Adopting a situated social cognition perspective, we relied on different methodologies—1 computational and 3 empirical studies—to investigate social group–related specificities pertaining to implicit gender-domain stereotypes, as measured by a mouse-tracking adapted Implicit Association Test (IAT) and IAT(-like) tasks. We tested whether the emergence of implicit stereotypes was partially determined by associations congruent with the self, by visuospatial features of the task and subsequent competition at both sensorimotor and abstract levels. We tracked human and simulated artificial participants’ hand movements among gender stereotypical (e.g., male engineers) and counterstereotypical (e.g., female engineers) social groups. In the computational study, data were simulated by a novel generative connectionist model integrating strengths from recent developments in embodied models of decision-making. Results support the self-congruency hypothesis and suggest the presence of competition at both levels. Discussion focuses on the generalizability of the self-congruency hypothesis and on the relevance of a situated perspective for implicit social cognition.