Verified by 
Key points
- You can track eye movements to detect what people are paying attention to.
- Seeing one action someone takes guides later eye movements.
- These eye movements are predictions about the future.
- When those predictions are wrong, it guides what is learned from a situation.
A critical function of the brain is to predict the future in order to react effectively to what happens around you. Suppose you are walking down the middle of a sidewalk and you see someone coming toward you in the opposite direction. Ideally, you would like to predict which side they will select to walk on so that you can pass by them without having to stop or bump into them.
That means that you are constantly making predictions about other people’s actions as a situation unfolds. It also suggests that you should be particularly interested in cases where your prediction fails, because that suggests a place where you need to update your beliefs to prepare for the future.
This idea was explored in a fascinating set of studies by Christopher Wahlheim, Michelle Eisenberg, David Sawarczyk, and Jeff Zacks in a May 2022 paper in the journal Psychological Science.
They were able to explore the way people react to events on a moment-by-moment basis using eye-tracking. When you look at the world, you get only a small area of clear vision (about the size of your thumbnail at arm’s length), because that is where you have the most densely-packed cells at the back of your eye in the retina. In order to build up a sense of the overall scene you’re looking at, you have to move your eyes to various locations of the scene.
You may not realize it, but you move your eyes a few times each second in order to build up your view of a scene. As a result, when you have a prediction of what is going to happen in the future, you are likely to move your eyes to an area that relates to what you predict will happen. Eye-tracking uses a device that follows these eye movements on a millisecond-by-millisecond basis.
The researchers had people watch two videos of a woman engaging in common daily home and work activities. They were told that each video reflected her activities on a day of her life.
In some situations, she performed a similar action each day, and did it in the same way. For example, she might open a kitchen cabinet and take out a small bowl each day. In other situations, she performed a similar action each day, but there was a key difference. For example, she might walk to her front door and unlock the lock on the doorknob one day, but unlock the lock above the doorknob the second day. In one study, there was also a control condition in which the woman performed some actions only on the second day.
In the first session of the study, participants watched both videos while their eyes were being tracked. Then, about a week later, participants were asked about their memory of the scenes. They were given a cue such as “What did the woman take out of the cabinet on the second day?” After responding, they were asked if that was the same or different from what she did on the first day. If they thought there had been a change, they were also asked for their memory of what she did on the first day.
There were several interesting findings in this study.
When people were watching the second video, their eyes typically moved to an object that the woman interacted with in the first video. So, they might look at the bowl as she approached the cabinet or the doorknob as she walked to the door. Once she began to perform the action, though, they looked at the object she engaged with in that action. These eye movements suggest that people are predicting what the woman is going to do, though they can also recover and detect when she has done something different on the second day than she did on the first.
Second, while watching the second video when people moved their eyes to look at the object the woman interacted with on the first day, they were better able to remember a change from the first day to the second than if they did not move their eyes to the correct object. That is, when people remembered what happened on the first day and moved their eyes in prediction, they had a great memory of the change and great memory of what the woman had interacted with on the first day.
Third, while watching the second video, when people did not look at the object the woman interacted with on the first day, their memory of the change was bad. That is, if they had a poor memory of the action on the first day, they were unable to predict what was going to happen next, and their memory of what actually happened was bad as well.
Putting all of this together, people use even a single encounter with someone to predict that person’s actions in the future. When that prediction is violated, people remember both what they expected someone would do as well as what that person actually did. It is particularly interesting that these predictions play out as an event unfolds and guide eye movements, which suggests that they really are being used on the fly to direct people’s reactions.
References
Wahlheim, C. N., Eisenberg, M. L., Stawarczyk, D., & Zacks, J. M. (2022). Understanding Everyday Events: Predictive-Looking Errors Drive Memory Updating. Psychological Science, 33(5), 765–781.
