Article
Reference
Androstadienone's influence on the perception of facial and vocal attractiveness is not sex specific
FERDENZI, Camille, et al.
Abstract
The androgen steroid androstadienone, an odorous compound emitted from the human axillary region, has recurrently been considered as a candidate compound involved in human chemical communication and mate choice. Although perception of androstadienone has been shown to influence several affective (mood), attentional, physiological and neural parameters, studies investigating its impact on human attractiveness remain unpersuasive because of incomplete designs (e.g., only female participants) and contradictory results. The aim of this study was to investigate how androstadienone may influence others' attractiveness.
Specifically, we used a complete design (male and female raters, male and female faces and voices) to determine whether androstadienone influences the perception of social stimuli in a sex-specific manner, which would favor pheromonal-like properties of the compound, or in a more general manner, which would suggest that the compound has broader influences on human psychological responses. After comparing the ratings of men and women who were exposed to androstadienone masked in clove oil with those of men and [...]
FERDENZI, Camille, et al . Androstadienone's influence on the perception of facial and vocal attractiveness is not sex specific. Psychoneuroendocrinology , 2016, vol. 66, p. 166-175
DOI : 10.1016/j.psyneuen.2016.01.016
Available at:
http://archive-ouverte.unige.ch/unige:89224
Disclaimer: layout of this document may differ from the published version.
1 / 1
Contents lists available atScienceDirect
Psychoneuroendocrinology
j o u r n a l h o m e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / p s y n e u e n
Androstadienone’s influence on the perception of facial and vocal attractiveness is not sex specific
Camille Ferdenzi
a,b,∗, Sylvain Delplanque
a,b, Reni Atanassova
a,b, David Sander
a,baSwissCenterforAffectiveSciences,CampusBiotech,UniversityofGeneva,Geneva,Switzerland
bLaboratoryfortheStudyofEmotionElicitationandExpression(E3Lab),DepartmentofPsychology,FPSE,UniversityofGeneva,Geneva,Switzerland
a r t i c l e i n f o
Articlehistory:
Received25June2015
Receivedinrevisedform15January2016 Accepted15January2016
Keywords:
Humanchemosignals Olfaction
Attractiveness Sexdifferences Menstrualcycle
a b s t r a c t
Theandrogensteroidandrostadienone,anodorouscompoundemittedfromthehumanaxillaryregion, hasrecurrentlybeenconsideredasacandidatecompoundinvolvedinhumanchemicalcommunication andmatechoice.Althoughperceptionofandrostadienonehasbeenshowntoinfluenceseveralaffective (mood),attentional,physiologicalandneuralparameters,studiesinvestigatingitsimpactonhuman attractivenessremainunpersuasivebecauseofincompletedesigns(e.g.,onlyfemaleparticipants)and contradictoryresults.Theaimofthisstudywastoinvestigatehowandrostadienonemayinfluenceothers’
attractiveness.Specifically,weusedacompletedesign(maleandfemaleraters,maleandfemalefacesand voices)todeterminewhetherandrostadienoneinfluencestheperceptionofsocialstimuliinasex-specific manner,whichwouldfavorpheromonal-likepropertiesofthecompound,orinamoregeneralmanner, whichwouldsuggestthatthecompoundhasbroaderinfluencesonhumanpsychologicalresponses.After comparingtheratingsofmenandwomenwhowereexposedtoandrostadienonemaskedincloveoil withthoseofmenandwomenwhowereexposedtocloveoilalone,wefoundthatandrostadienone enhancedtheperceivedattractivenessofemotionallyrelevantstimuli(opposite-sexstimuliinmenand infertilewomen).Responsetimesforcategorizingthestimuliasattractiveornotwerealsoaffectedby androstadienone,withlongerresponsetimesinmenandinfertilewomenandshorterresponsetimes innon-fertilewomen,irrespectiveofthestimulussex.Theresultsfavorthehypothesisofgeneraleffects oversex-specificeffectsofandrostadienone,thusquestioningtherelevanceoffocusingonthatparticular compoundinthestudyofhumanattractivenessthroughbodyodorandencouragingthesearchforother semiochemicalsthatmightbesignificantforhumanmatechoice.
©2016ElsevierLtd.Allrightsreserved.
1. Introduction
Pheromonalcommunicationhasbeendemonstratedinawide range of species acrosstheanimal kingdom (Wyatt 2014) (see the pioneer work on sexual attraction in the moth related to themolecule“bombykol”:Butenandtetal.,1959),includingsev- eralmammals(pups’attraction towardthemother’s mammary pheromone in therabbit:Schaal et al.,2003; the Darcineffect on sexual attraction in mice: Roberts et al., 2010). Research in humans, however has led to much more ambiguous and controversial results. Among the wide variety of substances excretedinhumanbodyfluids(urine,saliva,genitalsecretionsand sweat;Stoddart, 1990), severalandrogenderivatives present in
∗Correspondingauthor.Presentaddress:CentredeRechercheenNeurosciences deLyon,UniversitéLyon1,50avenueTonyGarnier,69366LYONCedex07,France.
E-mailaddresses:camille.ferdenzi@inserm.fr,camillechloe.lemaitre@gmail.com (C.Ferdenzi).
apocrinesweathavereceivedmuch attentionfromscientists in the quest to identify human pheromones: androstenone (5␣- androst-16-en-3-one; e.g., Kirk-Smith and Booth, 1980; Pause, 2004),androstenol(5␣-androst-16-en-3␣-ol;e.g.,Kirk-Smithetal., 1978;MaiwormandLangthaler,1992)and,inthemostrecentstud- ies,androstadienone(androsta-4,16-dien-3-one;e.g.,Bensafietal., 2004a;Hummer and McClintock,2009; Saxton etal., 2008).At leasttwomainhistoricalreasonscanbecitedforstudyingthese volatilesteroidsaspossiblehumanpheromones.First,theyhave beendirectly linkedwiththereproductive behavior ofanother mammal(lordosisinthefemalepig:Dorriesetal.,1995).Second, somestudieshaveshownthatthesesteroidsareemittedinasex- uallydimorphicmanner,withhigherconcentrations,onaverage, inmen(androstenone:BirdandGower,1981;Goweretal.,1985;
androstadienone:Brooksbanketal.,1972).Despitethefactthat thesejustificationsarehighlydebatable(e.g.,seeGoweretal.,1994;
foracounterexampleofawomansecretingmoreofthesecom- poundsthanmostmenwhoweretested),numerousstudieshave http://dx.doi.org/10.1016/j.psyneuen.2016.01.016
0306-4530/©2016ElsevierLtd.Allrightsreserved.
focusedonthesecompoundsaspotentialcandidatesforahuman pheromone.
