The term ‘pheromone’ is derived from ancient Greek pherein ‘to transfer’ and hormōn ‘to excite’. Pheromones are known as chemical signals that convey information between members of the same species (Karlson and Luscher, 1959). Less appreciated are their intricate interplays with the central neural and neuroendocrine systems. In insects and rodents, specific neuropeptides and hormones are critically involved in the production, release, as well as perception of pheromones (Cusson and McNeil, 1989; Lin et al., 2016; Mugford and Nowell, 1971; Raina et al., 1989; Tang et al., 1989). Little is known as to whether this link holds for non-rodent mammals or how it is manifested there. In humans, particularly, the very existence of pheromones has long been an issue of debate (Wyatt, 2015). Since the first report of human menstrual synchrony in 1971 (McClintock, 1971), nearly five decades of research has shown that human body secretions, particularly axillary sweat, subconsciously convey emotion and reproductive state (de Groot et al., 2015; Stern and McClintock, 1998; Zhou and Chen, 2009), which in turn are driven by complex neurochemical changes (Blouin et al., 2013; Jones and Lopez, 2013). Of the many components of human body secretions, two endogenous steroids, androsta-4,16,-dien-3-one, a non-androgenic derivative of gonadal progesterone (Gower and Ruparelia, 1993), and estra-1,3,5 (10),16-tetraen-3-ol, related to the estrogen sex hormones but with no known estrogenic effects (Thysen et al., 1968), emerge as candidates of human sex pheromones. Aside from affecting autonomic responses and mood states in women and men, respectively (Bensafi et al., 2004; Jacob and McClintock, 2000; Lundström et al., 2003a; Olsson et al., 2006), androstadienone and estratetraenol are found to effectively convey sex-specific information and activate the hypothalamus, a structure critically involved in sexual reproduction (Simerly, 2002), in distinct patterns. Specifically, androstadienone signals masculinity and estratetraenol signals femininity to their targeted recipients (Zhou et al., 2014). They also prime the identification of emotionally receptive states for the potential mates with whom they are associated (Ye et al., 2019). In parallel, androstadienone activates the hypothalamus in heterosexual women and homosexual men, but not in heterosexual men or homosexual women, whereas estratetraenol activates the hypothalamus in heterosexual men and homosexual women, but not in heterosexual women or homosexual men (Berglund et al., 2006; Savic et al., 2001; Savic et al., 2005).

The hypothalamus contains several sexually dimorphic nuclei and is well documented to coordinate neuroendocrine responses with sensory cues that regulate motivated behavior (Simerly, 2002). It produces, among various hormones, two structurally similar nonapeptides called oxytocin and vasopressin that function both as neuropeptides and hormones and have ancient roles in reproductive behaviors. Oxytocin and vasopressin are heavily implicated in sexual arousal and subsequent copulatory behavior in rats, rabbits, rams, bulls, and humans, with their homologs mediating comparable activities in nematodes, snails, and bony fish (Carter, 1992; Donaldson and Young, 2008). Moreover, the two peptides, especially oxytocin, are believed to underlie the evolution and expression of human sociality including trust, love, and cooperation (Carter, 2014; Donaldson and Young, 2008). Based on the animal literature, receptors for oxytocin and vasopressin are expressed in the olfactory system and other downstream sexually dimorphic nuclei under the modulatory influence of gonadal steroids, and are involved in the processing of social odors (Stoop, 2012). These findings have led us to suspect that the effects of androstadienone and estratetraenol, considered by some as putative human sex pheromones, could be regulated by oxytocin and/or vasopressin.

To examine the above hypothesis, we employed a gender identification task of point-light walkers (PLWs) (Figure 1A)—dynamic point-light displays portraying the gaits of walkers—that had been shown to be sensitive to the interactive effects of androstadienone, estratetraenol, and the recipients’ sex and sexual orientation (Zhou et al., 2014), and combined it with pharmacological manipulations of central oxytocin and vasopressin. We also measured each participant’s social proficiency (the opposite of autistic-like tendency) with the Autism Spectrum Quotient (AQ) (Baron-Cohen et al., 2001), in view of the positive association between social proficiency and endogenous oxytocin level (Koven and Max, 2014; Lancaster et al., 2015; Parker et al., 2014) and research showing that individual differences in social proficiency or AQ score can moderate the effects of exogenously administered oxytocin (Bartz et al., 2019; Bartz et al., 2010; Bartz et al., 2011). From the data we obtained psychometric curves that depicted the probability of making male judgments as a function of the physical gender of the PLWs (Figure 1B) under different combinations of olfactory stimuli and drug treatments. Systematic comparisons of these psychometric curves (Figure 1C) across five experiments totaling 1056 testing sessions enabled us to assess the roles of oxytocin and vasopressin in chemosensory communications of masculinity and femininity through androstadienone and estratetraenol in both heterosexual (Experiment 1, sample size n = 72) and homosexual (Experiment 2, n = 72) men (Figure 1D) as well as in socially less proficient (Experiments 3 and 4, n = 24 in each experiment) and socially proficient (Experiment 5, n = 24) individuals (Figure 1E).

Figure 1

Download asset Open asset

General procedures of the gender identification task have been described elsewhere (Zhou et al., 2014). In brief, genders of the PLWs were quantified as Z scores (Troje, 2002) and ranged in seven equal steps from feminine (−0.45 SD) to masculine (0.45 SD) with 0 marking the approximate gender-neutral point individually adjusted for each participant prior to the actual experiment in the absence of olfactory stimulus and drug treatment. In each trial of the task, participants viewed a PLW for 500 ms (0.5 walking cycle) and made a forced choice judgment on whether it was a male or a female walker (Figure 1A). To characterize how gender perception criteria were influenced by androstadienone or estratetraenol under different manipulations of oxytocin and vasopressin, we fitted gender judgments for all PLWs separately for each olfactory condition and each participant under each drug treatment condition with a Boltzmann sigmoid function that contained two parameters: point of subjective equality (PSE) and difference limen (Figure 1B). The PSE, an index of response criterion, is the Z score (which denoted PLWs’ physical gender) corresponding to a probability of 50% on the fitted psychometric function, where the participant perceived a PLW as equally masculine and feminine. The difference limen is half the interquartile range of the fitted function and indicates response sensitivity. For each drug treatment condition, participants’ PSEs when smelling androstadienone or estratetraenol—each dissolved in a clove oil carrier solution (1% v/v clove oil in propylene glycol)—were compared with those when smelling the clove oil carrier solution alone (control). A negative PSE shift relative to the carrier control condition corresponds to an overall leftward shift of the psychometric curve and reflects an increased tendency to judge the PLWs as male, hence a masculine bias in gender perception (Figure 1C). Conversely, a positive PSE shift indicates a feminine bias in gender perception.

Oxytocin, vasopressin, and subconscious chemosensory decoding of sex in heterosexual and homosexual men

We first recruited 72 heterosexual males (Kinsey scores = 0) in Experiment 1 and randomly assigned them to three groups of 24 each to receive different intranasal drug treatments: 24 IU oxytocin, 24 IU vasopressin, or 60 µg atosiban—a desamino-oxytocin analogue and a competitive antagonist of both oxytocin and vasopressin receptors (Manning et al., 2012). Nasally delivered oxytocin and vasopressin can directly access the cerebrospinal fluid (Born et al., 2002; Freeman et al., 2016); there has been some limited indication that intranasal atosiban is also centrally available (Liu et al., 2018; Lundin et al., 1986). In a double-blind procedure, each participant was tested in 3 sessions held at around the same time of the day on 3 consecutive days. During each session they performed the gender identification task (Figure 1A) while being continuously exposed to either androstadienone (500 μM, 5 ml), estratetraenol (500 μM, 5 ml) or their carrier solution alone (control condition, 1% v/v clove oil in propylene glycol, 5 ml), one per session in a counterbalanced manner, 35 min following drug administration (Figure 1D). The three olfactory stimuli were perceptually indiscriminable as assessed in a separate panel of 48 male participants (mean triangular discrimination accuracy = 0.33 vs. chance = 0.33, t₄₇ = 0.02, p>0.9), which was also in line with earlier reports (Ye et al., 2019; Zhou et al., 2014). Our previous study had shown that heterosexual males were affected by estratetraenol, but not androstadienone, in making gender judgments. Specifically, estratetraenol subconsciously biased them toward perceiving the PLWs as more feminine (Figure 2A; Zhou et al., 2014). We reasoned that the effect of estratetraenol, if regulated by the oxytocin/vasopressin system, would be blocked by intranasal atosiban and modulated by intranasal oxytocin and/or vasopressin. It was difficult to predict the directions of their effects (if any), as both positive and negative effects of oxytocin and vasopressin have been reported in the literature, depending on dose, context, and personal characteristics (Bartz et al., 2011; Carter, 2014; Donaldson and Young, 2008). It was also possible that only one of oxytocin and vasopressin plays a role, in which case the chemosensory effect would not be altered by the administration of the other.

Figure 2

Download asset Open asset

Figure 2—source data 1

Experiments 1 and 2.

Androstadienone- and estratetraenol- induced visual gender judgment biases after the nasal administrations of 60 µg atosiban, 24 IU oxytocin, and 24 IU vasopressin in heterosexual and homosexual male participants.

https://cdn.elifesciences.org/articles/59376/elife-59376-fig2-data1-v1.data.xlsx

Download elife-59376-fig2-data1-v1.data.xlsx

Indeed, in the atosiban-treated heterosexual men, we found that smelling estratetraenol relative to the carrier solution alone failed to influence their gender perception criteria (indexed by the PSEs, t₂₃ = 1.37, p=0.18; Figure 2B,E). Interestingly, this was also the case for those treated with 24 IU oxytocin (t₂₃ = 0.17, p=0.87; Figure 2C,E). Only in the participants treated with 24 IU vasopressin did estratetraenol induce a systematic bias towards perceiving the PLWs as more feminine (t₂₃ = 4.25, p<0.001, Cohen’s d = 0.87; Figure 2D,E). The size of the estratetraenol-induced gender perception bias under vasopressin was similar to that found earlier without drug treatment (Zhou et al., 2014), and a direct comparison of the two sets of data showed no significant difference (p=0.35). In other words, vasopressin did not significantly enhance or diminish the effect of estratetraenol on heterosexual males. On the other hand, smelling androstadienone relative to the carrier solution alone did not influence gender perception criteria regardless of drug treatments (i.e. 60 µg atosiban, 24 IU oxytocin, or 24 IU vasopressin, ps > 0.34; Figure 2B–E), consistent with earlier findings (Zhou et al., 2014).

These results presented a complex picture. Antagonizing oxytocin and vasopressin receptors with atosiban abolished the known effect of estratetraenol on heterosexual males (i.e. biasing them toward perceiving the PLWs as more feminine), yet 24 IU oxytocin appeared to produce the same effect. On the flip side, the administration of 24 IU vasopressin did not significantly alter the processing of estratetraenol in heterosexual males—they remained biased toward perceiving the PLWs as more feminine under the exposure of estratetraenol, to the same extent as when no drug was administered. We wondered if this pattern of drug influences would hold for the chemosensory decoding of masculine information carried by androstadienone. To this end, we turned to homosexual males in Experiment 2, who had been shown to be subconsciously biased by androstadienone, but not estratetraenol, in making gender judgments (Figure 2F; Zhou et al., 2014).

Except for the participants’ sexual orientation (homosexual males, mean Kinsey score ± SD = 5.26 ± 0.63), Experiment 2 was identical to Experiment 1, and revealed a similar pattern of drug effects. Gender perception criteria were unaffected by the exposure to androstadienone, relative to the carrier control, in the homosexual men treated with 60 µg atosiban (t₂₃ = 0.88, p=0.39; Figure 2G,J) as well as in those treated with 24 IU oxytocin (t₂₃ = −1.29, p=0.21; Figure 2H,J). By contrast, in the 24 IU vasopressin group, smelling androstadienone induced a systematic bias toward perceiving the PLWs as more masculine (t₂₃ = −2.99, p=0.007, Cohen’s d = 0.61; Figure 2I,J). The strength of the androstadienone-induced masculine bias under vasopressin was again not different from that found earlier (Zhou et al., 2014) in homosexual males receiving no drug treatment (p=0.18). Meanwhile, in line with earlier findings (Zhou et al., 2014), smelling estratetraenol did not influence gender judgements in these homosexual men irrespective of drug manipulations (ps > 0.58; Figure 2G–J).

