Baumeister has a book coming out on willpower. There's a NYTimes article about his work, summarized and discussed on LessWrong.

I've previously discussed much of this. But some of it is new research.

It's still possible that the only thing being depleted is glucose available to the brain (which is most readily available from stored liver glycogen, or food). But there may be other chemical resources involved. In any case, weighing and making choices taxes the brain; I'm not sure how much more or less than e.g. solving puzzles or reading and thinking critically.

Quoting from the LessWrong summary:

You spend the most willpower when you have to make AND implement your decisions:

which phase of the decision-making process was most fatiguing? To find out, Kathleen Vohs, a former colleague of Baumeister’s now at the University of Minnesota, performed an experiment using the self-service Web site of Dell Computers. One group in the experiment carefully studied the advantages and disadvantages of various features available for a computer — the type of screen, the size of the hard drive, etc. — without actually making a final decision on which ones to choose. A second group was given a list of predetermined specifications and told to configure a computer by going through the laborious, step-by-step process of locating the specified features among the arrays of options and then clicking on the right ones. The purpose of this was to duplicate everything that happens in the postdecisional phase, when the choice is implemented. The third group had to figure out for themselves which features they wanted on their computers and go through the process of choosing them; they didn’t simply ponder options (like the first group) or implement others’ choices (like the second group). They had to cast the die, and that turned out to be the most fatiguing task of all. When self-control was measured, they were the one who were most depleted, by far.

 

Willpower depletion makes you reluctant to make trade-offs:

Once you’re mentally depleted, you become reluctant to make trade-offs, which involve a particularly advanced and taxing form of decision making. In the rest of the animal kingdom, there aren’t a lot of protracted negotiations between predators and prey. To compromise is a complex human ability and therefore one of the first to decline when willpower is depleted. You become what researchers call a cognitive miser, hoarding your energy. If you’re shopping, you’re liable to look at only one dimension, like price: just give me the cheapest. Or you indulge yourself by looking at quality: I want the very best (an especially easy strategy if someone else is paying). Decision fatigue leaves you vulnerable to marketers who know how to time their sales, as Jonathan Levav, the Stanford professor, demonstrated in experiments involving tailored suits and new cars.

 

Willpower depletion makes you more likely to take the path of least resistance:

As they started picking features, customers would carefully weigh the choices, but as decision fatigue set in, they would start settling for whatever the default option was. And the more tough choices they encountered early in the process — like going through those 56 colors to choose the precise shade of gray or brown — the quicker people became fatigued and settled for the path of least resistance by taking the default option. By manipulating the order of the car buyers’ choices, the researchers found that the customers would end up settling for different kinds of options, and the average difference totaled more than 1,500 euros per car (about $2,000 at the time).

 

Testing willpower depletion in rural Indian villages:

Most of us in America won’t spend a lot of time agonizing over whether we can afford to buy soap, but it can be a depleting choice in rural India. Dean Spears, an economist at Princeton, offered people in 20 villages in Rajasthan in northwestern India the chance to buy a couple of bars of brand-name soap for the equivalent of less than 20 cents. It was a steep discount off the regular price, yet even that sum was a strain for the people in the 10 poorest villages. Whether or not they bought the soap, the act of making the decision left them with less willpower, as measured afterward in a test of how long they could squeeze a hand grip. In the slightly more affluent villages, people’s willpower wasn’t affected significantly.

 

Decision fatigue can be a factor in trapping people in poverty:

Spears and other researchers argue that this sort of decision fatigue is a major — and hitherto ignored — factor in trapping people in poverty. Because their financial situation forces them to make so many trade-offs, they have less willpower to devote to school, work and other activities that might get them into the middle class.

