1.1. Sex differences in psychological kinship

There are also several conceptually independent reasons to believe that the tendency to experience friendship as akin to kinship may be more pronounced for women than for men. First, consider sex differences in the functional consequences of social status: compared to female reproductive fitness, male reproductive fitness has been more highly dependent on social status (e.g., position in a dominance hierarchy; Buss, 1989). Consequently, male relationships are likely to be more competitive, whereas female friendships are likely to be comparatively cooperative, mirroring the nepotistic cooperativeness that occurs between kin. This line of reasoning applies primarily to same-sex friendships. A second line of reasoning applies more clearly to cross-sex friendships and draws on the logic of differential parental investment which suggests that, compared to men, women are likely to be more cautious and risk-averse in their approach to mating (Trivers, 1972). One consequence is that women may be much more sensitive than men to superficial signals that heuristically connote a poor mate choice. Thus, while men and women both respond negatively to the prospect of sexual intercourse with individuals who are explicitly known to be close kin (although even here women respond more negatively; Fessler & Navarrete, 2004, Lieberman et al., 2003), women may be much more likely than men to also respond negatively to the prospect of sexual relations with any individuals who are associated with superficial kinship cues of the sort that characterize many friendships (e.g., perceived phenotypic similarity, feelings of self-other overlap).

Consistent with these speculations are empirical results documenting sex differences in friendships. Women and men take somewhat different approaches to social bonding (Geary & Flinn, 2002, Kashima et al., 1995, Taylor et al., 2000). Whereas male coalitions tend to be transitory, hierarchical, and task-oriented, female coalitions are more tightly interconnected and focused on socioemotional bonds (Barth & Kinder, 1988, Baumeister & Sommer, 1997, Cross & Madson, 1997). Compared to male friendships, female friendships are marked by higher levels of smiling, emotional sharing, and self-disclosure (Baumeister & Sommer, 1997, Caldwell & Peplau, 1982). Attraction patterns between opposite-sex friends are also consistent with the notion of sex differences. Relative to men, women are less sexually attracted to their opposite-sex friends and less likely to consider potential sexual contact an important precursor to initiation of these friendships (Bleske-Rechek & Buss, 2001). Thus, whereas women generally show higher levels of intimacy within friendships, they explicitly desire less sexual intimacy with opposite-sex friends—a pattern entirely consistent with the conjecture that women, more than men, perceive friendships as akin to kinships.

While previous studies have documented many predictable sex differences in the nature of friendship relations, no prior research has attempted to make the explicit comparisons necessary to more directly test whether there is a sex difference in the extent to which friendships are experienced as akin to kinships. To do so, we conducted two studies that compared behavioral and affective responses to kin, friends, and strangers—and did so in two contexts that are directly relevant to the evolved psychology of kinship.

1.2. Overview of current research

It can be argued that any psychological indicators of kinship should feed into at least two different downstream mechanisms: (a) a “sexual value estimator” that inhibits incestuous mating (Lieberman et al., 2003) and (b) an estimator of “welfare–tradeoff ratio” that regulates the extent to which individuals will forego rewards in order to benefit another (Tooby, Cosmides, & Price, 2006). Our two studies followed from this conceptualization (also see Lieberman et al., 2007) and assume that, if friendships are akin to kinships, pertinent evidence should emerge in social contexts relevant to sexual intercourse and interpersonal generosity.

Study 1 focused on emotional responses to imagined sexual contact with opposite-sex kin, friends, or strangers. Previous research indicates that, within sexual situations, perceived kinship is associated with a strong disgust response (Fessler & Navarrete, 2004, Lieberman et al., 2003). Consequently, it is informative to examine disgust reactions specifically, as well as approach-oriented positive emotions (e.g., romantic love) that are functionally inconsistent with disgust. If women are more likely than men to treat friendships as kinships, then—compared to men—women would be expected to experience relatively greater disgust, and relatively lower levels of positive emotion, when imagining sexual contact with a friend.

Study 2 examined a particular form of attributional benevolence (i.e., the giving of credit) toward actual kin, friends, and strangers. The logic of nepotism implies greater favoritism towards kin than towards strangers. If women are more likely than men to treat friendships as kinships, then women would be expected to express greater attributional benevolence toward friends as well.

