Physiology and pillow talk

Abstract

This study investigates the association between individual differences in testosterone and communication after sexual activity. Two hundred and fifty-three young adult participants (78% women, M age = 21 years, 73% White) provided saliva samples (later assayed for testosterone) and subsequently, over a 2-week period, completed an online diary after each time they engaged in sexual activity. Individual differences in testosterone levels were inversely associated with perceived benefits of, and positively associated with perceived risks of, disclosing thoughts and feelings to one’s partner after sexual activity. When testosterone levels were higher, post sex disclosures were less intentional and less positive, and these associations were mediated by risk–benefit assessments. An interaction between testosterone and orgasm revealed that higher testosterone levels were associated with more negative post sex disclosures for those who did not orgasm, but not for those who experienced orgasm. This finding suggests that high testosterone/no orgasm individuals may be the least likely to experience the beneficial effects of post sex communication. Similar results were found both when biological sex was controlled for and when analyses were conducted separately for women and men. Implications for a biosocial model of post sex behavior and communication are discussed.

Keywords

Communication privacy management theory disclosure oxytocin pillow talk post sex communication steroid/peptide theory of social bonds testosterone

Communication serves an important function in sexual and relational satisfaction in romantic relationships (Byers, 2005; Byers & Demmons, 1999; Cupach & Comstock, 1990), but only recently have scholars turned their attention to the behaviors occurring immediately after sexual activity (e.g., Denes, 2012; Denes & Afifi, 2014; Kruger & Hughes, 2010, 2011; Hughes & Kruger, 2011; Muise, Giang, & Impett, 2014). Although the importance of post sex behavior has been noted for decades (e.g., Halpern & Sherman, 1979), work investigating the post coital time interval (Kruger & Hughes, 2010) has helped to revive the “science of post-coitus” (Ryder, 2012). This revival has resulted in greater acknowledgment of the importance of post sex communication in relational functioning. The behaviors immediately following sexual activity have important effects not only on the sexual episode, but also on couples’ relationship satisfaction (Muise et al., 2014). To date, studies provide important insights into the role of post sex behavior in relationship functioning, but the field has yet to investigate the physiological factors that might promote or restrain post sex bonding.

Decades of reproductive behavioral endocrinology research suggests that studying the physiological correlates of post sex communication could advance researchers’ understanding of close relationships. Studies with humans and other animals (see Nelson, 2011, for review) suggest that individuals may be inclined toward certain communication tendencies because of either intrinsic trait-like or acute state-specific or maturational changes in testosterone (T), a primary product of the activity of the hypothalamic–pituitary–gonadal axis. Communication and relationship scholars have explored T’s associations with affectionate communication, marriage, and parenting/birth (see Floyd, 2004, for review), indicating the importance of the hormone in understanding relationships and communication behavior. Such research suggests that lower T levels facilitate more social and less dominant behavior, as well as more nurturant behavior (Harris, Rushton, Hampson, & Jackson, 1996; van Anders, Goldey, & Kuo, 2011). It is also important to note that men generally have higher T levels than women (Nelson, 2011). Furthermore, several studies suggest that T may be positively associated with orgasm and sexual activity (e.g., Dabbs & Mohammed, 1992; Exton et al., 1999; van Anders, Hamilton, Schmidt, & Watson, 2007), making it a particularly relevant hormone in the context of post sex communication.

The relationship between T and pair-bond formation has also been the subject of recent theorizing. The steroid/peptide (S/P) theory of social bonds (van Anders et al., 2011) states that certain biological components work together to influence the formation of social bonds. Social bonds are also likely formed through communication between partners, and it is possible that certain hormones promote or restrain such communication. Using communication privacy management (CPM) theory (Petronio, 2000, 2002) in conjunction with S/P theory suggests that certain hormones may be associated with more permeable privacy boundaries, which allow individuals to disclose to their partners and form social bonds after sexual activity. Exploring the association between physiology and post sex communication provides a richer understanding of the many factors that influence decisions to disclose and the biological predispositions individuals bring to interpersonal interactions.

Post sex communication and T

Research on behavior after sexual activity has investigated a range of verbal and nonverbal communication that occurs post sex, including sleeping, snacking, and so on (Hughes & Kruger, 2011). Research on behavior after sexual activity has focused on the post coital time interval (Kruger & Hughes, 2010). However, we use the term post sex time interval (PSTI), as coitus privileges heterosexual penetrative sex and we aim to be inclusive of a range of sexual orientations and sexual practices.

Communication about sex and the communication that occurs during and after sexual activity have been linked to increased sexual and relationship satisfaction (Byers, 2005; Byers & Demmons, 1999; Cupach & Comstock, 1990; Muise et al., 2014). Of importance in the present investigation is the role of communication, and self-disclosure in particular, after sexual activity. Self-disclosure includes a wide range of topics and is not focused solely on sexual disclosures, such as revealing sexual desires/preferences or sexual health status. Rather, self-disclosure more broadly involves “any information exchange that refers to the self, including personal states, dispositions, events in the past, and plans for the future” (Derlega & Grzelak, 1979, p. 152). An important part of sexual activity involves the communication that occurs immediately afterward, during “pillow talk” (Denes, 2012). During this time, individuals experience heightened emotion and intimacy, which may facilitate pair-bonding behaviors such as disclosing feelings and sharing emotions with one’s partner (Halpern & Sherman, 1979; Hughes & Kruger, 2011; Veenestra, 2007). Post sex communication may therefore serve an important role in not only sustaining a satisfying sexual relationship between partners, but may also contribute to more general feelings of relationship satisfaction and closeness (Denes, 2015; Muise et al., 2014).

Prior research on pillow talk reveals that orgasm is an important predictor of post sex communication. Individuals who orgasm perceive greater benefits to disclosing than those who do not orgasm and are more likely to disclose positive feelings to their partners after sexual activity (Denes, 2012; Denes & Afifi, 2014). Compared to individuals who do not orgasm, those who orgasm are also more likely to disclose information of greater depth to their partners immediately after sexual activity (Denes & Afifi, 2014). Additionally, quantity of post sex affectionate behavior is positively associated with relationship satisfaction, and research reveals that this association is mediated by sexual satisfaction (Muise et al., 2014). Several of the above findings have also revealed sex differences in these associations, as women who orgasm disclose more positive feelings after sexual activity than men who orgasm (Denes, 2012), and men are more likely to report experiencing orgasm from sexual activity compared to women (Denes, 2012; Denes & Afifi, 2014). Denes and colleagues (Denes, 2012; Denes & Afifi, 2014) argue that the physiological experience of orgasm may play an important role in such communication behavior and underlie disclosure decisions during the PSTI, though physiological associations with pillow talk have yet to be tested. Thus, the present study makes a unique contribution to the growing line of research exploring post sex communication by investigating the association between T and pillow talk.

T is an important hormone to consider when investigating communication after sexual activity, as it has been linked to both sexual behavior and communication in close relationships (e.g., Bancroft, Sanders, Davidson, & Warner, 1983; Exton et al., 1999; Morris, Udry, Khan-Dawood, & Dawood, 1987; van Anders et al., 2007). Research investigating the association between T and communication behavior reveals that the hormone is related to husbands’ and wives’ social support (Cohan, Booth, & Granger, 2003). T has also been linked to sexual behavior. For example, one study investigated women’s and men’s T levels in the early evening and late evening on both “sex” and “nonsex” days (Dabbs & Mohammed, 1992). Although individuals’ early evening T levels were not significantly different on days when they did or did not have sex, both women’s and men’s late evening T levels were significantly higher on the days they had sex (which occurred in between the two time points). T has also been linked to orgasm, though the results are inconsistent. Several studies reveal a positive association between T and orgasm frequency and post orgasm T increases for men (e.g., Fox, Ismail, Love, Kirkham, & Loraine, 1972; Knussmann, Christiansen, & Couwenbergs, 1986 ). When investigating women, T is also positively associated with orgasm (i.e., women who orgasm have significantly higher T levels than women who do not orgasm) and T increases in women post orgasm (Exton et al., 1999; van Anders et al., 2007). Other studies, however, find no correlation between sexual behavior and women’s and men’s T levels (e.g., Kruger et al., 1998; Lee, Jaffe, & Midgley, 1974; Mantzoros, Georgiadis, & Trichopoulosm, 1995). For example, contrary to the findings above, van Anders, Hamilton, Schmidt, and Watson (2007) suggest that orgasm does not lead to increased T levels, as the change between pre- and post-T levels did not significantly differ for women in their study who did versus did not orgasm.

