Exposure to Female Fertility Pheromones Influences Men’s Drinking

Abstract

Research shows that humans consciously use alcohol to encourage sexual activity. The current study investigated whether decision-making about alcohol use and sex can be cued outside of awareness by recently revealed sexual signaling mechanisms. Specifically, we examined if males exposed without their knowledge to pheromones emitted by fertile females would increase their alcohol consumption, presumably via neurobehavioral information pathways that link alcohol to sex and mating. We found that men who smelled a T-shirt worn by a fertile female drank significantly more (non-alcoholic) beer, and exhibited significantly greater approach behavior toward female cues, than those who smelled a T-shirt worn by a non-fertile female. These findings reveal previously unknown influences on human alcohol consumption, augment the research base for pheromone cuing of sexual behavior in humans, and raise the possibility that other, as yet unknown, pathways of behavioral influence may be operating hidden from view.

Alcohol use and sexual activity have been linked throughout human history, from the Old Testament (e.g., Song Sol. 1:8) to recent popular culture (e.g., American Pie [Weitz et al., 1999]). That individual humans are aware of this linkage has been confirmed by scientific research: both males and females expect alcohol use to encourage and facilitate sex (e.g., Wilson, 1981), and individuals with more positive alcohol-sex expectancies are more likely to have sex after drinking (Patrick & Maggs, 2009). As yet unaccounted for in this research domain are recent findings that show human sexual behavior to be influenced by biological signals of which individuals seem completely unaware. In this study, we examine whether these recently uncovered sexual signals may serve as a never-before-considered pathway to alcohol consumption.

Female Fertility Cues and Men’s Sexual Behavior

Any single act of human sexual intercourse leads to pregnancy only 3% of the time (Tietze, 1960), largely because fertilization is most likely from only a few days before ovulation until the day of ovulation (Wilcox, Weinberg, & Baird, 1995). Given the significance of the timing of fertility, evolutionary theories suggest that humans have adopted strategies to promote mating behavior during this narrow window of fertility (Gangestad, Thornhill, & Garver-Apgar, 2005). In primates, some strategies of this kind are obvious; when female primates are most receptive to sexual approaches from males, they overtly display fertility signals—e.g., large hindquarter sexual skins or sexual swellings of female chimpanzees, bonobos, and other primates (Dixson, 1983; Nuun, 1999).

Although human females also have fertility displays that influence males, these displays are far subtler than those of primates. In fact, human fertility cues are so subtle that until recently, scientists thought that human females exhibited no fertility signals. (Very recent evolutionary accounts theorize that minimization of sexual signals in human females decreased aggressive competition among males, thereby permitting more cooperation and more complex social structures [Edgar, 2014]). Studies over the past few decades have shown that women do in fact give off several subtle cues of ovulation, such as changes in scent, skin color, voice pitch, body symmetry in paired soft tissue traits (e.g., breast shape), and waist-to-hip ratio (Kirchengast & Gartner, 2002; Pipitone & Gallup, 2008; Scutt & Manning, 1996; Singh & Bronstad, 2001; Thornhill et al., 2003; Van den Berghe & Frost, 1986). Furthermore, a number of studies have indicated that these female fertility cues influence men’s behavior. For example, males tipped significantly more money to fertile female lap dancers (Miller, Tybur, & Jordan, 2007), and acted more jealously and possessively when their significant other was fertile (i.e. “mate guarding”; Haselton & Gangestad, 2006). Even beyond the immediate social/sexual realm, fertility signals influenced male behavior; decisions by male participants became riskier in the presence of a fertile female confederate (Miller & Maner, 2011). Although the males in these studies may have been conscious, at some level, of their overt behavior, it was highly unlikely they were aware that these subtle cues of female fertility influenced their behaviors.¹

