Akrasia

[nyuanshin]

"The Fremen were supreme in that quality the ancients called spannungsbogen -- which is the self-imposed delay between desire for a thing and the act of reaching out to grasp that thing."
-- from The Wisdom of Muad'Dib by the Princess Irulan
Take a random sample of a few hundred 4-year-olds. Put them in a room with a marshmallow on a plate and some

Comments

[airstrip]

I must wonder what direction the causal arrow points. I wish we had a way to access what it is to get things wrong....

[the_kot]

Well... An interesting fact, thank you for the information.

[smandal]

What do you think is a possible economic niche for the segment of the population that is both highly intelligent and highly impulsive? My naive guess is in creative and risky enterprises, like entrepreneurship, the arts, celebrity-hood, etc.

[nyuanshin]

Those, and also maybe investigative journalism or stock trading. Anything high risk/high reward.

Marshmallows

[sethisalive]

Akrasia is a great word! I've been searching for it for...forever. Thanks!

I think one always has to be careful about averages, although they are interesting. Variance and distribution are important, too. For example, if 10% of the impulsive folk (as measured by the marshmallow effect) contribute to the entrepreneurial creativeness, invention, technology, and higher-end labor productivity that spurs 30% of economic growth, then the vector for the average and the 10% are not isomorphic, the marginal return on each is different, the average skews the truth. A similar effect is surely observed for intelligence, maybe something like this for example (with an arbitrary definition for "genius"): marshmallow-impulsive: 10% geniuses, 90% not geniuses -- non-marshmallow-impulsive: 1% geniuses, 99% not geniuses.

Re: Marshmallows

[nyuanshin]

Yeah, I can buy the idea that the high-functioning subset of impulsive people might punch well above their weight in terms of contributing to economic growth. But given the inverse relationship between IQ and delay discounting there'd have to be some weird distribution going on there for that set to be significantly large.

really?

[rixim]

this is maybe the most depressing post of yours i've read. psychometrics suck; In my many years of dealing with intelligence tests i've seen little to support that they have any merit at all. In turn, I find things that are compared to them in this way... unsettling. what about the four year olds that don't like marshmallows?

ya, rly

[nyuanshin]

Well, apparently you haven't seen enough. IQ predicts a bunch of sociological outcomes better than most variables and has numerous biological correlates; I have issues with psychometrics (I've been putting off writing about this subject for like almost a year now) but the prima facie value of intelligence tests is long established.

"Marshmallow" is just a dummy for "some food item the kid really likes". If taste was a significant factor in the Mischel & Shoda studies then this would tend to decrease the correlations they found, meaning controlling for taste would have made the test an even better predictor than it already is.

experimental design, where?

[rixim]

I always find it interesting when people are fans of IQ testing; it makes me wonder what their life experiences are like or what their idea of intelligent is.

"Marshmallow" is just a dummy for "some food item the kid really likes". If taste was a significant factor in the Mischel & Shoda studies then this would tend to decrease the correlations they found, meaning controlling for taste would have made the test an even better predictor than it already is.

are you sure about that? think about that for a minute. I've been looking at this for a while, and without looking at the studies themselves, it would seem to me that you're making an error about how correlation works. since the correlation is only between two variables, i'm not sure how this statement could possibly be true. as you described the experiment "taste" (or what i was suggesting, preference) is a confound, there is no way to assert what impact it would have on the correlation one way or another. Feel free to correct me if i'm wrong.

Re: experimental design, where?

[nyuanshin]

My assumption is that the taste factor would just add unsystematic noise to the data, which decreases correlations. I can't think of why there'd be *systematic* bias, though I'm all ears if you've got suggestions.

If anything my personal experiences with intelligence tests and those of people I know personally have been ambiguous. I was practically dragged kicking and screaming to where I am now by all the reading I did on the subject last year.