Human (Homo sapiens) Facial Attractiveness and Sexual Selection: The Role of Symmetry and Averageness

Abstract

We hypothesized from the parasite theory of sexual selection that men (Homo sapiens) would prefer averageness and symmetry in women's faces, that women would prefer averageness and symmetry in men's faces, and that women would prefer largeness (not averageness) of the secondary sexual traits of men's faces. We generated computer images of men's and women's faces and of composites of the faces of each sex, and then had men and women rate opposite-sex faces for 4 variables (attractive, dominant, sexy, and healthy). Symmetry, averageness, and the sizes of facial features were measured on the computerized faces. The hypotheses were supported, with the exception of the hypothesized effects of averageness of female and male faces on attractiveness ratings. This is the first study to show that facial symmetry has a positive influence on facial attractiveness ratings.

Full text

Journal
of
Comparative
Psychology
1994, Vol. 108,
No. 3,
233-242
Copyright
1994
by the
American
Psychological
Association,
Inc.
0735-7036W/$3.00
Human (Homo
sapiens}
Facial
Attractiveness
and
Sexual
Selection:
The
Role
of
Symmetry
and
Averageness
Karl
Grammer
and
Randy
Thornhill
We
hypothesized
from
the
parasite
theory
of
sexual
selection
that
men
(Homo sapiens) would
prefer averageness
and
symmetry
in
women's faces, that women would prefer averageness
and
symmetry
in
men's
faces,
and
that women would prefer largeness (not averageness)
of the
secondary sexual traits
of
men's faces.
We
generated computer
images
of
men's
and
women's
faces
and of
composites
of the
faces
of
each sex,
and
then
had men and
women
rate
opposite-sex
faces
for 4
variables (attractive, dominant, sexy,
and
healthy). Symmetry, averageness,
and the
sizes
of
facial features were measured
on the
computerized faces.
The
hypotheses were supported,
with
the
exception
of the
hypothesized
effects
of
averageness
of
female
and
male
faces
on
attractiveness ratings.
This
is the first
study
to
show that facial symmetry
has a
positive
influence
on
facial attractiveness ratings.
Although
adult facial attractiveness ratings
are
replicable,
even
cross-culturally
(see reviews
and
discussions
in
Jones
&
Hill, 1993,
and
Langlois
&
Roggman,
1990), there
has
been considerable controversy around attempts
to
identify
in
research
the
facial features that actually cause faces
to be
judged attractive
or
unattractive.
As
discussed
by
Langlois
and
Roggman, studies
of
individual
facial
features
(e.g.,
nose size)
often
have yielded inconsistent results between
studies. Faces created
by
combining individual faces into
composites have been shown
to be
more attractive than
the
individual
faces, which
is
felt
to be a
preference
for
average
facial
features (Langlois
&
Roggman, 1990; Symons,
1979). Averageness
effaces
can be
calculated metrically
or
constructed
photogrammetrically.
Gallon
(1879) con-
structed composites
of
individual pictures with
the
photo-
graphic
method
of
simply projecting them
one
over
the
other
on a
negative. According
to
Gallon,
this method
"enables
us to
obtain with mechanical precision
a
general-
ized picture;
one
that represents
no man in
particular,
but
portrays
an
imaginary
figure
possessing
the
average features
of
any
given group
of
man"
(1879,
p.
341).
Indeed,
Treu
(1914)
had the
impression that these composites
are
"sin-
gularly
beautiful"
(p.
441). However,
as
Alley
and
Cunning-
ham
(1991;
see
also Benson
&
Perrett,
1991) pointed out,
composites
are
also more symmetrical
and
rather
free
of
Karl
Grammer,
Ludwig
Boltzmann
Institute
for
Urban
Ethology,
Vienna,
Austria;
Randy
Thornhill,
Department
of
Biology,
Uni-
versity
of New
Mexico.
We
thank
Klaus
Atzwanger,
John
Dittami,
Steve
Gangestad,
Joy
Ingram,
Kurt
McKean,
and Don
Symons
for
helpful
comments
on
the
research.
Randy
Thornhill
thanks
Klaus
Atzwanger
and
John
Dittami
for
their
hospitality
in
Vienna.
Anne
Rice's
help with
the
preparation
of the
article
is
appreciated.
Correspondence
concerning
this
article
should
be
addressed
to
Karl
Grammer,
Ludwig
Boltzmann
Institute
for
Urban
Ethology,
Althanstrasse
14,
A-1090
Vienna,
Austria
or to
Randy
Thornhill,
Department
of
Biology,
University
of New
Mexico,
Albuquerque,
New
Mexico
87131-1091.
Electronic
mail
may be
sent
to
rthorn@carina.unm.edu.
facial
blemishes,
and
therefore
one of
these traits, rather
than averageness
per se, may be the
cause
of the
enhanced
attractiveness
of
composites.