Precise definitions have been proposed for the concept of pheromone(Beauchampetal.,1976;KarlsonandLüscher,1959;
Wyatt,2010),1butsofarstudiesinhumanshavenotmanagedto identifymoleculesfittingthesecriteria(Wyatt,2015)and some authorsareskepticalabouttheexistenceofpheromonesinhumans (Doty, 2010).Studiesinvestigatingputativepheromonalproper- ties of somecompounds(mostly those citedearlier, chosen on thebasis of questionablearguments) have tested, using varied methodologies,theirinfluenceonhumanbehaviorandphysiologi- calandemotionalstates.Ratherthanthoroughlytestingtheclassic definitionsofpheromones,studiesinhumanshaveinsteadhypoth- esizedthatifthesecompoundshadabiologicalfunctioninsexual behavior,then(1)perceptionofcandidatecompoundscouldbesex specific,inthattheycouldelicitresponsesdifferinginintensityorin qualitybetweenthesexes;(2)variationsshouldoccurduringthe menstrualcycleofwomen,withanenhancedeffectofthecom- poundsaroundovulation,whentheriskofconceptionishigher;
(3)sexualorientationshouldbeinfluential;and(4)age,acorrelate ofreproductiveability(youngadultsversuschildrenforexample), shouldalsobelinkedwithvariationsintheperceptionandeffects ofthesecompounds.
Mitigated evidence has been obtained so far on these four criteriafortheandrogencompoundthathasrecentlybeenreceiv- ingthegreatestattention:androstadienone. Psychophysicaland neuroscientificevidencesuggeststhatthisodorouscompoundis processedinasex-dimorphicmanner.First,ithasbeenshownthat womendetectandrostadienoneatlowerconcentrationsthanmen do(KoelegaandKöster,1974;Lundströmetal.,2003b).Thesedif- ferencesmayappearfrompubertyonwards(Hummeletal.,2005), buttheroleofsexualmaturationremainsuncertainbecauseother studiesalsofoundsexdifferencesinpre-pubertalchildren’sneu- ralresponsetothismolecule(Burkeetal.,2014).Whethersuch aphenomenonisspecifictoandrostadienonealsoremainsuncer- tain,becausesimilarpuberty-relatedchangeshavebeenfoundfor othermalodors(Chopraetal.,2008).Second,Savicetal.(2001) foundthatandrostadienoneinduceshypothalamicactivationina sex-specificmanner(higheractivationofthisareainwomencom- paredwithmen)andinasexualorientation-specificmanner(like heterosexualwomen,homosexualmenhavehigherhypothalamic activationthandoheterosexualmen:Savicetal.,2005).Thesex specificityofthesebrainactivationswas,however,questionedina studythatuseddifferentconcentrationsandfoundhypothalamic activationsinbothmenandwomen(Burkeetal.,2012).Regarding thevariationsduringthemenstrualcycle,womenwerefoundto bemoresensitivetoandrostadienonearoundovulationthanwere womeninthenon-fertilephaseorwomenwhowereusingoral contraceptives,whichwasnotthecaseforanothernon-bodyodor, phenyl-ethanol(Lundströmetal.,2006).
1AccordingtoKarlsonandLüscher(1959),pheromonesare“substanceswhich aresecretedtotheoutsidebyanindividualandreceivedbyasecondindividualof thesamespecies,inwhichtheyreleaseaspecificreaction,forexample,aspecific behaviororadevelopmentalprocess.”Theconceptofpheromonewasthenrede- finedbyBeauchampetal.(1976)tobetterfitthemammalianmodel:tothem,a pheromoneisasinglemolecule(oratmostamixofonlyafewcompounds)having awell-definedbehavioralorendocrinologicalfunctionthatisspeciesspecificand expressedthroughstereotypedresponsesthatdonotresultfromlearningorexpo- sureeffects.Morerecently,emphasizingthedifferencebetweenpheromonesand signaturemixtures(highlyvariableodorslearnedforindividual/familyrecognition), Wyatt(2010)proposedamodifiedversionofKarlsonandLüscher(1959)definition ofpheromones:“moleculesthatareevolvedsignals,indefinedratiosinthecaseof multiplecomponentpheromones,whichareemittedbyanindividualandreceived byasecondindividualofthesamespecies,inwhichtheycauseaspecificreaction, forexample,astereotypedbehaviororadevelopmentalprocess.”
Further,theeffectofandrostadienoneonhumanbehaviorand onphysiologicalandemotionalstateshasreceivedmuchattention.
Several studies have highlighted positive effects of androsta- dienone onparticipants’ mood and alertness: some in women only(Bensafietal.,2004a,b;JacobandMcClintock, 2000), some inbothsexes(HummerandMcClintock,2009;Jacobetal.,2002) and some in female participants without comparison to males (Grosser et al., 2000; Lundström et al., 2003a;Lundström and Olsson,2005).Discordantevidencehas,however,beenreported regarding physiological responses of the autonomous nervous system(sympathetic-likeeffects in Bensafiet al.,2004a;versus parasympathetic-likeeffectsinGrosseretal.,2000),andthesewere context dependent(varyingaccordingtothesexof theexperi- menter:LundströmandOlsson,2005).Althoughthesestudiesdid not specificallyinvestigatesexualbehavioror matechoice(but ithasbeenclaimedthat“apositivemoodisknowntofacilitate women’ssexualresponse, and increasedfocus improvessexual satisfaction,” Verhaeghe et al., 2013), other studies have more specificallytested theeffectof androstadienoneonthepercep- tionofsocialstimuli.First,androstadienonewasfoundtoenhance attentiontowardsexlessdrawingsofemotionalfacesinbothmen and women(Freyetal.,2012;Hummer and McClintock,2009).
Thespecificityofthiseffecttowardsocialstimuliversusnon-social objects is underdebate (nonspecific: Hummer and McClintock, 2009; specific: Parmaet al., 2012). Studiesthat have explicitly investigated theeffect of androstadienone onperceived attrac- tivenessarescarce,havinginvolvedonlyfemaleparticipantsand havingevaluatedonlymalestimuli(LundströmandOlsson,2005;
Saxtonetal.,2008)ormaleandfemalestimuli(Parmaetal.,2012).
Whereasearlierstudiessuggestedthatothersimilarcompounds suchasandrostenolorandrosteronecouldmodulatefaceattrac- tiveness(Kirk-Smithetal.,1978;MaiwormandLangthaler,1992), androstadienonewasfoundtohavenoimpactontheperceptionof faceattractiveness(LundströmandOlsson,2005),tohaveaneffect thatisnotreplicable(significantinoneofthreespeed-datingstud- ies:Saxtonetal.,2008),ortohaveaneffectontheperceptionof same-sexfacesonly(Parmaetal.,2012).