What is clearer is the association between T and social behavior. In general, individuals with lower levels of T demonstrate less aggressive, less dominant, and more social behaviors than individuals with higher levels of T (Harris et al., 1996; Mazur & Booth, 1998; Nelson, 2011). Taylor et al. (2002) also suggest that T suppresses the effects of oxytocin (OT), which is a hormone positively associated with social behavior, pair-bonding, affiliation, and intimate physical activity. OT is known to increase from baseline levels after sexual arousal and is linked to a number of prosocial behaviors (Blaicher et al., 1999; Carmichael et al., 1987). For example, OT is associated with increased perceptions of trust, enhanced social recognition abilities, increased ability to read nonverbal cues, enhanced approach behavior, decreased aggressive behavior, and decreased threat perception (Domes, Heinrichs, Michel, Berger, & Herpertz, 2007; Guastella, Mitchell, & Dadds, 2008; Huber, Veinante, & Stoop, 2005; Kirsch et al., 2005; Kosfeld, Heinrichs, Zak, Fischbacher, & Fehr, 2005; Lim & Young, 2006)¹. Thus, increases in OT due to sexual activity may facilitate greater pair-bonding between partners post sex and produce feelings of a “safe haven” for disclosing (Denes, 2012). Nevertheless, because T might suppress the effects of OT, it is possible that higher levels of T may restrain post sex communication and the positive effects of OT.

S/P theory of social bonds

The S/P theory of social bonds (van Anders et al., 2011) offers one framework for understanding the interpersonal influence of T. S/P theory focuses on the role of various physiological components involved in the formation of social bonds and the contexts that inhibit or promote these bonds from forming. The theory centers on three hormones: T, OT, and vasopressin—the first of which is the center of the present investigation. van Anders, Goldey, and Kuo (2011) define pair-bonds as “long-lasting affiliations involving intimacy, sexual contact, preferential proximity, and emotional attachment with relative exclusivity” (p. 1266). Similar to affection exchange theory (Floyd, 2006), S/P theory focuses on the adaptiveness of forming pair-bonds as a means by which evolutionary fitness is promoted. While van Anders et al. (2011) note that pair-bonds may be formed in a multitude of ways (e.g., multipoint bonds in which individuals form pair-bonds with multiple partners), van Anders et al. (2007) explain that two-point bonds may be hormonally unique from other types of pair-bonds. Two-point bonds refer to pair-bonds between two individuals, as opposed to multipoint bonds formed with multiple partners. Thus, the focus of the present analysis is on two-point bonds between sexual partners.

Pair-bonds can form in a multitude of relational types (see van Anders et al., 2011). van Anders et al. (2011) explain that pair-bonds may function through two physiological pathways: “a nurturant system evolved to support parent-offspring bonds and infant survival, and a sexual system evolved to support reproduction” (p. 1267). Pivotal to these pathways is the role of T, as it has been linked to both sexuality and nurturance, demonstrating a positive relationship with the former and a negative relationship with the latter (see van Anders et al., 2011, for review). S/P theory suggests that lower T levels facilitate nurturant behaviors (or loving warm contact), which can be implemented through such behaviors as pair-bonding and comforting. In terms of communication behaviors, it is possible that lower T levels will also be associated with increased perceptions of the benefits of disclosing and decreased perceptions of the risks of disclosing to one’s partner post sex, as disclosing may ultimately be perceived as a way to express nurturance to one’s partner. Disclosing personal information is often a primary way in which people establish and maintain close relationships (see Petronio, 2002).

S/P theory provides a useful framework for investigating physiological correlates of post sex communication, as it directly addresses sexual intimacy in relationships. The theory builds upon the challenge hypothesis (Wingfield, Hegner, Dufty Jr., & Ball, 1990), which posits that high T is linked to sexual activity, while low T is linked to social bonding. As van Anders et al. (2011) point out, however, the hypothesis “does not address links between sexuality and pair bonds directly … [though] sexual activity facilitates pair bonds” (p. 1269). The authors explain that an “intimacy paradox” (van Anders & Watson, 2006; van Anders et al., 2007) exists in which T increases due to sexual activity, but T can also inhibit bonding, even though sexual activity is thought to promote bonding.

S/P theory addresses this paradox by suggesting that two types of intimacy exist: sexual intimacy and nurturant intimacy, which align with the sexual and nurturant systems detailed above. Consensual sexual activity involves both forms of intimacy (van Anders et al., 2011), though sexual intimacy may be especially salient when engaging in sexual acts with one’s partner. However, because the PSTI is a time of high emotions and bonding immediately following sexual activity (Halpern & Sherman, 1979; Veenestra, 2007), it is also a likely context where nurturant intimacy is high. Together, then, S/P theory suggests that T levels should predict social bonding and intimacy in the context of sexual activity. One way that individuals may communicate intimacy and facilitate pair-bonding is through pillow talk. Sexual partners may disclose to their partners immediately after sexual activity as a way of reinforcing an intimate bond, but such communication may vary based on T levels. As S/P theory suggests, the lower the individuals’ T levels, the more likely they are to engage in nurturant behaviors such as pillow talk. While the biological basis of S/P theory is important for addressing the physiological correlates of communication, communication theory also contributes to understanding the disclosure process that likely occurs after sexual activity.

Communication privacy management

Several theories of communication have addressed the disclosure decision process and the ways that individuals manage private information (e.g., Afifi & Steuber, 2009; Altman & Taylor, 1973; Petronio, 2002). CPM theory (Petronio, 1991, 2000, 2002) is one such theory that addresses the ways that individuals reveal or conceal information. CPM theory focuses on privacy boundaries, explaining that individuals feel ownership over private information and establish different boundaries around private versus public information (Petronio, 2000, 2002). Boundary structures are formed around information and determine who is allowed access to the information, with private information contained within a thick, or more rigid, boundary, and public information contained within a thin, or more permeable boundary. Boundary permeability is partially determined based on the risk of revealing the information. Risky information tends to be contained within thick boundaries, while less risky information is contained within thin boundaries (Petronio, 2002).

According to CPM theory (Petronio, 2000, 2002) and other models of disclosure (e.g., Afifi & Steuber, 2009), individuals assess the risks and benefits of disclosing before making the decision to reveal or conceal information. Risks of disclosure may include loss of control, embarrassment, hurt, or rejection (Baxter & Montgomery, 1996), while benefits may include increased intimacy, closeness, and satisfaction. CPM theory provides a useful framework for understanding the process by which individuals make the decision to disclose, though it is possible that individuals’ biology also influences communication tendencies. The physiological effects of orgasm (i.e., the surge of OT that accompanies orgasm) could make individuals’ privacy boundaries more permeable. Conversely, because high overall levels of T are associated with decreased pair-bonding (as suggested by van Anders et al., 2011) and less social behavior, individuals with higher levels of T may also have more rigid privacy boundaries. In the context of post sex communication, this might translate into individuals who generally have higher T levels assessing greater risks and fewer benefits to disclosing after sexual activity, which could, in turn, influence various dimensions of their post sex disclosures. Although a multitude of dimensions of self-disclosure have been explored in previous literature (e.g., Altman & Taylor, 1973), four dimensions have been chosen as the focus of the present investigation.

Dimensions of disclosure

The current manuscript investigates four dimensions of post sex disclosures: intentionality, amount, magnitude, and valence (in line with Denes & Afifi, 2014). Intentionality focuses on the reasoning behind the communication, and intentional acts are those which involve control, planning, and executing behavior to bring about a desired outcome (Brunner, 1973; Denes & Afifi, 2014; Poulin-Dubois & Shultz, 1988 ; Shultz, 1980). Conversely, unintentional behavior is more mechanical, and unintentional disclosures occur when individuals behave in ways contradictory to their intended plans for managing the information (Denes & Afifi, 2014; Malle, Moses, & Baldwin, 2001). The second dimension of disclosure—magnitude—aligns with Petronio’s (2000, 2002) conceptualization of the thickness of privacy boundaries and Altman and Taylor’s (1973) idea of depth. Information of greater magnitude is that which is more personal or private and is contained within a thicker privacy boundary. The amount of information revealed is the third dimension of disclosure, which considers the breadth of information disclosed (Altman & Taylor, 1973). Finally, the valence of disclosures is investigated. Positively valenced information is that which conveys positive sentiment or emotion (Denes, 2012), while negatively valenced information communicates more unfavorable attitudes and feelings.

Research on post sex communication suggests that individuals vary in their disclosures along these dimensions. Denes and Afifi (2014) found that women and men who experienced orgasm disclosed information more intentionally, disclosed information of greater magnitude, and disclosed more positively valenced information than individuals who did not orgasm. Taken together with the predictions of S/P theory and CPM theory, it is likely that T levels predict the risks and benefits of disclosing after sexual activity as well as the dimensions of disclosure. Because individuals with lower T levels are more nurturing and demonstrate more social behaviors (Harris et al., 1996; van Anders et al., 2011), it should follow that these individuals would also see greater benefits and fewer risks to disclosing to their partners after sexual activity. Furthermore, because OT can create feelings of a safe haven, and T may suppress the effects of OT (Taylor et al., 2002), individuals with lower T levels may feel the effects of OT after sexual activity more strongly than individuals with higher T levels.

