A false recognition paradigm showed that spontaneous trait inferences are bound to the person performing a trait-implying behavior. In 5 experiments, participants memorized faces and behavioral sentences. When faces were paired with implied traits in a recognition test, participants falsely recognized these traits more often than a) unrelated traits paired with the same faces; b) the same traits paired with familiar faces or; c) paired with faces initially presented simultaneously with actors faces. Effect sizes (r) across experiments were .67 at the level of participants and .75 at the level of stimuli. Antonyms of the implied traits were less often falsely recognized than unrelated traits, suggesting that participants went beyond the initial trait inference. Participants also took longer to correctly reject implied traits. Explicit person-trait judgments accounted for 65% of the false recognition variance, which accounted in turn for 28% of the response time variance, completely mediating the relationship between explicit judgments and response times.