Spontaneous trait inferences (STIs) are ubiquitous and occur when perceivers spontaneously infer actor traits from actor behaviors. Previous research has elucidated the processes underlying STIs, but little work has focused on the functions of STIs. This article proposes that these unintentional early inferences serve a general approach or avoidance function. Two studies are reported in which external approach and avoidance motivations elicited via flexion–extension (Study 1) or physical warmth (Study 2) affect the encoding of trait-implying behavioral statements in a valence-matching manner. The results suggest that somatic states can act as cues that affect unintentional social information processing independently of the actual experience of the psychological states associated with those somatic states. Implications for a functional perspective on STIs are discussed.