Hormones and Behavior 42, 274–283 (2002) doi:10.1006/hbeh.2002.1826

Psychological Effects of Musky Compounds: Comparison of with and Muscone

Suma Jacob, Sheila Garcia, Davinder Hayreh, and Martha K. McClintock1 Department of Psychology, Institute for Mind and Biology, The University of Chicago, Chicago, Illinois 60637

Received June 7, 2001, revised January 18, 2002, accepted April 11, 2002

Previously, we have shown that ⌬4,16-androstadien-3- Social chemical signals influence physiology, psy- one modulates psychological state, reducing negative chological state, and behavior in many animal species mood and increasing positive mood (Jacob and Mc- (McClintock, 2000, 2002). Studying the effects of Clintock, 2000; Jacob et al., 2001a). In order to determine chemical signals on humans is particularly challeng- whether similar musky compounds also produce these ing because human behavior is multiply determined effects, we compared the effects of androstadienone to and there are myriad potential chemosignals. These those of androstenol and muscone, measuring the psy- include social odors, both conscious and unconscious, chological states of 37 participants. Androstenol and as well as (McClintock, 2001). Responses muscone were chosen because they too have a musky to these chemosignals can reflect their odor qualities, odor at high concentrations, while androstenol is a ste- roid like androstadienone and muscone is not. In a con- such as pungency, familiarity, hedonic valance, level trolled laboratory setting, we conducted a double-blind, of conscious detection, and recognized source of the within-subject, repeated-measures experiment counter- stimulus. Previously, we have demonstrated that a ⌬ balanced for order of presentation. Under each partici- particular , 4,16-androstadien-3-one (common pant’s nose, a nanomolar amount of each compound name: androstadienone), modulates psychological was presented, masked by clove oil to minimize percep- and physiological variables without conscious detec- tible olfactory differences. Participants completed a tion as an odor (Jacob and McClintock, 2000); Jacob et baseline psychological battery and twice again at 25-min al., 2001a). The goal of this study was to begin evalu- intervals after exposure. Androstadienone’s effects on ating the chemical and olfactory stimulus specificity of psychological state were unique in comparison with androstadienone. those of androstenol and with muscone. Exposure Nanomolar amounts of androstadienone affect gen- through passive inhalation, rather than dermal contact, eral emotional and arousal states, quantified by well- was sufficient for these effects. Although this is addi- validated measures of psychological state (POMS, tional evidence that androstadienone may be a phero- ARCI, and VAS scales; (Jacob and McClintock, 2000)). mone, it is yet to be determined whether humans exude Androstadienone prevents the drop in positive mood concentrations into the air adequate for social commu- and rise of negative mood, which typically occur un- nication or process this chemical information within nat- der the odor carrier condition during our experimen- ural social contexts. © 2002 Elsevier Science (USA) tal protocols. This suggests that androstadienone Key Words: chemosignals; ; odor; mood; modulates the physiological mechanisms of emotional musky compounds; . states and attention. Indeed, androstadienone alters cerebral glucose utilization in brain areas not typically associated with olfaction, including subcortical re- gions and widespread areas of the neocortex, includ- 1 To whom correspondence and reprint requests should be ad- dressed at Department of Psychology, The University of Chicago, ing the prefrontal cortex, superior parietal area, and 5730 S. Woodlawn Avenue, Chicago, IL 60637. Fax: 773-702-0320. visual cortex (Jacob et al., 2001b). These widely distrib- E-mail: [email protected]. uted changes are consistent with modulation of an