In
addition, Alley
and
Cun-
ningham
emphasize that there
is
considerable evidence
(e.g.,
Keating, 1985) that
the
most sexually attractive male
faces
are
those that show extremes,
not
averageness,
in
certain features (e.g., wide jaw),
felt
to be
perceived
as
dominance indicators.
Our
study
of the
cross-sex attractive-
ness ratings
of
faces
of
both sexes attempts
to
clarify
adult
facial
beauty
by
examining
the
roles
of
facial symmetry
and
averageness
and
their interaction
with
individual dimen-
sions
of the
face.
We
use the
theoretical framework
of
sexual selection.
The
prominent theory
of
sexual selection
is the
parasite theory,
which proposes that sexual selection favors those traits that
advertise resistance
to
parasites, both
microparasites,
such
as
bacteria
and
viruses,
and
macroparasites,
such
as
nema-
todes
and
protozoa (Hamilton
&
Zuk, 1982). There
is
con-
siderable evidence that parasite-resistant organisms
win in
competition
for
mates, both
in
intrasexual
competition (usu-
ally males competing
for
females)
and in
being
chosen
by
the
opposite
sex and
that secondary sexual traits advertise
parasite resistance (see partial reviews
in
Hausfater
&
Thornhill, 1990,
and
Zuk,
1992).
According
to the
parasite
theory
of
sexual selection, mate choice decisions include
medical examinations
of
potential mates,
and
parasite-
resistant
organisms
are
preferred
because
they
produce
ge-
netically
resistant
offspring
or
provide better parental care
to the
offspring.
Thus,
the
parasite theory proposes that
beauty
of
bodily
form
is
perceived
as a cue to
high parasite
resistance
by
animals
in
choosing mates.
Secondary sexual characters
are
evolved outcomes
of
sexual
selection. There
is a
link between parasite resistance
and
secondary sexual traits because
sex
hormones, espe-
cially
testosterone, lower
immunocompetence
(Folstad
&
Karter,
1992;
Wedekind,
1992). Whereas high
liters
of
testosterone
are
necessary
for the
production
of
large sec-
ondary
sexual trails, there
will
necessarily
be a
positive
correlation between developmenl
of
secondary sexual
233

234
KARL
GRAMMER
AND
RANDY
THORNHILL
traits
and the
quality
of the
immune system; only healthy
organisms
can
afford
the
high testosterone handicap
on
the
immune system that
is
necessary
for the
production
of
elaborate sexual traits (Folstad
&
Karter,
1992).
The hu-
man
face contains secondary sexual traits, that
is,
facial
features
that develop
or
increase
in
size
at
puberty under
the
influence
of the sex
hormones,
androgens
and
estro-
gens. Enlarged jaws, chins,
and
cheekbones
in men are
examples
of
facial secondary sexual traits that
are
influ-
enced
by
testosterone
(Enlow,
1990; Tanner, 1978),
and
Thornhill
and
Gangestad
(1993) hypothesized that large-
ness
in
these features
are
considered sexually attractive
because
of
advertised immunocompetence.
Genetic diversity
may be an
important defense against
parasites, both
at the
between-organisms level (i.e.,
the
population
level)
and at the
within-organisms level (see
review
in
Thornhill
&
Gangestad, 1993;
see
also Hamilton,
1982; Tooby, 1982). Within-organisms genetic diversity
is
dependent
on
individual genetic
heterozygosity.
For
herita-
ble
traits that
are
continuously distributed, heterozygosity
correlates positively
with
average trait expression
(Soule
&
Cuzin-Roudy,
1982). Facial attractiveness
is
continuously
distributed
and
probably
is
heritable (see Thornhill
&
Gangestad, 1993). This implies that average values
of
facial
features
reflect high heterozygosity. Thus, Thornhill
and
Gangestad hypothesized that facial averageness
is
attractive
because
of its
association with heterozygosity
and
thus
parasite
resistance. This hypothesized pattern
may
primarily
apply
to
female faces.
The
intrasexual sexual competition
component
of
sexual selection involving dominance
and
combat
has
worked more strongly
on
males than
on
females
in
human evolutionary history (Darwin,
1871;
Symons,
1979),
and
male faces have multiple, testosterone-facilitated
secondary sexual traits (viz., adult male chin, cheekbones,
brow
ridge, and
jaw) that
are
expected
by the
parasite theory
to
have evolved
to
signal health-related dominance
by
large-
ness.
Multiple male facial features
in
which dominance
is
signalled
by
enlarged features
may
take precedence over
averageness
in
men's facial attractiveness
to
women.
Symmetry
of
bilaterally represented traits
is
positively
correlated with heterozygosity
in
many animals, including
humans (see review
in
Thornhill
&
Gangestad,
1993).
