Book Review: Willpower

I.

Willpower: Rediscovering the Greatest Human Strength by Roy Baumeister and John Tierney attracted me with the following pitch: there are only two quantities in psychology that have been robustly linked to a broad range of important life outcomes. One is IQ and isn’t changeable. The other is willpower and is easily changeable. Therefore, study willlpower.

I expected the book to center around Baumeister’s groundbreaking experiments in “ego depletion” – where people who are forced to expend willpower on one task have less willpower left over for future tasks – and “ego repletion”, where people who have been depleted of willpower get it back after taking some glucose. I’d been left kind of confused by competing claims about those studies and I hoped this book would fill out my knowledge of them and settle my confusion.

Instead, it spent a couple of chapters mentioning their existence and praising them as revolutionary, and then got deep enough into Pop Science Self-Help Book Mode to be almost a self-parody. Chapter three is called “The To-Do List From God To Drew Carey” and illustrates why to-do lists are a great idea with anecdotes from Drew Carey’s life and the Bible. Chapter four is the same but with Eliot Spitzer. Chapter six is David Blaine, Chapter seven is H.M. Stanley, Chapter eight is Eric Clapton, and Chapter ten is Oprah. All of these people apparently have important lessons about willpower to teach us, of which the interesting ones are:

– Willpower is a limited resource that is depleted by use and restored by glucose

– Having to make too many small decisions in a day causes “decision fatigue”, leading you to be exhausted and make bad decisions.

– Using willpower a lot strengthens your willpower, allowing you to be more effective.

– People with more willpower do much better in life; for example, in the famous “marshmallow test”, children who were able to resist eating a marshmallow for a few minutes in order to get two marshmallows had better life outcomes twenty years later.

– Careful quantification of goals, setting precommitments, and publicizing your success or failure to other people helps you stick to resolutions. (If you are wondering exactly how to do this, this might be a good time to mention there’s a new ad on the sidebar for Beeminder, a company that manages this for you with some neat evidence-based tricks.)

– Religious people have more willpower than non-religious people for some reason. But you have to believe if you want to get this benefit – you can’t just hang out at church and go through the motions in order to reap the willpower gains.

– Chinese-Americans do better than white Americans on willpower tests from toddlerhood toward adulthood. This is the most likely reason Chinese people outperform whites in the real world, and in fact although average white and Chinese IQ are pretty similar, Chinese people can break into “elite” professions at a lower IQ threshold than whites because their increased self-control compensates. The book admits this may be partly genetic, but also attributes some of it to Chinese parents teaching their children discipline and setting hard goals, which they contrast with white parents who tell their kids to “have fun” and “have high self-esteem” and “be self-directed”. Obviously in order to believe this result you’d have to believe parenting styles can affect children, which this book takes on faith but which is on shaky ground.

– Amount of self-control does not affect body weight (!) or success at dieting (!!) very much.

Overall there were some interesting findings in here, even if I found the pop sci tone a little bit over-the-top after a while.

II.

But I was disappointed. The reason I got this book is that there’s a big debate going on over “willpower”, “ego depletion”, and their younger cousins “growth mindset” and “grit”. All of these are slightly different constructs, but they’re all measures of stick-to-it-ness, and all of their proponents make grand claims about how small interventions to make people have them will bring those people success at school, work, and life. On the other hand, there are a lot of other people who think this whole area is a load of bunk.

For example, Mischel’s marshmallow test started all this off by “proving” that children who were able to delay gratification longest had higher SAT scores, higher parent-rated competence, and better coping skills. But now the test is under fire as people question whether it just shows that people from good environments learn to be more trusting and so more likely to believe the researcher’s promise of extra rewards for delayed gratification. Other people ask if it just shows that kids who are smart enough to think of good strategies to distract themselves are also smart enough to get good SAT scores and all of the other positive correlates of high IQ. Still other people point out the very low sample size, the reverse correlation in other subgroups, and an apparent failure to replicate. Mischel fires back in an Atlantic article where he says of course he took these things into account, that the test was done on a homogenous upper-class population.

This book just says the marshmallow test proves willpower is important, and leaves it at that.

Or how about the idea that glucose is the limited resource that willpower depletes? Robert Kurzban very correctly points out that the metabolic math doesn’t come close to adding up – we know how much glucose things in the body use, and these short little willpower tasks aren’t really going to affect blood glucose levels at all. Also, turns out that if you rinse your mouth out with a tasty glucose solution, you get just the same amount of ego replenishment even though none of the glucose actually entered your body. And for that matter, how come I can’t get infinite willpower just by snacking while I work? How come M&Ms don’t work as a poor man’s Adderall?

