PsyFi Search

Wednesday 27 January 2010

Gambling, From Iowa to Soochow

Bodies and Brains

One of the more convoluted and, to the majority of the world, boring arguments among psychologists is around the extent to which we use our brains to learn. Of course a commonsense view has it that brains are fairly essential things for learning type activities but there’s also a view that bodies have a part to play as well.

Through a series of convoluted jumps this takes us from William James' research on phantom limbs, through the Somatic Marker Hypothesis and onto the odd findings of the Soochow Gambling Task. On the other, less invisible, hand it might be better for all of us if I simply skip to the nub of the problem: investors are addicted to gains, so much so that they’ll happily make overall losses just as long as they make lots and lots of small wins along the way.

From James to Damasio

To cover some history, albeit briefly, the idea that there’s more to life's experience than simply triggering activity in the brain has been around since before William James wrote The Consciousness of Lost Limbs in 1887. The sheer number of American Civil War amputees experiencing phantom limbs suggested to James that the body was more than simply a complex vessel for carrying about the brain – the body was, in effect, the way the brain experienced the world and, if the brain decided that it didn’t like the way the world was being experienced it was liable to change the way it regarded it.

Antonio Damasio is probably the most famous modern exponent of such theories and his Somatic Marker Hypothesis is, to cut a very long story short, a modern day neuroscientific version of James’ ideas. The body and the brain are intimately linked – the body experiences the world, the brain interprets the experiences and lays down so-called somatic markers which bias future interpretations and the body reacts. Emotions are the result of the combination of the experiences and the interpretation and, as Damasio has shown, people who don’t have emotions – usually because of brain injuries – don’t behave in ways we’d consider rational.

The Iowa Gambling Task

Rationality is, as usual, at the heart of this but the idea that people behave less rationally when they’re less emotional is, well, odd. As we’ve seen before, in Unemotional Investing is Best and Get an Emotional Margin of Safety, Damasio has shown that emotionally damaged people make better investing choices under ambiguous conditions than their uninjured counterparts. However, this comes at a price – in the real world such people are also likely to fail to appreciate situations of real danger. Somehow emotion is integrated with rational behaviour – which is a bit of a surprise really when we consider how emotional responses tend to destroy investors’ returns.

The classic experiment demonstrating this mind-body link is known as the Iowa Gambling Task. Participants are given a wad of cash to gamble with and need to take a card from one of four decks. Two of the decks are “bad” in that if you choose from them consistently you will lose $250. Two of the decks are “good” in that if you choose from them consistently you will win $250. Note, however, that the way you win or lose your $250 differs – deck A (loser) and deck C (winner) have 5 gains and 5 losses while deck B (loser) and deck C (winner) have 9 gains and 1 loss.

The results seem clear cut – the uninjured participants learn, after about 40 draws, to avoid the bad decks. The brain damaged participants don’t ever really learn. In addition the participants were wired up to a lie detector, measuring their physiological reaction to making a choice. The uninjured participants started showing a reaction to choosing from a bad deck after only 10 card flips but could only consciously report this after 50 card flips. Overall, this seemed to show that emotional responses are needed to help make decisions under risky conditions and that the bits of the brain that signal risk are engaged way before we are consciously aware of them.

From Iowa to Soochow

Now, if you’ve been following this journal you’ll probably have an uneasy feeling that there’s something not quite right about this, but not be able to put your finger on it. Largely, no doubt, because there are no free lunches in behavioural finance and partly because the ability of uninjured participants to learn to anticipate risk doesn’t seem to accord with the findings of Prospect Theory and Mental Accounting: normally investors are short-sighted, yet the Iowa Gambling Task suggests that they’re looking at long term outcomes.

You’d be right, of course.

More detailed analysis of the results revealed that the experimental design was potentially flawed. Although the overall results showed uninjured participants preferring the winning decks in aggregate the singly most preferred deck was one of the losing ones: deck B with 9 wins and 1 loss, but with the loss a mammoth one. So, is there anything else we can think of that resembles lots of small wins, alternated by the occasional whopping loss?

The Soochow Gambling Task

A group of Taiwanese researchers analysed and modified the Iowa Gambling Task in what’s now known as the Soochow Gambling Task (see also this research) and proposed an alternative explanation to Damasio’s hypothesis – one that is nothing to do with somatic markers and everything to do with the relative frequencies of wins and losses. The relative preponderance of wins in deck B was biasing people towards it. In simple terms, the more times they won from that deck the more likely they were to stick with it. The researchers went on to confirm their results in a later study.

This, of course, looks a hell of a lot like the typical loss aversion biases associated with stockmarket investors, a suggestion confirmed by a study of decision learning models by Ahn and colleagues. Typical loss aversion sees investors selling winners for small amounts and holding losers for large losses – both behaviours due to an attempt to avoid losses. Perhaps even more relevant are myopic loss aversion tendencies where investors discount the probability of extreme rare events to zero and then see years of small gains wiped out by a single, catastrophic market crash for which they’re psychologically unprepared.

Fascinatingly even when they pointed out the potential problems with high frequency winning decks participants continued to insist that they’d choose cards from those decks. Again, this seems uncannily like the behaviour of stockmarket investors choosing to chase gains knowing that they're bound to go wrong eventually but expecting, or hoping, to flip their winners before calamity strikes.

Accidental Behavioural Finance?

The especially interesting thing about the outcome of the Soochow Gambling Task is how it appears to validate behavioural finance completely by accident – an experiment originally designed to look at the difference in rational analysis between people with and without certain types of brain injury has produced evidence pointing in a completely different direction.

Although the original Iowa Gambling Task has led us away from the result originally expected – and noting that the alternative explanations are not without their critics – there is still plenty of evidence linking decision making and emotional states. It couldn’t really be any other way, we’re emotional in order to aid real-life decision making, not to prevent it. As long as our emotions help us get it right more often than not – and leaving aside what ‘right’ means in this context – then that’s as good as it gets. Just don’t expect your emotions to help you be a better investor, they won’t: the stockmarket is not real-life as we're designed to expect it.

Related Articles: Get an Emotional Margin of Safety, Loss Aversion Affects Tiger Woods, Too, Unemotional Investing is Best

1 comment:

  1. "Just don’t expect your emotions to help you be a better investor, they won’t: the stockmarket is not real-life as we're designed to expect it."

    I'm not so sure.

    The conventional investing advice today is that it is a bad thing to panic during a market crash. Andrew Smithers has a different take. He says that those who panic first panic best. Sometimes there is a good reason for panic. At times of insane overvaluation, feelings of panic are an investor's best friend.

    I think that emotion can be put to good uses or bad uses, just as reason can be put to good uses or bad uses. Lots of the "dumb" people who gave in to emotional feelings of panic will end up looking smart and lots of "rational" people who looked to "studies" saying to buy and hold will end up looking dumb.

    The full reality is that all human decision-making involves a mix of reason and emotion. There is no purely rational decision and there is no purely emotional decision. We always rationalize our emotional choices. And we always choose what "studies" to look at because we suspect that they are going to support our emotional predispositions.