Is Freedom Contagious? A Self-Regulatory Model of Reactance andSensitivity to Deviant Peers

N. Pontus LeanderUniversity of Groningen

Michelle R. vanDellenUniversity of Georgia

Judith Rachl-WillbergerUniversity of Groningen

James Y. Shah, Gavan J. Fitzsimons, andTanya L. Chartrand

Duke University

Psychological reactance is typically assumed to motivate resistance to controlling peerinfluences and societal prohibitions. However, some peer influences encourage behav-iors prohibited by society. We consider whether reactant individuals are sensitive tosuch opportunities to enhance their autonomy. We specifically propose a self-regulatoryperspective on reactance, wherein freedom/autonomy is the superordinate goal, andthus highly reactant individuals will be sensitive to peer influences that could enhancetheir behavioral freedoms. In 2 studies, we find that reactant individuals can becooperative in response to autonomy-supportive peer influences. Participants read ascenario in which a peer’s intentions to engage in substance use were manipulated toimply freedom of choice or not. Results indicated that highly reactant participants weresensitive to deviant peers whose own behavior toward alcohol (Study 1, N � 160) ormarijuana (Study 2, N � 124) appeared to be motivated by autonomy and thus affordedfree choice. Altogether, the results support a self-regulatory model of reactance,wherein deviant peer influence can be a means to pursue autonomy.

Keywords: reactance, autonomy, peer contagion, self-regulation

Psychological reactance motivates autonomyfrom controlling interpersonal and societal in-fluences (Brehm, 1966; Brehm & Brehm,1981). Although interpersonal and societal in-fluences often align, in some cases they conflict:

proximal peer influences to drink alcohol, forinstance, conflict with distal societal prohibi-tions against underage drinking. How do highlyreactant individuals respond to deviant peer in-fluences—do they resist, ignore, or cooperate?Accepting a peer’s influence might seem anti-thetical to an autonomously motivated individ-ual; however, deviant peers might also providethe means or inspiration to react against societalprohibitions. History is certainly rife with ex-amples of rebelliously minded individualsbanding together in the shared pursuit of free-dom. Former Vice President Hubert Humphreyeven went as far as to suggest, “freedom is themost contagious virus known to man.” (Thom-sett & Thomsett, 2008 p. 53) For reactant indi-viduals, seeking autonomy could be manifestedin sensitivity to everyday acts of deviance byothers.

In the present work, we consider how indi-viduals regulate their reactant tendencies in so-cial environments that often provide little roomfor it. We consider reactance to be part of a

N. Pontus Leander, Department of Social and Organiza-tional Psychology, University of Groningen; Michelle R.vanDellen, Department of Psychology, University of Geor-gia; Judith Rachl-Willberger, Department of Social andOrganizational Psychology, University of Groningen;James Y. Shah, Gavan J. Fitzsimons, and Tanya L. Char-trand, Department of Psychology and Neuroscience & TheFuqua School of Business, Duke University.

This research was supported by the National Institute onDrug Abuse (NIDA) Grant P30 DA023026 and Grant P20DA017589. Its contents are solely the responsibility of theauthors and do not necessarily represent the official views ofNIDA. This research also received support from the DukeInterdisciplinary Initiative in Social Psychology.

Correspondence concerning this article should be addressedto N. Pontus Leander, Department of Social and Organiza-tional Psychology, University of Groningen, 2/1 Grote Kruis-straat, 9712TS Groningen, the Netherlands. E-mail: n.p.leander@rug.nl

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Motivation Science © 2016 American Psychological Association2016, Vol. 2, No. 4, 256–267 2333-8113/16/$12.00 http://dx.doi.org/10.1037/mot0000042

256

broader motivational system that serves to bothprotect and enhance a person’s freedom ofchoice. As a result, highly reactant individualsmight resist some influences but not others—oreven accept certain influences that enhance theirbehavioral freedoms. We assume the need forautonomy is regulated much like any other goaland is thus sensitive to opportunities as well asthreats.

A self-regulatory model could help to recon-cile the seemingly competing roles of psycho-logical reactance and peer influences to engagein behaviors prohibited by society. Models ofpeer contagion generally suggest that adoles-cents “catch” their peers’ intentions toward un-healthy behaviors (Dishion & Dodge, 2005;Prinstein & Wang, 2005). College students whoare subliminally primed with the names of pro-drug peers, for instance, subsequently reportincreased motivation to use marijuana them-selves (Leander, Shah, & Chartrand, 2009); col-lege students who see a person select oversizedportions of candy also eat more candy them-selves (McFerran, Dahl, Fitzsimons, & Mo-rales, 2010). Yet it is unclear whether reactantindividuals are sensitive to these types of peercontagion; reactance typically serves to coun-teract influences that constrain them to one par-ticular course of action. However, it might beself-defeating to counteract influences that en-hance behavioral freedoms. Even Brehm’s(1966) seminal theorizing considered ways inwhich reactance to smaller freedoms can beattenuated when the alternative is losing a muchmore valued freedom. Perhaps reactance ofteninvolves the choosing of lesser evils, if thesuperordinate goal is autonomy and to experi-ence freedom of choice. Along these lines, wepropose the strength of people’s reactance mo-tivation makes them sensitive to deviant peers.

