Research indicates that impaired empathy is a risk factor of aggression and that social attention is important for empathy. The role of social attention in associations between empathy and aggression has not yet been fully elucidated. Therefore, indicators of affective empathy, cognitive empathy, social attention, and aggression were simultaneously assessed in children aged 45 months. A total of 61 mother–child dyads participated in a lab visit, during which maternal reports of aggression were obtained. Children watched three clips showing a sad, scared, and happy child, respectively, and a neutral social clip while heart rate was recorded. Heart rate change from nonsocial baseline clips to emotional clips was calculated as an index of affective empathy. Questions about the emotions of the children in the clips were asked to assess cognitive empathy. Social attention was defined as time spent looking at faces during the clips. Correlation analyses revealed negative associations between affective empathy and aggression and between social attention and aggression. Furthermore, multivariate linear regression analyses indicated that the association between affective empathy and aggression was moderated by social attention; the negative association between affective empathy and aggression was stronger in children with relatively reduced social attention. No association was found between cognitive empathy and aggression. Therefore, both affective empathy and social attention are important targets for early interventions that aim to prevent or reduce aggression.

Introduction
The stare-in-the-crowd (SITC) effect describes the ability to detect self-directed gaze in a crowd. Given the importance of gaze detection in initiating and maintaining social interactions, there is a need to better characterize the SITC effect.
Methods
Autistic and neurotypical young adults were presented with four SITC conditions. Eye tracking outcomes and arousal were compared by diagnosis and condition using repeated measures analysis of variance. Hierarchical regression was used to explore behavioral measures.
Results
Significant interaction of diagnosis and condition was found for eye tracking outcomes. Overall, autistic participants exhibited less looking than neurotypical participants. Interest area dwell time, fixation count, and second fixation duration were significantly higher for conditions with shifting gaze, as well as conditions with more self-directed gaze across participants. Two hierarchical regression models of gaze behaviors with advanced theory of mind as a predictor were significant.
Discussion
Autistic individuals respond to various gaze conditions in similar patterns to neurotypical individuals, but to a lesser extent. These findings offer important targets for social interventions.