Human Nature

, Volume 26, Issue 1, pp 28–43

Laughter’s Influence on the Intimacy of Self-Disclosure


    • Institute of Cognitive NeuroscienceUniversity College London
  • Brian Parkinson
    • Department of Experimental PsychologyUniversity of Oxford
  • Robin I. Dunbar
    • Department of Experimental PsychologyUniversity of Oxford

DOI: 10.1007/s12110-015-9225-8

Cite this article as:
Gray, A.W., Parkinson, B. & Dunbar, R.I. Hum Nat (2015) 26: 28. doi:10.1007/s12110-015-9225-8


If laughter functions to build relationships between individuals, as current theory suggests, laughter should be linked to interpersonal behaviors that have been shown to be critical to relationship development. Given the importance of disclosing behaviors in facilitating the development of intense social bonds, it is possible that the act of laughing may temporarily influence the laugher’s willingness to disclose personal information. We tested this hypothesis experimentally by comparing the characteristics of self-disclosing statements produced by those who had previously watched one of three video clips that differed in the extent to which they elicited laughter and positive affect. The results show that disclosure intimacy is significantly higher after laughter than in the control condition, suggesting that this effect may be due, at least in part, to laughter itself and not simply to a change in positive affect. However, the disclosure intimacy effect was only found for observers’ ratings of participants’ disclosures and was absent in the participants’ own ratings. We suggest that laughter increases people’s willingness to disclose, but that they may not necessarily be aware that it is doing so.


LaughterSelf-disclosureSocial relationshipsEndorphinsSocial bonding

Self-disclosure can be defined as verbally communicating personal information about oneself to another person. It may range from divulging highly sensitive information (such as personal fears and deeply held religious convictions) to exposing relatively superficial facts about oneself (such as one’s favorite food or where one lives). Self-disclosure has long been regarded as critical to relationship development and is typically considered as an exchange, where intimacies are traded as a means of deepening and developing relationships (Collins and Miller 1994; Derlega and Berg 1993). Indeed, people tend to like those to whom they disclose as well as those who disclose to them (Sprecher et al. 2013), and disclosure intimacy typically increases as relationships develop (for reviews, see Collins and Miller 1994; Greene et al. 2006).

There are many ways in which disclosure may impact on liking in the initial stages of the acquaintance process when the exact nature of the future relationship is still ambiguous (Collins and Miller 1994; Cozby 1973). For the recipient, receiving disclosure from another can inform them about the discloser’s character in ways that would otherwise have been off-limits, reducing uncertainty about his or her disposition and consequently increasing feelings of liking the other by virtue of this new-found familiarity (Berger and Calabrese 1975; Berscheid and Reis 1998), thereby improving the effectiveness and efficiency of the relationship. For the discloser, the process of self-disclosing can be intrinsically gratifying (activating reward regions of the brain: Tamir and Mitchell 2012), and these feelings may become associated with the target of one’s disclosure, generating affection toward the disclosed (Jourard 1971; Pennebaker 1985). Disclosers may also infer liking toward the disclosed by observing their own disclosures: one likes those to whom one discloses, and may likewise infer that one is liked when one is disclosed to (Bem 1972; Chaikin and Derlega 1974).

Although disclosing intimate information may be highly rewarding, self-disclosure needs to be carefully managed because too much disclosure may embarrass or hurt others and lead to rejection or social exploitation (DePaulo et al. 1996). As a result, there is always a tension between the need to disclose to others in order to build social bonds and the need to refrain from doing so in order to avoid negative social consequences. Knowing what, how much, when, and to whom one wishes to disclose can be exceedingly complex (Collins and Miller 1994; Cozby 1973; Derlega and Berg 1993; Greene et al. 2006). However, evolution may have built in various heuristics that prompt one to disclose under conditions that on average facilitate relationship development.

