Evolution and Human Behavior 21 (2000) 371±390
Non-verbal behavior as courtship signals: the role of control and choice in selecting partners Karl Grammer a,*, Kirsten Kruck b, Astrid Juette a , Bernhard Fink a a
Ludwig-Boltzmann Institut fu Anthropologie, gie, Althanstrasse Althanstrasse 14, È È r Stadtethologie c/o Institut fuÈ È r Anthropolo Vienna A-1090, Austria b Humanethologisches Filmarchiv der Max-Planck Gesellschaft, Von-der-Tannstrasse 3-5, Andechs D-82346, Germany
Received 10 December 1997
Abstract
In thi thiss work, work, we provid providee evide evidence nce based based on direct direct obser observat vation ion of behavi behavior or in encoun encounter terss of opposite-sexed strangers, that women initiate and ``control'' the outcome. In the first minute of these videotap videotaped ed 10-min 10-min interactio interactions, ns, neither neither female female ``solicita `solicitation' tion''' behavior behavior nor ``negative `negative''' behavior behavior is strongly related to professed interest in the man, while female ``affirmative'' behavior at this stage modulates male verbal output in later stages (4±10 min). Although the rate of female courtship-like behavior is significantly higher in the first minute, it is only in the fourth to tenth minute that the rate of female courtship-like behavior is correlated with professed female interest. We hypothesize that this serves as a strategic dynamic reflecting sexual asymmetry in parental investment and the potential cost of male deception to women. Ambiguous protean behavioral strategies veil individuals' intentions and make their future actions unpredictable. These behavioral strategies may result in men's overestimation of female sexual interest. D 2000 Elsevier Science Inc. All rights reserved. Mate selecti selection; on; Initia Initiatio tion n of sexual sexual encoun encounter ters; s; Courts Courtship hip;; Decepti Deception; on; Affirm Affirmati ative ve behavi behavior; or; Keywords: Mate Solicitation; Negative behavior; Interest; Protean behavior
According According to many social non-evolutionary non-evolutionary views, the sexes are identical identical in their behavior, behavior, and and if sex differ differenc ences es in behav behavior ior occu occur, r, they they are are a resul resultt of cultur culturee-sp speci ecifi ficc learn learning ing processes. These learning processes lead further to formation of gender stereotypes. In our culture, socially constructed gender stereotypes allegedly require men to take the initiative in * Corresponding author. E-mail address:
[email protected] (K. Grammer). 1090-5138/00/$ ± see front matter D 2000 Elsevier Science Inc. All rights reserved. PII: S 1 0 9 0 - 5 1 3 8 ( 0 0 ) 0 0 0 5 3 - 2
372
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
initial mix initial mixeded-sex sex encoun encounter terss more more often often than women women (Bruch (Bruch,, Giorda Giordano, no, & Pearl, Pearl, 198 1986; 6; Garcia, Stinson, Ickes, & Bissonnette, 1991). If one follows the logic of McCormick and Esser Esser (198 (1983) 3),, e.g., e.g., on onee find findss that that gende genderr relat relation ionss show show an asym asymmet metri ricc distr distribu ibutio tion n of ``control behavior'' (i.e., who initiates and determines the outcome of opposite sex interactions), which is almost completely norm-governed. According to this view, male-dominated societies allow men to use power in order to obtain sex, whereas women are only allowed to use power in order to avoid sex with unsuitable partners. Evolutionists have developed hypotheses that can be used to derive predictions about possible sex differences on the basis of asymmetric investment theory. Women having the higher investment in their offspring (Trivers, 1972) have more to lose from making a poor mate choice than men. Evolved female psychology is expected to include mechanisms that could motivate active female choice and strategic control of interactions with male strangers, thus leading towards the most favourable outcome for the female actors. This does not imply that men are not cautious in mate selection and that they do not choose actively or try to control female approaches. However, However, because of the asymmetric investment, investment, women should engage in control activities more often than men and should manipulate men relatively more often and with greater subtlety than men would manipulate women. Thus, we would expect a relative difference between male and female ``control'' behavior in friendly encounters with strangers of the opposite sex.
1. Non-verbal signals in human courtship
In recent years, a repertoire of courtship signals has been established and empirically validated (Grammer, 1990; Kendon & Ferber, 1973; Moore, 1985, 1995; Moore & Butler, 1989 19 89;; Schef Scheflen len,, 19 1965 65). ). Altho Althoug ugh h these these auth author orss used used diff differ erent ent settin settings gs (sin (single gle''s bars, bars, laboratory, and client±therapist interactions), they all observed similarities in the repertoire of female female behavio behaviors rs in the presenc presencee of male strange strangers, rs, includin including g ``come`come-onon-eff effects ects''' of these signals. According to Moore's (1985) observations in single's bars, a woman is able to affect male approaches by exhibiting or withholding non-verbal displays: ``they can elicit a high number of male approaches allowing them to choose from a number of available men, or they may direct solicitations at a particular man''. In this work, a courtship repertoire was established from from vari variati ation on in the the freq frequen uenci cies es of diff differ eren entt behav behavio iors rs of women women in sin singl gle' e'ss bars bars vs. vs. cafeterias, and an index of this repertoire was positively correlated with male approaches in a single's bar. Moore (1985) also showed that female interest could be assessed if a threshold number of behaviors occurred out of a catalogue of 52 behaviors, but there was no direct prediction of female interest in a man. Further Furthermor more, e, Grammer Grammer,, Honda, Honda, Juette, Juette, and Schmitt Schmitt (1999) (1999)1 show showed ed that that in the the very very beginning (the first 120 s) of an interaction, traditional ethological behavior analysis was
1
This study used the first 2 min from the same experiments experiments which are analyzed analyzed in this article. article. The study used different observation categories and methods. The data in this study were analyzed by K. Kruck.
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
373
not able to reveal behaviors related to professed female or male interest in Japanese and in German couples. In this analysis, it was only possible to show, with the help of digital image processing of body movements, that qualitative changes in body movements could predict professed female interest but only to a limited extent. These qualitative changes of movement were detected in the speed, size and number of movements. The analysis also show showed ed that that the the chan changes ges were were inde indepen penden dentt to the the cont content ent of behav behavio iorr. The The analy analysi siss did did not assess possible effects of ``control'' behavior and its effects on subsequent behavior in later stages of the encounter. In the folk psychology literature on the use of signals in courtship, we find a striking sex difference, which reflects the general relative difference in the mate selection psychology of the two sexes. Women seem to be exquisitely familiar with what occurs during courtship. They can describe in detail how they and other women flirt and pick up men. Even quite successful men seem to have no idea what happens during flirtation (Perper & Fox, 1980). ``I just know it works out'' was an answer often recorded by Kirkendall (1961) in his interviews with young men. This might be a result of covert female control strategies. Indeed, women are much better then men in encoding and decoding non-verbal behavioral cues. From the earliest childhood, females exhibit more and more expressive, non-verbal beh behav avio iorr than than males males do (e.g. (e.g.,, facia faciall expr express essio ions ns and touc touchi hing ng;; Argy Argyle, le, 19 1988 88;; Henl Henley ey & LaFrance, 1984). To quote DePaulo's (1992) review: ``Differences in ability, motivation, and spontaneous expressiveness all converge to produce what may be one of the most pervasive and and im impo port rtant ant of all all indi individ vidual ual diff differ eren ences ces in the use use of no nonn-ve verb rbal al beha behavi vior or for for selfself presentational purposes: Sex differencesF F F [Women's] body movements are more involved and more expressive'' (pp. 222±223). If this is the case, it seems reasonable to look for other strategies and ways in which meaning is created in non-verbal exchanges. Grammer, Kruck, and Magnusson (1998)2 found evidence of female manipulation of the man in a non-intrusive and non-obvious way. If a woman is interested in a man, female rhythmic body movements create ``hidden'' and highly complex patterns in synchrony with the male body movements. The man perceives changes in her interest but is not able to ascertain their source. The puzzling finding in this work was that female ``courtship signals'' were not predictive of the particular pattern of behavior observed.
2. Control behavior, deception, and proteanism
In sum, the logic of this line of reasoning about asymmetric investment, women's need to avoid male deception, and women's non-verbal skills, leads to the prediction that women are expected to control men without the men recognizing the manipulation directly. We predict that that in an initi initial al enco encoun unter ter,, when when the the go goals als are un unkn know own n (and (and po possi ssibl bly y conf conflic lictin ting) g),, the possibility of deception could play a major role in the interactions of both sexes. If female parental investment and any cue of the possible adaptive loss resulting from bad decision
2
See above.
