OUP user menu

From Firm Muscles to Firm Willpower: Understanding the Role of Embodied Cognition in Self-Regulation

Iris W. Hung , Aparna A. Labroo
DOI: http://dx.doi.org/10.1086/657240 1046-1064 First published online: 1 April 2011

Abstract

Across five studies, we show that firming one's muscles can help firm willpower and that firmed willpower mediates one's ability to withstand immediate pain, overcome food temptation, consume unpleasant medicines, and attend to immediately disturbing but essential information, provided that doing so is seen as providing long-term benefits. We draw on theories of embodied cognition to explain our results, and we add to that literature by showing for the first time that one’s body can help firm willpower and facilitate the self-regulation essential for the attainment of long-term goals.

For most people, self-control poses a dilemma. Such dilemmas arise in situations where a person must undergo immediate pain (and/or avoid immediate pleasure) to achieve long-term benefits (Fishbach, Friedman, and Kruglanski 2003; Fishbach and Labroo 2007). For instance, dieters face a self-control dilemma whenever they have to forgo immediately pleasurable indulgent food in favor of less tasty but healthy food or undertake the immediate pain of exerting themselves through exercise to achieve long-term health benefits. Similarly, academically inclined students face a self-control dilemma when they are tempted to party but know they should be studying for an important exam. To overcome such dilemmas, people try to firm their willpower (i.e., “regulatory strength” or “mental resolve”; Muraven, Tice, and Baumiester 1998) in a variety of ways—for instance, by summoning thoughts that bolster the value of long-term actions and by suppressing thoughts of temptation (Gollwitzer and Moskowitz 1996; Metcalfe and Mischel 1999). But willpower is a limited resource, and firming willpower is difficult (Baumeister, Heatherton, and Tice 1994). As a consequence, as people try to firm their resolutions, they may clench their fists, grit their teeth, and scrunch their muscles, reflecting the mental turmoil that comes from combating a self-control dilemma. But rather than reflecting the turmoil that results from combat, could such actions also help a person keep that turmoil at bay? In particular, might clenching one's fists, gritting one's teeth, or scrunching one's muscles also firm willpower and consequently improve self-control? And if so, why?

Recent findings in psychology are starting to establish that the relationship between the mind and the body is more complex than has previously been presumed. Moving beyond the idea that actions are a consequence of mental activities (Ajzen and Fishbein 1977; Alba and Hutchinson 1987; Srull and Wyer 1979), evidence on embodied cognition is starting to establish that the body too might be influencing the mind to a greater degree than has previously been recognized. Findings suggest that merely simulating many of the bodily actions that result from or accompany thought processes can also facilitate accessibility to those very thoughts. In this manner, the body can exert a powerful influence on the mind (Barsalou 2008).

Following the logic of those studies in embodied cognition, we argue that the firmed muscles that usually result from trying to recruit willpower—for example, clenched fists and stiffened muscles—also firm willpower, help one grasp long-range thoughts, and, consequently, help one engage more effectively in self-control. Across five experiments, we demonstrate that firm muscles can firm willpower and, as a consequence, increase self-control. Of course, simply engaging in such actions is not a cure for all self-control problems, but to the extent that these simple actions help firm willpower and facilitate self-control, something most of us spend much of our lives trying to master, our findings have important implications for consumer well-being.

Theoretical Background

The mind commands the body and it obeys. (St. Augustine)

Embodied Cognition

In the past, most philosophers, psychologists, and consumer researchers agreed that the mind is an engine that powers the body and enables us to act, steering us to accomplish whatever we perceive is in our best interest. As a consequence, actions are based on the thoughts that come to mind in any situation (Ajzen and Fishbein 1977) and on factors that influence thinking at that moment, including contextual factors, memory-based factors, or affective factors. The mind helps people attend to, acquire, interpret, and integrate new information with old information and with information contextually salient in that situation. Emerging research, however, shows that this mind-to-body relationship is not as one-directional as once presumed. Because cognition is “embodied,” the body exerts a powerful influence on shaping a person's thoughts (Barsalou 2008; Niedenthal et al. 2005).

The basic idea underlying these emerging findings on embodied cognition is that memories are composed of experiences that are multimodal and spread throughout the body, not amodal semantic nodes stored purely in the mind. Cognitions that reside in the mind are only one component of an experience that additionally includes an entire constellation of perceptions, movements, and sensations that make up the experience. These experiences are stored in the body as well as in the mind and are a combination of a person's thoughts and bodily states. When people retrieve experiences from their memories, these multimodal representations ensure that accompanying bodily states are activated along with the associated cognitions. For example, for those exerting self-control, resisting the temptation of ice cream might include not only attempts to focus on one's overarching health goals and to avoid thoughts about giving in to the temptations but also such bodily actions as stiffening the tongue, physically trying to not approach the temptation, and other sensations associated with exerting willpower. And one important consequence of memories being multimodal is that not only do cognitions generate accompanying bodily responses, but those very bodily responses that accompany thoughts can independently generate the associated cognitions and influence judgment (Briñol and Petty 2008).

For instance, while research shows that the solution of brainteasers involving spatial relations is accompanied by eye movements, it also shows that eye movements can significantly improve the rate at which one can solve such brainteasers (Thomas and Lleras 2007). Similarly, not only does moving their hands accompany children’s solving of math problems, but hand movements also facilitate math performance (Goldin-Meadow, Cook, and Mitchell 2009). Also, standing upright (vs. slumping over), which often results when one feels powerful, can also evoke sensations of power (Riskind and Gotay 1982). Using the dominant (vs. nondominant) hand to write helps individuals validate their thoughts (Briñol and Petty 2008); nodding (vs. shaking) one's head can increase how much one agrees with a persuasive message (Wells and Petty 1980) and can facilitate the recognition of positive (vs. negative) words (Förster and Strack 1996; Neumann and Strack 2000); and merely simulating a smile (vs. a frown) or physical approach (vs. avoidance) can elicit positive affect (Cacioppo, Priester, and Bernston 1993; Labroo and Nielsen 2010; Strack, Martin, and Stepper 1988). Physiological responses can also influence self-perceptions (Bem 1972), thereby affecting people's inferential processes and making them infer their attitudes from those responses. For example, individuals can infer their mood (e.g., happiness, anger) on the basis of the arousal that extraneous sources induce (Schachter and Singer 1962), and a sensation of effort can lead individuals to think that an outcome is goal relevant because people usually exert effort to accomplish important goals (Labroo and Kim 2009). In sum, motor actions accompanying performance on a task can facilitate performance on the task and can influence thoughts, feelings, and inference-making processes via self-perception.

The Current Research: Embodied Willpower Augmentation

In line with prior research suggesting that cognition is embodied, we propose that the body is also instrumental in self-regulation. It can help augment willpower by evoking nonconscious willpower-strengthening goals that boost and firm people's ongoing conscious attempts to engage willpower and thus facilitate their pursuit of long-term goals. As a result, an individual can better resist tempting but unhealthy food, consume unpleasant but necessary medicine, or attend to disturbing but essential information—that is, accomplish self-control when faced with the dilemma of choosing activities that are immediately aversive but beneficial in the long run. Anecdotal evidence amply suggests that muscle firming often accompanies endurance regimens. Indeed, in some cases, muscle firming might enhance pain tolerance or self-control. For instance, in the martial arts, fighters learn to contract their abdominal muscles to protect their organs. But in most situations, firmed muscles are believed to either simply accompany or be a consequence and not a cause of stretched endurance, a reflection of pain rather than its antidote. For example, when bodybuilders and weight trainers clench their facial, neck, arm, and leg muscles, the muscle clenching is usually a reflection of them stretching their endurance. Language, too, is full of metaphors that associate pumped fists and clenched jaws with overcoming challenges and with physical endurance. And people fighting to exert self-control may often tell themselves to “hold on to” that last shred of willpower. Thus, anecdotally, firmed muscles accompany willpower augmentation and may even help an individual endure physical pain and build resistance.

Similarly, if self-control is a battle that involves strengthening willpower to withstand immediate pain or resist immediate pleasure to attain long-term benefits, might firming one's muscles help firm willpower? If so, why? We argue that because firmed muscles usually accompany willpower strengthening, merely firming one’s muscles can also serve as a situational cue that primes willpower strengthening. As a consequence, among people who are consciously trying to strengthen willpower because they care about the long-term benefits of self-control but face an immediate temptation, this nonconscious priming of willpower-strengthening goals augments their conscious and ongoing efforts to strengthen willpower. In other research, related findings show that situational cues that activate a nonconscious pursuit of emotion-regulation goals augment emotion regulation among people who are consciously trying to regulate their negative emotions (Williams et al. 2009). These situational cues augment conscious emotion regulation, especially under conditions when conscious regulation is difficult (Williams et al. 2009). By a similar token, we propose that among individuals who face a self-control dilemma, merely firming their muscles can cue firmed willpower that augments their conscious efforts to firm willpower, and these augmented ongoing willpower efforts will in turn enhance self-control.