An omnibus ANOVA of the PSEs from both Experiments 1 and 2 (olfactory condition × drug treatment × sexual orientation) identified a significant interaction between olfactory condition and drug treatment (F_{4, 276} = 3.58, p=0.007, partial η² = 0.05), which reinforced that the effects of androstadienone and estratetraenol on gender judgments were modulated by the drug manipulations, and that this modulation was similar in both heterosexual and homosexual males (olfactory condition × drug treatment × sexual orientation: F_{4, 276} = 0.80, p=0.53). To facilitate comparison, we highlighted the central tendencies of the androstadienone- and estratetraenol-induced PSE shifts (x- and y- axes, respectively) under different drug treatments in Figure 3 (see also Figure 3—figure supplement 1), generated by using a standard bootstrapping procedure (Davison and Hinkley, 1997; see Materials and methods). Under 60 µg atosiban and 24 IU oxytocin, neither androstadienone nor estratetraenol induced any significant change in gender perception criterion in heterosexual (cyan dots) or homosexual males (lime dots), and the bootstrapped sample means of the two groups of men overlapped around the origin. Conversely, under 24 IU vasopressin, they formed two discrete clusters: estratetraenol biased heterosexual, but not homosexual, males toward perceiving the PLWs as more feminine (cyan dots fell around the vertical axis on the positive side), whereas androstadienone biased homosexual, but not heterosexual, males toward perceiving the PLWs as more masculine (lime dots fell around the horizontal axis on the negative side). In addition, the difference limens of the heterosexual and homosexual males did not differ (F_{1, 138} = 0.47, p=0.50) and were comparable across olfactory conditions and drug treatments (olfactory condition × drug treatment: F_{4, 276} = 1.19, p=0.31; olfactory condition: F_{2, 276} = 0.48, p=0.62; drug treatment: F_{2, 138} = 0.24, p=0.79). So were their self-reported mood states on the Profile of Mood States (McNair et al., 1971) (POMS; total mood disturbance: ps = 0.84, 0.51 and 0.90, respectively; all subscales: ps > 0.05, corrected). Thus, it was the criterion (reflected in the PSEs) rather than the sensitivity of gender judgment (reflected in the difference limens) or transient mood state that was swayed by the interplays between the chemosignals and the drug manipulations.

Figure 3 with 2 supplements see all

Download asset Open asset

Figure 3—source data 1

Experiments 1 and 2.

Degrees of gender perception biases induced by chemosensory sexual cues in high AQ and low AQ individuals.

https://cdn.elifesciences.org/articles/59376/elife-59376-fig3-data1-v1.data.xlsx

Download elife-59376-fig3-data1-v1.data.xlsx

The combined results of Experiments 1 and 2 thus suggested that the decoding of feminine information carried by estratetraenol and that of masculine information carried by androstadienone, while contingent upon the recipients’ sexual orientation, were subserved by similar neuroendocrine mechanisms that were disrupted by intranasal atosiban—the competitive antagonist of both oxytocin and vasopressin receptors, as well as by 24 IU oxytocin, and were unaffected by 24 IU vasopressin. Since vasopressin seemed to exert no effect (participants’ response patterns to the chemosignals under 24 IU vasopressin were comparable to those previously obtained without drug treatment), by deduction, such mechanisms involved oxytocin. The question remained as to why the administration of 24 IU oxytocin, like atosiban, exempted the participants from the influences of the chemosignals, and we explored it in more detail.

If, regardless of one’s sexual orientation, oxytocin plays a role in the processing of chemosensory sexual cues associated with the preferred sex, given the relationship between social proficiency and endogenous oxytocin level (Koven and Max, 2014; Lancaster et al., 2015; Parker et al., 2014) and the heterogenous effects of intranasal oxytocin on individuals with different levels of social proficiency (Bartz et al., 2011), specifically as assessed by the AQ (Bartz et al., 2019; Bartz et al., 2010), it follows that individuals with different AQ scores could differ in their susceptibility to such chemosignals as well as to the effect of exogenous oxytocin. Put differently, social proficiency could be related to the subconscious extraction of chemosensory sexual information, a link hitherto unsuspected. As an initial attempt to test this inference, we reexamined the data from Experiments 1 and 2 to see if high AQ and low AQ participants differed in their susceptibility to the chemosignals. The participants’ AQ scores ranged from 9 to 35. We adopted a relatively strict criterion—AQ scores ≥ 25, that is, 1 SD or more above the reported mean for males (Baron-Cohen et al., 2001)—for high AQ individuals in hopes to better capture the effects of social proficiency (Bartz et al., 2019). Whereas an AQ score above 32 is used as the cutoff for distinguishing individuals with clinically significant levels of autistic traits (Baron-Cohen et al., 2001), we did not opt for this stringent cutoff value here due to pragmatic considerations: Such individuals are rare in the general population and only 1 out of the 144 participants scored above 32. Those with AQ scores below 25 were classified as low AQ individuals. Overall, AQ scores were comparable among the drug groups and also between the heterosexual and the homosexual participants in Experiments 1 and 2 (drug treatment: F_{2, 138} = 0.48, p=0.62; sexual orientation: F_{1, 138} = 1.39, p=0.24; interaction: F_{2, 138} = 0.097, p=0.91). Since no significant effect of androstadienone or estratetraenol was observed under the 60 µg atosiban condition or 24 IU oxytocin condition, we focused our examination on the 24 IU vasopressin condition where the effects of the two chemosignals were comparable to those obtained earlier without drug treatment (Zhou et al., 2014). The majority (85.4%) of the participants were low AQ individuals (AQ scores < 25). Our supplementary analysis revealed that, relative to the high AQ participants (n = 7, 14.6%), estratetraenol and androstadienone induced larger gender perception biases in the low AQ participants, driving the overall effects of the chemosignals on gender perception (Figure 3—figure supplement 2). The results thus lent preliminary support to the postulated link between social proficiency and the subconscious processing of chemosignals, and led us to assess the effect of exogenous oxytocin separately in high AQ and low AQ individuals.

Dose-dependent modulation of chemosensory decoding of sex by oxytocin but not vasopressin in high and low AQ individuals

It has been reported that the effect of intranasal oxytocin is more pronounced in socially less proficient individuals as measured by the AQ (Bartz et al., 2019; Bartz et al., 2010; Bartz et al., 2011) and that a lower dose of oxytocin could exert a more positive effect than a higher dose (Cardoso et al., 2013; Quintana et al., 2017; Spengler et al., 2017). To better probe the role of oxytocin in the chemosensory processing of sexual information and not miss any possible effect of vasopressin, we recruited 24 heterosexual males with high AQ scores (range: 25–36) in Experiment 3 and assessed the extent to which their gender perception was biased by estratetraenol under different doses of oxytocin and vasopressin in a within-subject design (Figure 1E). Specifically, each participant was tested in 10 sessions held around the same time of the day on 10 days that comprised of five drug conditions (2 sessions each): 12 IU oxytocin, 24 IU oxytocin, 12 IU vasopressin, 24 IU vasopressin, and no drug treatment. Under each drug condition, they performed the gender identification task while being continuously exposed to estratetraenol in one session, and to the carrier solution alone in the other session. Androstadienone was not included as it consistently showed no effect on heterosexual men’s gender perception (Experiment 1) (Zhou et al., 2014).

In the absence of drug treatment, the high AQ heterosexual men failed to utilize the feminine information carried by estratetraenol in making gender judgments (t₂₃ = 0.70, p=0.49; Figure 4A,F), which echoed with the preliminary result from the high AQ individuals in Experiments 1 and 2 (Figure 3—figure supplement 2). Remarkably, this chemosensory ability was restored by the administration of 12 IU (t₂₃ = 2.72, p=0.012, Cohen’s d = 0.55; Figure 4B,F), but not 24 IU (t₂₃ = −0.36, p=0.72; Figure 4C,F), intranasal oxytocin. Meanwhile, the participants remained insusceptible to estratetraenol regardless of whether they were treated with 12 IU (t₂₃ = 0.76, p=0.46; Figure 4D,F) or 24 IU vasopressin (t₂₃ = −0.24, p=0.81; Figure 4E,F). The results thus more directly pointed to the involvement of oxytocin, but not vasopressin, in the processing of chemosensory sexual information. Critically, they suggested that the effect of oxytocin was not monotonic. At 12 IU, intranasal oxytocin facilitated the decoding of the feminine information carried by estratetraenol in high AQ heterosexual men; but at 24 IU, the effect disappeared.

Figure 4 with 1 supplement see all

Download asset Open asset

Figure 4—source data 1

Experiments 3 and 5.

Estratetraenol-induced visual gender judgment biases in the absence of drug treatment and after the nasal administrations of 12 IU oxytocin, 24 IU oxytocin, 12 IU vasopressin, and 24 IU vasopressin in high AQ and low AQ heterosexual male participants.

https://cdn.elifesciences.org/articles/59376/elife-59376-fig4-data1-v1.xlsx

Download elife-59376-fig4-data1-v1.xlsx

Figure 4—source data 2

Experiment 4.

Estratetraenol-induced visual gender judgment biases after the nasal administrations of saline, 12 IU oxytocin, and 24 IU oxytocin in high AQ heterosexual male participants.

https://cdn.elifesciences.org/articles/59376/elife-59376-fig4-data2-v1.xlsx

Download elife-59376-fig4-data2-v1.xlsx

The nonmonotonic effect of oxytocin on chemosensory decoding of sex was striking, and we sought to replicate it before continuation. In Experiment 4, we adopted a fully double-blind placebo-controlled within-subject design and recruited another 24 high AQ (range: 25–39) heterosexual men, who underwent three drug conditions: saline, 12 IU oxytocin, and 24 IU oxytocin. Their results mirrored those of Experiment 3 (Figure 4—figure supplement 1). The combined data from Experiments 3 and 4 affirmed that high AQ heterosexual men were unsusceptible to estratetraenol at baseline (no drug treatment/saline; t₄₇ = 0.95, p=0.35; with no difference between no drug treatment and saline, p=0.95), yet showed a robust estratetraenol-induced gender perception bias when treated with 12 IU oxytocin (t₄₇ = 4.10, p<0.001, Cohen’s d = 0.59) and not 24 IU oxytocin (t₄₇ = −0.025, p=0.98). There was overall a significant quadratic effect of intranasal oxytocin dose on estratetraenol-induced shift of gender judgment criterion in these participants (F_{1, 47} = 6.58, p=0.014, partial η² = 0.12). We hence deduced that oxytocin level and the chemosensory processing of sexual cues follow an inverted-U-shaped relationship. It follows that low AQ individuals, who presumably have higher levels of endogenous oxytocin (Koven and Max, 2014; Lancaster et al., 2015; Parker et al., 2014), would benefit less, if any, from 12 IU oxytocin, and not from 24 IU oxytocin. Furthermore, we predicted that their decoding of chemosensory sexual information would not be influenced by vasopressin regardless of the dose.

Our hypotheses were confirmed in Experiment 5, which was identical to Experiment 3 except that the participants were 24 heterosexual males with low AQ scores (range: 8–24). In sharp contrast to the high AQ men in Experiments 3 and 4, the low AQ individuals decoded the feminine information carried by estratetraenol in the absence of drug treatment (t₂₃ = 3.15, p=0.005, Cohen’s d = 0.64; Figure 4G,L), again in line with the preliminary result from the low AQ individuals in Experiments 1 and 2 (Figure 3—figure supplement 2). The chemosensory effect was nonetheless diminished following the administration of 12 IU oxytocin (t₂₃ = 1.30, p=0.21; Figure 4H,L) and was abolished by the administration of 24 IU oxytocin (t₂₃ = 0.47, p=0.64; Figure 4I,L); that is, exogenous oxytocin at these doses hampered rather than facilitated the utilization of chemosensory feminine information in these low AQ individuals. Vasopressin, as expected, exerted no influence. The participants remained capable of detecting the feminine information carried by estratetraenol under the treatments of 12 IU and 24 IU vasopressin (t₂₃s = 4.02 and 3.10, ps = 0.001 and 0.005, Cohen’s ds = 0.82 and 0.63, respectively; Figure 4J–L).