Giventheinconsistenciesintheseresultsandthemethodolog- icalshortcomingsofthesestudies(nocomparisonbetweenmale andfemaleresponsesinthepresenceofandrostadienone;men- strualcycletakenintoaccountinonlyonestudy),wedevisedanew studywithamorecompletedesigntoinvestigatehowandrosta- dienonemayinfluenceothers’attractiveness.Moregenerally,this studyaimedtopresentnewelementstodeterminewhetheritis relevant–for abetterunderstandingof humanchemosignaling in matechoice–tokeep focusingonthis particularcompound ratherthanonothers(Wyatt,2015).Withthisaim,wecollected attractivenessevaluationsofmaleandfemaleparticipants,thelat- ter beingallocated toa “fertile” groupor a “non-fertile” group accordingtothephaseoftheirmenstrualcycleatthetimethey participated inthestudy.We usedahighly standardizedsetof socialstimulithatwerevariedinattractiveness,includingmaleand femalefacesand–forthefirsttimeinandrostadienonestudies– voices,takenfromtheGEnevaFacesandVoicesdatabase(GEFAV:
Ferdenzietal.,2015).Weusedbothfacesandvoicesbecausewe wantedtotestwhethertheeffectsthathavemostlybeentested inthevisualdomainwouldreplicateinanotherrelevantmodality.
Abetween-subjectdesignallowedustocomparetheresponsesof participantsexposedtoandrostadienonewiththeresponsesofpar- ticipantsexposedtoacontrolodor.Weexamined(i)thespeedof processingoffaces/voices,measuredbyresponsetimetocatego- rizethestimulusasattractiveorunattractive,and(ii)thevalence offaces/voices,measuredbyattractivenessratings.
With this design, we tested whetherandrostadienone influ- ences the perception of others’ faces and voices and whether thiseffectissexspecific(regardingtheperceiverandtheperson
evaluated)or moregeneral.In thefirst hypothesis(sex-specific effects),androstadienonewouldactasasignalofmatequality(see ThornhillandGangestad,1999;foralinkbetweenbodyodorand matequality).Beingaby-product oftestosterone(Rennieetal., 1989),androstadienonecouldconstituteatestosterone-dependent sexualtrait:thesetraits,suchasdeepvoicesormasculinefacial features,signalbiological qualitytoa prospectivemate(Folstad andKarter,1992)andincreaseattractiveness(Collins,2000;Perrett et al., 1998). Consequently, men should be more attractive to womeninthepresenceofandrostadienone,andthiseffectshould beenhancedduringthefertilephaseoftheperceiver’smenstrual cycle(inaccordancewiththeincreasedpreferenceformasculine faces,voicesandbodiesaroundovulation:GangestadandThornhill, 2008).Notethatmasculinity/femininityratingswerecollectedto testhowthisdimensioncouldmediatepotentialeffectsonattrac- tiveness.Althoughandrostadienone should havenoeffect, or a negativeeffect,onwomen’sattractiveness,becausefemininity– notmasculinity–tendstodrivepreferencesinthiscase(Collinsand Missing,2003;Fraccaroetal.,2010),maleratersmaybeabletoinfer othermen’squalityfromtheodorofandrostadienoneassociated withthem(HuovialaandRantala,2013)andthereforeratethemas moreattractive.Becauseofthecostofattractivenessevaluationsin termsofattentionalresources(Jungetal.,2012)andthedecrease inprocessingspeedoffaceswithincreasingattractiveness(Kranz andIshai,2006),thetimeneededtocategorizethestimuliasattrac- tiveornotwouldbeexpectedtofollowattractivenesschanges,i.e., toincreaseifattractivenessincreasesinthepresenceofandrosta- dienoneandtodecreaseifattractivenessdecreases.Inthesecond hypothesis(moregeneral),androstadienonewouldhaveaninflu- encethat isnotsexspecific,butinsteadspecificallydirectedto emotionallyrelevantinformation(HummerandMcClintock,2009).
Here,themostemotionallyrelevantstimuliare,giventhenature ofthetasks(attractivenessjudgments)andthesexualorientation oftheparticipants(heterosexual),opposite-sexstimuli.Androsta- dienonemaythusspecificallymodulateattentiontotheemotional contentofopposite-sexstimuli:if,assuggestedbyHummerand McClintock(2009, p.556),“emotionalcontent[is] moredifficult toignorewhenandrostadienone[is]present,”thenopposite-sex stimulishouldcaptureattentionmoreandsubsequentlyincrease the time taken to process them in the categorization task. As suggested by Villemure and Bushnell (2007), it could be that only attention to the emotional content of these stimuli, not theemotionalcontentitself(attractiveness),maybeaffectedby androstadienone.
2. Method
2.1. Participants
Forty men and 40 women with a mean age of 23.1 years (SD=4.4)wererecruitedfromthestudents oftheUniversity of Geneva.Inclusioncriteria,basedonself-declaration,wereasfol- lows:18–35yearsofage;ofEuropeanorigin;heterosexual;anative Frenchspeaker;not aregular smoker;havingnormal olfaction, visionandaudition;and,forwomen,notbeingpregnantortaking anyformofhormonalcontraception.Onefemaleparticipantwho failedthetestfornormalolfaction(seeSection2.3.2)wasexcluded fromthedataanalysis.Participants wererandomlyallocatedto oneoftwoexperimentalgroups:onegroup(CA)wasexposedto androstadienonemaskedincloveoil,andtheother(control)group (C)wasexposedtocloveoilalone.TheCAgroupcomprised20men and20women,including8whowereinthefertilephaseoftheir menstrualcycle.TheCgroupcomprised20menand19women, including11fertile women.Fertility statuswasestimatedfrom theparticipants’reportsoftheonsetoftheircurrentcycle(first
dayoflastmenstruation)andoftheirusualcyclelength.Wefirst computedtheprobabledateoftheendofthecyclebeforeusinga backwardcountingmethod(Ferdenzietal.,2009;Thornhilland Gangestad,1999),estimating thefertile periodbetween14 and 20daysfromtheexpectedendofthecycle.Appointmentswere scheduledsoastoincludeapproximatelyasmanyfertileasnon- fertilewomen.TheCandCAgroupsdidnotdifferinage(t77=1.69;
p=.094)orinthenumberoffertile/non-fertilewomen(2=1.25, p=.26).Participantswereinformedthatweaimedtotesttheeffect ofcommonodorsontheperceptionoffacesandvoices,butnoinfor- mationwasprovidedregardingthenatureoftheodorousstimuli.
Participantsreceivedfinancialcompensationfor theirparticipa- tion.Theygave writteninformedconsentprior toparticipation, andtheCommitteeonResearchEthicsoftheFacultyofPsychology andEducationSciencesattheUniversityofGenevaapprovedthe study.
