Wednesday, September 21, 2005

Neuroeconomics 2005

This past weekend was the 4-day "Neuroeconomics 2005" conference on Kiawah Island, South Carolina. "Neuroeconomics" is the study of how people make decisions about things they value (or should value). One example is how people evaluate and trade stocks. Other examples include games that measure financial trust, honesty and cooperation, strategic decision making and asset allocation, and risk preferences. This conference is a largely academic affair with research presentations by neuroscience, psychology, and economics researchers throughout each day.

Attendees were largely professors and graduate students, though a few pracitioners such as Jason Zweig (Money Magazine), Arnold Wood (Martingale Asset Management), Shirley Mueller (Star Financial), and myself were also present.

I'm unable to report on details of the latest "neurofinance" research until it appears in publication. The Stanford researchers (Kuhnen and Knutson -- see post below) have additional data from their study, and researchers at Baylor College of Medicine (especially Terry Lohrenz) are doing a fascinating asset market experiment with fMRI. Lohrenz is a former energy trader and mathematician.

In general, it was a very enjoyable meeting. Some of the findings will be described in later posts as they are published.


Thursday, September 01, 2005

Greed and Fear sabotage investors in a Stanford study

"We simply attempt to be fearful when others are greedy and to be greedy only when others are fearful."
~Warren Buffet

A recent neuroimaging study from researchers at Stanford University reveals the brain origins and expensive consequences of emotions experienced during financial decision making. Graduate student Camelia Kuhnen (School of Business) and professor Brian Knutson (Psychology Department) performed brain imaging on Stanford graduate students while they engaged in an investment decision making task. They report on this study in a paper entitled “The Neural Basis of Financial Risk-Taking” (2005), published in the journal Neuron v47, n5 on September 1, 2005. This is the first study to monitor subcortical brain activity during an investment decision-making experiment.

The experimenters used functional magnetic resonance imaging (fMRI) to examine brain function. FMRI allows researchers to measure changes in brain activity over short time intervals (2 seconds) in small regions of the brain (2 cubic millimeters). Changes in brain activity are indicated by alterations in regional metabolism, tissue oxygen usage, and blood flow.

In their experimental task, called the Behavioral Investment Allocation Strategy (BIAS) task, Kuhnen and Knutson asked subjects to make an investment choice between three alternatives. Two of the alternatives were risky “stocks” (either "Stock A" or "Stock B"), one with a net positive payoff over the 10 trials, and the other with a net negative payoff over the ten trials. At the beginning of each block of 10 trials, subjects did not know which stock was profitable or which was money losing. The "good" stock had a payoff of +$10 (50%), $0 (25%), and -$10 (25%), for an expected value per trial of $2.50. While the "bad" stock had payoffs of +$10 (25%), $0 (25%), and -$10 (50%), for an expected value per trial of -$2.50. A third choice was a bond which yielded a constant return of $1 per trial.

Potential Choice Payoffs:
Bond = $1 per trial
Stock A = Averaged either +$2.50 or -$2.50 per trial.
Stock B = Averaged either +$2.50 or -$2.50 per trial.

While the goal of this task is to make as much money as possible, subjects can't know which stock is the high paying one until after the task has begun. Subjects made "irrational" stock choices when they did not maximize their profits (e.g., they chose a bond even though fairly certain of which is the high-paying stock).

Dear reader, try to decide how you would choose in this experiment. Would you start out choosing a stock? If not, how would you proceed? After how many trials would you know with some certainty which was the high-payoff stock? [Do this quickly before reading on].

While laying on their backs in the MRI scanner, subjects were allotted 12 seconds per trial, with 10 trials per block. For each trial:
1. Subjects would see the three options appear on a computer screen for 2 seconds (Stock A, Stock B, and Bond),
2. followed by a 2 second delay during which they were asked to make a choice,
3. then another 2 second delay interval,
4. then a report of the outcome of their choice 2 seconds,
5. then a report of the outcomes of all the possible choices for 2 seconds,
6. then a 2 second delay until the next trial.
Each of the trials lasted 12 seconds total. There were 20 blocks of 10 trials each. Subjects were encouraged to try to make as much money as possible.

After learning progressing through several the trials, one should optimally begin to choose the "good" stock when 70% certain of its identity. Of course, most of us don’t know Bayes theorem by heart. Truth be told, few readers have heard of Bayes theorem. And if you're an investor, and you actually use it, you're a rare breed.

The true genius of this paper lies in how the researchers analyzed the data they had obtained. They found that subjects tended to make irrational decisions, and that two areas of the brain, strongly linked to the experience of the emotions excitement and fear, were activated just before irrational decisions were made.