Among the human fertility cues that influence men’s sexual behavior, scent is well-supported by research (Ziegler, Schultz-Darken, Scott, Snowdon, & Ferris, 2005). For example, not only did men rate odors collected from women during the ovulatory period as more pleasant and sexier than odors from women not in the ovulatory period (Singh & Bronstad, 2001; Thornhill et al., 2003), but as shown by Miller and Maner (2010), these ovulatory cues could influence an important substrate of male mating behavior, men’s testosterone levels (Roney, Lukaszewski, & Simmons, 2007). Specifically, Miller and Maner (2010) showed that men exposed to control and nonovulatory odors significantly decreased in testosterone levels when compared to their own baseline testosterone level, whereas ovulatory odor exposure rendered this decrease non-significant, leading the authors to conclude that male endocrinological processes are sensitive to chemosensory cues of female ovulation. In a follow-up study, Miller and Maner (2011) also showed ovulatory scents influence cognitive processes; male participants exposed to scents of fertile women had increased accessibility to sexual concepts and judged women’s sexual arousal to be higher when compared to men exposed to scents of women who were not fertile.

Alcohol and Men’s Sexual Behavior

Male sexual activity and alcohol use often occur concomitantly; it is difficult to say if alcohol use leads to sex, or if the anticipation of a sexual opportunity leads to alcohol use (most likely both, in a reciprocal fashion). Much of this reciprocal influence may be as much expectancy-based as it is pharmacological in nature; at high doses, the effects of alcohol on sexual response may be primarily pharmacological, but at lower doses, the enhancing and disinhibiting effects of alcohol on sexual response are heavily influenced by expectancies (Crowe & George, 1989). In tightly controlled laboratory studies, male participants manifested significantly greater subjective and objective sexual arousal (measured by penile tumescence) when they were led to believe their drinks contained alcohol, regardless of the actual content of the drinks (Lansky & Wilson, 1981). Because men expected alcohol to decrease interpersonal inhibitions and to cause drinking women to be more open to sexual advances, experiments showed that men were more likely to approach intoxicated women (George, Cue, Lopez, Crowe, & Norris, 1995; Wilson, Perold, & Abrams, 1981).

The Interplay Between Female Fertility Cues and Men’s Alcohol Use: The Current Study

Yet to be investigated as triggers for alcohol use are those “hidden” fertility-pheromone cues only recently uncovered. If these cues were found to influence men’s drinking, a novel and important pathway to drinking would be exposed. This study was designed to begin the investigation of this relationship. Our hypothesis that pheromone-cued males would demonstrate increased alcohol consumption was derived from expectancy theories in a variety of behavioral domains (e.g., Huron, 2006), and in particular, recent elaborations of alcohol expectancy theory (Goldman & Reich, 2013; Reich & Goldman, in press). This group of theories suggests that detection of sexual pheromones by men should activate associational networks that store information about the anticipated contingency between alcohol use and sexual expression—i.e. that in the presence of sexual/fertility cues, alcohol use would increase the likelihood of sex—thereby increasing alcohol consumption. To strengthen the inference that increases in drinking were part of the fertility pheromone-sexual expression pathway, we also assessed men’s approach behavior toward stimuli suggesting the presence of a woman.

Method

Study Design

Performing a study of this kind presented unique challenges, largely centered on optimizing the effectiveness of pheromone samples from women. The women providing samples could not be using hormone-based contraception, had to have regular monthly cycles, and had to remain odor-free so that their pheromone samples (provided in the form of T-shirts they wore to sleep) were free of competing olfactory cues (i.e. women could not use scented deodorants, perfumes, or other cosmetics on days they were wearing the T-shirt, women had to wash their sheets with unscented detergent, T-shirts had to be new, and women could not sleep with sexual partners or pets while wearing the T-shirt to avoid odor contamination). Finally, to be effective as stimuli, the T-shirts had to be delivered to the laboratory by the female participants and used within one week of being turned in; the availability of the T-shirt then had to overlap with the availability of male volunteers for the study. Due to this substantial burden on the female participants, and the demands of coordinating fresh T-shirts with male schedules, we used a simple two-group design to minimize the number of females necessary to provide pheromone samples in this initial “proof of concept.” Despite this accommodation to scheduling, we optimized power by using more male participants (and more women to provide T-shirts) than any other study of this kind to date, and we also confirmed fertility with biological assays and not simply with self-reported monthly calendars (as had previous studies).

