Find an antecedent intervention (within the last 5 years) from this peer-reviewed behavioral journal list and describe/synthesize how that treatment could be applied/utilized to im

https://doi.org/10.1177/01454455221085249

Behavior Modification 2023, Vol. 47(1) 219 –246

© The Author(s) 2022 Article reuse guidelines:

sagepub.com/journals-permissions DOI: 10.1177/01454455221085249

journals.sagepub.com/home/bmo

Review

Applications of the Premack Principle: A Review of the Literature

Jessica L. Herrod1 , Sara K. Snyder1, Joseph B. Hart1, Sarah J. Frantz1, and Kevin M. Ayres1

Abstract The Premack principle states that any Response A can reinforce any other Response B if the independent rate of A is greater than the independent rate of B. This theory demonstrates reinforcer relativity, where the relative probabilities of responses can be more impactful than preference. Applying the Premack principle involves arranging the environment to restrict access to certain responses based on relative probabilities of a set of given responses. Though the Premack principle is described in modern behavior analytic texts, Konarski et al. identified a lack of empirical evidence to support its application. The purpose of the current paper is to systematically review the extant literature using the Premack principle and evaluate how and if researchers have applied reinforcer relativity as described by Premack and the subsequent effectiveness of these procedures. Additionally, we make recommendations for practitioners and future researchers based on our findings.

Keywords Premack principle, response deprivation hypothesis, disequilibrium theory

1University of Georgia, Athens, GA, USA

Corresponding Author: Jessica L. Herrod at the Center for Autism and Behavioral Education Research, University of Georgia, 850 College Station Road, Building II, Athens, GA 30605, USA. Email: [email protected]

1085249 BMOXXX10.1177/01454455221085249Behavior ModificationHerrod et al. review-article2022

220 Behavior Modification 47(1)

Premack’s probability-differential provided behavior analysts an alternative perspective to program for reinforcement. Structural descriptions of positive reinforcement have traditionally described it as access to a stimulus in tem- poral relation to a response (Skinner, 1938), whereas Premack made the dis- tinction that reinforcement could be achieved through access to engaging in a response (Premack, 1959). This led to the Premack principle, which involves arranging the environment to restrict access to certain responses based on relative probabilities of a set of given responses (Timberlake & Farmer-Dougan, 1991). This theory capitalizes on reinforcer relativity, which suggests the relative probabilities of responses can be more impactful than preference. The Premack principle states that any Response A can reinforce any other Response B if the independent rate of A is greater than the indepen- dent rate of B, which is greater than the independent rate of response C (Premack, 1959). Thus, a response can serve as a reinforcer in some environ- mental circumstances, but not others. For example, Response A will Reinforce response B and Response C. Response B will reinforce Response C, but Response C will not reinforce Responses A or B. Therefore, Response B both is and is not a reinforcer depending on the other available responses available in a situation (Premack, 1959). The Premack principle focuses on two sepa- rate responses. The first response, the one that serves as the reinforcer, is referred to as the “contingent response” and the second response, the one that grants access to the contingent response, is referred to as the “instrumental response.” In the example above, Response A is a contingent response to reinforce the instrumental Response B.

The introduction of the Premack principle was widely popular, as it made identification of reinforcers convenient and unintrusive (Timberlake & Farmer-Dougan, 1991). Furthermore, arranging contingencies in this way provides a variety of conveniences to interventionists in applied settings. For example, interventionists can arrange naturally occurring activities and/or responses already in an individual’s repertoire based on the Premack princi- ple to increase desired behavior without making substantial changes to their setting. Additionally, practitioners would not need to introduce reinforcers not already existing in the environment (e.g., edible items).