Thus,
facial
symmetry, like facial averageness,
may
display
un-
derlying
heterozygosity
and
parasite resistance (Thornhill
&
Gangestad, 1993). Also, body bilateral symmetry seems
to
reflect
overall quality
of
development, especially
the
ability
of
an
organism's
developmental machinery
to
resist genetic
perturbations
and the
numerous environmental perturba-
tions that occur during development
(Leary
&
Allendorf,
1989; Parson, 1990, 1992;
see
review
in
Watson
&
Thorn-
hill,
1994),
which implies that
a
symmetrical face displays
developmental homeostasis (Thornhill
&
Gangestad, 1993).
The
symmetry
of
bilateral secondary sexual characteristics
is
more sensitive
to
environmental perturbations during
their
development than
is
that
of
nonsexually
selected
bi-
lateral traits (see review
in
M011er
&
Pomiankowski,
1993).
This pattern
of
greater sensitivity
of
secondary sexual traits
is
seen
in
diverse animal taxa, including
nonhuman
primates
(Manning
&
Chamberlain, 1993). Parasites have been
shown
to
affect
differentially
the
development
of
symmetry
in
organisms' traits,
and the
symmetry
of
secondary sexual
traits
is
most negatively influenced (M011er, 1992; M011er
&
Pomiankowski, 1993; Watson
&
Thornhill, 1994). Symme-
try
of
facial secondary sexual traits
may
display immuno-
competence because
the
construction
of
such traits, espe-
cially large ones,
is
expected
to
require more
sex
hormone
(Thornhill
&
Gangestad, 1993).
In
this article
we
test
the
following hypotheses about
facial
attractiveness, which arise
from
these considerations:
(a) Men
prefer averageness
and
symmetry
in
women's
faces;
(b)
women prefer averageness
and
symmetry
in
men's faces;
and (c)
women prefer extreme expression
of
the
secondary sexual traits
of
men's faces.
Our
approach
uses
computer techniques
to
measure composite
and
indi-
vidual
faces
to
assess
the
influence
of
averageness, symme-
try,
and
facial dimensions
in
facial attractiveness judgments.
Method
Rating
Study
The
raters
of
computerized
faces
were
52
female
and 44
male
college
students
(Homo sapiens) from
a
German
university.
The
digitized
facial
pictures
were
presented
to
each
subject
individu-
ally
by an
interactive
computer
program.
Order
effects
of
presen-
tation
were
controlled
by
presenting
faces
to
raters
in a
randomized
order.
In a first run the
program
showed
all
pictures
to
each
subject
sequentially.
In a
second
run the
subject
rated
each
picture
for
11
adjectives
on a
rating
scale
from
1
(least)
to 7
(most).
The
adjectives
and
their
accompanying
rating
scales
were
displayed
randomly
with
each
picture
at the
bottom
of the
computer
screen.
The
subjects
were
allowed
to
view
pictures
as
long
as
they
wished.
For
this
study
only
4 of the
11
adjectives
were
used:
attractive,
dominant, sexy,
and
healthy.
The
subjects
were
trained
interac-
tively
by the
computer
before
making
ratings
until they
were
able
to
solve
the
rating task.
Generation
of
Individual
Test
Photographs
The
photographed
subjects were different
persons
from
the
raters.
Each
subject
(16
women,
age
M
± SD =
26.6
years
±
4.0,
and
16
men,
age M ± SD =
25.3
years
±
3.8)
was
seated
upright
in
a
chair
with
a
light
source
on
each
side
of the
face
in
order
to
prevent
shading.
We
carefully
positioned
each
subject
so
that
all
were
looking
directly
into
the
camera
without
any
tilt
of the
head.
Distance
to the
camera
was
constant.
The
picture
was
taken
with
a
high-resolution
video
camera
and
digitized
on a
computer.
The
picture
size
was 600 X 570
pixels
with
a
resolution
of 72
pixels/
2.54
cm.
Faces
were
standardized
in
size
on a
video
screen
by
cross-hairs
that
marked
the
horizontal
midline
of the
mouth,
the
horizontal
midline
of the
inner
and
outer
eye
corners,
and two
vertical
lines
at the
center
of the
pupil.
Faces
were
adjusted
to fit
in
the
cross-hairs
with
the
zoom
lens
of the
video-camera
(Figure
1;
also
see
Langlois
&
Roggman,
1990).
Generation
of
Composites
The 16
pictures
of
each
sex
were
randomly
paired,
and a
blended
gray
value
was
calculated
for the
corresponding
pixels
of
each
pair.
For
this
calculation
the
arithmetic
drawing
mode
blend

FACIAL ATTRACTIVENESS: SYMMETRY
AND
AVERAGENESS
235
Figure
1.
Pictorial
of
technique used
to
standardize facial size.
of
ColorQuickDraw
(part
of
System
7,
Apple Computers, Cuper-
tino,
CA) was
used.
