The authors apply an embodied account to mere exposure, arguing that through the repeated exposure of a particular stimulus, motor responses specifically associated to that stimulus are repeatedly simulated, thus trained, and become increasingly fluent. This increased fluency drives preferences for repeated stimuli. This hypothesis was tested by blocking stimulus-specific motor simulations during repeated exposure. In Experiment 1, chewing gum while evaluating stimuli destroyed mere exposure effects (MEEs) for words but not for visual characters. However, concurrently kneading a ball left both MEEs unaffected. In Experiment 2, concurrently whispering an unrelated word destroyed MEEs for words but not for characters, even when implemented either exclusively during the initial presentation or during the test phase and when the first presentation involved an evaluation or a mere study of the stimuli. In Experiment 3, a double dissociation between 2 classes of stimuli was demonstrated, namely, words (oral) and tunes (vocal). A concurrent oral task (tongue movements) destroyed MEEs for words but not for tone sequences. A concurrent vocal task (humming "mm-hm") destroyed MEEs for tone sequences but not for words. (Contains 1 figure and 3 tables.)