Cognitive Fluency Encourages Self-Disclosure

Understanding when people reveal unfavorable information about themselves is both practically and theoretically important. Existing research suggests that people tend not to adopt stable disclosure strategies, and consequently disclose too much information in some situations (e.g., embarrassing personal information on Facebook) and too little in other situations (e.g., risky sexual behavior to a physician during diagnosis of a possible sexually transmitted disease). We sought to identify a domain-general cue that predicts self-disclosure patterns. We found that metacognitive ease, or fluency, promoted greater disclosure, both in tightly controlled lab studies (Studies 1a, 1b, and 3) and in an ecologically valid on-line field study (Study 4). Disfluency tended to prime thoughts and emotions associated with risk, which might be one reason why people who experience disfluency are less comfortable with self-disclosure (Studies 2 and 3). We conclude by discussing the implications of these results for theory and clinical practice. Self-disclosure, ‘‘making the self known to others’’ (Jourard & Lasakow, 1958, p. 91), fosters interpersonal trust (Ensari & Miller, 2002; Turner, Hewstone, & Voci, 2007), dampens anxiety following trauma (e.g., Greenberg & Stone, 1992a, 1992b; Pennebaker, Kiecolt-Glaser, & Glaser, 1988), enhances the quality of social relationships (e.g., Collins & Miller, 1994), and often improves negotiation outcomes (e.g., Fisher, Ury, & Patton, 1991). Despite these benefits, self-disclosure is also inherently risky. People cannot reveal themselves to others without first trading the protection of privacy for enhanced scrutiny and potential criticism (e.g., Jourard, 1971). Indeed, resistance to disclosure is so strong that a majority of high-risk HIV patients fail to disclose enough information to their doctors to facilitate accurate diagnoses (Epstein et al., 1998). Self-disclosure also poses modern financial risks associated with identity theft, privacy breaches, and Internet fraud (Whitty & Joinson, 2009). Previous research has shown that people do not have stable disclosure strategies, and base their decision of whether or not to disclose information on transient environmental cues (John, Acquisti, & Loewenstein, 2009). Much of the work on disclosure elicitation has focused on how people respond to specific trustbased cues. For example, studies have shown that people are more likely to disclose information when the venue they are in seems private (Joinson & Paine, 2007), if their interaction partner has established a record of trustworthiness (Andrade, Kaltcheva, & Weitz, 2002), if prevailing laws effectively encourage disclosure (Galletly & Pinkerton, 2006), and if the costs of disclosure are relatively low (Paine, Joinson, Buchanan, & Reips, 2006). Although these cues seem useful in identifying safe settings, people also respond to cues that are potentially misleading. The same questionnaire administered on-line rather than in paper-and-pencil format tends to elicit more honesty and disclosure, and less socially desirable responding (Tourangeau, 2004). People are similarly more inclined to divulge information using e-mail or on-line instant messages, or while blogging, than when they communicate face-to-face (e.g., Harper & Harper, 2006; Whitty & Joinson, 2009). These findings are important because they suggest that people are particularly willing to selfdisclose in potentially dangerous settings—those that contain numerous anonymous viewers whose motives are difficult to ascertain. Although the literature describes specific social and situational factors that lead to disclosure, little work has explored the domain-general cognitive factors that lead people to divulge information. One potential candidate is fluency—the metacognitive experience of ease or difficulty associated with processing information. Fluency is a ubiquitous metacognitive cue; all cognitive tasks vary by how easy it is to process the relevant information (e.g., stimuli can be difficult to perceive, spaces can Address correspondence to Adam L. Alter, Marketing Department, Stern School of Business, Tisch Hall, 40 West 4th St., New York, NY 10012, e-mail: [email protected]. PSYCHOLOGICAL SCIENCE 1414 Volume 20—Number 11 Copyright r 2009 Association for Psychological Science be difficult to navigate, text can be difficult to process linguistically), and people therefore have access to fluency cues when making decisions across a broad range of situations (for a review, see Alter & Oppenheimer, 2009). Indeed, fluency influences a vast array of cognitive processes, including truth assessment (e.g., McGlone & Tofighbakhsh, 2000), stock purchasing decisions (Alter & Oppenheimer, 2006), currency valuation (Alter & Oppenheimer, 2008a), appraisal of psychological distance (Alter & Oppenheimer, 2008b), and judgments of confidence (e.g., Simmons & Nelson, 2006). Although researchers have not yet examined the effects of fluency on self-disclosure, there are good reasons to expect that disfluency discourages self-disclosure. Considerable evidence suggests, for example, that disfluency functions as an alarm that signals the need for greater scrutiny and deliberation before acting (Alter, Oppenheimer, Epley, & Eyre, 2007). Moreover, recent research has shown that people consider stimuli that are processed disfluently to be riskier than stimuli that are processed fluently (Song & Schwarz, 2009), which suggests that people should engage in less self-disclosure in disfluent contexts than in fluent contexts. Because of the ubiquity of fluency as a metacognitive cue, understanding how fluency influences self-disclosure is critical. This understanding will allow policymakers, researchers, and legislators to engineer situations that effectively discourage or encourage disclosure. The current studies extend the theoretical focus of fluency research from cognitive processing to an interpersonal, social-psychological domain. We conducted three lab studies and one on-line field study to examine the effect of fluency on self-disclosure, and to determine whether this effect might emerge in part because disfluency prompts thoughts and emotions associated with risk and concern. STUDY 1A: ADMITTING FLAWS, PART I In the first study, participants completed a social-desirability scale, which measured their willingness to disclose socially undesirable thoughts and behaviors. To examine the effects of fluency on self-disclosure, we presented the scale to participants in either a clear font or a difficult-to-read font.