A recent study provides putative evidence for such a mechanism. Forgas (2011) presented participants with a video clip either from Fawlty Towers (a British comedy program) or from a film that discussed death from cancer (My Life) in order to induce happiness or sadness (respectively). Following this, he instructed participants to imagine that they had met someone whom they found attractive and to write down five pieces of information they would like to tell this “other” about themselves. After having rated the disclosures for intimacy, abstractness (disclosing about concrete behaviors vs. abstract traits), and valence (positivity versus negativity of content disclosed), Forgas found that those who watched the comedy clip disclosed more intimate, more positive, and more abstract information about themselves (Forgas 2011). Forgas interpreted the difference in disclosure between the two conditions as the result of a change in affect induced by the videos. However, it is possible that the change in disclosure was not a product of affect per se, but a result of different laughter rates in response to the two clips, and the way laughter modulates relaxation via the endorphin mechanism. In other words, Forgas’s experimental design may have confounded affect with laughter. Although heightened positive affect is known to increase willingness to take risks (Trope et al. 2001) and may itself be important in social bonding (Spoor and Kelly 2004), we need to determine whether laughter (acting through the endorphin mechanism) influences self-disclosure rates independently of mood change.

Although a number of (not necessarily mutually exclusive) hypotheses have been suggested for the function of laughter (signaling social or mating interest: Grammer 1990; Grammer and EibI-Eibesfeldt 1990; Li et al. 2009; Martin and Gray 1996; Mehu and Dunbar 2008; emotional contagion: Bachorowski and Owren 2001; Owren and Bachorowski 2003; social bonding: Dunbar 2004; Dunbar et al. 2012; Provine 2001), laughter in humans is characteristically highly social and intensely contagious (Provine 2001). Laughter is 30 times more likely to occur when in company than when alone, and it occurs more frequently with friends than with strangers (Devereux and Ginsburg 2001; Provine 2001). In conversation, laughter certainly signals affiliation and agreement with the speaker (Vettin and Todt 2004), and in humor, people tend to laugh at jokes that align with their own implicit preferences (Curry and Dunbar 2013; Flamson and Barrett 2008; Lynch 2010). On the whole, the conditions that seem most conducive to laughter are those in which disclosure is least likely to result in negative social consequences and most likely to facilitate relationship development. Accordingly, laughter may serve as a cue to the presence of a potentially beneficial relationship partner (Flamson and Barrett 2008; Provine 2001) and, with the advent of language, may have come to signal the adaptive conditions for self-disclosure.

The act of laughing has been shown to increase endorphin release in much the same way that grooming does in nonhuman primates (Dunbar et al. 2012). This had led to the suggestion that laughter may have evolved not just as a signal to potential partners, but as a mechanism that enables humans to create social bonds with more individuals than primates can do by social grooming alone (Curley and Keverne 2005; Dunbar 2010, 2012; Keverne et al. 1989). In this view, laughter is conceived as a form of “grooming at a distance”; it exploits the same psychopharmacological mechanism (the release of endorphins) as social grooming does in primates, but it does so in such as a way as to permit at least a threefold increase in the number of individuals that can be affected by it at any one time (the one-to-one of grooming vs. the observed size of natural laughter groups: Dezecache and Dunbar 2012). Thus, laughter, more than just signaling the appropriate conditions for disclosure, may actually encourage disclosure through the physiological relaxation induced by the opioid response of the endorphin system. Just as primate grooming may alleviate the tension of interacting with another individual (relaxing the recipient of the grooming via endorphin release and hence encouraging the exchange of support: Dunbar 2010; Goosen 1981; Schino et al. 1988), the act of laughing may assuage the tension in humans through the same psychopharmacological mechanisms (Dunbar et al. 2012). Endogenous opioid activation has been shown to have a role in regulating fear (Chaijale et al. 2013; Ipser et al. 2013; Resendez and Aragona 2012), and laughter is known to have profound effects on stress reduction (Keltner and Bonanno 1997; Mehu 2011).

Accordingly, the present study directly compares positive affect inductions with and without laughter in order to determine whether laughter mediates any effects on disclosure independently of any changes in affect. We compared the characteristics (intimacy, abstractness, and valence) of self-disclosing statements produced by those who had previously watched one of three video clips that differed in the extent to which they elicited laughter and positive affect (neutral, positive affect only, and comedy). As in Forgas’s (2011) study, participants in this study were asked to write down five pieces of information they would like to reveal about themselves, and their statements were subsequently rated by two independent observers for intimacy, abstractness, and valence.