2.2. Stimuliandquestionnaires 2.2.1. Odorants
Participantstookpartintwosessions:themainevaluationses- sionandtheolfactoryscreeningsession(seeSection2.3).Inthe mainevaluationsession,toenablecomparisonwithotherstudies, weusedthesameconcentration,natureandadministrationofthe odorantsasthoseproposedbyJacobandMcClintock(2000)and laterusedbyotherresearchers(e.g.,Lundströmetal.,2003a;Saxton etal.,2008).Theexperimentalsolution(CA)wasa250Mconcen- trationofandrostadienoneinpropyleneglycolwith1%cloveoilas amaskodor,andthecontrolsolution(C)was1%cloveoilinpropy- leneglycol.The odoroussolutions wereplaced intwo identical flasks,AandB,byoneexperimenter(CF),whileanotherexperi- menter(RA),whowasblindtothecontentoftheodorflasks,ranthe experimentalsessions.Fortheolfactoryscreeningsession,three triplets ofodorpens (Sniffin’Sticks;BurghartGmbh,Germany) wereusedtotest1)theefficiencyoftheclovemask,2)anosmia toandrostadienoneand3)generalolfaction,respectively.Triplet1 comprisedtwopensfilledwithCandonewithCA,triplet2con- sistedoftwoblanksversusandrostadienoneatahighconcentration (5000M),andtriplet3consistedoftwoblanksversusacomplex strawberryaroma.AllsubstanceswereprovidedbyFirmenichSA, Geneva.
2.2.2. Facesandvoices
Allparticipantswerepresentedwiththesame120stimuli,con- sistingofpicturesofneutralfacesandvoiceexcerpts(sayingthe Frenchword“bonjour”)of30menand30womenfromtheGEneva FacesandVoicesDatabase(GEFAV;Ferdenzietal.,2015).All60 GEFAVdonorswerenativeFrenchspeakers,ofEuropeanoriginand 18–35yearsold.Raterswerenotinformedthatthefacesandthe voicescamefromthesamedonors.
2.2.3. Moodquestionnaire
Participantshadtoratetheircurrentmoodtwice,beforeand aftertheevaluationsession.Weselected12itemstakenfromthe validatedFrenchversionsoftheProfileofMoodStates(POMS-f;
Cayrouetal.,2003)andofthePositiveandNegativeAffectSched- ule(PANAS;Gaudreauetal.,2006):attentive,confused,energetic, sluggish,relaxed,nervous,cheerful,sad,enthusiastic,irritated,con- siderateandsensual.Ratingsweregivenbyplacingacursorona continuousvisualanalogscalefrom“notatall”to“extremely.”Par- ticipantswerealsoaskedtoratetheirgeneralmoodfrom“very negative”to“verypositive.”Thepositionofthecursorwasthen transformedintoascorerangingfrom0to100.
2.3. Procedure
Participantstookpartintwoexperimentalsessionsondifferent days.Duringthefirstsession(evaluationsession),theyratedvoices andfacesineithertheCorCAodorcondition,dependingonthe grouptheywereassignedto.Duringthesecondsession(olfactory screeningsession),theirolfactoryabilitiesweretested.Fertilitysta- tusofthefemaleparticipants(fertileornotfertile)hadtobethe sameduringtheevaluationandtheolfactoryscreeningsessions.
Forbothsessions,theparticipantswereinstructednottowearany perfume,nottosmokeonthedayoftheexperiment,andnottoeat ordrinkanythingbutwaterinthehourprecedingthetest.Incase ratershadacoldorablockednoseonthedayoftheexperiment, theyhadtoreschedulethesession.
2.3.1. Evaluationsession
CandCAgroupsperformedthesametaskondifferentdaysto avoidodorcontaminationofthetestingroom.Theroomwaswell ventilatedattheendofeachtestingday.Atthebeginningofthe ratingsession,theodoroussolutionwasappliedontheskinjust beneaththeparticipant’snostrilsbyusingacottonswab.Malepar- ticipantswerethusallowedtotakepartonlyiftheyhadshaved, so as not to interfere withodor application. Participants were thenseatedinfrontofacomputerequippedwithheadphonesand presentedwiththeratinginterfacedesignedwithE-Prime2.0(Psy- chologySoftwareTools,Pittsburgh,PA,USA).Amaximumoffour participantsunderwentthestudysimultaneously.First,theyhadto ratetheircurrentmoodonthe13itemsofthemoodquestionnaire presentedrandomly.Second,theywerepresentedwithtwoblocks ofstimuli:voicesfollowedbyfaces.Eachblockwasdividedinto twotasks:a“first-impression”taskfollowedbya“rating”task,each beingperformedwithopposite-sexstimulibeforesame-sexstimuli (Fig.1).Withineachsub-block(e.g.,firstimpressionofopposite-sex voices),stimuliwerepresentedrandomly.
Inthefirst-impressiontask,participantshadtocategorizethe stimuliaseitherattractiveorunattractiveasfastastheycouldand basedontheirgutfeeling,usingthedigits1and3ofthenumeric keypad(labelpositionswerecounterbalancedacrossparticipants).
Stimuluspresentationwasprecededbya1000msfixationcross toensurethattheparticipantswerepreparedfortheupcoming stimulus.Eachstimuluswas presentedonce for 100ms forthe facesandfor430–870msforthevoiceexcerpts(dependingonthe speaker;durationwasnotmodifiedtopreventalterationofattrac- tiveness:Ferdenzietal.,2013).Responsetimefromthebeginning ofthestimuluswasrecorded,witha3000msmaximumresponse timeallowedaftertheendofthestimuluspresentation.A3000ms blankscreenseparatedtheresponsefromthefixationcrossthat announcedthenextstimulus.Fourpracticetrialswithtwomale andtwofemalevoicesandfacesnotpresentinthemainexperiment wereconductedbeforetherealtaskstarted.
Intheratingtask,participantsratedthesamestimulithatthey saw/heardin thefirst-impressiontaskoncontinuous scalesfor attractiveness,from“notattractiveatall”to“veryattractive,”as wellasformasculinity/femininity,from“veryfeminine”to“very masculine”formalestimuli andfrom“verymasculine”to“very feminine”forfemalestimuli.Thepositionofthecursorwasthen transformedintoascorerangingfrom0to100.Responseswerenot limitedintime:thevocalstimulicouldbereplayedadlibitumand thefacialstimuliremainedvisibleuntiltheparticipantvalidated his/herresponse.
Afterhavingratedthefaces,participantsrated theircurrent moodagainonthe13itemsofthemoodquestionnaire.Attheendof theevaluationsession,participantswereaskedtoratehowpleas- antandintensetheyfoundtheodorappliedontheirupperlip(on a0–8scale)andtobrieflydescribeit.
2.3.2. Olfactoryscreening
The10-minolfactoryscreeningwasconductedindividuallyand tookplaceaftertheratingsession,withatleastonedaybetween them.Foreachofthethreetripletsofodorpensdescribedearlier (seeSection2.2.1),participantswereaskedtoindicatewhichpen smelleddifferentfromtheothertwo.Thiswasrepeatedfivemore timespertripletwiththepenspresentedinrandomorder,allowing ustocalculatethenumberofcorrectanswersbetweenzeroand six.Triplet1(testingefficiencyoftheclovemask)wasusedfirst becauseoftheodorants’lowerconcentration,whereastheother twotripletswereusedafterwardsinrandomorder.Nofeedback wasgiven.Attheendofthescreeningsession,participantswere paidanddebriefed.