Across five experiments, we demonstrate that firming one's muscles helps firm willpower and enhance self-control by augmenting willpower to facilitate self-control. Experiment 1 first demonstrates that firm muscles can help an individual focus on immediately aversive information to become a better person in the long term. Experiment 2 inflicts physical pain on respondents, making them each submerge a hand in an ice bucket, presumably in order to improve blood circulation, and finds evidence of a facilitative role of firming muscles (vs. muscle loosening or no action) on enduring physical pain. Experiment 3 demonstrates the facilitating effect of firming muscles by showing how it helped participants consume more of a nasty but healthy vinegar drink. These experiments additionally show that the augmented ongoing willpower thoughts that arise while a person is engaged in self-control mediate the facilitating effect of firming muscles on self-control. Although conscious inferences pertaining to self-perceptions of willpower also follow the observed pattern of self-control effects, these self-perceptions do not drive the impact of muscle firming on self-control. Experiment 4 provides field evidence showing that an effect of firming one's muscles extends beyond enduring immediate pain: firm muscles firm mental resolve to resist the immediate pleasure of tempting food, but, importantly, only among people who are likely to engage willpower because they value the long-term health benefits of resisting temptations, as we would predict. Finally, experiment 5 shows that firming muscles facilitates self-control only if people firm their muscles at the same time they face a self-control dilemma, which further suggests that willpower must be augmented the moment self-control needs to be exerted. If people augmented their willpower by firming muscles in a previous self-control task, they are less able to engage willpower in a subsequent self-control task and appear depleted unless the latter task draws on the same type of willpower that the previous task augmented.

Experiment 1: Clenched Fists but Open Hearts for Haiti

Method

Experiment 1 investigated whether firming their muscles can make people attend to a disturbing and immediately aversive charity appeal and part with their own money to help others in need. Past research has employed similarly disturbing charity materials and has used donation to such a charity as an indication of self-control: the materials are so unpleasant that donating to the charity is extremely unpleasant in the moment and does not make people feel good; rather, donors need to overcome their immediate temptation to ignore the aversive appeal in favor of trying to become a better person in the long term by helping others now (Fishbach and Labroo 2007; Labroo and Mukhopadhyay 2009). A pretest (n = 20) confirmed that the materials we used elicited immediate negative affective reactions (“Attending to this appeal makes me ____,” 1 = feel bad now, 7 = feel good now; “Ignoring this appeal makes me ____,” 1 = feel bad now, 7 = feel good now; Mattend = 2.00 vs. Mignore = 4.00; t(19) = 5.41, p < .001) but that the reactions to these materials were also associated with long-term goals of becoming a better person (“Attending to this appeal makes me become a better person in the long term,” 1 = not at all, 7 = very much; “Ignoring this appeal makes me become a better person in the long term,” 1 = not at all, 7 = very much; Mattend = 5.10 vs. Mignore = 2.25; t(19) = 7.03, p < .01).

Fifty-four undergraduate students participated for course credit in this study on multitasking. First, we randomly assigned them to one of two conditions—muscle firming versus control action. We instructed firming-condition participants to grasp a pen in their fists; those in the control condition held the pen between their index and middle fingers. We chose this manipulation of muscle firming because one could also argue that being “tightfisted” makes people mentally rigid and less willing to part with their money rather than willing to summon the necessary willpower to part with their money when thinking about the plight of others, as we predict. The experimenter asked participants to maintain the assigned position until they were asked to stop. Next, experimenters said they were advising the Red Cross International (RCI) on donation appeals. Participants were told they would evaluate an appeal from the RCI, and if they wished, they could place an anonymous donation in a sealed envelope. They were told that regardless of whether they made a donation, they should put the envelope in the donation box when they were done. The appeal solicited donations for rebuilding earthquake-devastated Haiti, a cause students cared deeply about, and contained disturbing images of wrecked homes and wailing and injured children.

Along with the appeal and an envelope from the RCI, participants received a survey for evaluating RCI (“I think RCI is ____,” 1 = negative, bad, dislike, 7 = positive, good, like). Then, as manipulation checks, all participants indicated the extent to which they followed instructions to firm their muscles when they were holding the pen (“I firmed my muscles while holding the pen,” 1 = not at all, 7 = very much) and how tight their muscles felt when holding the pen (“My muscles felt tight while holding the pen,” 1 = not at all, 7 = very much; r = .75, p < .01). As control measures, all participants indicated their general tendency to donate to charities (“I usually donate to charity ____,” 1 = not at all, 7 = very much), their feelings at the moment (nine items, each on 7-point scales: “Right now, I feel [happy, relaxed, content, calm, tense, worried, aroused, disappointed, discouraged],” 1= not at all, 7 = very much), whether they felt empathy (“I felt empathy toward the Haiti victims,” 1 = not at all, 7 = very much), their perceived negativity of the task (“This task is ____,” 1 = negative, bad, dislike, 7 = positive, good, like), and their general approach-avoidance tendency (1 = I generally approach desirable things/good things, 7 = I generally avoid undesirable things/bad things). Finally, as a suspicion check, we asked participants to guess the purpose of the study. No participant guessed correctly.

Results and Discussion

Tables 1 and 2 present all means and associated standard errors for manipulation checks, control items, and hypothesis tests.

View this table:
View this table:

Manipulation Checks and Control Measures

As expected, participants in the muscle-firming condition reported that they firmed their muscles to a greater extent than control participants (Mfirm = 4.38 vs. Mcontrol = 2.88; F(1, 52) = 22.63, p < .01). We observed no effects of muscle action on any of the control measures (participants' evaluation of the task, general tendency to donate to charities, believability of the cover story, general controllability over outcomes, each of the feelings, arousal, and general approach-avoidance tendencies; all p > .20; see table 2).

Real Monetary Donation

Ninety-two percent of respondents in the muscle-firming condition chose to make a donation to Haiti, whereas only 72.4% of participants in the control condition chose to do so, and the results of a binary logistic regression showed that this difference was significant (b = 1.48, SE = .85, Wald(1) = 3.05, p < .05). Among participants who decided to make a donation, however, muscle action did not significantly affect the amount of the donation even though the amounts tended to be in the predicted direction (Mfirm = $6.74 vs. Mcontrol = $5.01; F(1, 42) = 1.59, p > .10). This finding suggests that muscle action facilitated self-control in overcoming the initial aversion to viewing unpleasant materials and parting with money, but once people overcame their instinct to say no, the donation amount was based on how much the appeal persuaded them. Consistent with such a premise, the evaluation of RCI also did not vary on the basis of muscle action (F < 1). In sum, experiment 1 suggests that, as we expected, firming muscles increased participants' likelihood of overcoming the immediate instinct to avoid the unpleasant materials and making a real monetary donation; however, it did not affect their evaluations of the charity materials, the amount participants donated once they decided to make a donation, their perception of how often they help or the importance of being generally helpful, or their mood, arousal, or involvement. This finding is particularly impressive because we manipulated muscle firming by asking participants to clench their fists, and a clenched fist may have cued a restricted donation. However, clenching their fists did not make people behave in a tightfisted manner or restrict their actions. Instead, firm muscles appear to have augmented willpower and enabled people to act proactively and expansively in the face of negative information, in accordance with the goal of becoming a better person in the long term.

A possible alternative account for our findings is that participants who tightened their fists may have felt more pain than those who performed a control action and that feeling pain may have led to increased empathy for others. In line with such a possibility, a one-way ANOVA on the empathy measures revealed a significant effect of muscle firming (Mfirm = 5.56 vs. Mcontrol = 4.86; F(1, 52) = 6.34, p < .01). However, importantly, empathy failed to mediate the effect of muscle firming on a participant's likelihood of making a donation, suggesting that it was not their reason for making a donation. Also, empathy on its own did not increase the donation rate (p > .30). Thus, although muscle firming increased empathy and the likelihood of making a donation, the effect of muscle firming on each was independent of the other, and empathy did not increase the donation rate. This result was in line with our predictions because our argument is that making a donation to this charity required self-control. Additionally, when people feel pain, they usually try to address it, and addressing pain by adding to it—in the form of parting with their money or attending to aversive information—is unlikely. The muscle-firming and control conditions also did not differ with respect to any of the mood measures; pain should have made people feel worse. Thus we do not believe participants donated money in an attempt to distract themselves from pain.

