Content uploaded by John P Swaddle
Author content
All content in this area was uploaded by John P Swaddle
Content may be subject to copyright.
doi: 10.1098/rspb.2002.2165
, 2285-2289269 2002 Proc. R. Soc. Lond. B
John P. Swaddle and Gillian W. Reierson
attractiveness in human males
Testosterone increases perceived dominance but not
References
http://rspb.royalsocietypublishing.org/content/269/1507/2285#related-urls
Article cited in:
Email alerting service
hereright-hand corner of the article or click
Receive free email alerts when new articles cite this article - sign up in the box at the top
http://rspb.royalsocietypublishing.org/subscriptions go to: Proc. R. Soc. Lond. BTo subscribe to
This journal is © 2002 The Royal Society
on May 24, 2011rspb.royalsocietypublishing.orgDownloaded from
Received
9 May 2002
Accepted
15 August 2002
Published online
9 October 2002
Testosterone increases perceived dominance but
not attractiveness in human males
John P. Swaddle
*
and Gillian W. Reierson
Biology Department, College of William & Ma ry, Williamsburg, VA 23197-879, USA
Recent evidence suggests that certain f eatures on the human face indicate hormonal levels during growth,
and that women judge the attractiveness of potential partners based on the appearance of these features.
One entrenched notion is male facial features that are affe cted by testosterone are used as direct cues in
mate preference. Testosterone may be particularly revealing as it is purported to be an honest indicator
of male tness. Increased testosterone may impose an immunocompetence handicap on the bearer and
only the best males can carry this hand icap. To date, tes ts of this theory have been indirect, and have
relied on digital manipulations that represent unrealistic continuums of masculine and feminine faces. We
provide a much more direct test by manipulating digitally male faces to mimic known shape variation,
caused by varying levels of testosterone through puberty. W e prod uced a continuum of faces that ranged
from low to high levels of testosterone in male faces and asked women to choose the points on the
continuum that appeared most attractive and most physically domin ant. Our data indicate that high testos -
terone faces reveal dominance. However, there is no evidence of directional selection for increased (or
decreased) testosterone in terms of attractiveness to the opposite sex. We discuss the relevance and appli-
cability of evolutionary interpretations of our data and, contrary to predictions, provide evidence of stabil-
izing selection acting on testosterone through mate preferences.
Keywords: sexual selection;
Homo sapiens
; immunocompetence; handicap; mate preference
1. IN TRODUCTION
Several hypotheses suggest that human faces display mar-
kers of hormones, and that these cues affect judgements
of facial attractiveness ( Johnston & Franklin 1993; Perrett
et al.
1998; Thornhill & Gangestad 1999; Johnston
et al.
2001) and dominance (Mazur
et al.
1994; Mazur & Booth
1998). In particular, evolutionary psychologists have gen-
erated theories to pred ict that levels of testosterone
re ected in male facial features are (or were) important in
generating sexual dimorphism in humans, as well as
directly affecting mate preferences (Perrett
et al.
1998;
Penton-Voak
et al.
1999; Thornhill & Gangestad 1999;
Johnston
et al.
2001). It has been suggested that facial
testosterone may be an honest indicator of male tne ss
(Thornhill & Gangestad 1999), as increased testosterone
can impose an immunocompetence handicap on the
bearer (see discussions in Grossman 1985; Hasselquist
et
al.
1999; Duckworth
et al.
2001) and only the best males
can carry this handicap (Zahavi 1975; Hamilton & Zuk
1982).
Existing tests of testosterone-related hypotheses of
human attractiveness have focused on producing femi-
nine–masculine axes of facial shape in digital represen-
tations (by computer graphics manipulations) and
monitoring people’s responses to these faces (Perrett
et al.
1998; Penton-Voak
et al.
1999; Johnston
et al.
2001). In
some cases , more masculine, male faces are preferred
(Penton-Voak
et al.
1999; Johnston
et al.
2001), whereas
in others more feminine male faces are judged as attractive
(Perrett
et al.
1998). Part of this variation appears to be
correlated with the stage of ovarian cycle of women judg-
*
Author for correspondence (jpswad@wm.edu).
