Say you got passed up for promotion at work. You immediately blame your manager, who has always had it out for you — but how did you arrive at that conclusion, and so quickly?

Some philosophers suggest people determine who should take responsibility for a particular outcome by imagining what would have happened if a suspected cause had not intervened. This kind of reasoning, known as counterfactual simulation, is believed to occur in many situations. For example, soccer referees deciding whether a player should be credited with an “own goal” (a goal accidentally scored for the opposing team) must try to determine what would have happened had the player not touched the ball.

This process of blaming can be conscious, as in the soccer example, or unconscious, so that we are not even aware we are doing it, as in the blaming of the manager — and by using technology that tracks eye movements, cognitive scientists at MIT have now obtained the first direct evidence of the brain functions that underpin both.

Article continues below

Until now, studies of counterfactual simulation could only use reports from people describing how they made judgments about blame and responsibility, which offered only indirect evidence of how their minds were working. (It requires a high degree of self-awareness to be able to parse snap judgments and unconscious decisions, which most of us don’t have.)

The MIT team set out to find more direct evidence of how people decide what or who is to blame for an outcome, by tracking people’s eye movements as they watched two billiard balls collide. The researchers created 18 videos showing different possible outcomes of the collisions. In some cases, the collision knocked one of the balls through a gate; in others, it prevented the ball from doing so.

Before watching the videos, some participants were told that they would be asked to rate how strongly they agreed with statements related to ball A’s effect on ball B, such as, “Ball A caused ball B to go through the gate.” Other participants were asked simply what the outcome of the collision was.

As the subjects watched the videos, the researchers were able to track their eye movements using an infrared light that reflects off the pupil and reveals where the eye is looking. This allowed the researchers, for the first time, to gain a window into how the mind imagines possible outcomes that did not occur, and thus determine what or who is to blame for an outcome.

“What’s really cool about eye tracking is it lets you see things that you’re not consciously aware of,” said Josh Tenenbaum, a professor in MIT’s Department of Brain and Cognitive Sciences and the senior author of the study published in the journal Psychological Science. “When psychologists and philosophers have proposed the idea of counterfactual simulation, they haven’t necessarily meant that you do this consciously. It’s something going on behind the surface, and eye tracking is able to reveal that.”

The researchers found that when participants were asked questions about ball A’s effect on the path of ball B, their eyes followed the course that ball B would have taken had ball A not interfered. Furthermore, the more uncertainty there was as to whether ball A had an effect on the outcome, the more often participants looked toward ball B’s imaginary trajectory.

“It’s in the close cases where you see the most counterfactual looks. They’re using those looks to resolve the uncertainty,” Tenenbaum said.

Participants who were asked only what the actual outcome had been did not perform the same eye movements along ball B’s alternative pathway.

The researchers are now using this approach to study more complex situations in which people use counterfactual simulation to make judgments of causality and assign blame.

“We think this process of counterfactual simulation is really pervasive,” said Tobias Gerstenberg, a postdoc at MIT and lead author of the study. “In many cases it may not be supported by eye movements, because there are many kinds of abstract counterfactual thinking that we just do in our mind. But the billiard-ball collisions lead to a particular kind of counterfactual simulation where we can see it.”

One example the researchers are studying is the following: Imagine ball C is headed for the gate, while balls A and B each head toward C. Either one could knock C off course, but A gets there first. Is B off the hook, or should it still bear some responsibility for the outcome?

“Part of what we are trying to do with this work is get a little bit more clarity on how people deal with these complex cases. In an ideal world, the work we’re doing can inform the notions of causality that are used in the law,” Gerstenberg says. “There is quite a bit of interaction between computer science, psychology, and legal science. We’re all in the same game of trying to understand how people think about causation.”