However, to further ensure against the possibility that muscle firming is painful and thus sensitizes people to pain, or even the reverse—that people who firm their muscles engage in self-control to distract themselves from their pain—in experiment 2 we used ability to withstand real physical pain as the dependent variable. We required participants to submerge one hand in icy water for as long as they could, presumably because it is good for blood circulation, while they either firmed or did not firm their muscles in the hand that was not in the water. If the findings of experiment 1 occur because firming their muscles sensitizes people to pain, then we should observe in experiment 2 that people who are firming their muscles and are presumably more sensitive to pain will submerge their hands in icy water for less time than no-action control participants will. We, of course, predict the reverse: firming their muscles in one hand will enable people to endure physical pain in the other hand. This experiment also included a “loosen muscles” condition; if muscle action is a distraction, then loosening muscles should be a better distraction than firming muscles because relaxation should facilitate distraction from pain better than presumably painful firming should. Again, we make the opposite prediction: loosening muscles will not recruit willpower; thus, such people will be less able to tolerate the icy water than will those firming their muscles.

Experiment 2: Of Frozen Hands and Icy Resolve

Method

Forty-seven undergraduate students participated in this study for compensation. We assigned participants randomly to one of three conditions—muscle firming, muscle relaxing, or no instruction pertaining to their left hands—in a between-subjects design and asked them to immerse their right hands in an ice bucket for as long as they could. We ran participants individually in one-on-one sessions. As a cover story, we explained to all respondents that they were participating in a study on motor skills and that in the experiment they would have to immerse their right hands in an ice bucket for as long as they could. Experimenters told them that this ice bucket test was an indication of longevity because people with healthy circulation can immerse their hands in ice for a longer duration than can those with poor circulation. In line with this cover story and to ensure that all participants experienced a self-control dilemma, such that they believed immersing their hands in the ice bucket was important but were also tempted to give up because it was painful, we first asked participants to list one reason why being healthy was important and then to indicate how much they cared about being healthy (1= not at all, 7 = very much; M > 5 in all conditions).

We then assigned participants randomly to one of three embodied cognition conditions: in the muscle-firming condition, the experimenter instructed participants to clasp a pen between their left palms and fingers and to maintain the position while they put their right hands in the ice bucket. In the muscle-relaxing condition, the experimenters instructed participants to let their left hands rest freely while supporting a pen between their index fingers and thumbs. In the no-instruction control condition, participants did not hold a pen. Once participants immersed their right hands in the ice bucket, the experimenter used a stopwatch to measure how long participants were able to keep their hands in the ice bucket. Importantly, as measures of the underlying process, all participants were asked to vocalize all thoughts that came to mind for the duration that their hands were submerged, and the experimenter recorded these thoughts.

Once participants withdrew their right hands from the ice bucket, we asked those in the firming or relaxing conditions to release the pen. All participants then completed a questionnaire regarding the actions they had just performed. First, as additional measures of process, they indicated how much willpower they could summon while performing the ice-bucket task (“While submerging my hand in the ice bucket … I summoned my willpower,” 1 = very little, 7 = very much), how much mind control they felt (“I toughened my mental resolve,” 1 = very little, 7 = very much), and how mentally strong they felt (“I tried to be mentally strong,” 1 = very little, 7 = very much). Next, as manipulation checks, participants in the firming and relaxing conditions indicated the extent to which they followed instructions on how to hold the pen (“Holding the pen, I firmed my muscles,” 1 = not at all, 7 = very much) and how tight their muscles felt when holding it (“Holding the pen, my muscles felt tight,” 1 = not at all, 7 = very much). All participants indicated the extent to which they believed immersing their hands in an ice bucket could reflect good circulation (1 = not at all, 7 = very much). As control measures, all participants further indicated the extent to which they generally felt in control of their outcomes (“In general, I am in control of my outcome,” 1 = not at all, 7 = very much), their feelings at the moment (nine items, each on a 7-point scale, “Right now, I feel [happy, relaxed, content, calm, tense, worried, aroused, disappointed, discouraged],” 1= not at all, 7 = very much), the negativity of the ice-bucket task (“This ice bucket task is ____,” 1 = negative, bad, dislike, 7 = positive, good, like), how painful the task was (“Submerging my hand was ____,” 1 = somewhat painful, 7 = extremely painful; “Holding the pen was painful,” 1 = disagree, 7 = agree), their beliefs pertaining to the body-mind relationship (“healthy mind, healthy body” and “strong mind, strong body,” 1 = disagree, 7 = agree), their tolerance for pain (“I am ____,” 1 = unable to endure pain/less able to endure pain than others, 7 = able to endure pain/more able to endure pain than others), and their approach-avoidance tendency (1 = I generally approach desirable things/I approach good things, 7 = I generally avoid undesirable things/I avoid bad things). Finally, they tried to guess the purpose of the study, provided background information, and were debriefed.

Results and Discussion

Tables 1 and 2 present all means and associated standard errors for manipulation checks, control items, and hypothesis tests.

Manipulation Checks and Control Measures

A single-factor (muscle: firm vs. relaxed) between-subjects ANOVA on a muscle-firming index that was created by averaging the two measures reflecting how much participants contracted their muscles and how tight their muscles felt as they held a pen in their left hands (r = .45, p < .01) revealed the expected effect of condition (Mfirm = 3.97 vs. Mrelax = 2.27; F(1, 31) = 19.58, p < .01). Also as expected, a single-factor (muscle: firm vs. control vs. relaxed) ANOVA on the measure of believability of the cover story that immersing one's hand in ice can reflect good circulation revealed no significant effect (Mfirm = 4.17 vs. Mcontrol = 4.57 vs. Mrelax = 4.60; F < 1, p > .65). Finally, single-factor ANOVAs on each of the control measures (controllability over outcomes, feelings, arousal, task negativity, painfulness, body-mind relationship beliefs, pain endurance, and general approach-avoidance tendencies) revealed no significant effects (all p > .15).

Hand Immersion (Behavioral Measure)

Unlike experiment 1, in which participants could choose whether to make a donation, and thus their decision to make a donation reflected their self-control in overcoming an initial aversion to avoid the unpleasant appeal, we required everyone in experiment 2 to immerse their hands in the ice bucket. Thus the duration for which they were willing to immerse their hands in the ice bucket reflected their self-control. The data pertaining to the time participants immersed their right hands in the ice bucket were positively skewed and were thus log transformed (Fazio 1990). A single-factor (muscle: firm vs. control vs. relaxed) ANOVA on this measure revealed a significant effect of condition (F(2, 45) = 3.37, p < .05). Participants who clutched the pen in their left hands were able to immerse their right hands longer than were participants who held a pen loosely between their index fingers and thumbs (Mfirm = 126.89 seconds vs. Mrelaxed= 73.07 seconds; t(45) = 2.07, p < .05) and longer than the no-instruction control participants (Mcontrol = 69.53 seconds; t(45) = 2.41, p < .05).

Self-Report Willpower Index

A single-factor (muscle: firm vs. control vs. relaxed) between-subjects ANOVA on a willpower index created by averaging the three associated items (willpower summoned, mind control, mental strength; α = .81) revealed the expected effect of condition (F(2, 45) = 7.22, p < .01). Participants who clutched the pen in their left hands reported greater willpower than did participants who held a pen between their index fingers and thumbs (Mfirm = 5.05 vs. Mrelaxed = 4.02; t(45) = 3.47, p < .05) and more than the no-instruction control participants (Mcontrol = 4.17; t(45) = 2.84, p < .05).

Online Willpower Thoughts

We analyzed the type of online thoughts participants generated while they immersed their hands in the ice bucket. Two coders (r = .94) who were blind to the hypothesis coded these thoughts as willpower thoughts (e.g., “I will feel less cold after a while, I must try a little bit more,” “I must sustain as long as possible,” and “I should endure longer”) or other thoughts (e.g., “Can anybody do this for more than 10 seconds?” “It's very cold,” “Never knew this is what it feels like,” “My fingers are numb,” and “I hope it is not harmful to my hand”). A one-way ANOVA on willpower thoughts revealed a significant effect of condition (F(2, 45) = 14.05, p < .01). Participants who clutched the pen in their left hands reported more willpower thoughts than participants who held the pen loosely (Mfirm = 1.11 vs. Mrelaxed = 0.07; t(45) = 4.54, p < .05) and the no-instruction control participants (Mcontrol = 0.07; t(45) = 4.34, p < .05). Importantly, the total number of thoughts was not different across conditions (Mfirm = 1.56 vs. Mloosen = 1.20 vs. Mcontrol = 1.13; F(2, 45) = 1.47, p > .20), suggesting that elaboration was not different across conditions.