Participants

A priori sample size estimates indicated a minimum of 102 participants to detect a medium effect size (Cohen’s d = .50). To this end, we recruited 107 undergraduate men, of which 94 between the ages of 21 and 37 years met methodological requirements and were included in the final analyses (M = 23.56, SD = 3.48). This group included males within the age range of strong sexual interest who could consume alcohol legally. These men, who volunteered for course credit, identified themselves as White/Caucasian (53.2%), Hispanic/Latino (16.0%), Asian/Pacific Islander (13.8%), Black/African American (10.6%) or other (6.3%). Excluded were men under 21 years of age, non-beer drinkers, current smokers (smoking can lead to smell impairment; Vennemann, Hummel, & Berger, 2008), and those not exclusively heterosexual. The sample was limited to males over age 21 because the primary dependent variable included an invitation to consume what was presented to them as an about-to-be-marketed alcoholic beverage (beer). College undergraduates are among the heaviest drinking segment of the U.S. population (O’Malley & Johnston, 2002), and certainly use alcohol in circumstances in which the alcohol/sexual contingency is in operation (Paul, McManus, & Hayes, 2000).

Using a limited snowball sampling approach, a total of 19 women were recruited to provide T-shirts for the study. Of those, 17 women provided a total of 38 T-shirts. Each woman received $10 for each T-shirt provided. Due to scheduling issues, only 30 T-shirts (from 14 women) were used in the study; these women ranged in age from 20 to 29 years old (M = 24.14; SD = 2.54), were 50% White/Caucasian, 28.6% Asian/Pacific Islander, and 21.4% Hispanic/Latino.

Measures

Chemical Sensitivity Scale (CSS; Nordin, Millqvist, Lowhagen, & Bende, 2003)

The critical outcome measure in this study was dependent on detection of the pheromone odor in the T-shirts. Therefore, a self-report measure used in previous studies of this type (Miller & Maner, 2011) assessed possible baseline differences in smell sensitivity between prime conditions. The CSS includes 21 items with a 6-point Likert scale rating agreement to statements related to the participant’s sense of smell. Total scores were calculated with higher scores indicating greater sensitivity to, and awareness of, smells in the environment.

Visual Reaction Time (VRT) to Female Faces

We included a version of the VRT task (see Abel, Huffman, Warberg, & Holland, 1998) used previously as a dependent measure of attraction to females, but modified in our version to be part of the independent variable accommodating a different purpose. In our version, we included pictures of female faces with minimal sexual provocativeness, rather than pictures showing full female bodies set in sexually provocative circumstances (e.g., women in swimsuits, people embracing) as had been used before. Because males, in the real world, are never exposed to fertility pheromones in the absence of a woman, our intent was to carefully balance the need to couple a female cue with the pheromone stimulus, as had been done in previous studies, against the concern that the female cue might itself be sexually provocative (i.e., contribute independent influence on subsequent tasks). (Results showed that VRT ratings did not differ in our two groups, indicating that any subsequent group differences on drinking were not a function of any differential influence of the female pictures.) Participants were introduced to the task with two sample items, and then completed ratings for 12 “test” items, which were the same pictures in the same order for all participants. Once the participant made a rating of each picture on the computer, they were able to advance to the next picture. Longer viewing periods were taken as evidence of increased attractiveness of the women in the picture, and Likert scale ratings of attractiveness—from 1 (“highly unattractive”) to 7 (“highly attractive)—also were obtained.

