Journal ofBehavioralDecision Making J.Behav.Dec.Making,19: 87–98(2006) Published online 26July 2005in WileyInterScience (www.interscience.wiley.com).DOI: 10.1002/bdm.501 The Heat of the Moment: The Effect of Sexual Arousal on Sexual Decision Making DANARIELY1*andGEORGELOEWENSTEIN2 1MassachusettsInstituteofTechnology,USA 2CarnegieMellonUniversity,USA ABSTRACT Despite the social importance of decisions taken in the ‘‘heat of the moment,’’very littleresearchhasexaminedtheeffectofsexualarousalonjudgmentanddecisionmak- ing.Hereweexaminetheeffectofsexualarousal,inducedbyself-stimulation,onjudg- ments and hypothetical decisions made by male college students. Students were assignedtobeineitherastateofsexualarousaloraneutralstateandwereaskedto: (1)indicatehowappealingtheyfindawiderangeofsexualstimuliandactivities,(2) reporttheirwillingnesstoengageinmorallyquestionablebehaviorinordertoobtain sexual gratification, and (3) describe their willingness to engage in unsafe sex when sexuallyaroused.Theresultsshowthatsexualarousalhadastrongimpactonallthree areas of judgment and decision making, demonstrating the importance of situational forcesonpreferences,aswellassubjects’inabilitytopredicttheseinfluencesontheir ownbehavior.Copyright#2005JohnWiley&Sons,Ltd. keywords sexualarousal;situationalinfluences;predictionofpreferences INTRODUCTION Thesexdriveisavitallyimportantmotivationalforceinhumanbehavior,fromtheperspectiveofboththe individualandthesociety.Sexualmotivationplaysadirectroleinconsiderableeconomicactivity,including pornography and prostitution, and a less direct role in diverse industries and activities such as night-time entertainment,advertising,andfashion.Sexualmotivationandbehavioralsounderliesnumeroussocialills, including sexuallytransmitted disease, unwantedpregnancies, andsex-related crimes. Despitetheimportanceofthetopic,mostoftheinformationwehaveabouttheeffectofsexualarousalon judgment, choice, and behavior more generally, comes from personal or vicarious experience. Unlike the extensiveresearchon,forexamplefear(e.g.,LeDoux,1996;Lerner&Keltner,2001;Panksepp,1998),there hasbeenverylittleresearchtracingoutthediverseeffectsofsexualarousalonjudgmentanddecisionmak- ing.Inthispaper,weexaminetheeffectofsexualarousalinyoungmaleadultsonthreeaspectsofjudgment *Correspondenceto:DanAriely,MIT,E56-311,77MassachusettsAvenue,Cambridge,MA02139,USA.E-mail:[email protected] Copyright#2005JohnWiley&Sons,Ltd. 88 Journal ofBehavioral Decision Making and choice: (1) their preferences for a wide range of sexual stimuli and activities, (2) their willingness to engage in morally questionable behaviors in order to obtain sexual gratification, and (3) their willingness toengage inunsafe sexwhen sexuallyaroused. There aregood reasons, beyondintrospectionandcasualempiricism,tosuspect thatsexualarousal will affectthesedimensionsofjudgmentandchoice.Thesexualcircuitryofmenandwomenevolvednotonlyto orchestratesexualbehaviorbutalsotomotivateitinsuitablesituations(Buss,2003;Rolls,1999).Byexo- genously arousing male subjects we are, in effect, parasitizing men’s evolved psychological mechanisms, providing internal and external cues that would ordinarily be associated with increased odds of gaining access towhatBuss and Schmitt (1993) refer to as‘‘short-term opportunistic copulation.’’ Mostappetitivesystemsinthebrain,includinghungerandthirst,aredesignedtoincreasemotivationdur- ing times of opportunity (Rolls, 1999), and there is no reason to expect sex to be an exception to the rule. Whenthebrainreceivescuesthatarecommonlyassociatedwithopportunitiesforcopulation,whichwould includeexperiencingastateofhighsexualarousal,weshouldexpecttoobserveanincreaseinmotivationto havesex.1Thisincreaseinmotivationshould,inturn,havediverseconsequencesforjudgmentsanddecisions. Consistentwithsuchaprediction,priorresearchhasshownthatsexualmotivationcandistortjudgmentsof theriskofcontractingsexuallytransmitteddisease(Blanton&Gerrard,1997;Ditto,Pizarro,Epstein,Jacob- son,&MacDonald,2005),andthatitleadstosteepertimediscountinginmales(Wilson&Daly,2004). Thenextquestioniswhetherindividualscancorrectlyestimatetheeffectsofhighsexualarousalstateson their preferences and behavior. Based on prior research on ‘‘hot-cold empathy gaps’’ (Loewenstein, 1996; Loewenstein, O’Donoghue, & Rabin,2003;Van Boven & Loewenstein, 2003), we anticipated that