Mediation Analyses

Mediation analyses revealed that online willpower thoughts mediated the impact of firm muscles on how long participants immersed their hands in the ice. The muscle-firming condition significantly influenced the number of willpower thoughts that participants generated (b = .54, SE = .12, p < .001) and how long participants kept their hands immersed (b = 29.40, SE = 14.09, p < .04), and the number of willpower thoughts that participants generated significantly influenced the submersion time (b = 43.83, SE = 13.48, p < .01). When we excluded willpower thoughts as a covariate in the regression predicting an effect of the muscle-firming condition on immersion, the effect became nonsignificant (b = 8.32, SE = 15.97, p > .60), whereas the impact of willpower thoughts remained significant (b = 39.23, SE = 16.20, p < .02; Sobel z = 2.12, p < .03). Importantly, although participants were also able to accurately report having summoned willpower, presumably when they did so, only ongoing willpower thoughts mediated self-control. Unsurprisingly, participants could accurately self-report having employed willpower when they had done so, but importantly, those self reports did not mediate self-control. This finding, along with control measures showing no movements on self-perceptions of pain tolerance, suggests that explicit self-perceptions (e.g., “If I firmed muscles, I must have self-control”) did not drive self-control associated with firming muscles. Instead, the actual willpower that participants summoned while firming their muscles mediated self-control, showing that muscle firming augmented the ongoing summoning of willpower, which influenced self-control, as we proposed. Thus, muscle firming is inherently tied to summoning willpower and is embodied, is captured by verbalizations of the muscle actions that actual willpower evokes in the moment, and is not a consequence of deliberative self-perception inferences made after firming muscles.

To summarize, experiment 2 demonstrated that although participants across conditions found immersing their right hands in the ice bucket to be quite painful, those who were simultaneously firming muscles in their left hands (vs. not firming muscles) were able to summon more willpower and, as a consequence, keep their right hands in the ice bucket longer. The online willpower thoughts that muscle firming evoked augmented willpower and increased self-control, and self-perception is an unlikely explanation of our findings. Participants' arousal, mood, pain endurance, general approach-avoidance tendencies, or perceived controllability over outcomes in general or the extent to which they followed instructions also did not drive these effects. Participants across conditions believed our cover story. Additionally, the muscle-relaxing condition did not influence willpower differently from the control condition. This finding is in accordance with our theorizing: we do not expect relaxation to deplete willpower nor to summon willpower in the way muscle firming does. Also, the finding in experiment 1 that muscle firming did not influence the amount participants donated and the finding in experiment 2 that muscle firming did influence the time participants submerged their hands are not discrepant. Donation amount in experiment 1 was conditioned on having made a donation, and self-control was associated with the decision to make a donation. In contrast, in experiment 2, participants did not have any choice regarding whether to exert self-control; we required all of them to submerge their right hands in an ice bucket. Thus the overall time participants submerged their hands is a measure of self-control. Indeed, if we look at donation amount without conditioning on donation rate in experiment 1, we get a significant effect of muscle action on donation amount (p < .05).

A question that may arise at this stage is whether muscle firming assists people in augmenting their willpower when they are facing a self-control dilemma, as we claim, or whether it generally cues long-term goals in everyone and helps people withstand pain. We expect that only people who face a self-control dilemma—for instance, choosing long-term health versus the temptation to indulge—will benefit from the facilitative effects of muscle firming because only such individuals will need to summon willpower. This proposition is in line with findings in the emotion-regulation literature that show that situational cues that help augment emotion regulation are most effective when individuals consciously wish to regulate their emotions but find that doing so is difficult (Williams et al. 2009). In line with our proposition, in experiment 1 students valued becoming a good person, and in experiment 2 we specifically asked everyone to first think about why long-term health goals are important; therefore, we ensured that participants would wish to engage in self-control and overcome the immediate aversion of parting with money or placing a hand in an ice bucket for the long-term benefits of becoming a good person or becoming healthy. Thus, in these situations, participants valued engaging in self-control, and we believe that only because they valued self-control did muscle firming result in willpower firming. Also compatible with our premise, in experiment 2 we saw that the extent of online willpower thoughts mediated self-control. Additionally, why muscle firming should generally focus people on the long term or help them withstand pain is theoretically somewhat unclear, unless it is because they are summoning willpower. Still, to clearly implicate a role of self-control, in experiment 3 we manipulated the extent to which people value self-control, the idea being that only people who face a self-control dilemma will summon willpower and that muscle firming will help only such individuals exert self-control to attain long-term goals in the face of immediate temptation.

Experiment 3: Wanting Health and Consuming Vinegar

The goal of experiment 3 was to directly test for a role of embodied cognition associated with firming muscles in self-regulation. We required participants to consume as much of a nasty health tonic as they could while they performed a muscle firming (vs. control) action. Prior to this activity, however, we primed roughly half of the participants (but not the remainder) with long-term health goals. This priming manipulation allows us to test the idea that muscle firming somehow generally improves pain resistance and the pursuit of long-term goals against our premise that muscle firming benefits only those who need to summon willpower to resist the temptation because their long-term health goals are accessible. If firming of willpower to resolve a self-control dilemma underlies our effects, as we predict, then only participants who are primed with health goals will consume more of the health tonic when firming muscles (vs. performing a control action). However, if firming muscles generally cues pain resistance, then even participants not primed with health goals (who are unlikely to be recruiting willpower because they are not facing the self-control dilemma to persist drinking the nasty tonic even though they want to immediately stop) will benefit from muscle firming.

Method

Ninety-one undergraduate students participated in two successive studies, one on verbal skills and one on motor skills. We ran participants individually in one-on-one sessions. Once they were seated, we assigned participants randomly to one of four conditions in a 2 (health goal: high vs. low) × 2 (muscle: firming vs. control) between-subjects design. First, as a goal-priming manipulation, participants completed a sentence-scramble task (e.g., Labroo and Kim 2009) that comprised eight sets of five scrambled words. We explained that we were pretesting a verbal skills task and asked them to make a grammatically correct four-word sentence out of each set of five words. In the health-goal condition, the sentences were related to health goals (e.g., “Fitness is a virtue” from “virtue, a, is, to, fitness”). In the control condition, the sentences were neutral (e.g., “This ball is blue” from “this, blue, ball, is, be”). In line with the cover story and as filler measures, participants then evaluated the sentence-scramble task (1 = negative, paid little attention, 7 = positive, paid a lot of attention).

Next we explained to participants that they would participate in a motor-skills task in which they would have to maintain a certain posture while tasting a product. We placed a concoction of 10 parts water and one part Heinz white vinegar in three 7-ounce cups next to where the participant was seated (manipulation adapted from Vohs et al. [2008]) and explained that the drink was a health tonic that had been recently introduced to the market. The tonic was presumably healthy but tasted bad, and therefore they could consume as much or as little as they wanted. We then asked participants to assume a required body posture as they drank the vinegar water. In the muscle-firming condition, we asked participants to lift their heels off the floor to contract their calf muscles. In the control-action condition, we told participants to keep their feet flat on the ground. Experimenters provided a picture in each condition, illustrating how the bodily action should be performed (see appendix) and made sure the participants could perform the actions. Experimenters asked the participants to assume the position when they were ready to consume the tonic and to maintain this position for as long as they drank the tonic.

Experimenters recorded the volume of vinegar drink participants consumed and the thoughts participants had when consuming the drink. Participants verbalized all thoughts they had when consuming the health drink—they were unable to articulate thoughts while they had the tonic in their mouths, so they responded immediately after they were done and between sips. Next, each participant completed a questionnaire pertaining to the actions they had just completed and comprising items almost identical to those used in experiment 2, including process measures (summoned willpower, mind control, mentally strong; 1 = very little, 7 = very much), manipulation check measures (contracted leg muscles, made leg muscles tight; 1 = very little, 7 = very much), and control measures (drink negativity, positive feelings, negative feelings, outcome controllability, mind-body beliefs, perceived body strength, self-perceptions of tolerance of drinking or eating distasteful but healthy food; 1 = I cannot tolerate eating healthy food that tastes bad/I am less able to tolerate eating healthy food that tastes bad than others, 7 = I can tolerate eating healthy food that tastes bad/I am more able to tolerate eating healthy food that tastes bad than others; and general approach-avoidance tendency, with scales identical to those in experiment 2). Participants also evaluated the leg posture they had assumed (1 = bad, negative, dislike, painful, 7 = good, positive, like, painless). Finally, after completing a suspicion check, we debriefed participants. No participant correctly guessed the purpose of the study.

Results and Discussion

Tables 1 and 2 present all means and associated standard errors for manipulation checks, control items, and hypothesis tests.