To further characterize the role of oxytocin in the chemosensory processing of sexual information, we analyzed the pooled PSEs across different combinations of oxytocin doses (no drug treatment/saline, 12 IU oxytocin, 24 IU oxytocin) and olfactory stimuli (estratetraenol vs. carrier control) from both the high AQ and low AQ heterosexual men in Experiments 3–5. We found that relative to baseline (no drug treatment/saline), the high AQ and low AQ individuals showed opposite patterns when treated with 12 IU oxytocin, resulting in a significant three-way interaction among social proficiency, oxytocin treatment (no drug treatment/saline vs. 12 IU oxytocin) and olfactory condition (F_{1, 70} = 5.65, p=0.020, partial η² = 0.075), on top of a strong main effect of olfactory condition (F_{1, 70} = 20.63, p<0.0001, partial η² = 0.23) and a marginally significant interaction between social proficiency and oxytocin treatment (F_{1, 70} = 3.13, p=0.081). To illustrate these effects, we plotted in Figure 5A the central tendencies of the estratetraenol-induced PSE shifts (relative to the carrier control) at baseline (x axis) and under 12 IU oxytocin (y axis) from both the high AQ and low AQ heterosexual men in Experiments 3–5. The bootstrapped sample means of the two groups of men formed two discrete clusters: those of high AQ individuals (dark brown dots) fell around the vertical axis on the positive side, indicating that they utilized the feminine information carried by estratetraenol only after 12 IU oxytocin treatment, and those of low AQ individuals (light brown dots) fell around the horizontal axis on the positive side, indicating a significant effect of estratetraenol in them at baseline but not after 12 IU oxytocin treatment. A closer inspection of the data revealed that AQ score, while negatively correlated with estratetraenol-induced PSE shift at baseline (r₇₂ = −0.26, p=0.028), was positively correlated with the increase of estratetraenol-induced PSE shift post 12 IU oxytocin treatment (i.e. estratetraenol-induced PSE shift under 12 IU oxytocin minus that at baseline; r₇₂ = 0.29, p=0.014) (Figure 5B). In other words, one’s social proficiency was predictive of his response to 12 IU oxytocin treatment. Under 24 IU oxytocin, both groups of men became insusceptible to estratetraenol (olfactory condition × social proficiency: F_{1, 70} = 0.13, p=0.72; olfactory condition: F_{1, 70} = 0.11, p=0.74; social proficiency: F_{1, 70} = 0.072, p=0.79), consistent with the results of Experiment 1. Overall, as summarized in Figure 5C, the effect of exogenous oxytocin on the chemosensory processing of sexual cues was dose-dependent, nonmonotonic and contingent upon the recipient’s social proficiency. By contrast, exogenous vasopressin did not exert any significant impact.

Figure 5

Download asset Open asset

At the same time, the difference limens of the high AQ and low AQ heterosexual men in Experiments 3 and 5 did not differ (F_{1, 46} = 0.078, p=0.78) and were comparable across all olfactory conditions and drug treatments (i.e. 12 IU oxytocin, 24 IU oxytocin, 12 IU vasopressin, 24 IU vasopressin, and no drug treatment) (olfactory condition × drug treatment: F_{4, 184} = 1.48, p=0.21; olfactory condition: F_{1, 46} = 1.00, p=0.32; drug treatment: F_{4, 184} = 1.66, p=0.16). Their mood states, as reflected by self-reported ratings on the POMS, were also stable across olfactory conditions and drug treatments (total mood disturbance: ps = 0.76, 0.86, and 0.45, respectively; all subscales: ps > 0.05, corrected). We also specifically examined the difference limens and POMS ratings across different doses of oxytocin (no drug treatment/saline, 12 IU oxytocin and 24 IU oxytocin) and olfactory conditions for all the participants in Experiments 3–5, and obtained the same results (ps > 0.5). Moreover, these participants could not tell apart estratetraenol and the carrier solution alone by smell (mean accuracy = 0.36 vs. chance = 0.33, t₇₁ = 0.93, p=0.36). We therefore concluded that the chemosensory processing of sexual information took place below olfactory awareness. Through interactions with the oxytocin system, it shifted one’s criterion, but not sensitivity, of gender perception, without significantly altering his transient mood state.

The current study represents an initial effort to unravel the neuroendocrine mechanisms underlying human chemo-signaling of sex. Data collected through formal psychophysical testing of 216 individuals over a total of 1056 testing sessions jointly demonstrate that the decoding of chemosensory sexual cues, including that of estratetraenol in heterosexual men and of androstadienone in homosexual men, is modulated by oxytocin instead of vasopressin in a dose-dependent manner, and is blocked by atosiban, a competitive antagonist of both oxytocin and vasopressin receptors (Manning et al., 2012; Table 1).

We are aware of a recent criticism of studies using androstadienone and/or estratetraenol that show a positive effect, which states that these studies were underpowered and problematic due to small sample sizes, lack of a priori evidence of effects, and lack of full replication, and the results are likely false positives (Wyatt, 2015). We, along with other researchers in the field (Endevelt-Shapira et al., 2018), do not agree with this criticism and believe that the current study presents a strong counter-argument. We had clear a priori evidence of the effects of androstadienone and estratetraenol from an earlier study (Zhou et al., 2014). Using the same task as in that study, we consistently replicated the original findings (Table 1), namely, androstadienone signals masculinity to homosexual men (Experiment 2) and estratetraenol signals femininity to heterosexual men (Experiments 1, 3, 4, and 5). The overall effect size was comparable to that of gender adaptation using visually presented faces or bodies (Ghuman et al., 2010). Moreover, whereas the natural occurring concentration of estratetraenol in sweat has not been measured, the concentration of androstadienone (500 µM) presented to the participants was similar to that in freshly produced apocrine sweat (mean = 0.44 nmol/μl = 0.44 × 10⁻³ mol/l = 440 μM) (Gower et al., 1994) and thus arguably ecologically relevant (although likely significantly higher than those encountered in non-intimate social interactions). It was indeed the reliability of these chemosensory effects that allowed us to probe into the underlying neuroendocrine mechanisms and uncover their modulation by oxytocin rather than vasopressin.

Oxytocin and vasopressin differ from each other at only two amino acid positions. Both are strongly implicated in a range of reproductive and social behaviors in animals as well as humans (Carter, 1992; Carter, 2014; Donaldson and Young, 2008). Many of the roles of oxytocin have been associated with female-typical behaviors in animals, and many of the behaviors associated with vasopressin have been demonstrated in males (McCall and Singer, 2012). Nonetheless, the specific behaviors they influence show extensive variation among different species, and to which extent the sex-related differences hold for humans is scantly known (Donaldson and Young, 2008). Here, we found in heterosexual and homosexual men that the utilization of the feminine information carried by estratetraenol, while ‘male-typical’, was not modulated by vasopressin regardless of its dose or the recipient’s social proficiency, but was instead modulated by oxytocin, like the ‘female-typical’ utilization of the masculine information carried by androstadienone. These results hence argue against a clear-cut sex- or sexual orientation-based division of labor between the oxytocin and vasopressin systems in humans.

Importantly, we observed that the oxytocinergic modulation of human chemosensory decoding of sex was dose-dependent, non-monotonic, and roughly followed an inverted-U-shaped function, and that the dose effect of exogenous oxytocin depended on the recipient’s social proficiency. At 12 IU, nasally administered oxytocin restored the ability to utilize the feminine information carried by estratetraenol in socially less proficient heterosexual men, yet hampered the very ability in socially proficient ones. Overall, the less socially proficient a man was, the less he utilized chemosensory sexual information at baseline, the more he became to do so after 12 IU oxytocin treatment, and vice versa. Increasing the dose of oxytocin to 24 IU eliminated the processing of the chemosensory sexual cues irrespective of the recipient’s sexual orientation or social proficiency. It has been shown in rats that the dose-response curve for oxytocin is bell-shaped. Moderate doses of oxytocin facilitate, whereas high doses of oxytocin inhibit penile erections (Argiolas and Gessa, 1991). Similarly, moderate doses of oxytocin promote, whereas high doses of oxytocin disrupt social memory (Benelli et al., 1995). Moreover, a marked increase of oxytocin has been suspected to mediate or signal satiety and contribute to a refractory state in animals (Carter, 1992). In humans, there has also been hints that the effect of intranasal oxytocin is dose-dependent (Cardoso et al., 2013; Quintana et al., 2017). A recent study indicates that pretreatment blood oxytocin concentrations predict treatment response to a set dosage of intranasal oxytocin in children with autism (Parker et al., 2017). Several others note that the influence of intranasal oxytocin on socially proficient individuals is complicated, and can be null or even antisocial (Bartz et al., 2011). Our findings dovetail with these reports. Moreover, they provide strong behavioral evidence for a non-monotonic effect of intranasal oxytocin that interacts with the recipient’s social proficiency. While we tested healthy male volunteers, the non-monotonic property of oxytocin’s effects has significant implications in the design of oxytocin-based treatment protocols for conditions like autism spectrum disorder.

Following the work of Kosfeld et al., 2005, the majority of studies on the effect of oxytocin in humans have used a single intranasal dose of 24 IU and shown a prosocial effect. Most have not simultaneously evaluated the effect of vasopressin or an antagonist of oxytocin or vasopressin receptors. In the limited studies noting that a higher dose of oxytocin produces a smaller effect, the phenomenon is speculated to be a result of cross-binding to vasopressin receptors that ‘cancels out’ the oxytocinergic effect (Cardoso et al., 2013). In our study, only at 12 IU did intranasal oxytocin significantly facilitate the decoding of sexual information in socially less proficient individuals. There was no significant effect of vasopressin irrespective of the dose. Thus, the lack of a positive effect of 24 IU oxytocin could not be explained by cross-binding of oxytocin to vasopressin receptors. Rather, we argue that it reflects the dose-response characteristics of oxytocin. Moreover, in view of the well-documented prosocial effect of 24 IU intranasal oxytocin, our results suggest that the optimal oxytocin level for the decoding of chemosensory sexual information differs from that for the promotion of prosocial behavior. After all, social behavior evolves as a result of effects upon the reproductive competition among group members (Alexander, 1974).

In rats, oxytocin has been found to enhance social recognition by modulating cortical control of early olfactory processing (Oettl et al., 2016). What neural pathways subserve the observed oxytocinergic modulation of human chemosensory communication? How does oxytocin act on sexually dimorphic sensory processing? These interesting questions await further research to clarify. Given the evolutionary and anatomical intimacy between the olfactory system and the neuroendocrine system, particularly through the hypothalamus (Gorbman, 1995), analyses of their interplays will open up new avenues for the regulations of both sensory perception of the external world and physiological processes within the human body.

All data analyzed during this study are included in the manuscript and supporting files. Source data files have been provided for Figures 2, 3 and 4.