Another potential issue with Forgas’s procedure is that it focused entirely on how much participants were observed to disclose and thus failed to consider the degree to which the participants personally felt that they had disclosed. This is important because there may be a “disconnect between disclosing behaviors and perceptions” (Nguyen et al. 2012:109). In other words, people in a positive mood may, in reality, reveal more than those in a sad mood, but they may feel like they disclose more when sad than when happy. We therefore compared the rates of disclosure assessed by independent raters with those rated by the subjects themselves.

Although other hypotheses for the evolution of laughter might also predict disclosure, our study is explicitly motivated by the laughter-as-social-lubricant hypothesis, and we therefore phrase our discussion largely in these terms. We emphasize that this study is not a test of alternative hypotheses for laughter, but rather a test of whether laughter itself (rather than affect) is the factor that facilitates self-disclosure. We predicted that people in the comedy condition would disclose more intimate information about themselves than those in either the neutral or the positive affect conditions because they laughed more. Although we might expect people in the positive affect condition to disclose more intimately than those in the neutral condition, given the influence of positive affect on willingness to take risks (Trope et al. 2001), we predicted that, if laughter is the key factor, those in the comedy condition would reveal more intimate information about themselves than people in either the neutral or the positive affect conditions, since laughter, rather than positive affect per se, is hypothesized to have evolved specifically to facilitate the development of intense social bonds.


The experiment used a 3 × 2 between-subjects factorial design, with clip (comedy, positive affect, and neutral) and group sex (male vs. female) as the independent variables and self-reported and observer-rated disclosure (intimacy, abstractness, and valence) as the dependent variables.


Participants were 112 students (72 women, 40 men; age range 18–31; M = 22.43 years, SD = 3.34 years) who were compensated for their participation with £10 cash. Participants were recruited via mass email to the faculties and departments of two universities in Oxford, England. Participants signed up individually for the experiment and were randomly assigned to groups online. Because laughter in mixed-gender groups may serve a sexual rather than a cooperative function (Mehu and Dunbar 2008), each group was either all male or all female in composition. Each group of four participants was randomly assigned to one of three conditions (comedy, neutral, affect). Of the males, 12 individuals were assigned to the comedy condition, 16 to the neutral condition, and 12 to the positive affect condition; of the females, 24 individuals were assigned to each of the three conditions.


Four individuals arrived simultaneously at the meeting room of the social psychology laboratory, where they were welcomed by an experimenter and completed an informed consent form. None of the participants in any of these groups had met before, and all were requested not to speak to each other until asked to do so by the experimenter. Through the advertising of the study and through the participant information form, participants were led to believe that they would be given the opportunity to interact with the other participants at some later point during the study.


Each group of four participants first watched one of the 10-min mood-induction videos. The comedy film was taken from a performance by stand-up comedian Michael McIntyre (from his BBC television series Michael McIntyre’s Comedy Road Show). The show contains a compilation of inoffensive observational comedy accompanied by audience laughter, which has proved highly popular to UK audiences of both genders. The neutral film was taken from a golf instruction video (Paul Wilson’s Golf Instruction Tips and Schools). The positive affect video was an excerpt showing lush, pleasant scenes and animals from the “Jungles” episode of the BBC TV series Planet Earth. These films were previously used for similar purposes by Dunbar et al. (2012). The choice of a comedy clip to stimulate laughter was driven by the need to encourage Duchenne (relaxed, unforced laughter that is stimulus driven and emotionally valent) as opposed to non-Duchenne laughter (which is context-driven and emotionless). Importantly, only Duchenne laughter has been shown to increase positive affect and alleviate stress (Keltner and Bonanno 1997), and as a consequence only Duchenne laughter is believed to trigger endorphin release and serve the function of facilitating the development of social bonds. We did not directly monitor the frequency of Duchenne and non-Duchenne laughter; we relied on the fact that stand-up comedy reliably triggers Duchenne laughter (even if it may also sometimes trigger non-Duchenne laughter) and the fact that, in our previous experiments (Dunbar et al. 2012), laughter reliably increased pain threshold (a widely used proxy for endorphin activation) even though no distinction was made between types of laughter. In fact, it is likely that non-Duchenne laughter is rare in these contexts because subjects do not interact with each other (the normal context for non-Duchenne laughter) but instead sit in a line watching a video screen.