Manipulation Checks and Control Measures

The two versions of the scrambled sentence task did not differ in goodness or in attention participants paid in either condition (p > .25). Additionally, the evaluation of the leg posture (good, like, positive, painless) did not differ depending on the posture participants assumed (p > .75). As expected, participants who contracted their muscles reported they had done so (r = .74, p < .001) to a greater extent than control participants (Mfirm = 4.59 vs. Mcontrol = 3.42; F(1, 89) = 16.45, p < .01). Priming and muscle action did not influence any other control measures (extent to which participants followed instructions, believability of the cover story, controllability over outcomes, feelings, arousal, body-mind relationship beliefs, self-perceptions of tolerance of distasteful food, and general approach-avoidance tendencies; all p > .15). Everyone thought the drink tasted awful (M = 2.73; p > .30).

Vinegar Consumption (Behavioral Measure)

A 2 (goal prime) × 2 (muscle action) ANOVA on vinegar consumption revealed an expected main effect of goal (Mhealthy = 218.37 milliliters vs. Mnonhealthy = 95.73 milliliters; F(1, 87) = 18.59, p < .01), muscle action (Mfirm = 176.53 milliliters vs. Mcontrol = 127.02 milliliters; F(1, 87) = 3.74, p < .06), and a two-way interaction (F(1, 87) = 3.97, p < .05; see table 1). As expected, among participants who cared about long-term health, muscle firming increased vinegar consumption (Mfirm = 268.70 milliliters vs. Mcontrol = 160.50 milliliters; t(87) = 3.27, p < .01). Participants who did not care about long-term health goals did not show similar effects (Mfirm = 95.00 milliliters vs. Mcontrol = 96.59; t < 1); thus, muscle firming does not generally increase pain resistance.

Self-Report Willpower Index

A 2 × 2 ANOVA on the index created by averaging the summoned willpower measures the revealed expected main effects of goal (Mhealthy = 4.47 vs. Mnonhealthy = 3.17; F(1, 87) = 18.16, p < .01), muscle action (Mfirm = 4.06 vs. Mcontrol = 3.05; F(1, 87) = 4.85, p < .05), and a two-way interaction (F(1, 87) = 4.05, p < .05). Health-primed participants reported having summoned more willpower when they contracted their leg muscles (Mfirm = 5.04 vs. Mcontrol = 3.80; t(87) = 8.44, p < .01); expectedly, participants who did not care about health goals and did not need to summon willpower did not show similar effects (Mfirm = 3.19 vs. Mcontrol = 3.14; t < 1).

Online Willpower Thoughts

Willpower thoughts included, “It makes my throat itchy but I'm making up my mind to go on,” “It smells disgusting, but to keep drinking I'm trying to think I liked it,” and “This is a flavor I really don't like, but if it's good for me I will keep drinking it.” General thoughts included, “It will fail in the market for sure,” “Why am I doing this?” “Aftertaste is even worse than the taste,” and “To be successful they should make it taste like apple cider.” Two coders who were blind to the hypothesis coded these thoughts as willpower thoughts (r = .95) that augment long-range actions or banish temptation thoughts. Participants did not differ in the total number of thoughts they reported, based on goal and muscle action (all p > .15); however, a 2 × 2 ANOVA on the number of willpower thoughts revealed the expected main effects of goal (Mhealthy = .42 vs. Mnonhealthy = .08; F(1, 87) = 9.99, p < .01), muscle action (Mfirm = .33 vs. Mcontrol = .14; F(1, 87) = 4.15, p < .05), and a two-way interaction (F(1, 87) = 9.14, p < .01). Participants primed with health goals reported more willpower thoughts when they contracted their leg muscles (Mfirm = .65 vs. Mcontrol = .15, t(87) = 3.50, p < .01); we did not observe similar effects among non-health-primed participants (Mfirm = .04 vs. Mcontrol = .14, t < 1).

Mediation analyses revealed that willpower thoughts partially mediated the effect of muscle firming on tonic consumption. We observed a positive effect of goal × muscle action on consumption of the health tonic (b = 136.69, SE = 30.78, p < .001) and on willpower thoughts (b = .34, SE = .12, p < .01), and willpower thoughts positively influenced tonic consumption (b = 23.13, SE = 26.83, p < .01). When we included willpower as a covariate in the regression predicting the effect of goal × muscle action on consumption, the effect of willpower thoughts remained significant (b = 95.85, SE = 26.55, p < .001), but goal × action was reduced significantly (b = 104.02, SE = 30.27, p < .01; z = 2.29, p < .02). As in experiment 2, we did not observe a similar mediation with the explicit self-report willpower, which reflects self-perception ratings of willpower people had summoned.

Experiment 3 thus confirms our hypothesis that muscle firming helps summon willpower. By contracting their muscles, participants who valued becoming healthy in the long term—and only those participants—were able to consume more of a nasty tonic that presumably would fulfill a goal of becoming healthy. The generation of willpower thoughts mediated this effect. Furthermore, muscle firming does not generally cue long-term goals; rather, once those goals are accessible and people presumably consciously wish to recruit willpower to achieve those goals, muscle firming augments willpower and enables self-control.

Our goal in experiment 4 is threefold. First, in contrast to experiments 1–3, which find that muscle firming helps individuals withstand more pain in the service of long-term goals, in experiment 4 we investigate whether muscle firming also helps people resist immediate pleasures and avoid indulgences in the service of long-term goals. Second, although experiments 1–3 all employ real behavioral measures (donation, pain, and vinegar consumption), we conducted all of them in a lab setting; in experiment 4, we measure actual purchases by customers over a 10-minute period at a snack bar. Third, instead of manipulating the importance of long-term goals to distinguish between people who wish to momentarily recruit willpower in the service of those goals when faced with temptation and people who do not, we measure chronic differences among individuals pertaining to long-term goals.

Experiment 4: Field Study

Method

We recruited 66 participants for the study as they approached a snack bar on campus. The study employed a 2 (chronic health goal: high vs. low) × 2 (muscle action: firming vs. control) between-subjects design. First, experimenters asked participants to complete a short survey designed to measure their chronic health goals (1 = Fitness is a virtue/I stay in shape/I exercise every week/Health is more important than gratification, 6 = Indulgence is a virtue/I aim for enjoyment/I indulge every week/Gratification is more important than health; α = .89). In an effort to control for any self-generated validity or priming effects because of this measurement, for roughly half of the participants we took this measure at the end of the experiment (no order effects observed from counterbalancing). Then experimenters asked participants to proceed to the snack bar to purchase whatever they had planned to buy but to engage in a specific motor action while doing so. In the firming condition, experimenters asked participants to weave a pen between the stretched fingers of their nondominant hand; in the control condition, we asked them to weave a pen loosely between their index and middle fingers. The experimenter told participants that researchers were interested in studying multitasking in a shopping context, and they told them to come back to be compensated once they had purchased their snacks.

When the participants returned, the experimenter noted what food and drink items the participants had purchased, and participants completed a survey in which they reported how hungry they felt (“Right now, I feel hungry,” 1 = less than usual, 7 = more than usual), how satisfied they were with the purchase (“Right now, I am satisfied with my purchase,” 1 = less than usual, 7 = more than usual), how much they bought (“I bought ____,” 1 = less than usual, 7 = more than usual), how substantial the purchase was (“I bought many things,” 1 = not at all, 7 = very), and how full they thought they would feel once they were done with the food and drinks they purchased (“Consuming this food will make me feel full,” 1 = not at all, 7 = very). Then they indicated how much willpower, body control, and mind control they felt (1 = not at all, 7 = very much; α = .75) when making their food and drink choices. Similar to experiments 1–3, participants then reported muscle-firming manipulation checks (stiffened muscles, tightened muscles; 1 = not at all, 7 = very much) and control measures (control over outcomes, momentary feelings [happy, relaxed, content, calm, tense, worried, aroused, disappointed, discouraged], task negativity [positive, good, like], and body-mind belief [“healthy mind, healthy body,” “strong mind, strong body”]; 1 = not at all, 7 = very much), and general approach-avoidance tendency (1 = I generally approach desirable/good things, 7 = I generally avoid undesirable/bad things). Finally, as a suspicion check, experimenters asked participants to guess the purpose of the study. No participant guessed correctly.

Results

Tables 1 and 2 present all means and associated standard errors for manipulation checks, control items, and hypothesis tests.

Manipulation Checks and Controls

Participants in the muscle-firming condition reported that they firmed their hand muscles to a greater extent than those in the control condition (Mfirm = 4.70 vs. Mcontrol = 3.56; F(1, 61) = 9.37, p < .01). Regression analyses revealed, as one might expect, that indulgent respondents reported feeling hungrier than health-conscious respondents (b = .27, SE = .11, t = 2.46, p < .02), but, importantly, the main effect of muscle action or the interaction was not significant (all p > .45). They also spent less money than health-conscious respondents did (b = −.43, SE = .25, t = 1.74, p = .09), which is not surprising given that healthy items are generally more expensive than unhealthy ones, and they were less satisfied with their purchases (b = −1.56, SE = .36, t = 4.37, p < .01). The effect of muscle action or the interaction was not significant, however (all p > .45). The effects of goal, muscle action, or their interaction on whether respondents purchased more food than usual, how substantial the purchase was, and how full they thought they would feel after consuming their items were all nonsignificant (all p > .40). No other effects of any of the control measures were significant (followed instructions, outcome controllability, feelings, body-mind belief, approach-avoidance tendency; all p > .15).