Individuals with lower T levels may therefore share information more intentionally post sex, aware of the potential benefits of disclosing during this highly emotional time. Denes and Afifi (2014) found that individuals who experienced orgasm disclosed more intentionally post sex than individuals who did not orgasm. They explained that orgasm may provide individuals with a heightened perceptual ability that allows them to see the connection between their behavior (e.g., disclosing thoughts and feelings to one’s partner during pillow talk) and achieving a desired outcome (e.g., building closeness and intimacy with one’s partner). Similarly, individuals with lower T levels may be better able to see the benefits of disclosing (and also see fewer risks to disclosing), and therefore communicate in incredibly intentional ways to achieve an end goal of pair-bonding. Relatedly, individuals with lower T levels may perceive greater benefits to disclosing, and therefore disclose a greater amount of information, disclose information of greater magnitude, and disclose more positively valenced information. Furthermore, risk–benefit assessments should mediate the relationship between T and the dimensions of disclosure. Finally, because orgasm has previously been linked to post sex disclosure processes (Denes & Afifi, 2014), and orgasm is also associated with T levels (Fox et al., 1972; Knussmann et al., 1986), two research questions are posed regarding the potential interaction between T levels and orgasm. With this information in mind, the following hypotheses and research questions are presented:
H1: Individuals’ T levels will negatively predict their perceived benefits of disclosing and positively predict their perceived risks of disclosing, such that the higher the individuals’ T levels, the less they should perceive benefits and the more they should perceive risks to disclosing during the PSTI.

H2: The lower the individuals’ T levels, the more likely they are to (i) disclose intentionally, (ii) disclose a greater amount of information, (iii) disclose information of greater magnitude, and (iv) disclose more positively valenced information, as mediated by risk–benefit assessments.

RQ1: Does T interact with orgasm to predict the risks or benefits of disclosing?

RQ2: Does T interact with orgasm to predict individuals’ scores on the dimensions of disclosure (i.e., intentionality, magnitude, amount, and valence)?

Method

Participants

The sample included 253 individuals. Participants ranged in age from 18 to 45 years (M = 21.02, SD = 3.15). The sample consisted of 78% (n = 197) women and 22% (n = 56) men. Nearly the entire sample identified as heterosexual (n = 246), with five individuals identifying as bisexual. The sample was predominantly White (n = 185), though participants also identified as Latina/o (n = 28), Asian (n = 25), and mixed race (n = 8). Most participants identified as single/unmarried (n = 245); several participants indicated they were married (n = 5) or divorced (n = 3). The majority of participants indicated that they were in committed/monogamous relationships (70%; n = 164). The remainder of the sample identified as being in open/casual relationships (30%; n = 70). The average length of their relationships was 21 months (range = 0–144 months).

Procedure

Participants were gathered from two large public universities. At one university, students were recruited through a department-wide online system that allows students to register for research studies that count toward their course research requirement. At the other university, students were recruited through class announcements. They were informed that they must be in a current sexual relationship (i.e., engaging in sexual activity with another individual) to be eligible for the study, and were compensated for their time with course research credit or extra credit. Participants were told that sexual activity involved any activity “below the belt,” which could include penile–vaginal intercourse, oral sex, anal sex, or hand stimulation with a partner. If participants met this criterion, they registered for a time slot to come to the communication laboratory to provide saliva samples and complete a pre-survey. All saliva samples were collected in a 3-hour time frame between 2 p.m. and 5 p.m. to avoid capturing fluctuations in hormone levels that happen at different hours of the day. Participants were instructed not to eat or drink 2 hours prior to coming to the lab, not to consume alcohol, smoke, or exercise the day of their lab visit, and not to brush their teeth an hour before coming to the lab or go to the dentist 48 hours before coming to the lab, as all these activities could interfere with an accurate saliva sample. Upon arriving at the lab, participants were asked if they ate or drink within the previous 2 hours. If they did, participants were instructed to drink 8 ounces of water and wait for 15 minutes before proceeding with the study. Participants were excluded from the study if they were taking any medications or drugs (e.g., steroids or hormone replacement therapies) or had an illness that could influence their hormone levels.

Once they arrived at the laboratory, the procedures were explained to participants and consent was obtained. Participants were told that their saliva would be analyzed for a certain hormone (though they were not told exactly which hormone) and were invited to contact the primary investigator if they wanted more information about the study and a copy of the results upon completion of the project. Following Granger, Kivlighan, El-Sheikh, Gordis, and Stroud (2007b), participants were instructed to rinse their mouths out with water, wait 10 minutes, and provide 1 mL of whole saliva by passive drool method. Forty-five minutes later, after completing a pre-survey, participants provided a second saliva sample. Participants took approximately 3 to 5 minutes to fill the vial/tube. The saliva was frozen immediately until it was shipped overnight on dry ice to the Center for Interdisciplinary Salivary Bioscience Research (CISBR) at Johns Hopkins University (JHU) for assay.

Participants were then sent a link to an online diary hosted on SurveyMonkey.com. They were instructed to log on to the online diary website and complete a diary after every time they engaged in sexual activity with their partner over the next 2 weeks. Participants were informed (verbally and in writing) that sexual activity included anything below the belt, and they were asked to complete the diary within 2 hours of the sexual activity. This time frame was chosen to ensure that participants completed the survey as close to the time of the sexual activity as possible and to lessen the chances of participants engaging in other activities such as sleep (Denes, 2012; Denes & Afifi, 2014).

Measures

Nature of sexual activity

In each diary entry, participants were asked to indicate whether or not they experienced orgasm (recorded as “yes,” “no,” or “other—please specify”). They were also asked about the nature of their sexual activity (i.e., participants were asked to indicate the type of sexual activity they engaged in). Participants reported engaging in sexual intercourse (M = 3.39, SD = 2.38) and/or other sexual activity (M = 3.51, SD = 3.20) several times a week. All reported sexual activity involved a partner, and no participants reported engaging in solo sexual activity (e.g., masturbation).

Dimensions of disclosure

In each diary entry, participants were asked to complete items assessing the four dimensions of disclosure in the context of sexual activity (Denes, 2012; Denes & Afifi, 2014). Four items assessed the intentionality of individuals’ disclosures (e.g., “I didn’t mean to say the things I said to my partner”; “I wanted to tell my partner the things I said”). Higher scores on the intentionality measure indicated that individuals were less intentional/more unintentional in their disclosures. The Cronbach’s α for the intentionality scale was .75. Three items assessed the magnitude of individuals’ disclosures (e.g., “I told my partner things that few people know”; “I told my partner personal things”). The reliability for the magnitude scale was .83. Two items assessed the amount of information disclosed compared to normal communication patterns (e.g., “I disclosed more to my partner after sexual activity than I normally would”). The reliability for the amount scale was .86. Lastly, 4 items assessed the valence of individuals’ disclosures (e.g., “I expressed some positive feelings for my partner to him/her”; “I told my partner some negative thoughts I’ve been having about him/her”). Higher scores on the valence measure indicated more positively valenced disclosures. The Cronbach’s α for the valence scale was .75. A 5-point Likert-type scale was used (with 1 being strongly disagree and 5 being strongly agree) for all four measures.

Risk–benefit assessments

Participants were also asked to respond to 7 items in each diary assessing perceptions of the risks and benefits of disclosing to their partners. An adapted version of Denes and Afifi’s (2014) scale that captures individuals’ perceived risks and benefits of disclosing was used in the present study (e.g., “I felt safe disclosing how I felt about my partner to him/her”; “I saw some danger in expressing my feelings to my partner”). A 5-point Likert-type scale was used (with 1 being strongly disagree and 5 being strongly agree). These items loaded strongly onto two latent constructs (loadings of .61–.82). The Cronbach’s α was .84 for the benefit scale and .92 for the risk scale.

Testosterone

Immediately after collection, samples were frozen and then stored at −20°C for approximately 3 months until they were shipped on dry ice overnight to JHU CISBR for analysis. Saliva samples were assayed for T using a commercially available immunoassay without modification to the manufacturer’s recommended protocol (Salimetrics, Carlsbad, CA). Test volume was 25 µL with a range of sensitivity from 1 to 600 pg/mL. All saliva samples were tested in duplicate and intra- and inter-assay coefficients of variation were on average less than 10% and 15%, respectively. The correlation for T between the first and second saliva samples was r = .99, p < .01, and T values were averaged across saliva samples. The distribution of T values was positively skewed, and thus raw scores were subject to logarithm (Ln) transformation (Granger et al., 2007a). Ln scores were used in all analyses, but raw scores (pg/mL) are reported in Tables 1 –3 when noting the means for T for ease of understanding. As expected, women had lower T levels (M = 46.99 pg/mL, SD = 17.58) than men (M = 129.15 pg/mL, SD = 38.48), t(251) = 22.69, p < .01.

Table 1.
Correlations, means, and standard deviations for all continuous variables across the six diaries for the entire sample.

Variables M SD T Risk Benefit Intentionality Amount Magnitude Valence

T 65.18 41.61 –

Risk 1.96 0.97 .22 –

Benefit 3.83 0.77 −.15 −.43 –

Intentionality 1.80 0.64 .18 .46 −.51 –

Amount 2.23 0.95 .09 .26 .06 .17 –

Magnitude 2.91 1.00 −.06 −.16 .41 −.26 .39 –

Valence 4.15 0.72 −.13 −.54 .51 −.61 −.07 .31 –

Note. T = testosterone; M = mean; SD = standard deviation. All variables are measured on a scale from 1 to 5 except for T.

p < .01.

Table 2.
Women’s correlations, means, and standard deviations for all continuous variables across the six diaries.

