Sunday, July 01, 2007

I am the message centre and I am in control

Cognitive neuroscientists staged an experiment in which subjects were asked to make decisions on whether or not to invest money. It was the classic set-up that demonstrates risk-aversion bias: on each round, the subject could choose between keeping $1 and taking a 50 per cent chance of getting $2.50. Theoretically, you should always take the gamble, as the present value of the 50 per cent chance is greater than $1, but usually, people will only go for it 60 per cent of the time or thereabouts.

The clever bit, though: they asked some brain-damaged patients, who have lost part of the brain that deals with some emotions, to do the test as well as the normals who acted as the control group. And what happened?
Patients incapable of feeling emotions chose to invest 83.7 percent of the time, and gained significantly more money than normal subjects. They also proved much more resistant to the sting of losses, and chose to gamble 85.2 percent of the time after they lost a coin toss. In other words, losing money made them more likely to invest, as they realized that investing was the best way to recoup their losses. It is an irony of economic theory that it only excels at predicting the behavior of patients with serious brain injuries.
I can already see that there's a good sci-fi story in this, and possibly even a movie. Imagine the character, a near-future stockbroker or HFG who undergoes deliberate brain damage in order to improve their investments - and seriously impair their relations with other people, but hell, doesn't working 70 hours a week do that anyway? And there's serious money in it. And if we happen to run over a cad, we can pay for the damage if ever so bad. How pleasant it is to have money..

Then, it happens! and the plot is away, scrabbling over the rooftops...


Anonymous said...

hell, doesn't working 70 hours a week do that anyway?

less than you'd think ...

I'm planning on writing something about this general area at some point in the future, in the context of "The Black Swan" by Nicholas Taleb. Thing is, outside some very oddball specialised situations, in the good old financial markets, you're never offered a gamble that's a purely random draw from a known distribution. You're offered the opportunity to take a bet on a particular situation, which is almost always entirely shaped by a lot of human beings. All the really good players have a very sharp understanding of human nature indeed - for some reason, options traders are always backgammon players, while stockbrokers favour chess over poker and bond traders vice versa.

Anonymous said...

I remember that Adam Curtis' The Trap (end of the 2nd programme) asserted that the only two kinds of people who conform to rational behaviour as predicted by economic (game) theory are economists and psychopaths.

So: "It is an irony of economic theory that it only excels at predicting the behavior of patients with serious brain injuries."

sounds just about right then.

Anonymous said...

This is a major subplot in Peter Watts' excellent novel _Blindsight_. Came out last year, is up for a Hugo.

BTW, might it be time to remove the "Blair must go" banner? Just wondering.

Doug M.

JonHendry said...

It'd be a good basis for a sequel to Rain Man.

On the other hand, I get the impression a certain amount of schmoozing skill is required to make it in the financial world.

Alex said...

You haven't met Dsquared, clearly.

Anonymous said...

This is also a major idea in A Deepness in the Sky by Vernor Vinge.

Anonymous said...

The result seems completely reasonable to me. The brain-damaged would come out better only on average. A substantial proportion of them would come out worse while the normal subjects are certain to come out better from their choice.

Anonymous said...

The piece implies the players were allowed to make the decision several times. In that case, the best decision (the one that would be expected to give the best outcome) is to take the gamble (in this case) as the odds of not winning the same or more are slim:

1+1+1 = 3
(2.5+2.5+2.5)/2 = 3.75
The odds of winning zero times is 0.5^3 (1/8)
There are three ways of winning once ($2.5-- not too bad an outcome) so the odds are 3/8
twice ($5) is 3/8
three times ($7.5) is 1/8

So for three games, 7/8ths of the time you'll do OK or well, and 50% of the time you'll do better than if you took the safe option.

As you play more games, the odds of winning improve further. Though there are more suboptimal possibilities, in proportion there are slightly more outcomes that are better than taking the certain outcome. Further, the odds of not getting something near the expected amount get longer (this requires a separate explanation).

A fallacy (in a sense) here is that there is an option with no associated risk. Such things don't actually exist.

kostenloser Counter