Proc. R. Soc. Lond.
B (2002)
269
, 2285–2289 2285 Ó 2002 The Royal Society
DOI 10.1098/rspb.2002.2165
ing the faces (Penton-Voak
et al.
1999; Johnston
et al.
2001). However, there is no evidence to indicate that
these uctuations in female preferences affect actual part-
ner choice.
Previous experiments have also provided fairly un natu-
ral, and rather indirect, tests of the testosterone-related
hypotheses. First, there are many differences between
masculine and feminine faces, which are not accounted
for by testosterone alone (Perrett
et al.
1998). Hence, the
results of these studies are dif cult to interpret in terms
of testosterone expression in the male face. Second, the
faces produc ed through such manipulations result in male
facial shapes outside a natural male distribution (Penton-
Voak
et al.
1999; Johnston
et al.
2001), and are commonly
presented as faces without hairlines or necks (Perrett
et
al.
1998). Preferences for face shapes from such arti cial
distributions may not re ect preferences shown in society,
as the facial variation in normal male human populations
can be very different.
A more direct approach to testing testosterone-related
hypotheses is to produce facial variation that re ects
quanti ed variation in te stosterone levels. Therefore, we
manipulated the shape of male faces to mimic growth dif-
ferences resulting from varying levels of testosterone dur-
ing puberty. Women viewed these faces and made
judgements of facial attractiveness and dominance.
According to previou s reports, we predicted that women
should nd face s with higher testosterone expression most
attractive ( Thornhill & Gangestad 1999).
2. MATERIAL AND METHODS
We took digital photographs of 21 male Caucasians’ faces in
a (right-side) pro le and face-on orientation to the camera.
Models were 18–21 years old, had short hair, lacked beards or
on May 24, 2011rspb.royalsocietypublishing.orgDownloaded from
2286 J. P. Swaddle and G. W. Reie rson Testosterone and facial attractiveness
Table 1. Approximate relative trait size changes (i.e. distances between two points) resulting from w arp manipulations for the
100% high and 100% low testosterone treatments, for pro le and face-on photographs of male faces.
(Facial characters were determined from anthropometric de nitions (Farkas 1981) and their pos ition was estimated on each male
face. Points used were: nasion (N) midpoint of both nasal root and nasofrontal suture; pogonion (POG) anterior midpoint of
the chin; gonion (GO) most lateral point of the mandibular angle; menton (ME) lowest point on the mandible; ans (ANS)
approximated by the posterior extremity of the nose; articulare (ART) estimated by the posterior and ventral atta chment of the
ear; sella (S) estimated by the anterior and ventral attachment of the ear.)
pro le face-on
trait 100% high 100% low 100% high 100% low
N–ANS
1
0.01
2
0.01
1
0.01
2
0.01
N–ME
1
0.01
2
0.01
1
0.01
2
0.01
ART–ANS
1
0.01
2
0.01 0 0
ART–POG
1
0.02
2
0.02
1
0.02
2
0.02
ART–GO
1
0.04
2
0.04
1
0.04
2
0.04
S–ME
1
0.02
2
0.02
1
0.02
2
0.02
ART–ME
1
0.02
2
0.02
1
0.02
2
0.02
GO–POG
1
0.02
2
0.02
1
0.01
2
0.01
moustaches, were not wearing jewellery, and were told to adopt
a neutral facial expression while sitting for the photographs.
Photographs were taken in standardized lighting conditions
against a common background.
We compared how facial s tructure changes with natural levels
of testosterone during puberty, and in delayed puberty boys
treated with low-dose testosterone (Verdonck 1997; V erdonck
et al.
1999) to create vectors of facial trait changes that represent
realistic variation in plasma testosterone in uencing facial bone
growth elds (Enlow 1990; Silveira
et al.
1992). The relative
changes of the facial traits we manipulated are listed in table 1.
Brie y, face height increased and the lower jaw became larger
with increased testosterone. These manipulations were speci -
cally de signed to isolate the effects of te stosterone on facial skel-
etal characters, providing a more direct test of the effec ts of
testosterone on perception of the face.