Food and Drinks Purchased

Two coders labeled items that participants purchased as healthy (e.g., fresh fruit, green tea, yogurt) or unhealthy (e.g., ice cream, butter croissant, candy, chocolate; r = .95). Regression analyses using goal, muscle action, and their interaction as independent variables on the proportion of healthy items purchased revealed a main effect of goal (b = −.07, SE = .03, t = −2.21, p < .03) and a significant interaction (b = −.17, SE = .06, t = −2.73, p < .01). Consistent with findings in experiment 3, spotlight analysis revealed, as expected, a significant effect of muscle firming in increasing the purchase of healthy snacks among people with a health goal (b = .37, SE = .15, t = 2.54, p = .01) but not among people with an indulgence goal who presumably did not wish to summon willpower (b = −.19, SE = .15, t = −1.34, p > .15).

Self-Report Willpower Index

We created an index by averaging the three measures pertaining to willpower. Regression analyses using goal, muscle action, and their interaction as independent variables on willpower revealed a main effect of goal (b = 1.01, SE = .35, t = 2.87, p < .01) and a marginal interaction (b = −.36, SE = .21, t = −1.74, p < .09). Spotlight analysis revealed, as we expected, a significant effect of muscle firming on willpower among chronically health-oriented people (b = 1.63, SE = .50, t = 3.26, p < .01) but not among indulgence-oriented people who presumably did not wish to summon willpower (b = .39, SE = .50, t < 1, p > .40). As in experiments 2 and 3, although these willpower measures followed the predicted patterns, they did not mediate choice, possibly because experimenters collected these self-reports of recollections of willpower after the respondents had exerted willpower. Collecting online willpower thoughts in this study was not possible because participants were left on their own to shop as naturally as possible.

Taken together, the experiments thus far suggest muscle firming plays a role in summoning willpower by making people more able to resist mental pain by opening their minds and hearts to unfortunate others (experiment 1), by making them tolerate physical pain and submerging their hands in ice to improve health (experiment 2), and by consuming an unpleasant but healthy tonic (experiment 3). They are also able to resist momentary pleasures of immediate temptations in favor of long-term goals (experiment 4), provided they value such goals and thus wish to summon willpower in expending self-control. The goal in experiment 5 is to further investigate the role that muscle firming plays in augmenting willpower and enabling self-control and to answer, in particular, the question of when individuals should firm their muscles. Past research suggests that willpower is a limited resource and that once people expend it, they will become depleted and have a difficult time summoning it for future use (Baumeister et al. 2008; Muraven et al. 1998). At first glance, our findings may appear to conflict with what depletion researchers report, because if muscle firming augmented willpower, that effort should have depleted participants' willpower, leaving them too depleted to exert self-control. Notice, however, that participants in experiments 1–4 always firmed their muscles in tandem with exerting self-control; therefore, the summoned willpower was not wasted before it was needed and could instead be applied to the self-control task at hand. For this reason, we argue that when participants firm their muscles during a self-control task—rather than before it—firming will facilitate self-control. However, because willpower is a limited resource (Baumeister et al. 2008), expending it prior to the self-control task will deplete an individual's willpower and lead to poorer performance. Indeed, existing research shows that depletion is most likely to occur when participants infer that they already finished exerting self-control prior to the task, rather than when they see it as an ongoing task requiring self-control (Dewitte, Bruyneel, and Geyskens 2009). Thus a primary objective of experiment 5 was to test a differential effect of the timing of the muscle firming (during vs. prior) on self-control.

Experiment 5: Summoning versus Depleting Willpower

Method

Ninety-eight undergraduate students, all on a diet, received course credit to participate in this study, and experimenters assigned them randomly to one of four conditions in a 2 (action type: firm vs. loose) × 2 (action timing: during vs. before the target self-control tasks) between-subjects design. The key dependent variable was whether participants differentially would choose an indulgent (chocolate) or healthy (apple) snack based on their condition. On arrival at the lab, participants learned that the purpose of the study was to investigate perceptual motor skills. In the muscle-firming condition, experimenters instructed participants to contract their left biceps if they were right-handed, and their right biceps if they were left-handed. Experimenters instructed participants in the loose condition to loosen the respective bicep by pointing their hand downward. The experimenter demonstrated the motor action and made sure participants understood the instruction. All participants then read two self-control scenarios that were unrelated to each other in sequence, and they were assigned at random to engage in the muscle action while reading either the first or the second scenario. The first scenario was a filler 130-word scenario, described as boring but important because participants could be tested on it, thus requiring willpower, and the second (i.e., the target self-control tasks) was a choice scenario (adapted from Labroo and Patrick [2009]) in which participants imagined they were Pat and on a diet to look attractive and feel healthy. One day, Pat's close friend offered Pat a very tempting chocolate mousse cake, Pat's favorite. After reading the target scenario, participants evaluated whether they, as Pat, should eat the cake (1 = strongly disagree, 7 = strongly agree), whether eating the cake posed a self-control dilemma because it was against the diet plan (1 = strongly disagree, 7 = strongly agree), and whether eating the cake was socially appropriate (1 = strongly disagree, 7 = strongly agree). All participants were then instructed that they were now outside the scenario and that, as a token of appreciation for completing the study, they should now choose a snack for themselves: either a low-calorie apple or an indulgent chocolate bar. They then completed control items identical to those used in experiments 2 and 3.

We predicted that participants would be more likely to choose an apple over a chocolate bar when they previously firmed muscles in the Pat scenario because their own choice and Pat's choice are similar and can be seen as part of the same ongoing self-control dilemma (Dewitte et al. 2009). However, if participants had instead previously firmed muscles to attend to the boring scenario, they would now be able to expend less self-control while making their own choices because the boring scenario was dissimilar to the choice scenario. Participants would view the second self-control scenario as a different self-control task, and as they had already exerted self-control on the previous task, they would act as if they were depleted. Given that participants were completing two scenarios and that reporting thoughts aloud on one scenario might somehow influence their processing of the second scenario, we did not ask participants to report their thoughts. We also did not ask them to report how much willpower they, as Pat, were able to summon while making a decision on Pat's behalf because of the awkwardness of the question. In a suspicion check, no participant guessed the true purpose of the study.

Results and Discussion

Tables 1 and 2 present all means and associated standard errors for manipulation checks, control items, and hypothesis tests.

Manipulation Checks and Control Variables

Participants did not differ across conditions in their evaluation of the filler scenario (M = 3.24; all p > .45). Muscle-action type or action timing did not influence the extent to which participants agreed that eating the cake was against Pat's diet plan (all M > 5.68; p > .25) or that eating the cake was socially appropriate (all M > 3.56; F < 1). Also, none of the effects of any of the control measures was significant (followed instructions, task negativity, task involvement, a general controllability over outcomes, feelings, and general approach-avoidance tendencies; all p > .25).

Resisting Temptation

First, a 2 × 2 ANOVA conducted on the measure of whether Pat should eat the cake revealed a main effect of action timing, such that participants were more likely to indicate that Pat should eat the cake if they engaged in muscle firming on the boring task prior to reading the Pat scenario (Mbefore = 4.96 vs. Mduring = 4.02; F(1, 94) = 15.43, p < .01), and a significant interaction between action type and action timing (F(1, 94) = 10.82, p < .01). As expected, when participants engaged in muscle firming when reading the Pat scenario, the action reduced the extent to which they thought they, as Pat, would consume the cake (Mfirm = 3.70 vs. Mloose = 4.43; t(94) = 2.32, p < .01), but when muscle firming preceded the Pat scenario in the boring scenario, the action increased the extent to which participants thought they, as Pat, should consume the cake (Mfirm = 5.38 vs. Mloose = 4.59; t(94) = 2.30, p < .01). Thus, muscle firming has a beneficial effect on self-control only when individuals engage in it while trying to exert self-control.

