10.1177/0146167203259934 ARTICLEPERSONALITY AND SOCIAL PSYCHOLOGY BULLETINTamir et al. / MEANING OF UNCONSCIOUS CUESAre We Puppets on a String? The ContextualMeaning of Unconscious Expressive CuesMaya TamirUniversity of Illinois at Urbana-ChampaignMichael D. RobinsonNorth Dakota State UniversityGerald L. CloreUniversity of VirginiaLeonard L. MartinDaniel J. WhitakerUniversity of GeorgiaIn three studies, the authors show that unconscious expressivecues can lead to opposite evaluations, depending on the contextin which they occur. In Study 1, brow (vs. cheek) tension reducedpreferences in an easy judgment context but increased preferencesin a difficult context. In Study 2, head shaking (vs. nodding)either increased or decreased prosocial affect depending on thecontext in which the judged character was presented. In Study 3,a subliminal smile (vs. frown) led to higher self-ratings of perfor-mance when paired with one’s own actions but to lower self-ratings of performance when paired with a competitor’s actions.Together, these results suggest that the meaning of unconsciousexpressive cues is not fixed.Keywords: nonconscious processing; affect; evaluation; bodilyexpressionsThe informational value of expressive cues hasintrigued scholars throughout the history of psychology.James (1890) regarded the self-perception of one’s ownfacial expression (e.g., one’s own smile) as an importantcue to one’s subjective experience. Relatedly, Darwin(1872/1965) was concerned with the communicationvalue of another organism’s facial expression (e.g.,someone else’s smile) on behavior and emotional expe-rience. Modern research demonstrates the effects ofboth types of cues on evaluative outcomes.Research on self-produced facial actions demon-strates that such cues tend to elicit congruent affectiveresponses. For example, imitating a smile leads to posi-tive affect (Laird, 1974); furrowing the brow leads to feel-ings of effort and difficulty (Stepper & Strack, 1993); andhead nodding, in comparison to head shaking, elicitspositive reactions to persuasion attempts (Wells & Petty,1980). Similarly, research on affective priming (Murphy& Zajonc, 1993), facial mimicry (Dimberg, Thunberg, &Elmehed, 2000), and aversive learning (Öhman &Dimberg, 1978) demonstrates that external stimuli suchas subliminal smiles versus frowns tend to elicit congru-ent affective responses. Such effects can occur uncon-sciously, either by unobtrusive manipulation in the caseof self-produced expressions or by subliminalpresentation in the case of external stimuli.THE MEANINGS OF EXPRESSIVE CUESWhereas most research has focused on the relation-ship between the expressive stimulus and an evaluativeoutcome, the current studies explore the nature of theunderlying process. Our main goal was to differentiatebetween processing stages that might be involved intranslating perceptions into subjective reactions.237Authors’ Note: The research in this article was supported by grantfunding from the National Science Foundation (SBR 98-1764) to Rob-inson and the National Institute of Mental Health (MH50074) toClore. Weareindebtedto Adam Joncich for help with Study 2 and Stan-ley Colcombe for help with Study 3. Correspondence concerning thisarticle can be directed to Maya Tamir, Psychology Department, Univer-sity of Illinois, 603 E. Daniel St., Champaign, IL 61820; e-mail: [email protected], Vol. 30 No. 2, February 2004 237-249DOI: 10.1177/0146167203259934© 2004 by the Society for Personality and Social Psychology, Inc.Many existing models of information processing dis-tinguish between at least two broad levels of meaningassignment (e.g., Teasdale, 1993). One level involves theperceptual identification and categorization of the stim-ulus object; for example, a piece of metal of a certain sizewith serrated edges will be classified as a knife. Anotherlevel involves determining the evaluative implications ofthe given object. A knife brandished by a stranger in adark alley has very different implications than a knifeplaced beside one’s dinner plate. Thus, the perception-related meaning of an object (e.g., as a knife) must bedistinguished from the subjective meaning of the object.The present investigation in part seeks to dissociatethese two types of meaning in understanding the impactof unconscious expressive cues on feelings.One approach to unconscious expressive cues, whichwe refer to as the Default Implications Model, suggeststhat the link between perception and evaluative implica-tions is fixed. According to this model, perceptual inputsare directly translated into corresponding evaluativeimplications (for a related view, see Dijksterhuis &Bargh, 2001). For example, a subliminal smile necessar-ily has positive evaluative implications. Whereas theinfluence of expressive cues on evaluative implications isfixed, the model accords some flexibility in potentiallysuppressing these implications (e.g., because one has apreexisting evaluation of the stimulus) (Priester,Cacioppo, & Petty, 1996). Within this model, assimila-tion effects are the norm, contrast effects are unlikely,and suppression is an effortful process. Of additionalnote, this model essentially equates the perceptualmeaning of a stimulus (e.g., a smile) with its evaluativeimplications (e.g., good).A different approach to the link between unconsciousexpressive cues and evaluative implications, which werefer to as the Contextual Implications Model, differen-tiates the perceptual meaning of a stimulus from itsevaluative implications. Whereas a subliminal smile maybe a positive cue in terms of its perceptual meaning, theevaluative implications of this perceptual cue can differdepending on context. For example, a subliminal smilepaired with explicit erotic material might engenderembarrassment, arguably a negative (rather than posi-tive) subjective experience. Within this model, then,there is no direct translation from perceptual meaningto evaluative implications; instead, the translation isinferential, flexible, and contextual.The model is somewhat consistent with recent modelsconcerning the effects of mood states on evaluative judg-ment. Martin’s (2000, 2001) Mood-as-Input model andGendolla’s (2000) Mood-Behavior model both arguethat mood states carry no direct implications for behav-ior. Rather, their implications for behavior depend onone’s current goals as well as other contextual features.In one study, for example, Martin, Ward, Achee, andWyer (1993) found