 

Glucose restores willpower in humans and dogs:

To establish cause and effect, researchers at Baumeister’s lab tried refueling the brain in a series of experiments involving lemonade mixed either with sugar or with a diet sweetener. The sugary lemonade provided a burst of glucose, the effects of which could be observed right away in the lab; the sugarless variety tasted quite similar without providing the same burst of glucose. Again and again, the sugar restored willpower, but the artificial sweetener had no effect. The glucose would at least mitigate the ego depletion and sometimes completely reverse it. The restored willpower improved people’s self-control as well as the quality of their decisions: they resisted irrational bias when making choices, and when asked to make financial decisions, they were more likely to choose the better long-term strategy instead of going for a quick payoff. The ego-depletion effect was even demonstrated with dogs in two studies by Holly Miller and Nathan DeWall at the University of Kentucky. After obeying sit and stay commands for 10 minutes, the dogs performed worse on self-control tests and were also more likely to make the dangerous decision to challenge another dog’s turf. But a dose of glucose restored their willpower.

 

Ego depletion causes activity to rise in some parts of the brain and to decline in others:

The results of the experiment were announced in January, during Heatherton’s speech accepting the leadership of the Society for Personality and Social Psychology, the world’s largest group of social psychologists. In his presidential address at the annual meeting in San Antonio, Heatherton reported that administering glucose completely reversed the brain changes wrought by depletion — a finding, he said, that thoroughly surprised him.Heatherton’s results did much more than provide additional confirmation that glucose is a vital part of willpower; they helped solve the puzzle over how glucose could work without global changes in the brain’s total energy use. Apparently ego depletion causes activity to rise in some parts of the brain and to decline in others. Your brain does not stop working when glucose is low. It stops doing some things and starts doing others. It responds more strongly to immediate rewards and pays less attention to long-term prospects.

 

Good decision makers structure their lives so as to conserve willpower:

“Good decision making is not a trait of the person, in the sense that it’s always there,” Baumeister says. “It’s a state that fluctuates.” His studies show that people with the best self-control are the ones who structure their lives so as to conserve willpower. They don’t schedule endless back-to-back meetings. They avoid temptations like all-you-can-eat buffets, and they establish habits that eliminate the mental effort of making choices. Instead of deciding every morning whether or not to force themselves to exercise, they set up regular appointments to work out with a friend. Instead of counting on willpower to remain robust all day, they conserve it so that it’s available for emergencies and important decisions.

“Even the wisest people won’t make good choices when they’re not rested and their glucose is low,” Baumeister points out.

They had food-deprived subjects sit at a table with two types of food on it: cookies and chocolates; and radishes. Some of the subjects were instructed to eat radishes and resist the sweets, and afterwards all were put to work on unsolvable geometric puzzles. Resisting the sweets, independent of mood, made participants give up more than twice as quickly on the geometric puzzles. Resisting temptation, the researchers found, seemed to have “produced a ‘psychic cost.’”

Based on the reporter's description, I expected that the chocolate eaters merely tried harder because of the caffeine and sugar (both of which are known to help motivate or support mental labor).

But the actual  study was careful of this (see pg 1254, Method, and Table 1); a no-food no-temptation control group performed the same as the chocolate-indulging group. It was only the radish-eating chocolate-resisters (giving up 40% sooner due to the psychic pain of resisting a fresh-baked-chocolate-chip-cookie temptation).

It looks like people studying people's decline in vigilance-required tasks over time need to take care not to suggest to their subjects the idea that they will probably fatigue.  The theory is that tasks that specifically require "self control" can lead to specific fatigue in other "self control" tasks, as distinct from general mental fatigue, although it's been found that glucose availability to the brain explains most of this.

A new study, which is much more careful than past ones, gives a pretty strong idea that people's expectations for how they'll perform while willpower-taxed are the determining factor (at least for artifical, low-motivation psych-study tasks), and further that when these expectations are manipulated (by push polling), that this obliterates the effects typically reported in the ego-depletion literature. Because of the push polling affecting performance, you can't just say that it's 'the person's idiosyncratic "availability of willpower" after a demanding task that shapes idiosyncratic beliefs about willpower'.

I reproduce here my comments from this LessWrong discussion:

What's demonstrated: if you prime an excuse for doing poorly, you will do poorly. I think there's already some similar research (different types of excuses, though). They also show that self-reported exhaustion (not just "ego depleting" tasks) leads to a difference in performance that goes in exactly the direction that the subjects are primed to believe (either being reminded of an existing belief, or being tricked into holding it with biased questions).

