension
of the PP entail a number of difficulties. Danaher (1974) and others (Danaher and
Lichtenstein. 1974; Mahoney. 1970) have reviewed these and pointed out numerous
such difficulties. principle among them. the assessment of baseline rates while the re-
sponse is extrinsically maintained.

INDIRECT STUDIES
Some researchers (e.g. McDonald, 1972) have cited as support for the PP studies
in which a R,, reinforced a R,,, but without either an independent assessment of
response probability, or a reversal of the response-reinforcer relationship. Undoubtedly.
a perusal of the literature would uncover any number of investigations in which rc-
inforcement effects are demonstrated. and in which the reinforcer may be identified
in some sense as an R,,. Two of the most frequently cited studies in this category
are reviewed here as examples of investigations of this sort.
Bushell, Wrobel and Michaelis (1968) examined the effect on study behavior of partici-
pation in special events. Twelve children enrolled in a preschool earned tokens by work-
ing at various academic tasks. The tokens could then be used to purchase a special-event
ticket which allowed for participation in activities such as trips, movies. art projects.
stories. and the like. Through an ABA design the researchers clearly showed the effective-
ness of backing up the tokens with special events, that is, the special events increased
the percent of time the children devoted to study when studying produced tokens. It
is because the special events varied from day to day, but were always the sorts of
things children enjoy doing that they have been labelled R,,, and in turn the study
behavior labelled R,,. However. there is little justification for so conceiving the responses
of studying and special events since no independent assessment of their probabilities
or rates were undertaken. nor was there an attempt to reverse the response-reinforcer
relationship. In fact. the authors make no mention of the PP.
This is not true of another study in this category. In treating three cases of anorexia
nervosa. Blinder. Freeman and Stunkard (1971) used access to physical activity as a
reinforcer for gains in weight. Ward observations revealed high rates of physical activity
for these three anorexia nervous patients, and for such patients in general. Prior to
treatment. during the free access to passes period, the patients walked an average of
6.8 miles per day, this compares to the 4.9-mile average for a normal weight person
in the community. Rapid weight gain was produced for the three subjects in less than
The Premack Principle 145

a week after beginning treatment. Thus. it appeared as though activity acted as a rein-
forcer in these three patients. since their weight increased when gains were followed
by passes which allowed the patients to engage in various physical activities. In fact,
the authors cite the PP as justification for viewing high frequency behaviors as a source
of reinforcement. The design, however. does not allow a clear conclusion about physical
activity as a reinforcer since neither a reversal nor a multiple baseline design were
employed. All of the studies in this category suffer from similar defects making them
dubious as sources of support for the PP.

CONCLUSION
An examination of the available manipulation studies. case reports. and anecdotal
accounts seeking to demonstrate and support the PP in human experimental and applied
settings reveals several crucial weaknesses: (1) few of the investigations meet even the
minimum of the standard test conditions, principle among them an independent assess-
ment of the response probabilities; (2) those studies which do most closely approximate
the test technique and conceptual constraints have employed mechanical manipulandum
resembling those used in the rat and pigeon experimental laboratory, for example. com-
paring CRF and FR5 schedules; (3) only two studies utilized the appropriate R,, locked
control, in which the high probability behavior is simply prevented from occurring
rather than made contingent upon the R,, and those two studies have produced contra-
dictory results; (4) many of the studies are plagued by faulty experimental designs either
confounding group treatment conditions, or employing inappropriate or inadequate
single-subject procedures; (5) finally, no study has sought to demonstrate the reversibility
of the reinforcement relationship. which lies at the heart of the PP and Premack’s
account of reinforcement.
Thus, in view of the test conditions and conceptual constraints imposed by the PP
originator. it is difficult to find among the human experimental and applied setting
studies any definitive support for the contention that R,, contingent upon R,, will
produce reinforcement results, and R,, contingent upon R,, will yield punishment
results, where reinforcement and punishment refer to the respective increase and decrease
in response rate as a function of a stimulus consequence. Nor is there any support
for the more important reversibility of the reinforcement relationship.