374
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
making are higher for women than for men, then the possibility possibility of male deception deception should play a more important strategic role for women than for men. Male Male dece decept ptio ion n abou aboutt inte intent ntio ions ns or male male qu qual alit ity y may may be the the resu result lt of male± male± male male comp competi etiti tion on and and female female mate mate select selection ion stand standar ards ds.. The The conte content nt of male male adve advert rtise iseme ment nt is controlled through female and male perception of an optimal mate. Indeed, men report that they they attemp attemptt to decei deceive ve women women with with status status symb symbol olss (Bus (Buss, s, 19 1992 92), ), wher whereas eas women women use use deception to enhance their physical appearance. Men also use deception more often when they try to attain a short-term sexual relation. These low-investment copulations have been cited more often among men than among women. Men test continuously for opportunities to reduce investment and pursue women (Landolt, Lalumie re, & Quinsey, 1995; Townsend, Kline Kline,, & Wasser asserman mann, n, 19 1995 95). ). When When compa comparin ring g them them again against st the the back backgro groun und d of male male decepti deceptive ve mating mating tactics tactics,, women women sho should uld have have evolve evolved d psy psycho cholog logical ical mechan mechanism ismss that that motivate manipulation of men to reveal information about their intentions (what relationship is he seeking if any Ð short or long term?). This evaluation should take place without male awareness of female evaluation along this dimension (in which case male deceptive efforts might be intensified). A consequence of this strategic female influence is that men should be less aware of being manipulated for the purpose of revealing information about their mating intentions and interests and, indeed believe that it was them, rather than the woman, who initiated and guided the interaction. Because of these conflicting demands Ð the utility of advertising and of deception Ð two strangers who meet find themselves in a delicate situation. On the one hand, interest in the other person has to be communicated in order to signal approachability, while on the other hand, overt signalling could induce and raise the likelihood of deception. Yet, an ``overt'' approach might be dangerous for the woman in more respects. The approaching man could start to insist on interacting with the woman after she has decided that the information from the man did not suit her aspirations in mate selection; hence, this information can only be transmitted verbally. The woman could run into the problem of getting rid of this man again, or even worse, the man who received positive ``come ons'' could react aggressively to a subsequent subsequent rejection. rejection. The main challenge for communication in such a situation is that as soon as the man perceives some female interest, his tendencies for deception might rise. The strategic solution for the woman would be to influence the man to the extent of his not consciously processing the come-on and thus believe that he himself initiated the approach. Sabi Sabini ni and and Silv Silver er (198 (1982) 2) po post stul ulat ated ed that that the the esse essenc ncee of init initia iall op oppo posi site te-s -sex exed ed enco encoun unter terss is the the creat creation ion of ambig ambigui uity ty,, wher wheree bo both th sexes sexes tend tend to hide hide their their go goals als and and reve reveal al info inform rmat atio ion n abou aboutt po poss ssib ible le comm commit itme ment nt slow slowly ly.. Such Such a situ situat atio ion n enta entail ilss the the ``com `commun munica icatio tion n parado paradox' x''' (Gra (Gramme mmerr, Fieder Fieder,, & Filov Filova, a, 19 1997 97)) where where inten intentio tions ns have have to be comm commun unic icat ated ed with withou outt actu actual ally ly reve reveal alin ing g thes thesee inte intent ntio ions ns.. The The effe effect ctss of such such sublimi sub liminal nal contro controll strateg strategies ies could could lead to effects effects which which have have been describ described ed repeate repeatedly dly for for male±fem male±femal alee inte intera ract ctio ion ns. A nu numb mber er of stud studie iess have ave sho shown that that men men perc percei eive ve wome women n in a mo more re sexu sexual aliz ized ed way way than than vice vice vers versaa (Abb (Abbey ey,, 19 1982 82;; Abbe Abbey y, Cozz Cozzar arel elli li,, McLaughlin, & Harnish, 1987; Edmonson & Conger, 1985). In their terms, men usually femalee frien friendl dlin iness ess and and ov over er-r -rate ate femal femalee sexual sexual int inter erest est,, wher whereas eas wome women n misconceive femal tend tend to un unde dere rest stim imat atee male male sexu sexual al inte intere rest st.. This This rese resear arch ch is usua usuall lly y do done ne with with qu ques es--
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
375
tionnair tionn aires es and and ratin rating g scales scales or by inter intervi viewi ewing ng subje subjects cts who who have have spen spentt a certa certain in tim timee togethe tog etherr. Until Until tod today ay,, detaile detailed d analyse analysess of behavio behaviorr of interact interactions ions of ini initial tial encounte encounters rs have have no nott been been prov provide ided. d. The existence of such situations naturally leads to the development of ``mindreading'' with individ ind ividual ualss conceal concealing ing their their intenti intentions ons and deceivi deceiving ng actively actively (Krebs (Krebs & Dawkin Dawkins, s, 198 1984). 4). Miller Mill er (1997) (1997) identif identifies ies three three pos possibl siblee counter counter strategi strategies es against against deceptio deception: n: (a) hid hiding ing of intentions (poker face strategy); (b) tactical deception and misinformation (KGB strategy); and (c) adaptive unpredictability (the protean strategy). The latter concept was developed by Chance (1957) and elaborated by Humphries and Driver (1970) who called unpredictable behavior ``protean'', after the Greek river god who eluded capture by continually, unpredictably changing form. The adaptive logic of protean behavior lies in the fact that animals gener generally ally evolve evolve percep perceptua tuall and cognit cognitive ive capaci capacitie tiess to entrai entrain, n, track, track, and predi predict ct the move mo vemen mentt of othe otherr biolo biologi gical cally ly relev relevan antt anima animals ls such such as prey prey,, pred predato ators rs,, and and po poten tentia tiall mates. Such predictive abilities mean that unpredictable behavior will often be favoured in many many natural natural pursuit± pursuit± evasion evasion predato predatorr ± prey prey situ situatio ations. ns. Usually Usually,, hos hostile tile animals animals or conconspecif specifics ics capabl capablee of correc correctt predi predicti ction on pun punish ish predi predicta ctabil bility ity.. Thus, Thus, endowi endowing ng it with with characteristics that cannot be predicted by an opponent can enhance the effectiveness of almost any behavioral tactic. As outlined above, human courtship is one situation where deception and mindreading will play a role as outlined by Miller (1997) who expects a coevolutionary arms race in courtship between social prediction and social proteanism. However, even if concealment of intentions and ambiguity through social proteanism are creat created ed with with resp respect ect to the the go goals als of inter interact actio ions ns,, ``con `contr trol ol''' strat strateg egies ies are are neces necessar sary y for for managing social interactions in one's favour. We would expect asymmetry in the use of control strategies. The sex, which has the most to lose in an interaction with the other sex, shou should ld try try to contr control ol the inter interact actio ion n to a hig higher her degr degree ee in orde orderr to get the the info inform rmati ation on necessary for decision making and minimization of costs. In this view, men and women have different interests, which will find their expression in how ho w they they act act in enco encoun unte ters rs with with the the op oppo posi site te sex. sex. Ques Questi tion onss of who, who, when when,, and and ho how w interactions with a stranger of the opposite sex are initiated are thus central to any discussion of intersexual interactions and their outcome. We have studied encounters between strangers because they have numerous advantages. These situations have a high degree of uncertainty. If one of the interactants develops interest in his/her partner, it has to be communicated immediately because none of the interactants knows how long the interaction will last. In such a situation with such a short window of opportunity for courtship, the relevant information has to be conveyed very fast. This may look like a limitation of our approach but ``waiting room situations'', where people randomly meet strangers, are as natural as any other situation. This type of approach even has an advantage; it offers a wide range of combinations of degree of interest among the participants ranging from rejection to intense interest. If interest is then communicated, further courtship may take place (Grammer, 1990). Several hypotheses follow from our line of reasoning about behavioral interactions of opposite sex persons who have not previously met. (1) (1) Women, omen, when when conf confro ronte nted d with with a male male stran stranger ger,, will will try try to ``con `contr trol ol''' the man' man's behavior more often than the reverse.
376
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
(2) Obvious courtship and ``come on'' signals will play a role only in the later stages of an interaction when the woman has adequately assessed the man's intentions. In the very early stages, women will show protean behavior, behavior, i.e., try to signal interest and disinterest disinterest erratically in order to veil their intentions and avoid male tendencies for deception by ``blocking'' the man's mindreading attempts. (3) We also can assume that women will try to ``control'' men non-verbally, because nonverbal behavior is, unlike verbal behavior, behavior, non-binding non-binding (Grammer, (Grammer, 1992). 1992). As the man's man's goals are unknown to the woman, there is only one possibility when meeting a male stranger: female solicitation should elicit male self-presentation. (4) Under the pressure of male±male competition and an unknown time for interaction, men will be forced to act in a direct and swift manner. Men should therefore use verbal self presentation in order to communicate interest in the woman.
