What Is Willpower?

Monday, March 21, 2016
By dreeves

A kid grabbing a marshmallow off a plate

Our previous post, “Ego Depletion Depletion,” generated a lot of discussion and I found I was contradicting myself on the question of what willpower is exactly.

First a recap, hopefully in plainer English, about what all the fuss is about.

A big finding in psychology is that “willpower is like a muscle”. For example, exerting willpower to not eat pie for breakfast will “deplete your ego” and make it harder to resist watching youtube videos instead of working in the afternoon. That spawned lots of advice about picking your willpower battles in the course of a day. The willpower-as-muscle analogy also led to advice about building up the strength of your willpower by exercising it.

All that advice was built on a lie. The evidence for it now appears to be bogus.

This is a massively big deal for the whole field of psychology because Ego Depletion was a very well-established result — dozens of studies all agreeing — and if it’s wrong then can we really believe any results from the psychology literature? This is psychology’s replication crisis and to the field’s credit they’ve taken it dead seriously, like by launching the Reproducibility Project.

Donuts vs dessert

“But willpower-as-muscle matches my experience!” said multiple people I’ve debated this with, like Scott Alexander. “I can resist donuts at work in the morning much more easily than I can resist dessert after a long, stressful day.”

Well let me ask you this, partially hypothetical people: is it possible you simply value dessert more after a hard day than you do donuts in the morning? That’s what Occam’s razor might suggest. No difference in willpower, just a difference in circumstances/preferences. Maybe it makes sense to reward yourself after hard work, or to relax or indulge when you’re wiped out.

(Mark Forster offered another simple explanation for the donuts-vs-dessert example: you have more stable routines in the morning.)

But I’m not actually taking the hard line that willpower is constant. Maybe hyperbolic discounting gets noticeably more or less severe depending on your level of mental fatigue.

“Willpower = overcoming akrasia by pure introspection”

Let’s define willpower as overcoming akrasia by pure introspection. If we agree on that definition then I have two different claims:

  1. Willpower isn’t any more depletable than, say, ability to do mental arithmetic.
  2. If you align your short and long term preferences then no akrasia and no need for willpower.

I’m not that wedded to #1. I think the donuts/dessert example is more elegantly explained by a simple reassessment of priorities in light of a hard day. But disagreements on #1 may be hair-splitting.

Beeminder’s big claim is that it’s possible to route around willpower altogether.

No need for that hypothesis

Here’s what I mean when I say there’s no such thing as willpower, despite having just defined it. Paraphrasing Laplace, I can explain all behavior simply in terms of responding to incentives. You want this whole pie in your body right now, and also you want to be two sizes smaller by next summer. Conflicting preferences are normally no big deal. You just, y’know, weigh them, make your tradeoffs, and reach a decision. But when the preferences apply at different timescales (pie now, thinner later) humans suffer from a massive irrationality which philosophers call akrasia and economists call dynamic inconsistency and normal people call … being stupidly short-sighted, or in the case of time management: procrastination.

Commitment devices are a way to change your own incentives so that willpower is a non-issue. They make your short-term and long-term incentives line up. There are many less drastic things you can do as well.

Does it take willpower to set up a Beeminder goal?

That may be Beeminder’s Achilles heel. I think it’s similar to starting up any new productivity system, or making a new year’s resolution or whatnot. We often have these little bursts of motivation and the hard part is the follow through. So if you can seize on your next motivation burst to get a Beeminder goal set up then — if Beeminder works as advertised (and please, please talk to us if you feel it doesn’t!) — the follow-through will be in the bag.

Tags: , , , , , , ,

  • http://giannopoulos.net/ Markos Giannopoulos

    Just a short note.
    “All that advice was based on a lie”.
    I haven’t followed the subject a lot, but the word “lie” is a strong one. In my mind it implies intent to deceit, which a big accusation for a scientist.

  • http://beeminder.com Daniel Reeves

    True! I didn’t mean it that way. “Based on a falsehood” doesn’t have the same ring to it, but this correction is well noted. I’ll adjust it somehow.

  • http://beeminder.com Daniel Reeves

    Googling around more I think it’s idiomatic enough that it’s not in danger of being misinterpreted so I think I’ll take the disclaimer in the text back out and just reemphasize here that I’m definitely not accusing Baumeister or other psychology researchers of intentionally deceiving.

  • https://entirelyuseless.wordpress.com/ entirelyuseless

    “Willpower isn’t any more depletable than, say, ability to do mental arithmetic.”

    I think it is obvious that both willpower and the ability to do mental arithmetic are depletable, and in the same way, and for the same reasons. You get tired after doing arithmetic for a while, and you get worn out by using willpower.

    I don’t need studies to establish that; it is a fact of experience.

  • Dave Warnock

    “Here’s what I mean when I say there’s no such thing as willpower, despite having just defined it. Paraphrasing Laplace, I can explain all behavior simply in terms of responding to incentives. ”

    I found this helpful. However, what term do you use for the way that people seem to vary widely in their ability to handle conflicting preferences in favour of long term goals? In other words it seems obvious that while “humans suffer from a massive irrationality which philosophers call akrasia and economists call dynamic inconsistency” it varies tremendously between people.

