Autistic Individuals Do Not Alter Visual Processing Strategy During Encoding Versus Recognition of Faces: A Hidden Markov Modeling Approach.

Researchers watched where autistic and neurotypical teens looked while they learned new faces and later picked them out of a lineup.

The team used hidden Markov models to track tiny changes in eye movement patterns between the learning and testing phases.

Typical teens switched to a wider, more exploratory scan during recognition.

Autistic teens kept the exact same narrow scan path for both stages.

Their eyes did not adjust when the task changed, showing less flexible face processing.

Williams et al. (2002) first showed autistic adults scatter their gaze away from eyes, nose, and mouth. The new study adds that this scatter stays locked even when the job changes.

Goulardins et al. (2013) linked weak gaze-ERP coupling to poor emotion reading in autistic teens. Griffin et al. (2026) now shows the gaze side of that link is rigid across learning and recall.

Benson et al. (2016) found autistic adults only slow down later when spotting social oddities. The teen data look opposite at first—no change at all—but the tasks differ: one scans real scenes, the other scans single faces. Age and task type likely explain the gap, not a true clash.

When you run face-emotion drills, don’t assume the learner will naturally shift their looking strategy between study and test. Build in explicit cues like “check the eyes again” or use guided video that replays where to look just before the recall trial. A quick prompt can replace the missing self-adjustment and speed up social-skills gains.