0018-506X/02 $35.00 © 2002 Elsevier Science (USA) 274 All rights reserved. Androstadienone, Androstenol, Muscone 275 integrated neural network for regulation of emotional one, have a musky odor. Androstenol is in the same states and attention. The observed effects of androsta- steroid family of 16-androstenes, while muscone is a dienone are not sex-specific, as suggested by others 15-carbon ring with a ketone functional group. All (Monti-Bloch and Grosser, 1991; Monti-Bloch et al., three have been used in human perfumes as essential 1994); both men and women respond to androstadi- musky ingredients (Engen, 1991). Such compounds enone (Jacob and McClintock, 2000a; Jacob et al., are also stable and effective fixatives used in many 2001a). Moreover, effects on the autonomic nervous different perfumes (Berliner et al., 1991). Thus, there is system and mood are seen only in particular social the opportunity for generalization of the effects of contexts (Jacob et al., 2001a). androstadienone through its common associations Humans are, to date, the only species other than with musky odor and common perfumes and fra- pigs known to produce significant amounts of andro- grance products. stadienone. Androstadienone has been measured in Androstenol is a mating releaser pheromone in the peripheral plasma of men at 0.01–0.06 ␮g/100 ml pig (Sus scrofa) and a nonandrogenic 16-androstene unconjugated, 0.05–0.1 ␮g/100 ml sulfate-conjugated steroid like androstadienone. In humans, androstenol (Brooksbank et al., 1969), 98 ng/100 ml (Brooksbank et and androstadienone are part of the same metabolic al., 1972), and 2.05 pmol/ml (Fukushima et al., 1991). It pathway in humans (Gower and Ruparelia, 1993). In is also found in men’s sweat (Labows, 1988), semen addition, adrostenol is also excreted by humans in at (Kwan et al., 1992), and axillary hair (Nixon et al., 1988; least three isomers (Brooksbank and Gower, 1964; Rennie et al., 1990). In women, androstadienone is Brooksbank, 1970). Muscone is a territorial releaser measurable in plasma at 36 ng/100 ml (Brooksbank et pheromone in musk deer (Moschus moschiferus) (Par- al., 1972). Fragrance manufacturers claim to have iso- rott et al., 1985; Agosta, 1992) and is not produced by lated androstadienone from skin cells (Kodis et al., humans. Both androstadienone and androstenol have 1998), although there is no peer-reviewed publication been classified prematurely as human pheromones in stating how they were extracted, isolated, or identified marketing claims for fragrances and in some basic (Preti and Wysocki, 1999). Thus there are several sites research (Monti-Bloch and Grosser, 1991; Monti-Bloch of androstadienone production that would make it et al., 1994). These claims have been made prior to available during human interactions. amassing evidence that they fulfill all of the criteria Others have reported that androstadienone has required to be termed a pheromone in any species unique effects on the surface potentials of the vomer- (Beauchamp, 1976; reviewed in Jacob et al., 2002; Mc- onasal epithelium and speculated that its behavioral Clintock, 2002). Whether or not these claims are ulti- effects would also be unique and not generalize to mately substantiated, the effects of androstadienone similar compounds (e.g., other 16-androstenes). This might generalize to androstenol because they are pro- apparent stimulus specificity was used to justify call- duced together in humans, but not to muscone, which ing androstadienone a pheromone. We therefore is a not produced by humans. sought to evaluate the specificity of androstadienone’s We exposed each participant to androstadienone, effect on psychological state. We chose psychological androstenol, and muscone, masking their musky odor changes as our basis of comparison, deferring any with clove oil, the strong odor used in our previous questions or inferences about receptor specificity until research (Monti-Bloch and Grosser, 1991; Jacob and we had demonstrated functional effects. McClintock, 2000; Jacob et al., 2001a). Following the The effects of androstadienone could be associated comparison design typically used to compare the ef- with its musky odor, detectable at high concentra- fects of similar odorants (Lorig and Schwartz, 1988), tions, even though at low concentrations it altered we made independent comparisons of androstadien- psychological state without being consciously de- one with androstenol and with muscone (two re- tected as an odor. If so, other compounds that have a peated-measures within-subjects analyses). An odor similar musky odor might produce the same psycho- comparison design purposely omits “no-odor” carrier logical effects. conditions, in order to avoid several important con- In the current experiment, we tested this hypothesis founds. The dramatic difference between the odor by comparing androstadienone with two other musky conditions and the “control” condition, which obvi- compounds, the pig pheromone androstenol (5␣-an- ously has no odor, creates an unconscious “Haw- drost-16-en-3␣-ol) and the musk deer pheromone thorne effect.” In addition, dramatically different muscone (3-methylcyclopentadecanone or methylex- brain states between the no-odor carrier condition and altone). Androstenol and muscone, like androstadien- the presence of an odor can produce statistical artifacts 276 Jacob et al.