All videos were played using Windows media player and were projected onto a SMART Board SB680 (screen size: 188 × 117.2 cm). Because humans do not laugh readily when watching even the funniest performances alone (Freedman and Perlick 1979; Provine 2001), and because laughter is 30 times more likely to occur in social contexts than otherwise (Provine and Fischer 1989), all subjects were tested in groups of four.

Manipulation Checks

Laughter Rates and Funniness

Since it was important that the experimenter left the room during the manipulation (Chapman 1976) and since people are poor judges of how much they laugh (Vettin and Todt 2004), laughter in response to the video clips was recorded by a small audio recorder, discreetly placed behind the participants. At the end of the study, two raters independently counted the number of laughs recorded for the group for each 10-min session. These ratings were highly consistent (Cohen’s κ = 0.82). For analysis, we averaged the raters’ scores. Participants also rated how funny they found the clip on a scale of 1 (not at all) to 5 (a great deal).

Positive and Negative Affect Schedule (PANAS)

Immediately after the clip, the participants’ emotional state was assessed using the positive and negative affect schedule. The PANAS is a 20-item self-report measure of positive and negative affect developed by Watson et al. (1988). The PANAS requests that participants indicate the extent to which they are currently experiencing a range of positive and negative emotions on a scale of 1 (very slightly or not at all) to 5 (extremely). Ten items relate to positive affect (e.g., enthusiastic, interested, and excited) and ten items relate to negative affect (e.g., upset, ashamed, and irritable).

Self-Disclosure Task

After the manipulation checks, in an ostensibly unrelated experiment on “social communication,” participants were instructed to sit in separate corners of the room and were given a (randomly assigned) piece of colored card (red, blue, green, or yellow). They were asked to show the card to the other participants and to remember which individuals held which colored cards. They were then asked to face away from each other and to complete a questionnaire which instructed them to compose a message for one of the members of the group as follows:

Of the other people present whilst you watched the clip, please compose a message for the person holding the [ ] card so that they can get to know you better. Write down on the sheet provided the five pieces of information you would like to say about yourself to this person. Please use the same words you would use in the actual conversation. You will subsequently have the opportunity to interact with this person.

This was adapted from Forgas’s (2011) self-disclosure task. However, while Forgas requested that participants target their message to a hypothetical other, with whom they were instructed to imagine having a conversation, the present study asked participants to compose a message for someone they had just met and believed that they were about to interact with. This was intended to increase the ecological validity of the task.

All participants composed a message for someone who did not compose a message for them. Coupled with the fact that participants individually completed the questionnaire, this all but eliminated the possibility of the participants attempting to signal to their partners that they were writing a message for each other and hence manipulate their partner’s desire to disclose (e.g., catch their partner’s eye and smile).

Dependent Measures

Following the disclosure task, participants then rated their five self-disclosing statements on 10-point scales for intimacy, abstractness, and valence. Following Forgas (2011), these were defined to the participants as follows: (a) intimacy: revealing personal details (e.g., high-intimacy topics: health, body, finance, sex; low-intimacy topics: age, school, suburb, study area; Jourard 1971); (b) abstractness: information about specific versus general, abstract characteristics (e.g., disclosing about concrete behaviors vs. abstract traits); and (c) valence: positivity versus negativity of content disclosed (e.g., revealing successes vs. failures). Of these, intimacy is the key variable, since these kinds of disclosure are the only ones in which the speaker risks giving away personal secrets that would make them vulnerable, and in that respect might lay the foundation for a close, intimate relationship. Examples of highly intimate disclosure statements were “In January I broke my collarbone falling off a pole while pole dancing,” “Half of my favorite films are (embarrassingly) Disney films,” and “I’m currently living in squalor (with mice!).” Low-intimacy disclosure statements included “I am at Worcester College in my first year,” “I am from Cheltenham,” and “I love eating different foods from around the world.” By contrast, abstractness and valency are of much less relevance to the perceived quality of a relationship. We thus expect any differences that correlate with experimental condition to be reflected in intimate disclosures, with any effects for abstract or valent disclosures being much weaker or even absent altogether.