Additionally, a 2 × 2 binary logistic regression on participants' own choice of a healthy or indulgent snack revealed a main effect of action timing (χ2(1) = 7.20, p < .01) and of action type (χ2(1) = 6.97, p < .01) and a significant interaction (χ2(1) = 8.99, p < .01). When firming accompanied the Pat scenario, it subsequently decreased participants' choice of chocolate as their own snack (χ2(1) = 4.34, p < .04), but when it accompanied the boring scenario, it subsequently increased the choice of chocolate as their own snack (χ2(1) = 6.05, p < .01). In line with but also extending the findings of Dewitte et al. (2009), this finding suggests that muscle firming on a previous occasion can beneficially influence subsequent self-control, but only when the two self-control tasks are similar enough to be seen as part on the same ongoing self-control task. Thus these results provide added insight into when and how muscle firming can facilitate self-control. When people firm muscles prior to exerting self-control, subsequent self-control efforts are impaired, presumably because summoned willpower is exhausted and wasted. In contrast, when muscle firming summons willpower while one is exerting self-control, we observe facilitative effects on self-control (Pat task) and even, as Dewitte et al. (2009) would suggest, on a subsequent task (own choice) similar to the self-control task (Pat task) that employed muscle firming.

General Discussion

Five studies provide remarkable evidence that simply firming one's muscles can firm one's resolve and facilitate self-control. Much research over the past century has looked at the issue of self-control and has tried to uncover ways to help people overcome immediate temptations and instead accomplish what is good for them in the long term. Indeed, self-control dilemmas are omnipresent in most people's lives. Much of the time, people go about trying to work rather than relax, exercise rather than become a couch potato, and regulate their diets rather than give in to indulgent food choices—in general, pushing their endurance, physical and mental, further in an effort to become better people in the long term. Past research has suggested that people can exert better self-control if they try to boost thoughts about their long-term goals and avoid thoughts about immediate temptations. But this form of conscious self-control requires willpower, and willpower is a limited resource that can become easily exhausted, resulting in people giving in to immediate temptations (Baumeister et al. 2008). The current research is important because it suggests a powerful way to firm willpower: a person can simply firm her muscles. Put simply, steely muscles can lead to a steely resolve.

Across our studies, we triangulated on our premise that muscle firming facilitates self-control by firming willpower in several ways. We used multiple operations of muscle firming: clenching one’s fist, stretching one's fingers, tightening one's calf muscles, and firming one's biceps. We also employed different kinds of self-control tasks, ranging from those that required a person to take on immediate pain for future gain (e.g., considering immediately disturbing information, putting a hand in an ice bucket to improve blood circulation, and consuming a nasty but healthy vinegar tonic; experiments 1–3) to those that required avoiding immediate pleasure for future gain (e.g., avoiding tempting food in a snack bar and choosing a healthy apple over a chocolate bar; experiments 4 and 5). We found these effects in a natural field setting in which participants went about a shopping task they had come to perform (experiment 4), and we found them in controlled lab settings that tapped into various measures of underlying process (experiments 1–3, 5). We established the underlying mechanism by mediation (ongoing willpower thoughts; experiments 2 and 3) and by moderation (the effects occur only among people who want to recruit willpower, for chronic reasons [experiment 3] or situational reasons [experiment 4], and only when they have not depleted willpower on a previous occasion [experiment 5]). Importantly, online willpower thoughts that participants generated while engaging in the self-control task mediated increased self-control (experiments 2 and 3), which suggests that self-control depends on the actual willpower people summoned when engaged in self-control and that muscle firming augments willpower. Equally importantly, (a) muscle relaxing is similar to a no-action condition, as we would expect. We predict only that muscle firming increases willpower firming and not that muscle relaxation depletes willpower (experiment 2) and that (b) firming muscles by clenching a fist does not make people “tightfisted”; rather, it helps them open their hearts and wallets for others, even though doing so is upsetting in the moment. The fact that our muscle-firming manipulations did not influence our control measures, including those pertaining to feelings of control, affect or mood, arousal, goal orientation, self-perceptions and inferences regarding pain tolerance, and attention gives us additional confidence in our findings.

Theoretical and Practical Contributions

The current research is important for both theoretical and practical reasons. From a theoretical perspective, this article makes an important contribution to the embodied cognition literature by being the first to demonstrate the role that embodied cognition can play in self-regulation. It is the first to demonstrate that the body, not just the mind, can influence self-control and that, importantly, the body can influence self-control by engaging the mind. And we propose a new process by which the body influences the mind: the body helps automatically summon and firm willpower but does so in a strategic manner (Dewitte et al. 2009), depending on whether one desires to engage in self-control and whether a prior task that required self-control is compatible with a current task engaging self-control. The finding is consistent with research in emotion regulation showing that situational cues can evoke automatic regulation goals that help augment conscious goal pursuit, especially among people who really wish to engage in regulation but are tempted in the moment and so for whom conscious regulation could fail (Williams et al. 2009).

The current research also contributes to past research on self-regulation and self-control. Published research suggests that people have a limited supply of resources and that the ability to exert self-control depends on the availability of these resources (Baumeister et al. 2008). Much evidence suggests that individuals' capacity to exert self-control in long-range actions would be impaired if individuals engage in any prior self-regulation, unless the two self-control tasks are related and are seen as a single ongoing self-control activity (Dewitte et al. 2009). The present findings are consistent with and extend those findings in important ways. First, we showed that firming muscles while engaging in self-control facilitates self-control (experiments 1–5). We observed these enhancing effects only when the act of muscle tightening occurred among participants trying to engage willpower and only while they were engaging in willpower. Consistent with existing research showing that prior self-regulatory efforts lead to decrements in subsequent exertions of self-control, we observed depletion in self-control when individuals had already expended willpower to complete a previous task (experiment 5). For instance, when participants had expended willpower in paying attention to a boring task, they were less able to resist indulgences in a subsequent self-control task (experiment 5). Second, compatible with and extending the findings that Dewitte et al. (2009) report—that expending willpower on an initial task is depleting only when the two tasks are not similar in the type of self-control required—we also found that the similarity between the two self-control tasks qualified the depletion in willpower resulting from having exerted willpower in a previous task. In particular, when muscle-firming participants previously expended willpower to pay attention to a boring task, they were more depleted on a subsequent indulgence task (own choice), but when they had previously recruited willpower in an indulgence task (Pat's self-control choice), they showed high levels of self-control on a subsequent indulgence task (own choice).

From a public policy and consumer well-being perspective, the current research has important implications because it demonstrates subtle but powerful ways in which people's actions can alter their lives for the better. In this regard, to demonstrate the normative implications of our findings, we conducted a field study (n = 61). Participants who were on their way to the campus snack bar at lunchtime were intercepted and asked to indicate why health goals are important and why self control is a dilemma. Before they proceeded to the snack bar, we either told them that we are researchers whose findings show that “tightening muscles helps self-control” (correct information condition) or that “relaxing muscles helps self-control” (wrong information condition) or did not give them any information (baseline/control condition). We did not tell them that we would be coding what they purchased, but once they completed their purchases, they returned to our information desk on the pretext of completing a study for payment. After checking whether participants in the two information conditions believed the information and decided to follow it on their own (all did), to their surprise—and similar to experiment 4—we coded the proportion of healthy items participants purchased (interjudge correlation = .92). Notably, nine participants returned about 10 minutes later without purchasing anything (five with correct information, four controls). Results (excluding those nine participants’ data; coding their decision as healthy and including the data strengthens our results) showed an expected main effect of information condition (F(2, 49) = 6.66, p < .01); those who were given correct information purchased a greater proportion of healthy food (Mtighten_helps = .66) than did those given wrong information (Mrelax_helps = .33; t(49) = 2.39, p < .01) or those given no information (Mno_info =.14; t(49) = 3.38, p < .01; relaxing helps vs. no information: t(49) = 1.29, p > 20). There were no effects of information on the total number of food items bought.

Thus, our findings suggest a pervasiveness of the effect of embodied cognition on self-regulation and additionally present an easy but effective way to increase the probability of one's success at self-regulation. We also identify when such actions of muscle tightening may benefit self-regulation and when they may deplete self-control. When people tighten their muscles while exerting self-control, they can also strengthen willpower and thus further enhance their self-control. It is particularly impressive that the beneficial effects of muscle tightening were found with a number of manipulations (stretching fingers, making a fist, contracting leg muscles) and in a variety of self-control tasks (parting with money, tolerating the physical pain of inserting one's hand in an ice bucket, consuming a terrible drink of vinegar, and restricting indulgent actions). These multitudes of operations and decision contexts suggest a generality of situations wherein the observed effects may be useful in changing consumer behavior. Also important is that these effects arose mostly when people cared about self-regulation; firming their muscles did not increase self-control among people who were unwilling to exert willpower or did not have long-term goals in mind.