Taste-rating task (TRT)

To unobtrusively measure levels of alcohol consumption, we used a modified version of the procedure developed by Marlatt, Demming, and Reid (1973) for taste-rating. To avoid the possible pharmacological influence of actual alcohol on subsequent measures (approach task and expectancy assessment), non-alcoholic beers were used. Non-alcoholic beers have been employed effectively before for these purposes (e.g., Carter, McNair, Corbin, & Black, 1998), including in our laboratory (e.g., Roehrich & Goldman, 1995); we confirmed that most participants did not detect the true nature of these “beers” during debriefing. Participants were presented with two carafes of “beer,” marked A and B, and two glasses, marked A and B. They were told to pour the beer from the corresponding carafes into the glasses before consuming the “beers.” “Beers” were chilled, and identifying labels and containers were hidden. Water was provided so participants could rinse their mouths between tastings. Participants were told to take their time and to sample as much of each beverage as needed to arrive at a decision. Consistent with the rationale for the study presented to participants, a 7-point Likert scale was used to rate color, consistency, aroma, and taste of each “beer,” along with the likelihood of drinking the product at a party. To ensure that enough time would be taken by this assessment to permit ad-libitum sampling of drinking throughout the 10-minute period, open-ended questions about “product characteristics” also were included.

To reduce social constraints on drinking, the experimenter left the room after giving participants ratings sheets. Participants were unaware of the 10-minute time limit. Halfway through the 10-minute period, the experimenter returned to check on the participant’s progress, and at the end of the 10 minutes, the experimenter returned and had the participant continue to the next portion of the study. After each participant completed the study, the amount of “beer” left was measured, and the amount of “beer” consumed was calculated.

“Female” Approach task

To assess whether males exposed to female fertility pheromones (and given the sequencing of dependent measures, also exposed to female pictures and “beer” consumption) might display indications that they would be more likely to approach an actual woman, we used an established task (e.g., Smith & Bargh, 2008). The experimenter led participants into a room that contained five adjacent chairs in a row. A woman’s sweater hung on the back of the chair farthest to the left, and a woman’s purse and a clipboard were placed on the seat of the chair. The four chairs appearing to be “available” for the participant to sit on were coded from 1 to 4, with Seat 1 located next to the chair with the “phantom woman.” In previous work, the choice by the male to sit in lower number chairs was taken to reflect increased approach behavior.

Quantity Frequency Variability Index (QFVI; Cahalan, Cisin, & Crossley, 1969)

To rule out differences in the customary drinking practices of men in the two prime conditions as a possible (and confounding) source of group differences, customary drinking information was collected. The QFVI is a 13-item questionnaire that classifies drinking styles based on participant responses to questions about their typical amount of alcohol consumed per sitting, frequency of alcohol use, and the variability of alcohol consumption, including the modal amount, and the highest amount, of alcohol consumption.

Alcohol Expectancy Questionnaire (AEQ)

Because differences in the alcohol expectancies held by individual participants might be related to the responsiveness of their alcohol consumption to the sexual prime (i.e., the more they expected that alcohol was associated with sexual outcomes, the more they might consume), we assessed these expectancies using the AEQ. The AEQ contains six subscales; of particular interest to the current study were the Sexual Enhancement (7 items) and Social Assertiveness (10 items) subscales. To prevent contamination of the primary dependent variables by exposure to the items in this instrument (which could have primed consumption independent of the pheromone exposure), we administered the AEQ following other assessments. This presentation sequence precluded a test of whether differential expectancies activated prior to drinking may have influenced consumption levels (via either mediation or moderation), but allowed for evaluation of the consistency of assessed expectancy levels with the outcome of the experimental manipulation.

