No Sugar Coating Problems for the Glucose Model

So, I got a unique invitation from creator John Brockman back in September of last year. He was organizing an event which he dubbed HeadCon. He invited a group of scholars to give brief talks on the topic of our choice, to be recorded close up. The camera coming in tight on people’s heads was the origin of the “Head” part of HeadCon.

Our marching orders were to talk about what was new in our respective areas of social science and why whatever was new mattered, eschewing our “canned” presentation, keeping our remarks “conversational.” The talks, along with a pretty creepy video of a face, can be found on the HeadCon web page.

I used the opportunity to talk about a topic that I’ve been oddly obsessed with over the last few years, the idea that “willpower” is fueled by a mysterious resource, and the subsidiary idea that glucose is the mysterious resource in question. (I’ve discussed these ideas in prior blog posts. I and some of my colleagues at Penn have also recently provided an alternative account, for those interested.  Another account just came out in Trends in Cognitive Science.)

To get at this (narrow) topic I framed my remarks in the (broad) context of the current discussion in psychology about replication. (Indeed, the title on the web site of my talk is P-hacking and the Replication Crisis.) More or less, the theme I tried to hit – and now you don’t have to sit through the video of my presentation – is that there’s just a ton of reasons to think that the “resource model” of self-control is wrong and yet the model… Just. Won’t. Die.

At HeadCon, I told the story of my experiences with the willpower-as-resource model, which I quote here a bit, despite how unseemly it is to quote oneself.

… I started just talking informally with colleagues about this. I would go to give talks in places and, lo and behold, it turns out there’s this kind of background radiation—there’s the dark matter of psychology – which is a few people who fail to replicate and don’t publish their work and also don’t talk about it … It’s sort of like sex, it’s the thing that we’re all doing, we’re all replicating, we just don’t want to talk about it too much, right?…Once I did that I started getting the sense that I was fishing into literature where there’s no there there. … more and more work is coming out that’s very difficult to interpret under the willpower model.

I’m pleased to report that after discussing my frustration with the fact that the resource model refused to die its appropriate death, the good people at Edge chose to run with my suggestion that this could be a good Edge Question. To wit, the 2014 question was: “What scientific idea is ready for retirement?” The Times just ran a little piece on it.) (And, I admit, I basically just cribbed from my talk for my own answer. Sue me.)

In any case, it’s in this context that I thought I would discuss, briefly, a new paper out in Appetite. The (provocative) title is, Sweet Delusion: Glucose Drinks Fail to Counteract Ego Depletion, and the authors – Florian Lange and Frank Eggert – report two experiments (N = 70, N = 115) in which they tried to replicate the results of two studies (Gailliot et al., 2007; Hagger et al., 2010) purporting to show that drinking a sugary beverage improves subsequent performance on a “self-control” task.  The idea is that glucose is the fuel for self-control, so drinking a sugary drink will improve performance on a task the requires self-control.

Lange and Eggert begin their paper by pointing out a number of reasons to doubt the glucose-as-resource as model, including referring to the recent paper by Schimmack, which discussed some statistical worries about one of the key papers in this literature, the 2007 paper by Gailliot et al. in the Journal of Personality and Social Psychology. Lange and Eggert add a number of other worries regarding these findings, including pointing to some errors in a recent meta-analysis by Hagger et al., (2010), writing:

In sum, the original (Gailliot et al., 2007) and meta-analytic (Hagger et al., 2010) evidence is less compelling than suggested, illustrating that the effect of sugar supplementation on ego depletion is far from being an established research phenomenon.

In the first study, subjects drank either a regular sugary beverage or a sugar-free version, and were subsequently given a task that measured discounting, willingness to forgo a smaller, sooner payoff in favor of a larger, later one, with choosing the latter sorts of payoffs considered to indicate a greater ability to exert “self-control.” They found no effect of condition, despite their calculation that “an effect of glucose consumption on ego depletion as large as reported by Hagger et al. (2010) or Wang and Dvorak (2010) could have been detected with a probability close to 1 (1-β > .99).” A second experiment, in which subjects merely rinsed with the glucose solution, instead of drinking it, yielded a similar null result. The authors conclude:

In line with an increasing body of evidence demonstrating that (a) exerting self-control is unlikely to reduce blood glucose levels (Kurzban, 2010; Molden et al., 2012), and (b) performance on a second self-control task does not vary as a function of participants’ blood glucose levels (Dvorak & Simons, 2009; Schimmack, 2012), the present results question the validity of the glucose model of self-control. Especially in view of the above-mentioned shortcomings of studies supporting the role of blood glucose in ego depletion, the model’s major strength appears to lie in its ostensible support for “the folk notion of willpower” (Gailliot & Baumeister, 2007, p. 304) and not in its empirical corroboration… These findings require models positing a major role for glucose in self-control to be fundamentally revised if not completely abandoned.

