Link: "Don't: The secret of self-control", Jonah Lehrer. The New Yorker. May 18, 2009.

Article Summary

Walter Mischel, a psychologist at Columbia University, has spent a long time studying what correlates with failing or passing a test intended to measure a preschooler's ability to delay gratification. The original experiment, involving a marshmallow and the promise of another if the first one remained uneaten for fifteen minutes, took place at Bing Nursery School in the "late 1960's". Mischel found several correlates, none of them really surprising. He discovered a few methods that allowed children to learn better delay gratification, but it is unclear if the learning the tricks changed any of the correlations. He and the research tradition he started are now waiting for fMRI studies, because that's what the discriminating 21st century psychologist does.

Best line: "'I know I shouldn't like them,' she says. 'But they're just so delicious!'"

Article Criticism

The New Yorker is not exactly meant as a science journal for scientists, so it leads and weaves a couple human interest stories in with the boring science parts. We hear from a 'high delayer' (someone who did well at the test) and her 'low delayer' brother. This sort of opening seems terribly common in popular science writing, and it's so infuriating because it primes the reader with the conclusions of the science before the science is even discussed. The high delayer goes on to become a social psychologist; the low delayer goes to Hollywood and is now working on becoming a producer. So already we see anecdotal evidence that high delayers are successful, low delayers are unsuccessful, and that there's not a genetic component (because brothers and sisters share so much genetic material — right?).

At this point the author gives the impetus for Mischel's study as conversations he had with his daughters after they left Bing. Of course, that's not the whole story; that just explains the longitudinal part. According to Mischel's bio halfway through the article, he's been doing experiments like the marshmallow test every since his Ph. D. The article cites a chocolate bar test he did in Trinidad in 1955 that was intended to examine the racial stereotypes of Africans and East Indians.

The only technique mentioned that helped increase children's ability to delay gratification was "...a simple set of mental tricks — such as pretending the candy is only a picture, surrounded by an imaginary frame ...." However, buried on the next-to-last page of the article, we find that " remains unclear if these new skills persist over the long term."

Background Literature

I dug up the delay test article ("", W Mischel, Y Shoda, and MI Rodriguez (26 May 1989) Science 244 (4907), 933"), which those without subscription can find here, and found a few more practical applications. Mischel et. al. found, for instance, that "attention to the rewards consistently and substantially decreased delay time" (934). This suggests that if you would like to delay a reward, it's best not to see the reward directly. However, another experiment showed that images of the reward made delaying gratification easier. By far, though, the best technique discovered by the study was self-distraction. Even when the reward was present in the room with them, children who sung songs to themselves or played with toys did much better at the delay test than those who didn't.

In conclusion, I found neither the article nor the study itself practically helpful. I was annoyed by the human-interest propaganda that littered the article and the narrowness of the original study. As far as beating akrasia goes, the best advice I can give is for improving short-term gratification delay, practice self-distraction and avoid tempting yourself with the actual reward. If you think a visual reminder of the reward would help, settle for a picture.

New Comment
16 comments, sorted by Click to highlight new comments since: Today at 6:45 AM

Interesting in a similar way is the article "How To Make Your Own Luck".

We asked subjects to flip through a news-paper that had photographs in it. All they had to do was count the number of photographs. That's it. Luck wasn't on their minds, just some silly task. They'd go through, and after about three pages, there'd be a massive half-page advert saying, STOP COUNTING. THERE ARE 43 PHOTOGRAPHS IN THIS NEWSPAPER. It was next to a photo, so we knew they were looking at that area. A few pages later, there was another massive advert -- I mean, we're talking big -- that said, STOP COUNTING. TELL THE EXPERIMENTER YOU'VE SEEN THIS AND WIN 150 POUNDS [about $235].

For the most part, the unlucky would just flip past these things. Lucky people would flip through and laugh and say, "There are 43 photos. That's what it says. Do you want me to bother counting?" We'd say, "Yeah, carry on." They'd flip some more and say, "Do I get my 150 pounds?" Most of the unlucky people didn't notice.

A good example of verbal manipulation - instead of lucky and unlucky they could just as easily be called flaky and task-oriented.

Luck as discussed in this article sounds related to Openness to Experience from the Big Five.

If you're going to have an "in conclusion" at the end of your post, it would be nice if the post had a presented thesis that the rest of the material supports or at least is involved in analyzing.

I agree - I really liked the way this article was formatted, but perhaps it would make more sense if the "in conclusion" paragraph was part of the "criticism". Then you can just have a thesis for that section, with a relatively neutral summary and background literature.

On the other hand, perhaps it was supposed to be the lifehacking takeaway - in which case maybe another header was in order, and it should be opened with something other than "in conclusion".



avoid tempting yourself with the actual reward

This is a bit confused. When the "actual reward" is the result you seek by delaying, "tempting" yourself with it is the right idea. It's unrelated rewards that our brains have trouble with, or rewards for inaction.