Blend
applies
the
following formula
to the two
source pixels
in a
eight-bit drawing environment:
New
Pixel
=
Source
1 X
Weight
H-
65536
+
Source
2 X (1 -
Weight
H-
65536)
(1)
In
an
eight-bit drawing environment, there
are 256
gray values
and
the
weight
of
each photo
was set to
50%.
The
gray value
of the
blended pixel corresponds
to the
arithmetic mean
of the
gray
values
of the
original pair
of
pixels. Because
the
composites
created
by
this technique appear blurred,
the
original photos were
also
blurred
by
randomly introducing pixels. Finally,
the
gray
values
of the
composites were equalized
and
adapted
in
brightness
and
contrast
to the
original photos. This procedure presumably
controls
for the
strong
effect
of
skin texture
on
attractiveness
ratings (Benson
&
Perrett,
1991). This
was an
important control
in
our
study because
we
were trying
to
examine
the
effect
of
other
factors,
especially symmetry,
in
attractiveness.
For
each
sex
this generated eight composites, each constituted
from
two
originals
of
each
sex
(Figure
2). By
combining these
composites randomly
in
pairs,
we
created additional composites.
The final
test
set of
photographs
for
each
sex
consisted
of 16
individual
faces,
4
pictures combining
the
faces
of
four
persons,
2
pictures
combining
eight
faces,
and one
picture combining
all 16
individual
photos (Figure
3).
Thus,
the
test
set had 23
pictures
for
each
sex.
Facial
Measurements
Measurement
points.
The
facial
measurements were done
on
the
computer screen
with
IMAGE
1.41
(shareware, National
In-
stitute
of
Mental Health,
Bethesda,
MD) at 76% of the
original
photo. This program measures
and
saves
the
coordinates
of a
selected pixel.
We
used
13
points, which were defined
by
distinct
morphological
structures
of the
face
and
thus could
be
identified
reliably (Figure
4). The
reliability
of
these points
was
examined
by
two
methods.
First,
one of us
placed
the
points
on
several faces
three
different
times; point locations
did not
differ
by
more than
one
pixel. Second,
a
person unfamiliar with
the
research's
hypoth-
eses
placed
the
points
on
each
of
113
faces,
and
this
was
repeated
by
one of us
without
knowledge
of the
points
placed
by the
naive
assistant.
The
point locations
had
very high reliability:
The
zero-
order correlation between
the
facial symmetries calculated
from
the
points
of the two
raters
was .80 (p <
.0001).
Points used
included
the
outermost
(PI
and P2) and the
innermost
eye
corners
(P3 and
P4).
The
points
for
measuring
the
cheekbones
(P5
and P6)
were defined
as the
leftmost
and rightmost
pixels
of the
face
on a
horizontal
line beneath
the
eyes.
A
comparable definition
was
made
for the
points
for
measuring
the
nose:
P7 and P8
describe
the
leftmost
and rightmost
point
of the
nose
in the
lower nose region.
Jaw
width
(P9 and
P10)
was
measured
as
face width
at the y
coordinate
of the
mouth corners
(PI
1 and
P12).
A final
point
was
the
chin
boss
(PI
3), or the
lowest
point
of the
chin curvature.
In
persons with
a
W-shaped chin
boss,
the
middle point
was
used.
Calculation
of
asymmetry.
We
only dealt with horizontal
asymmetry
in
this study. Facial asymmetry
was
calculated
in two
ways.
Overall facial asymmetry (FA)
was
based
on the sum of all
possible
nonredundant
differences between
the
midpoints
of six
horizontal lines between
the
following pairs
of
points:
P1-P2,
P3-P4, P5-P6, P7-P8, P9-P10,
and
P11-P12.
These
six
lines
were
designated
Dl,
D2, D3, D4, D5, and D6,
respectively (Figure
5).
The
midpoint
of
each line
was
calculated with
the
formula,
([Left
Point
-
Right Point]
+
2) +
Right Point.
On a
perfectly
symmetrical
face,
all
midpoints
lie on the
same vertical line,
and
the sum of all
possible nonredundant midpoint differences
is
zero.
The
second measure
of
facial symmetry, which
we
call central
facial
asymmetry
(CFA),
focuses
on the
differences between mid-
points
of
adjacent lines, especially
in the
center
of the
face.
CFA
corresponds
to the sum of the
differences
of the
midpoints
of the
lines
Dl and D2, D2 and D3, D3 and D4, D4 and D5, and D5 and
D6. In our
analysis
we
assume
for
simplicity that facial symmetry
deviations
are
linear rather than nonlinear.
Calculation
of
averageness. Averageness
of a
face
was
deter-
mined
in the
following
way. First,
the
mean lengths
of
each
of the
lines
Dl
through
D6
were calculated
for the 16
normal photos
of
each
sex.