Ruling out Alternative Accounts

An account that muscle firming generally increases pain tolerance or cues long-term goals is unlikely to be a viable explanation for our findings. Aside from the issue that a possible theoretical reason for such a prediction that muscle firming generally enhances pain tolerance or cues of long-term concerns is unclear, we designed experiments 3–5 specifically to ensure against this possibility. Research has defined self-control as taking on immediate pain (or resisting immediate pleasure) in the service of goals that are of greater long-term importance (Baumeister et al. 1994; Gollwitzer and Moskowitz 1996; Metcalfe and Mischel 1999); and in line with past research, we expected people to summon willpower to engage in self-control only if they already valued the long-term benefits associated with taking on the immediate pain (or resisting the immediate pleasure). Thus, in experiments 3 and 4, we manipulated or measured the extent to which people value their long-term goals, and we found that only people who were facing a self-control dilemma because they valued the long-term benefits of self-control but were tempted to avoid pain (experiment 3) or to indulge (experiment 4) in the moment were better able to summon willpower when firming their muscles and to engage in self-control. Had muscle firming generally increased pain tolerance, we should have found that muscle firming, regardless of whether people faced a self-control dilemma, increased self-control. Additionally, if muscle firming generally increases pain tolerance (or indulgence resistance), then when people firmed their muscles in experiment 5 should not have mattered, and we should have simply observed a main effect of muscle firming. Also, our data provide direct evidence that summoning online willpower mediates self-control, and measures designed to directly capture general pain tolerance showed absolutely no movement based on muscle firming (see table 2). Thus an account that muscle firming generally increases pain tolerance or cues long-term goals or generally cues people to firm their self-control is quite unlikely.

For similar reasons, a suggestion that muscle firming somehow increased participants' ability to “do the right thing” is likely implausible. Again, aside from the issue that experiments 3–5 go against such an explanation for reasons similar to those described above, what doing the right thing is and whether self-control is necessarily the right thing are unclear. Our experiments show when respondents act in their long-term interests in favor of immediate temptations, but to say they are doing the right thing presumes that exerting self-control is the right thing to do, and that assertion is debatable. Research shows that exerting self-control can lead to regrets later in life (Kivetz and Keinan 2006), and in that sense, exerting self-control may actually be the wrong thing to do. But even if we were to agree that acting in one’s long-term interest is always the right thing to do, which we do not, our data (experiments 3–5) show that firming muscles does not always result in the right thing to do. Experiment 5 is particularly problematic for such an interpretation because it shows that firming muscles before a self-control task results in less self-control (depletion-like effects). If firming muscles primes “do the right thing,” then we are unclear on why the priming must be ongoing during a self-control task and, even more so, why priming “do the right thing” before the self-control task is depleting to participants. Future research could, however, additionally directly test these issues.

Self-perceptions or inferences participants made about themselves based on their muscle actions (i.e., self-perception; Bem 1972; Briñol and Petty 2008) also appear unlikely based on the evidence we provide. The basic premise underlying self-perception theory is that “individuals come to ‘know’ their own attitudes, emotions, and other internal states partially by inferring them from observations of their own overt behavior,” and this inference arises when “internal cues are weak, ambiguous, or uninterpretable, and the individual is functionally in the same position as an outside observer” (Bem 1972). These ideas pertaining to self-perception theory were extended to the embodied cognition literature by Labroo and Nielsen (2010), who suggested that psychological sensations of approach in space or time toward an outcome can make people infer that the outcome is more desirable than what they would otherwise infer. If the current effects were similarly based on self-perception, they would be compatible with those findings and would still meaningfully extend past findings by describing a novel effect and being the first to do so in the context of exerting self-control. However, self-perception theory does not appear to account for the current effects.

First, in this article, we found that online willpower thoughts mediated self-control but, interestingly, that self-reports of expended willpower, which should correspond more closely with self-perceptions, did not. This finding is particularly important because it suggests that although willpower does improve self-control, people are not able to accurately report how much willpower they are feeling. Thus the process by which the body influences the mind is more likely to be one in which muscle firming automatically activates mental resolve and facilitates automatic self-control, similar to the ways Williams et al. (2009) report. For a self-perception account to hold (Bem 1972), muscle firming should have influenced these self-reported inferences about how much mind and body control participants had, and these self-perceptions should have mediated self-control. We additionally found that muscle firming did not influence participants' self-reports of self-perceptions of pain tolerance; if muscle firming influenced self-perception, this inference should have been influenced. Experiment 5 also goes against a self-perception account: if firming muscles increases the self-perception that one is good at self-control or has abundant willpower, then why the timing of the muscle action should matter is unclear. Self-perception theory would predict a main effect of muscle action regardless of when the action is undertaken. By distinguishing that the mechanism underlying the impact of the body on the mind in the context of self-control is likely to be an automatic memory-activation process rather than a self-perception one, the current research further enriches the embodied cognition research. Future research should further investigate conditions when self-perception versus memory-activation processes may underlie the impact of the body on self-control. It should also provide more evidence of automatic cuing resulting in augmentation of conscious self-control pursuit and should further delineate situations when self-perception will facilitate self-control and when automatic augmentation of willpower will underlie self-control.

Second, measures of arousal also did not show movement; thus, arousal is unlikely to account for our effects. But even if firm muscles had influenced the measure, such that firm muscles increased arousal, arousal usually impedes self-control (Fedorikhin and Patrick 2010; Shiv and Fedorikhin 1999; Van den Bergh, Dewitte, and Warlop 2008). Thus, if anything, effects opposite from those we observe would have been predicted; thus, arousal is unlikely to reasonably explain our findings. Similarly, involvement also cannot account for our effects. Not only did we observe that participants across conditions did not vary in the actual number of thoughts or in self-reported attention, but why muscle firming should increase elaboration is unclear except for the reasons we suggest—that the body does influence one's thoughts—but in this case, not the number but the nature of one’s thoughts. The data also show that distraction does not account for the observed effects. First, we found mediation by online willpower thoughts that participants generated; participants did not generate distraction thoughts (e.g., “Oh, let me not think of the pain”). Second, we found that muscle action facilitated self-control in some but not all conditions (experiments 3–5); if muscle firming increases distraction, it is not clear why it should not help everyone facing pain distract themselves. Third, the control data (painfulness of tasks, general tolerance for pain, attention) also did not show differences based on muscle action; a distraction account could reasonably be expected to register movement on at least some of these measures. Additionally, in some experiments (e.g., experiments 4 and 5), participants had to make proactive choices (e.g., choose healthier options), and why distraction should increase self-control rather than reduce it in these choices is unclear (i.e., why distracted people would make healthier choices). Also, experiment 2 additionally included a muscle-relaxation condition, and why firming should be more distracting than relaxing is unclear.

Future Research

An area for future research might be an investigation of whether awareness of an influence of muscle firming on self-control lessens the benefits of that influence on self-control. Although all of our participants were obviously aware of engaging in self-control and engaging in a muscle action, we did not ever tell them that muscle action might be influencing their self-control or augmenting willpower. Past research, however, has shown that when people become aware of an external influence on their judgments and behaviors, they correct for the influence of that external source (Labroo and Nielsen 2010; Schwarz 2004). Thus, if they were to become aware that their body influenced their mind, they would correct for any such influence. We think such a possibility is unlikely. First, our field data from the experiment in which we gave participants the information showed that as long as participants believe that muscle firming enhances self-control, they are better at self-control. This finding is in line with the proposition that correction effects are usually observed for decrements in performance or judgment biases that participants wish to avoid. But self-control is something people would like to engage in more, and so conscious awareness most likely will not lead to a reduction in self-control. Second, such corrections are observed only when the bias also influences self-perceptions; but in our data, we did not find evidence of a self-perception account. Third, we support a view that muscle-firming cues willpower strengthening because it usually accompanies willpower strengthening, and this automatically cued willpower strengthening augments conscious willpower strengthening in self-control efforts. If willpower augmentation is the underlying process, then we would not expect source awareness to lead to correction effects.

Future research should employ additional methods to distinguish between the conscious and nonconscious aspects of willpower strengthening and the role of the body in either process and should investigate the impact of source awareness on willpower strengthening. Distinguishing the body’s role in engaging the nonconscious and the conscious, in this manner, is likely to be a particularly useful exercise because our data suggest that the nonconscious can serve as a powerful resource when it works in conjunction with the conscious to enable self-control among people who face the most intense self-control dilemmas. Thus, to the extent that our manipulations can nudge people who want to exert self-control to change their behaviors more easily and in a rather costless manner while not impairing the behaviors of those who choose not to engage in self-regulation, and to the extent that our manipulations can be used in situations when people wish to exert self-control but not when they choose not to engage in self-control, these findings are likely to enhance individual and social welfare and to have important policy implications.

Enhancement

Online Color Figures

Appendix Stimulus Materials (Experiment 3)

Figure A1.

Control Condition (No Muscle Tightening)

Figure A2.

Muscle-Tightening Condition

References

View Abstract