3. Method
Male Male and and fema female le stra strang nger erss (mea (mean n age age 18 18.5 .5 year years, s, age age rang rangee 18± 23 23)) were were sele select cted ed randomly from different high school classes visiting the research institute where they met in a ``hid `hidden den''' exper experime iment nt.. They They were were told told that that they they woul would d part partici icipa pate te in a vide videor orati ating ng experiment. The experimenter left the room in order to ``answer the phone call'', thus leaving the the dy dyad ad alone alone.. This This sit situat uatio ion n is a no nonn-ar artif tifici icial al ever everyd yday ay waiti waiting ng room room situa situatio tion. n. The The experimenter returned after 3 min. In this first 3 min, participants did not know how long they would interact (phase I: data analysis limited to the first minute). Upon returning, the male or female experimenter told the participants that the phone call would take another 10 min but actual interruption took place at the tenth minute (phase II: 4±10 min). During the whole time time,, they they were were vide videot otap aped ed thro throug ugh h a on onee-way way mi mirr rror or.. One One pair pair of the the orig origin inal ally ly 46 participants who suspected that they were being videotaped was excluded from the analysis resulting in n = 45 pairs. The interactions we used in this analysis are a subset of interactions, which have already been analyzed with regard to laughter by Grammer (1990) with a different different method and coding syst system em and and with with rega regard rd to the the repo report rt by Gram Gramme merr et al. al. (199 (1998) 8) with within in the the conc concep eptu tual al framework of interpersonal coordination. After the experiment, participants were debriefed and asked for their consent. None of the participants refused consent. This is a widespread method of obtaining comparable results on the interaction of two people called the ``dyadic interaction paradigm'' by Ickes, Bissonnette, Garcia, and Stinson (1990). For the analysis, we used three kinds of data. 1. Self-report. Self-report data included the results of a questionnaire assessment of the subj subjec ecti tive ve eval evalua uati tion on of the the situ situat atio ion n by the the part partic icip ipan ants ts at the the end end of 10 mi min. n. Participants rated the physical attractiveness of their partner [Likert scales: 1 (low)±6 (high)] and their level of interest in their partner [Likert scales: 1 (definitely not)±7 (definitely)] (definitely)] with the following statements: (1) I would give my telephone telephone number to the person present if I were asked for it, and (2) I would go out to the cinema with the person if I were asked to. The rank Spearman correlation between these two items was
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
377
0.63 for women (n = 45, p < 0.01, Cronbach' Cronbach'ss alpha 0.80 for the two-item index), and 0.68 for men (n = 45, p < 0.01, Cronbach' Cronbach'ss alpha 0.83). Both items were combined into one index of high and low professed interest in the partner as an independent variable for the determination of non-verbal correlates of professed interest. 2. Non-ve Non-verba rball behavio behaviorr. Analysi Analysiss of non non-ve -verba rball behavio behaviorr consist consisted ed of a microan microanalys alysis is of the behavior of 45 pairs. The microanalysis of behavior was done on an Apple Macinto Macintosh sh compute computerr runnin running g MAC-MA MAC-MAX X with with a frame-b frame-by-f y-fram ramee int interac eractive tive coding coding pro progr gram am,, deve develo lope ped d by the the auth author ors. s. The The rati ratio o of time time for for coll collec ecti tion on of orig origin inal al mate materi rial al to codi coding ng time time eff effort ort was was 1:1 1:110 resu result ltin ing g in 86 865 5 h of codi coding ng for for the the 45 dyads. Coding was completed for each of the 45 dyads and repeated for each of the 83 movement and speech categories. For the complete catalogue, see Grammer et al. (1998). For the present analysis, only a subset of categories was used: ``courtship'' signals sig nals by women, women, and for partner partners, s, an ind index ex of ``negati `negative' ve''' (reject (rejecting ing)) sign signalli alling ng and an index of ``affirmative'' behavior. Seventeen movements, which were defined as being being cour courtsh tship ip sig signa nals, ls, were were thos thosee ident identif ified ied by Mo Moor oree (198 (1985) 5),, Given Givenss (197 (1978) 8) and and oth other erss as flir flirtat tatio ious us or cour courts tship hip sig signa nals ls (see (see lis listt of beha behavi vior oral al un units its,, Table able 2). 2). Signals of rejection or negative female signals have not previously been described in the literature. There are basically two different strategies to indicate rejection: people can avoi avoid d talki talking ng to each each oth other er,, or use use speci specifi ficc negati negative ve sig signa nals ls for for rejec rejecti tion on.. To define ``negative'' signals, we used an index (the simple sum of frequencies) of the six behavioral units with the largest negative correlation with participants' professed inter interes estt in the other other perso person. n. Head Head no nods ds defi define ned d ``affir `affirmat mativ ive' e''' behav behavior ior.. Seve Several ral stud studie iess have have show shown n that that head head no nodd ddin ing g by the the list listen ener er can can po posi siti tive vely ly rein reinfo forc rcee a speak speaker' er'ss verb verbal al perfo perform rman ance ce (for (for a revie review w, see see Argy Argyle, le, 19 1988 88). ). This This means means that that a spea speake kerr who who watc watche hess a pers person on no nodd ddin ing g will will reac reactt by spea speaki king ng mo more re ofte often n and and long longer er.. Male Male and and female female head head no nods ds were were coded coded as ``aff `affir irmat mative ive''' behav behavio iorr on only ly if they took place in parallel to the partner talking. Intercoder reliability for the nonverb verbal al beha behavio viorr analy analysi siss was repo report rted ed (Gra (Gramme mmerr et al., al., 19 1998 98)) as 0.84 0.84 (per (percen centag tagee agreement) for five recorded dyads. 3. Verbal performance. The duration of speech was used as a simplified measure of the participants' verbal performance during the 10-min social interaction. Duration was the sum of each onset and offset of speech for each member of the dyad.
4. Results
4.1. Subjective experience of the situation
The The analy analysi siss of qu ques estio tionn nnair aires es show showed ed that that the deve develop lopmen mentt of selfself-rep repor orted ted male male interest in the woman depended on how attractive the man rated the woman. Professed interest shows sufficient variation from a minimum of four for men (women: two) to a maximum of 14 (women: 14). With such a range, we can analyze the episodes on a broad spect spectru rum m from from low low int inter erest est to high high inter interes est. t. As predi predicte cted, d, prof profes essed sed male male int inter erest est was correlated with female attractiveness. The more attractive the man rated the woman, the
378
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
higher his profess higher professed ed interes interestt (r = 0.37, n = 45, p = 0.004, 0.004, two-tai two-tailed) led).. Women omen did not sign signif ific ican antl tly y link link thei theirr inte intere rest st in the the man man with with male male attr attrac acti tive vene ness ss (r = 0.23 0.23,, n.s. n.s.). ). Furthermore, the correlation did not differ significantly from those of the men when tested with with a metho method d prop propos osed ed by Soka Sokall and Rohl Rohlff (199 (1995) 5).. Inter Interes estt in the part partner ner show showed ed an interesting sexual asymmetry. Professed male interest in women was significantly higher (n = 45, Mdn = 8.5 of 14 maximum) than professed female interest in men (Mdn = 6.5, Wilcoxon; p = 0.00 0.0018 18;; twotwo-ta tail iled ed). ). This This show showss that that men, men, on aver averag age, e, were were mo more re inter interes ested ted in women women than the the othe otherr way arou around nd,, and it sugg sugges ests ts that that mo more re women women are interesting for men than men are for women. This result corroborates the results found by Abbey, McAuslan, and Ross (1998). 4.2. Non-verbal behavior: female courtship signals
In this part of the analysis, we attempted to determine the relationship between female courtship signals and female-reported interest in her partner (Table 1). Cor Correla relati tion onss of the the beha behavi vior orss wer were calc calcul ulat ated ed firs firstt for for the the tota totall du dura rati tion on of the the obser ob serva vatio tion n period period (10 (10 min min,, Table able 1). 1). Head behavio vioral ral patte pattern rn wher wheree the Head akimbo akimbo is a beha hands are folded, the arms are moved up and the hands rest in a position behind the head with with the the shou should lder erss draw drawn n back back.. This his patt patter ern n is inte interp rprreted eted as ``loo `look k at my bo body dy'''
Table 1 Female solicitation signals and professed interest in the male Spearman correlation with professed female interest Signal Head akimbo Primp Head toss Hair flip Head tilt Breast presentation Palm Head down Shrug Coy smile Legs open Look through Short glance Illustrator Arms flex Smile Laugh Total
Total (10 min)
Phase I (first (first min)
Phase II (fourth±tenth min)
Author who described signal
À0.25
À0.20
À0.08
0.35* À0.17 À0.00 0.24 À0.01 0.07 0.10 0.14 0.34* 0.03 0.31* 0.51** 0.28 À0.21 0.13 0.16 0.45**
0.18 À0.09 À0.13 0.22 À0.06 0.01 0.01 0.20 ± À0.10 0.09 0.26 0.09 À0.16 0.20 À0.01 0.19
0.31* À0.13 À0.00 0.30* 0.04 0.15 0.17 0.18 0.34* 0.15 0.27 0.19 0.32* À0.13 0.13 0.17 0.38**
Goffmann (1979) Scheflen (1965) Moore (1985) Moore (1985) Moore (1985) Moore (1985) Moore (1985) Ellis (1992) Givens (1978) Eibl-Eibesfeldt (1995) Grammer (1990) Moore (1985) Moore (1985) Moore (1985) Moore (1985) Moore (1985) Moore (1985)
* Significant at p < 0.05, two-tailed. ** Significan Significantt at p < 0.01, two-tailed.