Averageness
of the
vertical dimension
for
each
sex was
Figure
2. A
two-face composite (center), with
the
original
photographs
on
either side.

236
KARL
GRAMMER
AND
RANDY
THORNHILL
B
Figure
3.
Sixteen-face composites
of
each sex.
calculated
by the
distance
between
P13 and the
midpoint
of
Dl
and
the
distance between
PI3
and the
midpoint
of D5.
Then
the
absolute differences between these means
for
each
sex and the
length
of
each
of the
same lines
in
each individual face
of the
same
sex
were summed.
Calculation
of
metric facial dimensions.
The
sizes
of
nine
facial features
were
calculated:
distance
between
outer
eye
corners
(Dl), distance between inner
eye
corners (D2), mean
eye
size
([P3
-
PI
+ P2 - P4]
-T-
2),
nose width (D4), facial width
at
cheekbones (D3), mouth width (D5),
and jaw
width (D6).
In
addition,
we
calculated average cheekbone prominence,
(P6 — P5
+ P6 —
P10)
H-
2.
Cheekbones
are
more prominent when values
are
positive
than when they
are
negative.
Finally,
lower face
proportions describe
the
relation between face length
and jaw
width,
Midpoint
Dl — P13
-r-
D6. A
high value indicates
a
long
face
with
a
narrow jaw;
a low
value indicates
a
face with
a
wide
jaw.
Statistical Tests
All
statistical significance levels reported
in
this study
are
two-tailed.
All
ratings
and
means
of
ratings
for the
faces follow
normal distributions.
The .05
level
is the
critical level
for
statistical
significance.
Results
Composites Versus Normals: Attractiveness
The
mean
ratings
of
composites
versus
normal
faces
of
each
sex by
opposite-sex
raters
were
compared
with
t
tests
(Table
1).
Female
composites
were
judged
significantly
more
attractive
and
sexier
than
normal
photos.
Also,
female
composites
were
judged
significantly
less
dominant
than
normal
female
faces.
There
was not a
significant
difference
in
the
health
ratings
of
female
composite
versus
normal
faces,
but
female
composites
were
rated
healthier
(p =
.10).
Normal
male
faces
were
rated
as
significantly
healthier,
sexier,
and
more
dominant
than
composite
male
faces;
the
same
pattern
is
seen
for the
adjective
attractive,
and
this
just
missed
statistical
significance
(p =
.06).
Thus,
making
Figure
4. The
landmark points used
for
facial measurements.

FACIAL ATTRACTIVENESS: SYMMETRY
AND
AVERAGENESS
237
Figure
5. The
lines used
for
facial measurements.
composites
from
individual female faces yielded pictures
that were
more
attractive than
the
originals,
but the
reverse
was the
case
for
male photos.
It is
possible that normal faces
of
males were rated more positively because they more
frequently
contained more extreme values
in
facial features.
Indeed,
male
composites
were
rated
as
less
dominant.
The
ages
of
normal subjects
and the
mean attractiveness
ratings
of
their faces showed statistically insignificant cor-
relations
in
each
sex (16
females,
r =
.13,
p > .5, and 16
males,
r =
.22,
p >
.05).
Composites
Versus Normals:
Averageness
and
Symmetry
As
pointed
out by
Alley
and
Cunningham
(1991),
in
comparison
with
normal photos, composites
are
more aver-
age in
facial features
but
also show higher values
in
other
traits,
such
as
symmetry
and
smoothness
of
complexion
(also
Benson
&
Perrett,
1991).
Thus,
we
measured
metric
averageness
per se in
composite
and
normal photos.
Be-
cause
of the
small number
of
composites (7),
a
nonpara-
metric
comparison
of
median averageness
for
composite
and
normal pictures
of
each
sex was
conducted.
For
each
sex
the
median deviation
from
averageness
for
composite
faces
was
less than
for
normal faces
but was not
statistically
significantly
(16
normal faces
vs. 7
composite faces
for
each
sex,
Mann-Whitney
[/test,ps
>
.10).
However,
the
median
for
the
differences
from
facial averageness
for the
individ-
ual
women's faces
was 52, but it was
only
36 for the
16-face
composite. This same pattern
was
seen
for
men. Individual
male faces showed
a
median
of 77, and the
16-face com-
posite
had a
median
of 54.
Thus,
our
results indicate that
computer averaging does
not
necessarily create statistically
higher
metrical
averageness
in
faces
and
that
the
averaging
effect
in
composites
may
depend
on how
many faces
are
added
to
form
a
composite.