A Model of Self-Regulatory Reactance

Fifty years ago, Brehm (1966) laid the theoret-ical groundwork for a self-regulatory approach toreactance: he predicted reactance motivationwould be moderated by key motivational factors,such as perceived importance (of the desired free-dom) and assessments of the feasibility of restor-ing the freedom. In a subsequent review, Brehmand Brehm (1981) updated reactance theory toconnect it to advances in motivation sciencesuch as helplessness and energization; they also

began to ponder connections between reactanceand control motivation—specifically, the needto have personal control over one’s outcomes(Brehm & Brehm, 1981). Thus, our notion thatreactance operates as part of a broader self-regulatory system focused on the ongoing pur-suit of autonomy is consistent with classic re-actance theory. Even in his seminal theorizing,Brehm (1966) depicted reactance as a means toan end, stating that individuals can better satisfytheir needs if they have the freedom to do whatthey want and do it when and how they want.Perhaps the same psychological need that facil-itates reactance, against controlling social influ-ences, also facilitates sensitivity to autonomy-supportive influences that increase one’sperceived freedom and autonomy.

To pursue this idea, we do not focus on just aparticular reactant state per se, but also on aperson’s general tendencies toward reactance.Traditionally, psychological reactance is con-sidered a motivational state that arises whenone’s behavioral freedom is threatened (Brehm& Brehm, 1981; Miron & Brehm, 2006). How-ever, reactance can be either state or trait, andwe consider trait-level reactance to represent aperson’s general tendencies toward their pursuitof autonomy, which refers to a superordinateneed for freedom of choice and self-determina-tion (Deci & Ryan, 2000). If autonomy is thesuperordinate goal, it becomes conceivable thatthe motivational underpinnings of reactance—whether trait reactance or state-induced reac-tance—promote sensitivity to opportunities toenhance one’s sense of autonomy. Dependingon circumstances, reactance could, at times,promote cooperativeness with deviant peer in-fluences.

A self-regulatory model of reactance couldfoster more nuanced predictions about whetherindividuals show counteraction or contagionwhen exposed to deviant peer influences. Thusfar, research mainly suggests that, although stu-dents are generally sensitive to goal contagionfrom peers (Aarts, Gollwitzer, & Hassin, 2004),reactant adolescents tend to resist goal conta-gion. For example, in one scenario-based study,students high in trait reactance were less likelyto “catch” a peer’s ostensible goal to spend aholiday helping in disaster relief (Leander,Shah, & Chartrand, 2011); in another study,reactant students who were subliminally primedwith the name of a controlling relationship

257SELF-REGULATORY REACTANCE

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

showed subsequent activation of an opposinggoal (Chartrand, Dalton, & Fitzsimons, 2007).A self-regulatory model assumes there may becases wherein reactance facilitates sensitivity tosuch influences.

We test this model with respect to how reac-tant individuals react to the perceived goal-directed behavior of others—namely, that of apeer who appears to be motivated to engage ina deviant behavior. Reactant individuals maynot be averse to such peer influences per se;they are averse to influences that restrict theirfreedom of choice. Reactant individuals may besensitive to—even inspired by—peers who ap-pear to be autonomously motivated. Indeed, re-search on goal contagion suggests people aresensitive to “catching” the goals of peers whenthey have a need for that goal themselves—aslong as the influence does not threaten theirother needs or values (Aarts et al., 2004; Lean-der et al., 2011). The issue, of course, is thatadolescents often report feeling pressured bytheir peers to engage in prohibited behaviors(Hays & Ellickson, 1990), and hence many de-viant peer influences only threaten one’s auton-omy in a different way. Given that state reac-tance and goal contagion are both triggered byinferences about a target person’s intentions(Aarts et al., 2004; Ringold, 2002), it is possiblereactant individuals are sensitive to autonomy-supportive cues when making inferences abouta deviant peer’s motivation toward a prohibitedbehavior. Yet for reactance to facilitate a coop-erative response, it may not suffice to simplyinfer that a peer intends to engage in a prohib-ited behavior; the reactant perceiver may alsoneed to infer the peer is motivated by autonomy.

In two studies, we test a self-regulatorymodel of reactance and outline how reactantindividuals navigate their social environmentsin pursuit of autonomy. We apply the model toyoung adults in context to behaviors often tar-geted by distal societal prohibitions—namely,alcohol and drugs. The studies test whether re-actance facilitates goal contagion when a peertriggers inferences of autonomy. Note that wepredict reactant individuals will only be sensi-tive to peer influences that trigger the pursuit ofautonomy, not peer influences that simply re-strict their choices toward engaging in the pro-hibited behavior.

Study 1

Reactant individuals may be sensitive to goalcontagion when a peer’s motivation to engagein a prohibited behavior appears to be aboutexercising free choice as opposed to some othermotive. This idea was tested in the context ofunderage drinking. Alcohol is prohibited in theUnited States to those under 21 years of age, yetpeer contagion is considered a significant con-tributor to underage drinking (Dishion &Dodge, 2005; Prinstein & Wang, 2005). In thisstudy, underage participants were exposed to apeer whose desire to drink alcohol either ap-peared to be motivated by the pursuit of auton-omy or not. Participants’ subsequent motivationto drink was assessed via their explicit ratings ofa series of advertisements for various alcoholicand processed beverages, as well as their behav-ior toward the ads, in terms of time spent look-ing at them. Reactant participants were ex-pected to demonstrate higher interest in the adswhen the deviant peer appeared to be motivatedby autonomy as opposed to not.

Method

Participants and design. One hundredsixty undergraduates (97 female) from a small,private southeastern university participated inexchange for course credit. All participantswere under the age of 21.1 The data were col-lected prior to analysis and data collectionstopped at the end of the semester.