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
379
(Gof (Goffm fman ann, n, 19 1979 79), ), bu butt for for ou ourr part partic icip ipan ants ts,, this this patt patter ern n corr correl elat ates es nega negati tive vely ly with with prof profess essed ed female female int intere erest. st. Primp refe refers rs to orde orderi ring ng on one' e'ss clo clothes thes with withou outt a visi visibl blee necessity. In Head toss, the head moves down, followed by a fast circular upward move, and then the head slowly returns to the original position. The Hair consis ists ts of the Hair flip flip cons same head movement as Head toss, the difference being that here, the hands are used to thro throw w the the hair hair back back.. Neck Neck pres presen enta tati tion on is a side sidewa ward rd Head suppos posedly edly sign signalli alling ng Head tilt tilt , sup submission or avoidance of the aggressive effect of the ``staring eyes'' when eye contact occurs. occurs. In Breast presentation shoulders are mov moved ed back simu simultan ltaneou eously sly.. In Palm, presentation, both shoulders bot both h palms palms are pres present ented ed up upwa ward rds. s. Head lowers rs the the head ead and and gaze gaze and and is also also Head down down lowe alleged to be a sign of submission in courtship. In Shrug , the the shou shoulde lders rs are are repe repeate atedly dly moved up and down. Coy smile is a smile followed immediately by a turning away and lowe loweri ring ng of the the head head.. Legs when sitt sitting ing is self-ex self-explan planator atory y. Look Legs open open when Look throug through h was defined as looking at the other person but not fixating on her or him and looking away imme im medi diat atel ely; y; ther theree is no paus pausee betw betwee een n the the mo move veme ment nt of look lookin ing g at the the part partne nerr and and looking away from the partner. Short glance is directed at the partner for less than 3 s. illustr trati ating ng hand hand mo movem vemen entt when when speak speakin ing. g. In Arms Illustrator is an illus Arms flex flex, one or both arms are flexed at the elbow and held in front of the body. Smile is a contraction of the Musculus zygomaticus major and Laugh is the same, but with the mouth open and sound is produced. produced. Only a few patterns showed a significant positive correlation (Primp, Coy smile, Look throug through, h, Short Short glance) glance) with profes professed sed female female int interes erestt (Table (Table 1). This This situ situatio ation n does does not change much when each of the behaviors was limited to either of the two phases. In order to find out whether there were any changes in the correlations between the two phases, the coefficients were z -transformed -transformed and tested with a paired t -test. -test. Overall, there was a significant increase in size of the correlation between phases I and II (t -test, -test, df = 16, t = À3.93, p = 0.001), with a higher correlation between professed interest and total solicitation score in phase II (r = 0.38) than in phase I (r = 0.19). An index of solicitation behavior (the frequency of solicitation behaviors) for the whole 10-min period was significantly positively correlated (r = 0.45) with professed female interest in her partner. 4.3. Entropy and variability of courtship behavior
The frequency of courtship behavior decreased significantly from phase I to phase II. Table 2 shows that most of the individual behaviors increased significantly in frequency from from ph phas asee I to ph phas asee II. II. This This appa appare rent nt cont contra radi dict ctio ion n is solv solved ed when when we look look at the the variabi variability lity of courtshi courtship p behavio behaviorr. Variabili ariability ty is the quo quotien tientt of the num number berss of differ different ent beh behav avior iorss perf perfor ormed med,, divid divided ed by the the nu numb mber er of all behav behavior iorss perf perfor ormed med.. A maxim maximum um variability of 1 would mean that each behavioral act was different. In phase I, variability was 0.39 and dropped in phase II to 0.13. This difference is highly significant (Wilcoxon test, Z = À5; p = 0.000 0.000,, two-t two-tail ailed) ed).. This This varia variabi bilit lity y could could be an indi indicat cator or of a high higher er degree of proteanism in phase I. In order to verify that proteanism could be reflected in variability, an analysis of entropy was conducted. The basis of this analysis is information theory as outlined by Shannon and Weaver (1949). Their approach was refined by Losey (1978) and applied to the analysis of
380
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
Table 2 Frequencies of female courtship behavior ( n = 45) Mean of behaviors/m behaviors/minut inutee (SEM) Signal
Total (10 min)
Phase I (first (first min)
Phase II (fourth±tenth min)
Wilcoxon test Z ( p) Phase I < Phase II
Head akimbo Primp Head toss Hair flip Head tilt Breast presentation Palm Head down Shrug Coy smile Legs open Look through Short glance Illustrator Arms flex Smile Laugh Total
1.04 (1.97) 3.98 (3.36) 1.93 (3.77) 3.15 (5.32) 12.87 (9.88) 0.33 (0.63) 0.28 (0.72) 60.41 (24.76) 2.85 (7.58) 0.07 (0.33) 2.26 (3.26) 1.78 (2.39) 60.98 (22.39) 6.52 (9.04) 6.70 (7.09) 10.15 (6.64) 11.67 (6.80) 18.7 (5.4)
0.26 (0.95) 0.98 (1.47) 0.33 (0.73) 0.54 (1.28) 1.93 (1.88) 0.02 (0.15) 0.04 (0.21) 7.26 (4.45) 0.57 (1.26) 0.00 (0.00) 0.46 (0.72) 0.46 (0.86) 0.67 (1.12) 0.57 (1.22) 0.93 (1.36) 1.41 (1.34) 1.57 (1.36) 18.0 (7.5)
0.61 (1.26) 2.20 (2.50) 1.11 (2.80) 2.04 (3.38) 8.09 (6.89) 0.17 (0.38) 0.22 (0.66) 40.96 (16.93) 1.46 (4.51) 0.04 (0.21) 1.35 (2.07) 0.96 (1.66) 4.37 (3.72) 5.00 (7.50) 4.22 (5.08) 6.33 (4.35) 7.26 (4.81) 14.3(4.8)
À2.0 (0.04) À2.8 (0.005) À2.3 (0.02) À4.1 (0.0001) À5.4 (0.0001) À2.0 (0.04)
n.s. À5.9 (0.0001) n.s. n.s. À2.7 (0.007) n.s. À4.9 (0.0001) À4.8 (0.0001) À4.0 (0.0001) À5.7 (0.0001) À5.6 (0.0001) Phase I > Phase II 1.8 (.05)
information information content in behavior behavior sequences by Hughes (1978). (1978). For this analysis, we developed developed a comp comput uter er prog progra ram m imp implem lemen entin ting g the the algor algorith ithms ms and and sour source ce code code prov provid ided ed by Pres Press, s, Teukolsky, Vetterling, and Flannery (1992). A measure of uncertainty (U ) of events is defined as a monotonic increasing function of the number of alternative behaviors available. U then is a logarithmic function of the number of alternatives and it is the expectation of the function log P i (the probability of behavior behavior i); it is thus analogous to Shannon's measure of information and: U À
i
P i log P i X
i1
U max max is the maximum uncertainty possible under a given repertoire. Thus, for a range of n behaviors, U max max = log2n. In our case, this is 4.087. Hughes (1978) was able to show with this method that behavior sequences in children's play show higher entropy and thus are more freely variable than in exploration. In our case, U was calculated for each participant and then the means from the first minute and the last minute of the observation period were compared. The last minute was used because the time range of observation has to be the same; thus, the expected probability of occurrence for each behavior is the same. In the first minute, U = 1.50 and in the last minute, -test for pairs, df = 44, t = 4.93, p < 0.001) 0.001).. Hence, U = 1.08. This difference was significant (t -test
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
381
the uncertainty in the performance of behavior dropped significantly from the beginning of the encounter to its end. In the the firs firstt mi minu nute te (pha (phase se I) court courtsh ship ip beha behavi vior or was more more varia variabl blee and and no nott stro strong ngly ly relat related ed to prof profess essed ed female female int inter erest est in the initi initiat ation ion of the op oppo posit sitee-sex sexed ed enco encoun unter ters. s. Yet, they seemed seemed to become become more more strong strongly ly related related in later later stages stages (``T (``Total otal''' sig signal nal phase phase II; II; Table able 1). 1). This difference difference raises the question question of how women with low professed interest use courtship signals. In order to answer this question, the female participant's reports of interest in their male partners were divided into three categories of professed interest. Three groups were chosen because this yielded enough cases for each group. We then compared the low-interest group to the high-interest group. In phase I, women with low professed interest sent 15.5 courtship signals per minute, whereas 19.1 came from highly interested women (Median test, n.s.). The difference was statistically significant in phase II (11.4 for low interest vs. 15.5 courtship signals for high professed interest; Median test, exact probability p = 0.007). It would appear that women with low professed interest cannot be discriminated on the basis of their their cour courts tship hip behavi behavior or from from thos thosee with with high high prof profess essed ed inter interest est in ph phas asee I, bu butt can be discriminated by phase II. 4.4. Non-verbal behavior: negative signals