We used the warp functi on in Gryphon Software’s Morph
program to alter the shape of male faces according to the trait
size c hanges listed in table 1. We also digitized 10 points around
the eyes, ve on the nose, and ve on the lips to ensure that this
part of the face did not change under the warp manipulation.
The warp function moves the reference points to a speci ed pos-
ition without altering colour or tone ( gure 1). In addition to
producing the high and low testosterone treatments, we pro-
duced nine, equally spaced, intermediate warps between the
control and high testosterone faces, and between the control and
low testosterone faces. This resul ted in 2 1 representations of a
single face that ran ged from the low treatment to the high treat-
ment with equal warp differences betwe en all images, and where
the median face was the control (i.e. non-manipulated face).
The sequence of 21 representations was arranged in ascending
or descending order in a PowerPoint presentation, with one
image on each slide. For half the faces the rst slide contained
the lowest testosterone treatment; for the remaining faces the
rst slide was the highest testosterone treatment. Each slide also
displayed a sequential number so that viewers could tell us
which slide they chose.
Thirty females, age 18–21 years old, viewed each of the 42
sequences on a computer screen. Subjects were asked to choose
the most sexually attractive, and most physically dominant look-
ing, face in each sequence. One of the experimenters advanced
(or went back through) the slides at a regular pace of one slide
Proc. R. Soc. Lond. B (2002)
approximately every 0.5 s. In ad dition, viewe rs instructed the
experimenter to advance, reverse, and stop the sequence when-
ever the viewer wished. The order of male faces viewed by
females was random ized, except that all the pro le faces were
presented in sequence, as were the face-on presentations. Half
of the viewers saw the pro le sequences rst, and vice versa.
Viewers were paid for their participation. If a viewer recognized
one of the male faces, she informed the researchers and the data
for that face were discarded.
3. RESULTS AND DISCUSSION
The frames chosen by females as being most attractive,
for both face-on and pro le views, were faces with testos-
terone expression very similar to that of the non-manipu-
lated face (paired
t
-tests of the most attractive versus the
control face: face-on,
t
2 0
= 1.94,
p
= 0.067; pro le,
t
2 0
= 1.86,
p
= 0.078). The frequency diagram of most
attractive frame choice illustrates that women’s prefer-
ences centre on existing levels of testosterone in men’s
faces ( gure 2). S uch distributions are consistent with sta-
bilizing selection acting on testos terone expression in
men’s faces rather th an the previously proposed direc-
tional selection for reduced or increased testosterone
(Perrett
et al.
1998; Johnston
et al.
2001).
It is important to point out that individual female raters
used the entire range of faces for attractivenes s choices.
For face-on views, the mean ± s.e. (among-rater) range of
frame chosen was 18.63 ± 0.54 (out of a possible 21) for
face-on views, and 19.50 ± 2.54 for pro le views. This
indicates that females did not feel constrained to only
using the central portion of the animation sequence s, and
that our result is not an artefact of a rater central tendency
independent of attractiveness judgements.
In order to explore further the effects of testosterone on
facial attractivenes s, we calculated an index of testosterone
in non-manipulated male faces (the ratio of AR T–POG
length to N–S length, where a high value indicates higher
facial testosterone expression; see table 1 for de nitions)
and examined whether natural variation in testosterone
affected choice of frames. Linear measurements were
taken directly from the digital photographs of pro le faces.
on May 24, 2011rspb.royalsocietypublishing.orgDownloaded from
Testosterone and facial attractiveness J. P. Swaddle and G. W. Reierson 2287
frame 1
low
frame 11
control
frame 21
high
(i)
(ii)
(iii)
(i)
(ii)
(iii)
Figure 1. The extremes of the manipulations p roduced for (
a
(i)–(iii)) face-on and (
b
(i)–(iii)) pro le views of the same face.
Low, low testosterone treatment; control, unmanipulated face; high, high testosterone treatment.