Symmetry
in
composites
and
normal faces
was
also com-
pared. Composites showed less asymmetry than
the
normal
photos
(Mann-Whitney
U
tests):
FA for
normal male faces,
Mdn
= 67, and for
composite male faces,
Mdn
=
41,
p =
.08;
FA for
normal female faces,
Mdn = 54, and for
composite female faces,
Mdn = 25, p =
.03;
CFA for
normal male faces,
Mdn =
18.5,
and for
composite
male
faces,
Mdn = 12, p =
.05;
CFA for
normal female faces,
Mdn
=
13.5,
and for
composite
female
faces,
Mdn =
8.5,
p
=
.05. Apparently, symmetry
is
reached with fewer faces
than
is
metrical
averageness
when faces
are
combined
in
composites.
Overall then, male composite faces were less attractive
than
normal male faces, whereas composite female faces
were more attractive than normal female faces.
In
both
sexes
the
composites were more symmetrical,
but
metrical
averageness seemed
to be
present only
for the
16-face
composites
in
both sexes. However,
it
must
be
emphasized
that
the
symmetry
and
averageness comparisons
of
normal
versus
composite
faces
are
based
on
only
7
composite
faces.
Asymmetry
and
Facial Attractiveness
We
used parametric partial correlation analysis
to
assess
the
effects
of
facial averageness
and
facial asymmetry
in-
dependently
on the
ratings
of
normal faces
for
each
of the
four
adjectives. Mean ratings
for
each adjective
on
each face
were used
in the
analysis. When facial averageness
was
partialed
out of
male
ratings
of
normal female faces
(Table
2), CFA and FA
correlated
significantly
negatively with
mean ratings
of the
faces
for the
adjectives attractive
and
sexy.
The
mean ratings
of the
female faces
for the
adjectives
dominant
and
healthy
also
showed negative
partial
correla-
tions
with
asymmetry (both
CFA and
FA),
but
these corre-
lations
did not
reach statistical significance. Controlling
averageness
in the
ratings
of
normal male photos
by
females
resulted
in
significant negative partial correlations between
CFA and
mean ratings
of
attractive, sexy
and
healthy.
FA in
the
normal male faces correlated significantly negatively
with
ratings
of
attractive
and
healthy,
but not
with sexy
(Table
2). The
zero-order correlations
are
also shown
in
Table
2. The
similarity
of the
zero-order
correlations
and the
partial correlations
for
female faces rated
by men
reveals
that
partialing
out
facial averageness
has
little
effect
on the
magnitude
of the
relation between facial symmetry
and the
ratings
of the
face. However, partialing
out
averageness
does seem
to
improve
the
relation
for
male faces rated
by
women.

238
KARL
GRAMMER
AND
RANDY
THORNHILL
Table
1
Ratings
by Men and
Women
for
Normal
and
Composite Faces
Ratings
of
women's
faces
by men
Ratings
of
men's
faces
by
women
Normal
Characteristic
M
SD
Composite
M
SD
Normal
Composite
'(43)
M
SD
M
SD
«53)
Attractive
Dominant
Sexy
Healthy
2.79
3.29
2.57
4.16
0.80
0.93
0.78
0.91
3.26
3.12
3.18
4.34
1.00
0.85
0.94
1.14
3.59
2.04
4.69
1.66
.001
.048
.0001
.103
3.12
3.48
2.81
4.68
0.85
0.88
0.86
0.84
2.88
3.02
2.51
4.42
1.07
0.99
0.98
0.99
1.92
5.23
2.85
3.17
.060
.0001
.006
.003
Note.
Fifty-two
women
and 44 men
rated
16
normal
and 7
composite
faces
of
each
sex.
Probabilities
are
two-tailed.
Averageness
and
Facial Attractiveness
When either
CFA
or FA was
partialed
out of the
men's
ratings
of
female faces,
measured
female facial
averageness
did not
correlate significantly with
the
means
of any of the
ratings.
A
different result
for
averageness
was
seen
for
women's ratings
of
male faces when symmetry
was
con-
trolled.
In
this
case,
statistically significant
(ps
^
.05)
negative partial correlations were
found
between facial
av-
erageness
and
attractive (CFAS,
r =
-.45;
FAS,
r -
-.49),
dominant
(CFAS,
r
=
-.49; FAS,
r =
-.50)
and
sexy
(CFAS,
r =
-.53; FAS,
r =
-.53).
In
this analysis,
averageness scores were multiplied
by
—
1.
Thus,
the
neg-
ative
correlations mean that more average faces were less
attractive. Both
photometrical
averageness (seen
in
compos-
ites)
and
metrical averageness seem
to
have
a
negative
effect
on the
ratings
of
male faces
by
women.
Facial
Features
and
Attractiveness
In
order
to find out
which facial characteristics play
a
role
in
attractiveness ratings,
we
correlated
the
nine facial mea-
sures with
the
ratings
of all
pictures, that
is,
ratings
of
both
normal
and
composite
faces. Also,
we
partialed
out FA and
examined
the
relations between mean ratings
of
faces
and
the
nine facial features, because
FA may
have
had an
effect
on
the
judgments
of the
features. Zero-order correlations
and
partial correlations show similar patterns.