In order to determine a baseline of negative signalling, we calculated an index of the six behavioral categories showing the lowest correlation with professed interest (negative signall sig nalling; ing; Table 3). Here, Here, the results results of our study contra contradict dict the literatu literature re on courtsh courtship ip sign signals als.. The The codes codes are are selfself-exp explan lanato atory ry with with the the excep exceptio tion n of Deictograph. This is a simpl simplee po poin intin ting g mov moveme ement nt with with a stre stretch tched ed-o -out ut finger finger towar towards ds an ob objec ject. t. Two of the courts courtship hip sign signals, als, Head correlate ate negat negative ively ly with with prof profess essed ed Head toss toss and Head Head akim akimbo bo, correl inte intere rest st and and are are thus thus a part part of the the inde index x for nega negati tive ve sign signal alss for for wome women. n. Nega Negati tiv ve sign signall alling ing seems seems to be stable stable thro throug ugho hout ut bo both th ph phase asess for for women women.. It drop dropss sign signif ifica icantl ntly y for for men men from from ph phas asee I to ph phas asee II. II. For For nega negati tive ve sign signal alli ling ng,, ther theree is an inte intere rest stin ing g sex sex diff differ eren ence ce in ph phas asee II. II. Wom omen en sign signal al nega negati tive vely ly mo more re ofte often n in ph phas asee II than than men men do (1.7 (1.7 vs. vs. 1.1: 1.1: Wilcox ilcoxon on test, test, p < 0.00 0.001, 1, twotwo-tai tailed led). ). The The ov over erall all freq frequen uencie ciess of negat negativ ivee signalling are astonishingly low when compared to courtship signals [see Total in Table Table 3 Spearman rank correlations negative signals with professed interest ( n = 45, both sexes) Behavior
Males
Behavior
Females
Roll sleeves up Look around Deictograph Palm Close legs Smile Index
À0.18
Head akimbo Knees toward body Move legs Cross legs Arms parallel Head toss
À0.25
À0.19 À0.20 À0.19 À0.36* À0.20 À0.40*
* Significant at p < 0.05, two-tailed.
À0.18 À0.32* À0.31* À0.17 À0.17 À0.22
382
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
6 1 . 0 r
6 2 . 0
7 2 . 0
À
) ) . n 1 i I 1 ± I h m ( e 0 r t s t 1 u h . a n 1 o e h f P ( t
) 3 7 . 0 ( 8 7 . 0
) 4 . 0 1 ( 7 . 1 2
) 9 6 8 s 1 0 1 . . . e 0 0 0 x e r À s h t ) ) o . ) ) 2 b . 1 8 n 0 r i ( . 1 I o 2 1 m ( 4 ( f e t , s 2 2 s 5 . . a i . r 7 4 h f 1 1 0 P ( 2 = n ( 2 4 3 2 2 2 r . . . e 0 0 0 n t r À r a p ) ) 3 7 ) e . 9 ) h s . ( 0 9 . t n e 0 i ( 9 l ( n 5 i a l a m 0 8 3 t . . t . m s e 1 0 1 o 0 e 1 r F T ( 2 e t n i * * * * d 0 4 6 e 4 3 3 s . . . s 0 0 0 e f À r o r p ) ) 1 h ) t 0 8 ) i . . n I 1 ( 0 4 . ± i ( w I h 9 m ( s t e 8 0 n s r h 5 7 1 u t o a . . . i n o h 1 0 t 1 f e a P ( t 2 l e r r 5 2 1 o 1 1 2 c . . . 0 0 0 ) r r À ( k ) ) ) n 5 9 a . . ) 2 . r 2 1 8 n i ( ( ( n I m a e 8 6 2 m s t r 2 9 s 9 . . a . r a i 2 h f 1 7 e P ( 1 p S d * n * * a 0 9 3 2 3 4 ) . . . 0 0 0 D S r À ( s ) e i 1 ) ) r 8 o . . ) 8 g . 0 0 8 n e ( ( i 8 t ( s a l 8 7 e c l a m 8 5 7 t . . . 0 r a o 1 1 0 0 o 2 i M T ( v a h n / i e , b m n n / o f i s o t e l a i t n v a 4 o i i h a s n t n r m c i n e a i g i m r u / e l a g s f e d m s b e e f p a T M N A S a
a
a
b
. . 1 5 0 . 0 . 0 0 < . < p . 5 5 0 p t , 0 . . , s 0 0 t s e r e < < e r t e p p t i n n , t , i t s t h s e h i e t t t i w n n w n o o n o x x o i o i t o c a c t l i l a l i l e r W e r W r r o , , I c o I I I c k e e n s k s a n a a r a h h P r n P n a < a m > I I r m a r e e e s a s a a e p h h p S P P S * * *
a
b
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
383
2 (18.7) as compared to total negative signals in Table 4 (1.0)]: the ratio of negative to pos posit itiv ivee sign signal alss (the (the nu numb mber er of nega negati tive ve sign signal alss divi divide ded d by the the nu numb mber er of po posi siti tive ve sign signal als) s) is ver very low: low: it was was 0.09 0.09 in ph phas asee I and and drop droppe ped d to 0.07 0.07 in ph phas asee II, II, a no nonnsignificant significant change. 4.5. Affirmation and speech
The frequency of nodding was lower for both sexes in phase II than in phase I (Wilcoxon test, p < 0.01, two-tailed, two-tailed, Table Table 4), and there was no significant significant sex difference. difference. In contrast to non-verbal behavior, the amount of male speech increased significantly from phase I to phase II and only male speech duration correlated positively with male professed interest in phase II. Female professed interest showed no significant relation to female affirmation behavior and speech duration. 4.6. Female non-verbal control
Non Non-v -ver erba ball cont contro roll coul could d be exer exerte ted d by mean meanss of sign signal als, s, whic which h are are no nott dire direct ctly ly linke lin ked d to sexu sexual al sig signa nalli lling ng.. One One such such sig signa nall is affi affirm rmati ation on thro throug ugh h no nodd ddin ing. g. This This can happen happen in every every interact interaction ion ind indepen ependen dentt of who is interact interacting. ing. Further Furthermor more, e, there there mig might ht be be othe otherr beha behavi vior ors, s, whic which h are are used used in the the same same mann manner er,, bu butt no nodd ddin ing g is the the on only ly on onee wher wheree we have have a theo theore reti tica call and and expe experi rime ment ntal al rati ration onal alee for for the the assu assume med d func functi tion on of nodd no ddin ing g in the the cont contro roll of spee speech ch flow low in soci social al inte intera ract ctio ions ns.. Affi Affirm rmat atio ion n thro throu ugh nodd no ddin ing g by a sign signal al send sender er coul could d lead lead to a rise rise in freq freque uenc ncy y of the the rece receiv iver er talk talkin ing, g, but but ther theree are are sever several al caveat caveats. s. Firs First, t, the the sig signa nall sende senderr canno cannott no nod d if the the recei receiver ver is no not t speaking. Then, it is not clear if the signal sender nods because the receiver is speaking or the recei receiver ver is speak speaking ing becaus becausee the the sign signal al send sender er no nods ds.. Even Even when when we exten extend d this this to no nodd ddin ing g and and spea speaki king ng at diff differ eren entt time timess of the the inte intera ract ctio ion, n, ther theree coul could d be an inte intera ract ctio ion n betw betwee een n spea speaki king ng at time time 1 and and spea speaki king ng at time time 2. The The same same is true true for for nodding. Thus, the interactions between speaking and nodding at different times have to be be cons consid ider ered ed.. In orde orderr to test test this this,, we used used the the foll follow owin ing g appr approa oach ch as prop propos osed ed by Aiken Aiken and West (1996) (1996).. In the the firs firstt mo mode dell (Tab (Table le 5, mo mode dell A), A), male male spee speech ch in ph phas asee II is the the dep depende endent nt vari variabl able, e, while while the indep independ enden entt variab variables les were were female female no nodd ddin ing g in ph phas asee I, male male speec speech h in ph phase ase I and and female female no nodd ddin ing g in ph phase ase II. II. Addit Additio ional nally ly,, the the three three inter interact action ionss among among the indepen independen dentt variabl variables es were were entered entered into the regres regressio sion n equatio equation. n. The The resu resulti lting ng regr regres essio sion n mod model el show showed ed that that affi affirm rmati ative ve behav behavio iorr was was sign signif ifica icantl ntly y affe affect cted ed by male male spee speech ch in ph phas asee I, by fema female le no nodd ddin ing g in ph phas asee I, by the the inte intera ract ctio ion n between male speech in phase I and female nodding in phase I, and by female nodding in phases I and II. Model A shows that male speech duration in phases I and II were sign signif ifica icant ntly ly and and po posit sitive ively ly assoc associat iated ed.. Furt Furthe herm rmor ore, e, males males spok spokee mo more re in ph phase ase II in relat relation ion to femal femalee no nodd ddin ing g frequ frequen ency cy in ph phas asee I. Female Female no nodd ddin ing g and and male male speec speech h in pha phase se I ampl amplif ifie ied d each each othe otherr and and this this inte intera ract ctio ion n faci facili lita tate ted d male male spee speech ch in ph phas asee II. II. High High spee speech ch scor scores es for for men men and and the the freq freque uenc ncy y of fema female le no nodd ddin ing g led led to even even high higher er male male speec speech h score scoress in ph phas asee II. II.
