Stanford neuroscientists have discovered that collective brain activity can forecast the movement of actual stock prices, even in instances when conscious decision-making and traditional indicators such as price volatility did not.
The study, published March 8 in the Journal of Neuroscience, built on previous research showing that group brain activity can foreshadow collective choice, including how often and how long people looked at videos on YouTube. Its findings raise questions about whether investors can use information about choice to beat the market, recalling economist John Maynard Keynes' famous claim that people often act on the basis of irrational "animal spirits."
"We think that that is not a ridiculous metaphor," said Dr. Brian Knutson, a professor of psychology and neuroscience at Stanford University and an author of the study. "Is there a bull and a bear in the brain? Can we see what they're doing?"
The team focused on two regions deep within the brain, the anterior insula and the nucleus accumbens, both of which have been conserved across millions of years of evolution. The nucleus accumbens is part of the brain's reward circuitry, while the anterior insula is part of what Knutson informally calls the brain's "risk circuitry."
"Instead of pushing you towards something, it's pushing you away from it," Knutson explained. "It's part of the circuit that anticipates bad things happening."
In the first of two experiments, 34 participants were scanned while completing a series of asset pricing tasks. They had to choose whether or not to invest a dollar in a stock based on a trend line showing its price over time. Knutson and his colleagues used actual historical stock data from 2015. They randomly chose 14 stocks from the S&P 500 Index, conducting 10 trials per stock for a total of 140 trials.
A second replication experiment recruited 39 new participants while using different historical stock data. The researchers also sought to eliminate the effects of negative autocorrelation — in other words, the tendency of stocks that went up one day to go down the next.
The scientists paid participants $20 an hour. They also gave participants a $10 endowment with which to trade, adding or subtracting the money participants gained or lost. Knutson explained that they started with $10 rather than zero to avoid cutting into participants' hourly pay. Afterward, they chose a random subset of trials to count for real, adding or subtracting how much participants made or lost from the $20 hourly payment. In the language of economics, this made the experiments "incentive-compatible," Knutson said, meaning that participants did best if they acted on their real preferences.
"We wanted this to matter to participants," Knutson explained.
The results were striking. In the first experiment, activity in the anterior insula averaged across the group predicted changes to the direction of stock prices, while averaged group activity in the nucleus accumbens predicted stock price directions. In the second experiment, nucleus accumbens activity no longer predicted the direction of stock prices, meaning that this brain region might gain useful information from negative autocorrelation. But the anterior insula was still predictive, lighting up in ways that foreshadowed shifts in the direction of stock prices.
Incredibly, group brain activity predicted stock price movements even when group choices did not — the participants essentially broke even in their bets.
"How is it possible that average brain activity can forecast when average behavior does not?" Knutson asked.
The participants were mostly students at Stanford and skewed heavily male. According to the paper, this was "consistent with the sex imbalance typically observed in professional traders." Knutson says this study could still be considered representative.
"Most people think of representativeness in terms of individuals, but we've started to think of representativeness in terms of cognitive processes that support choice," said Knutson. "Some of the processes might generalize more across people."
He noted that his team obtained some preliminary data from the Robinhood trading app a few years ago. Their initial analysis suggested that Robinhood traders acted very differently from traders on Wall Street.
Knutson agrees that this research calls to mind the "PreCrime" division depicted in "Minority Report," which relied on clairvoyant humans to stop crimes before they occurred. "Do we have mutants? No," joked Knutson.
Yet there is a tantalizing possibility that individual differences in cognition, personality or other factors could make some brains better predictors.
"When I talk to people in finance, the first thing they think of is experience — that experienced traders may have better behavior or better brain activity," said Knutson. "I regard it as an empirical question. It could be better. It could be worse. We don't know."
Knutson believes the study's results might have implications for how we think about financial markets.
"People in finance say there's a difference between gambling and finance," said Knutson. "I'm not so sure that's the case. I think it's maybe a distinction of degree rather than quality."
He also thinks his results could offer guidance to average day traders hoping to maximize their gains.
"You shouldn't necessarily follow your immediate emotional impulses — and certainly, those can be magnified by what everybody else is doing," said Knutson. "When everybody else's anterior insula is on, and something's going up, that's when the price is most likely to flip."
"The problem there is, how do you measure everybody's anterior insula?" he added. "What people say may not be what's going on in their brains."
"People may very well know what Warren Buffett does — and they may be able to tell you that — and yet when you see what they do, it's not the same," said Knutson.
The study, "Brain activity foreshadows stock price dynamics," published in the Journal of Neuroscience on March 8, was authored by Mirre Stallen, Nicholas Borg and Brian Knutson, Stanford University.