For
men's
rating
of
female faces (Table
3),
certain extremes
in the eye
region were rated
as
less attractive
or
sexy. Greater distance
between
outer
eye
corners
was
rated
as
significantly nega-
tively
attractive
or
sexy.
The
size
of
eyes showed
the
same
pattern
(p =
.06).
The
distance between both outer
and
inner
eye
corners
were
negatively
correlated
with
ratings
of
dominant.
Cheekbone prominence showed
a
significant
positive correlation
with
sexy.
A
different
pattern
was
seen
for
women's
rating
of
male
faces.
Here extremes
in the
lower face were rated most
favorably
(Table
4). A
large
jaw and a
wide mouth were
rated
as
attractive.
A
large nose
and a
wide mouth were
rated
as
healthy. Also,
a
broader face,
in
comparison with
its
length
(lower face proportions),
was
attractive, sexy,
healthy,
and
dominant. Cheekbone prominence
was
posi-
tively,
but not
significantly
(p =
.11),
correlated with
the
male facial attractiveness ratings.
As
predicted,
in
male
faces,
certain extreme traits seem
to
play
a
role
in
attrac-
tiveness,
especially
the
testosterone-facilitated traits that
signal
social
dominance.
Discussion
Despite
our
small samples
of 16
normal faces
and 7
composites
of
each sex,
we
obtained some statistically sig-
nificant
relations, and,
in
general,
the
results were consistent
with
the
hypotheses derived
from
the
parasite theory
of
sexual
selection. Before
we
discuss
the
results,
we
deal with
our
methods, some
of
which
are
different
from
the ap-
proaches that have been used traditionally
in the
study
of
facial
features
and
attractiveness. Some
of our
methods
may
represent improvements.
Table
2
Zero-Order
and
Partial Correlations,
With
Metric Facial Averageness Controlled,
Between Mean Ratings
of
Normal Faces
and
Central Facial Asymmetry (CFA)
and
Overall
Facial Asymmetry (FA)
Ratings
of
women's
faces
by men
CFA
Characteristic
Attractive
Dominant
Sexy
Healthy
r0(14)
-.55*
-.26
-.51*
-.21
rp(13)
-.54*
-.26
-.50*
-.21
FA
r0(14)
-.54*
-.41
-.49*
-.18
rp(13)
-.53*
-.41
-.48*
-.19
Ratings
of
men's
CFA
r0(14)
-.56*
.03
-.38
-.45
rp(13)
-.64*
.07
-.51*
-.53*
faces
by
women
FA
r0(14)
-.48
.06
-.27
-.41
rp(13)
-.60*
-.15
-.45
-.54*
Note.
Fifty-two
women
and 44 men
rated
16
normal
faces
of
each
sex.
Probabilities
are
two-tailed.
*p
<
.05.

FACIAL ATTRACTIVENESS: SYMMETRY
AND
AVERAGENESS
239
Table
3
Zero-Order
and
Partial Correlations Between Mean Ratings
by Men of
Women's Normal
and
Composite Faces
and
Facial
Features
Attractive
Characteristic
Distance between outer
eye
corners
Distance between inner
eye
corners
Mean
eye
size
Nose width
Cheek width
Jaw
width
Mouth
width
Cheekbone prominence
Lower face proportions
r0(21)
-.32
.01
-.30
-.02
-.23
-.07
-.17
.45*
-.01
rp(20)
-.44*
.01
-.42
.04
-.25
-.07
-.01
.25
-.01
Dominant
r0(21)
-.45*
-.53*
-.07
.10
-.39
.00
-.07
.38
-.16
rp(20)
-.49*
-.55*
-.09
.09
-.39
.00
.00
.38
-.18
Sexy
r0(21)
-.38
.04
-.37
-.14
-.33
-.12
-.25
.27
.06
rp(20)
-.51*
.04
-.49*
-.19
-.37
-.14
-.08
.48*
.06
Healthy
r0(21)
-.10
-.11
-.02
.32
.15
.00
-.09
.14
.02
rp(20)
-.12
-.11
-.03
.31
-.14
.00
-.02
.13
.01
Note.
Forty-four
men
rated
23
normal
and
composite women's faces. Partial correlations
are
controlled
for
asymmetry. Probabilities
are
two-tailed.
*
p
<
.05.
It
is
customary
to
standardize facial size
across
faces
by
the
full
length
of the
face (chin
to
hairline;
see
Cunningham,
1986). This
is
problematic because
of the
subjectivity
of
defining
the
hair line
in
many facial photos.
Our
method
involved making
all
faces equal
in the
vertical distance
between
the
midline
of
eyes
and the
midline
of
mouth,
and
equal
in the
distance between pupils. Thus,
the
reference
lines
for
standardization
of
face
size
can be
determined
objectively.