2.4. Dataanalysis
First,weexaminedhowtheexperimentalodorswereperceived, namely whether someparticipants were anosmic to androsta- dienone and whether androstadienone was efficiently masked.
Discrimination performances of CA participants between both experimentalodors,andbetweenahighconcentrationofandrosta- dienoneandblanks(olfactoryscreeningsession),wereanalyzed withasinglesamplet-testtocomparethescoreswithchancelevel (i.e.,2)andwithaone-wayanalysisofvariance(ANOVA)tocom- pareratersexgroups.Two-wayANOVAswereconductedtotest theeffectofodorcondition(C,CA)andratersex(men,non-fertile women,fertilewomen)ontheperceivedintensityandpleasant- nessof theexperimentalodor(evaluationscollectedattheend oftheratingsession).PosthocanalyseswereTukeyHSDtestsat
˛=.05.
Second,wetested theeffectofodorconditionandratersex onmood ratingsand onresponsestothefacesand voices. For mooddata,we conductedtwo ANOVAs:(1)a two-way ANOVA withodorcondition(C,CA)andratersex(men,non-fertilewomen, fertilewomen)asbetween-subjectsfactorsongeneralmoodrat- ing,and(2)arepeated-measureANOVAwithmooditem(12mood adjectives) as a within-subjects factor, and odor conditionand rater sex as between-subjects factors.For face and voice data, scoreswereaveragedbystimulus,andanalyseswereconducted forsame-sexandopposite-sexstimuliseparately:stimulussexwas notincludedasafactorbecauseopposite-sexstimuliwerealways presentedfirst,thusmakingthecomparisonirrelevant.Analyses werethusperformedonmatricescomposedof90cases(30stim- uli×3ratersexes:men,fertilewomen,andnon-fertilewomen)by 2averagescores(ineachoftheCandCAgroups),separatematri- cesbeingusedforsame-andopposite-sexstimuli.Therefore,we usedrepeated-measuresANOVAswithodorconditionasawithin- subjectsfactor(“subject”being“stimulus”here)andratersexas abetween-subjectsfactortotesttheeffectsofthesefactorson(i) log-transformedresponsetimesinthefirst-impressiontask,(ii)the percentageofratersineachsexgroupwhocategorizedthestimulus asattractiveinthefirst-impressiontask,(iii)averageattractiveness ratingsand(iv)averagemasculinityratingsintheratingtask.For thelatter,becausetheratingscalewasinvertedforfemalestim- uli,ratingsweretransformedinto100minusthefemininityrating.
Notethat,becauseresponsetimesinthefirst-impressiontaskdid notfollowanormaldistribution,welog-transformedthemafter removingoutliersdefinedonanindividualraterbasisbyaduration equaltothemean±3standarddeviations(i.e.,1.4%ofthefacetrials and.7%ofthevoicetrials).Posthocplannedcomparisonswereper- formedwhensignificant(p<.05)ornearsignificant(p<.10)effects werefound.AllstatisticaltestswereconductedwithStatisticav.12 (StatsoftInc.,Tulsa,OK,USA).
Fig.1.Diagramoftheexperimentaldesign.OS:opposite-sexstimuli,SS:same-sexstimuli.
3. Results
3.1. Detectionandperceptionofandrostadienone 3.1.1. Anosmiatoandrostadienone
Results of the discrimination test for the blanks and high-concentration androstadienone triplet (olfactory screening session)revealedthatCAparticipantswereabletodetectandrosta- dienoneat abetterthan chancelevel(3.6correct detectionsof sixtrials,t39=5.29,p<.001).Takenseparately,womenclearlyhad thisability(fertilewomen:t7=3.79,non-fertilewomen:t11=4.84, ps<.01),whereasmenonlytendedtohaveit(t19=1.96,p=.065).
Therewasa significanteffect of ratersex (F(2,37)=5.02,p<.05) becausewomen(fertileandnon-fertile,notdifferent)performed betterthanmen(posthoctests).Consequently,beforeperforming theanalysespresentedafter,weremovedthedatafromanosmic participants,thatis, 11 participantsfromtheCAgroup(9men and2non-fertilewomen)and9participantsfromtheCgroup(3 men,2non-fertilewomenand4fertilewomen)basedonthefact thattheyhad onlyzero,oneortwocorrect detectionsof sixin thescreeningtestsinvolvingandrostadienone(masked,andhigh- concentrationandrostadienone).Fortheremainingparticipantsof theCAgroup(N=29),thedetectionofpureandrostadienonewas unaffectedbyratersex(F(2,26)=1.45,p=.252)andoccurredatabet- terthanchancelevelinfertilewomen(t7=3.79,p<.01),non-fertile women(t9=6.33,p<.001),andmen(t10=3.01,p<.05)(seeFig.2) 3.1.2. Efficiencyofandrostadienonemasking
Perceptualratingsoftheexperimentalodorcollectedattheend oftheratingsessionweresimilarinCandCAgroups.Indeed,there wasnomaineffectofexperimentalconditionontheodor’sper-
Fig.2. Androstadienonedetectioninfertilewomen(N=8),non-fertilewomen (N=10)andmen(N=11)forthenon-anosmicparticipantsexposedtoandrosta- dienone(CAgroup).Theaveragenumberofsuccessfuldetectionsofsixtrialsisin whitebarsformaskedandrostadienone(i.e.,successfuldiscriminationofasample with250Mandrostadienoneincloveoilfromtwosampleswithcloveonly)andin greybarsforpureandrostadienone(i.e.,successfuldiscriminationofasamplewith 5000Mofandrostadienonefromtwosampleswithsolventonly).Comparisonwith chancelevel(chancelevel=2):***p<.001,**p<.01,*p<.05,nsnon-significant.
ceivedpleasantness(C:4.7±1.8;CA:4.3±1.9;F(1,52)=.99,p=.324) orintensity(C: 3.9±1.9;CA:3.8±1.9;F(1,53)=.00, p=.976). For thesetwo variables,wefoundnomaineffectofratersex(men, non-fertilewomen,fertile women:ps>.298)and nointeraction betweenodorconditionandratersex(ps>.352).Descriptionsof theexperimentalodorsweresimilarinbothgroups. Theywere mostlyrelatedtocloveandotherspicesorplants,tothedentist andmedication,ortosweetsandfruits.Perfumeanddeodorant wereevokedtwiceineachgroup,andtheonlyreferencestonatu- ralhumanbodyodor(“masculineodor”and“earwax”)camefrom twonon-fertilewomenbelongingtotheCAgroup.