Natural levels of facial testosterone were negatively corre-
lated with choice of attractive frame (face-on: Pearson
r
=
2
0.636,
n
= 21,
p
= 0.0011; pro le:
r
=
2
0.710,
n
= 21,
p
= 0.000 15). Inspection of the numerical values
of these correlations indicated that the most preferred ver-
sions of naturally high testosterone faces were those that
reduced apparent testosterone. Whereas the most pre-
ferred representations of naturally low testosterone faces
were those that increased apparent testosterone. The net
effect of such pre ferences is a convergent selection press-
ure on average testosterone expression across the popu-
lation, which provides further evidence of stabilizing
selection on tes tosterone. Although these correlations are
strong, we need to emphasize that there is a possibility
that raters’ preferences could alter with their oestrus cycle
(cf. Penton-Voak
et al.
1999; Johnston
et al.
2001). In
further studies, it would be interesting to investigate
whether females’ preferences for facial testosterone change
systematically according to the possibility of conception.
The choice of the most dominant frame was very differ-
ent to the choice of the most attractive frame. In agree-
ment with previous evidence (Perrett
et al.
1998; Johnston
et al.
2001), the most dominant frames were those with a
high degree of testosterone expression ( gure 2). This pat-
tern he lps to co n rm that our manipulation was realistic,
as links betwee n testosterone and male aggression and
dominance are generally well supported (Mazur &
Booth 1998).
It was also evident th at choices made by viewers when
judging males face-on were different to when males were
in pro le. Females chose faces signi cantly higher in tes-
tosterone as most attractive when viewing a face in pro le,
Proc. R. Soc. Lond. B (2002)
compared with the same male face-on (
t
2 0
= 3.16,
p
= 0.0049). Conversely, females chose faces signi cantly
lower in testosterone as most dominant for pro le, versus
face-on, views of males (
t
2 0
= 3.65,
p
= 0.0016). These dif-
ferences clearly demo nstrate that female judgement of a
male face changes with viewing angle. This implies that
further experiments need to account for three- dimensional
viewing of faces if re searchers are going to document
realistic relations between facial characters and socially (or
evolutionarily) important parameters. However, in the
context of our experiment, we observed the same general
relationships between attractiveness (or dominance) and
facial testosterone expression in pro le and face-on.
A recent resurgence in adaptive explanations of human
behaviour and morphology has stimulated interest in
whether researchers can demonstrate evolutionarily
important relations hips through preference and rating
experiments (Perrett
et al.
1998; Scheib
et al.
1999; John-
ston
et al.
2001). To demonstrate a current evolutionary
effect a research programme must meet certain criteria. In
the context of this study, we would need to demonstrate
evidence of selection pres sure acting on traits, and heri-
table variation in those traits. In terms of the former, o ur
data could indicate stabilizing selection if the re is a link
between differential reproductive success and rating of
facial attractiveness. There is evidence that facial attract-
iveness positively affects the ways in which others treat
them in many social contexts (Kalick 1988; Zebrowitz
1997). However, links between facial attractiveness and
health or reproductive success are equivocal or non-
existent (Kalick
et al.
1998; Shackelford & L arsen 1999;
Rhodes
et al.
2001). Women may place less importance
on May 24, 2011rspb.royalsocietypublishing.orgDownloaded from
2288 J. P. Swaddle and G. W. Reie rson Testosterone and facial attractiveness
6
5
4
3
2
1
0
7
6
5
4
3
2
1
0
1
3
5
7
9 11 13 15 17 19 21
frame chosen
frequency frequency
a
)
b
)
Figure 2. (
a
) Frequency histogram for frame choice of most
attractive (open bars) and most dominant ( lled bars) frames
for face-on views; (
b
) and for pro le v iews of male faces. In
both distributions, the choice of dominant frame represented
a much higher degree of testosterone expression than the
choice of attractive frame (face-on,
t
20
=
11.75,
p
,
0.001;
pro le,
t
20
=
3.05,
p
=
0.0064).
on physical attractiveness when choosing a mate than men
do (Waynforth 2001). This implies a low probability that
male facial attractiveness will relate to reproductive suc-
cess or tness. However, it does not preclude that attract-
iveness was related to reproductive success at some time
in our evolutionary history. There is some indication that
increased facial dominance is associated with increased
sexual activity in you ng men (Mazur
et al.