(i.e., biasing omnibus F tests) simply as the result of of the community, but did not enable comparisons of the difference in information between an odorant- ethnic differences. Participants were in good health, containing condition and a no-odorant condition. This with no sinus problems, and no history of any respi- comparison design is also used for clinical treatment ratory or reproductive disease. Subjects provided trials, in which the gold standard is showing signifi- written consent, as required by the Institutional Re- cant differences from “comparable” substances or pro- view Board at the University of Chicago, and were tocols. Nonetheless, it is important to verify that the paid for their participation. effect sizes in this study are comparable to those in studies where there were carrier controls. Therefore, Testers and Testing Facility we confirmed that androstadienone in the current study had effects similar in size to those in previous Two testers (one male age 24 years, of Scottish/ studies (Jacob and McClintock, 2000). Indian descent, and one female age 22 years, of Fili- Finally, in our previous experiments, we had always pino descent) ran the data-collection sessions. Testers applied androstadienone directly on the skin under were assigned randomly to subjects and ran all three the nose, enabling continuous exposure. This expo- sessions of a given subject. The testers were trained to sure method has always left open the possibility that adhere closely to a written script and demonstrated our observations derived from either direct transcuta- high inter-tester reliability before being approved for neous transmission of the steroid into the bloodstream data collection by the primary study directors (S.J. and or the sustained exposure to the steroid remaining on M.M.). The testers wore no strongly scented products the skin. In this experiment, we determined whether during study sessions. We confirmed that there were brief passive inhalation of the air space above a small no significant differences in any of our measures be- amount androstadienone solution was sufficient to tween the two testers (0.32 Յ all P values Յ 0.96 based produce its psychological effects. We exposed each on analyses of data with tester as a between-subjects subject to the odorants using two methods: (1) simply variable). holding a small amount of the stimulus solution near Participants were tested in an isolated testing room the nose on a swab for only 4 s and (2) applying the (10 ϫ 10 ft in size, 74.3 Ϯ 2°F, with 15 air changes per solution to the skin under the nose with a soaked hour) that was carpeted, with neutral color walls, and swab, as we had done in our previous studies. well lit with fluorescent light. Two desks were posi- tioned side by side facing the wall to the left of the entrance. Throughout the session the participant sat MATERIALS AND METHODS by the desk farther from the doorway in a comfortable chair. Participants Stimulus Solutions and Odor Masking Thirty-seven nonsmoking volunteers (18 women and 19 men) participated in a study that we described Because we successfully masked awareness of ste- as investigating a variety of common household odor- roid chemosignals in our previous studies using clove ants, including compounds in perfumes. Participants oil as a strong-odor olfactory mask (Jacob and Mc- were blind to our specific hypotheses and the identity Clintock, 2000), we decided to use clove oil as the of all compounds. The participants were between the olfactory masking component in this experiment as ages of 18 and 35 years (means: women, 25.7 years; well. The three musky compounds were androsta- men, 22.2 years). Participants were students or associ- dienone (⌬4,16-androstadien-3-one, Steraloids, Inc.), ated with the medical science community at the Uni- androstenol (5␣-androst-16-en-3␣-ol, Sigma-Aldrich versity of Chicago. All participants identified them- Co.), and muscone (3-methyl-cyclopentadecalactone, selves as heterosexual and all but 3 were right- Firmenich). The stimulus solutions were composed of handed. Eight women and eleven men described one of the musky compounds at 250 ␮M concentration themselves as Caucasian or White, 4 women and 3 in a 10% clove oil/90% propylene glycol (by volume) men described themselves as African-American, carrier solution. The participants could not discrimi- Black, or of African descent, 4 women and 3 men nate the three solutions as different odors, either in described themselves as of Asian descent, while the open-ended verbal description or in olfactory charac- rest were of Latin descent or had multiple ethnic back- teristics. The subjects described the three solutions as grounds. This sample represented the ethnic diversity smelling “the same.” To further validate our olfactory Androstadienone, Androstenol, Muscone 277 masking of all three compounds, we asked partici- subscales (McNair et al., 1971; Haertzen, 1974a,b). Both pants to use Likert scales to describe each solution’s the ARCI and the POMS are widely used and vali- intensity (0 to 5), familiarity (0 to 5), and pleasantness dated measures. In contrast, the VAS (Bond and (Ϫ5 to 5). There were no significant differences in Lader, 1974; de Wit et al., 1990) measures do not have perception of these olfactory qualities (0.33 Յ all P’s Յ any preestablished subscales. 