Tofurthertestwhetherthemaskingofandrostadienone was efficient,weanalyzedtheresultsoftheCversusCAdiscrimination task(triplet 1oftheolfactoryscreeningsession)of participants fromtheCAgroup.AveragecorrectdetectionofCAwas2.7±1.4 trialsof6,whichwassignificantlyhigherthanchance(t28=2.44, p<.05). Taken separately, men, non-fertile women and fertile womendidnotansweratabetterthanchancelevel(ps>.094)and didnotdiffer(F(2,26)=.36,p=.697;Fig.2).Theseresultsthuspro- videmixedevidenceregardingtheefficiencyofandrostadienone maskingintheCAgroup.
3.2. Effectsofandrostadienone 3.2.1. Effectsonmood
Asexpected,anoveralldegradationofmoodoccurredduring theexperiment(increaseofnegativemooditemsanddecreaseof positiveandgeneralmooditems).Analysisofthe12moodratings atthebeginningoftheexperiment,ofthe12moodratingsatthe endoftheexperiment,andofthedifferencebetweentheendand thebeginningrevealednoeffectofodorcondition(ps>.804forthe threevariables),noeffectofratersex(i.e.,nodifferencebetween men,non-fertilewomenandfertilewomen:ps>.218),nosignifi- cantinteractionbetweenodorconditionandratersex(ps>.253), andnosignificantinteractionsbetweenodorconditionorratersex andmooditems(ps>.134).Analysisofgeneralmoodratingalso didnotshowanysignificantmaineffectsorinteractionsofthese factors(allps>.055).
3.2.2. Effectsonfirstimpressions
Inthefirst-impressiontask,analysesofresponsetime(thetime takento categorizethestimuli aseither attractiveor unattrac- tive)revealedsignificantinteractionsbetweenodorconditionand ratersexforallstimulustypes(Table1).Theseinteractionsare illustratedin thefirstcolumn ofFig.3 withresultsofthe post hoctests.Theyarecharacterizedbyaslowingeffectofandrosta- dienoneforopposite-sexfacescategorizedbyfertilewomenand forallstimuliexceptsame-sexfacescategorizedbymen.Inversely fornon-fertilewomen,allstimulustypeswerecategorizedmore quicklyinthepresenceofandrostadienonethaninthecontrolcon- dition.Withrespecttothepercentageofraterscategorizingthe stimuliasattractive,therewasnosignificantinteractionbetween odorconditionandratersex(Table1).
Table1
Resultsoftherepeated-measuresANOVAswithodorcondition(C,CA)andratersex(men,non-fertilewomen,fertilewomen)asbetween-subjectsfactors,onthevariables recordedinthefirst-impressiontask(Responsetime:averagelog-transformedresponsetimeperstimulus,andAttractiveness(categ.):numberofraterscategorizingeach stimulusasattractive)andintheratingtask(N=59raters).InteractionsareillustratedinFig.3.
FirstImpressiontask Ratingtask
Responsetime Attractiveness(categ.) Attractiveness Masculinity
Faces F p F p F p F p
Opposite-sex
(1)Odor F(1,87)=2.8 .100 F(1,87)=1.6 .205 F(1,87)=50.9 <.0001 F(1,87)=18.2 <.0001
(2)Ratersex F(2,87)=1.1 .350 F(2,87)=2.8 .068 F(2,87)=0.0 .984 F(2,87)=9.2 <.001
(3)Odor×ratersex F(2,87)=22.8 <.0001 F(2,87)=2.8 .067 F(2,87)=22.8 <.0001 F(2,87)=6.1 <.01 Same-sex
(4)Odor F(1,87)=5.3 <.05 F(1,87)=2.6 .111 F(1,87)=14.5 <.001 F(1,87)=9.8 <.01
(5)Ratersex F(2,87)=1.4 .262 F(2,87)=1.1 .330 F(2,87)=1.9 .158 F(2,87)=17.0 <.0001
(6)Odor×ratersex F(2,87)=6.1 <.01 F(2,87)=1.4 .241 F(2,87)=12.3 <.0001 F(2,87)=15.3 <.0001
FirstImpressiontask Ratingtask
Responsetime Attractiveness(categ.) Attractiveness Masculinity
Voices F p F p F p F p
Opposite-sex
(7)Odor F(1,87)=0.2 .619 F(1,87)=2.6 .148 F(1,87)=8.3 <.01 F(1,87)=1.0 .311
(8)Ratersex F(2,87)=2.1 .130 F(2,87)=2.0 .110 F(2,87)=3.2 <.05 F(2,87)=33.7 <.0001
(9)Odor×ratersex F(2,87)=15.8 <.0001 F(2,87)=0.8 .473 F(2,87)=3.9 <.05 F(2,87)=1.7 .195 Same-sex
(10)Odor F(1,87)=0.1 .736 F(1,87)=0.8 .387 F(1,87)=1.4 .245 F(1,87)=25.3 <.0001
(11)Ratersex F(2,87)=1.1 .324 F(2,87)=2.2 .122 F(2,87)=1.8 .178 F(2,87)=39.3 <.0001
(12)Odor×ratersex F(2,87)=6.2 <.01 F(2,87)=0.7 .490 F(2,87)=2.5 .085 F(2,87)=21.6 <.0001
3.2.3. Effectsonstimulusratings
Intheattractivenessratingtask,thereweresignificantinterac- tionsbetweenodorconditionandratersexforallstimulustypes exceptsame-sexvoices(Table1).Theseeffectsoccurredbecause,in thepresenceofandrostadienone,higherattractivenessjudgments weremadebyfertilewomen(foropposite-sexstimuli),bynon- fertilewomen (forfacesonly)and bymen(opposite-sex voices only),comparedwiththecontrolcondition(Fig.3).
Contrarytoourexpectations,thetypicallymasculinecompound androstadienonedidnotreliablyincreaseperceivedmasculinityof thestimuli(Table1andFig.3).Itdidsoonlyforfacesevaluated bynon-fertilewomen,whichmighthaveaccountedforthehigher perceivedattractivenessofmalefaceswhenthesewomenwere exposedtoandrostadienone.However,inallotherinstances,no reliablelinkcouldbeestablishedbetweentheeffectofandrosta- dienoneonmasculinityanditseffectonattractiveness.
4. Discussion
Theaimofthisstudywastoinvestigatehowandrostadienone influencesothers’attractiveness.Morespecifically,weusedacom- pletedesign(maleandfemaleraters,maleandfemalefacesand voices)todeterminewhetherandrostadienoneinfluencestheper- ceptionofsocial stimuliin asex-specific manner,which would favorpheromonal-likepropertiesofthecompound,orinamore general manner,which would suggest that the compound has broaderinfluencesonhumanpsychologicalresponses.Usingapro- tocolof androstadienonemasking (JacobandMcClintock, 2000) usedbyseveralresearchgroups,wefoundthat,comparedwitha controlcondition,androstadienone:(1)slowedthecategorization ofattractivenessinfertilewomenandinmenbutaccelerateditin non-fertilewomen,and(2)hadapositiveeffectonattractiveness ratingsmadebywomen,butalsoonattractivenessratingsmade bymeninoneinstance(opposite-sexvoices)(seesummary,Fig.4).