The five disclosing statements were also rated on the same dimensions (intimacy, abstractness, and valence) on 10-point scales by two independent raters blind to the experimental condition. The constructs were defined to the raters in the same way that they were defined to the participants. Because the two raters showed acceptable levels of agreement across each dimension of disclosure (Cohen’s κ = 0.76), their ratings were averaged for the analysis.

Control Measures

Participants were also requested to provide information about traits that have previously been shown to modify self-disclosure in interactions (Greene et al. 2006). Participants were instructed to complete the 10-item personality inventory (TIPI), a psychometrically reliable and valid instrument for testing openness, conscientiousness, agreeableness, extraversion, and neuroticism (Gosling et al. 2003).


Debriefing featured a series of open-ended questions about what participants thought about the study. Responses suggested that participants had not been aware of the manipulations.


Manipulation Checks

Laughter Rates and Funniness

To establish that laughter rates differ across the conditions as predicted, we tested for an effect of video type on laughter while watching the video. There was a main effect of condition on the extent to which participants laughed during the clip (F2, 109 = 28.45, p < 0.001, η2 = 0.34) (F2, 109 = 28.45, p < 0.001, η2 = 0.34)1 and on the extent to which they found the clips funny (F2, 109 = 17.72, p < 0.001, η2 = 0.25). Scheffé post hoc tests confirmed (a) that both laughter and rated funniness were significantly higher in the comedy condition than in either of the other two conditions (p < 0.001) and (b) that there were no significant differences in laughter and rated funniness between the neutral and the positive affect conditions (p > 0.05). For the means and standard deviations, see Table 1. Thus, at least as far as making subjects laugh is concerned, the clips did what they should have done.
Table 1

Descriptive statistics for laughter, funniness, and affect as a function of condition


Comedy (n = 36)

Positive affect (n = 36)

Neutral (n = 40)





















Positive Affect







Negative affect














a Laughter rates were calculated by two independent raters who showed high inter-rater agreement (Cohen’s κ = 0.82). All other measures were self-reported

There was no main effect of video type on either positive affect (F2, 109 = 1.21, p > 0.05, η2 = 0.02) or negative affect (F2, 109 = 0.87, p > 0.05, η2 = 0.02). Although the absence of a main effect of condition on positive affect was contrary to expectation (and the same “affect” video had yielded an increase in positive affect scores in previous experiments; Dunbar et al. 2012), the fact that there are no differences in positive affect scores between the three videos but there were differences in laughter rates nonetheless allows us to dissociate these two effects. There was a main effect of video clip on how interested participants reported to be feeling (condition: F2, 109 = 6.04, p < 0.05, η2= 0.10). Scheffé post hoc tests revealed (a) that interest was significantly higher in the comedy and positive affect condition than in the neutral condition (p < 0.05) and (b) that there was no significant difference in reported interest between the comedy and the positive affect conditions (p > 0.05). Thus, at least one aspect of positive affect was influenced as intended by the manipulation. For the means and standard deviations see Table 1.

The Effect of Laughter on Disclosure

The Relationship Between Self-Report and Observer-Rated Disclosure

To test whether self-reported disclosure and observer-rated disclosure are independent measures of disclosure, the three disclosure dimensions were correlated across these two forms of measurement. Self-reported and observer-rated valence correlated significantly (r112 = 0.27, p < 0.05) as did self-reported and observer-rated intimacy (r112 = 0.20, p < 0.005). However, the modest size of these correlations suggests that raters and disclosers had partly discrepant perceptions of these dimensions. Further, the correlation between self-reported and observer-rated abstractness was non-significant and close to zero (r112 = 0.06, p > 0.05).

Laughter and Self-Reported Disclosure

Considering self-reported disclosure only, the three facets of disclosure (intimacy, abstractness, and valence) were correlated with one another to test their independence. There were no significant relationships between any of the three facets of self-reported disclosure (p > 0.05).