1994), but
increased sexual activity does not nec essarily result in
increased numbers of offspring (Mueller & Mazur 1998).
Additionally, extreme dominance can negatively in uence
reproductive success (Mueller & Mazur 1998).
In terms of the ‘heritability’ c riteria, there is some e vi-
dence of signi c ant heritability of the male facial traits we
assessed (upper estimates range from 10% to 36%
heritability) (Enlow 1990) and of circulating testosterone
levels in men (upper estimates of
ca
. 60% heritability)
(Harris
et al.
1998). Overall, interpreting our experiment
and those using similar methodologies (Perrett
et al.
1998;
Penton-Voak
et al.
1999; Scheib
et al.
1999; Johnston
et
al.
2001) in terms of evolutionary processes is problem-
atic, and generally unsupported by empirical evidence. It
is more likely that judgements of particular facial shapes
are mediated by cultural norms and individual decision-
making (Tooby & Cosmides 1989).
Even if some researchers insist on evolutionary
interpretations, we cannot nd evidence for directional
Proc. R. Soc. Lond. B (2002)
selection to red uce or increase testosterone through female
preferences. Previous studies that have documented
female preference s for masculine ( Johnston
et al.
2001) or
feminine (Perrett
et al.
1998) male faces cann ot be inter-
preted in terms of the effects of testosterone alone. Other
hormonal differences between the sexes must (in part)
explain previously d ocumented preferences. It may be that
selection acting on oestrogen and on female faces may
equally explain human evolutionary changes (Thornhill &
Grammer 1999). The comparative literature has demon-
strated that both sexes can be the target of directional sex-
ual selection and that both sexes change through
evolutionary time, resulting in dimorphism (Karubian &
Swaddle 2001). Our data also suggest t hat social domi-
nance may underlie the evolution of human sexual dimor-
phism, as increased testosterone is positively associated
with perceived dominance. However, increased perceived
dominance does not enhance male attractiveness.
The authors thank Adam Rubenstein and three anonymous
reviewers for helpful comments on an earlier draft, as well as
the 21 male and 30 female students of William & Mary who
agreed to take part in this study. J.P.S. was supported by the
NSF (IBN-0133795) and the Thomas F. and Kate Miller
Jeffress Memorial Trust. G.W.R. was supported by the
Howard Hughes Medical Institute.
REFERENCES
Duckworth, R. A., Mendonca, M. T. & Hill, G. E. 2001 A
condition dependent link between testosterone and disease
resistance in the house nch.
Proc. R. Soc. Lond.
B
268
,
2467–2472. (DOI 10.1098/rspb.2001.1827.)
Enlow, D. H. 1990
Facial growth
. Philadelphia, PA: Har-
court Brace.
Farkas, L. 1981
Anthropometry of the head and face in medicine
.
New York: Elsevier.
Grossman, C. J. 1985 Interactions between the gonadal ster-
oids and the immune system.
Science
227
, 257–261.
Hamilton, W. D. & Zuk, M. 1982 Heritable true tness and
bright birds: a role for parasites?
Science
218
, 384–387.
Harris, J. A., Vernon, P. A. & Boomsma, D. I. 1998 The heri-
tability of testosterone: a study of Dutch adolescent twins
and their parents.
Behav. Genet.
28
, 165–171.
Hasselquist, D., Marsh, J. A., Sherman, P. W . & Wing eld,
J. C. 1999 Is avian humoral immunocompetence suppressed
by testosterone?
Behav. Ecol. Sociobiol.
45
, 167–175.
Johnston, V. S. & Franklin, M. 1993 Is beauty in the eye of
the beholder?
Ethol. Sociobiol.
14
, 183–199.
Johnston, V. S., Hagel, R., Franklin, M., Fink, B. & Grammer,
K. 2001 Male facial attractiveness: evidence for hormone-
mediated a daptive design.
Evol. Hum. Behav.
22
, 251–267.
Kalick, S. M. 1988 Physical attractiveness as a status cue.