Resultsonfacialandvocalstimuliwererelativelyconcordant,and theeffectsof androstadienoneonattractivenessratingsdidnot seemtobemediatedbymodulationofperceivedmasculinityor mood.
Thepositiveeffectofandrostadienoneonperceivedattractive- ness (see Fig. 4) is consistent withresults of previous studies inwomenevaluatingmale(withandrostadienone:Saxtonetal., 2008)orfemalestimuli(withandrostenol:Kirk-Smithetal.,1978).
Ourstudythuspresentsadditionalevidenceforaneffectthathas proveddifficulttoreplicate(Saxtonetal.,2008;Lundströmand Olsson, 2005).Could thiseffect betheresultof a generalposi- tivityofaffectivedispositionstriggeredbyandrostadienone?This questionmakessenseregardingpreviousresearchthatshowsboth moodimprovementafterexposuretothiscompound(Jacobetal., 2002;JacobandMcClintock,2000;Lundströmetal.,2003a)and a favorable impact of good mood on theevaluation of others’
attractiveness (Verhaeghe etal., 2013).However, theansweris
“no”becauseandrostadienonedidnothaveanysignificantinflu- enceonmoodinourstudy.Thiswasprobablyduetotheabsence ofcontextualinfluences:thefemaleexperimenterstayedbehind a one-way mirrorduring the experiment, and stimuli had bal- ancedemotionalvalence:sameversusoppositesex(seeHummer andMcClintock,2009;LundströmandOlsson,2005;Bensafietal., 2004a, for theimportance of thecontext onandrostadienone’s effects).Howthencanthepositiveeffectofandrostadienoneon perceivedattractivenessthatwefoundinourstudybeinterpreted?
In accordancewithKirk-Smithetal.(1978),we foundthatthis effectwasnotsexspecific,which helpsruleoutthehypothesis ofpossiblepheromone-likepropertiesofthiscompound:notonly women,butalsomengavehigherattractivenessratingswhenthey wereexposedtoandrostadienone.Moreover,infertilefemaleand maleraters,thisappliedonlytostimulioftheoppositesex.This findingisinaccordancewiththehypothesisthatandrostadienone influencesresponsestoemotionallysalientstimuli(Hummerand McClintock,2009),i.e.,inthiscase,facesand/orvoicesofindivid- ualsoftheoppositesex.Consistentwiththis,theboostingeffect ofandrostadienoneinnon-fertilefemaleratersappliedtostimuli ofbothsexes,probablybecausethediscrepancybetweentherele- vanceofmaleandfemalestimuliisattenuatedduringthatphaseof themenstrualcycle(aphaseinwhichperceivingtheattractiveness ofmalesmayloseitsadaptivevalue).
2.50 2.75 3.00 3.25 3.50
Ferle women
Non-ferle women
Men
Logged responseme
*** *** *
ResponseTime (log)
2.50 2.75 3.00 3.25 3.50
Ferle women
Non-ferle women
Men
Logged responseme
*** ns ns
2.50 2.75 3.00 3.25 3.50
Ferle women
Non-ferle women
Men
Logged response me
*** *** ns
2.50 2.75 3.00 3.25 3.50
Ferle women
Non-ferle women
Men
Logged responseme
ns * *
Faces Opp.-sex
Faces Same-sex
Voices Opp.-sex
Voices Same-sex
0 20 40 60 80 100
Ferle women
Non-ferle women
Men Percentage of raters categorizing the smuli as aracve
0 20 40 60 80 100
Ferle women
Non-ferle women
Men Percentage of raters categorizing the smuli as aracve
0 20 40 60 80 100
Ferle women
Non-ferle women
Men Percentage of raters categorizing the smuli as aracve
0 20 40 60 80 100
Ferle women
Non-ferle women
Men Percentage of raters categorizing the smuli as aracve
Aracveness (categorizaon)
0 20 40 60 80 100
Ferle women
Non-ferle women
Men
Aracveness
*** *** ns
0 20 40 60 80 100
Ferle women
Non-ferle women
Men
Aracveneess
*** ns ns
0 20 40 60 80 100
Ferle women
Non-ferle women
Men
Aracveness
** ns *
0 20 40 60 80 100
Ferle women
Non-ferle women
Men
Aracveness
Aracveness (rang)
0 20 40 60 80 100
Ferle women
Non-ferle women
Men
Masculinity
ns ***
ns
0 20 40 60 80 100
Ferle women
Non-ferle women
Men
Masculinity
*** ** ***
0 20 40 60 80 100
Ferle women
Non-ferle women
Men
Masculinity
0 20 40 60 80 100
Ferle women
Non-ferle women
Men
Masculinity
*** * ns
Masculinity (rang)
Control (C) Androstadienone (CA)
Fig.3.Averageresponsetime(inseconds,log-transformed)tocategorizethestimuliasattractiveornot,proportionofraterscategorizingthestimuliasattractive(%), attractivenessratings(0–100)andmasculinityratings(0–100)(incolumns)foropposite-sexfaces,same-sexfaces,opposite-sexvoicesandsame-sexvoices(inlines)by fertilewomen,non-fertilewomenandmen(N=59raters).***p<.001,**p<.01,*p<.05,nsnotsignificant:resultsoftheposthocplannedcomparisonsconductedonodor condition×ratersexinteractions(onlywhentheseinteractionsweresignificant).
Using another task, we showed that, in fertile women and men,androstadienoneslowedtheprocessofcategorizing aface ora voiceas attractive ornot. This canbe related toprevious studiesontheroleofandrostadienoneinattentiontoemotional visual stimuli. Indeed, in addition to increasing women’s and men’sfeelingsofbeingfocused(HummerandMcClintock,2009;
LundströmandOlsson,2005), exposuretoandrostadienonewas
foundtoeffectivelyincreaseattentionalcapturebyhumanfaces (butnotbyneutralnon-socialstimuli:LundströmandOlsson,2005;
Parmaetal.,2012).Morespecifically,androstadienoneenhances attentiontoemotionalfacesversusexpressionlessfaces(Hummer and McClintock, 2009), facilitates emotion recognition in faces (Freyetal.,2012)andincreasesfaceviewingtime(Parmaetal., 2012).In ourstudy,responsetime dependedontheamountof
Fig.4. Summaryoftheeffectsofandrostadienoneon(A)categorizationtimeinthefirstimpressiontask,and(B)attractivenessjudgmentsintheratingtask.:increasein thepresenceofandrostadienonecomparedwiththecontrolcondition;decreaseinthepresenceofandrostadienonecomparedwiththecontrolcondition;v:forvoices only;f:forfacesonly.
attentional resources put into theprocessing of the emotional stimulusorattendingtoitandintooutputtingaresponsetothis stimulus(Yiend,2010).Moreover,determiningwhetherafaceis attractiveornotrequiresattentionalresources(Jungetal.,2012).
