The roles of testosterone and cortisol in friendship formation

Highlights

• Higher testosterone associated with lower closeness.

• Lower cortisol and dynamic cortisol decreases associated with greater closeness.

• Partners of those who had lower cortisol desired more closeness.

Abstract

Although research has investigated the neuroendocrine correlates of romantic relationships, the neuroendocrine correlates of friendship formation are largely unexplored. In two conditions, participants’ salivary testosterone and cortisol were measured before and after a high versus low closeness activity with another same-sex participant. In the high closeness task, participants took turns answering questions that fostered increases in self-disclosure. The low closeness task fostered low levels of self-disclosure. Dyadic multilevel models indicated that lower basal testosterone and decreases in testosterone were associated with increased closeness between recently acquainted strangers. Our results suggest that people high in testosterone felt less close to others and desired less closeness. Further, lower basal cortisol and dynamic cortisol decreases were associated with greater closeness and desired closeness in the high closeness condition. Finally, we found that the partners of those who had lower cortisol desired more closeness. These findings suggest that lower testosterone and cortisol are linked to the facilitation of initial social bonds and that these social bonds may, in turn, be associated with changes in these hormones.

Introduction

Closeness is crucial to friendships and romantic relationships, and the drive to become close to others has been theorized as a fundamental human motivation or need (Aron and Aron, 1986, Baumeister and Leary, 1995). In terms of the self, closeness represents the degree to which aspects of another are included in one’s own identity. This ‘inclusion of other in the self’ (IOS) can result in an expansion of the self-concept, where both members think of themselves as a cognitive unit (Aron and Aron, 1986, Aron et al., 1991, Aron et al., 1992). Recent approaches to the study of closeness in humans are beginning to examine the biological mechanisms underlying interpersonal closeness, particularly through investigating how hormones such as testosterone and cortisol are linked to social bonds (Ellison and Gray, 2009, van Anders and Gray, 2015).

One hormone that is potentially pivotal in the formation and function of close relationships is testosterone, a steroid hormone regulated by the hypothalamic-pituitary-gonadal (HPG) axis (Gettler et al., 2011, Gray et al., 2006, Edelstein et al., 2011, Perini et al., 2012, van Anders and Goldey, 2010). For instance, higher testosterone appears to facilitate sexual relationships and mating (Gettler et al., 2011, Peters et al., 2008, Slatcher et al., 2011, Welling et al., 2008). Outside of short-term mating contexts, however, higher testosterone may be associated with inhibited social connections and lower relationship quality (Edelstein et al., 2014). Indeed, coupled men and women tend to have lower testosterone than singles (Gray et al., 2004, van Anders and Goldey, 2010). Evidence suggests that nurturant behaviors and contexts may contribute to a decline in testosterone (Gettler et al., 2011; Gettler and Oka, 2016, Gray et al., 2004, van Anders et al., 2012, Wingfield et al., 1990). Further, testosterone may impact the quality of an existing romantic relationship. Those with lower testosterone in committed relationships have reported higher relationship satisfaction, and the partners of those with lower testosterone also report higher relationship satisfaction (Edelstein et al., 2014). Additionally, high testosterone individuals have been found to be more avoidant, lonely, and disconnected from others (Turan et al., 2014). These results suggest that higher testosterone may be associated with decreased closeness and desire for closeness.

A majority of research on testosterone and relationships considered thus far indicates that hormones shape social behaviors and interactions. Do social interactions, in turn, shape hormones? For example, cortisol reactivity is a marker of the hypothalamic-pituitary-axis (HPA) response to stress. Cortisol and testosterone both change in response to the social environment. Measuring cortisol and testosterone reactivity increases our understanding of how social interactions shape our hormones. Further, individual levels of hormone reactivity are socially relevant physiological markers beyond basal hormone levels. For instance, nurturant responses to babies are associated with reactive decreases in testosterone (van Anders et al., 2012), and interest in babies is inversely related to testosterone reactivity to sexual stimuli (Zilioli et al., 2016). Further, young men’s testosterone increases after a brief social interaction with a member of the opposite sex (Roney et al., 2007). Together, these findings suggest that testosterone reactivity may be associated with social relationship processes.

