Willpower Is Not a Resource
The idea that willpower is "drained" is compelling. And wrong.
Posted August 25, 2011
September 1st is the official release date of Willpower: Rediscovering the Greatest Human Strength by Roy Baumeister and John Tierney. I haven't read the book, but from the description along with Tierney's piece on Sunday in the New York Times, I feel comfortable inferring that the book argues - drawing on the work and ideas of Baumeister and his colleagues over the last decade or so - that willpower is like a muscle, getting exhausted with use. (Another metaphor used in this area is that willpower is a resource that gets depleted with use.)
This idea has a visceral, compelling feel to it - it does feel like I'm tired and drained after making decisions - and the idea has received a tremendous amount of attention from scientific and lay communities alike, perhaps partially explaining the duo of scientist/journalist on the dust jacket.
There are a number of problems with the model, and in this post, I'll talk about one of them: the model has been falsified a number of times, including by research performed in the lab of the proponents of the model. (In later posts, I'll discuss other problems.)
The basic finding in this research area is that if subjects do one "self-control" task,1 such as not eating tempting brownies placed in front of them, these subjects are worse at a subsequent task requiring "self-control," such as trying to figure out unsolvable anagrams. The explanation is that the first task exhausted the muscle/drained the resource.
If indeed the correct explanation for poor performance were that a resource has been depleted, then people's beliefs about that resource shouldn't matter. If I'm out of gas but I think the tank is full, the car still won't go. Two studies (Martijn et al. 2002; Job, Dweck and Walton, 2011) investigating this found that beliefs do matter. In one study, people told that the resource model was false actually did better than controls, a finding difficult for a resource model to accommodate.
Another line of evidence that undermines the resource account comes from the Tice/Baumeister lab itself. Again using the two-task method, Tice et al. (2007) showed that they could eliminate the effect of doing the first self-control task if, in the middle, subjects were shown a funny video or given a surprise gift. The only way that this can be accommodated by the model is if we grant that getting a gift or watching a funny video can replenish the resource, which makes the resource a strange sort of thing.
The resource view seems to entail that incentives shouldn't matter: a runner who has hit her wall might well not be able to power through, no matter how important winning the race is. But in "self control" tasks, incentives do matter. Participants in studies who are told that their performance on the second of two tasks will benefit themselves or others eliminates the performance decrement induced by the previous task (Muraven & Slessareva, 2003). Similarly, the story you tell subjects surrounding the task matters: framing a laboratory task such as squeezing a handgrip as long as possible as a test of a subject's "willpower" improves performance compared to a neutral framing (Magen & Gross, 2007). If you take the resource account as a model that suggests that performance is literally constrained by a resource that gets depleted by a "self control" task, these findings undermine the model. (See also Ackerman et al., 2009).
None of this is to deny that there's an interesting phenomenon to be explained in other studies, and that certainly some explanation is needed. (One caveat: there are published reports of failures to replicate the basic effect.) The repeated failures of the model to explain the data suggests the resource model is at best incomplete and at worst just wrong.
Now, you might object that I've asked too much of the model, and what the model really says is that there's some resource that limits performance, but other factors, such as "motivation" also matter, and that once the resource is depleted a little bit, people husband their supply of it.
And you would be justified in this objection. Baumeister and colleagues, it seems, don't actually think that people's performance on these tasks is limited by a putative resource. As Baumeister and Vohs (2007) put it, using a reservoir metaphor, "If the tank were truly and thoroughly empty, it is unlikely that increasing incentives would counteract depletion" (p. 11). Similarly, Baumeister, Vohs, & Tice (2007) suggested that "people can exert self-control despite ego depletion if the stakes are high enough."(p. 352) and, further - and this is the key part - they say that their work has never actually shown that people were literally unable to "control themselves," that "[p]ragmatic and ethical limitations have prevented us from showing this in laboratory work thus far" (p. 353). That is, these researchers seem to be saying that none of their data can be explained by depletion in itself, that all of their subjects could have controlled themselves; they chose not to.
In short, they seem to be saying the resource in question is not necessary to exert self-control, that self-control does not require the resource, an apparent substantial departure from the original statement of the model (in 2000), the very first assumption of which was (p. 248, my emphasis):
Self-control strength is necessary for the executive component of the self (i.e., the aspect of self that makes decisions, initiates and interrupts behavior, and otherwise exerts control) to function. Acts of volition and self-control require strength.
That's obviously fine. People should be free to change their minds, and they should as new data come in. But a second consequence of taking this position is more serious; there is now no pattern of data that can possibly falsify the revised model. If reductions are observed after an act of self-control, a victory is scored for the model. If no reduction is observed - as in the work described above - the model is declared intact because the resource isn't required to exert self-control.
This problem is not unique to the self-control model. Navon (1984), writing about resource models, in general, observed that the
frequent cases in which the predictions do not bear out are dismissed by resorting to built-in escapes in the theory, such as, data limits, operation below full capacity, disparate resource composition, and so forth. This is probably the source of the self-reinforcing nature of the concept and the unfalsifiable status of the theory (p. 231, emphasis added).
So, to the extent that the model can be falsified, it has been. I leave to others the interesting question of why reviewers, editors, and publishers don't seem to care about this. Next time, I'll address the claim that glucose is the resource. (For a preview, I have a paper on the topic, and material in Chapter 8 of WEEH.)
1 I put "self control" in quotation marks there because I do not really understand what, exactly, makes a task a "self control" task. The tasks that come under this heading are very diverse.
Ackerman, J. M., Goldstein, N. J., Shapiro, J. R., & Bargh, J. A. (2009). You wear me out: The vicarious depletion of self-control. Psychological Science, 20(3), 326-332.
Baumeister, R. F., Vohs, K. D., & Tice, D. M. (2007). The strength model of self-control. Current Directions in Psychological Science, 16(6), 351-355.
Job, V., Dweck, C. S., & Walton, G. M. (2010). Ego depletion-Is it all in your head? Implicit theories about willpower affect self-regulation. Psychological Science, 21(11), 1686-93.
Magen, E., & Gross, J. J. (2007). Harnessing the need for immediate gratification: Cognitive reconstrual modulates the reward value of temptations. Emotion, 7(2), 415-428.
Martijn, C., Tenbult, P., Merckelbach, H., Dreezens, E., & de Vries, N. K. (2002). Getting a grip on ourselves: Challenging expectancies about loss of energy after self-control. Social Cognition, 20(6), 441-460.
Muraven, M., & Slessareva, E. (2003). Mechanisms of self-control failure: Motivation and limited resources. Personality and Social Psychology Bulletin, 29(7), 894-906.
Navon, D. (1984). Resources - A theoretical soup stone? Psychological Review, 91(2), 216-234.
Tice, D. M., Baumeister, R. F., Shmueli, D., & Muraven, M. (2007). Restoring the self: Positive affect helps improve self-regulation following ego depletion. Journal of Experimental Social Psychology, 43(3), 379-384.
Robert Kurzban. Copyright 2011. All rights reserved.