REFERENCES*
ADDI~KI~ R. M. and HOhlhft L. E. (1966) The reinforcing event (RE) menu. h’atn. Sot. Proyr. Iu.srr. J. 5.
g-9.
ALLISON J. and TIMBCRLAK~ W. (1975) Instrumental and contingent saccharin licking in rats: Response
deprivation and reinforcement. Lcurnrrrc) Motir. 6. 121-142.
A~LLO~ T. and AZRIS N. (196X)T/u, Token Ecofro/n!,: A Moficario& Spsre/n jdr Tlrrrtrpr ad Re/~ahilitafio~~.
Appleton-Century-Crofts. New York.
BASDURA A. (1969) Primiplez or Bchucrnr Modificariorl. Holt, Rinehart & Winston, New York.
BAU~RMEISTERJ. J. and SCHA~FF~K R. W. (1974) Reinforcement relation: Reversibility within daily experimental
sessions. Bull. Ps~chorr. SK.. 3. 206208: also see BAUERMEISTERJ. J. (1970) Positive remforcement: Further
tests of the Premack theory. Doctoral dissertation. Florida State University, Ann Arbor: Xerox University
Microfilms. No. 7 l-6962.
BERGER R. J. (1965) A test or the Premack hypothesis with responses to audio and visual stimuli. Doctoral
dissertatton. Arizona State University. Ann Arbor: Xerox University Microfilms, No. 65-10.371.
BLINDER B. J.. FRKMAS D. M. A. and STUNKARI> A. J. (1970) Behavior therapy of anorexia nervosa: Effective-
ness of activity as a reinforcer of weight gain. A/n. J. Psychiur. 126. 1093-1098.

* Four additional studies have become available since the original draft of this manuscript (BAT~MA~ S.
(1975) Application of Premack‘s generalization on reinforcement to modify occupational behavior in two
severely retarded individuals. .4rn. J. .EIerrr. Defic. 79. 604-610; EDDY J. B. (1974) Application of the Prcmack
hypothesis to the verbal hehavtor of retardates. Doctoral Dissertation, University of Nevada: GLOSS G. G.
(1971) An application of the Premack contingency of reinforcement principle to the behavior of three conduct-
disordered chtldren. Doctoral dissertation. Case Western Reserve University. Ann Arbor: Xerox University
Microfilms. No. 7&?5. 960; SCHAKFFER R. W.. and NOLAN R. J. (1974) Verbal learning and reinforcement:
.4 reexamination of the Premack hypothesis. Bull. Psych. Sot., 2. 431-433). but do not alter the conclustons
drawn.
146 T. J. KNAPP