1. 1. Alexander RD

   (1974) The evolution of social behavior

   Annual Review of Ecology and Systematics 5:325–383.

   https://doi.org/10.1146/annurev.es.05.110174.001545

   • Google Scholar
2. 1. Argiolas A
   2. Gessa GL

   (1991) Central functions of oxytocin

   Neuroscience & Biobehavioral Reviews 15:217–231.

   https://doi.org/10.1016/S0149-7634(05)80002-8

   • PubMed
   • Google Scholar
3. 1. Baron-Cohen S
   2. Wheelwright S
   3. Skinner R
   4. Martin J
   5. Clubley E

   (2001) The autism-spectrum quotient (AQ): evidence from asperger syndrome/high-functioning autism, males and females, scientists and mathematicians

   Journal of Autism and Developmental Disorders 31:5–17.

   https://doi.org/10.1023/a:1005653411471

   • PubMed
   • Google Scholar
4. 1. Bartz JA
   2. Zaki J
   3. Bolger N
   4. Hollander E
   5. Ludwig NN
   6. Kolevzon A
   7. Ochsner KN

   (2010) Oxytocin selectively improves empathic accuracy

   Psychological Science 21:1426–1428.

   https://doi.org/10.1177/0956797610383439

   • PubMed
   • Google Scholar
5. 1. Bartz JA
   2. Zaki J
   3. Bolger N
   4. Ochsner KN

   (2011) Social effects of oxytocin in humans: context and person matter

   Trends in Cognitive Sciences 15:301–309.

   https://doi.org/10.1016/j.tics.2011.05.002

   • PubMed
   • Google Scholar
6. 1. Bartz JA
   2. Nitschke JP
   3. Krol SA
   4. Tellier P-P

   (2019) Oxytocin selectively improves empathic accuracy: a replication in men and novel insights in women

   Biological Psychiatry: Cognitive Neuroscience and Neuroimaging 4:1042–1048.

   https://doi.org/10.1016/j.bpsc.2019.01.014

   • Google Scholar
7. 1. Benelli A
   2. Bertolini A
   3. Poggioli R
   4. Menozzi B
   5. Basaglia R
   6. Arletti R

   (1995) Polymodal dose-response curve for oxytocin in the social recognition test

   Neuropeptides 28:251–255.

   https://doi.org/10.1016/0143-4179(95)90029-2

   • PubMed
   • Google Scholar
8. 1. Bensafi M
   2. Brown WM
   3. Khan R
   4. Levenson B
   5. Sobel N

   (2004) Sniffing human sex-steroid derived compounds modulates mood, memory and autonomic nervous system function in specific behavioral contexts

   Behavioural Brain Research 152:11–22.

   https://doi.org/10.1016/j.bbr.2003.09.009

   • PubMed
   • Google Scholar
9. 1. Berglund H
   2. Lindström P
   3. Savic I

   (2006) Brain response to putative pheromones in lesbian women

   PNAS 103:8269–8274.

   https://doi.org/10.1073/pnas.0600331103

   • PubMed
   • Google Scholar
10. 1. Blouin AM
    2. Fried I
    3. Wilson CL
    4. Staba RJ
    5. Behnke EJ
    6. Lam HA
    7. Maidment NT
    8. Karlsson KÆ
    9. Lapierre JL
    10. Siegel JM

    (2013) Human hypocretin and melanin-concentrating hormone levels are linked to emotion and social interaction

    Nature Communications 4:1547.

    https://doi.org/10.1038/ncomms2461

    • PubMed
    • Google Scholar
11. 1. Born J
    2. Lange T
    3. Kern W
    4. McGregor GP
    5. Bickel U
    6. Fehm HL

    (2002) Sniffing neuropeptides: a transnasal approach to the human brain

    Nature Neuroscience 5:514–516.

    https://doi.org/10.1038/nn0602-849

    • PubMed
    • Google Scholar
12. 1. Cardoso C
    2. Ellenbogen MA
    3. Orlando MA
    4. Bacon SL
    5. Joober R

    (2013) Intranasal oxytocin attenuates the cortisol response to physical stress: a dose-response study

    Psychoneuroendocrinology 38:399–407.

    https://doi.org/10.1016/j.psyneuen.2012.07.013

    • PubMed
    • Google Scholar
13. 1. Carter CS

    (1992) Oxytocin and sexual behavior

    Neuroscience & Biobehavioral Reviews 16:131–144.

    https://doi.org/10.1016/S0149-7634(05)80176-9

    • PubMed
    • Google Scholar
14. 1. Carter CS

    (2014) Oxytocin pathways and the evolution of human behavior

    Annual Review of Psychology 65:17–39.

    https://doi.org/10.1146/annurev-psych-010213-115110

    • PubMed
    • Google Scholar
15. 1. Cusson M
    2. McNeil JN

    (1989) Involvement of juvenile hormone in the regulation of pheromone release activities in a moth

    Science 243:210–212.

    https://doi.org/10.1126/science.243.4888.210

    • PubMed
    • Google Scholar
16. Book

    1. Davison AC
    2. Hinkley DV

    (1997) Bootstrap Methods and Their Application

    Cambridge: Cambridge University Press.

    https://doi.org/10.1017/CBO9780511802843

    • Google Scholar
17. 1. de Groot JH
    2. Smeets MA
    3. Rowson MJ
    4. Bulsing PJ
    5. Blonk CG
    6. Wilkinson JE
    7. Semin GR

    (2015) A sniff of happiness

    Psychological Science 26:684–700.

    https://doi.org/10.1177/0956797614566318

    • PubMed
    • Google Scholar
18. 1. Donaldson ZR
    2. Young LJ

    (2008) Oxytocin, vasopressin, and the neurogenetics of sociality

    Science 322:900–904.

    https://doi.org/10.1126/science.1158668

    • PubMed
    • Google Scholar
19. 1. Endevelt-Shapira Y
    2. Perl O
    3. Ravia A
    4. Amir D
    5. Eisen A
    6. Bezalel V
    7. Rozenkrantz L
    8. Mishor E
    9. Pinchover L
    10. Soroka T
    11. Honigstein D
    12. Sobel N

    (2018) Altered responses to social chemosignals in autism spectrum disorder

    Nature Neuroscience 21:111–119.

    https://doi.org/10.1038/s41593-017-0024-x

    • PubMed
    • Google Scholar
20. 1. Freeman SM
    2. Samineni S
    3. Allen PC
    4. Stockinger D
    5. Bales KL
    6. Hwa GG
    7. Roberts JA

    (2016) Plasma and CSF oxytocin levels after intranasal and intravenous oxytocin in awake macaques

    Psychoneuroendocrinology 66:185–194.

    https://doi.org/10.1016/j.psyneuen.2016.01.014

    • PubMed
    • Google Scholar
21. 1. Ghuman AS
    2. McDaniel JR
    3. Martin A

    (2010) Face adaptation without a face

    Current Biology 20:32–36.

    https://doi.org/10.1016/j.cub.2009.10.077

    • PubMed
    • Google Scholar
22. 1. Gorbman A

    (1995) Olfactory origins and evolution of the brain-pituitary endocrine system: facts and speculation

    General and Comparative Endocrinology 97:171–178.

    https://doi.org/10.1006/gcen.1995.1016

    • PubMed
    • Google Scholar
23. 1. Gower DB
    2. Holland KT
    3. Mallet AI
    4. Rennie PJ
    5. Watkins WJ

    (1994) Comparison of 16-androstene steroid concentrations in sterile apocrine sweat and axillary secretions: interconversions of 16-androstenes by the axillary microflora--a mechanism for axillary odour production in man?

    The Journal of Steroid Biochemistry and Molecular Biology 48:409–418.

    https://doi.org/10.1016/0960-0760(94)90082-5

    • PubMed
    • Google Scholar
24. 1. Gower DB
    2. Ruparelia BA

    (1993) Olfaction in humans with special reference to odorous 16-androstenes: their occurrence, perception and possible social, psychological and sexual impact

    Journal of Endocrinology 137:167–187.

    https://doi.org/10.1677/joe.0.1370167

    • Google Scholar
25. 1. Insel TR
    2. Young L
    3. Witt DM
    4. Crews D

    (1993) Gonadal steroids have paradoxical effects on brain oxytocin receptors

    Journal of Neuroendocrinology 5:619–628.

    https://doi.org/10.1111/j.1365-2826.1993.tb00531.x

    • PubMed
    • Google Scholar
26. 1. Jacob S
    2. McClintock MK

    (2000) Psychological state and mood effects of steroidal chemosignals in women and men

    Hormones and Behavior 37:57–78.

    https://doi.org/10.1006/hbeh.1999.1559

    • PubMed
    • Google Scholar
27. Book

    1. Jones RE
    2. Lopez KH

    (2013)

    Human Reproductive Biology (Fourth Edition)

    Academic Press.

    • Google Scholar
28. 1. Karlson P
    2. Luscher M

    (1959) Pheromones': a new term for a class of biologically active substances

    Nature 183:55–56.

    https://doi.org/10.1038/183055a0

    • PubMed
    • Google Scholar
29. 1. Kosfeld M
    2. Heinrichs M
    3. Zak PJ
    4. Fischbacher U
    5. Fehr E

    (2005) Oxytocin increases trust in humans

    Nature 435:673–676.

    https://doi.org/10.1038/nature03701

    • PubMed
    • Google Scholar
30. 1. Koven NS
    2. Max LK

    (2014) Basal salivary oxytocin level predicts extra- but not intra-personal dimensions of emotional intelligence

    Psychoneuroendocrinology 44:20–29.

    https://doi.org/10.1016/j.psyneuen.2014.02.018

    • PubMed
    • Google Scholar
31. 1. Lancaster K
    2. Carter CS
    3. Pournajafi-Nazarloo H
    4. Karaoli T
    5. Lillard TS
    6. Jack A
    7. Davis JM
    8. Morris JP
    9. Connelly JJ

    (2015) Plasma oxytocin explains individual differences in neural substrates of social perception

    Frontiers in Human Neuroscience 9:132.

    https://doi.org/10.3389/fnhum.2015.00132

    • PubMed
    • Google Scholar
32. 1. Lin HH
    2. Cao DS
    3. Sethi S
    4. Zeng Z
    5. Chin JSR
    6. Chakraborty TS
    7. Shepherd AK
    8. Nguyen CA
    9. Yew JY
    10. Su CY
    11. Wang JW

    (2016) Hormonal modulation of pheromone detection enhances male courtship success

    Neuron 90:1272–1285.

    https://doi.org/10.1016/j.neuron.2016.05.004

    • PubMed
    • Google Scholar
33. 1. Liu R
    2. Yuan X
    3. Chen K
    4. Jiang Y
    5. Zhou W

    (2018) Perception of social interaction compresses subjective duration in an oxytocin-dependent manner

    eLife 7:e32100.

    https://doi.org/10.7554/eLife.32100

    • PubMed
    • Google Scholar
34. 1. Lundin S
    2. Akerlund M
    3. Fagerström PO
    4. Hauksson A
    5. Melin P

    (1986) Pharmacokinetics in the human of a new synthetic vasopressin and oxytocin uterine antagonist

    Acta Endocrinologica 112:465–472.

    https://doi.org/10.1530/acta.0.1120465

    • PubMed
    • Google Scholar
35. 1. Lundström JN
    2. Gonçalves M
    3. Esteves F
    4. Olsson MJ

    (2003a) Psychological effects of subthreshold exposure to the putative human pheromone 4,16-androstadien-3-one

    Hormones and Behavior 44:395–401.

    https://doi.org/10.1016/j.yhbeh.2003.06.004

    • PubMed
    • Google Scholar
36. 1. Lundström JN
    2. Hummel T
    3. Olsson MJ

    (2003b) Individual differences in sensitivity to the odor of 4,16-androstadien-3-one

    Chemical Senses 28:643–650.

    https://doi.org/10.1093/chemse/bjg057

    • PubMed
    • Google Scholar
37. 1. Manning M
    2. Misicka A
    3. Olma A
    4. Bankowski K
    5. Stoev S
    6. Chini B
    7. Durroux T
    8. Mouillac B
    9. Corbani M
    10. Guillon G

    (2012) Oxytocin and vasopressin agonists and antagonists as research tools and potential therapeutics

    Journal of Neuroendocrinology 24:609–628.

    https://doi.org/10.1111/j.1365-2826.2012.02303.x

    • PubMed
    • Google Scholar
38. 1. McCall C
    2. Singer T

    (2012) The animal and human neuroendocrinology of social cognition, motivation and behavior

    Nature Neuroscience 15:681–688.

    https://doi.org/10.1038/nn.3084

    • PubMed
    • Google Scholar
39. 1. McClintock MK

    (1971) Menstrual synchorony and suppression

    Nature 229:244–245.

    https://doi.org/10.1038/229244a0

    • PubMed
    • Google Scholar
40. Book

    1. McNair DM
    2. Lorr M
    3. Droppleman LF

    (1971)

    Profile of Mood States

    San Francisco: Educational and Industrial Testing Service.

