Autism Spectrum Disorder (ASD) is characterized by difficulties in social interaction and communication. However, the underlying psychological mechanisms remain unclear. While some research suggests alexithymia (difficulty identifying emotions) and empathy deficits contribute to these difficulties, these studies often focus on one or two personality measures without considering other covarying factors, leading to incomplete understanding and biased samples. This study aimed to address this by characterizing the position of autistic traits within a comprehensive social-affective personality space and identifying personality dimensions that account for atypical social trait judgments from faces in individuals with ASD. The study leverages a dimensional or 'trans-diagnostic' approach, considering a wider range of personality traits and using validated behavioral and neurophysiological measures to connect self-reported personality to observable behaviors. Specifically, the study examined the relationships between autistic traits and other social-affective personality dimensions (including empathy, anxiety, prosociality, and antisociality) and how these dimensions relate to social trait judgments (warmth, trustworthiness, and competence) derived from facial expressions. This comprehensive approach addresses limitations of previous research by examining multiple personality constructs simultaneously and connecting self-reported personality with social-affective behavior.

Prior research on the social-affective difficulties in ASD has yielded mixed findings. Some studies suggest deficits in emotion recognition, particularly from static facial images, potentially due to atypical attention patterns toward faces. However, the broader realm of social-affective performance extends beyond basic emotion recognition to include rapid judgments of social traits (trustworthiness, warmth, competence) from faces, which have significant implications for social interaction. Studies investigating social trait judgment in ASD have produced inconsistent results, with some reporting abnormalities and others finding normal judgments. This inconsistency highlights the need for a larger, more comprehensive study that considers multiple personality measures to investigate the link between personality, social judgment and neural processing.

This study involved a large online sample of 89 self-identified individuals with ASD and 307 neurotypical controls. Participants completed a battery of 10 social-affective personality questionnaires encompassing 33 subscales covering affective deficits, antisocial traits, the Big Five personality dimensions, and other-oriented and empathic traits. Factor analysis was used to identify underlying latent personality dimensions from these questionnaire subscales. Participants also completed a face judgment task, rating 500 images of celebrities on 10 social traits (warmth, trustworthiness, competence, etc.) using a 7-point Likert scale. Rigorous exclusion criteria were applied to ensure data quality. Additionally, single-neuron recordings from the amygdala and hippocampus of 8 neurosurgical patients (with implanted electrodes) were collected while they viewed the same 500 face images. Representational similarity analysis (RSA) was used to compare the pattern similarity between social trait judgments and neural responses. Further RSA was employed to investigate the relationship between individual differences in social trait judgments and personality dimensions, examining both overall associations and group differences in these associations. Network analyses were also conducted to visualize the relationships between different personality measures and the position of autistic traits within the broader social-affective personality space.

Factor analysis revealed a four-dimensional personality space: 'Autistic trait and social avoidance,' 'Empathy and prosociality,' 'Antisociality,' and 'Social agreeableness.' ASD participants scored significantly higher on the 'Autistic trait and social avoidance' factor but did not differ significantly from controls on the other three factors. Behavioral results showed that ASD participants tended to give marginally higher ratings of trustworthiness, warmth, practical, strong and youthful than controls in the face judgment task. However, RSA of neural data revealed that for trustworthiness, the social trait DM for participants with ASD was less correlated with the neural response DM from the neurosurgical patients compared to controls. This pattern, while less pronounced, was also observed for warmth. Additional regression analysis demonstrated that while the correlation patterns between factor scores and social trait judgments were qualitatively similar across groups, the strengths of correlations differed. Specifically, the association between empathy/prosociality and trustworthiness judgment was stronger in controls than in the ASD group. The association between antisocial traits and trustworthiness judgment was negative in the ASD group but not in controls. Similarly, the association between social agreeableness and warmth judgments was stronger in controls than in the ASD group. Representational similarity analysis further supported these findings, showing that similarity in empathy and prosociality dimensions was more strongly related to similarity in trustworthiness judgments in the control group than in the ASD group, and the opposite pattern for the antisocial dimension. The study also showed that the autistic traits and social avoidance dimension was not associated with social trait judgments, suggesting that social avoidance and anxiety alone do not necessarily lead to these social judgment differences.

This study provides novel insights into the neurocognitive underpinnings of social difficulties in ASD. The findings challenge the notion that ASD is characterized by a fundamental lack of empathy or prosocial tendencies. Instead, the data suggest that atypical associations between social-affective personality and social trait judgments, possibly rooted in distinct neural encoding of social traits, might contribute significantly to the social challenges faced by individuals with ASD. The weaker association between prosocial personality and the perception of trustworthiness and warmth in ASD suggests a disconnect between internal disposition and external social perception. This disconnect might lead to difficulties in initiating and maintaining social interactions, even in individuals who self-report prosocial tendencies. The distinct neural encoding of social traits further supports this hypothesis, highlighting the complex interplay between personality, perception, and neural mechanisms in ASD.

This study integrated neuronal recordings, social trait judgments, and a dimensional approach to personality to shed light on social difficulties in ASD. Autistic traits were strongly linked to social anxiety and emotional understanding difficulties, but were independent of prosociality and empathy. Importantly, altered associations between personality dimensions and social trait judgments, particularly trustworthiness and warmth, were identified in ASD. These findings suggest that the challenges faced by individuals with ASD in social interaction may stem from a disconnect between internal disposition and external social perception, possibly underpinned by distinct neural processing of social cues. Future research should investigate the mechanisms underlying this disconnect and explore interventions targeting social perception and neural encoding.

One limitation is the self-reported nature of the ASD diagnosis. While the online sample’s scores on autism questionnaires were comparable to a smaller group of clinically diagnosed individuals, verification of diagnosis was not possible for all participants. The relatively smaller sample size of the ASD group, although larger than in previous studies, could also limit the generalizability of findings. Future studies should address these limitations by including a larger, clinically diagnosed sample, enhancing the robustness and generalizability of the results. The use of celebrity faces as stimuli could also introduce bias, as prior knowledge might influence judgments. While a control analysis addressing this issue was performed, a more comprehensive assessment of semantic knowledge of the celebrities may improve understanding.