people whowerenotarousedwouldunderestimatetheinfluenceofemotionalarousalontheirpreferencesanddeci- sions.Previousresearchhasdemonstratedhot-coldempathygapsacrossseveralemotionalstates.Forexam- ple,peoplewhodonotownanobjectunderestimatehowattachedtheywouldbetoitandhowmuchmoney they would require to part with the object if they owned it (Loewenstein & Adler, 1995; Van Boven, Dunning, & Loewenstein, 2000). People who are about to exercise predict they would be less bothered by thirst if they were lost without food or water than do people who have just exercised and are thirsty and warm (Van Boven & Loewenstein, 2003). Peoplewho are sated because they have just eaten are less likelytochooseahigh-caloriesnacktoconsumeatawell-definedtimeinthefuturethanhungrypeoplewho havenoteaten(Read&vanLeeuwen,1998),andpeoplewhoarehungrybecausetheyhavenoteatenexpect tobemoreinterestedineatingaplateofspaghettiforbreakfastthanpeoplewhoaresated(Gilbert,Gill,& Wilson, 2002). Heroin addicts who are not currently craving because they just received a ‘‘maintenance’’ doseofopioidagonist,valuegettinganextradoseaweeklaterabouthalfashighlyasthoseaskedtovalue theextradoseanhourearlier,beforetheyhavereceivedtheirmaintenancedose(Giordanoetal.,2002).And, inthepriorresearchmostobviouslyrelevanttotheresearchreportedhere,menwhoarenotsexuallyaroused predicted they would be less likely to engage in sexually aggressive behavior than men who are sexually aroused as aresultof viewingphotographsof nudewomen (Loewenstein, Nagin, & Paternoster, 1997). Afindingthatsexualarousalaffectspredictionsoftheindividual’sownjudgmentsandbehaviorwouldnot onlysupporttheideathatarousalinfluencesdecisionmaking,butalsosuggestthatpeoplehavelittleinsight intotheseeffects.Ifpeoplewereawareofhowtheirjudgmentsand(hypothetical)decisionswerebeinginflu- encedbytheirownstateofarousalthentheycouldcompensateforsuchinfluencesintheirjudgmentsand decisions. When not aroused, if they appreciated how being aroused would influence their responses and expectedtobearousedinthesituationaskedaboutinthequestion,theycouldadjusttheiranswersaccord- ingly.Thesamelogicwouldapplytothosewhowerearousediftheyfullyappreciatedhowthearousalwas influencing their responses. 1Wemightalsoexpecttoobserveacommensuratedecreaseinotherdimensionsofmotivation(see,Brendl,Markman,&Messner, 2003). Copyright#2005JohnWiley&Sons,Ltd. JournalofBehavioralDecisionMaking,19,87–98(2006) D. Ariely and G.Loewenstein The EffectofSexualArousal onSexualDecision Making 89 Theexperimentpresentedbelowwasdesignedtojointlytestifastateofsexualarousalinfluencesthese threeaspectsofjudgmentandchoice(preferencesforawiderangeofsexualstimuli;willingnesstoengage inmorallyquestionablebehavior;andtheirwillingnesstoengageinunprotectedsex),andwhetherourpar- ticipants can accurately predict theseinfluences. THE STUDY Method Researchparticipantsweregivenalaptopcomputerandwereaskedtoansweraseriesofquestionsusinga small handheld keypad. The keypad and the program that administered the questions were designed to be operatedeasilyusingonlythenon-dominanthand.Inthecontrol(non-aroused)treatment,subjectsanswered thequestionswhileintheirnatural,presumablynothighlyaroused,state.Inthearousaltreatment,subjects werefirstaskedtoself-stimulatethemselves(masturbate),andwerepresentedwiththesamequestionsonly after they had achieveda high butsub-orgasmiclevelof arousal. The screen of the computer, as it appeared in the arousal treatment, was divided into three panels, each with a different function(see Figure1).Three keyson the topleft corner ofthe keypadswitched between thesethreepanels,withtheactivatedpanelindicatedonthescreenbyabrightredborder.Thepanelsonthe Figure1. Anillustrationofthescreenwithanon-representativecartoonimage.Therightpanelandthetopleftpanel wereonlyavailableinthearousalcondition.Thekeypadbelowillustratesthecontrolsoftheinterface Copyright#2005JohnWiley&Sons,Ltd. JournalofBehavioralDecisionMaking,19,87–98(2006) 90 Journal ofBehavioral Decision Making right and on the top left part of the screen were displayed only in the arousal treatment. The panel on the right,whenactivated,displayedan‘‘arousalthermometer’’withregionscoloredfrombluetoredrepresent- ingincreasinglevelsofarousal.Twokeysonthekeypadallowedtheusertomovetheprobeonthearousal metertoindicatetheirmomentarylevelofarousal.Thepanelonthetopleftoccupiedthelargestpartofthe screen, displaying diverse erotic photographs. When this panel was activated, the same two keys used to move the probe on the arousal thermometer allowed the subject to scroll forward and backward through the photographs. The panel at the bottom of the screen presented the series of questions that the subjects answered. This panel was visible in both treatments, but, in the arousal treatment, the questions could be answeredonlywhentheself-declaredlevelofarousalwas75%orhigheronthearousalthermometer.