    • Google Scholar
41. 1. Mugford RA
    2. Nowell NW

    (1971) Endocrine control over production and activity of the anti-aggression pheromone from female mice

    Journal of Endocrinology 49:225–232.

    https://doi.org/10.1677/joe.0.0490225

    • Google Scholar
42. 1. Oettl LL
    2. Ravi N
    3. Schneider M
    4. Scheller MF
    5. Schneider P
    6. Mitre M
    7. da Silva Gouveia M
    8. Froemke RC
    9. Chao MV
    10. Young WS
    11. Meyer-Lindenberg A
    12. Grinevich V
    13. Shusterman R
    14. Kelsch W

    (2016) Oxytocin enhances social recognition by modulating cortical control of early olfactory processing

    Neuron 90:609–621.

    https://doi.org/10.1016/j.neuron.2016.03.033

    • PubMed
    • Google Scholar
43. 1. Olsson MJ
    2. Lundström JN
    3. Diamantopoulou S
    4. Esteves F

    (2006) A putative female pheromone affects mood in men differently depending on social context

    European Review of Applied Psychology 56:279–284.

    https://doi.org/10.1016/j.erap.2005.09.010

    • Google Scholar
44. 1. Parker KJ
    2. Garner JP
    3. Libove RA
    4. Hyde SA
    5. Hornbeak KB
    6. Carson DS
    7. Liao CP
    8. Phillips JM
    9. Hallmayer JF
    10. Hardan AY

    (2014) Plasma oxytocin concentrations and OXTR polymorphisms predict social impairments in children with and without autism spectrum disorder

    PNAS 111:12258–12263.

    https://doi.org/10.1073/pnas.1402236111

    • PubMed
    • Google Scholar
45. 1. Parker KJ
    2. Oztan O
    3. Libove RA
    4. Sumiyoshi RD
    5. Jackson LP
    6. Karhson DS
    7. Summers JE
    8. Hinman KE
    9. Motonaga KS
    10. Phillips JM
    11. Carson DS
    12. Garner JP
    13. Hardan AY

    (2017) Intranasal oxytocin treatment for social deficits and biomarkers of response in children with autism

    PNAS 114:8119–8124.

    https://doi.org/10.1073/pnas.1705521114

    • PubMed
    • Google Scholar
46. 1. Quintana DS
    2. Westlye LT
    3. Hope S
    4. Nærland T
    5. Elvsåshagen T
    6. Dørum E
    7. Rustan Ø
    8. Valstad M
    9. Rezvaya L
    10. Lishaugen H
    11. Stensønes E
    12. Yaqub S
    13. Smerud KT
    14. Mahmoud RA
    15. Djupesland PG
    16. Andreassen OA

    (2017) Dose-dependent social-cognitive effects of intranasal oxytocin delivered with novel breath powered device in adults with autism spectrum disorder: a randomized placebo-controlled double-blind crossover trial

    Translational Psychiatry 7:e1136.

    https://doi.org/10.1038/tp.2017.103

    • PubMed
    • Google Scholar
47. 1. Raina AK
    2. Jaffe H
    3. Kempe TG
    4. Keim P
    5. Blacher RW
    6. Fales HM
    7. Riley CT
    8. Klun JA
    9. Ridgway RL
    10. Hayes DK

    (1989) Identification of a neuropeptide hormone that regulates sex pheromone production in female moths

    Science 244:796–798.

    https://doi.org/10.1126/science.244.4906.796

    • PubMed
    • Google Scholar
48. 1. Savic I
    2. Berglund H
    3. Gulyas B
    4. Roland P

    (2001) Smelling of odorous sex hormone-like compounds causes sex-differentiated hypothalamic activations in humans

    Neuron 31:661–668.

    https://doi.org/10.1016/S0896-6273(01)00390-7

    • PubMed
    • Google Scholar
49. 1. Savic I
    2. Berglund H
    3. Lindström P

    (2005) Brain response to putative pheromones in homosexual men

    PNAS 102:7356–7361.

    https://doi.org/10.1073/pnas.0407998102

    • PubMed
    • Google Scholar
50. 1. Scheele D
    2. Plota J
    3. Stoffel-Wagner B
    4. Maier W
    5. Hurlemann R

    (2016) Hormonal contraceptives suppress oxytocin-induced brain reward responses to the partner's face

    Social Cognitive and Affective Neuroscience 11:767–774.

    https://doi.org/10.1093/scan/nsv157

    • PubMed
    • Google Scholar
51. 1. Simerly RB

    (2002) Wired for reproduction: organization and development of sexually dimorphic circuits in the mammalian forebrain

    Annual Review of Neuroscience 25:507–536.

    https://doi.org/10.1146/annurev.neuro.25.112701.142745

    • PubMed
    • Google Scholar
52. 1. Spengler FB
    2. Schultz J
    3. Scheele D
    4. Essel M
    5. Maier W
    6. Heinrichs M
    7. Hurlemann R

    (2017) Kinetics and dose dependency of intranasal oxytocin effects on amygdala reactivity

    Biological Psychiatry 82:885–894.

    https://doi.org/10.1016/j.biopsych.2017.04.015

    • PubMed
    • Google Scholar
53. 1. Stern K
    2. McClintock MK

    (1998) Regulation of ovulation by human pheromones

    Nature 392:177–179.

    https://doi.org/10.1038/32408

    • PubMed
    • Google Scholar
54. 1. Stoop R

    (2012) Neuromodulation by oxytocin and vasopressin

    Neuron 76:142–159.

    https://doi.org/10.1016/j.neuron.2012.09.025

    • PubMed
    • Google Scholar
55. 1. Tang JD
    2. Charlton RE
    3. Jurenka RA
    4. Wolf WA
    5. Phelan PL
    6. Sreng L
    7. Roelofs WL

    (1989) Regulation of pheromone biosynthesis by a brain hormone in two moth species

    PNAS 86:1806–1810.

    https://doi.org/10.1073/pnas.86.6.1806

    • PubMed
    • Google Scholar
56. 1. Thysen B
    2. Elliott WH
    3. Katzman PA

    (1968) Identification of estra-1,3,5(10),16-tetraen-3-ol (estratetraenol) from the urine of pregnant women (1)

    Steroids 11:73–87.

    https://doi.org/10.1016/S0039-128X(68)80052-2

    • PubMed
    • Google Scholar
57. 1. Troje NF

    (2002) Decomposing biological motion: a framework for analysis and synthesis of human gait patterns

    Journal of Vision 2:2–387.

    https://doi.org/10.1167/2.5.2

    • Google Scholar
58. 1. Wyatt TD

    (2015) The search for human pheromones: the lost decades and the necessity of returning to first principles

    Proceedings of the Royal Society B: Biological Sciences 282:20142994.

    https://doi.org/10.1098/rspb.2014.2994

    • Google Scholar
59. 1. Ye Y
    2. Zhuang Y
    3. Smeets MAM
    4. Zhou W

    (2019) Human chemosignals modulate emotional perception of biological motion in a sex-specific manner

    Psychoneuroendocrinology 100:246–253.

    https://doi.org/10.1016/j.psyneuen.2018.10.014

    • PubMed
    • Google Scholar
60. 1. Zhou W
    2. Yang X
    3. Chen K
    4. Cai P
    5. He S
    6. Jiang Y

    (2014) Chemosensory communication of gender through two human steroids in a sexually dimorphic manner

    Current Biology 24:1091–1095.

    https://doi.org/10.1016/j.cub.2014.03.035

    • PubMed
    • Google Scholar
61. 1. Zhou W
    2. Chen D

    (2009) Fear-related chemosignals modulate recognition of fear in ambiguous facial expressions

    Psychological Science 20:177–183.

    https://doi.org/10.1111/j.1467-9280.2009.02263.x

    • PubMed
    • Google Scholar

Acceptance summary:

We are so pleased that you have made a necessarily complex paradigm involving several areas of expertise understandable to the general reader in service of an exciting advance of our understanding of chemosensory communication in humans. Clearly we are not moths. Yet even though human chemo-communication is not dramatic it nonetheless exists and influences perception, decisions, behavior in nuanced ways such as the compelling example that you demonstrate here.

Decision letter after peer review:

Thank you for submitting your article "Oxytocin mediates human chemosensory communication of sex in a dose-dependent manner" for consideration by eLife. Your article has been reviewed by four peer reviewers including Peggy Mason as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen Catherine Dulac as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Gün R. Semin (Reviewer #2).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, we are asking editors to accept without delay manuscripts, like yours, that they judge can stand as eLife papers without additional data, even if they feel that they would make the manuscript stronger. Thus the revisions requested below only address clarity and presentation.

All reviewers found the work to be interesting, fundamental and innovative. Hormones (OXT, VAS), gender perception using point light walking, and chemosensation intersect in this study making it inherently complex and somewhat challenging to the reader. The first point of revision is to clarify and hand-hold a bit more through the results and also to explain some of the expectations which were not obvious to the reviewers and are unlikely to be obvious to the general reader.

Beyond clarification, please also respond to the specific suggestions in the full reviews attached below.

Reviewer #1:

This innovative study examines the role of oxt and vas in gender perception using point light walking and administered female and male smells. The study is complex and difficult to keep track of but that is inherent to the design and the authors do a good job. I found the data compelling and the interpretations interesting.

In Figure 2E,J and Figure 3, the graphs should show some metric of the data rather than the shift so that the no drug condition and its variability can be illustrated.

Subsection “Oxytocin, vasopressin and subconscious chemosensory decoding of sex in heterosexual and homosexual men” “it follows that individuals with different endogenous oxytocin level would differ in their susceptibility to such chemosignals as well as to the effect of exogenous oxytocin.” Why does this follow?

Reviewer #2:

Overall, this is a very impressive study that logically builds across a set of 5 experiments a convincing case to uncover the neuroendocrine mechanisms involved in the chemosensory communication of sex and show that oxytocin and vasopressin have different roles to play in this process.

Some trivial observations:

1) I would have appreciated information about which the criteria they used -if any – to decide about sample sizes.

2) It would have been useful if the authors would have argued as to why they did not use heterosexual and lesbian women participants?

Reviewer #3:

In this research, the authors investigated the biological substrate(s) of chemosensory decoding of femininity, specifically focusing on the oxytocin and, closely related, vasopressin systems based on their roles in reproductive and social behavior. Results showed that oxytocin, but not vasopressin, plays a causal role in chemosensory communication in humans. The paper was well-written and I particularly liked how they used an oxytocin agonist AND antagonist to support their claims, as well as vasopressin to speak to discriminant validity. This work makes an important contribution to our understanding of the biological mechanisms that support human social information processing.

My first concern has to do with sample size, which seems on the small side. That said, I do appreciate how difficult it is to conduct these kinds of drug administration studies, and the fact that the authors ran multiple studies and reported consistent effects across studies, so I am torn. Perhaps the authors could provide additional information on statistical power. The authors do address statistical power (subsection “Participants”):

"…Sample sizes (n = 24 in each subgroup) were determined by G*Power to be adequate to detect a moderate effect of androstadienone or estratetraenol (d ≈ 0.6), at 80% power. The effect size was estimated based on an earlier study that employed almost identical stimuli and psychophysical testing procedures to those in the current study (Zhou et al., 2014)."

However, it seems that they are only reporting the power to detect the effect of androstadienone and estratetraenol on chemosensory communication, NOT the moderation by drug, or the moderation by individual differences (AQ, sexual orientation). Can the authors please speak to these issues? Especially for Study 3 where there appear to be 10 conditions (12 IU OT, 24 IU OT, 12 IU AVP, 24 IU AVP, PL x estratetraenol vs. carrier).

My second concern has to do with the author's assertion that less socially proficient individuals have lower levels of endogenous and the reason why oxytocin should be helpful to them is BECAUSE they have lower levels of oxytocin. For example, as the authors write:

Abstract: "…and contingent upon the recipients' social proficiency – a partial manifestation of their endogenous oxytocin level."

Discussion section: "…such that the dose effect of exogenous oxytocin depended on the recipient's social proficiency, which in turn partially reflected his endogenous oxytocin level (Parker et al., 2014)."

Discussion section: "Moreover, they provide strong behavioral evidence for a non-monotonic effect of intranasal oxytocin that interacts with the recipient's social proficiency or endogenous oxytocin level."

This last statement is particularly problematic given that the authors did not actually measure participants endogenous OT, but the statement makes it seem like they did.

While there is evidence linking endogenous OT levels with social proficiency, to my knowledge, no one has demonstrated that exogenous OT selectively benefits less socially proficient individuals because it alters endogenous OT levels. Given this, the authors might want to re-think their rationale for Experiments 3 and 4. Actually, there is plenty of empirical evidence for the selectively beneficial effects of oxytocin for individuals who are less socially proficient (e.g., Luminet et al., 2011; Radke and de Bruijn; 2015; Feeser et al., 2015), specifically as assessed by the AQ (Bartz et al., 2010; 2019)-I think citing that research is sufficient justification to look at AQ as a moderator in Experiment 3 and 4. Of course, the authors can mention the endogenous OT-ASD findings; I just wouldn't make explicit claims about mechanism as it seems like they are (unnecessarily) going out on a limb.

Reviewer #4:

This study aims to test whether intranasal administration of the neuropeptides Oxytocin (OXT) and Vasopressin (AVP) can mediate the effect of two putative human chemosignals – androsta-4,16,-dien-33-one (AND) and estra-1,3,5(10),16-tetraen-3-ol (EST). To test this hypothesis, the authors used a behavioral task named PLW, in which participants determine the gender (female or male) of a dot-figure. In the manuscript, they detailed 5 experiments which provide evidence for a link between OXT and the effect of EST.