Procedure. After giving informed consent,participants were randomly assigned to one oftwo conditions (inferred goal: free choice vs.drinking only). Participants then completed agoal contagion manipulation via computer. Par-ticipants read a scenario about a friend’s plansto go drinking, which was framed to trigger aninference that the friend was motivated to drinkeither for reasons related to autonomy or forunrelated reasons. Participants read:

Imagine that it’s near the end of the day and a friend ofyours is trying to decide what to do. One option is to goto a party happening tonight where there will be plentyof alcohol, as it’s been a long time since your friendhas �had vs. made� the choice to go out and drink.

1 Some additional participants were recruited but wereexcluded for being 21� years old (and therefore drinkingalcohol would be considered a legal behavior for them).

258 LEANDER ET AL.

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Changing the one word (had vs. made)changed the goal inference: stating how long ithas been since the friend had the choice to drinkimplies a drinking goal but one motivated by thepursuit of autonomy (e.g., because of a priorlack of opportunity); stating how long it hasbeen since the friend made the choice implies adrinking goal only and nothing about freechoice (e.g., the friend has always had opportu-nity but only recently chose to drink).

Participants’ subsequent motivation towarddrinking was assessed using an indirect self-reportmeasure and an implicit behavioral measure, so asto circumvent self-presentation issues with askingreactant participants to self-report their drinkingintentions. Immediately after the scenario, partic-ipants were given a cover story that the research-ers were interested in the potential market effec-tiveness of a series of beverage ads. There were 20full-color ads, half were for alcoholic beveragesand the rest were sugary sodas and other pro-cessed beverages one might encounter at a party.The ads were presented in random order and par-ticipants gave a subjective rating for each ad(“How effective do you think this ad will be?”: 1[not at all] to 7 [extremely]). Two dependentmeasures were derived: First was the subjec-tive rating of the ads’ effectiveness (M �4.03, SD � 0.66, � � .83), which could bepositively biased by reactant participants’ ex-posure to the autonomy-motivated peer. Thesecond dependent measure was time spent onthe ads—a measure of goal-directed behavior.Activated goals tend to draw attention towardgoal-relevant stimuli (Moskowitz, 2002), andthis may apply to the amount of time perceiv-ers spend looking at product advertisements(Celsi & Olson, 1988). To address nonnor-mality typical for reaction time (RT) data,each score was log-transformed and outlierswere removed (i.e., RTs 3 SDs beyond themean; 2.6% of all responses). A mean scorewas calculated for the number of secondsspent per ad, Muntransformed � 5.29, SD �0.88, � � .92.

Participants then completed a series of ques-tionnaires, including the Hong Reactance Scale,M � 3.03, SD � 0.54, � � .74 (Hong & Faedda,1996). There was also an eight-item measure ofself-regulatory effectiveness to help explorewhether the predicted reactance effect was indeedassociated with self-regulation. Sample items in-clude, “I usually judge what I=m doing by the

consequences of my actions,” “It’s hard for me tonotice when I’ve ‘had enough’ (alcohol, food,sweets)” [reverse coded] rated as 1 (not at all) to7 (extremely), M � 4.99, SD � 0.76, � � .68 (seeBrown, Miller, & Lawendowski, 1999; Leander etal., 2009).2 Participants then reported their demo-graphics, indicated their suspicions about thestudy, and were fully debriefed. No participantsreported how the scenario might have influencedtheir responses to the task.

Results and Discussion

Subjective ratings. An initial regressionanalysis predicted participants’ advertisementratings from their peer influence condition (in-ferred goal: free-choice vs. drinking only[coded 1, �1]), trait reactance (standardized),and the interaction of these two variables. Re-sults indicated a crossover interaction, B � .16,t(156) � 2.96, p � .004, 95% CI [0.05, 0.27],and no direct effects (ts � 1). As illustrated inFigure 1, relatively reactant participants (1 SDreactance) gave increased ratings when theirfriend was motivated by free choice, B � 0.94,t(156) � 1.75, p � .082, 95% CI [�.02, .28]; incontrast, relatively nonreactant participantsgave increased ratings when their friend wasmotivated to drink for reasons unrelated to freechoice, B � �0.19, t(156) � �2.55, p � .012,95% CI [�.33, �.04]. Note that at �1.17 SDreactance, the positive effect of the free-choicecondition crossed the threshold for significance(p � .050). Altogether, relatively reactant par-ticipants showed higher motivation towarddrinking when it was associated with autonomy.

Goal-directed behavior. A regression anal-ysis predicted time spent on the ads (log-transformed) from their goal inference condition,trait reactance, and the interaction of these twovariables. Results again indicated a two-way in-teraction, B � .007, t(156) � 2.13, p � .034, 95%CI [0.001, 0.014], as well as a marginal positivedirect effect of free choice, B � .005, t(156) �1.66, p � .100, 95% CI [�0.001, 0.012]. Asillustrated in Figure 2, relatively reactant (1 SD)participants spent more time on the beverage adswhen their friend’s goal to drink was motivated byfree choice as opposed to not, B � .012, t(156) �2.63, p � .010, 95% CI (0.003, 0.022). Relatively

2 Trait reactance and self-regulatory effectiveness werenegatively correlated, r � �.26, p � .001.

259SELF-REGULATORY REACTANCE

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

nonreactant (�1 SD) participants did not differ intime spent on the ads (t � 1).