(This criteriawassettocreatehighlevelofarousalbutatthesametimenottoohighinordertoavoidejaculation.) Thesametwokeysusedtomovetheprobeonthearousalthermometer,andtoscrollforwardandbackward throughthephotographs,allowedthesubjecttomoveaprobetoindicatetheiranswerstothequestionspre- sented inthis panel. Adifferent keywas used to submit the answer andprogress tothe nextquestion. Thequestionstooktheformofstatements,whichparticipantscouldreacttoonavisual-analogscalethat stretchedbetween‘‘no’’ontheleftto‘‘possibly’’inthemiddleto‘‘yes’’ontheright.Becausethemovement oftheprobeonthevisual-analogscalewascarriedoutbyrepeatedlypressingkeys,weusedadiscretescale withtwenty-sixstepsalongthevisual-analogscale.Responseswereconvertedtoa0–100scale,where0is the mostextremenegativeresponse and 100 is the most extreme positiveresponse. Thequestionswerepresentedinthreemodules,eachdesignedtoaddressoneoftheissuesmentionedear- lier:theattractivenessofdifferentsexualactivitiestotherespondent,thelengthstherespondentwouldgoto inorder to obtain sexualgratification, and their attitude towardsexualrisks inthe heat of passion. Thesetofquestionsthataskedsubjectstoevaluatetheattractivenessofdifferentsexualstimuliandactiv- itiesincludedquestionsabouttheattractivenessofwomen’sshoes,a12-year-oldgirl,ananimal,a40-,50-, and 60-year-old-woman, a man, an extremely fat person, a hated person, a threesome including a man, a woman who was sweating, cigarette smoke, getting tied up by their sexual partner, tying up their sexual partner, awoman urinating, getting spanked by a woman, spanking a woman, anal sex, contacts with ani- mals, havingsexwith the lights on, and reactions to‘‘just’’ kissing. Thesetofquestionsthataskedsubjectstoassessthelengthstheywouldgotoprocuresexincludedques- tionsaboutwhethertheywouldencourageadatetodrink,slipheradrug,takehertoafancyrestaurantortell hertheylovedher(whentheyinfactdidnot),inallcaseswiththegoalofhavingsex,andalsowhetherthey would try tohave sexevenafter the persontheywere dating said ‘‘no.’’ Theitemsdealingwithsexualrisk-takingelicitedtherespondent’sself-reportedlikelihoodofusingbirth controlandthelikelihoodofnegativeconsequencesifonefailedtodoso.Stillotheritems,notdiscussedin thispaper,askedsubjectstomakeaseriesofhypotheticalintertemporalandriskychoicesformonetaryout- comes.2 Subjects Research participants were 35 University of California, Berkeley male undergraduates recruited with ads placedaroundcampus,whoreceived$10persessioninexchangeforparticipating.Beforetheexperimental sessionsubjectswereinformedabouttheexperiment,includingthefactthatitwouldinvolvemasturbation,3 signed a consent form,and were randomly assigned toone ofthe threeorder-conditions. 2Thegeneralfindingfromtheseitemswasthatarousedsubjectstendedtobemorerisk-seekingandmoreshort-sighted.Wearecurrently followingupthispartofthestudybypresentingarousedandnon-arousedsubjectswithchoicesinvolvingrealconsequences. 3This,andthewithin-subjectnatureoftheexperiment,avoidedtheproblemofdifferentialdropoutbetweenconditions. Copyright#2005JohnWiley&Sons,Ltd. JournalofBehavioralDecisionMaking,19,87–98(2006) D. Ariely and G.Loewenstein The EffectofSexualArousal onSexualDecision Making 91 Table1. Thethreeexperimentalorder-conditions Condition Treatment1 Treatment2 Treatment3 Analysis N(n 11) Non-aroused1 — ¼ AN(n 12) Aroused2 Non-aroused3 2versus3 ¼ NAN(n 12) Non-aroused4 Aroused5 Non-aroused6 5versus(4 6)/2 ¼ þ Design Theexperimentaldesignwasamixedwithin-andbetween-subjectsdesign(asdescribedinTable1).IntheN condition,subjects answeredthequestionsinthenon-arousedtreatment,returnedthelaptop totheexperi- menteronthefollowingday,andwerepaid$10.IntheANcondition,subjectsfirstparticipatedinthearousal