Variables M SD T Risk Benefit Intentionality Amount Magnitude Valence

T 46.99 17.58 –

Risk 1.93 0.97 .23 –

Benefit 3.86 0.75 −.06 −.41 –

Intentionality 1.77 0.62 .08 .43 −.48 –

Amount 2.21 0.94 .004 .25 .04 .19 –

Magnitude 2.95 0.99 .05 −.13 .36 −.20 .37** –

Valence 4.18 0.73 −.11* −.56 .51 −.60 −.08 .27 –

Note. T = testosterone; M = mean; SD = standard deviation. All variables are measured on a scale from 1 to 5 except for T.

**p <* .05; p < .01.

Table 3.
Men’s correlations, means, and standard deviations for all continuous variables across the six diaries.

Variables M SD T Risk Benefit Intentionality Amount Magnitude Valence

T 129.15 38.48 –

Risk 2.09 0.99 .41 –

Benefit 3.64 0.81 −.18 −.47 –

Intentionality 1.96 0.70 .28 .55 −.58 –

Amount 2.30 0.97 .33 .28 .17 .07 –

Magnitude 2.74 1.06 −.09 −.26* .56* −.42 .49 –

Valence 4.06 0.66 −.22* −.45 .55 −.64 −.002 .45 –

Note. T = testosterone; M = mean; SD = standard deviation. All variables are measured on a scale from 1 to 5 except for T.

**p <* .05; **p < .01

Data analysis

This study used a longitudinal design to explore patterns of post sex communication over a 2-week time frame. Multilevel modeling (MLM) was used to capture within-person fluctuation, which “refers to more undirected variation over repeated assessments and is seen in contexts in which one would not expect any systematic change, and in which time is simply a way to obtain multiple observations per person (rather than serving as a meaningful index of a causal process)” (Hoffman, 2015, p. 8). This is in opposition to within-person change, which predicts that time has a meaningful effect on the variables of interest. Although orgasm (the moderating variable) was expected to vary at each time point (T was only measured at one time point), there was no expected linear or quadratic effect of time on individuals’ responses across the diaries. Model testing confirmed that there was no significant effect for time (explained in more detail below). Within-person fluctuation does not assume that time is a meaningful predictor, but instead uses other time-varying predictors to assess patterns of behavior. In the present study, T and orgasm (as a moderator) are expected to influence risk–benefit assessments and the dimensions of disclosure.

Using MLM allows for the analysis of between-persons and within-person differences. Between-persons differences focus on differences between individuals or interindividual variation. Between-persons variation is time invariant and considered “macro” or Level 2 analyses (Hoffman, 2015). Within-person variation captures differences within one individual over the course of the 2 weeks and is therefore time variant, or micro/Level 1 analyses (Hoffman, 2015). Diary entries capture Level 1 data, which are nested within Level 2 data. In the current study, orgasm, risk–benefit assessments, and the dimensions of disclosure were considered Level 1 data, which were assessed in each diary. These variables were embedded within individuals and their T scores, which were Level 2 variables.

Additional analyses

The studies cited throughout the literature review suggest a number of similarities in women’s and men’s sexual behaviors and pillow talk. An analysis of previous meta-analyses reveals that sex differences in communication are overestimated, and sex (i.e., female/male) accounts for only 1% of the variance in individuals’ communication (see Canary & Hause, 1993). Therefore, we approached the present analyses without the assumption of sex differences and therefore did not predict differences in women’s and men’s associations between T and post sex behavior. However, previous research (e.g., Denes, 2012; Nelson, 2011) has revealed differences in women’s and men’s T levels and post sex communication (when investigated separately). Therefore, for the sake of thoroughness and to rule out alternative explanations, we conducted additional follow-up analyses to explore potential differences in women’s and men’s post sex communication (by (1) controlling for sex, (2) testing women’s and men’s associations separately, and (3) testing the associations using T scores standardized for women and men separately).

Results

Data preparation and preliminary results

During the 2-week period of the study, participants completed between 1 and 19 diaries (M = 2.65, SD = 1.88). In total, 637 diaries were completed by participants. The missing data generated from the few people who entered 19 diaries caused issues with model convergence. Participants completed three diaries on average; however, the MLM file is structured to include potential data for the maximum number of diaries (which was designated as 19, as that was the maximum number of diaries a participant completed). However, for individuals who, for example, only completed three diaries, this resulted in 16 missing diary entries. This large amount of missing data made it difficult for the models to converge. Therefore, a maximum of six diary entries were retained for each participant, which only affected six participants’ data. The data obtained from the diary entries were analyzed using MLM using SPSS 18.0 MIXED. As a first step, unconditional models or empty models (with no predictors) were estimated. Next, a series of model comparisons took place using restricted maximum likelihood estimation to find out which alternative covariance structure (ACS) models best fit the data. The ACS model analyses revealed that a compound symmetry model should be used to estimate the variances and covariances for benefit assessments and the disclosure dimensions of intentionality, magnitude, and amount. The analyses also revealed that a Toeplitz (TP) model should be used for estimating risk assessments and a TP heterogeneous model should be used for estimating the valence of the disclosures.

Descriptive statistics and correlations for all variables for the entire sample are presented in Table 1. Tables 2 and 3 contain the descriptive statistics and correlations for women and men separately. Intraclass correlations were calculated from an empty means random intercept model for all dependent variables. These results are presented in Table 4. The percentages presented in Table 4 indicate the amount of variance that is due to either between-persons or within-person differences for each dependent variable. Because time was not found to be meaningful in any of the models, as indicated by the fixed linear time random intercept model, it was not included in any of the analyses.

Table 4.
Intraclass correlations.

Dependent variable Between-persons differences, % Within-person differences, %

Benefit 59 41

Risk 61 39

Intentionality 50 50

Amount 58 42

Magnitude 42 58

Valence 37 63

Note. Percentages indicate the amount of variance that is due to either between-persons or within-person differences.

Primary analyses

H1 predicted that the higher the individuals’ T levels, the less they would perceive benefits and the more they would perceive risks to disclosing during the PSTI. The results revealed a significant main effect between T levels and benefit assessments (β = −.24, p < .01). As individuals’ T levels increase, their assessments of the benefits of disclosing are expected to be significantly lower. After controlling for sex (female/male), this relationship was approaching significance (β = −.21, p = .09). The results also revealed a significant main effect between T levels and risk assessments (β = .32, p = .01). As individuals’ T levels increase, their assessments of the risks of disclosing are expected to be significantly higher. After controlling for sex, this relationship was still significant (β = .60, p < .01). H1 was largely supported.

H2 predicted that the lower the individuals’ T levels, the more likely they would be to (i) disclose intentionally, (ii) disclose a greater amount of information, (iii) disclose information of greater magnitude, and (iv) disclose more positively valenced information. First, intentionality was tested. Results revealed a significant main effect for T and intentionality (β = .21, p < .01). The lower the individuals’ T levels, the more intentional they were in their disclosures during the PSTI. After controlling for sex, this relationship was still significant (β = .26, p < .01).

Next, the amount of information disclosed was tested. There was no significant finding for the dimension of amount, even after controlling for sex. The magnitude of the disclosures was explored next. Similarly, there was no significant finding for the dimension of magnitude, even after controlling for sex. Lastly, the valence of the disclosures was tested. Results revealed a significant main effect for T and valence (β = −.18, p = .01). The lower the individuals’ T levels, the more positively valenced information they disclosed during the PSTI. After controlling for sex, this relationship was still significant (β = −.26, p < .01).

Additionally, separate analyses were conducted for women and men, though they revealed few differences. For women, the main effects between T and risk assessments (β = .54, p < .01), intentionality (β = .21, p = .03), and valence (β = −.23, p = .03) remained significant. Similar to the findings above, there were also no significant main effects for amount or magnitude. However, unlike the results above, there was no significant main effect between T levels and benefit assessments (β = −.18, p = .19). For men, the main effects between T and risk assessments (β = 1.04, p < .01) and intentionality (β = .58, p < .05) remained significant, though the valence model would not converge. Similar to the findings above, there were also no significant main effects for amount or magnitude. However, unlike the results above, there was no significant main effect between T levels and benefit assessments (β = −.44, p = .21).

Lastly, as done in previous work, analyses were conducted using T scores standardized separately for women and men (i.e., women’s scores were centered on the mean for women, and men’s scores were centered on the mean for men; Mehta & Josephs, 2010). The results were similar to the original analyses. The main effects between T and risk assessments (β = .59, p < .01), intentionality (β = .26, p < .01), and valence (β = −.25, p = .01) remained significant. Similar to the findings above, there were also no significant main effects for amount or magnitude. However, unlike the results above, there was no significant main effect between T levels and benefit assessments (β = −.20, p = .11).

Mediation analyses

To test for mediation, bootstrapping was used to estimate the indirect effects between the predictor variable (i.e., T) and the outcome variable (i.e., intentionality, amount, magnitude, and valence) through the mediating variable (i.e., risk–benefit assessments). Four separate models for intentionality, amount, magnitude, and valence were tested using Preacher and Hayes’ (2008) bootstrapping (indirect) macro in SPSS, which estimates the total, direct, and indirect effects of a predictor variable on an outcome variable through the proposed mediating variable. Because the above analyses revealed that the associations between T, risk–benefit assessments, and the dimensions of disclosure are similar for women and men, the indirect effects models were tested using the entire sample. All four dimensions of disclosure were tested, as indirect effects testing using bootstrap analysis does not assume an initial significant total effect (Preacher & Hayes, 2004). Mediation has occurred when the indirect effects are significant (Mallinckrodt, Abraham, Wei, & Russell, 2006). Given the limitations of testing mediation in MLM models investigating within-person fluctuation and models with time-invariant predictors, mediation models were tested using individuals’ first set of diary responses.