0.97). Moreover, to assess possible trigeminal nerve As expected, factor analyses and varimax rotation effects, we asked participants to rate any sensations of confirmed the robustness of previously derived factor itching (0 to 5) or burning (0 to 5). Again, there were structures. Only one of the factors was unique to this no significant differences among all three stimuli in study (anxiousness-irritability) and it did not reveal perception of these somatosensory qualities (0.45 Յ all any treatment effects. We confirmed that the factor P’s Յ 0.55). analysis output corresponded to strongly intercorre- lated clusters of questionnaire scores (all linear asso- ciation and correlation P values less than 0.0001). Psychological Measures and Testing Procedures A measure for each derived factor was calculated We employed psychological measures that have de- using methods described previously (Jacob and Mc- tected effects of androstadienone exposure in several Clintock, 2000). A factor score for each time interval previous studies (Jacob and McClintock, 2000; Jacob et was obtained by calculating a mean of the measures al., 2001a). These were the Addiction Research Center for each of the factors that had a varimax rotated Index (ARCI, cf. (Haertzen, 1974a,b) composed of 49 factor loading Ն0.5. Measures with a negative factor True/False responses which collapse into scores for loading were subtracted from the sum of the measures five families of drug effects); the Profile of Mood States with positive loadings. This sum was then divided by (POMS, McNair et al., 1971) composed of 72 5-point the total number of measures within each factor. The Likert scale responses which are scored as eight emo- individual factors and measures used to calculate each tion/arousal factors; the De Wit Visual Analog Scale factor score are listed in Tables 1 and 2. (VAS) items derived from earlier androstadienone (Ja- Finally, the psychological responses were quantified cob and McClintock, 2000) and psychopharmacologi- as a change from their value at baseline, prior to cal (de Wit et al., 1990) research. Each VAS is a 100-mm exposure to the olfactory stimuli. A change in factor line paired with an adjective. Its ends are labeled score from the baseline was calculated for each subse- “extremely” and “not at all.” A subject indicates how quent time interval and these values were used in much the adjective applies to their state by placing a analyses. Since the focus of our experiment is on the mark on the line at the appropriate point; Bond VAS psychological effects of androstadienone, one repeated- items created from the first reported use of VAS mea- measures ANOVA was done in comparison with an- sures (Bond and Lader, 1974). These are the same as drostenol and a second in comparison with muscone. the De Wit VAS, but, rather than representing a con- tinuum of intensity for a single adjective, they are Status marked with antonymic adjectives at either end of the line. To maximize the ability to discriminate odors, we The four psychological measures yielded 62 scores tested subjects during the periovulatory phase of their (5 ARCI subscales ϩ 7 POMS subscales ϩ 20 De Wit menstrual cycle when olfactory sensitivity is highest VAS items ϩ 30 Bond VAS items ϭ 62 total scores). To (Doty, 1981). Menstrual cycle information was gath- minimize potential Type I errors from statistically test- ered by all 18 women for two consecutive cycle ing each score individually, the data were combined months based on dates of menses onset. The 18 into discrete factor scores. Based on our previous find- women also collected detailed daily basal body tem- ings (Jacob and McClintock, 2000; Jacob et al., 2001a), perature, vaginal secretions, menses, and urinary LH we expected there to be several factors covering pos- data. Basal body temperature was measured before itive and negative emotion as well as states of arousal getting out of bed in the morning. Vaginal secretions or drug-like euphoria. To confirm these factors in the and LH measures were collected at 6 PM in the evening current study, factor analyses were performed sepa- to maximize detection of LH rise (Stern and Mc- rately on the ARCI/POMS data and on the Lab/Bond Clintock, 1996). During the first cycle month, we de- VAS data. These two data sets were analyzed inde- termined the day when their preovulatory LH surge pendently because they are different types of mea- began (LH surge onset) using a commercially avail- sures. The ARCI and POMS have well-established able kit (OvuQuik) and used this information to 278 Jacob et al.