Klatt and Morris (2001) cite activity as a reinforcer dating back to 1922 in work by Richter examining rat behavior. Premack (1962) demonstrated that the reinforcement relation between two activities could be “reversed” depend- ing on how he arranged exercise and drinking. Premack and Premack (1963) conceptualized the restriction of eating more as a loss of an activity than the physiological state of the animal after demonstrating that restriction and free access to an exercise wheel covaried with food intake (i.e., when the wheel was available, the animals ate less; when the wheel was unavailable, they ate

Herrod et al. 221

more). Nevin (2019) described this interchangeability in response-reinforcer relations reported by Premack as changing the way reinforcement is viewed. The 50 years of work since Premack’s seminal work have seen an evolution of this understanding. Therefore, a review and synthesis on the extant litera- ture can provide guidance on implementation variation, parameters, and con- texts in which this has been effective.

The basic work of Allison and Timberlake who began to demonstrate fur- ther how response deprivation can drive instrumental responding to Timberlake and Farmer-Dougan who illustrated how to translate this into applied settings. Premack principle is often applied in practical scenarios through first – then statements, which describe a behavioral contingency where the “first’’ component specifies the targeted behavior and the “then” component specifies the consequence contingent upon the targeted behavior occurring (Mechner, 2008). Practitioners use first – then statements prior to placing a demand that specifies the response requirement to gain access to a reinforcer (Trump et al., 2018). This contingency arrangement may be pre- sented using a visual support, such as a first-then board or a visual schedule (e.g., Warren et al., 2021).

Though a prominent development in the field, the Premack principle has some limitations. Premack described that the only way to determine response probability was collecting duration data on unrestricted responding, but this creates challenges when determining probability of discrete behaviors (Konarski et al., 1981). For example, some practitioners may want to evalu- ate response probability of behaviors with restricted opportunities to occur, such as responding to teacher directed academic instruction. Additionally, the responses possible to serve as reinforcers are limited in that the contingent response must always be of higher probability than the instrumental response. If not, Premack suggested that the subject being forced to engage in the lower probability response would theoretically serve as a punisher (Premack, 1959).

Timberlake and Allison (1974) expanded upon Premack’s work with the response deprivation hypothesis, which subsequently introduced the disequi- librium model (Timberlake & Farmer-Dougan, 1991) for effective use of the Premack principle. This suggests that an instrumental response, even if it is the lower probability behavior, will still serve as the contingent response if the schedule of reinforcement satisfies response-deprivation below independent levels of responding (Timberlake & Allison, 1974). This differs from Premack’s initial work in which he described that for a response to be a rein- forcer, independent levels of responding of the contingent response had to be higher than the instrumental response. Timberlake and Allison (1974) present their approach using the equation I/C >O / OI C . In this equation, I represents the scheduled amount of instrumental responding required to obtain C amount

222 Behavior Modification 47(1)

of the contingent response.OI andOc represent the operant levels of instru- mental and contingent responding during an independent baseline with no contingencies in place. Similarly, they present the equation I/C >O / OI C to demonstrate that the environment can be arranged so that, through restricting access to the lower probability behavior, the lower probability can serve as a reinforcing, contingent response with the high probability behavior as the instrumental response. Essentially, this suggests that a response of any proba- bility can serve as a reinforcer of another response if the practitioner is able to restrict access to the response. With this essential component, researchers can manipulate the schedule of reinforcement and adapt to motivation that changes based on the imposed contingency schedule (Timberlake & Allison, 1974).

Timberlake and Farmer-Dougan (1991) reported that the Premack princi- ple was described in a variety of textbooks used by the behavior analytic community since the 1970s (e.g., Donnellan et al., 1988; Kazdin, 1980; Sulzer-Azeroff & Mayer, 1977). However, Konarski et al. (1981) cited a lack of empirical support for the Premack principle at that time. Furthermore, the extant data often selected the contingent response to be a reinforcer based on anecdotal reports, not on the probability of a response (Konarski et al., 1981). The gap in the literature that Konarski et al. identified is relevant to both practitioners and researchers and warrants further investigation given that the Premack principle is still described in many contemporary behavior analytic texts, such as Cooper et al. (2020) and Alberto and Troutman (2013). The purpose of the current review of the literature is to assess the extent to which researchers have evaluated Premack principle and re-examine the data sup- porting its applied use in the 40 years since Konarski et al.’s reported the lack of published, empirical evidence. Further, this review weighs the rigor of the extant literature in an effort to identify best practices for future evaluation by researchers and practitioners.