We ran a multivariate analysis of variance (MANOVA) with the three self-rated facets of disclosure (intimacy, abstractness, and valence) as the dependent variables and gender (male/ female) and condition (comedy, positive affect, and neutral) as the two independent variables. There was no main effect of gender (Pillai’s trace V = 0.059, F3, 104 = 2.190, p > 0.05, η2 = 0.06) or of condition (V = 0.051, F6, 210 = 0.924, p > 0.05, η2 = 0.03), nor was there an interaction effect between the two (V = 0.073, F6, 210 = 1.32, p > 0.05, η2 = 0.04, see Fig. 1). These results held, even when separately and concurrently controlling for the effects of personality traits that have previously been shown to influence disclosure rates (i.e., extraversion, neuroticism, and openness) and when controlling for the rated funniness of the clip.
Fig. 1

Mean score (±SE) for each self-rated disclosure facet (intimacy, abstractness, and valence) as a function of video condition (comedy, neutral, and positive affect)

Overall individual positive affect scores (as measured by the PANAS) did not predict self-reported intimacy, abstractness, or positivity of disclosure (p > 0.05). Although individual positive affect scores did significantly predict self-reported disclosure positivity within the neutral condition (R2 = 0.191, F1,38 = 15.559, p = 0.003), it failed to do so for the comedy or positive affect conditions (p > 0.05).

Laughter and observer-rated disclosure

As with self-rated disclosure, we first tested whether the different facets of observer-rated disclosure (intimacy, abstractness, and valence) were indeed independent of one another. Only observer-rated intimacy and abstractness were significantly related (r112 = 0.47, p < 0.01).

We then ran a second MANOVA with each of the three observer-rated facets of disclosure (intimacy, abstractness, and valence) as the dependent variables and gender (male/female) and condition (comedy, positive affect, and neutral) as the two independent variables. The multivariate main effect of condition was significant (V = 0.240, F6, 210 = 4.76, p < 0.05, η2 = 0.12) but the main effects of gender (V = 0.037, F3, 104 = 1.33, p > 0.05, η2 = 0.04) and the interaction effect (V = 0.047, F6, 210 = 0.85, p > 0.05, η2 = 0.02) were not. There were also significant univariate main effects of condition for observer-rated intimacy (F2, 106 = 5.81, p < 0.05, η2 = 0.10) and abstractness (F2, 106 = 6.187, p < 0.05, η2 = 0.11, see Fig. 2), but not on observer-rated valence (F2, 106 = 1.17, p > 0.05, η2 = 0.02).
Fig. 2

Mean score (±SE) for each observer-rated disclosure facet (intimacy, abstractness, and valence) as a function of video condition (comedy, neutral, and positive affect)

Scheffé post hoc tests for observer-rated intimacy revealed (a) that intimacy was significantly higher in the comedy (M = 5.24, SD = 1.72) than in the neutral (M = 4.04, SD = 1.47) condition (p < 0.05), (b) that the difference in observer-rated intimacy between the comedy and the positive affect (M = 4.54, SD = 1.54) condition was non-significant (p > 0.05) and (c) that the difference in observer-rated intimacy between the neutral and the positive affect condition was non-significant (p > 0.05). Scheffé post hoc tests for observer-rated abstractness revealed (a) that the difference in observer-rated abstractness between the comedy condition (M = 4.07, SD = 1.24) and the neutral condition (M = 3.64, SD = 1.63) was non-significant (p > 0.05) and (b) that abstractness was significantly higher in the positive affect condition (M = 5.06, SD = 1.24) than in the comedy or the neutral conditions (p < 0.05). Separately and concurrently controlling for the effects of personality traits that have previously been shown to influence disclosure rates (i.e., extraversion, neuroticism, and openness) did not affect the pattern of significant results.

Overall, individual positive affect scores (as measured by the PANAS) did not predict observer-reported intimacy or abstractness. However, they did significantly predict observer-rated disclosure valence (R2 = 0.067, F1,109 = 8.889, p < 0.01). This effect was echoed within the positive affect condition (with individual PA significantly predicting disclosure valence: R2 = 0.089, F1,33 = 4.329, p < 0.05) but not within the comedy or the neutral condition (p > 0.05). No other effects were found. Controlling for the effect of funniness ratings had no effect on the pattern of results, except to make the difference in the frequency of abstract disclosures between the comedy and affect conditions non-significant.