BUSHELL D.. WR~BEL P. A. and MICHAELIS M. L. (lY68) Applymg “group” contingencies to the classroom
study behavior of preschool children. J. appl. B&r. Anal. 1. 5561.
CAUTELA J. R. (1966) Treatment of compulsive behavior by covert sensitization. Psvcltol. Rec. 16. 33-41.
DALEY M. F. (1969) “The reinforcing menu”: Fmding effect&e reinforcers. In: Bel~a&ral Counsrl/i,ly: Cases
and TechrMm (Eds. J. D. DRUMBOL~Z and C. E. THORESON) DD. I. 4245. Holt. Rmehart & Winston.
New York.’
DANAHER B. G. (1974) The theoretical foundations and clinical applications of the Premack Principle: A
review and critique. B&r. Thrrupy. 5. 307-324.
DANAHER B. G. and LICHTESSTEIN E. (1974) An experimental analysis of coverant control: Cueing and consequa-
tion. t&srrrrl Ps~*chological .4ssociation. San Francisco.
EISFNRERGER R.. KARPMAN M. and TRATTNEK J. (1967) What is the necessary and suficlent condition for
remforcement in the contingency situation’! J. ezp. Psj~hol. 74. 342-350.
FRAIR C. M. (1968) Behavioral moddication of trainable mentally retarded children. Doctoral dissertation.
Utah State University. Ann Arbor: Xerox University Microfilms. No. 70-2396.
GL.THRIE E. R. (1952) The Psycholog_v qf Lrurnitg. Harper, New York
HARRISON R. G. C1970) Investigations of Premackian reinforcement theory. Doctoral dissertation. Florida
State University. Ann Arbor: Xerox Microfilms, No. 71-7028.
HARTIE J. C. (1973) Premackian reinforcement of classroom behavior through topic sequencing. J. Psycho/.
84. 61-74.
HOLSTEIN S. and HUNDT A. G. (1965) Reinforcement of intracranial self-stimulation by licking. Psychm.
sci. 3. 17-18.
HOLT G. L. (1971) Systematic probability reversal and control of behavior through reinforcement menus.
Psychol. Rec. 21. 465-469.
HOMME L. E. (1965) Control of coverants. the operants of the mind. Psycho/. Rec. 15. 501-51 I.
HOMME L. E. (1966) Contiguity theory and contingency management. Psychol. Rec. 16. 233-241.
HOMME L. E. and DEBACA C. (1965) Contingency management on a psychiatric ward. In: Brharior Technology:
Motivation and Corrtingrnq Munugemenr (Eds. L. E. HOMME and D. T. TOSTI) (1971) pp. 83-92. Individual
Learning System. San Rafael. CA.
HOMMI L. E.. I)I.BA(A C.. DI \,\I P.. STI INHOST J. V. and RIC~LRT E. J. (1963) Use of the Premack Principle
in controlling the behavior of nursery school children. J. erp. anal. B&w. 6. 544.
HOMME L. E. and Tosn D. T. (1965) Some considerations of contingency management motivation. R;ar,7.
Sot. Progr. Insrr. J. 4. 14-16.
HORAN J. J. and JOHNSON R. G. (1971) Coverant conditioning through self-management application of the
Premack Principle: Its effect on weight reduction. J. Eehau. The-. e.up. Psychiat. 2. 243-249: also see
HORAN J. J. (1970) The effect on weight reduction of coverant conditioning through self-management
application of the Premack Principle. Doctoral dissertation, Michigan State University, Ann Arbor: Xerox
University Microfilms. No. 71-18.223.
HUNDT A. G. and PREMACK D. (1963) Running as both a positive and negative reinforcer. Science 142.
1087-1088.
INCE L. P. (1969) A behavioral approach to motivation in rehabilitation. Psychol. Rrc. 19. lO%lll.
JOHANSSON S.. LFWINSOHN P. M. and FLIPPO J. F. (1969) An application of the Premack Principle to the
verbal behavior of depressed patients. Associarion,for thr Adr~~ncrrn~ntof B&urior Thrrup~~. Washington.
D.C.: also see Johunsson S. (1969) An application of the Premack Principle to the verbal behavior of
depressed suhjccts. Master’s thesis. University of Oregon.
JOH~.SON W. G. (1971) Some applications of Homme’s coverant control therapy: Two case reports. Brhut:.
Thrrapy 2. 24s-248.
KANFER F. H. and PHILLIPS J. S. (1970) Lrurning Fow~durions ofBehuoior Therapy. Wiley. New York.
KUAI’I’ T. J. and SHOI)AI+I. S. A. (1974) Ben Franklin as a behavior moditier. Brhur. Thcrupy. 5. 656--660.
LAWSOL D. M. and MA) R. B. (1970) Three procedures for the extinction of smoking behavior. Psychol.
Rec. 20. 151-157.
LEWINX)HU P. M.. WLIYSTIII\ M. and SHAW D. (19691 Depression: A clinical research approach. In: Adcanccs
in Brhurior Thrrupj, (Eds. R. D. RL~BIN and C. M. FRANKS) (1969). pp. 231-240. Academic Press. New
York.
LIRERMA~VR. (1972) .4 Glridv ro Brhaciorul Amlysis urld Therapy. Pergamon Press. New York.
MACDO~~;ALL)M. L. (1972) The modification of the play behavior of preschool children using the Premack
Principle. Master’s Thesis. University of Illinois.
MAHONEY M. J. (1970) Toward an experimental analysis of coverant control. Behuc. Therupy. 1. 510 521.
MAHONEY M. J. (1971) The self-management of covert behavior: A case study. B&c. Thrrupy 2. 575-57X.
MCINTIRL: R. W. (1963) Reinforcement and verbal learning: A test of the Premack hypothesis. Psycho/. Rep.
12. 99- IO?.
MITCHLLL W. S. and STAFF~LMAYR B. E. (1973) Application of the Premack Principle to the behavioral
control of extremeI> inactive schizophrenics. J. uppl. Brhuu. Anal. 6. 419423.
PRI M,\CK D. (1959) Toward empirical behavioral laws: I. Positive reinforcement. Psychol. Rrr. 66. 219-133.
PKI MACE D. (1962) Revcrsibihty of the reinforcement relation. Scir!fce 136. 255-257.
PRLMAC~ D. (1963) PredictIon of the comparative reinforcement values of running and drinking. SL.~L’IZ(.(,
139. 1067-- 1063.
PRCMAC~CD. (1963) Rate differential reinforcement in monkey manipulation. J. exp. unul. Behu~~ 6. 81-89.
PREMACK D. (19651 Reinforcement theorv. In: Nebraska Swnoosiurn_ . on Motivation (Ed. D. LEVINE), .pp. 123-180.
University of Nebraska Press. Lincoln.
PREMACK D. (1971) Catching up with commonsense or two sides to a generalization: Reinforcement and
punishment. In: T/II, Nutwe of Rcirlforcernrnt (Ed. R. GLASER), pp. 121-150. Academic Press, New York.
PREMACK D. (1972) The effect on extinction of the preference relations between the instrumental and contingent
events. In: Rcirlforc,c,r,lr/lr: A Brhu~iorul Analysis (Eds. R. M. GILBERT and J. R. MILLLNSON). pp. 51-65.
Academic Press. Ne% York.
The Premack Principle 147

