We introduce a nonverbal "visceral" measure of hunger (i.e., squeezing a handheld dynamometer) and provide the first evidence of verbal overshadowing effects in this visceral domain. We presented 106 participants with popcorn and recorded their hunger levels in one of three conditions: (1) first report hunger using a traditional self-report rating scale (i.e., verbal measure) and then indicate hunger by squeezing a dynamometer (i.e., nonverbal measure), (2) first indicate hunger verbally and then indicate hunger nonverbally, or (3) indicate hunger only nonverbally. As hypothesized, nonverbal measures of hunger predicted subsequent eating behavior when they were uncontaminated by verbal measures—either because they preceded verbal measures of hunger or because they were the sole measure of hunger. Moreover, nonverbal measures of hunger were a better predictor of eating behavior than verbal measures. Implications of the study for communicating embodied experiences in a way that escapes the confines of symbolic representations are discussed.