384
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
Table 5 Results of backward multiple regression analysis (criterion for exclusion: probability of F -value -value ! 0.050) Model
Trait
Coefficient
Standard coefficient
t
p
4.556 3.026 2.279 À2.121
0.000 0.004 0.028 0.040
5.259
0.000
0.464
3.588
0.001
0.419 0.404
3.184 3.065
0.003 0.004
0.480
3.626
0.001
0.283
1.958
0.057
Affirmative behavior
Multiple R = 0.650, p = 0.000 A
B
a
Male speech II Male speech I 0.594 À3.783 Female nodding I 460.047 0.410 Male speech I Â Female nodding I 1.657 0.320 Female nodding I Â Female nodding II À41.455 À0.319 Multiple R = 0.694, p = 0.000 a Female nodding II Female nodding I 2.137 0.595
Female negative behavior
Multiple R = 0.534, p = 0.001 C
D
a
Male speech II Male speech I
2.953 Multiple R = 0.528, p = 0.001
Female negative behavior IIa Male speech I Male speech I Â Male speech II
0.002 1.698
Female Female solicitation solicitation
Multiple R = 0.480, p = 0.001 E
F
Male speech II Male speech I
a
Female solicitation IIa Female solicitation I
3.056 Multiple R = 0.283, p = 0.057 0.182
Indepe Independe ndent nt variab variables les (predi (predicto ctors) rs) are listed listed below below.. Only Only sig signif nifica icant nt traits traits from from the final final regres regressio sion n equati equations ons are report reported. ed. a Male speech in phase II Ð fourth to tenth minute Ð (male speech II), female nodding in phase II (female nodding II), female negative behavior in phase II and female solicitation in phase II as dependent variables. Model A: Predictors: (Constant), Male speech I, Female nodding I, Female nodding II, Male speech I±Female nodding I, Female nodding I±Female nodding II, Male speech I±Female nodding II. Model Model B: Predictors: (Constant), Male speech I, Male speech II, Female nodding I, Male speech I±Female noddin nod ding g I, Female nodding nodding I ± Male Male speech speech II, Male Male speech speech I ± Female Female nod noddin ding g I. Model C : Predictors: (Constant), Male speech I, Female negative behavior I, Female negative behavior II, Male speech I±Female negative behavior I, Female negative behavior I±Female negative behavior II, Male speech I± Female negative behavior II. Model D: Predictors: (Constant), Male speech I, Male speech II, Female negative behavior I, Male speech I± Female negative behavior I, Female negative behavior I±Male speech II, Male speech I±Male speech II. Model E : Predictors: (Constant), Male speech I, Female solicitation I, Female solicitation II, Male speech I± Female solicitation I, Female solicitation I±Female solicitation II, Male speech I±Female solicitation II. Model F : Predictors: (Constant), Male speech I, Male speech II, Female solicitation I, Male speech I±Female solicitation I, Female solicitation I±Male speech II, Male speech I±Male speech II.
385
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
As is the the case case of male male spee speech ch,, fema female le no nodd ddin ing g in ph phas ases es I and and II ampl amplif ifie ied d each each othe otherr and and male male speec speech h score score in ph phas asee II. II. This This sugg suggest estss the follo followin wing g inter interpr preta etatio tion. n. In ph phase ase I, the the male male star starts ts spea speak king, ing, the the fema female le is lik likely ely to no nod d. This This lead leadss to high higher er freq freque uenc ncie iess of male male spee speech ch in ph phas asee II. II. Thus Thus,, fema female le no nonn-ve verb rbal al beha behavi vior or in ph phas asee I cont contro rols ls male male speec speech h prod produc uctio tion n in ph phase ase II. II. In order to test whether male speech could determine female nodding and female nodding determine male speech, we tested whether female nodding in phase II could be predicted by male speech in phase I (Table 5, model B), but the only significant predictor for female nodding in phase II was female nodding in phase I. In orde orderr to asse assess ss the the im impa pact ct of fema female le nega negati tive ve beha behavi vior or,, we made made an anal analys ysis is of fema female le nega negati tive ve beha behavi vior or in ph phas asee I and and male male spee speech ch in ph phas asee II (Tab (Table le 5, mo mode dels ls C and D). In this case, the interactions and female negative behavior in phase II and male speec speech h in ph phas asee I were were enter entered ed int into o the the mod model. el. Inter Interes estin tingly gly,, it is on only ly male male spee speech ch in phase I that had a significant effect on male speech in phase II. When we tested for the reve revers rsee effe effect ct and femal femalee negat negative ive beha behavio viorr in ph phas asee II, II, it was was clear clear that that this this behav behavio ior r depe depend nded ed on male male speec speech h in ph phas asee I. In oth other er word words, s, the the wome women n react reacted ed nega negativ tively ely if the the man man spok spokee a lot. lot. As we had had pred predic icte ted d on the the basi basiss of prot protea ean n theo theory ry,, soli solici cita tati tion on beh behav avio iorr had had no effe effect ct ov over er time time on male male spee speech ch (Tab (Table le 5, mo mode dels ls E and and F). F). There remains the question of possible correlations between our criterion variables and the problem of interpreting interaction effects between them. In order to address this issue, we present a correlation matrix among the several criterion variables. Table 6 shows that the dependent variables we used in the regression equations are somewhat correlated. We have have cons consid ider ered ed seve severa rall ways ways for for a po poss ssib ible le solu soluti tion on (e.g (e.g., ., stru struct ctur uree equa equati tion on modelli mod elling ng for a more more causal causal int interp erpreta retatio tion n of our argumenta argumentation tion)) and finally finally decided to use use the the prese resent nted ed mu mult ltip iple le regr regres essi sion on mo mode dell to clea clearr up the the prob proble lem m of havi having ng seem seemin ingl gly y inde indepe pend nden entt test testss of mu mult ltip iple le corr correl elat ated ed pred predic icto tors rs.. The The anal analys ysis is of ou our r observ obs ervatio ations ns sho shows ws that behavio behaviorr remains remains rather rather constant constant during during a particu particular lar time span. span. Thus Thus,, we assu assume med d that that alte altern rnat ativ ivee meth method ods, s, like like caus causal al mo mode dell llin ing, g, did did no nott meet meet the the needs of our argument, as these other models require greater independence between the
Table 6 Spearman rank correlations of dependent (criterion) variables in the final regression equations ( n = 45) Male spee speech ch II Male speech II Female nodding II Female negative behavior II Female solicitation II
±
Femal emalee nodd odding ing II 0.35* ±
Female negative behavior behavior II 0.25 0.19 ±
Female solicitation II 0.29* 0.35* 0.52** ±
Male speech II = Male speech in phase II Ð fourth to tenth minute, Female nodding II = Female nodding in phase II, Female negative behavior II = Female negative behavior in phase II, and Female solicitation II = Female solicitation in phase II. * Significant at p < 0.05, two-tailed.
386
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
pre predi dict ctor or vari variab able less than than is the the case case.. Furt Furthe herm rmor ore, e, such such an appr approa oach ch woul would d gene genera rate te artifac artifacts ts in the int interpr erpretat etation ion of int interac eractio tion n effects effects..