Kurzban goes further and says he doesn’t believe in willpower as a limited resource at all. He notes that even when you’ve stopped studying because you’re too ego-depleted and exhausted to make yourself go on any further, if I offer you a million dollars to study another hour then you’ll do it. Guess that resource wasn’t so depleted after all. He proposes a different model of willpower, where it’s your brain’s way of nagging you about the opportunity cost of your actions – “you’ve been studying three whole hours, don’t you think you could use that lobe of your brain for something else now?” But this strikes me as ridiculous – my brain is very concerned that it has better things to do than study, but is perfectly happy with me playing Civilization IV: Fall From Heaven forever or simply lying in bed doing nothing?

Finally, Carol Dweck finds that willpower is only depletable if you think it is, which sounds like exactly the sort of thing Carol Dweck would find. If Carol Dweck ever became an oncologist, we would have to revise all the medical textbooks to say that people only get cancer if they think they will.

There is a meta-analysis of about a hundred studies said Baumeister was basically right about everything. On the other hand, Baumeister’s theory failed what sounds like a formal replication. So it’s complicated.

I really want to know what willpower is. It seems like one of the big challenges of my life; I never have enough willpower for everything I want to do, and I’d at least like to have a theory of what I’m up against. This book did not give me enough information to navigate the controversy. Instead, it totally denied there was any controversy and spent chapter after chapter on cute pieces of trivia interspersed with stories about Drew Carey and Oprah.

III.

So let me end this with some thoughts that a good explanation of willpower should take into account.

First, mental willpower seems a lot like physical willpower – by which I mean our ability to push through exhaustion to keep exercising. Kurzban complains that willpower can’t be a limited resource, because even after it’s all depleted you can still force yourself to keep going for a big enough reward. But the same is true of exercise. I’ll get exhausted and stop running after a certain number of miles, but if you offer me $1 million to run another one I’ll probably make it. And the studies showing that rinsing your mouth out with glucose gives you the same willpower boost as actually drinking is identical to the results of similar studies measuring exercise duration.

So it’s probably worth asking what exactly causes exercise fatigue. This is less firmly known than I expected, but it seems to be a combination of decreasing levels of inputs (especially glycogen stores in the muscles), buildup of toxic metabolic waste (including heat, lactic acid, etc), and cellular damage. Exercise long enough and your muscles need time to replenish their stores, clear away waste, and repair themselves.

Second, mental willpower seems to do something a lot like budgeting. Money, despite being pretty much the classic example of a limited resource, shares some of the features of willpower that Kurzban mocks. A few days ago a pipe burst, my house flooded, and I may have to spend several thousand dollars unflooding it. This means I will have no money for the next forever and then some, and if you ask me to meet you at a fancy restaurant I’ll probably refuse on financial grounds. But if my mother is on her deathbed and her final wish is to see me one last time, I will find the money to get an expensive flight to California. How’s that any different from me not having enough willpower to keep studying until you offer me a million dollars to do so?

Likewise, if tomorrow my boss offers me a $5000 bonus to be handed out next month, I’ll probably relax my budgetary constraints and start buying myself nice things even before I get the paycheck. How’s that any different from the body reacting to sugar when it’s in the mouth but not the bloodstream?

Finally, the question that I as a psychiatrist find most interesting – how come drugs can change willpower so dramatically? The guy who can’t concentrate on a project for more than five minutes straight will pull a whole week of all-nighters when he’s on Adderall or modafinil. Those certainly aren’t increasing blood glucose, so what’s up?

The model that makes the most sense to me is of a stupid default system running on short-term reinforcement learning, plus an evolutionarily novel (and therefore poorly implemented) executive system that can overrule the default. The executive system’s overrule isn’t a simple veto, but a constant action, the same way holding your hand high in the air for a long period requires constant action by your muscles. This effort is metabolically costly in the same way that using muscles is metabolically costly, and so your body runs a general-purpose budgeting function on it that convinces you to turn it off before it overheats. Given enough incentive, you can let it overheat, but then it’s going to be damaged and need to repair itself for a few days before you can use it effectively again. This seems to fit all the evidence except the drugs, which I interpret as acting on the default system so that you don’t need to bring in the executive planner. I freely admit this is sort of cheating.

Overall I recommend Willpower if you want a quick and fun survey of a bunch of loosely connected psychology topics, but not if you want a deep and balanced exploration into the literature of anything.