Moderation by self-regulatory effective-ness. Exploratory regression analyses testedwhether scores on the self-regulation measure

further moderated the effects. If reactance isself-regulatory, goal contagion might only oc-cur among effective self-regulators. Separateregression analyses predicted participants’ sub-jective ratings and implicit behavior from the

Figure 1. Mean rating of beverage advertisements as a function of participants’ goalinference condition (free choice vs. drinking only) and trait reactance (Study 1).

Figure 2. Mean time spent on ads as a function of participants’ goal inference condition(free choice vs. drinking only) and trait reactance (Study 1). Note the y-axis starts at 5 s.

260 LEANDER ET AL.

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

goal inference condition, trait reactance (stan-dardized), self-regulatory effectiveness (stan-dardized), and all possible interactions. On thesubjective ratings, there were no additional ef-fects of self-regulatory effectiveness (ts � 1).However, there was a marginal three-way inter-action for time spent on the ads, B � .006,t(156) � 1.96, p � .052, 95% CI [0.00, 0.01].The behavioral pattern illustrated in Figure 2only occurred among more effective self-regulators (1 SD). The moderation analysishelps to illustrate that reactance is self-regulatory.

The results suggest reactant individuals distin-guish opportunities from threats and regulate theirsensitivity to deviant peers accordingly. Partici-pants with higher trait reactance showed goal con-tagion when they could infer that the peer’s mo-tivation to drink was about exerting free choice, asindicated by their subjective ratings and behaviortoward the ads. Reactant participants were espe-cially likely to show increased goal-directed be-havior if they were effective self-regulators. Incontrast, participants with lower trait reactanceshowed goal contagion when the peer’s motiva-tion to drink was unrelated to autonomy, as indi-cated by the subjective ratings. The results supportthe idea that reactant individuals are sensitive topeers who trigger autonomy goals.

Study 2

The aim of this study is to demonstrate that thesuperordinate goal served by reactance is auton-omy. To reactant individuals, the appeal of devi-ant peer influences is not necessarily to engage inthe prohibited behavior itself, but to have thefreedom to choose whether or not to engage in thebehavior. Classic reactance theory (Brehm, 1966)has long considered ways in which people seekother, often indirect, means of restoring freedomthat are not necessarily focused on engaging in theone behavior being threatened—which suggestsengagement in the behavior itself is not the super-ordinate goal. Thus, exposure to a friend whoappears to be autonomously motivated may nottrigger a goal to engage in any one behavior inparticular; it could simply trigger motivation tojoin a friend in their pursuit of autonomy and freechoice.

This was tested in the context of marijuana useamong college students in the Netherlands. Dutchlaw technically prohibits marijuana possession,

but it is commonly available via so-called “coffeeshops.” As with underage drinking, peer conta-gion is a contributor to motivation for marijuanause; Leander et al. (2009) found that participantswho were subliminally primed with the names ofpro-marijuana friends subsequently showedheightened accessibility of marijuana-related con-cepts in memory via a euphemism-listing task.Yet in those studies, there were no contextual cuesto trigger autonomy goals. The present study in-stead used context cues to trigger reactance andthen behavioral cues to indicate whether a peerwas restricted by a focal motivation to use mari-juana, or was instead more broadly motivated togo out—and thus afford free choice—about usingmarijuana or not.

A second aim of this study is to providefurther evidence that reactance is self-regula-tory. We sought to manipulate reactance moti-vation in advance to show that it can be acti-vated in the same way as other goal states(Chartrand & Bargh, 1996). Participants alsoreported their chronic marijuana use to testwhether they would only show goal contagionwhen marijuana is a valued behavioral freedom,with the idea that implicit motivational influ-ences are often moderated by motivational self-relevance of the behavior (Aarts et al., 2004;Leander et al., 2009). Participants should notshow goal contagion if marijuana is not a valuedfreedom (see also Brehm, 1966).

Method

Participants and design. One hundredtwenty-six undergraduates from a Dutch univer-sity participated in exchange for course credit.Participants were in an English-speaking psy-chology program (88 German, 12 Dutch, 26other nationalities; 104 female). All data werecollected prior to analysis and data collectionstopped at the end of a 3-week lab reservationperiod.

Procedure and materials. After giving in-formed consent, participants were randomly as-signed to one of four conditions in a 2 (sublim-inal prime: reactance vs. control) � 2 (goalinference: free-choice vs. marijuana use) be-tween-subjects design. They first gave informedconsent and then completed a subliminal prim-ing procedure to either prime reactance motiva-tion or not. Based on similar paradigms (e.g.,Leander et al., 2009), participants were given a

261SELF-REGULATORY REACTANCE

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

focal task to occupy their conscious attention asthey were subliminally primed (16 ms) witheither reactance words (rebel, free, oppose, re-volt, independent) or control words (game,cake, flower, finger, atmospheres). The focaltask was simply to decide whether a numberthat appeared in the center of the screen waseven or odd (e.g., “768” or “745”) and to pressthe F key for even numbers and the J key forodd numbers. Before the number appeared, astring of asterisks (��������) first directed par-ticipant’s attention to the center of the screen.This was followed by a word prime in one of thescreen quadrants (i.e., their parafoveal field).Participants completed 106 trials and were thenforwarded to the goal inference scenario.

Participants were instructed to imagine it wasa Friday night and they were having dinner witha friend. After a while, the friend says,

Hey, I was thinking of what we could do later tonight.I know this great coffee shop where they sell goodquality marijuana. I am really up for smoking tonight.This is what we �are going to do / could do� today.I am �not / also� in the mood for anything else likedancing or watching a movie.