Odor Collection

Procedures for odor collection were informed by previous studies (see Gangestad & Thornhill, 1998; Miller & Maner, 2010), but were enhanced with the addition of commercially available biological ovulation tests (“First Response”) to detect fertility. Women who indicated regular menstrual cycles (approximately 26–34 days) and who were not on hormonal contraceptives wore a white cotton Hanes crew-neck T-shirt during the nights of the late follicular phase of their menstrual cycle (typically Days 13–15) or the nights of the luteal phase of their menstrual cycle (typically Days 20–22). Ovulation tests verified T-shirt assignments. Women were given verbal and written instructions to remain odor neutral by: a) showering with Dove unscented soap, Stonybrook Botanicals unscented shampoo and conditioner, b) refraining from using perfumes, deodorants and antiperspirants, c) avoiding consumption of foods with pungent odors (e.g., garlic, vinegar), d) abstaining from smoking cigarettes, drinking alcohol, and using drugs, and e) refraining from engaging in sexual activity and sleeping in the same bed with another human or animal. Women in the ovulation condition were instructed to start wearing the T-shirt once the ovulation test was positive. For the control condition, women were instructed to start wearing the T-shirt one week after the ovulation test was positive. Women also took an ovulation test each day they wore the T-shirt to verify their ovulation status. After each night they wore the T-shirt, women were instructed to place the T-shirt in a sealed Ziploc bag, and after three nights, the women returned the T-shirt and completed questionnaires on whether she had followed the instructions to remain odor neutral. Women were compensated with $10 for each T-shirt they provided. T-shirts were kept in a freezer when not in use, and all shirts were used within one week of being worn. One T-shirt was not used due to the participant not completing the protocol properly, and 7 T-shirts were not used due to not having any male participant sign up for the study within a week of the T-shirt being returned. After these exclusions, a total of 30 T-shirts from 14 different women were used.

Study Procedures

A female experimenter informed the male participants that the purpose of the study was to examine consumer preferences, which would be assessed using their senses of smell, sight, and taste. Female experimenters were all using hormonal contraceptives at the time of the experimenter so that the ovulation status of experimenters remained consistent across participants. Participants first completed a demographic questionnaire (age, ethnicity, gender, year in college, sexual orientation) and the CSS. After filling out the questionnaires, participants were primed using a T-shirt; this shirt represented either the follicular or ovulatory phases of the participating women’s menstrual cycles. Recall from the introduction that obtaining fresh T-shirts (an obviously scarce resource), and matching them in a timely fashion with the scheduling of male volunteers was challenging. Because T-shirts came from whichever woman could offer a fresh T-shirt that reflected one of the two usable windows during their ovulatory cycle, and male participants were assigned whichever condition happened to be available shortly after they signed up, no aspect of T-shirt assignment was in any way systematic (biased). In about half the cases, both kinds of T-shirts were available to a particular male participant; in these cases, males were assigned randomly to the ovulation prime or control prime condition. Experimenters were blind to the prime condition. Participants were told they would smell a T-shirt that had been washed with a laundry detergent and subsequently would be asked to rate the product. Further instructions were given to each participant to put his nose into the plastic bag containing the T-shirt and take three large inhalations. After smelling the T-shirt, participants rated the product on smell appeal, likelihood of use, and likelihood of purchase.²

About 2–5 minutes after being exposed to the prime, the VRT procedure assessed the participant’s attraction to female faces. Following the VRT task, the experimenter told the participant that they would taste and rate beverages that might include sparkling water, soda, coffee or beer. Participants began the TRT about 10–15 minutes after being exposed to the prime. After the 10-minute TRT, the participant was led into another room, which was set up for the “female” approach task. The experimenter told the participant they could sit in any available chair to finish the experiment by completing more questionnaires. After the participant sat down, they filled out the QFVI and the AEQ, and the experimenter surreptitiously noted the chair chosen by the participant. The total duration of the experimental session was typically between 25 and 35 minutes.

Participants were debriefed before being released. The debriefing session included manipulation checks for the entire procedure, including the credibility of the beverage as an authentic beer. Participants reported what they thought was the purpose of each task in the study, as well as the study as a whole. In this open-ended portion of the debriefing, no participant indicated suspicion about the nature of the “beers.” To push the limits of the credibility of the “beers,” they were asked if they thought the beers contained alcohol. Only four of 98 participants indicated that they did not think alcoholic beer was used and data from these participants was excluded.