Indirect effects were tested with 5,000 bootstrap replications and a 95% bias corrected and accelerated bootstrap confidence interval (CI). Bootstrap estimates indicated that the hypothesized indirect effect on intentionality was significant for benefit assessments, (95% CI [.03, .18], p < .05), and for risk assessments, 95% CI: [.03, .16], p < .05. Because zero is not included in the 95% CI, it can be concluded that the indirect effect is significantly different from zero (Preacher & Hayes, 2004). Bootstrap estimates indicated that the hypothesized indirect effect on amount was not significant for benefit assessments, 95% CI: [−.08, .02], p = not significant (n.s.), but was significant for risk assessments, 95% CI: [.02, .16], p < .05. Next, bootstrap estimates indicated that the hypothesized indirect effect on magnitude was significant for benefit assessments, 95% CI: [−.23, −.03], p < .05, and for risk assessments, 95% CI: [−.14, −.02], p < .05. Lastly, bootstrap estimates indicated that the hypothesized indirect effect on valence was significant for benefit assessments, 95% CI: [−.21, −.03], p < .05, and for risk assessments, 95% CI: [−.22, −.04], p < .05. H2 was supported for the dimensions of intentionality, magnitude, and valence, and partially supported for the amount of information disclosed. The full results of the bootstrap analyses are presented in Figures 1 –4.

Figure 1.
Model of effect of T on intentionality through risk–benefit assessments. The indirect effect was significant for benefit assessments (β = .10, SE = .04, p < .05, CI: [.03, .18]) and risk assessments (β = .09, SE = .03, p < .05, CI: .03, .16). The value outside of the parentheses represents the direct effect, from bootstrapping analyses, of IV on DV after the mediating variable is included (c′-path). The value in parentheses represents the total effect of IV on DV prior to the inclusion of the mediating variable (c-path). p < .05; p < .01; p < .001. Note. T = testosterone; SE = standard error; CI = confidence interval; n.s. = not significant; IV = independent variable; DV = dependent variable.

Figure 2.
Model of effect of T on amount through risk–benefit assessments. The indirect effect was not significant for benefit assessments (β = −.01, SE = .02, p = n.s., CI: [−.08, .02]), but it was for risk assessments (β = .07, SE = .03, p < .05, CI: [.02, .16]). The value outside of the parentheses represents the direct effect, from bootstrapping analyses, of IV on DV after the mediating variable is included (c′-path). The value in parentheses represents the total effect of IV on DV prior to the inclusion of the mediating variable (c-path). *p < .05; **p < .01. Note. T = testosterone; SE = standard error; CI = confidence interval; n.s. = not significant; IV = independent variable; DV = dependent variable.

Figure 3.
Model of effect of T on magnitude through risk–benefit assessments. The indirect effect was significant for benefit assessments (β = −.12, SE = .05, p < .05, CI: −.23, −.03) and risk assessments (β = −.06, SE = .03, p < .05, CI: −.14, −.02). The value outside of the parentheses represents the direct effect, from bootstrapping analyses, of IV on DV after the mediating variable is included (c′-path). The value in parentheses represents the total effect of IV on DV prior to the inclusion of the mediating variable (c-path). Note. T = testosterone; SE = standard error; CI = confidence interval; n.s. = not significant; IV = independent variable; DV = dependent variable. p < .05; p < .01; p < .001.

Figure 4.
Model of effect of T on valence through risk–benefit assessments. The indirect effect was significant for benefit assessments (β = −.11, SE = .04, p < .05, CI: [−.21, −.03]) and risk assessments (β = −.12, SE = .04, p < .05, CI: [−.22, −.04]). The value outside of the parentheses represents the direct effect, from bootstrapping analyses, of IV on DV after the mediating variable is included (c′-path). The value in parentheses represents the total effect of IV on DV prior to the inclusion of the mediating variable (c-path). *p < .05; ***p < .001. Note. T = testosterone; SE = standard error; CI = confidence interval; n.s. = not significant; IV = independent variable; DV = dependent variable.

Additional analyses were conducted using the subject-mean scores (i.e., the mean scores for each individual across their diary entries) for risk–benefit assessments and the dimensions of disclosure. Similar to the results above, all of the indirect effects were significant except for the indirect effect of T on the amount of disclosure through benefit assessments.

RQ1 asked if T would interact with orgasm to predict the risks or benefits of disclosing after sexual activity. Orgasm was not centered, as 0 is already a meaningful value. The interaction was tested by entering T, orgasm, and their interaction as the three predictors in the model. The interaction between T and orgasm did not significantly predict risk or benefit assessments.

Additional analyses revealed similar results when examining women and men separately, as well as when using T scores mean centered separately for women and men. The only notable difference was that the interaction between T and orgasm was significant for risk assessments for women (β = −.40, p < .05). Two follow-up tests were used to investigate the moderating effects of orgasm on the association between T and risk assessments for women. The results revealed a stronger association between T and risk assessments for women who did not orgasm (β = .90, p < .01) than for those who experienced orgasm (β = .29, p < .05).

RQ2 asked if T would interact with orgasm to predict the dimensions of disclosure after sexual activity. The interaction between T and orgasm was significant only for the dimension of valence (β = .24, p < .02). Two follow-up tests were also used to investigate the moderating effects of orgasm on the association between T and valence. The results revealed a significant negative association between T and valence for those who did not orgasm (β = −.49, p < .01), but not for those who experienced orgasm (β = −.10, p = n.s.). For individuals who did not orgasm, the higher their T levels, the less positively valenced information they disclosed during the PSTI.

Additional analyses revealed similar patterns when examining women and men separately, as well as when using T scores mean centered separately for women and men. The only notable differences involved men’s responses. First, the men’s risk model was unable to converge. Second, men did not report differences in the valence of their disclosures depending on whether they did or did not orgasm. Third, the interaction between T and orgasm was significant for the dimension of magnitude (β = 3.47, p < .02). Two follow-up tests were used to investigate the moderating effects of orgasm on the association between T and magnitude for men. The results revealed a significant negative association between T and magnitude for men who did not orgasm (β = −3.56, p < .03), but not for those who experienced orgasm (β = −.22, p = n.s.).

Finally, all of the models were tested for possible interaction effects with sex (i.e., female/male) and relationship length. There were no significant interactions between T and sex or T and relationship length.

Discussion

This study investigated the physiology of communication during the PSTI by exploring the relationship between T levels and aspects of the disclosure decision process after sexual activity. As predicted, when individuals’ T levels were higher, their assessments of the benefits of disclosing after sexual activity were lower and their assessments of the risks of disclosing were higher. Additionally, the higher the individuals’ T levels, the less intentional and less positively valenced their disclosures were during the PSTI. In accordance with CPM theory (Petronio, 2000, 2002) and other models of disclosure (e.g., Afifi & Steuber, 2009), risk–benefit assessments mediated the relationship between T and intentionality, magnitude, and valence. These findings and their implications are discussed in greater detail below.

T and communication during the PSTI

The findings regarding T correspond with previous research indicating the suppression effects of T on the hormone OT. OT is a prosocial hormone (Kuchinskas, 2009) that may facilitate positive feelings and connection between partners during the PSTI and potentially influence post sex communication (Denes, 2012). T, however, is thought to suppress the effects of OT. Therefore, it makes sense that individuals with higher levels of T disclose less positive feelings to their partners during the PSTI, as such individuals may not feel the warm, positive effects of OT after sexual activity. The current findings also support research on T, which suggests that individuals with lower levels of the hormone demonstrate less aggressive and more social behaviors (Harris et al., 1996; Nelson, 2011).

Higher levels of T have also been associated with risk-taking, dominance behaviors, and competiveness (e.g., Evans & Hampson, 2014; Mazur & Booth, 1998; Stanton, Liening, & Schultheiss, 2011), suggesting that lower levels might facilitate healthier interpersonal communication. Risk-taking in T research is typically examined in terms of competitiveness or economic tasks rather than communicative behaviors with intimate partners (e.g., Evans & Hampson, 2014; Stanton et al., 2011). Self-disclosing private information, however, is a different kind of risk—it is a social risk because it can make people emotionally vulnerable and potentially lead to personal and social rejection. At the same time, disclosure is often a primary means of facilitating closeness and attachments (Altman & Taylor, 1973; Petronio, 2002). Even though people with higher T levels might be more likely to take risks in other contexts, in terms of intimate relationships, it could leave them feeling too vulnerable and intimately close. Correspondingly, lower T levels predicted increased benefit and decreased risk assessments, as individuals with lower levels of T may be inclined to perceive more positive outcomes to PSTI communication and see the potential for pair-bonding. S/P theory (van Anders et al., 2011) similarly posits that lower T levels facilitate nurturant behaviors, which can include loving, warm contact such as that which may occur during the PSTI. In terms of communication behavior, it seems that individuals with lower T levels are more likely than individuals with higher T levels to use pillow talk as a way to express nurturance to their partners, as they disclosed more positive feelings for their partners after sexual activity.