It surprises me that, of the people who don't claim to expect to flag when fatigued, those who report being exhausted by the depletion task actually make less errors than those who don't. Unless this is just due to warming up their inhibition/vigilance (both the initial and final tests require it) while, it suggests that positive expectations can boost performance, not just that available excuses can harm it.

I like that they demonstrated that errors on IQ problems tracks errors on mundane rule-following, vigilance type tasks, but it's amusing to me that people who believe they'll do worse when fatigued, actually test as smarter (less IQ test errors) when fresh, whereas those primed to believe they won't effectively fatigue improve slightly, but are still worse than the "limited resource" believers initial performance. This effect is still there, but probably not significant, for the simple but tiresome "willpower" testing (Stroop) task. I assume the "limited"-believers are more engaged by an IQ-proving question, either for signaling or entertainment, compared to the boring Stroop task. Disclaimer: these differences, from figures in pg 5 of the 

paper. aren't strongly significant (N ~= 50), so maybe I shouldn't conclude anything (the authors don't pin anything on them).

It seems reasonable to me that push polling about someone's future behavior will lead them to act consistently with the signal they just sent in the poll - like in Cialdini's Influence, where people are polled on whether they like to go to opera, or give charitably, by some attractive person they want to impress, and then after affirming are ambushed with a sales pitch (they thought it was an innocent poll but are trapped by their answers). So it seems reasonable to assume that those who were push-polled into believing they will become either sloppier, or more accurate, with fatigue, would act consonantly.

But I don't think this objection is likely the whole story. The simplest explanation is that people's stated expectations of their performance do shape their performance - the power of positive thinking, and obviously, negative. (possibly unvoiced/persistent expectations as well as explicitly declared, although of course it's nearly impossible to measure such things surreptitiously).

From the paper in the previous post:

When people expect to have to exert self-control later, they will curtail current performance more severely than if no such demands are anticipated.

Cool.

Consistent with the conservation hypothesis, people can exert self-control despite ego depletion if the stakes are high enough. Offering cash incentives or other motives for good performance counteracts the effects of ego depletion.

I also expressed skepticism that experimental subjects were really motivated.

Inducing a state of positive emotion such as humor seems to have that effect [moderating or counteracting the effects of ego depletion] (Tice, Baumeister, Shmueli, & Muraven, 2007). Having implementation intentions — formulating ‘‘if–then’’ statements about how to behave in a situation prior to entering it — seems to be effective most likely because such intentions operate as behavioral plans and guidelines that reduce the need for executive control (Webb & Sheeran, 2003). To be sure, none of these procedures clearly counteracts the depleted state in the sense of replenishing the depleted resource. Rather, they may all operate by inducing the person to expend more of the depleted resource. In contrast, there is some reason to think that replenishing glucose in the bloodstream does actually rectify the depletion by restoring the depleted resource (Gailliot et al., 2007).

Having a comprehensive plan in place means you don't have to decide anything. Does making such a plan cost as much as making a decision in real time?

Logical reasoning, extrapolation, and other controlled processes depend on control by the self, and performance on these tasks dips sharply when people are depleted (Schmeichel, Vohs, & Baumeister, 2003).

I wonder what types of items generally used to measure IQ ("G") don't ego-deplete.

Recent studies indicate that the same energy is used for effortful decision making, as well as for active rather than passive responses (e.g., Vohs et al., 2007).  These seem to correspond to what laypersons understand as ‘‘free will,’’ namely the ability to override impulses, behave morally, show initiative, and behave according to rational choices (Baumeister, in press).

If I remember this research, contemplating the decision (e.g. imagining consequences) isn't depleting; making it is.

Success at building self-control through exercises has been inconsistent.

Agreed.

Identifying the biological substrates of self-control depletion (and replenishment) would be another helpful direction for further work.

I mentioned this earlier.