treatment;then,whentheyreturnedthelaptopthefollowingday,theywereinformedaboutthesecondses- sionandwereaskedtoparticipateinthenon-arousedtreatment.Aftercompletingthenon-arousedtreatment, thesubjectsreturnedthelaptoptotheexperimenterandwerepaid$20.IntheNANcondition,thesubjects first answered the questions in the non-aroused treatment; then, when they returned the laptop, they were askedtoparticipateinthearousedtreatment.Whentheyreturnedthelaptopagain,theywereaskedtopar- ticipate in another non-aroused session. After completing this second non-aroused treatment subjects returned the laptop to the experimenter and were paid $30. In all cases therewas at least a one-day delay between participation inthe different sessions. Results First, we examined the impact of the arousal treatment on the reported levels of online arousal. Figure 2 presentsthemeanreportedlevelofarousalinthearousaltreatmentasafunctionofthesession’sduration. NotethattherangeofthescaleinFigure2isfrom75to100,sinceparticipantswerenotpermittedtoanswer questions until their self-reported arousal level, as indicated on the arousal thermometer, reached the required level of 75%. As can be seen in Figure 2, reported momentary arousal kept increasing during theexperiment.Itisimportanttonotethatallthesubjectscompletedthesessions,andnoonereportedthat Figure2. Momentaryself-reportedarousalinthearousalcondition.Notethattheabscissaisintermsofquestions,and thatinthiscondition,subjectscouldnotansweranyquestionsuntiltheirarousalwasatleast75 Copyright#2005JohnWiley&Sons,Ltd. JournalofBehavioralDecisionMaking,19,87–98(2006) 92 Journal ofBehavioral Decision Making theyaccidentallyejaculatedduringthesession(subjects were instructedtopressthetab keyiftheyejacu- lated, which would have ended the session). Theexperimentaldesignincludedsomecomparisonsbetween,andsomewithin,thesubjects(seeTable1). Theminimumdelayofonedaybetweentheexperimentalsessionswasdesignedtominimizecontamination effectsofrepeatedexposuretothequestions.Totestwhetherrepeatedexposuretothequestionsinfacthad anyimpact on the responses, three tests were conducted. Inthefirst-ordertestallfournon-arousedtreatmentswerecomparedinatwo-factormixedANOVA,with allthedifferentresponsesbythesamesubject(allthequestions)asawithin-subjectfactor,andthefourtypes of non-aroused treatments as a between-subject factor. We used this model to examine the effect of the different non-aroused treatments controlling for the different questions and for individual subjects effects. This analysis revealed that none of the non-aroused treatments were statistically different from the others [F(3,43) 0.17, p 0.92], that the questions were statistically different from each other [F(32,1376) ¼ ¼ ¼ 60.88, p<0.001], andthat the interaction between them was notsignificant[F(96,1376) 0.65, p 1]. ¼ ¼ Asecond-ordertestfocusedonthecomparisonofthefirstandsecondnon-arousedtreatmentsintheNAN condition.Ifbeingarousedmadesubjectspermanentlyawareoftheeffectofarousalontheirattitudesand behaviors, we would expect a difference between the two non-aroused treatments—running in the same direction as the difference between the aroused and the first non-aroused treatment. This test was a two factorial within ANOVAwith the first and second non-aroused treatments as one repeated factor, and all thedifferentresponsesbythesamesubject(allthequestions)asasecondwithin-subjectfactor.Theanalysis revealedthatthetwonon-arousedtreatmentswerestatisticallyindistinguishablefromeachother[F(1,11) ¼ 0.1,p 0.76],thatthequestionswerestatisticallydifferentfromeachother[F(32,352) 19.53,p<0.001], ¼ ¼ andthattheinteractionbetweenthemwasnotsignificant[F(32,352) 0.91,p 0.61].Thethird-andfinal- ¼ ¼ ordertestcomparedresponsesforthearousaltreatmentwhenitwasfirst(AN)towhenitfollowedthenon- arousedtreatment(NAN).Thesetwoarousaltreatmentswerecomparedinatwo-factormixedANOVA,with all the different responses by the same subject (all the questions) as a within-subject factor, and the two order-conditions as a between-subject factor. This analysis revealed that the two arousal treatments were statisticallyindistinguishablefromeachother[F(1,22) 0.50,p 0.49],thatthequestionswerestatistically ¼ ¼ different from each other [F(32,704) 24.78, p<0.001], and that the interaction between them was not ¼ significant[F(32,704) 0.69,p 0.90].Overalltheseresultssuggestthatwhilethereweresystematicdif- ¼ ¼ ferencesbetweenthedifferentquestions(whichissomewhatobviousgiventhedifferentnatureoftheques- tions),therewerenosystematicordereffectsforeitherthenon-arousedorthearousedtreatments.Moreover, therewerenointeractionsbetweentheorderandthetypeofquestion—indicatingthatitisnotthecasethat the order of the treatments systematically influenced the responses for somequestions butnotothers. Based on this lack of order effects, and for simplicity of presentation, we present the main results as a within-subjects ANOVA, focusing on the comparison between the aroused and non-aroused treatments. ForthisanalysistheNconditionwaseliminatedfromtheanalysis(sinceithasnowithinsubjectscompar- ison),andtheresponsesofeachsubjectinthetwonon-arousedtreatmentsintheNANconditionwereaver- aged,resultinginawithin-subjectcomparisonacrossthearousalandnon-arousedconditionsforeachofthe questions(see lastcolumn ofTable 1). Thetaskincluded20differentitemsthatelicitedsubjects’preferencesforsexualstimuliandactivities.To examinetheeffectofthestateofarousalontheseitems,weanalyzedthemwitha20(questions) 2(stateof # arousal) fully within ANOVA. The overall model revealed a significant main effect for the arousal state [F(1,23) 53.81, p<0.001], an overall significant main effect for the questions factor ¼ [F(19,437) 43.31, p<0.001], and an overall significant interaction between arousal state and the ques- ¼ tionsfactor[F(19,437) 1.98,p 0.009].Theseresultssuggestthatthatthearousalstatehadasignificant ¼ ¼ effectontheresponsestothedifferentquestions,thatthequestionswerenotallthesame,andthatthearousal statehadadifferentialeffectonthedifferentquestions.Therefore,wefollowedtheseanalyseswithasetof 20independentpairedt-testsforeachofthequestions.AscanbeseeninTable2,subjectsfoundavarietyof Copyright#2005JohnWiley&Sons,Ltd. JournalofBehavioralDecisionMaking,19,87–98(2006) D. Ariely and G.Loewenstein The EffectofSexualArousal onSexualDecision Making 93 Table2. Meanresponse,standarddeviations,anddifferencesforthedifferentquestionsontheattractivenessofdifferent activities Question Non-aroused Aroused Difference p[t(23)] Arewomen’sshoeserotic? 42(5.9) 65(4.06) 23 <0.001 Canyouimaginebeingattractedtoa12-year-oldgirl? 23(4.11) 46(6.08) 23 <0.001 Canyouimaginehavingsexwitha40-year-oldwoman? 58(3.32) 77(2.07) 19 <0.001 Canyouimaginehavingsexwitha50-year-oldwoman? 28(4.80) 55(4.69) 27 <0.001 Canyouimaginehavingsexwitha60-year-oldwoman? 7(2.55) 23(4.61) 16 <0.001 Canyouimaginehavingsexwithaman? 8(2.47) 14(3.78) 6 0.14(ns) ¼ Coulditbefuntohavesexwithsomeonewho 13(4.30) 24(5.29) 11 <0.05 wasextremelyfat? Couldyouenjoyhavingsexwithsomeoneyouhated? 53(6.04) 77(3.59) 24 <0.001 Ifyouwereattractedtoawomanandsheproposed 19(4.97) 34(7.10) 25 <0.005 athreesomewithaman,wouldyoudoit? Isawomansexywhenshe’ssweating? 56(3.1) 72(5.62) 16 <0.01 Isthesmellofcigarettesmokearousing? 13(3.88) 22(6.00) 9 <0.03 Woulditbefuntogettiedupbyyoursexualpartner? 63(5.09) 81(4.49) 18 <0.005 Woulditbefuntotieupyoursexualpartner? 47(3.22) 75(3.89) 28 <0.001 Woulditbefuntowatchanattractivewomanurinating? 25(5.57) 32(5.53) 7 <0.03 Wouldyoufinditexcitingtospankyoursexualpartner? 61(5.35) 72(4.70) 11 <0.1 Wouldyoufinditexcitingtogetspankedbyan 50(3.40) 68(5.29) 18 <0.003 attractivewoman? Wouldyoufinditexcitingtohaveanalsex? 46(4.91) 77(3.58) 31 <0.001 Canyouimaginegettingsexuallyexcitedby 6(2.55) 16(4.19) 10 <0.02 contactwithananimal? Doyouprefertohavesexwiththelighton? 52(5.84) 50(5.15) 2 0.46(ns) $ ¼ Isjustkissingfrustrating? 41(4.43) 69(4.37) 28 <0.001 Note:Eachquestionwaspresentedonavisual-analogscalethatstretchedbetween‘‘no’’ontheleft(0)to‘‘possibly’’inthemiddle(50) to‘‘yes’’ontheright(100). potentialsexualactivitiestobemoreattractiveunderhigharousalthantheydidunderlowarousal.Fromthe 20questionsofthistypeonlyoneactivity(Doyouprefertohavesexwiththelighton?)wasviewedasless appealingbythoseinthearousalcondition thanbythoseinthenon-arousedcondition,thoughnotsignifi- cantlyso(p 0.46).Oneactivity(Canyouimaginehavingsexwithaman?)wasviewedasmoreappealing, ¼ butnotsignificantlyso(p 0.14)bythosewhowerearoused,andonequestionwasonlymarginallysignif- ¼ icant(Wouldyoufinditexcitingtospankyoursexualpartner?p 0.1).Theremaining16questionswereall ¼ significantlydifferentinthepredicteddirection.Aone-samplesigntestoverall20itemsrevealedahighly significanteffect(p<0.001),indicatingthat,overall,arousalfundamentallyincreasespredictedenjoyment and liking for the diverse activities that we queried