2.1. Methods

Study 1 employed a 2 (Participant Sex)×3 (Partner Relationship) between-subjects design. A total of 241 self-reported heterosexual undergraduate students (172 female, 69 male) were recruited in classes and participated in exchange for course credit.

Participants received a guided visualization task asking them to imagine interacting with an opposite-sex person of similar age identified as a sibling, a close friend of at least 1 year (with whom participants had never been romantically involved), or a newly met stranger. After this, participants spent one full minute contemplating the following scenario: “Imagine having a sexual relationship with this person. Concentrate on what it would be like to have sex with this person for the first time.” Questionnaire items then assessed emotional responses recorded on 8-point rating scales (not at all to very much). These included ratings of Disgust, as well as six additional negative affective states (Anger, Fear, Sadness, Shame, Confusion, Guilt), and three positive approach-oriented affective states (Happiness, Romantic love, Excitement).

2.2. Results

Given the conceptual importance of disgust as a kinship cue, one set of analyses focused specifically on ratings of disgust. Two additional emotion composite indices were created. A negative affect index was created by averaging ratings on the six negative emotions excluding Disgust (Cronbach's alpha=.90). A positive affect index was created by averaging ratings on the three positive emotions (Cronbach's alpha=.94). Separate 2×3 analyses of variance (ANOVAs) were conducted on Disgust ratings and the two composite indices. Additional simple-effect analyses were conducted using Bonferroni corrections (see Table 1 for means).

Table 1. Emotional reactions to imagined sexual contact (Study 1)

		Relationship type
		Kin	Friend	Stranger
	Disgust
	Women	6.88 (0.47)a	3.09 (2.41)b	1.77 (2.21)c
	Men	6.90 (0.31)a	1.73 (2.63)c	1.00 (1.98)c
	Negative affect
	Women	4.48 (2.16)a	2.65 (1.54)b	2.32 (2.00)b
	Men	4.53 (1.74)a	1.85 (1.65)b	1.43 (1.68)b
	Positive affect
	Women	.17 (0.44)a	2.52 (2.11)b	4.26 (1.89)c
	Men	.28 (0.60)a	4.29 (2.40)c	5.04 (1.14)c

Within each emotion set, means not sharing a common subscript differ by at least p<.05.

On Disgust ratings, there emerged main effects for Relationship Type [F(2,235)=141.53, p<.001, np2=.55] and Sex [F(1,235)=6.44, p=.01, np2=.03]. The sex difference was strongest in the Friend condition; indeed, it was only in the Friend condition that there emerged a statistically significant simple effect, whereby women expressed greater disgust than men [F(1,235)=8.83, p<.01, np2=.04].

These effects were less pronounced on other negative emotions. Although there did emerge the same two main effects on the negative affect composite (for Relationship Type, F(2,235)=36.24, p<.001, np2=.24; for Sex, F(1,235)=4.25, p<.05, np2=.02) these effects were weaker, and the sex difference specific to the friend condition was only marginally significant [F(1,235)=3.31, p=.07, np2=.01]. In order to further distinguish the role of the incest-relevant emotion (Disgust) from general negative affect, two analyses of covariance were performed. When the negative affect composite was entered as a covariate into an ANOVA on Disgust, the sex difference in the Friend condition persisted [F(1,234)=5.62, p<.02, np2=.02]. But when Disgust was covaried from an ANOVA on the negative affect composite, no sex difference emerged (F<1). These results suggest a sex difference specific to disgust rather than to other negative emotions.

The arousal of disgust inhibits approach-oriented positive emotions. Consequently, it was anticipated that there should also emerge predictable sex differences on the positive affect composite. Results on the positive affect index indicated main effects of both Relationship Type [F(2,235)=117.18, p<.001, np2=.50] and Sex [F(1,235)=14.13, p<.001, np2=.06], as well as a significant interaction [F(2,235)=4.18, p<.02, np2=.03]. The interaction attests to an especially strong sex difference within the Friend condition: only in the Friend condition was there a statistically significant simple effect whereby women expressed less positive affect than men [F(1,235)=20.56, p<.001, np2=.08].