Response to my post on ego depletion from Unnamed at LessWrong (who also pointed me to the research in that post):

people come into the lab, they do one task that drains their willpower, then they get some intervention that might restore their willpower, then they do another task that requires willpower. This review by Baumeister, Vohs, and Tice (pdf) lists a few that have worked and gives citations:

 

• Humor and laughter
• Other positive emotions
• Cash incentives
• Implementation intentions (‘‘if ... then’’ plans)
• Social goals (e.g., wanting to help people; wanting to be a good relationship partner)

I'll read the pdf later, but for now titrating a sweet food or drink seems like a cool willpower-boosting trick (I find that when I play beach volleyball for hours, I do better if I can have some chocolate, soda, or banana between games, but that's probably helping with the more obvious muscle glycogen depletion).

It's well known that exercise of self-control drains a "battery" - studies have shown that a sequence of willpower-taxing tasks become effectively more challenging, and that the obvious alternative explanations don't apply - the depleted resource is specific to self-regulation. One alternative I think should still be examined: repeated demands for self-control which aren't commensurately rewarded, should rationally lead to apathy. But at least one study had a real consequence for lack of willpower: subjects overpaid with their own money, so it wasn't the case that the tasks and their rewards were merely artificial tokens.

Baumeister et. al survey the evidence that willpower can also be generally improved by practice. It's pretty weak, but if you're hoping to self-improve, you can find some reasons to hope.

Best results were obtained among participants who were assigned to improve their posture whenever they thought of it. Other participants kept track of what they ate, and these showed some improvement. The group that was assigned to try to improve their mood state whenever they felt bad did not show any benefit from this exercise.

But maybe improving your posture or diet builds your willpower reservoir directly (suppose you were to do this due to pleasure in playing a sport, or having someone you love prepare better food for you, or you for them ...). Still, good news either way (if it's easy to improve posture and diet, and it leads to more willpower, fantastic).

(Oaten & Cheng, 2004a) enrolled participants in physical exercise programs for 2 months. The exercise programs included weightlifting, resistance training, and aerobics, and each participant received a program designed by a gym staff member specifically suited for him or her. A staggered waiting list control group design was used so that all participants eventually received the exercise program. The hypothesis was that adhering to an exercise program requires self-regulation, and so 2 months of such regular effort would improve the capacity for self-regulation in general.

This substantially reduced the ego depletion (where one challenging self-regulation task decreases performance on another such task immediately following it). Again, I'd think it may be physical exercise that improves the willpower battery's capacity or usage rate, not the use of willpower in order to perform the exercise (substitute some more fun exercise program, like sex, or a sport the subject enjoys, and see what happens).

people who performed the exercise routines became more successful at reducing their cigarette smoking, alcohol use, and caffeine consumption. They ate less junk food and ate more healthy food. They reported improvements in emotional control and a reduction in impulsive spending. They reported studying more and watching less television. Even some domestic habits (e.g., washing dishes instead of leaving them in the sink) also improved across the exercise program

I would explain this with consistency and identity effects, not willpower. But this is still useful. Since this was a prospective study with (I assume) no dropouts, we can reasonably guess that the exercise in fact caused this.

only some of these behaviors could plausibly be directly linked to the exercise itself

If the above survey-based claims of generalized self-control improvement (which were not directly measured). It's obvious that you can't separate survey results about whether the respondent has positive qualities from changes in their tendency to signal positive things about themselves (if they exercise, they're physically more attractive and perhaps more confident in credibly claiming virtue, and also honestly view themselves as more virtuous).

Participants signed up for a 4-month program on financial monitoring. Each participant met with the experimenter, individually, at the start, and together they reviewed the participant’s bills and spending habits and devised a personal money management plan. [Through use of diaries and logs], most participants improved substantially in regulating their use of money.

Again, the ego-depletion effect decreased in magnitude after just 1 month of following the budgeting and record-keeping regimen. And, again, survey responses indicated that people were improving their diet, reducing drug use, cleaning their house, keeping promises, and studying. This time they measured stress (PSS), distress (GHQ), and self-efficacy(GSES) also, and there was no significant difference. Identity/consistency effects can't be ruled out, but this is finally doing a pretty good job at indicating that regularly practicing control over some part of your life has a generalized self-control training effect.

Studies showing a reduction in ego-depletion from some non-exercise, non-diet, "responsible" self-monitoring intervention do not necessarily make people feel significantly better (students following a structured studying program did feel less stress, but the budgeting exercise left people feeling about the same), and definitely don't significantly improve GSES (self-efficacy). So the improvement is more focused, and it may be fair to call it an improvement in general self-regulation. One reason for caution: all these studies are by the same folks (Oaten & Cheng) who seem to really like the idea that willpower can be trained.

