For the first time, scientists have shown that they can predict when people will make risky decisions based on brain activity patterns. Could this lead to a world where we consult brain scans to predict whether we're making a risky choice or not?
Our lives revolve around choices, where we must sometimes choose between a "safe" option and a "risky" one. We typically know what the outcome of the safe option will be, but the risky option often has multiple — and sometimes unknown — outcomes.
For example, if you get drunk at a bar, you could choose to drive home or take a cab. If you take a cab, you can reasonably assume that you'll get home with little complication. If you drive, you may make it home in one piece and save on the cab fare, but you could also get into an accident or get pulled over by a police officer. But what's going on in our brains to make us choose the risky option over the safe one? Perhaps not surprisingly, the brain scans suggest that risk-taking is linked to poor impulse control.
"There has been quite a lot of research done on the neural correlates of risk, and on how the brain responds to different types of risk," explained Sarah Helfinstein, a neurobiologist at the University of Texas at Austin. Specifically, researchers have determined that there's a large network of brain regions that are "sensitive" to risk, including the striatum, thalamus and insula. And the stronger the rewards an option presents, the more likely we will choose it. "But what hasn't been known is how all of that stuff determines what kind of decisions people make," Helfinstein told io9.
If people are faced with a decision that's associated with a given amount of risk, what happens in their brain to make them choose the risky option over the safe option? And could that brain activity be used to actually predict people's choices? Helfinstein and her colleagues decided to find out.
For the study, the researchers had 108 participants perform a naturalistic risk-taking task called the Balloon Analog Risk Task, which requires them to pump up a virtual balloon. Participants receive points for each pump, which they can then "cash out" at any time, ending the balloon round. But if the balloon pops before they cash out, they lose all of their accumulated points. To further increase the stakes, the balloon is set to randomly explode between the 1st and 12th (final) pump, so each pump carries a risk of exploding and losing points.
The participants each spent 9 minutes on the balloon experiments while inside of an fMRI scanner, which recorded their brain activity. The researchers fed a subset of the brain scan data from some of participants into a "machine classification" algorithm. "This works by giving the computer the data, and saying, 'These samples here came from trials where subjects made a risky choice, and these are from trials where subjects made a safe choice,'" Helfinstein said. "'Now look at data from brain activity patterns and try to discriminate between the decisions.'"
The team made sure to only give the algorithm fMRI data up to the pump trials that immediately preceded the trials where the participants made their choices. "We didn't want the classifier to be able to tell the difference between the choices based on things that weren't relevant," Helfinstein explained. For example, if the computer had the brain maps from the decision trials, it could possibly discriminate between trials based on motor or reward effects in the brain, rather than the cognitive processes that lead up to the risky or safe decision.
After teaching the algorithm, the researchers gave it the rest of the brain scan maps. "We asked it, 'From looking at this data, can you tell us if the subjects are going to make risky or safe decision?'" Helfinstein said. Amazingly, the algorithm accurately predicted participant choices about 72 percent of the time.
In addition to using a whole-brain classifier, Helfinstein and her colleagues conducted a "searchlight analysis," in which they gave the computer data from only a tiny chunk of the brain at a time. This process allowed them to see which brain regions are most involved in making risky and safe choices. The regions identified in the searchlight analysis, they found, were those that are involved in cognitive control — areas involved with controlling your behavior and choices. Interestingly, these regions were more active when people made safe choices than when they made risky choices, suggesting that risky decisions may arise when the control systems fail to initiate the safe choice.
Helfinstein doesn't see any direct, practical applications of the research. After all, people don't spend their lives in fMRI scanners, so it's not as if we can tell when people are going to make a risky decision in their day-to-day activities. However, the research may someday help habitual risk-takers make safer decisions. "Maybe we can develop ways of helping people cultivate control, as a way of helping them make safer decisions," she said.
Check out the full study over in the journal PNAS.