The present finding that observer-rated intimacy increased in the comedy condition relative to the neutral condition supports Forgas’s (2011) result that people who watched a comedy clip were more likely to disclose intimate information than those who watched a sad clip. However, although Forgas attributed the difference in disclosure to a change in affect, the present study found that the induction of positive affect alone was insufficient to produce significant changes in the intimacy of disclosure. Although inducing positive affect did increase the personal intimacy of disclosure relative to the neutral condition, it did not do so significantly; only the comedy condition increased the disclosure rate significantly above the neutral baseline. This suggests that Forgas’s results may have been a consequence of disparate laughter rates rather than changes in mood per se. Indeed, the fact that there were significant differences in disclosure across the three video conditions despite the fact that there were no differences in positive affect argues against the interpretation that a change in mood is responsible for the differences in disclosure intimacy.

We suggest that one likely reason why the comedy condition elicited significantly more intimate disclosures than the neutral condition is that the higher laughing rate led to an increase in endorphin activation. Given laughter’s ability to trigger endorphin activation (Dunbar et al. 2012) and the role of endorphins in the formation of social bonds (Machin and Dunbar 2011), laughter may increase willingness to disclose intimate information because the opioid effect of endorphins makes individuals more relaxed about what they communicate. Since the effect of endorphins may be limited to behaviors associated with social bonding (Machin and Dunbar 2011), the present finding that the abstractness and valence of disclosures were not significantly increased in the comedy condition relative to the neutral or the positive affect conditions is particularly interesting and may reflect their lack of importance in relationship development. Certainly, when we think about disclosure as a tool for building relationships, how abstract or positive disclosures are seems far less important than their personal intimacy (Collins and Miller 1994; Derlega and Berg 1993). Similarly, laughter may have influenced observer-rated but not self-rated disclosure intimacy, as the former may be more critical for relationship development. In other words, the extent to which one feels one has disclosed may not be as important for relationship development as the extent to which others judge one to have disclosed (Nguyen et al. 2012; Schiffrin et al. 2010).

We have framed our study largely in terms of the laughter-as-social-lubricant hypothesis, but it is worth noting that our conclusions here are not specifically dependent on which of the many functional hypotheses offered for laughter (summarized briefly in the Introduction) is true. It might well be possible to predict an increase in intimate disclosure rates following laughter for all of them. However, the causal mechanisms would be very different in each case. The social lubricant hypothesis depends explicitly on the endorphin mechanism that laughter is known to engage and is fundamental to the bonding processes of anthropoid primates. The signaling hypotheses explicitly require that individuals signal to each other, and it is less obvious how this might work in a context where subjects were (a) not facing or interacting with each other while watching the videos (they were sitting side by side) and (b) not expecting to interact with each other at the time they were laughing (and in Forgas’s experiment did not in fact interact with each other after watching the videos). Similarly, explanations that laughter modulates and coordinates emotions would seem to be less plausible given the absence of any effect on affect. That being said, our concern here is not to test between alternative hypotheses for the function of laughter (some of which may, in any case, be by-products rather than causes of the evolution of laughter) but rather to test a particular confound that we have noted in Forgas’s experimental design—a confound that happens to be particularly relevant to the social lubrication functions of laughter. At the very least, our results would seem to provide prima facie evidence for that suggestion.

Precisely how endorphins work to increase disclosure intimacy is a question for future research. One possibility is that endorphins influence disclosure intimacy (and hence facilitate the development of social bonds) by allowing one to shift attention away from oneself. Since excessive self-focus increases access to negative attitudes and interferes with performance in social situations (Ickes et al. 1973; Mor and Winquist 2002), endorphins may serve to facilitate interaction through reducing self-directed attention; alleviating concerns about disclosing too much information or coming across as “weird” or unlikable, and, in turn, promoting the exchange of intimacies.

That endorphins may increase disclosure intimacy by shifting attention away from the self suggests an alternative (though not necessarily mutually exclusive) explanation for why laughter influenced observer-rated but not self-rated disclosure intimacy—namely, that laughter reduces self-focus, and that this in turn reduces awareness of how intimately one is disclosing. This raises several intriguing questions for further study. First, it predicts that manipulating self-focus should affect the extent to which there is a discrepancy between self-rated and observer-rated disclosure. Second, it suggests the possibility that laughter will lead people to be inaccurate in their judgments of their own disclosure, yet accurate in the extent to which they judge other people’s disclosures. And finally, it suggests that people with psychiatric disorders characterized by excessive self-focus, such as social anxiety disorder (Clark and Wells 1995; Ingram 1990), might benefit from practicing laughter when in the company of others.