The above findings may also help to explain the inverse association between T and relationship satisfaction revealed in previous work (e.g., Edelstein, van Anders, Chopik, Goldey, & Wardecker, 2014). Several studies have revealed that individuals’ T levels are inversely associated with relationship quality and relationship-maintaining behaviors (e.g., Bos, Panksepp, Bluthé, & van Honk, 2012; Edelstein et al., 2014). Furthermore, Edelstein, van Anders, Chopik, Goldey, and Wardecker (2014) found that partners’ T levels were inversely associated with relationship quality. They suggest that individuals and partners with higher T levels may demonstrate fewer nurturant behaviors, which may influence relationship quality. Such a suggestion also aligns with S/P theory’s (van Anders et al., 2011) assertion that higher T is associated with less nurturant behavior. The findings of the present study contribute to this line of research by suggesting that disclosure may also play an important function in these associations. For example, it is possible that individuals with higher T levels disclose less to their partners, which in turn decreases relationship satisfaction. Future research would benefit from further investigating whether self-disclosure (both after sexual activity and in general) mediates the association between T and relationship quality.

The results of this study also revealed a significant interaction between T and orgasm. For individuals who did not orgasm, the higher their T levels, the less positively valenced information they disclosed during the PSTI. However, this same link was not found for individuals who did orgasm. This finding is similar to that of Denes and Afifi (2014), who found that for individuals who did not orgasm, increased alcohol consumption was associated with disclosing less positive feelings for one’s partner after sexual activity, but that same association was not significant for those who experienced orgasm. Denes and Afifi (2014) suggest that “orgasm may provide a buffer between the ‘downer’ effect of alcohol and postcoital communication” and that increases in OT from orgasm “may counteract the negative effects of alcohol consumption, and thus nullify the relationship between alcohol and negatively valenced disclosures” (p. 353). Orgasm may have a similar effect on the association between T and the valence of individuals’ post sex disclosures, as individuals who orgasm may feel the desire to disclose positive feelings to their partners regardless of their T levels. This finding may suggest that high T/no orgasm individuals are the least likely to experience the beneficial effects of post sex communication, as they neither experience the positive effects of OT surges from orgasm nor experience the nurturant intimacy of low T individuals.

T also predicted the intentionality of post sex disclosures. The lower the individuals’ T levels, the more likely they were to disclose intentionally, and this association was mediated by risk–benefit assessments. Individuals with lower T levels may perceive greater benefits and fewer risks to disclosing, and subsequently communicate in intentional ways aimed at achieving the goal of pair-bonding. These results also align with those of Denes and Afifi (2014), which revealed that individuals who experienced orgasm disclosed more intentionally than individuals who did not orgasm. Similarly, lower T levels may enable individuals to more clearly see potential pair-bonding opportunities during the PSTI and execute behaviors that will help them achieve a closer connection with their partners. In this way, lower T levels may help individuals to intentionally communicate messages of positive regard and affection during the PSTI. Future research should also explore the possible relational outcomes of such communication. For example, does more intentional communication predict greater relationship satisfaction?

Contrary to what was expected, T was not directly associated with either the amount of information disclosed during the PSTI or the magnitude of the disclosures, though T was indirectly associated with magnitude through risk–benefit assessments and indirectly associated with amount through risk assessments. Similar to Denes and Afifi (2014), this finding may have emerged due to beliefs about appropriate conversation topics during the PSTI. Because the PSTI is a time of increased intimacy and closeness, partners may limit their communication to topics involving their partner and their relationship. This narrow scope of appropriate topics may result in individuals disclosing a smaller amount of information than they might in other contexts. More sophisticated measures are necessary that can capture such finite distinctions.

While Denes and Afifi (2014) found that the magnitude of post sex disclosures was predicted by orgasm, T levels do not appear to directly influence this dimension of communication during the PSTI. Considering that the average relationship length for participants in the current study was 21 months, it is possible that individuals have already expressed deeper, more personal feelings to their partners and thus do not focus on the magnitude of disclosures during pillow talk. Instead, participants with lower T levels may simply reiterate positive sentiments previously expressed to their partners, as indicated by the significant negative association between T and the valence of post sex disclosures. It is important to note, though, that magnitude was associated with T through risk–benefit assessments, indicating that individuals with lower T levels perceive greater benefits and fewer risks to disclosing, which is then associated with disclosing information of greater magnitude during the PSTI.

The present study has important implications for communication theory. More specifically, this study emphasizes the importance of integrating communication theories with physiological theories. S/P theory (van Anders et al., 2011) and CPM theory (Petronio, 2000, 2002) together contribute to the understanding of physiological correlates of the disclosure decision process during the PSTI. S/P theory underscores the importance of investigating T levels when exploring pair-bond formation. The theory supports the notion that individuals with lower T levels may have heightened nurturant behavior, which appears to influence post sex communication. The present study also adds a new application to CPM by exploring physiological correlates of the disclosure decision process. The results revealed that individuals’ T levels are negatively associated with their perceived benefits of disclosing during the PSTI and positively associated with perceived risk. Consequently, this study suggests that T levels may predispose individuals to certain boundary rules for managing information during the PSTI. More specifically, individuals with lower T levels may have more permeable boundaries and be more likely to share positive information with their partners during the PSTI.

Limitations and future directions

Although this study makes an important contribution to our understanding of post sex behavior and physiological correlates of communication, there are a few important limitations that should be addressed. First, T is only one of many hormones that is likely associated with communication during the PSTI. It is possible that the other hormones detailed in S/P theory (van Anders et al., 2011), such as OT and vasopressin, may also predict post sex communication. Future research would benefit from investigating not only these additional hormones, but also potential interactions between hormones. A second limitation is that this study primarily focused on college students’ post sex communication. Individuals in long-term, committed relationships (e.g., marriage or civil union) may engage in pillow talk differently and have different goals for post sex interactions. Relatedly, long-term couples may develop nuanced post sex behaviors that are not predicted by physiological components and have instead uniquely developed within their relationships over time. It would also be worthwhile for future research to examine the links between physiology and post sex communication in older couples. Given that T levels decrease over the life span (see Kelsey et al., 2014; Nelson, 2011), the associations between T and pillow talk may depend on age.

Additional limitations involve the design and procedures of the study. First, T was only assessed at one time point, separate from the sexual episode. Although Granger, Shirtcliff, Booth, Kivlighan, and Schwartz (2004, p. 1238) found that “individual differences in salivary T levels were remarkably stable within the day, across days, and across one year,” it is possible that measuring T levels immediately before and after sexual activity may reveal unique associations with post sex disclosures. For example, increases in T that occur immediately before and after sexual activity (van Anders et al., 2007) may negatively influence post sex communication, and this effect may be amplified for individuals who already have higher baseline T levels. Thus, future research would benefit from measuring T at different time points before and after the sexual episode. It is also possible that greater amounts of PSTI communication affect T levels and/or that this association is mutually influential. An additional potential limitation involves the time of sexual activity. Many participants reported engaging in sexual activity late in the evening. Thus, it is possible that individuals were tired and/or fatigued, which may have influenced their responses in their diary entries.

A final limitation of this study is that only one partner’s perceptions of post sex communication were assessed. Future research would benefit from exploring both partners’ perceptions of communication during the PSTI to better address the interdependent nature of dyadic communication. It would also be worthwhile to explore the different motivations guiding partners’ post sex disclosures. For example, how do partners negotiate their PSTI behavior when one partner has the goal of verbal communication, pair-bonding, and increasing intimacy, while the other has the goal of nonverbal affection and sleep? How do conflicting post sex goals influence pillow talk and/or relationship satisfaction afterward?

Conclusion

This study contributes to research investigating the antecedents and consequences of post sex communication and the relationship between biology and communication. T appears to influence people’s disclosure decisions after sexual activity. Future research should examine the effects of such decisions and communication patterns on relationship quality and sexual satisfaction, especially given that orgasm interacted with T to affect the valence of the disclosures. The findings underscore the utility of using physiological and communication theories together to better understand the relationship between biology and communication and point to several valuable directions for future research investigating the physiology of pillow talk. Not only does the current study contribute to researchers’ theoretical understanding of sexual communication, privacy boundaries, and biology, but it could also have broader practical implications for romantic relationships.