subjects about. Fiveotherquestionsaskedsubjectsabouttheirwillingnesstoengageinmorallyquestionablebehaviorto procuresex.Toexaminetheeffectofthestateofarousalontheseitems,weanalyzedthemwitha5(ques- tions) 2 (state of arousal) fully within ANOVA. The overall model revealed a significant main effect for # arousal state [F(1,23) 26.40, p<0.001], an overall significant main effect for the questions factor ¼ [F(4,92) 55.70,p<0.001],andanoverallnon-significantinteractionbetweenarousalstateandtheques- ¼ tionsfactor[F(4,92) 0.41,p 0.8].Theseresultssuggestthatthearousalstatehadsignificanteffectonthe ¼ ¼ responsestothedifferentquestions,thatthequestionswerenotallthesame,butthatthearousalstatehada similareffectonthedifferentquestions.Toexaminetheseeffectsinmoredetail,wefollowedtheseanalyses withasetoffiveindependentpairedt-testsforeachofthequestions.AscanbeseeninTable3,arousalhada strong influence on participants’ self-reported likelihood of engaging in a set of morally questionable Copyright#2005JohnWiley&Sons,Ltd. JournalofBehavioralDecisionMaking,19,87–98(2006) 94 Journal ofBehavioral Decision Making Table3. Meanresponse,standarddeviations,anddifferencesforthedifferentquestionsonthelikelihoodtoengagein immoral‘‘date-rape’’likebehaviors(astrictorderofseverityisnotimplied) Question Non-aroused Aroused Difference p[t(23)] Wouldyoutakeadatetoafancyrestauranttoincreaseyour 55(5.86) 70(3.83) 15 0.01 chanceofhavingsexwithher? Wouldyoutellawomanthatyoulovedhertoincreasethe 30(5.40) 51(4.54) 21 0.001 chancethatshewouldhavesexwithyou? Wouldyouencourageyourdatetodrinktoincreasethechance 46(5.80) 63(2.87) 17 <0.005 thatshewouldhavesexwithyou? Wouldyoukeeptryingtohavesexafteryourdatesays‘‘no.’’ 20(4.32) 45(3.44) 25 <0.001 Wouldyouslipawomanadrugtoincreasethechancethat 5(2.51) 26(3.65) 21 <0.001 shewouldhavesexwithyou? Note:Eachquestionwaspresentedonavisual-analogscalethatstretchedbetween‘‘no’’ontheleft(0)to‘‘possibly’’inthemiddle(50) to‘‘yes’’ontheright(100). behaviors to increase the likelihood of obtaining sex. In this set of questions, all five individual questions were statisticallysignificantinthe expecteddirection. Finally, the survey included eight questions that elicited subjects’ willingness to engage in risky sexual practices.Toexaminetheeffectofarousalontheseitems,weanalyzedthemwithan8(questions) 2(state # of arousal) fully within ANOVA. The overall model revealed a non-significant main effect for the arousal state[F(1,23) 0.05,p 0.82],anoverallsignificantmaineffectforthequestionsfactor[F(7,161) 47.6, ¼ ¼ ¼ p<0.001], and an overall significant interaction between arousal state and the questions factor [F(7,161) 5.90, p 0.009]. These results suggest that the arousal state did not have a significant main ¼ ¼ effectontheresponsestothedifferentquestions,thatthequestionswerenotallthesame,andthatthearousal statehadadifferentialeffectonthedifferentquestions.Wefollowedtheseanalyseswithasetofeightinde- pendentpairedt-testsforeachofthequestions.AscanbeseeninTable4,theresultsforthesequestionsare notasclear-cutastheothertwosetsofquestions.Therewerenodifferencesbetweenthearousedandnon- arousedtreatmentsforthefourquestionsdealingwiththeperceivedefficacyofcoitus-interruptus,acknowl- edgmentthatafriendcantransmitSTDs,trustingsomeonetheyhavejustmet,andassigningresponsibility Table4. Meanresponse,standarddeviations,anddifferencesforthedifferentquestionsonwillingnesstoengagein,and outcomesof,sexuallyunsafebehaviours Question Non-aroused Aroused Difference p[t(23)] Ifyoupulloutbeforeyouejaculate,awomancanstill 92(2.95) 92(4.55) 0 0.97(ns) ¼ getpregnant(N) Awomanwhoisagoodfriendcangiveyouasexually 86(4.44) 89(4.89) 3 0.21(ns) $ ¼ transmitteddisease(N) Wouldyoutrustawomanyou’vejustmetwhosayssheis 29(5.14) 25(5.51) 4 0.47(ns) $ ¼ usingbirthcontrol?(A) Birthcontrolisthewoman’sresponsibility(A) 34(5.41) 44(5.69) 10 0.051 ¼ Acondomdecreasessexualpleasure(A) 66(3.88) 78(3.73) 12 <0.002 Acondominterfereswithsexualspontaneity(A) 58(5.21) 73(4.45) 15 <0.01 Wouldyoualwaysuseacondomifyoudidn’tknowthesexual 88(2.66) 69(6.33) 19 <0.003 historyofanewsexualpartner?(N) Wouldyouuseacondomevenifyouwereafraidthatawoman 86(3.04) 60(4.32) 26 <0.001 mightchangehermindwhileyouwenttogetit?