Thus, in both conditions the friend was mo-tivated to use marijuana, but only the “coulddo/also” condition afforded free choice.

Cognitive accessibility of marijuana.Participants then completed a euphemism list-ing task previously used to measure the cogni-tive accessibility of marijuana (Leander et al.,2009). An ability to generate more “marijuanawords” from ambiguous stimuli is associatedwith greater cognitive accessibility (e.g., goalactivation) and predicts engagement in the be-havior (Stacy, Ames, Sussman, & Dent, 1996).Participants were instructed to generate as manyeuphemisms for the word “marijuana” as pos-sible (e.g., weed, reefer). The number of euphe-misms generated by participants represented itscognitive accessibility and thus whether a mar-ijuana goal was activated; five outliers werewinsorized, M � 3.41, SD � 2.25.3

Subjective rating of motivation. Participantswere later asked, “How likely is it that you joinyour friend?” rated as 1 (not at all) to 7 (ex-tremely), M � 3.94, SD � 1.99. This itemassessed general motivation to join their friendwithout specifying they would use marijuana.Participants who were motivated to exert freechoice may be more motivated to join the friend

even if they do not report any heightened mo-tivation to use marijuana per se. Two subse-quent items assessed perceptions of social pres-sure to smoke marijuana or to join their friend,but these items had no bearing on the results.Only their motivation to join the friend wascorrelated with cognitive accessibility of mari-juana, r � .36, p � .001; the other two itemswere not (rs � .04, ps � ns).

Participants’ history of chronic marijuanause, which would turn out to be a critical mod-erator of the predicted effects, was assessed viathree questions used in previous research (Le-ander et al., 2009). Participants reported howmany times they used marijuana in the last 30days (free response), lifetime (0 � never to 8 �100� times), and over the last 6 months (scaleresponse, 0 � no use to 8 � more than once perday). Responses to these items were standard-ized and combined (� � .89).

Suspicions regarding the nature of the studywere assessed before participants were thankedand fully debriefed. Three participants identi-fied at least one of the prime words, but exclud-ing these participants did not significantlychange the results.

Results and Discussion

The predicted effects were only observed in athree-way interaction with chronic marijuanause, so we focus on those results. Also, note thepattern of the data unexpectedly differed be-tween the two dependent measures, but did so ina theoretically consistent way.

Motivation to join the friend. A regres-sion analysis predicted motivation to join thefriend from participants’ reactance condition(reactance vs. control [coded 1, �1]), goal in-ference condition (free choice vs. marijuana useonly [coded 1, �1]), chronic marijuana use(standardized), and all possible interactions. Re-sults indicated a direct effect of chronic mari-juana use, B � 1.46, F(1, 118) � 77.74, p �.001, 95% CI [1.13, 1.79], and a three-wayinteraction of the goal inference, reactance, and

3 Participants also completed a modified measure of self-reported motives to use marijuana (see Leander et al., 2009).The motives correlated with chronic marijuana use, r � .44,p � .001, cognitive accessibility of marijuana, r � .23, p �.009, and motivation to join the friend, r � .56, p � .001.There were no effects of the manipulations on this measure.

262 LEANDER ET AL.

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

chronic marijuana use, B � �0.34, F(1, 118) �4.21, p � .042, 95% CI [�0.67, �.01]. Noother effects approached significance (Fs �1.65).

As illustrated in Figure 3, chronic marijuanausers generally reported a high likelihood ofjoining their friend. However, a slight goal con-tagion effect still emerged among chronic usersprimed with reactance when their friend’s mo-tivation toward marijuana afforded free choice.The positive effect of the free choice conditionwas marginally significant at 1 SD chronic mar-ijuana use, B � .58, t(118) � 1.74, p � .084;the simple slope crossed the threshold for sig-nificance (p � .050) at �1.82 SD chronic mar-ijuana use. Otherwise, reactant participants re-sisted the influence of a peer whose only goalwas to use marijuana, suggesting it threatenedtheir autonomy. Although marijuana had to bean attractive option in order for peer contagionto occur, it could not be the only option—it stillhad to be presented as a free choice.

Cognitive accessibility of marijuana. Totest whether reactant participants’ motivation tojoin the friend was focused on marijuana use ornot, a second regression analysis predicted par-ticipants’ cognitive accessibility of marijuanafrom their reactance condition (reactance vs.control), goal inference condition (inferredgoal: free choice vs. marijuana use only),chronic marijuana use (standardized), and all

possible interactions. Results again indicated adirect effect of chronic marijuana use, B � 1.37,F(1, 118) � 48.13, p � .001, 95% CI [0.98,1.76], and a significant three-way interaction—but this time in the opposite direction, B � 1.37,F(1, 118) � 5.09, p � .026, 95% CI [�0.84,�.054]. No other direct effects or interactionswere significant (Fs � 2, ps � .06).

As illustrated in Figure 4, chronic marijuanausers generally showed high accessibility ofmarijuana, but a goal contagion effect emerged,which exacerbated this accessibility, amongchronic users who were not primed with reac-tance and who inferred that the friend was re-stricted to marijuana use. This was indicated bya significant simple slope of the goal inferencecondition, B � �1.30, t(118) � �3.52, p �.001. This is a classic peer contagion effect thathas been observed before (Leander et al., 2009).Notably, participants primed with reactance didnot show increased accessibility of marijuana(ts � 1), nor did controlling for accessibilityalter their subjective ratings. Altogether, wefound no evidence that reactance-primed partic-ipants were specifically motivated to use mari-juana: although they reported increased motiva-tion to join the friend whose motivationafforded free choice of whether to use mari-juana or not, they maintained their default levelof accessibility of marijuana-related constructs.This suggests marijuana use was not their focal

Figure 3. Motivation to join friend as a function of participants’ priming condition (reac-tance vs. control), goal inference condition (free choice vs. marijuana use only), and chronicmarijuana use (Study 2).