Method

Search Procedures

The first author conducted a search of the databases PsychINFO, ERIC, PsychARTICLES, and PubMed in March 2020. They used the search terms “Premack principle” in quotation marks and “response deprivation” in two separate searches. Neither truncation nor wildcards were used in the search. Studies were included in the review based on the following criteria: (a) human participants, (b) contains the word “Premack,” (c) written in English. After duplicates were removed, this initial search yielded 79 studies. The first author evaluated the 79 studies using the additional inclusion criteria of

Herrod et al. 223

(d) applied research (evaluating behaviors of importance to humans or soci- ety as opposed to theory; Baer et al., 1968; Cooper et al., 2020) and (e) experimental (comparison of a phenomenon of interest under two or more conditions; Cooper et al., 2020). The additional inclusion criteria reduced the included studies to a total of 33. These studies included both published articles as well as unpublished theses and dissertations to reduce the likeli- hood of publication bias influencing review findings (King et al., 2020).

The first author then conducted an ancestral search on the references of the 33 studies from the initial search that added 12 more studies in the review. Next, the first author conducted a forward search of Premack’s (1959) article on PsychINFO. This yielded 76 studies; of these 7 studies met all inclusion criteria and were added to the review. Additionally, the authors completed a hand search of the two journals most represented in the 52 included studies, Journal of Applied Behavior Analysis and Behavior Therapy. After reviewing the tables of contents of these journals from the years 2000 to 2020, no addi- tional articles were added to the review. This resulted in a total of 52 studies with 61 separate experiments to be evaluated in the current review of the lit- erature. These search procedures are described in a PRISMA Flowchart in Figure 1 (Page et al., 2021).

Coding Procedures

The reviewers coded descriptive participant and experiment characteristics for all experiments. Participant characteristics included gender, age, diagno- sis, setting, and intellectual ability. Experiment characteristics included experimental design, dependent variable, independent, and contingent responses, evidence of independent responding, and potential of a causal relation. Additionally, coders evaluated the appropriateness of each article for inclusion considering the following criteria: (a) human participants, (b) con- tains the word “Premack,” (c) written in English, (d) applied research, and (e) experimental. All coders were trained on procedures and provided with oper- ational definitions for each coding category.

Participant and setting characteristics. Reviewers coded participant and setting characteristics to examine the variety of circumstances in which researchers applied the Premack principle. Participant and setting characteristics included gender, age, diagnosis, setting, and intellectual ability. Reviewers coded any age description of the participants or if the specific ages were not reported. Participant diagnosis codes included autism spectrum disorder (ASD), intel- lectual disability (ID), developmental delays (DD), multiple diagnoses (coded separately from other diagnoses that might have been part of their

224 Behavior Modification 47(1)

multiple diagnoses), other diagnoses, or none (i.e., participants were typi- cally developing). Experimental settings included hospital rooms, therapy rooms, classrooms, vocational settings, or living spaces.

Figure 1. PRISMA flowchart.

Herrod et al. 225

Experimental characteristics. Experimental characteristics included experi- mental design, dependent variable, evidence of response probability, and ability to demonstrate a causal relation. Reviewers coded the experimental designs as multiple probe/multiple baseline, alternating treatment design, group design, or other design. Dependent variables were coded in the catego- ries of problem behavior, academic responses, communicative response, vocational skills, activity allocation, and “other” variables that did not fit the definitions of the other, specific dependent variable categories. Reviewers coded for evidence of response probability assessment to determine if researchers presented baseline data to support the identified high and low probability responses. To accomplish this, researchers had to conduct a free operant paired baseline where the responses evaluated were simultaneously available without restriction (Timberlake & Allison, 1974). Experiments were coded as “with evidence” if the researchers presented data demonstrat- ing independent levels of responding and coded “without evidence” if researchers did not have data on independent levels of responding or assumed probability based on preference or anecdotes.