2.3. Discussion

Thoughts of incest typically arouse disgust and the inhibition of approach-oriented positive emotions. This was clearly evident in our results: Men and women both expressed high levels of disgust and low levels of positive affect when imagining sexual contact with a sibling. Sex differences on these emotional responses did emerge when imagining sexual contact with nonkin but only when imagining sexual contact with a close friend and not a stranger. Women showed a more aversive emotional response to the thought of sex with a friend. For men, there was no significant difference between the Stranger and the Friend conditions, but for women, the difference between Strangers and Friends was more substantial—indicating that, for women, more than for men, sex with a familiar friend is more disgusting and less alluring than sex with an imagined stranger. The overall profile of results indicates that while neither men nor women respond to close friends exactly as they respond to kin, there is a clear tendency for women, more than men, to respond to close friends in a more kin-like way.

Although these results are consistent with the hypothesis that women are more likely than men to perceive friends as psychologically akin to kin, the results might also be seen as consistent with an alternative explanation that has nothing to do with kinship, per se. This alternative explanation draws on previous research indicating that men often seek to form cross-sex friendships as a strategic means for gaining access to potential mates (Bleske-Rechek & Buss, 2001). It could be argued that the key sex difference—in the Friend condition—reflects not so much a tendency for friendship to connote kinship (for women more than for men) but, rather, a tendency for friendship to connote a preliminary stage in a desired mating relationship (for men more than for women). Of course, this alternative explanation is specific to the mating domain. In contrast, if friendship really does implicitly connote kinship for women more than men, this sex difference should be observed in other (mating-irrelevant) contexts within which kinship is functionally important—such as the domain of interpersonal generosity. Study 2 focused on this domain and tested the hypothesis that, compared to men, women respond to friends in a more benevolent manner mimicking nepotistic responses to actual kin.

3.1. Methods

Study 2 employed a 2 (Participant Sex)×3 (Relationship Type) between-subjects design. Participants included 166 undergraduate students recruited from a psychology subject pool. Participants in the Stranger condition were paired with another unfamiliar participant. Participants in the Friend and Kin conditions (blind to all experiment details) were asked to recruit either a close friend (with whom participants had never been romantically involved) or a kin member to participate with them. This resulted in the recruitment of 78 additional participants for a total of 244 participants (134 female, 110 male), who were paired in condition-specific dyads. All students received course credit for their participation.

Preliminary measures assessed participants' perceptions of dyad partners along three dimensions bearing on interpersonal intimacy: feelings of closeness, perceptions of similarity, and liking. For all three measures, ratings were recorded on 9-point scales.

Procedures for the main task were adapted from Campbell, Sedikides, Reeder and Elliot (2000). Dyad partners, seated at separate computers, each completed an online exam that was ostensibly used by major corporations to predict future professional success (items were adapted from the analytical portion of the Graduate Record Examination). Following the exam, participants were informed that because the researchers were interested in group processes, only an average dyad score would be reported. They were informed that their dyad score was in the 93rd percentile, and they were given explicit feedback that their dyad had done well.

The dependent variables consisted of responses made on 9-point rating scales to two questions assessing attributions about the source of their joint success: (1) “Who was most responsible for the outcome of this test?” and (2) “Who made the greatest positive contribution to this test?”

3.2. Results

Preliminary analyses tested for differences between same-sex and cross-sex dyads. None were observed (Fs<1) [there also was no difference between male and female participants in their likelihood of being paired with a male or female partner, χ2(1)=1.25, nonsignificant]. All analyses reported below collapse across this variable.

Potential statistical dependency between dyad members' attribution ratings was tested by means of intraclass correlation coefficient. Values ranged from .02–.06, indicating that dyad partner responses were independent. Analyses were therefore conducted at the level of the individual. The two attribution ratings were averaged to form a composite index (Cronbach's alpha=.91). Lower values on this index indicate more self-serving attributions; higher values indicate more benevolent attributions.

Fig. 1 portrays mean responses on the attribution index. This pattern of means indicates that: (a) within the Stranger condition, both men and women were relatively self-serving in their attributions; (b) within the Kin condition, both men and women were relatively benevolent in their attributions; and (c) within the Friend condition, men were relatively self-serving, while women were relatively benevolent.