263SELF-REGULATORY REACTANCE

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

goal; rather, their focal goal was to join theirautonomy-supportive friend.

The results suggest reactant individuals aresensitive to goal contagion from peers who trig-ger an autonomy goal rather than a goal toengage in the prohibited behavior per se. Whenchronic marijuana users were primed with reac-tance, they showed heightened motivation tojoin the friend who afforded freedom of choice.However, they did not become specifically mo-tivated to use marijuana. Only those chronicusers who were not primed with reactanceshowed increased motivation for marijuana use.In sum, reactant individuals showed peer con-tagion under different circumstances and, ap-parently, the goal they “caught” also differed.Importantly, this study helps to show that free-dom/autonomy is the superordinate goal in theminds of highly reactant individuals, not thespecific behavior per se.

These results support the idea reactance isself-regulatory. The reactance manipulationtriggered a slight shift in responding and theeffects only occurred among chronic marijuanausers (Leander et al., 2009). That the effectswere only observed among chronic marijuanausers suggests the inferred goal (i.e., the behav-ioral freedom) had to be self-relevant to triggergoal contagion (Brehm, 1966; Brehm & Brehm,1981). For reactant individuals, the prohibitedbehavior has to be an attractive option—but notthe only option, and the goal that is “caught” isthe pursuit of freedom/autonomy, not to engage

in the prohibited behavior per se. In sum, whenreactance facilitates sensitivity to implicit peerinfluences, autonomy is the goal that is trig-gered.

General Discussion

The present studies suggest reactant individ-uals can be influenced by others in their pursuitof autonomy. Reactant participants did not re-flexively react against every influence they en-countered—they were sensitive to opportunitiesas well as threats. They were sensitive to devi-ant peers who appeared motivated by autonomyand afforded freedom of choice. When reac-tance facilitated goal contagion, it was alsomainly among effective self-regulators (Study1) and the goal “caught” from their peers was toexercise free choice, not to engage in the pro-hibited behavior per se (Study 2). Notably, re-actant individuals who showed sensitivity todeviant peers were not simply motivated to getdrunk or high; their motivation toward the pro-hibited behavior was a means to an end. Tothem, the superordinate goal was autonomy andthey were sensitive to social opportunities toexercise freedom of choice.

Theoretical Implications

This research considered how reactance op-erates as part of a broader self-regulatory sys-tem focused on the ongoing pursuit of auton-

Figure 4. Cognitive accessibility of marijuana as a function of participants’ priming con-dition (reactance vs. control), goal inference condition (free choice vs. marijuana use only),and chronic marijuana use (Study 2). Higher scores indicate greater accessibility.

264 LEANDER ET AL.

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

omy. Our logic is based partly in earlytheorizing by Brehm (1966; Brehm & Brehm,1981), wherein he noted the important of reac-tance by stating that individuals can better sat-isfy their needs if they have the freedom to dowhat they want and do it when and how theywant. We built upon this idea, and consideredwhether the same motivational concerns thatcould make one react against controlling socialinfluences could also facilitate sensitivity to in-fluences that enhance freedom and autonomy.Our approach is in keeping with a classic per-spective on reactance. For example, whenBrehm and Brehm (1981) reviewed the ad-vances of the theory since its original 1966conceptualization, they considered the possibil-ity that reactance is connected to control moti-vation—namely, having control over one’s ownbehavior (i.e., autonomy). If reactance is indeeda manifestation of a superordinate autonomyneed, new predictions for reactance could bederived from other motivation theories: Self-Determination Theory (Deci & Ryan, 2000)could give insight on how to further connectreactance to the need for autonomy (or perhapsother trait-level motives—see Jonason & Fer-rell, 2016); alternatively, Goal Systems Theory(Kruglanski et al., 2002) could give insight intothe specific operation of reactance as a means toan superordinate end.

The findings of our two studies also illustratehow a self-regulatory perspective could explainthe nuances of reactance in response to differenttypes of influence. For example, the presentresults are in harmony with research showingthat people typically only resist goal contagionwhen the influence interferes with other needs(Leander et al., 2011). This may help to recon-cile the seemingly contradictory roles of reac-tance and peer contagion in predicting motiva-tion to engage in prohibited behaviors. It alsosuggests reactance could increase motivation toengage in a prohibited behavior via at least tworoutes: by reacting against the societal prohibi-tion via increased attraction to the restrictedbehavior—the traditional route, or in sensitivityto interpersonal influences that provide meansto engage in the behavior—a novel and indirectroute. Little is known about the interpersonalroute or its implications, but the present studiessuggest it could explain a range of health-related behaviors connected to peer influence.

The findings are also consistent with recentresearch showing that individuals may be sen-sitive to social influence as long as its control-ling aspects are not made salient (Laurin, Kay,Proudfoot, & Fitzsimons, 2013). This is remi-niscent of Brehm’s idea that reactance is onlytriggered when there is a perceived intent toinfluence (Brehm, 1966). Indeed, goal conta-gion occurred among reactant participants aslong as the prohibited behavior was not thefocal motivation, which may have reduced thesalience of its influence.