Variables	M	SD	T	Risk	Benefit	Intentionality	Amount	Magnitude	Valence
T	65.18	41.61	–
Risk	1.96	0.97	.22**	–
Benefit	3.83	0.77	−.15**	−.43**	–
Intentionality	1.80	0.64	.18**	.46**	−.51**	–
Amount	2.23	0.95	.09**	.26**	.06	.17**	–
Magnitude	2.91	1.00	−.06	−.16**	.41**	−.26**	.39**	–
Valence	4.15	0.72	−.13**	−.54**	.51**	−.61**	−.07	.31**	–

Variables	M	SD	T	Risk	Benefit	Intentionality	Amount	Magnitude	Valence
T	46.99	17.58	–
Risk	1.93	0.97	.23**	–
Benefit	3.86	0.75	−.06	−.41**	–
Intentionality	1.77	0.62	.08	.43**	−.48**	–
Amount	2.21	0.94	.004	.25**	.04	.19**	–
Magnitude	2.95	0.99	.05	−.13**	.36**	−.20**	.37**	–
Valence	4.18	0.73	−.11*	−.56**	.51**	−.60**	−.08	.27**	–

Variables	M	SD	T	Risk	Benefit	Intentionality	Amount	Magnitude	Valence
T	129.15	38.48	–
Risk	2.09	0.99	.41**	–
Benefit	3.64	0.81	−.18	−.47**	–
Intentionality	1.96	0.70	.28**	.55**	−.58**	–
Amount	2.30	0.97	.33**	.28**	.17	.07	–
Magnitude	2.74	1.06	−.09	−.26*	.56*	−.42**	.49**	–
Valence	4.06	0.66	−.22*	−.45**	.55**	−.64**	−.002	.45**	–

Dependent variable	Between-persons differences, %	Within-person differences, %
Benefit	59	41
Risk	61	39
Intentionality	50	50
Amount	58	42
Magnitude	42	58
Valence	37	63

Footnotes

Note

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the Foundation for the Scientific Study of Sexuality Student Research Grant, Kinsey Institute for Research in Sex, Gender, and Reproduction Student Research Grant Program, University of California President’s Dissertation Year Fellowship, and University of California Humanities and Social Sciences Research Grant Program.

References

Afifi

T. D.

Steuber

(2009). The risk revelation model (RRM) and strategies used to reveal secrets. Communication Monographs, 76, 144–176. doi:10.1080/03637750902828412

Altman

Taylor

(1973). Social penetration: The development of interpersonal relationships. New York, NY: Holt, Rinehart and Winston.

Bancroft

Sanders

Davidson

Warner

(1983). Mood, sexuality, hormones, and the menstrual cycle. III. Sexuality and the role of androgens. Psychosomatic Medicine, 45, 509–516. doi:10.1097/00006842-198312000-00005

Baxter

L. A.

Montgomery

B. M.

(1996). Dialogues and dialectics. New York, NY: Guilford.

Blaicher

Gruber

Bieglmayer

Blaicher

A. M.

Knogler

Huber

(1999). The role of oxytocin in relation to female sexual arousal. Gynecological and Obstetric Investigation, 47, 125–126. doi:10.1159/000010075

Brunner

(1973). The organization of early skilled action. Child Development, 44, 1–11. doi:10.2307/1127671

Boksem

M. A.

Mehta

P. H.

Van den Bergh

van Son

Trautmann

S. T.

Roelofs

… Sanfey

A. G.

(2013). Testosterone inhibits trust but promotes reciprocity. Psychological Science, 24, 2306–2314. doi:10.1177/0956797613495063

Bos

P. A.

Panksepp

Bluthé

R. M.

van Honk

(2012). Acute effects of steroid hormones and neuropeptides on human social–emotional behavior: A review of single administration studies. Frontiers in Neuroendocrinology, 33, 17–35. doi:10.1016/j.yfrne.2011.01.002

Byers

E. S.

(2005). Relationship satisfaction and sexual satisfaction: A longitudinal study of individuals in long-term relationships. Journal of Sex Research, 42, 113–118. doi:10.1080/00224490509552264

10.

Byers

E. S.

Demmons

(1999). Sexual satisfaction and sexual self-disclosure within dating relationships. The Journal of Sex Research, 36, 1–10. doi:10.1080/00224499909551983

11.

Canary

D. J.

Hause

K. S.

(1993). Is there any reason to research sex differences in communication? Communication Quarterly, 41, 129–144. doi:10.1080/01463379309369874

12.

Carmichael

M. S.

Humbert

Dixen

Palmisano

Greenleaf

Davidson

J. M.

(1987). Plasma oxytocin increases in the human sexual response. Journal of Clinical Endocrinology and Metabolism, 64, 27–31. doi:10.1210/jcem-64-1-27

13.

Cohan

C. L.

Booth

Granger

D. A.

(2003). Gender moderates the relationship between testosterone and marital interaction. Journal of Family Psychology, 17, 29–40. doi:10.1037/0893-3200.17.1.29

14.

Cupach

W. R.

Comstock

(1990). Satisfaction with sexual communication in marriage: Links to sexual satisfaction and dyadic adjustment. Journal of Social and Personal Relationships, 7, 179–186. doi:10.1177/0265407590072002

15.

Dabbs

J. M.

Jr Mohammed

(1992). Male and female salivary testosterone concentrations before and after sexual activity. Physiology & Behavior, 52, 195–197. doi:10.1016/0031-9384(92)90453-9

16.

De Dreu

C. K. W.

Greer

L. L.

Van Kleef

G. A.

Shalvi

Handgraaf

M. J. J.

(2011). Oxytocin promotes human ethnocentrism. Proceedings of the National Academy of the Sciences, 108, 1262–1266. doi:10.1073/pnas.1015316108

17.

Denes

(2012). Pillow talk: Exploring disclosures after sexual activity. Western Journal of Communication, 76, 91–108. doi:10.1080/10570314.2011.651253

18.

Denes

(2015). Toward a post sex disclosures model (PSDM): Exploring the associations among orgasm, self-disclosure, and relationship satisfaction. Communication Research. doi: 10.1177/0093650215619216 (published online).

19.

Denes

Afifi

T. D.

(2014). Pillow talk and cognitive decision making processes: Exploring the role of orgasm and alcohol on communication after sexual activity. Communication Monographs, 81, 333–358. doi:10.1080/03637751.2014.926377

20.

Derlega

V. J.

Grzelak

(1979). Appropriateness of self-disclosure. In Chelune

G. J.

(Ed.), Self-disclosure: Origins, patterns, and implications of openness in interpersonal relationships (pp. 151–176). San Francisco, CA: Jossey-Bass.

21.

DeWall

C. N.

Gillath

Pressman

S. D.

Black

L. L.

Bartz

J. A.

Moskovitz

Stetler

D. A.

(2014). When the love hormone leads to violence: Oxytocin increases intimate partner violence inclinations among high trait aggressive people. Social Psychological and Personality Science, 5, 691–697. doi:10.1177/1948550613516876

22.

Domes

Heinrichs

Michel

Berger

Herpertz

(2007). Oxytocin improves “mind-reading” in humans. Biological Psychiatry, 61, 731–733. doi:10.1016/j.biopsych.2006.07.015

23.

Edelstein

R. S.

van Anders

S. M.

Chopik

W. J.

Goldey

K. L.

Wardecker

B. M.

(2014). Dyadic associations between testosterone and relationship quality in couples. Hormones and Behavior, 65, 401–407. doi:10.1016/j.yhbeh.2014.03.003

24.

Eisenegger

Naef

Snozzi

Heinrichs

Fehr

(2009). Prejudice and truth about the effect of testosterone on human bargaining behaviour. Nature, 463, 356–359. doi:10.1038/nature08711

25.

Evans

K. L.

Hampson

(2014). Does risk-taking mediate the relationship between testosterone and decision-making on the Iowa Gambling Task? Personality and Individual Differences, 61, 57–62. doi:10.1016/j.paid.2014.01.011

26.

Exton

M. S.

Bindert

Krüger

Scheller

Hartmann

Schedlowski

(1999). Cardiovascular and endocrine alterations after masturbation-induced orgasm in women. Psychosomatic Medicine, 61, 280–289. doi:10.1097/00006842-199905000-00005’

27.

Floyd

(2004). An introduction to the uses and potential uses of physiological measurement in the study of family communication. Journal of Family Communication, 4, 295–317. doi:10.1080/15267431.2004.9670137

28.

Floyd

(2006). Communicating affection: Interpersonal behavior and social context. New York, NY: Cambridge University Press.

29.

Fox

C. A.

Ismail

A. A. A.

Love

D. N.

Kirkham

K. E.

Loraine

J. A.

(1972). Studies on the relationship between plasma testosterone levels and human sexual activity. Journal of Endocrinology, 52, 51–58. doi:10.1677/joe.0.0520051

30.

Granger

D. A.

Cicchetti

Rogosch

F. A.

Hibel

L. C.

Teisl

Flores

(2007a). Blood contamination in children’s saliva: Prevalence, stability, and impact on the measurement of salivary cortisol, testosterone, and dehydroepiandrosterone. Psychoneuroendocrinology, 32, 724–733. doi:10.1016/j.psyneuen.2007.05.003

31.

Granger

D. A.

Kivlighan

K. T.

El-Sheikh

Gordis

Stroud

L. R.

(2007b). Salivary alpha-amylase in bio-behavioral research: Recent developments and applications. Annals of the New York Academy of Sciences, 1098, 122–144. doi:10.1196/annals.1384.008

32.

Granger

D. A.

Shirtcliff

E. A.

Booth

Kivlighan

K. T.

Schwartz

E. B.

(2004). The “trouble” with salivary testosterone. Psychoneuroendocrinology, 29, 1229–1240. doi:10.1016/j.psyneuen.2004.02.005

33.

