Back

Heightened gaze responsiveness and mutual gaze instability at close distances characterize dyadic coordination in a marmoset model of autism

Nakamura, M.; Sato, T.; Kurotani, T.; Sakai, K.; Hadjiev, A. M.; Sasai, S.; Mimura, K.; Kawai, N.; Shimmura, T.; Ichinohe, N.

2025-12-09 animal behavior and cognition
10.1101/2025.11.17.688777 bioRxiv
Show abstract

Autism spectrum disorder (ASD) is associated with differences in social interaction, including alterations in gaze behavior and interpersonal distance. Although most assessments of ASD remain individual-centric, emerging human studies indicate that social difficulties may arise from atypical dyadic coordination rather than from reduced social interest alone. We examined dyadic coordination in a non-human primate model of ASD, focusing on gaze and inter-individual distance during minimally constrained first encounters between prenatal valproic acid-exposed marmosets (VPA) and unexposed conspecifics (UE). Markerless pose estimation quantified frame-by-frame gaze direction and spacing, comparing VPA-UE with UE-UE dyads. VPA-UE pairs exhibited a higher proportion of mutual-gaze events than UE-UE pairs. VPA individuals showed more rapid looks in response to the partners gaze, yet spent less time looking at the partner when unobserved. VPA-UE dyads maintained shorter inter-individual distances than UE-UE dyads, and closer spacing was associated with decreased stability of mutual gaze. These findings reveal a dyadic coordination pattern in VPA-UE pairs, with rapid responsiveness coupled with distance-dependent fragility, paralleling features reported in human ASD. Our dyad-centric, markerless framework yields ecologically grounded readouts and potential intervention targets that leverage intact social responsiveness while accommodating altered spatial coordination.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.