Another notable finding is that the autonomythreats were not real—they occurred entirely inparticipants’ minds. This supports past findingsshowing that the imagined presence of otherscan suffice to trigger reactance to implicit mo-tivational influences (Chartrand et al., 2007;Leander et al., 2011). Yet reactance effects canalso occur simply from perceiving threats toothers’ autonomy, even if there are no implica-tions for the perceiver’s freedoms (Andreoli,Worchel, & Folger, 1974). From a self-regulatory perspective, reactance may motivatea kind of vigilance that inflates one’s assess-ments of threat, leading even imaginary andinferred threats to trigger a reactant response.

Limitations and Future Directions

It is important not to overstate the results—the sample sizes were small and some of theeffects were only marginally significant. Ourtentative conclusion from these data is that thereare elements to reactance that fit a self-regulatory model. The results also do not indi-cate that reactance motivates pursuit of prohib-ited behaviors; rather, it motivates sensitivity tosocial opportunities to enhance one’s freedomand autonomy.

The studies also only focused on immediateresponses in an experimental setting, whichmay not always map onto long-term outcomes.Although reactant individuals may be drawn topeers whose influence facilitates their pursuit ofautonomy, they could eventually be repelledwhen the influence becomes repetitive. Theoriz-ing on fatal attractions suggests that the qualitythat initially brings two people together is oftenthe same quality that later pushes them apart(Felmlee, 1995). A question for future researchis how long or often reactant individuals willaccept the influence of a deviant peer.

265SELF-REGULATORY REACTANCE

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

It is also worth noting that Study 1 used ameasure of trait reactance that is operationalizedprimarily in terms of threat sensitivity. Al-though past work suggests reactance can indeedbe assessed as a unidimensional trait (e.g., Jo-nason, Bryan, & Herrera, 2010), there are con-cerns about the validity and usefulness of sucha measure (Jonason, 2007; Miron & Brehm,2006). The present work may signal the needfor a trait reactance measure that distinguishesautonomy-enhancing opportunities from threatsto autonomy.

Along the same lines, one might questionwhether the motivation adopted by participantsis indeed the personal pursuit of freedom/autonomy, or one more akin to vicarious reac-tance (arousal because of a threat to the target’sfreedom, see Sittenthaler, Traut-Mattausch, &Jonas, 2015). Although we assume the former,there is work to suggest that merely perceivinga threat to another person’s freedom has impli-cations for vicarious reactance (Andreoli et al.,1974; Sittenthaler, Jonas, & Traut-Mattausch,2016). Perhaps a distinction can be made be-tween a motivational contagion mechanism anda vicarious reactance mechanism by identifyingwhat, exactly, perceivers notice about the targetperson—is it their goal-directed behavior or thethreat to their freedom? A goal contagion ex-planation may be apt when perceivers attend toa target person’s behavior (rather than the threatto their freedom), especially given our findingsthat the goal “caught” was not necessarily toengage in the restricted behavior. More researchis needed to distinguish the specific implica-tions of exposure to others’ threats to freedomversus their behavior to restore it.

Altogether, the present work highlights thepotential self-regulatory nature of reactance,and perhaps also the illusory nature of any au-tonomy that one can attain in social situations.Despite the seeming shortsightedness of manyreactant behaviors, the need for autonomy maybe fundamental. Thus, societies and govern-ments may be obligated to accommodate thisneed and craft regulations in a way that makeshealthy choices also feel like autonomouschoices.

References

Aarts, H., Gollwitzer, P. M., & Hassin, R. R. (2004).Goal contagion: Perceiving is for pursuing. Jour-

nal of Personality and Social Psychology, 87, 23–37. http://dx.doi.org/10.1037/0022-3514.87.1.23

Andreoli, V. A., Worchel, S., & Folger, R. (1974).Implied threat to behavioral freedom. Journal ofPersonality and Social Psychology, 30, 765–771.http://dx.doi.org/10.1037/h0037529

Brehm, J. W. (1966). Social psychology: A series ofmonographs, treatises, and texts. New York, NY:Academic Press.

Brehm, S., & Brehm, J. (1981). Psychological reac-tance: A theory of freedom and control. New York,NY: Wiley.

Brown, J. M., Miller, W. R., & Lawendowski, L. A.(1999). The self-regulation questionnaire. In L.VandeCreek & T. L. Jackson (Eds.), Innovationsin clinical practice: A source book (Vol. 17, pp.281–289). Sarasota, FL: Professional ResourcePress.

Celsi, R. L., & Olson, J. C. (1988). The role ofinvolvement in attention and comprehension pro-cesses. Journal of Consumer Research, 15, 210–224. https://www.jstor.org/stable/2489526

Chartrand, T. L., & Bargh, J. A. (1996). Automaticactivation of impression formation and memoriza-tion goals: Nonconscious goal priming reproduceseffects of explicit task instructions. Journal of Per-sonality and Social Psychology, 71, 464 – 478.http://dx.doi.org/10.1037/0022-3514.71.3.464

Chartrand, T. L., Dalton, A., & Fitzsimons, G. J.(2007). Nonconscious relationship reactance:When significant others prime opposing goals.Journal of Experimental Social Psychology, 43,719–726. http://dx.doi.org/10.1016/j.jesp.2006.08.003

Deci, E. L., & Ryan, R. M. (2000). The “what” and“why” of goal pursuits: Human needs and theself-determination of behavior. Psychological In-quiry, 11, 227–268. http://dx.doi.org/10.1207/S15327965PLI1104_01

Dishion, T. J., & Dodge, K. A. (2005). Peer conta-gion in interventions for children and adolescents:Moving towards an understanding of the ecologyand dynamics of change. Journal of AbnormalChild Psychology, 33, 395–400. http://dx.doi.org/10.1007/s10802-005-3579-z

Felmlee, D. H. (1995). Fatal attractions: Affectionand disaffection in intimate relationships. Journalof Social and Personal Relationships, 12, 295–311. http://dx.doi.org/10.1177/0265407595122009

Hays, R. D., & Ellickson, P. L. (1990). How gener-alizable are adolescents’ beliefs about pro-drugpressures and resistance self-efficacy? Journal ofApplied Social Psychology, 20, 321–340.