Subjects would occasionally choose the bond after the experiment was well underway and they should have had adequate experience to choose the right stock. In the situation where the subjects had adequate information about the stocks to make a profitable decision, but they chose the bond anyway, they were making a "risk-aversion" mistake. Activation in an area of the brain called the anterior insula was correlated with risk-aversion errors (see image below). This image is from one of our unpublished studies on expected utility.

Anterior insula activation occurs when we are experiencing pain, anxiety, or disgust. After choosing one of the stocks, and losing, subjects felt pain about their loss and anxiety about losing again. Following such as loss, they were more likely to choose the safety of the bond, thus making a "risk-aversion mistake". This is a similar result to that in the "brain-damaged investors" study described in a previous blog post where, immediately following a loss, 20% fewer subjects accepted a subsequent gamble of the same, postive, odds.

When subjects in the experiment took too much risk, a different area of the brain was activated. Subjects who chose one of the stocks before the odds were known to be in their favor had higher activation of an area of the brain called the nucleus accumbens (see image below):

The nucleus accumbens is famous for several interesting characteristics:
1. The Pleasure Center. Rats with a wire placed in the nucleus accumbens would press a lever, to electrically stimulate themselves, repeatedly until they collapsed from exhaustion or died (Olds and Milner, 1954). Neurosurgeons found that people reported intense sensations of well-being (and some had orgasms) when stimulated in the nucleus accumbens. More recently, Knutson et al (2001) demonstrated that nucleus accumbens activation is linearly correlated with subjective reports of positive emotionality.
2. Dope and Drug Abuse. All drugs of abuse activate dopamine neurons in the mesolimbic dopamine circuit, which has terminals in the nucleus accumbens.
3. Reward Anticipation. The nucleus accumbens is activated on fMRI scans when people are anticipating or expecting to make money (Knutson et al, 2001) as well as other satisfying rewards.

Kuhnen and Knutson demonstrated that insula activation predicts errors of risk-aversion, and nucleus accumbens activation predicts errors of risk-seeking. Because the insula and nucleus accumbens generate emotions, their activations give rise to feelings. We can use our physical and emotional feelings as signals to indicate when we are likely to make errors in our investment decision making.

In most areas of life, feelings are necessary for making good decisions. For examples, see Malcolm Gladwell's new book Blink (2005). We need to feel in order to have an intuitive sense of which decisions are likely to be successful and which are likely to be damaging.

According to the press release: "On average, the participants in the study made rational choices 75 percent of the time and made mistakes 25 percent of the time, Knutson said. And the brain areas lit up even when rational choices were being made, just not as much. " This implies that emotions are part of both rational and irrational choice behavior. It's the extremes of emotion that lead to irrationality.

Per the press release: "These findings may also explain why casinos employ "reward cues" such as free drinks and surprise gifts as anticipation of other rewards that may activate the NAcc and lead to changes in behavior, Knutson added. Insurance companies might employ the opposite strategy, using strategies that would activate the anterior insula, he said."

Danger arises when feelings are either excessive or misattributed. Intense feelings such as greed and fear lead to the risk-aversion and risk-seeking mistakes seen in the Kuhnen and Knutson study. Misplaced feelings are attributed to incorrect causes. Being given small gifts by a casino may "prime" the reward system and put us in a risk-seeking state of mind. Viewing scenes of recent hurricane destruction prompts fears of other possible catastrophes and makes us risk-averse.

Investors should be careful to not let such biases affect their decision discipline. They can monitor their decisions to notice how they may be biased by recent experiences. "The bigger message may be a common-sense one: Whenever you're facing a big decision, step back a moment and think it over."

In particular, if a person has experienced a recent loss and notes that he or she is feeling nervous and is exhibiting other signs of irrational risk avoidance behavior such as 1) hesitation in entering new positions, 2) deliberating about further potential losses, or 3) seeing more financial threats than usual, then he or she must take special care to not let that increase in anxiety affect future discipline in trading decisions. Conversely, if a financial market participant has recently made large gains and is feeling 1) celebratory, 2) extremely intelligent, and 3) wants to let his position ride, then he or she must make sure not to focus solely upon potential returns and ignore the risk control and monitoring aspects required by his or her trading discipline when making future decisions.

The following paper is a summary of techniques investors can use to identify when their decisions may be biased by damaging emotions:
Top Ten Tips from the Neuroscience of Investing
My own homepage is here: Richard L. Peterson M.D.