Results

The possibility that groups differed in key ways prior to the experimental manipulation was examined first using t-tests (for continuous variables) and Chi-square tests (for categorical variables). No significant baseline differences were found between groups in age (ovulation: M = 23.32, SD = 3.30, control: M = 23.78, SD = 3.64, t(92) = −0.64, p > .250), ethnicity (ovulation: 50% Caucasian, control: 56% Caucasian, χ²(4) = 3.63, p > .250), year in college (ovulation: M = 3.70, SD = 0.86, control: M = 3.80, SD = 1.02, t(90) = −0.50, p > .250), or CSS total score (ovulation: M = 2.63, SD = 0.66, control: M = 2.66, SD = 2.66, SD = 0.73, t(92) = −0.27, p > .250). Groups also were roughly equivalent on the particularly important variable of drinker type; 70.4% (31 of 44) and 68% (34 of 50), respectively, were heavy/moderate drinkers in the ovulation and control groups, χ²(3) = 2.42, p > .250. The absence of differences between the two groups minimized concerns about third variable biases in further analyses.

VRT task

Although all participants spent longer looking at pictures of women who they thought were more attractive (r = .14, p < .001), no significant differences between groups were found either on attractiveness ratings (ovulation: M = 3.97, SD = 0.93, control: M = 3.95, SD = 0.74, t(92) = 0.12, p > .250), or seconds spent looking at the female faces, (control: M = 4.69, SD = 2.32, ovulation: M = 4.62, SD = 1.71, t(92) = 0.16, p > .250). Therefore, although the women’s faces may have psychologically combined with the pheromone scents to influence subsequent dependent variables (“beer” consumption and female approach behavior), differences in attractiveness were not responsible for any differences that might have appeared on these measures.

“Beer” consumption

Descriptive statistics for the primary dependent variable (“beer” consumption) can be found in Table 1. Consumption in both groups was relatively normally distributed (with a slight positive skew in the ovulation prime group), and consistent with our prediction that exposure to female fertility pheromones would lead to increased “beer” consumption in males, consumption minimums in the ovulation prime group began higher and rose to higher maximum levels than the consumption range for the control prime group. Figure 1 further shows that participants mean consumption (in milliliters) in the ovulation prime group was significantly higher than that of participants in the control prime group, t(92) = 2.17, p = .032 (for unequal variances), Cohen’s d = .44.

Table 1.

Descriptive Statistics for “Beer” Consumption

Condition	Mean	N	SD	Min.	Max.	Observed Range	Potential Range	Kurtosis	Skewness
Ovulation	216.05	44	175.55	29.00	710.00	681.00	710.00	1.33	1.45
Control	154.08	50	93.46	15.00	375.00	360.00	710.00	−0.41	.71
Total	183.09	94	140.77	15.00	710.00	695.00	710.00	3.36	1.75

Figure 1.

Means and 95% confidence intervals of amount of “beer” consumed by group.

Approach behavior to the “phantom female”

It can be seen from the distributions of seating choices in Table 2 that participants in the ovulation prime condition generally sat closer to the phantom female as compared to participants in the control prime condition who generally sat further from the phantom female. In fact, Figure 2 shows that the mean chair selected by participants in the ovulation prime group was significantly lower than the mean chair selected by participants in the control prime group (i.e., they sat themselves significantly closer to the phantom female), M (ovulation group)= 2.21, SD = 0.74; M (control group)= 2.54, SD = 0.79, t(91) = −2.07, p = .041, Cohen’s d = .43. This result was consistent with our prediction that exposure to female fertility pheromones would result in increased female approach behavior.

Table 2.

Frequency Table for Chair Position

		Prime Condition		Total
		Ovulation	Control
Chair Number	1.00	5	3	8
	2.00	27	23	50
	3.00	8	18	26
	4.00	3	6	9
Total		43	50	93

Figure 2.