TABLE 1 Factor Loadings (Varimax Rotation) of ARCI/POMS Subscale Responses to Androstadienone, Androstenol, and Muscone by Women and Men

Rotated factor loadings

Negative-confusion Elation-vigor Euphoric-stimulant (Factor 1) (Factor 2) (Factor 3) Rotated variable Eigenvalue: 5.62790 2.21673 1.10372 Uniqueness

Anxiety (POMS) 0.88472 ——0.14538 Depression (POMS) 0.91899 ——0.07420 Anger (POMS) 0.90218 ——0.10500 Fatigue (POMS) 0.69119 ——0.33678 Confusion (POMS) 0.83143 ——0.18067 Vigor (POMS) — 0.86228 — 0.17408 Friendliness (POMS) — 0.67307 — 0.42081 Elation (POMS) — 0.82285 — 0.21888 Intellectual efficacy (BG-ARCI) ——0.74104 0.25530 Stimulant (A-ARCI) ——0.87428 0.18383 Euphoric (MBG-ARCI) ——0.76266 0.23800

Note. Chart of items from the Addiction Research Center Index (ARCI) and Profile of Mood States (POMS) questionnaire items grouped by factor membership. Factors were identified based on scree plot inspection. Items within each factor were included given a varimax rotated factor loading Ն0.5 without rounding.

schedule the testing during the subsequent cycle ing/demonstrating the passive inhalation exposure (Stern and McClintock, 1996). The testing sessions procedure, two tubes contained 50 ␮l of the solution were scheduled during the late follicular/periovula- for the passive inhalation, and two tubes contained tory phase of the menstrual cycle (i.e., within a few 130 ␮l of the odorant for epidermal application. The days of LH surge onset). six tubes were placed on a rack outside the human subject room. Each session followed the same protocol. When par- Design and Procedures ticipants arrived for a session, they first washed their We used a within-subjects, double-blind design that hands with the nonodorous liquid soap. They were was randomized for order of exposure to the three then seated in the human subject room approximately stimulus solutions. Participants were asked to come to 1.25 m to the right of the tester, who gave them intro- our human subject room for three sessions, scheduling ductory instructions. Participants were asked to limit 1 day between sessions. With the exception of the movement and talking throughout the session, but stimulus solutions, sessions were identical in proce- were free to make observations specific to the sessions. dure, environment, time of day (Ϯ1 h), and experi- The sessions were exactly timed and were divided menter interaction. Each participant was tested with into three intervals: baseline, passive inhalation (non- the same tester for each of the three sessions. Which epidermal application), and epidermal application. In tester ran a subject’s sessions was determined on the the baseline interval, the participants were adminis- basis of availability. We confirmed that this resulted in tered a baseline battery of psychological state ques- random assignment of participants to each of the two tionnaires presented on a computer screen. If the par- testers. ticipants completed the tests before the predetermined One half hour before each scheduled session, odor- end of the baseline interval, the tester provided them ants were aliquotted by a laboratory member not in- with light nonstimulating reading material. volved in the study. Six standard 1.5-ml polypro- After 30 min, the passive inhalation exposure pro- pylene, flat top microcentrifuge tubes were used in a cedure was demonstrated to reduce its novelty and session. These have a surface area less than 5 mm in increase awareness and expectations. Using the empty diameter. All tubes were labeled by colored tape, iden- tubes, each of two cotton swabs, one to be held under tifying them according to when in the session they each nostril, was dipped into its own microcentrifuge were to be used. Two tubes were empty for practic- tube for 3 s. Each swab was then held approximately Androstadienone, Androstenol, Muscone 279