PKL~ZA~K D. and PK~MACK A. J. (1963) Increased eating in tats deprived of running. J. e\-p. unal. B&r.
6. 209%’ 1’.
PREMACK D.. SCHAEFFERR. W. and HCNDT A. G. (1964) Reinforcement of drinking by running: Effect of
fixed ratio and reinforcement time. J. tsp. a&. B&n,. 7. 91-96.
RACHLIN H. (1970) Introduction to Modern Behaviorism. Freeman. San Francisco.
ROI~ERTSA. E. (1969) Development of self-control using Premack’s differential rate hypothesis: A case study.
Be/m. Rrs. & 77wapy 7. 341-344.
ROBINSON J. D. and LEWISSOHN P. M. (1973) Experimental analysis of a technique based on the Premack
Principle changing verbal behavior of depressed individuals. Psycho/. Rep. 32. 199-210.
S~HAEFFER R. W. ( IW?I The reinforcement relation as a function of the instrumental response rate. J. rsp.
Ps~cl~ol. 69. 419425.
SCHAEFFEK R. W. (1967) Positive reinforcement: An extension of the Premack Principle to response chains.
PsychoIl. Sci. 8. 31-32.
SCHAEFFER R. W.. BAUERMEISTERJ. J. and DAVID J. H. (1973) A test of Premack’s “indifference principle.”
Bull. P.sych. Sot. 1. 399-401.
SCHAEFFER R. W.. HANNA B. and Russo P. (1966) Positive reinforcement: A test of the Premack theory.
P.~ycho~i. Sci. 4. 7-8.
SI IT/. F. C. (1971) A behavior modification approach to depression: A case study. Psychok 8. 58-63.
Sr LII K B. and MA) I K G. ( 1972) B&urior Morl~ficorron Proc~durcs ./or SC/IOO/Pcrsor~twl. Dryden Press. Hins-
dale. III.
SHERMAN A. (1973) Brhcior Mod$carim: Theory md Practice. Brooks/Cole. Monterey. CA.
T’ERHCISTJ. and PMMACK D. (1970) On the proportionality between the probability of not-running and
the punishment effect on being forced to run. Lrarning Morir. 1. 141-147.
THARP R. G. and WETZEL R. J. (1969) Beharior Modificariorl in the Natural Emhrwwnt. Academic Press.
New York.
TIMBERLAKE W. and ALLISON J. (1974) Response deprivation: An empirical approach to instrumental perfor-
mance. Psycho/. Rer. 81. 146164.
TODD F. J. (1972) Coverant control of self-evaluative responses in the treatment of depression: A new use
for an old principle. B&r. Thrrapy 3. 91-94.
T~IILEY J. T. and PRATT S. (1967) An experimental procedure for the extinction of smoking behavior. Psychol.
Rrc. 17. 209-218.
TYLER V. 0. and STRALGHAN J. H. (1970) Coverant control and breath holding as techniques for the treatment
of obesity. PSK-ho/. Res. 20. 473478.
WASIK B. H. (1969) The effects of fixed ratio and contingent time on human lever-pressing behavior. Psychol.
KC. 19. 95-104.
N Aslh B. H. (1969) The effects of fixed ratio and contingent time on human lever-pressing behavior. Ps!xhoL
Rec. 9. 95-104: also see WASIK. B. H. (1967) Tests of the applicability of Premack’s generalizations
of reinforcement to human lever-pressing behavior. Doctoral dissertation. Florida State University. Ann
Arbor: Xerox University Microfilms. No. 68-385: and WASIK. B. H. (1968) A post-contingency test of
the effectiveness of reinforcement. Psyclto~~. Sci. 13. 87-88.
WASK B. H. (1970) The application of Premack’s generalization on reinforcement to the management of
classroom behavior. J. ezp. Child Psycho/. 10. 3343.
WEISMAS R. G. and PREMACK D. (1966) Reinforcement and punishment produced by the same response
depending upon the probability relation between instrumental and contingent responses. Psychottovlic
So&f!, MrrfirxJ. St. Louis.
WHALEY D. L. and MALOTT R. W. (1971) E1rnlrntar.r Prirlciples qf B&vior. Appleton-Century-Crofts. New
York.
WHITE 0. R. (1971) A Glossary qf Behmioral Trrrnirmlogy. Research Press. Champaign. Illinois.

You might also like