Hong, S. M., & Faedda, S. (1996). Refinement of theHong Psychological Reactance Scale. Educationaland Psychological Measurement, 56, 173–182.http://dx.doi.org/10.1177/0013164496056001014

266 LEANDER ET AL.

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.

Jonason, P. K. (2007). Further tests of validity for theone-dimensional Hong’s psychological reactancescale. Psychological Reports, 101, 871–874.

Jonason, P. K., Bryan, A., & Herrera, J. (2010).Trimming the fat reveals a one-factor structure forHong’s Psychological Reactance Scale. IndividualDifferences Research, 8, 220–228.

Jonason, P. K., & Ferrell, J. D. (2016). Lookingunder the hood: The psychogenic motivationalfoundations of the Dark Triad. Personality andIndividual Differences, 94, 324–331. http://dx.doi.org/10.1016/j.paid.2016.01.039

Kruglanski, A. W., Shah, J. Y., Fishbach, A., Fried-man, R. S., Chun, W. Y., Sleeth-Keppler, D., &Zanna, M. P. (2002). A theory of goal systems. InM. Zanna (Ed.), Advances in experimental socialpsychology (Vol. 34, pp. 331–378). San Diego,CA: Academic Press.

Laurin, K., Kay, A. C., Proudfoot, D., & Fitzsimons,G. J. (2013). Response to restrictive policies: Rec-onciling system justification and psychological re-actance. Organizational Behavior and Human De-cision Processes, 122, 152–162. http://dx.doi.org/10.1016/j.obhdp.2013.06.004

Leander, N. P., Shah, J. Y., & Chartrand, T. L.(2009). Moments of weakness: The implicit con-text dependencies of temptations. Personality andSocial Psychology Bulletin, 35, 853–866. http://dx.doi.org/10.1177/0146167209334784

Leander, N. P., Shah, J. Y., & Chartrand, T. L.(2011). The object of my protection: Shieldingfundamental motives from the implicit motiva-tional influence of others. Journal of ExperimentalSocial Psychology, 47, 1078–1087. http://dx.doi.org/10.1016/j.jesp.2011.04.016

McFerran, B., Dahl, D. W., Fitzsimons, G. J., &Morales, A. C. (2010). I’ll have what she’s having:Effects of social influence and body type on thefood choices of others. Journal of Consumer Re-search, 36, 915–929. http://dx.doi.org/10.1086/644611

Miron, A. M., & Brehm, J. W. (2006). Reactancetheory—40 years later. Zeitschrift für Sozialpsy-chologie, 37, 9–18.

Moskowitz, G. B. (2002). Preconscious effects oftemporary goals on attention. Journal of Experi-mental Social Psychology, 38, 397–404. http://dx.doi.org/10.1016/S0022-1031(02)00001-X

Prinstein, M. J., & Wang, S. S. (2005). False con-sensus and adolescent peer contagion: Examiningdiscrepancies between perceptions and actual re-ported levels of friends’ deviant and health riskbehaviors. Journal of Abnormal Child Psychology,33, 293–306. http://dx.doi.org/10.1007/s10802-005-3566-4

Ringold, D. J. (2002). Boomerang effect: In responseto public health interventions: Some unintendedconsequences in the alcoholic beverage market.Journal of Consumer Policy, 25, 27–63. http://dx.doi.org/10.1023/A:1014588126336

Sittenthaler, S., Jonas, E., & Traut-Mattausch, E.(2016). Explaining self and vicarious reactance: Aprocess model approach. Personality and SocialPsychology Bulletin, 42, 458–470. http://dx.doi.org/10.1177/0146167216634055

Sittenthaler, S., Traut-Mattausch, E., & Jonas, E.(2015). Observing the restriction of another per-son: Vicarious reactance and the role of self-construal and culture. Frontiers in Psychology, 6,1052. http://dx.doi.org/10.3389/fpsyg.2015.01052

Stacy, A. W., Ames, S. L., Sussman, S., & Dent,C. W. (1996). Implicit cognition in adolescentdrug use. Psychology of Addictive Behaviors, 10,190–203. http://dx.doi.org/10.1037/0893-164X.10.3.190

Thomsett, M. C., & Thomsett, J. F. (2008). War andconflict quotations (p. 53). London, NY: McFar-land & Company, Inc.

Received October 20, 2016Revision received December 8, 2016

Accepted December 8, 2016 �

267SELF-REGULATORY REACTANCE

Thi

sdo

cum

ent

isco

pyri

ghte

dby

the

Am

eric

anPs

ycho

logi

cal

Ass

ocia

tion

oron

eof

itsal

lied

publ

ishe

rs.

Thi

sar

ticle

isin

tend

edso

lely

for

the

pers

onal

use

ofth

ein

divi

dual

user

and

isno

tto

bedi

ssem

inat

edbr

oadl

y.