I was recently surprised to get a phone call from someone involved with organizing the 2014 American Psychological Society (APS) convention in San Francisco in May. The idea was to have a public debate between me and Roy Baumeister about the depletion model of self-control. I agreed to participate in the debate, but was recently informed that the other party declined the opportunity to participate. So, I’ll just present my own thoughts during a session at APS this year. Should be fun.


Dvorak, R. D., & Simons, J. S. (2009). Moderation of resource depletion in the self control strength model: Differing effects of two modes of self-control. Personality and Social Psychology Bulletin, 35(5), 572-583.

Gailliot, M. T., Baumeister, R. F., DeWall, C. N., Maner, J. K., Plant, E. A., Tice, D. M., Brewer, L.E., & Schmeichel, B. J. (2007). Self-control relies on glucose as a limited energy source: willpower is more than a metaphor. Journal of personality and social psychology92(2), 325.

Hagger, M. S., Wood, C., Stiff, C., & Chatzisarantis, N. L. D. (2010). Ego depletion and the strength model of self-control: A meta-analysis. Psychological Bulletin, 136(4), 495-525.

Kurzban, R., Duckworth, A., Kable, J. W., & Myers, J. (2013). An opportunity cost model of subjective effort and task performance. Behavioral and Brain Sciences36(06), 661-679.

Lange, F. & Eggert, F. (2014). Sweet delusion: Glucose drinks fail to counteract ego depletion. Appetite.

Molden, D. C., Hui, C. M., Noreen, E. E., Meier, B. P., Scholer, A. A., D’Agostino, P. R., &

Martin, V. (2012). The motivational versus metabolic effects of carbohydrates on selfcontrol. Psychological Science, 23, 1130–1137.

Wang, X. T., & Dvorak, R. D. (2010). Sweet future: Fluctuating blood glucose levels affect future discounting. Psychological Science, 21(2), 183-188.

16. January 2014 by kurzbanepblog
Categories: Blog | 7 comments

Comments (7)

  1. What’s to debate? There is statistical evidence that one of the original JPSP packages was incredible. The data from Study 1 are missing. The effect seems difficult to replicate (one
    by a colleague of mine posted on the Psych File Drawer) and others have provided plausible alternative explanation for the positive hits in the literature.

    If the crew on the other side still believes in the model, then let’s see a preregistered large sample study done completely in the light of day to test some key predictions. If they have come to see the veracity of the critiques, why not just declare that the idea has not survived empirical and conceptual scrutiny so the field can move on. I read some piece in the APS Observer talking about glucose yesterday and it made me sad.

    Meehl and Lykken seem to have been spot on – theories almost never truly die in psychology. This is such a good example of this phenomenon in 2014.

    Mostly I just want to say “Thanks” for taking this on!

  2. I think it’s a shame that so many people have had to waste so much time and effort trying to empirically disprove a theory that never had a chance of being right in the first place. As Rob has pointed out previously, the glucose hypothesis is in principle the “wrong kind of explanation,” analogous to attributing a slow-running software program on a computer to a depleted battery. My own question about the theory has always been why it should require so much energy for the self-control “muscle” to suppress spontaneous impulses or urges, but it doesn’t require energy to generate and maintain those impulses or urges in the first place. Or to put it another way, why doesn’t glucose strengthen the to-be-suppressed impulses as much as,or more than, the self-control processes postulated to suppress them?

    Once you remove the glucose-as-mechanism part of the equation, the “self-control” model doesn’t seem to be anything more than a new bottle for Freud’s id-versus-superego wine — yet another example of Standard-Social-Science-Model Psychology going in circles.

  3. Great post Rob! And it is too bad that Baumeister won’t debate or discuss these issues with you. I’m disappointed.

    I see two major problems for this area of research in general, and that’s without talking about glucose, which is so obviously incorrect on so many levels (as first detailed by you in 2010). The first problem is with the demonstration of the phenomenon itself and the second is with the explanation of the phenomenon (if there is any phenomenon to explain).

    The second problem may be easier to tackle than the first, so maybe we can discuss it quickly. Given the many problems with the resource account (what is the resource? so many studies that are inconsistent with it, etc), what can explain the apparent limits of control at Time 2 after cognitive work at Time? You’ve offered a nice opportunity cost model; we have offered a complimentary account (thanks for linking it). There may be others that emerge, and I see this as a good thing because the resource explanation simply does not hold up.

