Gambling monkeys give insight into neural machinery of risk

Duke University Medical Center neurobiologists have pinpointed circuitry in the brains of monkeys that assesses the level of risk in a given action. Their findings -- gained from experiments in which they gave the monkeys a chance to gamble to receive juice rewards -- could give insights into why humans compulsively engage in risky behaviors, including gambling, unsafe sex, drug use and overeating.

The researchers, Michael Platt and Allison McCoy, published their findings in the advanced online version of Nature Neuroscience, posted August 14, 2005. The research was sponsored by the National Institutes of Health, the EJLB Foundation, and the Klingenstein Foundation.

In their experiments, the researchers gave two male rhesus macaque monkeys chances to choose to look at either of two target lights on a screen. Looking at the "safe" target light yielded the same fruit juice reward each time. However, looking at the "risky" target light might yield a larger or smaller juice reward. The average juice reward delivered by looking at either target was the same.

To their surprise, the monkeys overwhelmingly preferred to gamble by looking at the risky target. This preference held, regardless of whether the scientists made the risky target reward more variable, or whether the monkeys had received more or less fruit juice during the course of the day.

"There was no rational reason why monkeys might prefer one of these options over the other because, according to the theory of expected value, they're identical," said Platt. The researchers also tested whether the monkeys were simply responding to the novelty of the risky target.

"We wondered whether the monkeys preferred the risky target because the experiment was dull and boring, and they wanted the variability," said Platt. "But when we made the task more interesting by changing the color of the lights on each trial, the monkeys didn't care anything about it."

In fact, when the researchers made the average payoff for the risky target less than for the safe target, "we found that they still preferred the risky target," said Platt. "Basically these monkeys really liked to gamble. There was something intrinsically rewarding about choosing a target that offered a variable juice reward, as if the variability in rewards that they experienced was in itself rewarding."

Even when the researchers subjected the monkeys to a string of "losses," the high of a "win" appeared to keep them going, said Platt.

"If they got a big reward one time on the risky choice, but then continued to get small rewards, they would keep going back as if they were searching or waiting or hoping to get that big payoff. It seemed very, very similar to the experience of people who are compulsive gamblers. While it's always dangerous to anthropomorphize, it seemed as if these monkeys got a high out of getting a big reward that obliterated any memory of all the losses that they would experience following that big reward," said Platt.

Confident that they had developed a valid animal model that would reveal insights into the brain mechanism for assessing risk, the researchers next explored the neural circuitry that governed that assessment. They threaded hair-thin microelectrodes into a brain region called the posterior cingulate cortex, which studies in humans and animals had implicated in the processing of information on rewards. They then measured the electrical activity of neurons in the region as they administered the same behavioral task to the monkeys.

"We found that the neurons behaved very similarly to the monkeys," said Platt. "That is, as we increased the riskiness of a target, the neurons' activity would go up in the same way the monkey's frequency of choosing that target would go up. It was amazing the degree to which the activity of these neurons paralleled the behavior of the monkeys. They looked like they were signaling, in fact, the monkeys' subjective valuation of that target," he said. Further analysis of the neuronal activity indicated that, indeed, the neurons were reflecting the risk value the monkeys placed on the target, rather than an after-the-fact response to the payoff.

While Platt and McCoy believed they have isolated one component of the neural machinery of risk, they do not believe they have mapped the entire circuitry.

"We don't think the posterior cingulate cortex is by any means the only area that's important for assessing risk, for deciding what's valuable and for actually making a choice based on that valuation," said Platt. "We think that this is just part of a whole circuit that's involved in that process." However, he said, pinpointing a key region involved in risk assessment will enable further studies to map that circuitry.

"It's going to be interesting to trace this circuitry to see which parts of the brain are signaling something about subjective utility and which parts of the brain are signaling information about true reward and punishment experiences," said Platt.

He emphasized that such animal studies are a highly useful complement to human studies and genetic studies using mice. Neuroimaging studies in humans performing such tasks can identify brain regions involved in making decisions based on risk, he said.

"And then, using these animals, we can do electrophysiological studies that allow us to understand how the fundamental processing units of the brain -- single nerve cells -- actually process information about reward and risk and uncertainty; and how that information might contribute to the actual decision process that results in the monkey's choice," said Platt. What's more, he said, the monkey studies allow manipulation of the circuitry using drugs to determine how the circuitry might malfunction in human disorders.

"For example, it is believed that people who have low levels of the neurotransmitter serotonin might be more risk prone and impulsive," said Platt. "Disturbances in such neurotransmitter systems might be the basis of pathological conditions like pathological gambling, obsessive-compulsive disorder and depression. We can do pharmacological manipulation of the serotonin system in monkeys to see how it influences risk perception and risk preferences, and whether we see changes at the level of the single neurons that we're studying."

What's more, said Platt, the studies with monkeys can guide studies in mice, in which scientists can make genetic alterations in the mice and study the behavioral effects of those alterations. Such studies could contribute to understanding of the genetic basis of compulsive behaviors and other such behavioral disorders.

Source : Duke University Medical Center