Dangerous Models: When Constructed Realities Dehumanize

Human cognition does not passively mirror reality; it actively constructs models that organize perception, guide action, and shape moral judgment. This paper examines how such constructive processes can become pathological when they generate dehumanizing representations of others. Drawing on cognitive constructivism, social psychology, neuroscience, and social epistemology, it argues that dehumanization is best understood as a breakdown in reciprocal modelhood: the failure to recognize others as co-equal centers of agency and perspective. The analysis distinguishes adaptive cognitive simplification—an unavoidable feature of categorization—from rigid, self-sealing models that resist revision, suppress feedback, and erase interpersonal symmetry. Dehumanizing constructions are shown to display characteristic markers of model pathology, including inflexibility, closed feedback loops, loss of self–other reciprocity, and collective reinforcement through echo chambers and memetic amplification. These dynamics help explain how constructed realities can legitimize cruelty while remaining internally coherent to their adherents. Against relativistic interpretations of constructivism, the paper develops a form of constraint realism: although all knowledge is constructed, it remains bounded by shared biological capacities, empirical feedback, and interdependence. These constraints provide normative grounds for distinguishing viable from pathological models. The paper concludes by proposing an ethic of epistemic responsibility centered on maintaining revisability, anomaly-sensitivity, and recognition of others as fellow model-builders. In doing so, it positions cognitive constructivism not merely as a theory of knowledge, but as a framework for diagnosing and counteracting one of the central moral failures of modern societies: the systematic construction of realities that deny human dignity.