If the kids had to do something long and complicated to get their two marshmallows, it'd be no big deal. The problem is hard because they have to refrain from going after a reachable goal to get another, more abstract goal... and there's nothing else for them to DO. If the instructions said they had to finish assembling a toy before the experimenter got back, in order to get two marshmallows, I predict a much higher rate of success would result.

Basically, the kids would choose two marshmallows over one, before their thoughts of the immediately-available marshmallow reached monoidealistic focus, and they would then quickly become monoideal on the activity of assembling the toy.

Even if they finished before the experimenter came back, their investment of time and effort would now weigh against giving in (especially if the instructions were to make a choice up-front).

Anyway, from a practical perspective, you're better off using intrinsic than extrinsic rewards for your motivation. Extrinsic rewards dampen intrinsic motivation, and don't distract you from the goal. Thinking about the inherent rewards of your tasks being completed (assuming these thoughts reach monoideal status) is motivating enough that you will actually do them "against your will" (or at least without any need for conscious will).

Once we resolve our own confusion and understand that the reward is the second marshmallow and not the first, we can ask if subjects have that confusion and how it affects the result. That is, in this case, focusing on the reward, the second marshmallow, which the subject associates with the first marshmallow because they are two instances of the same thing, causes the subject to focus on the first marshmallow (not the actual reward), which is counterproductive.

I wonder how the results would change, if the reward for not eating the marshmallow for 15 minutes were instead a cookie, so there would be less association between the reward and the treat the subject should avoid eating. Though this has the unfortunate effect of complicating the protocol of the subject choosing the treat they value, which seems to work well in the original experiment.

While this may not be practically helpful since one can do little about one's IQ, some might find interesting this recent study showing correlation between cognitive skills and patience, as well as other economically valuable preferences.

If this is safe, the patient marshmallow kids' 210-point SAT advantage is unsurprising (if less than revealing in terms of causation).

(A quick search didn't turn this up on LW or OB, but excuse me if it's a repeat.)

From the abstract:

Individuals with better CS are more patient, in both short- and long-run. Better CS are also associated with a greater willingness to take calculated risks. Second, CS predict social awareness and choices in a sequential Prisoner's Dilemma game. Subjects with better CS more accurately forecast others' behavior and differentiate their behavior as a second mover more strongly depending on the first-mover's choice. Third, CS, and in particular, the ability to plan, strongly predict perseverance on the job in a setting with a substantial financial penalty for early exit."

Ungated pdf:

"While this may not be practically helpful since one can do little about one's IQ,"

I think you can change one's IQ. I have believed that ever since I read Dweck.

I think you can change one's IQ.

I recently stumbled upon a paper published last year that suggests that fluid intelligence can be trained and that the training effect is dosage-dependent:

Improving fluid intelligence with training on working memory Jaeggi et al. PNAS. 2008.

Fluid intelligence (Gf ) refers to the ability to reason and to solvenew problems independently of previously acquired knowledge. Gf is critical for a wide variety of cognitive tasks, and it isconsidered one of the most important factors in learning. More- over, Gf is closely related to professional and educational success, especially in complex and demanding environments. Although performance on tests of Gf can be improved through direct practice on the tests themselves, there is no evidence that training on any other regimen yields increased Gf in adults. Furthermore, there is a long history of research into cognitive training showing that, although performance on trained tasks can increase dramatically, transfer of this learning to other tasks remains poor. Here, we present evidence for transfer from training on a demanding working memory task to measures of Gf. This transfer results even though the trained task is entirely different from the intelligence test itself. Furthermore, we demonstrate that the extent of gain in intelligence critically depends on the amount of training: the more training, the more improvement in Gf. That is, the training effect is dosage-dependent. Thus, in contrast to many previous studies, we conclude that it is possible to improve Gf without practicing the testing tasks themselves, opening a wide range of applications.

Speaking of IQ, Linda S. Gottfredson recently published Logical fallacies used to dismiss the evidence on intelligence testing which might be of interest to some of the readers.

For anybody who's interested in trying the "dual n-back" task mentioned in Jaeggi et al., there is Brain Workshop, a free, open-source implementation, with a lively community of users.

Hey thanks for this. I'm always up for an article in intelligence research.

I agree that IQ is somewhat malleable, but I'm by no means confident that all attempts to raise IQ, e.g. test-taking skills, will reduce akrasia, or improve planning, or risk taking decisions.

What do you mean by 'practice self-distraction'? Can you give an example?

If you're hungry but, for some reason, think you should refrain from eating, you can go do something sufficiently absorbing that you stop thinking about how hungry you are. Such as play video games.