Means and 95% confidence intervals of chair number by group.

Relationship Between Pheromone Effect and Expectancies

Given that the presumptive pathway for the influence of sex pheromones on alcohol consumption included activation of sexual (or possibly social) expectancies, we assessed the association between downstream expectancies and subsequent alcohol consumption separately for the ovulatory pheromone and control groups. (Recall that expectancy assessment upstream of drinking assessment might have independently influenced drinking.) A significant correlation between expectancies measured downstream of drinking and previous alcohol consumption in the pheromone group would be consistent with this presumptive pathway of influence. No evidence was found of the activation of social expectancies. For AEQ sexual expectancies, however, the downstream indication that exposure to sexual pheromones had influenced the relationship between sexual expectancy activation and drinking was evident: the correlation between sexual expectancy scores and drinking was essentially zero for the control group (r = −.082, p > .250), but was statistically significant for the ovulatory pheromone group (r = .30, p = .048).

Discussion

This study tested a new and never-before-considered pathway to human alcohol consumption that incorporates recently uncovered sexual signaling mechanisms operating outside awareness into the matrix of well-established connections between alcohol use and sex. Specifically, very recent research led us to anticipate that scents related to fertility in human females would activate sexual expression pathways in human males. It was further thought that these activated pathways would go on to network with acquired associational information (expectancies) about the utility of alcohol as a facilitator of mating/sexual experiences. Based on this conceptual model, we hypothesized that males exposed to scents (pheromones) from fertile (ovulating) females would increase both their alcohol use and their approach behavior toward females without awareness of the controlling contingencies.

The finding that men exposed to scents from fertile females drank more “beer” than men not so exposed supports the hypothesized pathway described above. The difference in “beer” consumption in the two groups, 61.97 mL, represents approximately 1/5 of a standard beer. Considering that this difference was manifested over only a 10-minute window in a study of beverage preferences carried out in a university laboratory (not a context usually associated with alcohol consumption), this finding is not insubstantial. (Manipulation checks showed that almost all participants believed that what they were drinking was actual beer; those few who did not were removed from analyses.)

Because the males exposed to ovulatory pheromones (and who drank more) also chose seats closer to the “phantom female” was consistent with the hypothesized mechanism of influence, the activation of sexual expression pathways. Given that this assessment of male approach behavior toward females came after the drinking assessment, it is plausible that both outcomes were influenced by the same underlying biopsychological processes. That is, this outcome suggests that both increased drinking and mating-related behavior were elicited by exposure to the ovulatory pheromones. It must be acknowledged, however, that given the sequencing of the dependent variables, it also was possible that differences in approach to the “phantom” female might have been influenced solely by the differences in the drinking experiences of the two groups. Such differences could not have issued from differential pharmacological effects, however, because only non-alcoholic beer had been consumed. Instead, differential psychological/expectancy effects associated with beverage consumption could have carried over to the males’ approach behavior. Findings from many studies in the expectancy domain suggest that “believing” that they had consumed alcohol might have activated sexual associations in the male drinkers and diminished social anxiety about approaching females.

Because activation of alcohol expectancies was a key part of the theoretical underpinning for this study, we of course were interested in probing whether expectancies in fact had been activated after pheromone exposure and before the opportunity for ad-libitum alcohol consumption (thereby serving in a meditational role). As noted earlier, however, to avoid the possibility that assessment of expectancies might independently stimulate (prime) drinking, we chose not to assess this theorized meditational pathway at the critical (mid-) point in the sequence of measures. Instead, we assessed expectancies at the end of the string of dependent variables to determine if expectancy patterns detected at that point at least were consistent with the theorized pathway. As reported in the results section, the pattern of correlations that might have been anticipated if sexual expectancies had been activated was, in fact, obtained. Drinking amounts were related to sexual expectancies in the group exposed to the ovulatory pheromones, but not in the controls. Given that the basic relationship between pheromones and drinking has been established in the present study, future studies with more complex designs could more conclusively examine the meditational role of expectancies in the relationship between ovulatory cues and drinking behavior.