Gailliot et al 2007 test ego-depletion by presuming (evidence from Gordijn et al 2004) that overriding prejudice uses the same willpower resource; the initial task is to talk about a typical day of a particular hypothetical obese or homosexual person, while "avoiding stereotypes" (whatever that means - basically, second guess and filter your creativity when it's too quintessential, which should be tiring); the follow up task which should suffer from ego-depletion is anagram deciphering performance (anagrams are thought to require executive function, because trying all possibilities is taxing?). People who showed low in motivation to avoid prejudice did worst on anagrams.

This doesn't tell me much because both activities are globally tiring. Baumeister et al assumes that the thing that's tired out is exactly "willpower", which seems questionable to me, unless they also show, as was done with other ego-depletion measures, that performance on other metrics (e.g. arithmetic) wasn't also degraded. Also, they seem to presume that it's not the case that the people who tend to be good at inventing non-stereotypically are simply smarter, and thus better at anagrams. They attribute entirely to ego depletion the difference in performance between the better and worse stereotype-suppressors (the worse ones must be "trying harder"). Very sketchy.

The Gailliot willpower-practice exercise was to use the off-hand (non-dominant left/right) for everyday activities: brushing teeth, stirring drinks, computer mousing, carrying, eating, opening doors, etc. Also explored was verbal self regulation: avoid slang (e.g. "yes" not "yeah"), don't start sentences with "I", and don't curse. Both exercises (followed for 2 weeks) gave an improvement in anagram performance compared to the no-exercise group, but only for those who had trouble with the stereotype suppression exercise. Perhaps this gives some evidence that the anagram task really does measure ego depletion. Perhaps this means that the only people who benefit from willpower training are those who have an unusually low amount of practice exercising it; or it may only mean that the draining preliminary task (making up a story) is just really easy for some people, so their willpower might have improved from the exercises, but the test wasn't challenging enough to show any difference.

My largest objection to this area of research is that there's no proposed physical mechanism or detection of these purported changes. I don't need that kind of evidence to believe something about human behavior, but it's extremely easy to draw the wrong conclusions from purely behavioral studies.

My next objection is that the control groups (who are assigned NO regular exercises) are poorly designed given that the followup ego-depletion test is a repeat of the initial one; those who are assigned exercises are often remembering the ego-depletion test; the control group is never reminded at all. So the control group isn't controlling for task-specific learning/priming effects (you can definitely learn or refresh memories by visualizing/reviewing, not only by doing).

Summary: I now believe (70% likely) that people who are easily ego-depleted can train to become significantly more normal, if they're able to institute some self-regulation habits over a long period of time (2 weeks or more). It's unknown (50% likely) whether people who already have regular self-regulation exercise can benefit from more volume or intensity in that area. Also, (60% likely) improvements in exercise and diet (I would guess sleep also) may give better ego-depletion capacity, supplementary the fact that applying them involves exercise of self-regulation.

I have no idea if habits that involve self-monitoring and adjustment, that are easy to follow because they're immediately rewarding or self-image-enhancing, are just as good for willpower training as things that are subjectively grueling and painful. I definitely won't embark on any heroic efforts in that direction without stronger evidence; what I will do is ensure that I'm regularly working on changing something in my life (whatever seems most productive at the time).

There's also some evidence that willpower can be buttressed with sugar (the brain needs glucose):

Acts of self-control deplete relatively large amounts of glucose. Self-control failures are more likely when glucose is low or cannot be mobilized effectively to the brain (i.e., when insulin is low or insensitive). Restoring glucose to a sufficient level typically improves self-control. Numerous self-control behaviors fit this pattern, including controlling attention, regulating emotions, quitting smoking, coping with stress, resisting impulsivity, and refraining from criminal and aggressive behavior. Alcohol reduces glucose throughout the brain and body and likewise impairs many forms of self-control. Furthermore, self-control failure is most likely during times of the day when glucose is used least effectively.