Accordingtotheliteraturecitedearlier,itislikelythatandrosta- dienone delayed decision making because of greater attention captureby–andthuslongerexaminationtimeof–thepercep- tualcharacteristicsofthefaces/voiceswhentheywereprocessed intheworkingmemory.Thisresultmayalsorelatetothefactthat, intheratingtask,stimuliwereperceivedasbeingmoreattractivein thepresenceofandrostadienone:responsetimesareknowntobe longerformoreattractivefaces(e.g.,Bensafietal.,2002;Kranzand Ishai,2006).Thefactthat,again,mendisplayedthesamebehav- ioralresponsetoexposuretoandrostadienoneasfertilewomendid reinforcesthehypothesisofthegeneraleffectofthismoleculeon attentiontorelevantemotionalinformation,namelystimuliofthe oppositesex,servingdecisionalprocessesinmatechoice.Thesame effectwasfoundformalevoicescategorizedbymaleraters(see Fig.4):ifandrostadienoneisanindicatorofmalequality,assug- gestedelsewhere(HuovialaandRantala,2013),thenitislikelyto increasethelevelofintrasexualcompetitionandthereforethelevel ofvigilance–andthusexamination–ofthecompetitors.Despite thefactthatsimilareffectsrelatedtofemale-femalecompetition havebeensuggestedpreviously (Parmaetal.,2012), theywere notreplicatedhere.Finally,thereverseeffectofandrostadienone occurredinnon-fertilewomen(accelerationofattractivenesscat- egorization):thepresentedstimulihadloweremotionalrelevance thantheydidforfertilewomenandformenbecause–unlikethese groups–non-fertilewomenareatlowriskofconception.Ofthetwo componentsoftheresponsecitedearlier(attendingtothestimulus versusprovidingaresponsetothatstimulus),thefirstonemayhave remainedintactbecauseitislessrelevantfornon-fertilethanfor fertilewomen,whereasthesecondonemayhavebeenfacilitated in accordancewithpreviousobservations that androstadienone increasesaparticipant’sfocus(Lundströmetal.,2003a).
Thepresentstudyaddsevidencetotheexistingliteraturethat androstadienoneinfluenceshumanbehaviorandpresentsnewevi- denceofageneralbutnotsex-specificinfluenceonpsychological responses.However,somelimitationsmustbementioned.First, we didnot elucidate themechanisms governingtheactivityof androstadienone.Toenablecomparisonsbetweenstudies,weused askin-applicationprotocolthatiswidelyusedinandrostadienone studies.Thisproceduredoesnotenlightenusaboutwhicholfac- tory(throughstimulationofthenasalmucosa)orpharmacological (throughitspassageintothebloodstream)modeofaction isin play.Theecologicalvalidityofsuchaprotocolisquestionableas well,sincetransdermaldiffusionmaynotbethepreferentialway that androstadienone actsin day-to-day interpersonalrelation- ships(rather,itmaybeengagedduringclosephysicalcontact).
Second,itisstillunclearhowchangesinattractivenessorinatten- tiontosocialstimuli,asmeasuredinapsychologicalexperiment suchasours,influenceactualsexualbehaviororsocialbehavior betweenodorreleaserandreceiver.Finally,ourstudyhasmethod-
ologicallimitations.Itcanbearguedthatthepresentationorderof opposite-sexandsame-sexstimuli,andofvoicesandfaces,should becounterbalancedratherthatfixed.However,itispossiblethat theconditionsinfluenceeachother(forexample,seeingfacesor not before evaluatingvoices couldmodulate theattractiveness ratingsofthevoices;likewiseforopposite-versussame-sexstim- uli).Wethusbelievethatarandomizedblockdesignmighthave introducedsomenoiseintothedatathatcouldhavemaskedthe effectoftheandrostadienone.Additionally,theuseofabetween- subjectsdesign,chosenbecauseoflogisticalconstraints,isnotthe mostpowerfuldesign,sincewecannotexcludethepossibilitythat thetwogroups(withandwithoutandrostadienone)differedon variablesinfluencingtheirresponsestothetask.Thesepotential differences werelimited, though,because(i) both groups were homogeneousintermsofage,sex,occupation(mostlystudents) andsexualorientation;(ii)allocationtoa groupwasperformed randomly;(iii)theprotocolwasdouble-blind;and(iv)bothgroups equallyrespectedtheinstructionsgiventopreventalterationsof thesenseofsmell(suchasnotusingperfumedcosmeticproducts, notsmoking,etc.).
Apartfromthesereservations,wecanconcludethattogether, our results on response time and attractiveness ratings raise doubts about the proposition that androstadienone has sex- specific,pheromone-likeeffects.Rather,theybetterfitthemodel ofamoregeneraleffectofandrostadienoneonemotionalstimuli, asadvocatedbyHummerandMcClintock(2009),eventhoughwe didnottesttheeffectsonnon-socialstimuli.Indeed,itwasonly whenthefaces/voiceswererelevanttothemate-choicecontext thattheattentionpaidtothemandtheirvalencesweremodulated byandrostadienone:attentioncaptureby,andattractivenessof, opposite-sexfaces/voicesincreasedinperceiverswhotheoretically wereabletoreproduce(i.e.,menandfertilewomen).Non-fertile women reactedvery differentlytoandrostadienone:in particu- lar,androstadienoneinfluencedresponsestobothopposite-and same-sex stimuli, which likely reflectsthat both categories are emotionallyequivalentduringthisphaseofthemenstrualcycle.
Ourconclusionswerealwaysconsistentbetweenmodalities,even thoughtheydidnotalwaysapplytobothfacesandvoices.More- over,thefactthatinourstudymaleraterswerenon-anosmicto androstadienone isimportant. Asmentendtobeless sensitive tothiscompound(Hummeletal.,2005;Lundströmetal.,2003b;
ourstudy), it is likely that a design that includes menregard- lessoftheirabilitytosmellandrostadienonewouldconcludethat androstadienonehasnobehavioraleffectonthem.Thismaygen- erateerroneousinferencesaboutthefunctionofthiscompound,in particularaboutitssexspecificity.Ourdesigndiscreditedsexspeci- ficityinaccordancewiththeresultsofotherstudies(Hummerand McClintock,2009),butthefactthat,inreal-lifeconditions,asub- stantialpartofthepopulationisanosmictoandrostadienone(25%
oftheparticipantsinourstudy)mustalsobetakenintoconsider- ationwhendiscussingtheroleofandrostadienone.Indeed,thisis anotherreasontodoubtthatandrostadienoneactsasapheromone (Wyatt2015).