J.
Exp. Soc. Psychol.
24
, 469–489.
Kalick, S. M., Zebrowitz, L. A., Langlois, J. H. & Johnson,
R. M. 1998 Does human facial attractiveness honestly adver-
tise health? Longitudinal data on an evolutionary question.
Psychol. Sci.
9
, 8–13.
Karubian, J. & Swaddle, J. P. 2001 Selection on females can
create ‘larger males’.
Proc. R. Soc. Lond.
B
268
, 725–728.
(DOI 10.1098/rspb.2000.1407.)
Mazur, A. & Booth, A. 1998 Testosterone and dominance in
men.
Behav. Brain Sci.
21
, 353–397.
Mazur, A., Halpern, C. & Udry, J. R . 1994 Domina nt looking
male teenagers copulate earlier.
Ethol. Sociobiol.
15
, 87–94.
Mueller, U. & Mazur, A. 1998 Reproductive constraints on
dominance competition in male
Homo sapiens
.
Evol. Hum.
Behav.
19
, 387–396.
on May 24, 2011rspb.royalsocietypublishing.orgDownloaded from
Testosterone and facial attractiveness J. P. Swaddle and G. W. Reierson 2289
Penton-Voak, I. S., Perrett, D. I., Castles, D. L., Kobayashi,
T., Burt, D. M., Murray, L. K. & Minamisawa, R. 1999
Menstrual cycle alters face preference.
Nature
399
, 741–742.
Perrett, D. I., Lee, K. J., Penton-Voak, I., Rowland, D., Yoshi-
kawa, S., Burt, D. M., Henzl, S. P., Castles, D. L. & Akam-
atsu, S. 1998 Effects of s exual dimorphism on facial
attractiveness.
Nature
394
, 884–887.
Rhodes, G., Zebrowitz, L. A., Clark, A., Kalick, S. M., High-
tower, A. & McKay, R. 2001 Do facial averageness and sym-
metry signal health ?
Evol. Hum. Behav.
22
, 31–4 6.
Scheib, J. E., Gangestad, S. W. & Thornhill, R. 1999 Facial
attractiveness, symmetry and cues of good genes.
Proc. R.
Soc. Lond.
B
266
, 1913–1917. (DOI 10.1098/rspb.1999.
0866.)
Shackelford, T. K. & La rsen, R. J. 1999 Facial attractiveness
and physical health.
Evol. Hum. Behav.
20
, 71–76.
Silveira, A. M., Fishman, L. S., Subteiny, J. D. & Kassebaum,
D. K. 1992 Facial growth during adolescence in early, aver-
age and late maturers.
Angle Orthod.
52
, 185–190.
Thornhill, R. & Gangestad, S. W. 1999 Facial attract iveness.
Proc. R. Soc. Lond. B (2002)
Trends Cogn. Sci.
3
, 452–460.
Thornhill, R. & Grammer, K. 1999 The body and face of
woman: one ornament tha t signals quality ?
Evol. Hum.
Behav.
20
, 105–120.
Tooby, J. & Cosmides, L. 1989 Evolutionary psychology and
the generation of culture. Part I: theoretical considerations.
Ethol. Sociobiol.
10
, 29–4 9.
Verdonck, A. 1997
Effect of testosterone on craniofacial growth in
the human and the rat
. Leuven: Katholieke Universiteit te
Leuven.
Verdonck, A., Gaethofs, M., Carels, C. & de Zegher, F. 1999
Effect of low-dose testosterone treatment on craniofacial
growth in boys with delayed puberty.
Eur. J. Orthod.
21
,
137–143.
Waynforth, D. 2001 Mate choice trade-offs and women’s pref-
erence for physically attractive men.
Hum. Nat.
12
, 207–219.
Zahavi, A. 1975 Mate selection: a selection for a handicap.
J.
Theor. Biol.
53
, 205–214.
Zebrowitz, L. A. 1997
Reading faces: windows to the soul?
Boulder, CO: Westview Press.
on May 24, 2011rspb.royalsocietypublishing.orgDownloaded from
