Prosocial behavior is pivotal to our society. Guilt aversion, which describes the tendency to reduce the discrepancy between a partner’s expectation and his/her actual outcome, drives human prosocial behavior as does well-known inequity aversion. Although women are reported to be more inequity averse than men, gender differences in guilt aversion remain unexplored. Here, we conducted a functional magnetic resonance imaging (fMRI) study ( n = 52) and a large-scale online behavioral study ( n = 4723) of a trust game designed to investigate guilt and inequity aversions. The fMRI study demonstrated that men exhibited stronger guilt aversion and recruited right dorsolateral prefrontal cortex (DLPFC)-ventromedial PFC (VMPFC) connectivity more for guilt aversion than women, while VMPFC-dorsal medial PFC (DMPFC) connectivity was commonly used in both genders. Furthermore, our regression analysis of the online behavioral data collected with Big Five and demographic factors replicated the gender differences and revealed that Big Five Conscientiousness (rule-based decision) correlated with guilt aversion only in men, but Agreeableness (empathetic consideration) correlated with guilt aversion in both genders. Thus, this study suggests that gender differences in prosocial behavior are heterogeneous depending on underlying motives in the brain and that the consideration of social norms plays a key role in the stronger guilt aversion in men.

The Autism-Spectrum Quotient (AQ) is a widely used index to quantify autistic traits. Previous studies using clustering analysis have indicated that people showing high AQ and other autism indices may comprise several subtypes.  However, those studies mainly focused on Big5 personality traits and analyzed a limited number of samples (participants). The inclusion of a larger set of personality traits and attributes would contribute not only to understanding autism more deeply, but also to characterizing AQ more precisely. In addition, an analysis of larger general populations would give new perspectives on people with high AQ scores. To address these issues, here we conducted a personality traits-based clustering of 1364 subjects who had an AQ score of 33 or higher (high AQ group) collected online. We identified three subclasses (subtypes): Group 1, characterized by emotional instability, low self-esteem, high hostility, anxiety, depressive tendency and schizotypal traits; Group 2, characterized by high BIS for an inhibitory attitude towards novelty, and high Interpersonal Reactivity Index (IRI) and Group 3, characterized by personality traits and attributes consistent of the average of the general population. Thus, this study provides empirical results showing subtypes of the high AQ population.