Cortisol reactivity may also play a role in interpersonal relationships. Dyadic closeness and relationship quality tend to decrease under stress (Karney and Bradbury, 1995, Lavee and Ben-Ari, 2007). Further, providing social support to a close other and being socially accepted by peers has been shown to decrease cortisol (Blackhart et al., 2007, Smith et al., 2009). Partners in a relationship may even shape each other’s stress response. For instance, evidence indicates that a partner’s attachment style in a romantic relationship may influence the other’s stress reactivity, as measured via salivary cortisol (Laurent and Powers, 2007, Powers et al., 2006). Together, these results point to a potential association between cortisol reactivity and closeness.

Researchers have begun to posit that hormones, rather than acting in isolation, may be part of a multisystem approach where the joint contributions of hormones produced by the HPG and HPA axes work in concert (Bauer et al., 2002, Laurent et al., 2013). From an evolutionary context, there is empirical support indicating an interaction of the HPG and HPA axes in predicting social behaviors (see Salvador, 2012 for a review). In several species, dominant behaviors are closely related to social hierarchies. Studies investigating the role of these two steroid hormones and dominance have primarily focused on non-human animals, but have begun to be researched in humans. For example, evidence suggests that testosterone and cortisol may interact with each other at both the within-person and within-dyad level (Marceau et al., 2015). The dual-hormone hypothesis posits that testosterone’s effect on social behavior is dependent on the steroid hormone cortisol (Mehta and Josephs, 2010, Mehta and Prasad, 2015). A majority of findings related to the dual-hormone hypothesis support the theory that testosterone’s role in social behavior may be moderated by cortisol levels. Terburg et al. (2009) propose a neuro-evolutionary model that posits that social aggression is motivated by an imbalance in the levels of cortisol and testosterone.

Testosterone may have a differential influence on social behaviors depending on the level of cortisol. For example, testosterone is negatively associated with empathy when basal cortisol is low, but positively associated with empathy when basal cortisol is high (Zilioli et al., 2015). In contrast, a recent study found that males with higher testosterone and lower cortisol were more popular and more likely to act as social connectors (Ponzi et al., 2016). Given the prior associations found among testosterone and relationship satisfaction, emotional support, and nurturance, it is worth exploring whether testosterone and cortisol might jointly predict interpersonal closeness.

Growing evidence suggests that testosterone and cortisol are associated with behavior and functioning in social relationships. So far, the evidence suggests that high testosterone facilitates initial sexual and romantic relationships, yet is associated with lower long-term relationship satisfaction. Given the prior findings on links among testosterone and relationship processes, it is worth exploring its role in relationship formation. A systematic starting point is to investigate the role of testosterone during friendship formation. In friendships, testosterone may play a facilitative role, as it does with early-stage romantic love. Alternatively, given the potential inhibitory role testosterone plays in trust, empathy, and relationship satisfaction, (e.g. Bos et al., 2010, Edelstein et al., 2014, Hermans et al., 2006) testosterone may be associated with less closeness in friendly dyadic social interactions. Further, cortisol potentially affects and is potentially affected by interpersonal relationships. One study looked at the association between friendship formation and hormones. Across three laboratory-based intergroup friendship meetings, cortisol was measured before and after a friendship formation activity. Cortisol levels decreased (Page-Gould et al., 2008), indicating the potential for a hormonal response to a laboratory-based closeness induction task. To our knowledge, very few studies to date have examined the role of testosterone or cortisol in interpersonal closeness outside of a romantic context. Further, there is little research on dyadic effects of testosterone or cortisol (i.e., partner effects of one person’s hormone levels associated with the feelings of closeness felt by the other person in a dyad) in interpersonal relationships. Researchers have found that testosterone’s influence on social behaviors such as empathy may depend on concentrations of cortisol (Zilioli et al., 2015). The present study will examine such hormonal effects in the context of friendship formation.

The research described below investigated how testosterone and cortisol are related to closeness and the desire for closeness during friendship formation. A central aim of the present research was to explore the association of basal testosterone and cortisol and testosterone and cortisol changes with closeness and desire for closeness in same-sex dyadic social interactions and whether these interactions occurred as actor effects or partner effects (these effects consider the role each member plays in a dyad and the interaction of the two members in a dyad). Second, we investigated how a closeness induction paradigm modulates testosterone and cortisol changes. Third, we conducted an exploratory analysis to see how testosterone and cortisol might jointly predict closeness. To achieve these aims, we assessed testosterone and cortisol before and after a lab-based high or low closeness induction task.