Fig. 1. Mean attributions of responsibility (and standard errors) for successful dyadic performance in Study 2. The dotted line represents an equal attribution of credit within a dyad. Ratings greater than 5 indicate relatively benevolent attributions, while ratings less than 5 indicate relatively self-serving attributions.

Inferential analyses were accomplished through a contrast analysis for predicted pattern testing; this analysis is Type-1 error-responsive but more precise and parsimonious than an omnibus ANOVA (Levin & Neumann, 1999). Three cells received contrast weights of −1 (Kin–Female, Kin–Male, Friend–Female), and another three cells received contrast weights of 1 (Friend–Male, Stranger–Female, Stranger–Male). This contrast was significant [F(1,229)=14.15, p<.001, np2=.06]. The test of the residual contrast was not significant [Fresidual(4,229)=.048, nonsignificant], indicating that departures from the predicted pattern were statistically negligible.

Additional analyses focused on sex differences within each relationship type. No sex differences were found in the Stranger (p>.66) or Kin (p>.39) conditions, but there was a significant sex difference within the Friend condition [F(1,229)=4.14, p<.05, np2=.02].

Comparisons within Sex are also informative. Men made somewhat more benevolent attributions to kin than to friends, although this difference was nonsignificant [F(1,229)=2.09, p=.14] while responding in an equally self-serving way toward both strangers and friends (F<1). Women made more benevolent attributions to friends than to strangers [F(1,229)=4.74, p<.05, np2=.02] and responded in an equally benevolent way toward both friends and kin (F<1).

Ancillary analyses examined the ratings of interpersonal closeness, similarity, and liking. A 2×3 (Participant Sex×Relationship Type) multivariate ANOVA on these three variables revealed a main effect for Relationship Type [F(6,456)=105.86, p<.001, np2=.58] as well as an interaction [F(6,456)=2.58, p<.02, np2=.03]. Decreasing partner relationship was associated with less perceived intimacy, and this decline was especially strong with male participants. Multivariate contrasts indicated that significant sex differences emerged only in the Friend condition [F(3,228)=3.94, p<.02, np2=.05]. Given these results, it is important to know whether the sex difference in attributional benevolence toward friends resulted specifically from sex differences in perceived closeness, similarity, and liking. We addressed this question with a series of regression analyses designed to test whether these three intimacy variables mediated the relationship between sex and attributional benevolence (in the Friend condition). Results showed no evidence of significant mediation [ps>.12).

3.3. Discussion

The logic of nepotism implies that people should make more benevolent attributions to kin than to unrelated strangers. This was clearly the case. If friends are viewed as somewhat akin to kin, psychologically, it also follows that people should make more benevolent attributions to friends than to strangers. This occurred only for women. Men responded to friends as though they were strangers, not kin. In contrast, women responded to friends in a much more benevolent way—so benevolent, in fact, that these responses were not statistically distinguishable from their responses to kin.

Also, compared to men, women reported greater feelings of closeness, similarity to, and liking for their friends. These feelings of intimacy did not, however, mediate the relationship between sex and benevolent attributions. This lack of mediation is important: it argues against an interpretation based on the plausible, but conceptually uninteresting, possibility that women responded more generously than men toward their friends simply because women were more likely than men to perceive their friends as truly intimate friends (rather than more distant acquaintances). Women did show a greater tendency toward perceptions of intimacy, but this difference was insufficient to account for the finding that women, but not men, responded to their friends in a manner that mimicked exactly their responses to kin. Interestingly, true kinship also produces unique effects on nepotistic behavior over and above indices of intimacy like empathic concern and felt oneness (Korchmaros & Kenny, 2001, Kruger, 2003). Our findings suggest that (consistent with the workings of actual kinship in which genetic relatedness precedes felt intimacy) certain mechanisms elicit psychological kinship, which, in turn, may produce interpersonal intimacy.

So, exactly, what underlying psychological mechanism or mechanisms account for the greater tendency for women to treat their friends as akin to kin? We discuss several possibilities below.