Eye Tracking Caught the Knowledge Experts Don't Know They Have

Experts are famously bad at explaining how they do what they do. A master glassblower can demonstrate the exact pressure to apply to molten glass, but ask them to put it in words and the description will be frustratingly vague. A radiologist can spot a suspicious shadow on an X-ray that a resident walks right past, but articulating the visual reasoning often reduces to "it just looks wrong." This kind of embodied expertise - what philosopher Michael Polanyi called tacit knowledge - has long been considered the one form of human skill most resistant to transfer.

A team at MIT has found a way to pull some of it out.

What the eyes give away

In a study published in the Journal of Neural Engineering, researchers tasked 30 volunteers with classifying a series of images into one of two groups. Each image showed two shapes on opposite sides - squares, triangles, circles, in various colors and patterns - but only one side was actually relevant to the correct classification. The volunteers were given no rules or guidance. They had to figure it out through trial and error over more than 120 images.

While they worked, the team tracked two channels of data: where participants' eyes went (using cameras), and where their cognitive attention landed (using EEG sensors tuned to detect brain waves synchronized with the flickering of each shape). Together, these measurements produced maps of each participant's visual and cognitive focus at every stage of the experiment.

The maps told a clear story. Early in the experiment, when participants were essentially guessing, their attention spread across both sides of each image. As their accuracy improved, their focus converged - without their noticing it - onto the side that was actually informative. They had learned something, but they had not learned it consciously.

"They were unconsciously focusing their attention on the part of the image that was actually informative," said Alex Armengol-Urpi, a research scientist in MIT's Department of Mechanical Engineering and the study's lead author. "So the tacit knowledge they had was hidden inside them."

Making the invisible visible

When the researchers asked participants to describe how they approached the task, the volunteers consistently said they tried to look at the whole image. They were not lying - they genuinely did not know their attention had shifted. The tacit knowledge was inaccessible to introspection.

The researchers then showed each participant the maps of their own gaze and attention, side-by-side comparisons of their early novice phase and their later expert phase. When participants were then given additional images to classify, their accuracy improved beyond the expert level they had already reached.

This finding is the key one. It is not merely that tacit knowledge exists and can be detected - that had been suggested before. It is that making it explicit, by showing experts a visual record of their own unconscious behavior, enhances their subsequent performance. The knowledge, once surfaced, becomes usable in a different and more effective way.

Where this might actually matter

Armengol-Urpi is quick to note that the study involved a controlled, abstract visual task. Whether the same principle generalizes to more complex real-world expertise is a research question, not an established fact. But the underlying logic is suggestive: any domain where skill involves learned perceptual focus - where experts have learned to look at the right thing, even if they cannot say what that thing is - might be amenable to this kind of gaze-map feedback.

The team is currently extending the approach to glassblowing, table tennis, and medical imaging interpretation. All three involve long training periods, high performance variance between novices and experts, and the kind of embodied perceptual skill that resists verbal instruction.

The potential application in medical training is particularly significant. Radiology, pathology, and dermatology all require practitioners to notice subtle visual features that experienced clinicians have learned to see and novices miss. If expert gaze maps could be extracted and shown to trainees as part of their education, it might compress the training time currently required to reach competence.

"We as humans have a lot of knowledge, some that is explicit that we can translate into books, encyclopedias, manuals, equations. The tacit knowledge is what we cannot verbalize, that's hidden in our unconscious," Armengol-Urpi said. "If we can make that knowledge explicit, we can then allow for it to be transferred easier, which can help in education and learning in general."

The research was supported in part by Takeda Pharmaceutical Company.