In general, the manuscript is novel, and provides an important contribution, yet there are a few points which should be addressed:

1) The manuscript is a bit hard to follow. Though it is divided to sets of experiments per test focus, it is hard to follow the line of thought which lead to each experiment, what was the hypothesis raised and the way they wanted to test it.

2) In the main Experiments (1-5) the statistical method used is paired t-test. Data of this sort should be statistically tested using one statistical test (e.g. ANOVA instead of multiple t-tests), with factors of odor and participant.

3) Were there corrections for multiple comparisons applied? Please mention this in the text.

4) It is not clear whether all experiments were double-blind. Were both-experimenter and subject not aware of the drugs and odors administrated in all experiments? If not, this should be clearly acknowledged.

5) According to what did the authors choose a cut-off of 25 score for AQ? There is no supporting for this cut-off in the reference they provided, and in a previous study by the same group (2018) they chose a cut-off of 20. This should be clarified.

6) In regards to the olfactory stimuli, in the Materials and methods section it is mentioned "Participants were instructed to hold the jar with their non-dominant hand, position the nosepieces inside their nostrils and continuously inhale through the nose and exhale through their mouth throughout each block of the experiments". Did the authors verify using breathing measurements (e.g. spirometer) whether the participants actually inhaled from their nose and exhaled from their mouth? If not, this could add variability to the results and should be mentioned.

7) In the Results section it is mentioned as "The three olfactory stimuli were perceptually indiscriminable as assessed in a separate panel of 48 male participants (mean triangular discrimination accuracy = 0.33 vs. chance = 0.33, t47 = 0.02, p > 0.9)". However, in the methods section, it is mentioned "The effectiveness of the clove oil carrier solution as a masker for the odors of androstadienone and estratetraenol was verified beforehand in an independent group of 48 healthy male nonsmokers (22.71 {plus minus} 2.37 yrs; triangular test, mean accuracy {plus minus} SD = 0.33 {plus minus} 0.15 vs. chance = 0.33, p > 0.99)….…Since the psychological effects of androstadienone is unrelated to one's sensitivity to its odor (Lundström et al., 2003) and estratetraenol is generally regarded as odorless (Lundstrom, Hummel and Olsson, 2003), we did not assess individuals' sensitivities to androstadienone or estratetraenol alone". It is not clear whether the authors did actually test, or how did they test, if participants could discriminate or perceive the smell of these odorants, and was this tested separately per each odor stimulus? There is evidence that AND has an odor [Keller et al., 2007] and can be perceived as unpleasant to participants. If there were perceptual differences this may affect the interpretation of the results.

8) Throughout the manuscript the authors refer to "social proficiency" and participants' basal OXT levels, this without measuring their basal OXT levels. This issue has further implications due to their claim that OXT affects the response to EST, however, this claim becomes circular when basal levels are not measured nor taken into account when reaching conclusions of the effect of OXT.

9) Please rephrase the sentence where you suggest that using homosexual men is a replacement for women. (Discussion section)

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your article "Oxytocin mediates human chemosensory communication of sex in a dose-dependent manner" for consideration by eLife. Your revised article has been reviewed by three peer reviewers including Peggy Mason as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Catherine Dulac as the Senior Editor.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, we are asking editors to accept without delay manuscripts, like yours, that they judge can stand as eLife papers without additional data, even if they feel that they would make the manuscript stronger. Thus the revisions requested below only address clarity and presentation.

This manuscript has been responsive to the reviews. However, the writing remains dense and unfortunately errors in referring to the figures do not help the reader to decode this revision. Please take this opportunity to clean up the text and figures so that the revision can be fully appreciated and evaluated.

Reviewer #1:

This study remains quite complex and unfortunately errors in referring to the figures do not help the reader to decode this revision.

I would like the errors to be fixed so that a clear read can be had. Critical errors are in the referring to Figure 2, which while only one of 5 figures, is foundational to the story. Confusing the reader during the establishment of the basic paradigm is problematic.

First, it would appear that the men and women labels at the bottom of Figure 2F are reversed. If they are not, then I am confused. I looked at this easily 5-10 times and with those labels the figure does not compute.

Results section:

Figure 2B should be 2F?

should be Figure 2B G E J

Should be Figure 2D,I not G I

Why focus on VAS instead of OXT?

Reviewer #3:

Overall I think the authors did a fine job responding to the reviewers' comments, questions and suggestions. I just have a few remaining items:

1) Clarity about participants and experiments:

I raised this issue before-and I appreciate the changes the authors made-but I still feel they need to be more transparent about participants/studies/experiments and sessions:

"Psychophysical data collected from 216 heterosexual and homosexual men over 1,056 sessions consistently showed that such chemosensory communications of sex were blocked by a competitive antagonist of both oxytocin and vasopressin receptors called atosiban, administered nasally"

This still isn't entirely transparent as the sample size collapses across 5 experiments, thus giving the impression of greater statistical power than is warranted (per experiment/test). Please present sample size for each individual experiment here, or at least indicate that the 216 ps comprise 5 experiments.

"Systematic comparisons of these psychometric curves across 5 experiments totaling 1,056 sessions enabled…"

As per above, please state number of ps here for transparency.

Also, I find the term “sessions” to be confusing. Maybe this is a convention specific to their research area, but, to me, “sessions” suggests testing sessions, whereas it seems like the authors are referring to trials on the task?

I think reviewer 4 made a similar comment, but I also found the overall design to be unclear:

Are Experiments 1 and 2 identical expect that Experiment 1 has heterosexual men and Experiment 2 has homosexual men as participants? If so, why are they called two different experiments since the results are discussed side by side and, in fact, the authors combine the data anyway with the omnibus ANOVA?

More generally, is this one “Study” with multiple “Experiments” embedded within the overall Study? The Participants section is written this way (i.e., "A total of 216 young male adults participated in the main study…") but the way the Experiments are presented in the main text I had the impression that they were separate studies, with different participants and different designs, not one overall study, with experiment referring to testing different effects.

2) Predictions about oxytocin:

As detailed below, I was a bit surprised by some of the findings and I think the authors could do a better job walking readers through. Note: I am in no way suggesting that the authors should state hypotheses that they did not have a priori, but I think they need to take more care laying the groundwork for readers. See my suggestions below:

"…could be affected in either direction by the administration of oxytocin or vasopressin."

I was surprised that the authors did not have specific predictions for OT, given that they did have predictions about atosiban. Since OT is an agonist, wouldn't one expect the opposite effect of the antagonist atosiban? Again, I am not advocating that the authors make predictions post hoc, but I think they need to clarify *why* they did not have predictions for OT since it's not obvious (i.e., given they had predictions for atosiban).

"Smelling estratetraenol and androstadienone, relative to the carrier solution alone, failed to respectively influence the gender perception criteria (indexed by the PSEs) of the atosiban-treated heterosexual and homosexual men (t23s = 1.37 and 0.88, ps = 0.18 and 0.39, respectively; Figure 2C, D, I, J). This appeared to be the case for those treated with 24 IU oxytocin as well (t23s = 0.17 and -1.29, ps = 0.87 and 0.21, respectively; Figure 2E, F, I, J)."

I was also surprised that oxytocin and atosiban produced the same effect, one being an agonist and the other being an antagonist. I appreciate that the authors address this in the discussion, and I basically find their rationale to be compelling, but my sense is that they gloss over this surprising finding here and elsewhere (see below). I think they could do a better job of guiding the reader along by NOT glossing over this point. E.g., something like "Interestingly, this appeared to be the case for those treated with 24 IU oxytocin as well (t23s = 0.17 and -1.29, ps = 0.87 and 0.21, respectively; Figure 2E, F, I, J)."

I have similar issues with the following statements:

"…were subserved by similar neuroendocrine mechanisms that were disrupted by intranasal atosiban, the competitive antagonist of both oxytocin and vasopressin receptors, and unaffected by intranasal vasopressin."

Didn't oxytocin exert the same effect as atosiban? But oxytocin is not mentioned.

"The question remained as to why the administration of 24 IU oxytocin exempted the participants from the influences of the chemosignals, and we explored it in more detail."

I would suggest the following modification for transparency: "the question remained as to why the administration of 24IU oxytocin exerted the same effect as atosiban…"

Rather than glossing over this surprising finding, I think it would be more effective to acknowledge it head on.

All reviewers found the work to be interesting, fundamental and innovative. Hormones (OXT, VAS), gender perception using point light walking, and chemosensation intersect in this study making it inherently complex and somewhat challenging to the reader. The first point of revision is to clarify and hand-hold a bit more through the results and also to explain some of the expectations which were not obvious to the reviewers and are unlikely to be obvious to the general reader.

We have clarified the rationales, hypotheses, and results of the experiments in the revised manuscript. Please also refer to our responses to reviewer 1’s points 1 and 2, reviewer 3’s points 2 and 8, and reviewer 4’s points 1-3 below.

Reviewer #1:

This innovative study examines the role of oxt and vas in gender perception using point light walking and administered female and male smells. The study is complex and difficult to keep track of but that is inherent to the design and the authors do a good job. I found the data compelling and the interpretations interesting.

We thank the reviewer for the positive assessment of our work.

1) In Figure 2E,J and Figure 3, the graphs should show some metric of the data rather than the shift so that the no drug condition and its variability can be illustrated.

The PSE shifts in Figure 2E,J and Figure 3 are relative to their respective control conditions (clove oil carrier solution alone). We have now included a new figure in the revised manuscript (Figure 3—figure supplement 1) that illustrates the central tendencies of the PSEs under each combination of olfactory (androstadienone, estratetraenol, carrier control) and drug conditions (60 μg of atosiban, 24 IU oxytocin, 24 IU vasopressin). We note that the PSEs under the exposure to the carrier solution alone were comparable across drug treatments and between heterosexual and homosexual men (drug treatment: F_{2, 138} = 0.092, p = 0.91; sexual orientation: F_{1, 138} = 0.011, p = 0.92; interaction: F_{2, 138} = 0.22, p = 0.80).

2) Subsection “Oxytocin, vasopressin and subconscious chemosensory decoding of sex in heterosexual and homosexual men” “it follows that individuals with different endogenous oxytocin level would differ in their susceptibility to such chemosignals as well as to the effect of exogenous oxytocin.” Why does this follow?

We have changed the sentence to “If oxytocin plays a role in the processing of estratetraenol (in heterosexual males) and androstadienone (in homosexual males), given the heterogenous effects of oxytocin on individuals with different levels of social proficiency (Bartzvet al., 2011), specifically as assessed by the AQ (Bartz et al., 2019; Bartz et al., 2010), it follows that individuals with different AQ scores could differ in their susceptibility to such chemosignals as well as to the effect of exogenous oxytocin” in subsection “Oxytocin, vasopressin and subconscious chemosensory decoding of sex in heterosexual and homosexual men” of the revised manuscript. Please also refer to our response below to a related point raised by reviewer 3 (reviewer 3’s point 2).

Reviewer #2:

Overall, this is a very impressive study that logically builds across a set of 5 experiments a convincing case to uncover the neuroendocrine mechanisms involved in the chemosensory communication of sex and show that oxytocin and vasopressin have different roles to play in this process.

We appreciate the reviewer’s positive remarks.

Some trivial observations:

1) I would have appreciated information about which the criteria they used -if any – to decide about sample sizes.

In essence, for each drug condition, we examined whether there was a significant effect of androstadienone or estratetraenol in a subgroup of 24 participants. Sample sizes (n = 24 in each subgroup) were determined by G*Power to be adequate to detect a moderate effect of androstadienone or estratetraenol (d ≈ 0.6), at 80% power. The effect size was estimated based on an earlier study that employed almost identical stimuli and psychophysical testing procedures to those in the current study (Zhou et al., 2014). This is clarified in the Materials and methods section of the revised manuscript. See also our response below to a related point raised by reviewer 3 (reviewer 3’s point 1).

2) It would have been useful if the authors would have argued as to why they did not use heterosexual and lesbian women participants?

The effect of oxytocin in women could be affected by menstrual phase (fluctuations of gonadal steroids) and the use of hormonal contraceptives (Insel et al., 1993; Scheele et al., 2016). Oxytocin also causes uterine contraction, which could be particularly problematic for women during early pregnancy. It is thus pragmatically difficult to examine the effect of oxytocin in women. These are now clarified in the Materials and methods section of the revised manuscript.