One unexpected finding from the present study is that people in the positive affect condition made significantly more abstract disclosures than those in either the neutral or comedy conditions. Since participants in the positive affect manipulation were no higher on positive affect scores than those in other conditions, perhaps something about the positive affect condition, besides its effect on mood, caused people to disclose more abstractly. Since this difference disappeared when the funniness of videos was partialed out in the statistical analysis, laughter might again be relevant. Laughter rates were higher, albeit not significantly so, in this condition than in the affect condition (Table 1), and this might have been sufficient to encourage a limited increase in disclosures. Alternatively, the content of the positive affect video may have raised more abstract themes than the more specific observations of Michael McIntyre (comedy video) or the more specific advice about golf provided by Paul Wilson (neutral video). Thus, participants may have been primed in the positive affect condition, via exposure to abstract content, to subsequently reveal more abstract disclosures (Gilead et al. 2013). Further experiments seem warranted.

The present study also found no difference between the neutral condition and the positive affect and comedy conditions on self-rated or observer-rated disclosure positivity (valence). Although we are inclined to interpret this result at face value as indicating a genuine lack of an effect of comedy and positive affect on disclosure positivity, it could be a product of the strength of the present manipulation of mood. Given that the positive affect and comedy conditions did not differ significantly from the neutral condition on overall positive or negative affect, perhaps our manipulation was not strong enough to influence participants’ emotional states, and consequently, the change in mood in the comedy and positive affect conditions was insufficient to cause any differences in disclosure positivity between people in these conditions and those in the neutral condition. Further experimental manipulation might be warranted.

Although the present study improved upon the ecological validity of Forgas’s (2011) self-disclosure task by having participants compose a message for another participant as opposed to an imaginary other, the act of disclosing via a written message may not reflect disclosure in a conversational context. One reason may relate to communication mode richness (Daft and Lengel 1986). In comparison to face-to-face communication, sending a message to another via written text provides a substantial reduction in communication channels (notably facial expressions and voice intonations). If people prefer to communicate difficult concepts and emotionally sensitive information via “rich” media (Daft et al. 1987) and/or find face-to-face interaction more satisfying than text-based communication (Vlahovic et al. 2012), perhaps the current lack of an effect of laughter on the abstractness and the valence of disclosures is a consequence of the communication medium rather than a lack of effect of laughter rates per se, since laughter is less likely to occur in text-based interactions (Vlahovic et al. 2012). Future research should therefore address whether similar results are obtained when participants interact with one another FTF as opposed to via written text.

Regardless of whether disclosure via the written word corresponds to disclosure in a conversational context, the present study’s use of text as a medium for participants’ disclosures has value in itself. With the rise of social networking sites, such as Facebook and Twitter, many people now frequently reveal a significant amount of personal information via text (Coleman et al. 1999; Naaman et al. 2010). Although it may not be the “richest” medium for communicating personal information (Daft and Lengel 1986), a growing number of people use it to do so every day (Kock 2005). To the extent that the present study’s disclosure task mimics communication via social networking sites, one may wonder how applicable the current result of a “disconnect” between actual and perceived disclosure is to such interaction.

The principal goal of the present research was to test whether the social lubricant theory of laughter rather than affect offered a better explanation for Forgas’s (2011) findings that people will disclose more intimate information about themselves after positive affect. Through controlled experimental conditions, we found support for the role of laughter in encouraging more intimate disclosures to strangers but not more abstract or positive disclosures. Of particular interest was the fact that this effect was limited to observer ratings of disclosure and not self-ratings of disclosure. This would seem to be in line with the notion that laughter is linked specifically to fostering behaviors that encourage relationship development, since observer-ratings of disclosure may be more important for relationship development than how much one feels one is disclosing. These results suggest that laughter should be a serious topic for those interested in the development of social relationships.


Since participants completed the study in groups, the data were not strictly independent. For this reason, all analysis reported here were repeated again at the group level using aggregated data (n = 28). The pattern of significant findings was identical in all cases.



AG was supported by the Crewe Graduate Scholarship, Lincoln College Oxford. RD’s research is supported by a European Research Council Advanced grant.

Copyright information

© Springer Science+Business Media New York 2015