Have you ever heard the phrase ‘the eyes are a window to the soul’? New research from Dr. Robert Sekuler’s Vision Lab suggests that the eyes may be a window to the brain as well. In an article published in this month’s issue of the Journal of Vision, Neuroscience grad student Jessica Maryott (PhD ’09) and Psychology grad student Abigail Noyce showed that as participants learn, their eye movements change in a way that lets scientists investigate how that learning takes place, specifically in response to unexpected events.

Participants in the study watched as a disk moved on a computer screen in a zig-zag path; they then reproduced its trajectory from memory. Each path was repeated several times, allowing the researchers to examine the learning process as participants became familiar with the pattern and more accurate at reproducing it. Researchers also measured participants’ eye movements as they watched the disk move, and examined learning-related changes in those as well. The results suggest that eye movements reflect the participant’s level of learning by actually predicting where the disk will be going next.

Sometimes, part of the disk’s path changed after several repetitions going in the opposite direction (a 180 degree change, shown in the green trace on the figure), without warning to the participant. This caused participants to make a prediction error: the actual motion of the disk no longer matched the pattern they had learned, but their eyes moved in the direction of the expected movement (positive velocity) until they were able to correct the error (this is when the green trace reverses velocity and goes below 0 in the figure). After such a prediction error, when the pattern appeared again, participants’ eye movements showed that the previous prediction error produced fast ‘one-shot’ learning, and participants now expected to see the new version of the path (shown in the blue trace, which goes in the new expected direction, 180 degrees from the old – thus showing a negative velocity). The researchers concluded that unexpected events (like the induced error) have high salience for learning. These results suggest that humans have a cognitive system which monitors how well sensory input matches predictions, and responds to errors with sudden, strong learning about the new situation.