Reviewer #3:

In this research, the authors investigated the biological substrate(s) of chemosensory decoding of femininity, specifically focusing on the oxytocin and, closely related, vasopressin systems based on their roles in reproductive and social behavior. Results showed that oxytocin, but not vasopressin, plays a causal role in chemosensory communication in humans. The paper was well-written and I particularly liked how they used an oxytocin agonist AND antagonist to support their claims, as well as vasopressin to speak to discriminant validity. This work makes an important contribution to our understanding of the biological mechanisms that support human social information processing.

We thank the reviewer for the positive assessment of our work and the constructive suggestions.

1) My first concern has to do with sample size, which seems on the small side. That said, I do appreciate how difficult it is to conduct these kinds of drug administration studies, and the fact that the authors ran multiple studies and reported consistent effects across studies, so I am torn. Perhaps the authors could provide additional information on statistical power. The authors do address statistical power (subsection “Participants”):

"…Sample sizes (n = 24 in each subgroup) were determined by G*Power to be adequate to detect a moderate effect of androstadienone or estratetraenol (d ≈ 0.6), at 80% power. The effect size was estimated based on an earlier study that employed almost identical stimuli and psychophysical testing procedures to those in the current study (Zhou et al., 2014)."

However, it seems that they are only reporting the power to detect the effect of androstadienone and estratetraenol on chemosensory communication, NOT the moderation by drug, or the moderation by individual differences (AQ, sexual orientation). Can the authors please speak to these issues? Especially for Study 3 where there appear to be 10 conditions (12 IU OT, 24 IU OT, 12 IU AVP, 24 IU AVP, PL x estratetraenol vs. carrier).

As we were unaware of any previous study that had examined the modulation of chemosensory decoding of sex by oxytocin/vasopressin or individual differences, it was not possible to estimate a priori the effect sizes of their interactions. Our study was therefore powered to detect the effect of androstadienone or estratetraenol in each subgroup of participants (n = 24) under different drug treatments. Put differently, for each drug condition, we examined whether there was a significant effect of androstadienone or estratetraenol in a subgroup of 24 participants. This is now clarified in the Materials and methods section of the revised manuscript.

In Experiments 1 and 2, drug treatment served as a between-subjects factor, our analyses of the interactions among olfactory condition, drug treatment, and sexual orientation (Figure 2 and Figure 3) were based on the data from a total of 144 participants and 432 testing sessions. In Experiments 3-5, drug treatment served as a within-subjects factor, our analyses of the interactions among olfactory condition, drug treatment, and social proficiency (Figure 4 and Figure 5) were based on the data from a total of 72 participants and 624 testing sessions. We would like to note that our sample sizes are comparable to, if not larger than, those in some earlier studies that addressed the interplays between drug and individual differences (e.g., (Bartz et al., 2019; Bartz et al., 2010; Feeser et al., 2015; Scheele et al., 2014; Spengler et al., 2017)).

2) My second concern has to do with the author's assertion that less socially proficient individuals have lower levels of endogenous and the reason why oxytocin should be helpful to them is BECAUSE they have lower levels of oxytocin. For example, as the authors write:

Abstract: "…and contingent upon the recipients' social proficiency – a partial manifestation of their endogenous oxytocin level."

Discussion section: "…such that the dose effect of exogenous oxytocin depended on the recipient's social proficiency, which in turn partially reflected his endogenous oxytocin level (Parker et al., 2014)."

Discussion section: "Moreover, they provide strong behavioral evidence for a non-monotonic effect of intranasal oxytocin that interacts with the recipient's social proficiency or endogenous oxytocin level."

This last statement is particularly problematic given that the authors did not actually measure participants endogenous OT, but the statement makes it seem like they did.

While there is evidence linking endogenous OT levels with social proficiency, to my knowledge, no one has demonstrated that exogenous OT selectively benefits less socially proficient individuals because it alters endogenous OT levels. Given this, the authors might want to re-think their rationale for Experiments 3 and 4. Actually, there is plenty of empirical evidence for the selectively beneficial effects of oxytocin for individuals who are less socially proficient (e.g., Luminet et al., 2011; Radke and de Bruijn; 2015; Feeser et al., 2015), specifically as assessed by the AQ (Bartz et al., 2010; 2019)-I think citing that research is sufficient justification to look at AQ as a moderator in Experiments 3 and 4. Of course, the authors can mention the endogenous OT-ASD findings; I just wouldn't make explicit claims about mechanism as it seems like they are (unnecessarily) going out on a limb.

We fully agree and would like to thank the reviewer for directing us to these studies. We have made changes to the text in the Abstract and in the Discussion section of the revised manuscript accordingly to be more stringent with our expressions. Some of the mentioned studies are also cited in the revised manuscript.

Reviewer #4:

This study aims to test whether intranasal administration of the neuropeptides Oxytocin (OXT) and Vasopressin (AVP) can mediate the effect of two putative human chemosignals – androsta-4,16,-dien-33-one (AND) and estra-1,3,5(10),16-tetraen-3-ol (EST). To test this hypothesis, the authors used a behavioral task named PLW, in which participants determine the gender (female or male) of a dot-figure. In the manuscript, they detailed 5 experiments which provide evidence for a link between OXT and the effect of EST.

In general, the manuscript is novel, and provides an important contribution, yet there are a few points which should be addressed:

We thank the reviewer for the overall positive evaluation of our work and the detailed comments and suggestions.

Essential revisions:

1) The manuscript is a bit hard to follow. Though it is divided to sets of experiments per test focus, it is hard to follow the line of thought which lead to each experiment, what was the hypothesis raised and the way they wanted to test it.

The rationales and hypotheses for Experiments 1-2, 3, 4, and 5 are now respectively stated in the Results section of the revised manuscript. Please also refer to our responses above to two related points raised by reviewer 3 (points 2 and 8).

2) In the main experiments (1-5) the statistical method used is paired t-test. Data of this sort should be statistically tested using one statistical test (e.g. ANOVA instead of multiple t-tests), with factors of odor and participant.

The results of ANOVAs are now reported in subsection “Oxytocin, vasopressin and subconscious chemosensory decoding of sex in heterosexual and homosexual men” (Experiments 1 and 2) and in subsection “Dose-dependent modulation of chemosensory communication of sex by oxytocin but not vasopressin in high and low AQ individuals” (Experiments 3-5) of the revised manuscript.

3) Were there corrections for multiple comparisons applied? Please mention this in the text.

As mentioned in our response to reviewer 3’s point 1, our study was designed to examine whether there was a significant effect of androstadienone or estratetraenol in each subgroup of participants under each drug condition. The comparisons between androstadienone/estratetraenol and carrier control were planned (see also our response to this reviewer’s point 1) and hence did not require corrections for multiple comparisons. Moreover, the reported effects in each experiment would remain statistically significant if we were to apply Bonferroni correction.

4) It is not clear whether all experiments were double-blind. Were both-experimenter and subject not aware of the drugs and odors administrated in all experiments? If not, this should be clearly acknowledged.

As mentioned in the Materials and methods section of the original manuscript, both the odor solutions and the drug solutions were coded by an individual not involved in the study. All experiments were double-blind. In the “no drug” condition in Experiments 3 and 5, where both the experimenter and the participants knew no drug was administered, they were still blind to the identities of the olfactory stimuli.

5) According to what did the authors choose a cut-off of 25 score for AQ? There is no supporting for this cut-off in the reference they provided, and in a previous study by the same group (eLife 2018) they chose a cut-off of 20. This should be clarified.

At the group level, 24 IU oxytocin exempted the heterosexual men in Experiment 1 from the influence of estratetraenol and the homosexual men in Experiment 2 from the influence of androstadienone. Considering that oxytocin exerts a more pronounced and typically positive effect on socially less proficient individuals as measured by the AQ (Bartz et al., 2019; Bartz et al., 2010; Bartz et al., 2011), we adopted a relatively stringent criterion for high AQ individuals — AQ scores ≥ 25, i.e. 1 SD or more above the reported mean for males (Baron-Cohen et al., 2001) — in our supplementary analysis of the data from Experiments 1 and 2 and in Experiments 3 and 4 in hopes to better capture the effects of social proficiency and oxytocin in chemosensory communication of sex. This is now clarified in subsection “Oxytocin, vasopressin and subconscious chemosensory decoding of sex in heterosexual and homosexual men” of the revised manuscript.

We have also re-examined the data from Experiments 3-5 using 20 as the AQ cut-off and obtained similar patterns of results as those reported in the manuscript.

6) In regards to the olfactory stimuli, in the Materials and methods section it is mentioned "Participants were instructed to hold the jar with their non-dominant hand, position the nosepieces inside their nostrils and continuously inhale through the nose and exhale through their mouth throughout each block of the experiments". Did the authors verify using breathing measurements (e.g. spirometer) whether the participants actually inhaled from their nose and exhaled from their mouth? If not, this could add variability to the results and should be mentioned.

We verified that the participants inhaled through the nose and exhaled through the mouth by monitoring the surface of the liquid in the odor jar via a camera. As mentioned in our response to reviewer 3’s point 7, the order of olfactory conditions was counterbalanced across participants and that of drug conditions randomized across participants. The procedure controlled the effects of nuisance variables.

7) In the Results section it is mentioned as "The three olfactory stimuli were perceptually indiscriminable as assessed in a separate panel of 48 male participants (mean triangular discrimination accuracy = 0.33 vs. chance = 0.33, t47 = 0.02, p > 0.9)". However, in the methods section, it is mentioned "The effectiveness of the clove oil carrier solution as a masker for the odors of androstadienone and estratetraenol was verified beforehand in an independent group of 48 healthy male nonsmokers (22.71 {plus minus} 2.37 yrs; triangular test, mean accuracy {plus minus} SD = 0.33 {plus minus} 0.15 vs. chance = 0.33, p > 0.99)….…Since the psychological effects of androstadienone is unrelated to one's sensitivity to its odor (Lundström et al., 2003) and estratetraenol is generally regarded as odorless (Lundstrom, Hummel and Olsson, 2003), we did not assess individuals' sensitivities to androstadienone or estratetraenol alone". It is not clear whether the authors did actually test, or how did they test, if participants could discriminate or perceive the smell of these odorants, and was this tested separately per each odor stimulus? There is evidence that AND has an odor [Keller et al., 2007] and can be perceived as unpleasant to participants. If there were perceptual differences this may affect the interpretation of the results.

We verified the effectiveness of clove oil carrier solution (1% v/v clove oil propylene glycol solution) as a masker for the odors of androstadienone (500 μM) and estratetraenol (500 μM) in an independent group of 48 healthy male nonsmokers. Specifically, we tested whether they could discriminate among androstadienone in clove oil carrier solution, estratetraenol in clove oil carrier solution, and clove oil carrier solution alone, using a standard triangular test (6 trials per participant). Detailed testing procedures have been described elsewhere (Ye et al., 2019; Zhou et al., 2014). Briefly, in each trial, blindfolded participants were presented with three smells, two identical (comparison) and the other one different (target), and reported which one was the odd one out. Each smell served as the comparison in 1/3 of the trials and the target in 1/3 of the trials. Results showed that the three olfactory stimuli were perceptually indiscriminable, mean accuracy = 0.33 vs. chance = 0.33, p > 0.9, consistent with earlier reports (Ye et al., 2019; Zhou et al., 2014). Put differently, the odor of androstadienone or estratetraenol (if any) was effectively masked by clove oil. We did not assess individuals’ thresholds to androstadienone or estratetraenol without an odor mask.

These are now clarified in the Materials and methods section of the revised manuscript.

8) Throughout the manuscript the authors refer to "social proficiency" and participants' basal OXT levels, this without measuring their basal OXT levels. This issue has further implications due to their claim that OXT affects the response to EST, however, this claim becomes circular when basal levels are not measured nor taken into account when reaching conclusions of the effect of OXT.

This point is related to reviewer 3’s point 2. Please refer to our response to that point above. We would also like to note that several studies have found a link between social proficiency and endogenous oxytocin level (Koven and Max, 2014; Lancaster et al., 2015; Parker et al., 2014).

9) Please rephrase the sentence where you suggest that using homosexual men is a replacement for women. Discussion section

We have removed this sentence in the revised manuscript.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Reviewer #1:

This study remains quite complex and unfortunately errors in referring to the figures do not help the reader to decode this revision.