    Some, for example Olympic athletes show an amazing ability to keep a goal that is years ahead guiding their daily choices. When someone is suffering from depression they might find long term goals very hard to prioritise.

    I found the “willpower is like a muscle” argument helpful for this because you can imagine becoming fitter through exercise and see that different people may have different sized muscles.

    So if we are rejecting “willpower is like a muscle” then how are the differences between people described?

  • http://beeminder.com Daniel Reeves

    I have 2 answers!

    1. People discount the future differently, ie, they differ in their ability to see the big picture and not focus on the metaphorical pie under their nose. Which sounds a lot like a difference in willpower.

    But more importantly…

    2. People set up their incentives differently. Using Beeminder is the obvious one (for us) but people do this in many ways. Like the olympic athlete has a coach and teammates and structures their life in such a way that they don’t actually have to think years ahead to get themselves to the gym/track/rink/road every day. They have people counting on them to do that Right Now.

    Immediate incentives are inordinately powerful. So find ways to make your immediate incentives match your long-term incentives and willpower needn’t come into play at all!

    (I hope I didn’t skirt the question too much. As a handwavy analogy I don’t even really object to “willpower is like a muscle”.)

  • Andrew Lu

    This turned out to be a long post, so here’s a short summary:

    1. moral licensing may be another explanation for the donuts-dessert phenomenon (which I think is over-determined anyway)

    2. purposefully “working out” your willpower muscle never made much sense to me; they’re already overworked as it is!

    3. the willpower-as-muscle analogy didn’t come out of nowhere; there is an
    element of straining against some opposing resistance that the muscle
    analogy captures; in other words, willpower is not simply introspection
    and weighing preferences, but involves an essential element of being in
    opposition to something

    4. in addition to the commitment device aspect you mention, Beeminder also reduces the need for willpower by providing additional short-term incentives and thus increasing your motivation to do the task. Beeminder provides little opportunities throughout my day for an immediate dopamine boost when I enter a positive data point and see my safety buffer increase. It’s a bit like getting to reward yourself with a piece of chocolate every time you do something good, but without the hit to your waistline and without the alternative of simply eating the chocolate regardless (since the whole reason the positive data points are rewarding is because they’re a sign of progress and accomplishment and lessening the danger of derailing, and false datapoints would make you feel bad for lying and not actually making progress).

    —-longer version below with more details——-

    First, I’d like to add an additional explanation for the donuts-dessert phenomenon: moral licensing! Of course, I have a suspicion that this is over-determined, so no single explanation is correct.

    Similarly, when I accepted the “willpower as muscle” analogy, it was one of several reasons I tackle more cognitively and psychologically demanding tasks earlier in the day, and I will continue to do so despite these findings.

    Beyond that, I always thought the advice
    given in some self-help books to purposefully “work out” your willpower
    muscle was trite and seriously misguided. I mean, most people are
    already sufficiently working out their willpower muscle in their daily
    lives! To extend the analogy, it’d be like we’re all Olympic-level
    powerlifters who lift so much weight so often every day that we’re
    having trouble going for another set! The problem is not that our
    muscles are atrophied, but that they’re overworked! The solution is obviously not to do some more “easy” exercises, but to look for ways to avoid using willpower altogether! (Enter Beeminder).

    I totally agree that Beeminder is so wonderful because it mostly avoids the kind of white-knuckle willpower that we commonly picture as “willpower”. But I disagree that willpower does not exist at all and that all there is different incentives. For example, when I’m sprinting and push myself to hold my sprint for just a few more seconds, it sure feels like I’m exercising willpower (something like “conscious cognitive control over behavior in the face of countervailing temptations/impulses/desires”, which is basically the textbook definition of willpower), not weighing short-term and long-term preferences.

    It’s also not mere “introspection” to properly feel your true preferences, which is more neutral, not fighting against something. Willpower is more like using a muscle to push or pull something heavy (you can see why the muscle analogy seemed so natural and correct), or focusing closely on some complicated task involving fine motor skills. Another closely related feeling is when you’re angry or upset and trying to emotionally self-regulate. You’re fighting against an opposing inclination.

    This is all academic though. At the end of the day, Beeminder works! :)

    I wouldn’t go so far as to say it completely routes through willpower, but rather that it reduces the level of willpower needed by increasing your short-term motivation (aligning them more with long-term goals and preferences). In addition to the commitment device aspect, it also creates a little opportunity for an immediate dopamine boost when you enter a positive data point and see your safety buffer increase. It’s a bit like getting to eat a piece of chocolate every time you do something good, but without the hit to your waistline and without the temptation to just eat the chocolate regardless (it’d be absurd to keep entering positive datapoints if you’re not actually doing them because the whole reason the data points are rewarding is because they’re a sign that you’re making progress towards your goal and also lessening the danger of derailing, and entering false datapoints would make you feel bad for lying and not actually making progress).