Explaining and simulating judgment biases as an aggregation phenomenon in probabilistic, multiple-cue environments

Many judgment tasks refer to distal entities (e.g., risk, value) that cannot be perceived directly but must be inferred from probabilistic cues. The BIAS model (Brunswikian Induction Algorithm for Social cognition) assumes a noisy stimulus matrix, with columns for stimuli and rows for mediating cues. Aggregation over multiple stimuli (matrix columns) serves to extract systematic information and to cancel error variance. Factors leading to extended aggregation (e.g., discussion, mental elaboration) facilitate the extraction process. The model simulates various judgment biases, including the mere-thinking effect, group polarization, illusory correlations, self-related biases, accentuation, out-group homogeneity, range-frequency effect, and category-split illusions.