TABLE 2 Factor Loadings (Varimax Rotation) of VAS Item Responses to Androstadienone, Androstenol, and Muscone by Women and Men (DeWit Unipolar Scale, Bond Bipolar Scale)

Rotated factor loadings

Clearheaded-lucid Anxious irritability (Factor 1) (Factor 2) Rotated variable Eigenvalue: 15.61533 3.83290 Uniqueness

Alert/(drowsy) 0.86643 — 0.18117 Clear-headed/(fuzzy) 0.86333 — 0.12044 Clear-headed/(light-headed) 0.82417 — 0.18773 Sharp 0.77590 — 0.24290 Energetic 0.67552 — 0.25321 Light-limbed/(heavy) 0.59988 — 0.39754 Stimulated 0.55563 — 0.32992 Incompetent/(proficient) Ϫ0.52578 — 0.20386 Sedated Ϫ0.55626 — 0.37952 Clumsy/(well-coordinated) Ϫ0.58305 — 0.25693 Bored/(interested) Ϫ0.62878 — 0.26281 Stimulated/(sedated) Ϫ0.63037 — 0.25427 Spacy Ϫ0.66088 — 0.26230 Dreamy/(attentive) Ϫ0.75140 — 0.18719 Mentally slow/(quick-witted) Ϫ0.75599 — 0.16933 Lethargic/(energetic) Ϫ0.77844 — 0.15754 Discontented/(contented) — 0.53272 0.30262 Irritated — 0.54226 0.31374 Excited/(calm) — 0.60360 0.40230 Anxious — 0.60526 0.38443 Antagonistic/(amicable) — 0.61784 0.26146 Tense/(relaxed) — 0.75969 0.24224 Tranquil/(troubled) — Ϫ0.74731 0.23617 Peaceful/(irritated) — Ϫ0.72438 0.23701 Mellow/(irritable) — Ϫ0.69088 0.32690 Well/(sick) — Ϫ0.52124 0.31989

Note. Chart of items from the Visual Analog Scale (VAS) questionnaire items grouped by factor membership. Items listed with a slash (/) between them are ends of bipolar VAS items. Single descriptor items are from VAS items which rated a single adjective from “not at all” to “extremely.” Factors were identified based on scree plot inspection. Items within each factor were included given a varimax rotated factor loading Ն0.5 without rounding.

0.5 in. underneath the nostril for 4 s. This baseline 130 ␮l of solution and then swiped twice on the upper presentation contained no odorant and the subjects lip below the nose, once with each side of the swab were asked to continue breathing normally. The par- head. The participants continued to breathe normally ticipants were then asked a series of questions about and were then given the same battery of psychological what they smelled. tests. After each application, the tester placed the used After a minute, the same exposure procedure was cotton swabs and closed microcentrifuge tubes in a repeated but with swabs dipped into tubes with 50 ␮l sealed plastic bag to prevent any additional ambient ex- of the test solution assigned for that session and held posure. The entire session lasted approximately 90 min. just under each nostril. The participants were again asked to breathe normally without sniffing, resulting in passive inhalation of the stimulus solution. They RESULTS were then given the second battery of questionnaires. Main Effects of Treatment Condition Twenty-five minutes after the initial exposure, the participants were exposed again, except that the two Although participants could not identify any olfac- swabs were dipped into microcentrifuge tubes with tory differences between the three masked com- 280 Jacob et al. pounds, androstadienone had unique effects on posi- Effect of Sex of Subject tive and negative moods as well as mental acuity (see Independent of treatment, the sex of the subject was Fig. 1a for androstadienone vs androstenol effects and significant in only one analysis. When androstadien- Fig. 1b for androstadienone vs muscone). The magni- one and muscone data were analyzed together (i.e., tude of the effect of androstadienone was consistent when androstenol data were excluded), male subjects with that in our previous studies, i.e., it prevented the reported a greater average increase in Anxious irrita- drop in positive mood observed during the androste- bility during our study than did the female subjects nol and muscone conditions, which was the same as (F(1, 36) ϭ 6.13, P Յ 0.02). As a result, we included the drop seen during exposure to the odor carrier sex of subject as a potential interaction factor in our alone (Jacob and McClintock, 2000). In this study, analyses of treatment effect. However, there were no androstadienone prevented the drop in elation-vigor significant sex differences in response to the compounds that did occur in the presence of androstenol (main for any measure (0.04 Յ all Fs Յ 1.55, 0.22 Յ all P ϭ Յ effect of treatment, F(1, 35) 6.6, P 0.01) and values Յ 0.85). muscone (main effect of treatment, F(1, 35) ϭ 10.4, P Յ 0.003). This change mirrors the changes in the POMS Positive Mood measure observed in earlier work and DISCUSSION was a virtually identical effect size (Jacob and Mc- Clintock, 2000; a minimal drop in the androstadienone Androstadienone had unique effects on the partici- condition vs a 0.35 standard unit drop in the odor pants’ mood in comparison with androstenol and carrier condition). Similarly, androstadienone pre- muscone, even though they could not distinguish any of the compounds as an odor. It prevented both the vented the rise in clearheaded-lucid factor that oc- drop in positive mood and the rise in the negative curred in the presence of androstenol (main effect of mood that has previously occurred with our experi- treatment, F(1, 35) ϭ 5.0, P Յ 0.03) and muscone mental protocol (Jacob and McClintock, 2000). These (main effect of treatment, F(1, 35) ϭ 4.3, P Յ 0.04). results on mood correspond in both type and effect Again, this was an effect size similar to those of pre- size with our previous findings that compared andro- vious studies. stadienone in its strong clove-scented carrier to its Androstadienone also affected negative mood. An- carrier alone. Androstadienone also had unique effects drostadienone prevented the increase in negative-con- on mental acuity in this study. It should be noted, fusion that occurred during exposure to muscone ϭ Յ however, that our previous studies have not consis- (main effect of treatment, F(1, 35) 6.2, P 0.02), tently identified similar factors measuring mental although its effects were not significantly different acuity, arousal, or alertness, nor has androstadienone ϭ Յ from those of androstenol (F (1, 35) 0.84, P 0.37). had consistently significant effects on this aspect of This effect was specific to negative-confusion, as an- psychological function (Jacob and McClintock, 2000; drostadienone had no effect on anxious-irritability Jacob et al., 2001a). ϭ (main effect of treatment, androstenol, F(1, 36) 0.03, During sustained exposure, androstadienone has lo- Յ ϭ Յ and P 0.85; muscone, F(1, 36) 0.58, P 0.45). calized but widely distributed cerebral effects in areas associated with emotion and attention (Jacob et al., 2001b). The degree to which these effects were distrib- Effect of Exposure Methods uted throughout the brain suggests that the neural effects of androstadienone cannot be conceptualized Passive inhalation and dermal application had in- strictly in terms of olfactory perception and categori- distinguishable effects on mood, indicating that direct zation. The current findings provide additional evi- skin contact and prolonged exposure are not neces- dence that androstadienone is a chemosensory stimu- sary for the psychological effects of androstadienone. lus categorically distinct from simple odorants, with Moreover, exposure method did not interact with psychologically unique effects on humans different treatment in any of the factors. This occurred when from those of two other musky compounds, similar in male and female subjects were combined in a single odor at high concentrations and use in common per- analysis, both with and without a sex of subject inter- fumes and fragrances. It is noteworthy that the effects action term, and when males and female were ana- of androstadienone did not generalize to another 16- lyzed separately (0.0003 Յ all F’s Յ 3.32; .077 Յ all androstene, which is similar in structure and also P’s Յ 0.99). found in a variety of human secretions. Androstadienone, Androstenol, Muscone 281