(N) Note:Eachquestionwaspresentedonavisual-analogscalethatstretchedbetween‘‘no’’ontheleft(0)to‘‘possibly’’inthemiddle(50) to‘‘yes’’ontheright(100). TheexpectedhigherresponseismarkedwithAforarousedorNfornon-aroused. Copyright#2005JohnWiley&Sons,Ltd. JournalofBehavioralDecisionMaking,19,87–98(2006) D. Ariely and G.Loewenstein The EffectofSexualArousal onSexualDecision Making 95 for birth control towomen (which was marginally significant). Answers to all four questions dealing with condomswere,however,stronglyinfluencedbysexualarousal.Forallfourcondom-relatedquestions,sub- jectsinthearousedtreatmentindicatedalowerlikelihoodofusingacondomcomparedwithsubjectsinthe non-arousedtreatment.Thisdifferencebetweenthetwotypesofquestionsmightsuggestthatarousaldoes not change the general knowledge of individuals about the risks of unprotected sex, but when it comes to concrete steps involvingcondoms,sexualarousal changesone’s perceptions ofthe tradeoffs betweenben- efitsanddisadvantages ina fashion that decreases the tendencytouse them. GENERAL DISCUSSION Thisstudyexaminedtheeffectofhighlevelsofsexualarousalonthesubjectiveattractivenessofdifferent activities, on self-reported willingness to take various morally dubious measures to procure sex, and on willingnesstoengageinriskysexualactivities.Ourresultsonattractivenessofactivitiessuggestthatsexual arousalactsasanamplifierofsorts.Activitiesthatarenotperceivedasarousingwhenyoungmalesarenot sexuallyarousedbecomesexuallychargedandattractivewhentheyare,andthoseactivitiesthatareattrac- tiveevenwhennotaroused,becomemoreattractiveundertheinfluenceofarousal.Byshowingthat,when aroused, the same individual will find a much wider range of activities sexually appealing than when not aroused,thesefindingsweighinagainsttheviewofsexualpreferencesasbeingpurelyanindividualdiffer- encevariable—i.e.,asdispositionallyratherthansituationallydetermined.Certainly,therearerobustindi- vidualdifferencesinsexualpreferencesandinthelikelihoodofengaginginvariousbehaviors,buttherealso seem to be striking intra-individual differences caused, in our study, by externally caused variations in arousal level. Ourresultsfurthersuggestthatthechangeinattractivenessinfluencestheintensityofmotivationtohave sex relativeto othergoals. Specifically, the increase in motivation to have sex produced by sexual arousal seemstodecreasetherelativeimportanceofotherconsiderationssuchasbehavingethicallytowardapoten- tial sexual partner or protecting oneself against unwanted pregnancy or sexually transmitted disease. Like otherdrive-states(Loewenstein,1996),andalsosomewhatanalogoustotheeffectsofalcohol(Dittoetal., 2005;Steele &Josephs,1990), sexualarousal seems tonarrow the focus ofmotivation, creating akind of tunnel-visionwheregoalsotherthansexualfulfilmentbecomeeclipsedbythemotivationtohavesex(c.f., Blanton& Gerrard, 1997). As noted in the introduction, a secondary implication of our findings is that people seem to have only limitedinsightintotheimpactofsexualarousalontheirownjudgmentsandbehavior.Suchanunder-appre- ciationcould beimportant for both individual and societal decisionmaking. At the individual level, there is aconsiderable research showing that one’s meta-understanding of one’s ownpreferencescaninmanysituationsbealmostasimportantasthepreferencesthemselves.Forexample, asO’DonoghueandRabin(2003)show,theimpactofhyperbolictimediscountingonactualintertemporal choicebehaviordependscriticallyonwhetheroneisna¨ıveorsophisticatedaboutthefactthatonewillface self-control problems in the future. Ariely and Wertenbroch (2002) likewise found, in a study of students taking a class, that those who were aware of their own tendency to procrastinate, and hence voluntarily set deadlines for themselves, got higher coursegrades than thosewho did not. Self-insight when it comes tosexualarousalandsexualbehaviorissimilarlylikelytobeimportantfordecisionmaking.Forexample, themosteffectivemeansofself-controlisprobablynotwillpower(whichhasbeenshowntobeoflimited efficacy),butratheravoidingsituationsinwhich onewillbecomearousedandlosecontrol.Anyfailureto appreciate the impact of sexual arousal on one’s own behavior is likely to lead to inadequate measures to avoidsuchsituations.Similarly,ifpeopleunder-appreciatetheirownlikelihoodofhavingsex,theyarelikely to fail to take precautions to limit the potential damage from such encounters. A teenager who embraces ‘‘just say no,’’ for example, may feel it unnecessary to bring a condom on a date, thus