I would like the errors to be fixed so that a clear read can be had. Critical errors are in the referring to Figure 2, which while only one of 5 figures, is foundational to the story. Confusing the reader during the establishment of the basic paradigm is problematic.

1) First, it would appear that the men and women labels at the bottom of Figure 2F are reversed. If they are not, then I am confused. I looked at this easily 5-10 times and with those labels the figure does not compute.

The x-axis in each of Figure 2A-D and F-I, Figure 4A-E and G-K, and Figure 4—figure supplement 1A-C represents PLW’s physical gender. As mentioned in the Introduction and the Materials and methods section of the original main text, PLW’s gender was indexed by a normalized Z score on an axis that differentiated between actual male and female walkers in terms of a linear classifier, and ranged from -0.45 (female-like) to 0.45 (male-like). Overall, the more masculine a PLW was, the more likely it was judged as “male” (larger value on the y-axis that represents the proportion of “male” responses). We have clearly labeled the x-axis in these figures in the revised manuscript.

We realize that the PLW’s physical gender and the male and female symbols, appearing in the original Figure 2E and J, Figure 3, Figure 3—figure supplement 1, Figure 4F and L, Figure 4—figure supplement 1D, and Figure 5A, are confusing. The male and female symbols were meant to represent a masculine bias (a bias towards perceiving the PLWs as more masculine, i.e., a negative PSE shift) and a feminine bias (i.e., a positive PSE shift) in participants’ gender perception, respectively. We have since removed the male and female symbols from all figures. In addition, we have clarified and illustrated the relationships among PLW’s physical gender, proportion of “male” responses, PSE, and PSE shift in the Results section of the revised main text and in the revised Figure 1B and C

2) Results section:

Figure 2B should be 2F? Should be Figure 2B,G,E,J; Should be Figure 2D,I not G,I

We apologize for the errors. We tried different ways to organize and present the results of Experiments 1 and 2 and in this process mistakenly referred to the wrong subfigures in the text. Per reviewer 3’s suggestions (reviewer 3’s points 1d and 2), we have reorganized this part of the Results section and have made sure that the correct figures are referred to in the text.

3 Why focus on VAS instead of OXT?

As mentioned in the original manuscript, we set out to explore if high AQ and low AQ participants in Experiments 1 and 2 differed in their susceptibility to the chemosignals. Since no significant effect of androstadienone or estratetraenol was observed under the 24 IU oxytocin condition or 60 µg atosiban condition, we focused on the 24 IU vasopressin condition where the effects of androstadienone and estratetraenol were comparable to those obtained earlier without drug treatment (Zhou et al., 2014). This is now clarified in the Results section of the revised main text.

Reviewer #3:

Overall I think the authors did a fine job responding to the reviewers' comments, questions and suggestions. I just have a few remaining items:

1) Clarity about participants and experiments:

I raised this issue before-and I appreciate the changes the authors made-but I still feel they need to be more transparent about participants/studies/experiments and sessions:

1a) "Psychophysical data collected from 216 heterosexual and homosexual men over 1,056 sessions consistently showed that such chemosensory communications of sex were blocked by a competitive antagonist of both oxytocin and vasopressin receptors called atosiban, administered nasally"

This still isn't entirely transparent as the sample size collapses across 5 experiments, thus giving the impression of greater statistical power than is warranted (per experiment/test). Please present sample size for each individual experiment here, or at least indicate that the 216 ps comprise 5 experiments.

The Abstract has a word limit of 150 words. We have stated in the revised Abstract that psychophysical data were collected from 216 heterosexual and homosexual men across 5 experiments.

1b) "Systematic comparisons of these psychometric curves across 5 experiments totaling 1,056 sessions enabled…"

As per above, please state number of ps here for transparency

We have included the sample size of each experiment in the Introduction of the revised main text.

1c) Also, I find the term “sessions” to be confusing. Maybe this is a convention specific to their research area, but, to me, “sessions” suggests testing sessions, whereas it seems like the authors are referring to trials on the task?

“Sessions” does mean testing sessions. We have made it explicit in the revised manuscript. Specifically, as mentioned in the original manuscript, each participant was tested on multiple days (3 days in Experiments 1 and 2, 10 days in Experiments 3 and 5, 6 days in Experiment 4) and completed 1 testing session per day. Each session consisted of 12 blocks (5 baseline blocks and 7 experimental blocks) and each block consisted of 70 trials. The results reported in our manuscript are based on data collected from a total of 1,056 testing sessions or 1,056 person-days, which amount to 887,040 trials.

1d) I think reviewer 4 made a similar comment, but I also found the overall design to be unclear:

1d-i) Are Experiments 1 and 2 identical expect that Experiment 1 has heterosexual men and Experiment 2 has homosexual men as participants? If so, why are they called two different experiments since the results are discussed side by side and, in fact, the authors combine the data anyway with the omnibus ANOVA?

Experiment 2 was conducted after we analyzed the results of Experiment 1, which presented a complex picture. Antagonizing oxytocin and vasopressin receptors with atosiban abolished the known effect of estratetraenol on heterosexual males (i.e. biasing them towards perceiving the PLWs as more feminine), yet 24 IU oxytocin appeared to produce the same effect. On the flip side, the administration of 24 IU vasopressin did not significantly alter the processing of estratetraenol in heterosexual males –– they remained biased towards perceiving the PLWs as more feminine under the exposure of estratetraenol, to the same extent as when no drug was administered. We wondered if this pattern of drug influences would hold for the chemosensory decoding of masculine information carried by androstadienone. To this end, we turned to homosexual males in Experiment 2, who had been shown to be subconsciously biased by androstadienone, but not estratetraenol, in making gender judgments (Zhou et al., 2014). Experiment 2 was identical to Experiment 1 except for the participants’ sexual orientation. These are now clarified I the Results section of the revised main text.

We presented the results of Experiments 1 and 2 side by side in the original manuscript in hopes to make the section more concise and also to make it easier to compare the results of heterosexual and homosexual men. We have realized that this only made the rationales and results of the experiments hard to unpack. We have rewritten this part of the Results section to clarify the rationale and results of each experiment. With regard to the use of omnibus ANOVAs on the pooled data of two or more experiments, please refer to our response below to this reviewer’s point 1d-ii.

1d-ii) More generally, is this one “Study” with multiple “Experiments” embedded within the overall Study? The Participants section is written this way (i.e., "A total of 216 young male adults participated in the main study…") but the way the Experiments are presented in the main text I had the impression that they were separate studies, with different participants and different designs, not one overall study, with experiment referring to testing different effects.

The distinction between “study” and “experiment” can be subtle. We chose to present our work as 1 study with 5 experiments because, as mentioned in the original manuscript, we had one single goal that was to examine the roles of oxytocin and vasopressin in chemosensory communications of sex through androstadienone and estratetraenol. Except for Experiment 1, the design and interpretation of each experiment was guided by the findings of the previous experiment(s). The combined results of all 5 experiments, encapsulated by the omnibus ANOVAs on the pooled data of Experiments 1 and 2 and of Experiments 3-5, jointly demonstrated that the decoding of chemosensory sexual cues is modulated by oxytocin instead of vasopressin in a dose-dependent manner.

2) Predictions about oxytocin:

As detailed below, I was a bit surprised by some of the findings and I think the authors could do a better job walking readers through. Note: I am in no way suggesting that the authors should state hypotheses that they did not have a priori, but I think they need to take more care laying the groundwork for readers. See my suggestions below:

2a) "…could be affected in either direction by the administration of oxytocin or vasopressin."

I was surprised that the authors did not have specific predictions for OT, given that they did have predictions about atosiban. Since OT is an agonist, wouldn't one expect the opposite effect of the antagonist atosiban? Again, I am not advocating that the authors make predictions post hoc, but I think they need to clarify *why* they did not have predictions for OT since it's not obvious (i.e., given they had predictions for atosiban).

We had no specific prediction for oxytocin or vasopressin as we did not know whether one or both of them play a role in the processing of chemosensory sexual cues. It was also difficult to predict the directions of their effects (if any), as both positive and negative effects of oxytocin and vasopressin have been reported in the literature, depending on dose, context, and personal characteristics (Bartz et al., 2011; Carter, 2014; Donaldson and Young, 2008). We predicted that the administration of atosiban would block the effects of the chemosignals if such effects were regulated by the oxytocin/vasopressin system, because atosiban is a competitive antagonist of both oxytocin and vasopressin receptors. These are now clarified in the Results section of the revised main text.

2b) "Smelling estratetraenol and androstadienone, relative to the carrier solution alone, failed to respectively influence the gender perception criteria (indexed by the PSEs) of the atosiban-treated heterosexual and homosexual men (t23s = 1.37 and 0.88, ps = 0.18 and 0.39, respectively; Figure 2C, D, I, J). This appeared to be the case for those treated with 24 IU oxytocin as well (t23s = 0.17 and -1.29, ps = 0.87 and 0.21, respectively; Figure 2E, F, I, J)."

I was also surprised that oxytocin and atosiban produced the same effect, one being an agonist and the other being an antagonist. I appreciate that the authors address this in the discussion, and I basically find their rationale to be compelling, but my sense is that they gloss over this surprising finding here and elsewhere (see below). I think they could do a better job of guiding the reader along by NOT glossing over this point. E.g., something like "Interestingly, this appeared to be the case for those treated with 24 IU oxytocin as well (t23s = 0.17 and -1.29, ps = 0.87 and 0.21, respectively; Figure 2E, F, I, J)."

Suggestion well taken. We have done so in the Results section of the revised main text. See also our response to this reviewer’s point 1d-i.

2c) I have similar issues with the following statements:

"…were subserved by similar neuroendocrine mechanisms that were disrupted by intranasal atosiban, the competitive antagonist of both oxytocin and vasopressin receptors, and unaffected by intranasal vasopressin."

Didn't oxytocin exert the same effect as atosiban? But oxytocin is not mentioned.

"The question remained as to why the administration of 24 IU oxytocin exempted the participants from the influences of the chemosignals, and we explored it in more detail."

I would suggests the following modification for transparency: "the question remained as to why the administration of 24IU oxytocin exerted the same effect as atosiban…"

Rather than glossing over this surprising finding, I think it would be more effective to acknowledge it head on.

We have revised these sentences accordingly in subsection “Oxytocin, vasopressin and subconscious chemosensory decoding of sex in heterosexual and homosexual men” of the revised main text. The sentences now read as follows:

“… were subserved by similar neuroendocrine mechanisms that were disrupted by intranasal atosiban –– the competitive antagonist of both oxytocin and vasopressin receptors, as well as by 24 IU oxytocin, and were unaffected by 24 IU vasopressin. Since vasopressin did not seem to play a role (participants’ response patterns to the chemosignals under 24 IU vasopressin were comparable to those previously obtained without drug treatment), by deduction, such mechanisms involved oxytocin. The question remained as to why the administration of 24 IU oxytocin, like atosiban, exempted the participants from the influences of the chemosignals, and we explored it in more detail.”

Please also refer to our responses to this reviewer’s points 1d-i, 2a and 2b.

Author details

1. Kepu Chen

   State Key Laboratory of Brain and Cognitive Science, CAS Center for Excellence in Brain Science and Intelligence Technology, Institute of Psychology, Chinese Academy of Sciences, Beijing, China

   Contribution

   Data curation, Formal analysis, Validation, Investigation, Writing - original draft

   Contributed equally with

   Yuting Ye

   Competing interests

   No competing interests declared

2. Yuting Ye

   Department of Psychology, University of Chinese Academy of Sciences, Beijing, China

   Contribution

   Data curation, Formal analysis, Validation, Investigation, Writing - review and editing

   Contributed equally with

   Kepu Chen

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-9068-8045

3. Nikolaus F Troje

   Centre for Vision Research, York University, Toronto, Canada

   Contribution

   Software, Writing - review and editing

   Competing interests

   No competing interests declared

4. Wen Zhou

   1. State Key Laboratory of Brain and Cognitive Science, CAS Center for Excellence in Brain Science and Intelligence Technology, Institute of Psychology, Chinese Academy of Sciences, Beijing, China
   2. Department of Psychology, University of Chinese Academy of Sciences, Beijing, China
   3. Chinese Institute for Brain Research, Beijing, China

   Contribution

   Conceptualization, Supervision, Funding acquisition, Methodology, Writing - original draft, Writing - review and editing

   For correspondence

   zhouw@psych.ac.cn

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-6730-2116

• 2,163

  Page views

• 158

  Downloads

• 3

  Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.