Eye-tracking research explores earlier recognition of autistic characteristics in infants

A growing body of research suggests that differences in how babies view social scenes — measurable through eye-tracking technology — may help recognize autistic developmental patterns as early as 12 months old. Three recent peer-reviewed studies highlight both the potential and current limitations of these objective measures for earlier identification.

Efficient assessment approaches

The brief test, which observes where infants focus when viewing social scenes, recognized 78% of autistic children in the study while excluding 85% of non-autistic peers.

A March 2024 study in Nature found that a 2-minute eye-tracking protocol called Gazefinder could identify distinct gaze patterns in 12-month-olds who were later confirmed to be autistic. The brief test, which observes where infants focus when viewing social scenes, recognized 78% of autistic children in the study while excluding 85% of non-autistic peers.

An April 2024 JAMA Network Open study analyzed eye-tracking data from 350 toddlers in primary care settings. Researchers found that specific gaze patterns — including varied attention to eyes and different pacing in shifting between social and non-social stimuli — showed correlation with later autism confirmation by age 3.

Consistent research patterns

These studies align with a 2024 meta-analysis in ScienceDirect reviewing 35 earlier papers. The analysis confirmed that infants who later receive autism confirmation consistently show measurable differences in social attention, particularly in how they observe faces and distribute gaze across complex scenes.

Ongoing research at institutions like UCSD and through the SPARK study is exploring whether eye-tracking markers might be observable before 12 months. Preliminary results suggest some infants show distinct gaze patterns by 6-9 months, though individual outcomes remain difficult to predict.

While not conclusive by themselves, these tools might eventually complement existing screening methods like the M-CHAT questionnaire. 'What's promising is the potential for objective measures that don't depend entirely on parent reports or clinician observations,' noted one researcher involved in the JAMA study, referencing current median evaluation ages of 4-5 years in the U.S.