Because our aim was to explicitly focus on studies evaluating Premack as outlined in 1959, we isolated those studies that evaluated baseline levels of response allocation. These studies would have established higher and lower probability allocation and could therefore, based on evidence, arrange contin- gencies based on Premack. This allowed us to concentrate our analysis on those studies that could truly be said to evaluate Premack. Including those studies that did not present data related to baseline level evaluation of response alloca- tion would have risked incorporating data that did not represent Premack and thus obscured our conclusions. For this group of studies that could be deter- mined to have truly used Premack, reviewers coded for the potential of a causal relation demonstration based on criteria described by Gast and Ledford (2018), which involves replication of experimental effects at least three times. If exper- iments met this criteria, they were scored as potentially demonstrating a causal relation. If not, the study was coded as not having the potential to demonstrate a causal relation (e.g., some studies used an A-B-C-B arrangement which does not meet contemporary standards for direct replication within a study). If an experiment met the criteria for the probability of a causal relation, reviewers used the Single Case Analysis and Review Framework (SCARF; Ledford et al., 2016) to evaluate the rigor of experiments. The SCARF data collection tem- plate specifies that its use is to evaluate rigor of studies designed with at least three potential demonstrations of effect, similarly to the qualifications for a causal relation as described by Gast and Ledford (2018). For those familiar with What Works Clearinghouse (WWC) standards for single case design (Kratochwill et al., 2010), SCARF was designed to build on the strengths of

226 Behavior Modification 47(1)

those standards and address some of the WWC omissions (e.g., requirement of procedural fidelity). The SCARF template uses response options of yes/no/not applicable or scaled responses from 0 to 4 to systematically code information about each individual experiment in a study. The information collected from SCARF is displayed graphically to present a scatter plot of the included stud- ies’ quality and rigor. For experiments that demonstrated evidence of response probability and evaluated Premack principle using a group design, authors evaluated rigor using the WWC standards.

Interrater Reliability

Interrater reliability (IRR) was conducted by the second, third, fourth authors on all components of the review’s search and coding procedures. All coders were graduate students who received training on the search procedures, cod- ing procedures, and coding definitions. The second author conducted an inde- pendent initial search following the search procedures described above. The third author then analyzed the articles retrieved from the initial search and used the inclusion criteria to determine if they were appropriate for review. The second, third, fourth authors coded 100% of the included studies for inclusion criteria, participant descriptions, and experiment characteristics. Additionally, the second author conducted an independent SCARF coding of 100% of the articles which met the criteria to be coded in SCARF.

IRR results. IRR was calculated by dividing agreements by the number of agreements plus disagreements and multiplying by 100. The independent lit- erature search, conducted by the second author, resulted in 100% agreement (i.e., identical 79 studies to the first author’s initial search). Additionally, the third author’s evaluation of these 79 studies meeting inclusion criteria removed 43 studies, which was identical to the first author’s evaluation resulting in 100% IRR. The second, third, fourth authors took IRR data on the studies included from the initial, ancestral, and forward search results meet- ing inclusion criteria (a) to (e) with 100% agreement compared to the first author’s assessment. Participant and experiment characteristics IRR, also conducted by the second through fourth authors, resulted in 97.42% agree- ment when averaged across articles and 96.55% agreement as well when averaged across coding categories. The second author’s SCARF scoring resulted in 98.21% IRR when compared to the first author’s scoring.

Results

The review included 52 studies containing 61 experiments spanning the years 1959 to 2017. Tables 1 and 2 present participant information, dependent

227

T ab

le 1

. St

ud ie

s W

ith E

vi de

nc e

of R

es po

ns e

Pr ob

ab ili

ty .