5. Discussion
Inter Interact action ionss betwe between en stran stranger gers, s, especi especiall ally y tho those se betwe between en the the two two sexes, sexes, seem seem to be governed by specific rules. Our results, like that of other researchers, suggest a general sex difference in the professed interest in the other sex. Men are more interested in women than the other way around. Women seem more reluctant than men to make contact. Our second aim was to generat generatee predict prediction ionss from from evolut evolution ionary ary-ba -based sed meta-th meta-theor eories ies which which predict predict behavioral tendencies under a wide range of motivational factors. The The main main meth method odol olog ogic ical al im impl plic icat atio ion n of this this stud study y is the the util utilit ity y of the the conc concep eptt of proteanism in uncertain high-risk social situations. After performing many types of behaviors, wome women n seem seem to mo move ve to a mo more re cons consis iste tent nt repe repert rtoi oire re at the the end end of the the enco encoun unte terr. Furthermore, women seem to exert more active control over male behavior: men produce more verbal revelations than women, and the quantity of male self-presentation varies with the female's behavior (predictions 1, 3, and 4, Section 2). This is suggested by the fact that there was a clearer dependence of male behavior on the preceding female acts than vice versa. We foun found d that that women women solic solicite ited d male male verb verbal al react reactio ions ns activ actively ely and and that that femal femalee negat negativ ivee behavior was influenced by male verbal behavior. In addition, female engagement in verbal interaction did not necessarily signal interest. Many theories assume that the sexes are essentially equal in their behavioral propensities and that learning processes lead to social stereotypes that compel men to take the initiative in initial mixed-sex encounters more often than women. We find that the women themselves take control by sending subtle signals. To ou ourr kn know owled ledge, ge, this this is the the firs firstt tim timee that that these these effect effectss have have been been supp suppor orted ted with with quantit quantitativ ativee analyse analysess of obs observ ervatio ational nal data. data. Our results results sup suppor portt the hyp hypoth otheses eses that sex diff differe erenc nces es are are prese present nt and and that that they they foll follow ow the dire directi ction on pred predict icted ed by ou ourr evolu evolutio tionnmind mi nded ed hy hypo poth thes eses es:: wome women n try try to elic elicit it info inform rmat atio ion n from from men men and and try try to avoi avoid d the the possibility of deception. When the situation is highly uncertain and time is limited, both sexes start signalling immediately. There is no direct and strong initial relation between interest and signalling. Interestingly, women use signals described in the literature as solicitation signals irrespective of their professed interest. Wom omen en with with low low and and with with high high prof profes esse sed d inte intere rest st bo both th send send the the same same amou amount nt of solicitation signals, but only highly interested women maintain the rate in phase II. This view is underlined by the positive positive relation between female interest and an index of solicitation in phase II (Prediction 2, Section 2). This interpretation is strengthened by the correlation between affirmation and male speech duration. Female affirmation (Head nod ) seems to amplify male speech production. This does not merely involve a simple feedback in which affirmation affirmation triggers male speech production, production, which in turn causes female affirmation. affirmation. Rather, Rather, women appear to control male speech production through the use of non-verbal behavior. This is consistent with our proposed strategy theory, which holds that women try to elicit
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
387
information from the man in order to make a decision. In the first stages of the contact, women showed higher variability in their courtship behavior. This fact could be explained by different reasons: either women behave in a protean way and send signals independent of their interest, or women signal a complete repertoire and try to find out which behavior modulates a man's response. In this view, every pair would develop its own non-verbal code for courtship. However, this explanation is not sufficient to explain the fact that there was no clear and distinct negative relation between solicitation behavior in the first stages and an absence of professed interest. It would not be necessary to signal solicitation behavior, in order to establish a common communicative repertoire, if no interest is present. We predicted that women would tend to use covert strategies in order to avoid male deception. At a first glance, this does not seem to be the case. Rather, decisions are made quite early in an encounter as evidenced by a high correlation in behavior between the two pha phases ses.. Inde Indeed, ed, we kn know ow that that first first imp impre ress ssio ions ns are are made made very very qu quick ickly ly and and that that tho those se impressions are generally quite accurate (Ambady & Rosenthal, 1992). Although decision points can be very early, people still have to verify this first impression. This puts the signal sender in a difficult situation: if the sender is interested but needs verification of information, he/she has to be deceptive. In order to avoid detection in such a case, the signal sender (the woman) should start signalling positively and try to veil her intentions. However, deceit woul would d selec selectt for for effi effici cien entt mi mind ndre reade aders rs who who try try to find find mo more re subt subtle le cues, cues, bu butt we have have experimental evidence that men are not as skilled as women in deciphering non-verbal cues (Argyle, 1988). It seems possible that men ignore female solicitation in early interactions. Another male strategy would be to probe the signaller and demand evidence from additional signals, signals, but we currently currently have scarce evidence for such a strategy strategy. The protean strategy strategy would also explain the fact that men continuously overestimate female sexual interest which seems more likely than mindreading a protean woman. An objection against a protean/deceptive explanation is the possibility that women might test out men and then find out that the man is not interested and accordingly adjust their professed interest to disinterest. If professed interest is the same as genuine interest, then others present should be able to detect this. Our results on solicitation behavior replicated Moore's (1985). Moore also found that an index of only 10 out of 52 courtship behaviors could predict ``come on'' effects. Although Moore did not make this point, such a finding of any 10 of 52 behaviors would result in 1.6 Â 1010 different signal combinations. In general, non-verbal signals are ambiguous and it could be the quality of motor performance that denotes interest in such a situation as shown by Grammer et al. (1999). Moreover, if the combinations of different signals have to be taken into account by the observer, the complexity of the task becomes very challenging. On the same data, with different analytic methods, Grammer (1990) showed that female interest can be depicted from combinations of several signals performed simultaneously, i.e., laughter combined combined with body posture. Thus, the general ambiguity of non-verbal signalling signalling could lead to the variab variabili ility ty in female female sig signal nallin ling. g. There There are are oth other er po possi ssibl blee explan explanati ation onss for for the deployment of solicitation signals by women, including emphasis on gender role identification, and enhancement of their attractiveness to men. Women typically take longer than men to evaluate man's acceptability. One possible interpretation of our results is that the women were uncertain and thereby signalled conflicting cues. Shyness could be coupled with age.
388
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
Our participants (18.5 years) could be considerably younger than the participants from other studies who where observed in bars and nightclubs. Furthermore, cultural variations in the incidence and dynamics of solicitation behavior are likely. Most of the researches cited in this paper were carried out in the United States or Canada. Grammer et al. (1999) report the same repertoire for Japanese as Germans, but lower rates of solicitation behavior with opposite-sexed strangers in Japan than in Germany. There are also many known methodological problems in our study. The first problem is sample size. The negative findings can be a result of our small sample size, because the statistical power (the ability to detect true effects) for the analysis of our data was low. Other speculations speculations for the discrepancy discrepancy between signalling solicitation solicitation and professed professed interest are that the 10-min encounter encounter with analysis analysis based on 1 min was too short. Yet, Yet, the question question remains as to why there were more courtship signals in the very beginning than later stages in the 10-min encounter. Also, the situation could be artificial, although this is not actually the case. A waiting room situation is an everyday situation and interest in the other sex can occur in any situation, so people should be familiar with this kind of encounter. One of the surprising facts in this study was that women, if not interested, do not send clear rejection signals. Thus, the man is likely to feel that he himself has started the interaction. In fact, women show affirmation, engage in verbal exchange, and send sexually explicit signals without having much interest in the man. Female negative behavior is performed with astonishingly low frequencies. This can be interpreted as a result of the woman's attempt to prod men to reveal information about themselves. Negative signals were never strongly related to female interest at any time of the interaction. interaction. Negative behavior was apparently a reaction to male speech production: production: men who produced too much in phase I were highly likely to encounter female negative behavior later. Possible Possible goal incompatibility incompatibility in encounters between complete strangers strangers has great potential potential for conflict for both sexes. Cultural norms could function to prevent people from signalling negative emotions too directly and thereby reduce conflict. This argument is consistent with our findings of low rates of female negative signals. The necessity for manipulation, and for information, along with possible deception, leads to a highly ambiguous situation. Here men might well perceive female behavior as an aggressive ``come on'', especially in light of missing negative signals. In solving this dilemma, women attempt to create an ambiguous situation by using non-verbal behavior, while simultaneously trying to control the man. As soon as the highly discriminative woman determines that the man Ð who is under the impression of being accepted Ð is not suitable, she does not clarify the situation. This behavior can be interpreted as being ``aggressive'' in the sense that it manipulates men's behavior and their social perceptions. The The resu results lts of this this analy analysis sis show show many many incon inconsi siste stenci ncies, es, bu butt also also make make clear clear that that the conc concep eptt of prot protea eani nism sm in situ situat atio ions ns with with high high soci social al risk riskss mi migh ghtt be usef useful ul in futu future re researc research. h. Furthe Furtherr stud studies ies sho should uld entail entail ratings ratings by third third party party partici participan pants ts on interest interest and on the possible perception of interest. We would predict from protean theory that neither men nor women should be able to predict female interest from the very early stages of the the encou encount nter ers. s. Other Other dire directi ction onss for for futu future re resea researc rch h invo involv lvee comp compari ariso sons ns of the socia sociall inter interact actio ions ns in sim simila ilarr enco encoun unter terss in differ different ent conte context xtss and and with with pers person onss varyin varying g in age age and marital marital status. status.