Beyond the specific focus of this study on hidden sexual signaling mechanisms and alcohol consumption, these findings also add to the research support for the existence of adaptations in both human males and females that maximize reproductive success by encouraging mating during a woman’s monthly period of peak fertility. The particular adaptation receiving attention in the present study was, of course, sensitivity by males to female olfactory cues for fertility. The present findings not only replicated previous work, but strengthened the evidence base with methodological supplements. For example, the methods used in the present study were adapted from a series of seminal studies by Miller and Maner (2011) cited in the introduction. Two of their studies overlapped with our methodology in the use of women wearing T-shirts as the source of the pheromone cues, but in these studies their dependent variable was a word-stem completion task and did not include any direct tests of male-female behavior. In the third of their studies, a behavioral test of men’s decision-making was included, but this test used an actual female (confederate) as part of the experimental protocol. Their use of a specific female to assess the olfactory input could have blurred the distinction between the influence of fertility scents in general and the attraction to a specific confederate. We suggest, therefore, that our findings be considered in concert with the earlier work of Miller and Maner (2011) to increase confidence in the overall conclusion that hidden biological signaling pathways can profoundly influence sexual behavior, and concomitant behaviors that are seen to reinforce sexuality (i.e., alcohol use).

Despite these converging findings, we remain cautious in our reporting of these results given the reported problems (Wood et al., 2014) with replication in studies in the general domain of (female) sexual preferences and sexual signaling (although not in connection with studies of male sexual behavior). At the same time, however, we do not wish to downplay the substantial implications of a previously unknown biobehavioral pathway that might critically influence drinking practices. The importance of this pathway is amplified by the possibility that its period of peak influence might coincide with the developmental periods of adolescence and young adulthood. Why a pathway of this kind might be so influential during this developmental range is suggested by evolutionary theory; sexual/mating behavior is thought to become privileged during this time. That is, behavioral routines in this age range are theorized to be heavily oriented toward the competition for high status mates (Masten, Faden, Zucker, & Spear, 2008). If alcohol, long known to interface with sexual/social behavior, is supported in further research as a potentiator of even these most fundamental evolutionary processes, this newly identified pathway might be an important influence on drinking over this developmental period of high alcohol consumption, and perhaps over a lifetime.

These findings also underscore the complexity of the processes that may underlie what superficially appears to be conscious behavioral decision-making. Based on these results, a complex neurobehavioral pathway that travels across both females and males can be inferred: a cascade of brain-initiated neuroendocrine signals in the female triggers the internal manifestations of fertility (Stern & McClintock, 1998), and triggers the emission of scent-based (chemical) signals. The pathway then crosses from the females to the males via dispersion of the scents in the air. These scents then attract males to the fertile females at a time when mating might maximize reproductive success. In the males who receive these signals, olfactory sensory activation then leads to associational linkages operating outside awareness, which in turn influences alcohol consumption (because the associations link alcohol with a sexual advantage) and approach behavior toward women. Even beyond the drinking/sexual domain, these findings are noteworthy; not only do they suggest that yet unknown pathways of behavioral influence may be operating hidden from view, but they also indicate that the female signaler, and the male recipient of the signal, may comprise an evolutionary “unit” that optimizes procreation.

Future Directions

The obvious need for replications should include tests for generalizability in other groupings of males, the use of alcoholic beer, and use of other types of alcoholic beverages (e.g., wine, mixed drinks). It is possible that exposure to an ovulation cue in the context of actual pharmacological effects of alcohol may potentiate the effects reported in this study. A limitation to our study is that many variables that could have influenced the effects reported in our study--- general thirst of the participant, time and day of the experiment, gender of the experimenter, characteristics of the T-shirt wearer, etc.--- could not be tested in a single study. Future research also should examine if similar findings could be elicited from females, based on as yet unidentified signaling pathways flowing in the opposite direction.