FIG. 1. Effects of androstadienone, androstenol, and muscone on psychological factors derived from the test battery. Responses are expressed as change from baseline and standardized within subject (mean Ϯ SEM). No effects of exposure method were observed. Data from male (N ϭ 19) and female (N ϭ 18) subjects are combined in these figures as there were no significant sex differences in response. (a) Comparison of androstadienone and androstenol. (b) Comparison of androstadienone and muscone. 282 Jacob et al.

The psychological effects did not require direct epi- female hamsters affect ovarian function in rats (Wei- dermal application (Jacob and McClintock, 2000) or zenbaum et al., 1977). Thus, the fact that androstadi- presentation at the opening of the vomeronasal organ enone has specific effects on humans does not pre- (Monti-Bloch and Grosser, 1991). Airborne presenta- clude that possibility that it may play a part in sow’s tion of nanomolar amount during passive inhalation pheromonal responses to boars along with androstenol, was sufficient. In addition, we found no evidence that particularly because it shares a metabolic pathway. androstadienone has exclusive effects on women, as In sum, our findings provide additional evidence suggested by others (Berliner et al., 1991; Monti-Bloch that androstadienone may be a human pheromone. and Grosser, 1991; Monti-Bloch et al., 1994). This neg- Nonetheless, it still needs to be determined whether ative finding is consistent with our previous studies its psychological effects play a role in human social comparing androstadienone with the odor carrier behavior in the context of everyday life. If it does not alone (Jacob and McClintock, 2000). , function under normal, nonexperimental social condi- which has psychological effects similar to androstadi- tions, then it is not a pheromone. enone (Jacob and McClintock, 2000), may well show If so, what do we call the effects of androstadienone similar stimulus specificity in comparison to androste- when it is presented in nanomolar quantities? In this nol and muscone. miniscule amount, that which is contained in 1/10th Given the current findings and previous work, an- of a drop of sweat, it cannot be called or classified as drostadienone is clearly a human social chemosignal. a social odor because subjects were unable to perceive Others have claimed that it is a pheromone (Berliner et it as a smell or odor, either unmasked or masked, al., 1991; Monti-Bloch and Grosser, 1991; Monti-Bloch et under the conditions of this and all previous experi- al., 1994). We believe this claim to be premature without ments. It could be termed an “unconscious odor” or directly testing each of the various criteria that a com- “a low concentration odorant of which there is pound must meet to be called a mammalian pheromone no conscious awareness” (Lorig et al., 1990; Lorig, (reviewed by (Beauchamp et al., 1976; Jacob et al., 2002)). 1991). However, “unconscious odor” and “unconscious The current study is the first to test the criterion that odorant” are self-contradictory terms. An odor is by androstadienone has specific effects that are unique in definition a conscious percept in humans and so can- comparison with similar compounds, not just different not also be unconscious. Thus, we use the term “va- from the carrier solution (a criterion set forth by sana” as a category for the effects of chemosignals that Beauchamp, 1976). Indeed, androstadienone’s effects function neither as odors nor pheromones, yet do have were unique in comparison with two similar com- effects on psychological state and behavior without pounds, even though they were chemically similar being consciously detected (McClintock, 2001, 2002). (androstenol) or had similar musky odors at high con- How can further work determine whether andro- centrations (muscone and androstenol). stadienone in nanomolar amounts functions as a mod- An additional criterion is that animal pheromones ulatory pheromone or a social signal or is simply a operate independently of their odor properties (crite- vasana? One step will be to compare the reactivity of rion set forth by Beauchamp et al., 1976; McClintock, normal participants, anosmics, and individuals with 2000). Although it is true that some people are able to specific anosmias for androstadienone or other non- detect androstadienone at higher concentrations as an androgenic steroids (Labows and Wysocki, 1984; Pause odor, in this study and in all previous studies, andro- et al., 1999). Another will evaluate the natural dispersal of stadienone produced its effects without being de- androstadienone as a chemosignal and verify its role in tected as an odor. normal human social interactions. Whichever type of Androstenol and muscone are both animal phero- chemosignal androstadienone ultimately proves to be, mones and did not have effects on humans compara- this study has demonstrated uniquivocally that brief ble to those of androstadienone. This observation is passive inhalation of androstadienone, simply from important for evaluating previous claims that andro- the headspace above a small surface, is sufficient ex- stenol is a human as well as a pig pheromone. How- posure to produce its psychological effects. ever, although species-specificity has traditionally been a criterion for classifying a compound as a pher- omone, current data indicate that a number of phero- ACKNOWLEDGMENTS mones may serendipitously exert effects on multiple species, and so this additional classic criterion should This work was supported by NIH MERIT Award R37 MH41788 be dropped. For example, airborne chemosignals from and a grant from the Mind-Body Network of the John D. and Androstadienone, Androstenol, Muscone 283