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
389
Acknowledgments
Many thanks to Ivo Ponocny from the Institute of Psychology at the University of Vienna for his valuable help with the statistical analysis in this paper. References Abbey, A. (1982). Sexual differences in attributions for friendly behavior: Do males misperceive female's friendliness? Journal of Personality and Social Psychology, 42, 830±838. Abbey, A., Cozzarelli, K., McLaughlin, K., & Harnish, R. (1987). The effects of clothing and dyad sex composition on perceptions of sexual intent: Do women and men evaluate these cues differently? Journal of Applied Social Psychology, 17 , 108±126. Abbey, A., McAuslan, P., & Ross, L. T. (1998). Sexual assault perpetration by college men: The role of alcohol, misperception of sexual intent and sexual beliefs and experiences. Journal of Social and Clinical Psychology, 17 , 167±195. Aiken, L., & West, S. (1996). Multiple regression: Testing and interpreting interaction . Newbury Park, CA: Sage Publ. Ambady, N., & Rosenthal, R. (1992). Thin slices of expressive behavior as predictors of interpersonal consequences. consequences. A meta-analysis meta-analysis.. Psychological Bulletin , 111, 256±274. Argyle, M. (1988). Bodily communication. London: Methuen. Bruch, M. A., Giordano, S., & Pearl, L. (1986). Differences between fearful and self-conscious shy subtypes in background and adjustment. Journal of Research in Personality, 20, 172±186. Buss, D. M. (1992). Mate preference mechanisms: Consequences for partner choice and intrasexual com petition. In L. Cosmides, J. Tooby, & J. Barkow (Eds.), The adapted mind (pp. 249±266). Oxford: Oxford Univ. Press. Behavioral Science, 2, 30±45. Chance, M. R. A. (1957). The role of convulsions in behavior. Behavioral DePaulo, B. M. (1992). Non-verbal behavior and self-presentation. Psychological Bulletin , 111, 203±243. Edmonson, C., & Conger, J. (1985). The impact of mode of presentation on gender differences in social perception. Sex Roles, 32, 169±183. Eibl-Eibesfe Eibl-Eibesfeldt, ldt, I. (1995). (1995). Die Biologie des menschlichen Verhaltens. MuÈ nchen: Piper. Ellis, Ellis, B. J. (1992). The evolution evolution of sexual attraction: attraction: Evaluative Evaluative mechanisms in women. In L. Cosmides, Cosmides, J. Tooby, & J. Tooby (Eds.), The adapted mind (pp. 195±231). Oxford: Oxford Univ. Press. Garcia, S., Stinson, L., Ickes, W., & Bissonnette, V. (1991). Shyness and physical attractiveness in mixedsex dyads. Journal of Personality and Social Psychology, 61, 35±49. Givens, G. (1978). The non-verbal basis of attraction: Flirtation, courtship and seduction. Psychiatry, 41, 346±351. Goffmann, E. (1979). Gender advertisements. London: Macmillan. Grammer, K. (1990). Strangers meet: Laughter and non-verbal signs of interest in opposite sex encounters. Journal of Non-verbal Behavior , 14, 209±236. Grammer, K. (1992). Human courtship behavior: Biological basis and cognitive processing. In A. E. Rasa, C. Vogel, & E. Voland (Eds.), The sociobiology of sexual reproductive strategies (pp. 147±169). London: Chapman & Hall. Grammer, K., Fieder, M., & Filova, V. (1997). The communication paradox and possible solutions. In A. Schmitt, Schmitt, K. Atzwanger, Atzwanger, K. Grammer, Grammer, & K. SchaÈfer Èfer (Eds.), (Eds.), New aspects of human ethology (pp. 91±120). New York: Plenum. Gramme Grammer, r, K., Honda, M., Juette, Juette, A., & Schmit Schmitt, t, A. (1999) (1999).. Fuzzin Fuzziness ess of non non-ver -verbal bal courts courtship hip comcomJournall of Person Personali ality ty and Social Social Psychol Psychology ogy, munication munication unblurred by Motion Motion Energy Energy Detection. Detection. Journa 77 (3), (3), 487 487 ± 508. 508. Grammer, K., Kruck, K., & Magnusson, M. (1998). The courtship dance: Patterns of non-verbal synchronization in opposite sex encounters. Journal of Non-verbal Behavior , 22 (1), 3±29.
390
K. Grammer et al. / Evolution and Human Behavior 21 (2000) 371±390
Henley, N., & LaFrance, M. (1984). Gender as culture: Difference and dominance in non-verbal behaviour. In A. Wolfgang (Ed.), Non-verbal behaviour. Perspectives, applications, intercultural insights (pp. 351± 372). Lewiston, NY: Hogrefe. International onal Journal Journal of Behavioral Behavioral Hughes, M. (1978). Sequential analysis of exploration and play. Internati Development , 1, 83±97. Humphries, D. A., & Driver, P. M. (1970). Protean defense by prey animals. Oecologia, 5, 285±302. Ickes, W., Bissonnette, V., Garcia, S., & Stinson, L. (1990). Using and implementing the dyadic interaction paradigm. In C. Hendrick, & M. Clark (Eds.), Review of personality and social psychology: Research methods in personality and social psychology, vol. 11 (pp. 16±44). Newbury Park, CA: Sage. Kendon, A., & Ferber, A. (1973). A description of some human greetings. In R. P. Michael, & J. H. Crook (Eds.), Comparative ecology and behavior of primates (pp. 592±668). London: Academic Press. Kirkendall, L. (1961). Premarital intercourse and interpersonal relationships . New York: Julian Press. Krebs, Krebs, J. R., & Dawkins Dawkins,, R. (1984). (1984). Animal Animal sig signal nals: s: Min Mindre dreadi ading ng and manipu manipulat lation ion.. In J. R. Krebs, Krebs, Behaviora orall ecology ecology:: An evolut evolution ionary ary appro approach ach (2nd & N. B. Davi Davies es (Eds (Eds.) .),, Behavi (2nd ed., ed., pp. pp. 380± 380± 402). 402). Oxford: Oxford: Blackwell. Blackwell. Landolt, M. A., Lalumie re, M. L., & Quinsey, V. L. (1995). Sex differences in intra-sex variations in human Ethology and Sociobiolog Sociobiologyy, 16 , 3±25. mating tactics: An evolutionary approach. Ethology Losey, G. S. (1978). Information theory and communication. In P. W. Colgan (Ed.), Quantitative ethology (pp. 43±78). New York: Wiley. McCormick, N. B., & Esser, C. J. (1983). The courtship game. In E. R. Allgeier, & N. B. McCormick (Eds.), Changing boundaries: Gender roles and sexual behavior (pp. 64±86). Palo Alto: Mayfield. Miller, G. F. (1997). Protean primates: The evolution of adaptive unpredictability in competition and courtship. In A. Whiten, & R. W. Byrne (Eds.), Machiavellian intelligence: II. Extensions and evaluations (pp. 312±340). Cambridge: Cambridge Univ. Press. Moore, M. M. (1985). Non-verbal courtship patterns in women: context and consequences. Ethology and Sociobiology, 6 , 237±247. Moore, M. M. (1995). Courtship signaling and adolescents: Girls just wanna have fun? Journal of Sex Research, 32 (4), 319±328. Moore, M. M., & Butler, D. L. (1989). Predictive aspects of non-verbal courtship behavior in women. Semiotica, 3 (4), 205±215. Perper, T., & Fox, V.S. (1980). Flirtation and pickup pattern in bars. Paper presented at the Eastern Conference on Reproductive Behavior, New York, June 1980. Press, W. H., Teukolsky, S. A., Vetterling, W. T., & Flannery, B. P. (1992). Numerical recipes in Fortran (2nd ed.). Cambridge: Cambridge Univ. Press. Sabini, J., & Silver, M. (1982). Moralities of everyday life . Oxford: Oxford Univ. Press. Scheflen, A. E. (1965). Quasi-courtship behavior in psychotherapy. Psychiatry, 28, 245±257. Shannon, C. E., & Weaver, W. (1949). The mathematical theory of communication. Urbana: University of Illinois Press. Townsend, J. M., Kline, J., & Wasserman, T. H. (1995). Low-investment copulation: sex differences in Ethology and Sociobiolog Sociobiologyy, 16 , 25±51. motivation and emotional reactions. Ethology Trivers, R. L. (1972). Parental investment and sexual selection. In B. Campbell (Ed.), Sexual selection and the descent of man 1871±1971 (pp. 136±179). Chicago: Aldine.