    The first problem is more worrying. Yes, depletion studies seem to replicate, but given what we now know about p-hacking (intentional or otherwise), the problems with publication-bias, and the “conference-rumors” about many people’s failure to replicate the basic effect, there are hard questions about how replicable or real this phenomenon is. As you said, “is there any there there?” When conducting a bias-corrected meta-analysis, Mike McCullough, suggests the effect is either not there or perhaps rather small. Now I have my quibbles about these bias-correction estimates (e.g., there are numerous ones out there and they don’t necessarily converge), but the point about robustness is fair…and a real problem. Part of my unease is the fact that “depletion” (a term I hate and think should be forgotten given that it implies a structure that is simply not there) is nothing more than short-term mental fatigue. And, call me crazy, but I think that fatigue IS real, robust, replicable, etc. So how can fatigue be real, but depletion not? Perhaps the sequential-task paradigm needs to be rethought? Perhaps better measures and manipulations needs to be used? Perhaps powerful repeated measures and within-subject designs need to be implemented? I’m really not sure. One thing I do know is that if something isn’t done about nailing down the phenomenon so that all researchers can reproduce it time and time again, depletion will go down in history as yet again another phlogiston.

    • Thanks for your remarks, Michael… on the first point, I’m skeptical that there is a limit of control at Time 2 after work at Time 1, following, as you indicate, McCullough. I do, in contrast, think that fatigue is genuine, as in the vigilance literature… that is, both performance reductions and the phenomenology of fatigue… so my view is that fatigue can be real but explained with something along the lines that we (and you) have proposed. But, generally, I think that you and I are in broad agreement. Having said that, a Google Scholar search for [“self-control” and “depletion” and “strength”] (I added the last search term to limit false positives) for just 2013 yields 1,700 results. Searches from prior years imply the idea is waxing, not waning.

  4. I really liked that Edge talk, because I think it raises larger issues about social psychology research as a field. One of them is the idea that it takes evidence several orders of magnitude greater to disprove something in print than to get the original article accepted. Brent Donnellan has an interesting post on this regarding a recently accepted paper debunking a correlation between loneliness and preference for hotter showers:

    I’m kind of excited about the fact that EHB now has a dedicated statistician reviewing articles; but another thing that your post raises is the issue of what standard of evidence an original article needs to pass in order to get published. If there are data that just seem screwy–like a glass of lemonade making you better at solving mazes or a majority of people self-reporting one shower a week or even that the laws of cause and effect have somehow been reversed (Bem)–shouldn’t we require a study with more than 50 people? Even if the p-value is .01, it doesn’t seem convincing to me that we would want to accept things that fly in the face of previous evidence and normal experience unless that case is really airtight. And p-values just don’t cut it as an ultimate arbiter of truth.

  5. The whole of these conceptions or preconceptions
    about mental evolution is education and can be summarized by the Christian
    scripture “In the beginning there was the word and the word became flesh”.
    Obviously this passage connotes intelligent design of sorts and it makes little
    difference if the transition from word to flesh was instantly or took a
    trillion years. It still connotes intelligent design as we have known the
    concept. I am a person of limited knowledge,
    at lease in philosophy but was trained in the physical sciences where cause and
    effect has been proven to be a valid reality, and accepted by most of us. Thus,
    I believe that evolution after creation and our present conditions are a
    product of our design projected through cause and effect–our learned proficiencies,
    and the documented evolutions, all caused by our god given ability to think –
    our progress as we have learned to define it.

    The real problem of our concepts
    of mental evolution lies with our attempts to combine mental evolution with
    what we are calling cognition especially the learning aspects of cognition
    where we have applied most of our energy to concepts involving the biological aspects
    such as race which does not exist, and other mental aspects thought to be related
    to our physical evolution, not realizing trillions of such studies have served
    to only distort our findings.

    The only concept of learning as
    related to cognition is that we learn by doing. By doing we evolve via
    reasoning (proven validity), language, memory (all aids to communications),
    problem solving and perception (realizing what we already know). The rest is
    our attempt to complicate things- a concept involving self – actualization, a
    concept we created with our god given ability to think and become the best we
    can be.

    Of all the animal creatures of the
    present earth, only we humans can transfer knowledge from one generation to
    another. Thus our mental evolution was born. SOLVE THE SELF ACTULATION AND

    To solve this problem, our most practical
    approach is to change the pattern of mental evolution. There are multiple approaches to how this can
    be done but education is the most

  6. Pingback: Two Tales of Marshmallows and their Implications for Free Will | Practical Ethics

Skip to toolbar