Catherine T. MacArthur Foundation to Martha K. McClintock, as Kodis, M., Moran, D., et al. (1998). Love Scents. Dutton, New York. well as the Olfactory Research Fund’s Tova Fellowship and NIH Kwan, T. K., Trafford, D. J., et al. (1992). GC-MS studies of 16- MD/Ph.D. Training Grant HD-07009 to Suma Jacob. We thank Tyler androstenes and other C19 steroids in human semen. J. Steroid Lorig for insightful discussions of olfactory study design. Biochem. Mol. Biol. 43(6), 549–556. Labows, J. (1988). Odor detection, generation and etiology in the axilla. Antiperspirants and Deodrants, In C. Felger and K. Laden REFERENCES (Eds.), pp. 321–343. Dekker, New York. Labows, J. N., Wysocki, C. J., (1984). Individual differences in odor Agosta, W. C. (1992). Chemical Communication: The Language of Pher- perception. Perfumer Flavorist 9, 21–26. omones. Sci. Am. Library, New York. Lorig, T. S. (1991). The effects of low concentration odors on EEG Beauchamp, G. K., Doty, R. L., et al. (1976). The pheromone concept activity and behavior. Psychophysiology 5, 69–77. in mammalian chemical communication: A critique. In R. L. Doty Lorig, T. S., Herman, K. B., et al. (1990). EEG activity during admin- (Ed.), Mammalian Olfaction. Reproductive Processes, and Behavior, istration of low-concentration odors. Bull. Psychonomic Soc. 25(8), pp. 144–157. Academic Press, New York. 405–408. Berliner, D. L., Jennings-White, C., et al., (1991). The human skin: Lorig, T. S., and Schwartz, G. E. (1988). Brain and odor I: Alteration Fragrances and pheromones. J. Steroid Biochem. Mol. Biol. 39(4B), of human EEG by odor administration. Psychobiology 16(3), 281– 671–679. 284. Bond, A., and Lader, M. (1974). The use of analogue scales in rating McClintock, M. K. (2000). Human pheromones: Primers, releasers, subjective feelings. Br. J. Med. Psychol. 47(3), 211–218. signalers or modulators? In K. Wallen and J. E. Schneider (Eds.), Brooksbank, B. W., Wilson, D. A. A., et al. (1972). Fate of androsta-4, Reproduction in Context, pp. 355–420. MIT Press, Cambridge. 16-dien-3-one and the origin of 3-alpha-hydroxy-5-alpha-androst- McClintock, M. K. (2001). Pheromones and vasanas: The functions 16-ene in man. J. Endocrinol. 52, 239–251. of social chemosignals. In J. A. French, A. C. Kamil, and D. W. Brooksbank, B. W. L. (1970). Labelling of steriods in axillary sweat Leger (Eds.), Evolutionary Psychology and Motivation, Vol. 48, pp. after administration of 3H-D5- and 14C-progester- 75–112. Univ. of Nebraska Press, Lincoln. one to a healthy man. Experentia 26, 1012–1014. 2 McClintock, M. K. (2002). Pheromones, odors, and vasanas: The Brooksbank, B. W. L., Cunningham, A. E., et al. (1969). The detection neuroendocrinology of social chemosignals in humans and of androsta-4, 16-dien-3-one in peripheral plasma of adult men. animals. In D. Pfaff, A. Arnold, A. Etgen, R. Rubin, and S. Fahr- Steroids 13(1), 29–50. bach (Eds.), Hormones, Brain & Behavior, pp. 797–870. Academic Brooksbank, B. W. L., and Gower, D. B. (1964). “The use of thin- Press, San Diego, CA. layer and gas-liquid chormatography in the identification of 5␤- McNair, D. M., Lorr, M., et al. (1971). Profile of Mood States (Manual). androsten-16-en-3␣-ol and androsta-5,16-dien-3␤-ol in human Educational and Industrial Testing Service, San Francisco. urine.” Steroids 4, 787–800. Monti-Bloch, L., and Grosser, B. I. (1991). Effect of putative phero- de Wit, H., Metz, J., et al. (1990). The use of PET to study psycho- mones on the electrical activity of the human vomeronasal organ active drugs: Relationships between mood and metabolic rate. and the olfactory epithelium. J. Steroid Biochem. Mol. Biol. 39(4B), Committee on Problems of Drug Dependence, Richmond, VA. 573–582. Doty, R. L. (1981). Olfactory communication in humans. Chem. Monti-Bloch, L., Jennings-White, C., et al. (1994). The human vom- Senses 6(4), 351–376. eronasal system. Psychoneuroendocrinology 19, 673–686. Engen, T. (1991). Odor Sensation and Memory. Praeger, New York. Nixon, A., Mallet, A., et al. (1988). Simultaneous quantification of 5 Fukushima, S., Akane, A., et al. (1991). Simultaneous determination odorous steroids (16-androstenes) in the axillary hair of men. J. of and androstadienone (sex attractant) in human Steroid Biochem. 29(5), 505–510. plasma by gas chromatography-mass spectrometry with high- Parrott, R. F., Booth, W. D., et al. (1985). Aggression during sexual resolution selected-ion monitoring. J. Chromatogr. 565, 35–44. encounters between hormone-treated gonadectomized pigs in the Gower, D. B., and Ruparelia, B. A. (1993). Olfaction in humans with presence or absence of boar pheromones. Aggressive Behav. 11(3), special reference to odorous 16-androstenes: Their occurrence, 245–252. perception and possible social, psychological and sexual impact. J. Pause, B. M., Krauel, K., et al. (1999). Body odor evoked potentials: Endocrinol. 137, 167–187. A new method to study the chemosensory perception of self and Haertzen, C. (1974a). Addiction Research Inventory. National Insti- non-self in humans. Genetica 104, 285–294. tute of Mental Health, Washington, DC. Preti, G., and Wysocki, C. J. (1999). Human pheromones: Releasers Haertzen, C. (1974b). An Overview of the Addition Research Center or primers. In R. E. Johnston, D. Mu¨ ller-Schwarze, and P. W. Inventory (ARCI): An Appendix and Manual of Scales. National Sorensen (Eds.), Advances in Chemical Signals in Vertebrates. Vol. 8, Institute on Drug Abuse, Rockville, MD. pp. 315–331. Kluwer Academic/Plenum, New York. Jacob, S., Hayreh, D. J. S., et al. (2001a). “Context-dependent effects Rennie, P. J., Holland, K. T., et al. (1990). 16-Androstene content of of steroid chemosignals on human physiology and mood.” apocrine sweat and microbiology of the human axilla. In D. W. Physiol. Behav. 74(1–2), 15–27. MacDonald, D. Muller-Schwarze and S. N. Natynczuk (Eds.), Jacob, S., Kinnunen, L., et al. (2001b). “Sustained human chemosignal Chemical Signals in Vertebrates, pp. 55–60. Oxford Univ. Press, unconsciously alters brain function.” Neuroreport 12(11), 2391–2394. Oxford. Jacob, S., and McClintock, M. K. K. (2000). “Psychological state and Stern, K. N., and McClintock, M. K. (1996). Individual variations in mood effects of steroidal chemosignals in women and men.” biological rhythms. In M. F. Jensvold (Ed.), Psychopharmacology of Horm. Behav. 37, 57–58. Women, pp. 393–413. Am. Psychiatry Press, Washington, DC. Jacob, S., Zelano, B., et al. (2002). Assessing putative human phero- Weizenbaum, F., McClintock, M. K., et al. (1977). Decreases in vag- mones. In C. Rouby, et al. (Eds.), Olfaction, Taste, and Cognition, pp. inal acyclicity of rats when house with female hamsters. Horm. 178–195. Cambridge University Press, Cambridge. Behav. 8, 342–347.