Guastella

A. J.

Mitchel

P. B.

Dadds

M. R.

(2008). Oxytocin increases gaze to the eye region of human faces. Biological Psychiatry, 63, 3–5. doi:10.1016/j.biopsych.2007.06.026

34.

Halpern

Sherman

(1979). Afterplay: A key to intimacy. New York, NY: Stein & Day Publishers.

35.

Harris

J. A.

Rushton

H. J.

Hampson

Jackson

D. N.

(1996). Salivary testosterone and self-report aggressive and pro-social personality characteristics in men and women. Aggressive Behavior, 22, 321–331. doi:10.1002/(SICI)1098-2337(1996)22:5<321:: AID-AB1>3.0.CO;2-M

36.

Hoffman

(2015). Longitudinal analysis: Modeling within-person fluctuation and change. New York, NY: Taylor and Francis.

37.

Huber

Veinante

Stoop

(2005). Vasopressin and oxytocin excite distinct neuronal populations in the central amygdala. Science, 308, 245–248. doi:10.1126/science.1105636

38.

Hughes

S. M.

Kruger

D. J.

(2011). Sex differences in post-coital behaviors in long- and short-term mating: An evolutionary perspective. Journal of Sex Research, 48, 496–505. doi:10.1080/00224499.2010.501915

39.

Kelsey

T. W.

L. Q.

Mitchell

R. T.

Whelan

Anderson

R. A.

Wallace

W. H. B.

(2014). A validated age-related normative model for male total testosterone shows increasing variance but no decline after age 40 years. PLoS ONE, 9, e109346. doi:10.1371/journal.pone.0109346

40.

Kirsch

Esslinger

Chen

Mier

Lis

Siddhanti

… Meyer-Lindenberg

(2005). Oxytocin modulates neural circuitry for social cognition and fear in humans. Journal of Neuroscience, 25, 11489–11493. doi:10.1523/JNEUROSCI.3984-05.2005

41.

Knussmann

Christiansen

Couwenbergs

(1986). Relations between sex hormone levels and sexual behavior in men. Archives of Sexual Behavior, 15, 429–445. doi:10.1007/BF01543113

42.

Kosfeld

Heinrichs

Zak

P. J.

Fischbacher

Fehr

(2005). Oxytocin increases trust in humans. Nature, 435, 673–676. doi:10.1038/nature03701

43.

Kruger

D. J.

Hughes

S. M.

(2010). Variation in reproductive strategies influences post-coital experiences with partners. Journal of Social, Evolutionary, and Cultural Psychology, 4, 254–264.

44.

Kruger

D. J.

Hughes

S. M.

(2011). Tendencies to fall asleep first after sex are associated with greater partner desires or bonding and affection. Journal of Social, Evolutionary, and Cultural Psychology, 5, 239–247.

45.

Kruger

Exton

M. S.

Pawlak

von zur Muhlen

Hartmann

Shedlowski

(1998). Neuroendocrine and cardiovascular response to sexual arousal and orgasm in men. Psychoneuroendocrinology, 23, 401–411. doi:10.1016/S0306-4530(98)00007-9

46.

Kuchinskas

(2009). The chemistry of connection: How the oxytocin response can help you find trust, intimacy, and love. Oakland, CA: New Harbinger Publications.

47.

Lee

Jaffe

Midgley

(1974). Lack of alteration of serum gonadotropins in men and women following sexual intercourse. American Journal of Obstetrics and Gynecology, 120, 985–987.

48.

Lim

M. M.

Young

L. J.

(2006). Neuropeptidergic regulation of affiliative behavior and social bonding in animals. Hormones and Behavior, 50, 506–517. doi:10.1016/j.yhbeh.2006.06.028

49.

Malle

B. F.

Moses

L. J.

Baldwin

D. A.

(2001). Intentions and intentionality: Foundations of social cognition. Cambridge, MA: MIT Press.

50.

Mallinckrodt

Abraham

T. W.

Wei

Russell

D. W.

(2006). Advances in testing statistical significance of mediation effects. Journal of Counseling Psychology, 53, 372–378. doi:10.1037/0022-0167.53.3.372

51.

Mantzoros

C. S.

Georgiadis

E. I.

Trichopoulosm

(1995) Contribution of dihydrotestosterone to male sexual behaviour. British Medical Journal, 310, 1289–1291. doi:10.1136/bmj.310.6990.1289

52.

Mazur

Booth

(1998). Testosterone and dominance in men. Behavioral and Brain Sciences, 21, 353–363. doi:10.1017/S0140525X98321226

53.

Mehta

P. H.

Josephs

R. A.

(2010). Testosterone and cortisol jointly regulate dominance: Evidence for a dual-hormone hypothesis. Hormones and Behavior, 58, 898–906. doi:10.1016/j.yhbeh.2010.08.020

54.

Morris

N. M.

Udry

J. R.

Khan-Dawood

Dawood

M. Y.

(1987). Marital sex frequency and midcycle female testosterone. Archives of Sexual Behavior, 16, 27–37. doi:10.1007/BF01541839

55.

Muise

Giang

Impett

E. A.

(2014). Post sex affectionate exchanges promote sexual and relationship satisfaction. Archives of Sexual Behavior, 43, 1391–1402. doi:10.1007/s10508-014-0305-3

56.

Nelson

R. J.

(2011). An introduction to behavioral endocrinology (4th ed.). Sunderland, MA: Sinauer Associates, Inc.

57.

Petronio

(1991). Communication boundary management: A theoretical model of managing disclosure of private information between marital couples. Communication Theory, 1, 311–335. doi:10.1111/j.1468-2885.1991.tb00023.x

58.

Petronio

(2000). The boundaries of privacy: Praxis of everyday life. In Petronio

(Ed.), Balancing the secrets of private disclosures (pp. 37–49). Mahwah, NJ: Lawrence Erlbaum.

59.

Petronio

(2002). Boundaries of privacy. New York, NY: SUNY Press.

60.

Poulin-Dubois

Shultz

T. R.

(1988). The development of the understanding of human behaviors: From agency to intentionality. In Zelazo

P. D.

Astington

J. W.

Olson

D. R.

(Eds.), Developing theories of intention: Social understanding and self-control (pp. 109–125). Mahwah, NJ: Lawrence Erlbaum Associates.

61.

Preacher

K. J.

Hayes

A. F.

(2004). SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behavior Research Methods, Instruments, & Computers, 36, 717–731. doi:10.3758/BF03206553

62.

Preacher

K. J.

Hayes

A. F.

(2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40, 879–891. doi:10.3758/BRM.40.3.879

63.

Ryder

(2012). After the climax. Retrieved from http://www.salon.com/2012/03/15/after_the_climax/

64.

Shultz

T. R.

(1980). Development of the concept of intention. In Collins

W. A.

(Ed.), Development of cognition, affect, and social relations. The Minnesota Symposia on Child Psychology (Vol. 13, pp. 131–164). Hillsdale, NJ: Erlbaum.

65.

Stanton

S. J.

Liening

S. H.

Schultheiss

O. C.

(2011). Testosterone is positively associated with risk taking in the Iowa Gambling Task. Hormones and Behavior, 59, 252–256. doi:10.1016/j.yhbeh.2010.12.003

66.

Taylor

S. E.

Lewis

B. P.

Gruenewald

T. L.

Gurung

R. A. R.

Updegraff

J. A.

Klein

L. C.

(2002). Sex differences in biobehavioral responses to threat: Reply to Geary and Flinn (2002). Psychological Review, 109, 751–753. doi:10.1037/0033-295X.109.4.751

67.

van Anders

S. M.

Goldey

K. L.

Kuo

P. X.

(2011). The steroid/peptide theory of social bonds: Integrating testosterone and peptide responses for classifying social behavioral contexts. Psychoneuroendocrinology, 36, 1265–1275. doi:10.1016/j.psyneuen.2011.06.001

68.

van Anders

S. M.

Hamilton

L. D.

Schmidt

Watson

N. V.

(2007). Associations between testosterone secretion and sexual activity in women. Hormones and Behavior, 51, 477–482. doi:10.1016/j.yhbeh.2007.01.003

69.

van Anders

S. M.

Watson

N. V.

(2006). Relationship status and testosterone in North American heterosexual and non-heterosexual men and women: Cross-sectional and longitudinal data. Psychoneuroendocrinology, 31, 715–723. doi:10.1016/j.psyneuen.2006.01.008

70.

van Honk

Montoya

E. R.

Bos

P. A.

van Vugt

Terburg

(2012). New evidence on testosterone and cooperation. Nature, 485, E4–E5. doi:10.1038/nature11136

71.

Veenestra

(2007). Afterglow. In Malti-Douglas

(Ed.), Encyclopedia of sex and gender (Vol. 1, pp. 39–40). Detroit, MI: Macmillan Reference.

72.

Wingfield

J. C.

Hegner

R. E.

Dufty

A. M.

Jr Ball

G. F.

(1990). The “challenge hypothesis”: Theoretical implications for patterns of testosterone secretion, mating systems, and breeding strategies. American Naturalist, 136, 829–846.