Blame through my eyes: How first-person perspective shapes social attention and attribution in harmful interactions -...

Rigid attributional patterns are a key component of social cognitive deficits linked to aggression. Prior eye-tracking research using a third-person perspective revealed that, compared to community adults, incarcerated violent offenders oriented less to faces and showed a stronger correlation between perceived intentionality and assigned blame (Intentionality/Blame isomorphism), a metric of attributional inflexibility. The present study investigated how adopting an immersive first-person victim perspective affects both I/B isomorphism and facial attention.

A total of 184 adults participated in our study, including 119 community members (60 female, 59 male) and 65 incarcerated violent offenders (32 female, 33 male). During the eye-tracking task, participants viewed 27 static visual scenes depicting ambiguous interpersonal harm between two actors - a harm doer and a victim - and were instructed to imagine themselves as the victim in each scenario. The scenes included both hostile (e.g., a harmful act) and non-hostile (e.g., mitigating expressions or gestures) cues. Independent samples t-tests were used to compare group differences in I/B isomorphism and face dwell time, and Pearson's correlations were used to examine the relationships between variables.

Consistent with our hypothesis and contrary to findings from third-person studies, incarcerated violent offenders exhibited significantly lower I/B isomorphism than community adults under a first-person perspective. Furthermore, no significant differences in face dwell time were observed between these two groups.

These results suggest that, among incarcerated violent offenders, adopting a victim's perspective may disrupt rigid attributional patterns and normalize attention to social cues. These findings suggest that perspective-taking interventions could support improvements in empathy, cognitive flexibility, and overall interpersonal functioning.