C ita

tio n

N D

ia gn

os is

D V

Se tt

in g

R es

ea rc

h D

es ig

n

A es

ch le

m an

a nd

W

ill ia

m s

(1 98

9) 2

m al

es 1

fe m

al e

A ge

s 17

–1 9

ye ar

s

ID D

ur at

io n

en ga

ge m

en t

in lo

w p

ro ba

bi lit

y ta

sk s

T he

ra py

ro

om M

B/ M

P

A lle

n an

d Iw

at a

(1 98

0) 6

m al

es 4

fe m

al es

M ea

n ag

e 42

ye ar

s

ID Ex

er ci

se c

om pl

et io

n C

la ss

ro om

O th

er

*A m

ar i e

t  al

. (1

99 5)

1 fe

m al

e A

ge 1

5 ye

ar s

ID , E

pi le

ps y

K et

og en

ic fo

od

co ns

um pt

io n

N ot

r ep

or te

d A

T D

Ba te

m an

( 19

75 )

1 m

al e

1 fe

m al

e A

ge s

28 –3

ye ar

s

ID , D

ow ns

Sy

nd ro

m e

% o

f t im

e in

le ss

pr

ef er

re d

ac tiv

ity V

oc at

io na

l W

/R

Bi rc

h et

 a l.

(1 98

2) 5

m al

es 7

fe m

al es

Pr es

ch oo

l a ge

N ot

r ep

or te

d Ju

ic e

pr ef

er en

ce T

he ra

py

ro om

O th

er

D ou

gh er

( 19

83 )

2 m

al es

A ge

s 41

ye ar

s Sc

hi zo

ph re

ni a

Fr eq

ue nc

y of

s oc

ia l

re sp

on se

s H

os pi

ta l

ro om

W /R

Ed dy

( 19

75 )

1 m

al e

1 fe

m al

e A

ge s

6– 9

ye ar

s

ID %

o f i

nt er

va ls

w ith

co

m m

un ic

at io

n T

he ra

py

ro om

W /R

G up

to n

an d

Le Bo

w (

19 71

) 2

m al

es A

ge s

no t

re po

rt ed

N ot

r ep

or te

d %

o f s

al es

V oc

at io

na l

W /R

*H an

le y

et  a

l. (2

00 3)

4 m

al es

A ge

s 23

, 2 6,

4 2,

5 0

ye ar

s ID

, A SD

% o

f i nt

er ac

tio n

T he

ra py

ro

om W

/R

(c on

tin ue

d)

228

C ita

tio n

N D

ia gn

os is

D V

Se tt

in g

R es

ea rc

h D

es ig

n

H ol

bu rn

a nd

D

ou gh

er

(1 98

6)

1 fe

m al

e A

ge 2

1 ye

ar s

ID Fr

eq ue

nc y

of p

ro bl

em

be ha

vi or

T he

ra py

ro

om O

th er

H os

ie e

t  al

. (1

97 4)

Ex pe

ri m

en t

1 14

p ar

tic ip

an ts

Si xt

h gr

ad e

Ex pe

ri m

en t

2 20

p ar

tic ip

an ts

Fi ft

h– si

xt h

gr ad

e

N ot

r ep

or te

d T

im e

to c

om pl

et e

w ri

tt en

r ep

or t

C la

ss ro

om G

ro up

d es

ig n

K on

ar sk

i e t 

al .

(1 98

2) 4

m al

es A

ge s

7– 10

ye ar

s ID

D ur

a

Collepals.com Plagiarism Free Papers

Are you looking for custom essay writing service or even dissertation writing services? Just request for our write my paper service, and we'll match you with the best essay writer in your subject! With an exceptional team of professional academic experts in a wide range of subjects, we can guarantee you an unrivaled quality of custom-written papers.

Get ZERO PLAGIARISM, HUMAN WRITTEN ESSAYS

Why Hire Collepals.com writers to do your paper?

Quality- We are experienced and have access to ample research materials.

We write plagiarism Free Content

Confidential- We never share or sell your personal information to third parties.

Support-Chat with us today! We are always waiting to answer all your questions.