greatly increasing Copyright#2005JohnWiley&Sons,Ltd. JournalofBehavioralDecisionMaking,19,87–98(2006) 96 Journal ofBehavioral Decision Making thelikelihoodofpregnancyortransmissionofSTDs ifhe/she endsupgettingcaughtupintheheatofthe moment. Thesamelogicappliesinterpersonally.Ifpeoplejudgeothers’likelybehaviorbasedonobservingthem whentheyarenotsexuallyaroused,andfailtoappreciatetheimpactofsexualarousal,thentheyarelikelyto becaughtbysurprisebytheother’sbehaviorwhenaroused.Suchapatterncouldeasilycontributetodate- rape.Indeed,itcancreatetheperversesituationinwhichpeoplewhoaretheleastattractedtotheirdatesare mostlikelytoexperiencedate-rapebecausebeingunarousedthemselvestheycompletelyfailtounderstand orpredict the other (aroused)person’s behavior. Atasociallevelthefailuretoappreciatetheinfluenceofsexualarousalwhenoneisunarousedcanhave diverseconsequences.Forexample,judgesandjurors,whoaregenerallyunarousedwhenmakingdecisions of guilt and punishment, may be excessively condemnatory and punitive toward sexual offenders because they make their decisions in a sexually unaroused state and fail to appreciate how intense sexual arousal would alter eventheir own decision making in potentially compromising circumstances. The result is that decisionswillbestigmatizedasimmoralmisbehaviorevenbypeoplewhowouldthemselvesmakethesame choice when in an aroused state. It should be clear that such effects of arousal cannot justify any sexual exploitation, but they can make such behaviors somewhat more understandable. From the perspective of thelegalsystemitispossiblethatsexualarousalshouldbegivenmorecreditasapartiallymitigatingfactor thanitwouldnormallyreceive.Moreover,understandingtheseeffectscanhelpguideindividuals(sexoffen- dersforexample)suchthattheywillbelesslikelytosexuallyexploitorre-exploit.Finally,asalludedtoin thediscussionofindividualdecisionmaking, thefailure toappreciatesexualarousalbythosewhoarenot themselvesimmediatelyarousedcanalsohelptoexplaintheenactmentofmisguidedandineffectivepolicies suchas‘‘justsayno’’,leavingyoungadultsunpreparedtolimitthepotentialdamagefromtheirownbeha- vior inthe heat of the moment. Limitations Asaninitialinvestigationintotheeffectofsexualarousalonjudgmentanddecisionmaking,ourstudyinevi- tably suffers from serious limitations. For example, it is important to note that we did not observe actual behavior. It is therefore possible that the effect of sexual arousal was not to change the desirability of dif- ferentactionsandactivities,buttomakerespondentsmorewillingtoadmittotheirfeelings.Ifthiswerethe case,however,weshouldexpecttoseeastrongereffect foritemsthatpeopleareembarrassed about(e.g., findinga12-year-oldgirlattractive,orbeingexcitedbyanimals),buttheeffectswerefairlysimilaracross thesetypesofitemsandthosethatwereunlikelytodrawmuchshame(e.g.,beingattractedtoa40-year-old). A second limitation incumbent in not observing actual behavior is that we have no way to ascertain whether respondents’ predictions of their own behaviors are more accurate when subjects respond under treatments of arousal or non-arousal. Based on previous research on hot-cold empathy gaps (Bouffard, 2002;Loewenstein,Nagin,&Paternoster,1997),whichshowsthatpeopleoftenmispredicthowtheywould behaveinanaffectivestatedifferentfromtheonetheyarein,wesuspectthatbehavioralpredictionsmade understatesofarousalmoreaccuratelypredictbehaviorintheheatofthemomentthandopredictionsmade when respondents are not aroused. However, without observing actual behavior in the situations we ask about, we haveno ability toascertain whetherthis is infact the case. Athirdlimitationconcernsthelackofcontrolthatwehadovertheexperimentalsetting.Wehadsubjects conducttheexperimentintheprivacyoftheirownresidencesoastoprovideprivacyandreduceinhibitions, but this limited our ability to ensure that they carefully and conscientiously carried out the instructions. Indeed,ourinitialexpectationswerethatatleastsomeofthesubjectsinthearousalconditionwouldindicate bypressingthetabkeythattheyhadaccidentallyejaculated,butnonedid.Finally,ourexperimentalsetup did not allow us to measure subjects’ arousal using physiological methods, so we instead relied on self- reports of arousal, which have been shown to be fallible (Janssen, 2002). However, since we manipulated Copyright#2005JohnWiley&Sons,Ltd. JournalofBehavioralDecisionMaking,19,87–98(2006)