It is
possible,
however, that
our
method intro-
duces some unknown artifacts into
the
results.
For
ratings
of
faces
in our
study,
the
rater
was first
presented with
all
faces
in the set one at a
time
in a
randomized order.
In a
second step,
the
rater rated
the
randomly ordered pictures. This eliminated order
effects
but
may
have influenced
the
mean ratings.
Our
mean ratings,
however,
were similar
to
those reported
by
Langlois
and
Roggman
(1990).
We
calculated facial asymmetry
in two
ways
in an at-
tempt
to
begin looking
at
whether people assess symme-
try
in the
entire face
or
only
by
comparing adjacent
parts.
The
results
for the two
measures
of
asymmetry
were similar.
Female
composite faces
are
more attractive than individ-
ual
female faces,
but
individual male faces
are
rated more
favorably
than male composite faces.
Our
composite faces
possessed more bilateral symmetry than individual faces,
and
our
study
reveals
that both
sexes
view symmetry
in
opposite-sex individual faces
as
attractive. Why, then,
are
individual
male faces more attractive
to
women than com-
posite male faces?
The
answer apparently
is
that symmetry
is
not the
only trait that
affects
facial
attractiveness. Com-
posite male faces were
not
statistically more average than
individual
male faces.
We
found
that overall averageness
of
male facial features among individual male faces correlates
negatively
with attractiveness. Also,
we
found
that certain
male facial features
are
more
attractive
when
large,
espe-
cially
those that probably
are
associated with displaying
dominance
and its
correlates, such
as
parasite resistance,
in
Table
4
Zero-Order
and
Partial Correlations Between Mean Ratings
by
Women
of
Men's Normal
and
Composite Faces
and
Facial
Features
Attractive
Characteristic
Distance between outer
eye
corners
Distance between inner
eye
corners
Mean
eye
size
Nose width
Cheek width
Jaw
width
Mouth
width
Cheekbone prominence
Lower face proportions
r0(21)
.17
< 01
.23
.37
.21
.45*
.45*
.34
-.46*
rp(20)
.11
-.07
.21
.34
.23
.47*
.47*
.34
-.54*
Dominant
r0(2D
-.02
-.15
.08
.27
.07
.28
.12
.29
-.42*
rp(20)
.02
-.10
.10
.31
.06
.28
,12
.29
-.42*
Sexy
r0(2l)
.23
.12
.23
.26
.19
.32
.31
.19
-.41*
rp(20)
.20
.08
.21
.24
.19
.32
.32
.19
-.44*
Healthy
r0(2
1)
.16
-.01
.22
.45
.17
.31
.43
.20
-.43*
rp(20)
.10
-.06
.19
.44*
.18
.32
.45*
.20
-.50*
Note.
Fifty-two
women rated
23
normal
and
composite men's faces. Partial correlations
are
controlled
for
asymmetry. Probabilities
are
two-tailed.
*
p
<
.05.

240
KARL
GRAMMER
AND
RANDY
THORNHILL
human
evolutionary history.
A
broad male jaw, which
is a
testosterone-facilitated feature,
was
attractive
in our
sample.
Other studies have also
found
that wide jaws
in men are
viewed
as
attractive
to
women (see Alley
&
Cunningham,
1991; Cunningham,
Barbee,
&
Pike, 1990)
and are
viewed
as
a
trait implying
social
dominance (Keating, 1985). Keat-
ing, Mazur,
and
Segall
(1981)
showed that
men
with
big
jaws
are
judged
across
many diverse cultures
as
socially
dominant. Mazur, Mazur,
and
Keating
(1984)
showed that
the
rank ultimately attained
by
West Point cadets
is
predict-
able
from
the
dominance rating
of
their facial photo.
Prominent
cheekbones
in
women were viewed
as
sexy.
The
attractiveness
of
prominent cheekbones
in
women
was
shown
by
Cunningham
(1986).
It is
unclear whether prom-
inent
cheekbones
in
women display immune system com-
petence.
The
sexual dimorphism
in
cheekbone growth dur-
ing
puberty,
for
which
men
show much greater growth
(Enlow,
1990), suggests that female cheekbones
may
signal
information
other than testosterone tolerance. Apparently,
estrogens
also
handicap
the
immune system (see Thornhill
&
Gangestad,
1993;
Wedekind,
1992). Highly estrogenized
female
facial features
are
attractive (Johnston
&
Franklin,
in
press). Perhaps cheekbone prominence
in
women reflects
the
effects
of
estrogen
and
thereby
honestly
advertises
immunocompetence.
Our
findings
on
female facial features show some incon-
sistency
with
certain previous studies. This
is
most clearly
the
case
for
female
eye
size (see Alley
&
Cunningham,
1991; Jones
&
Hill,
1993).
Our
results suggest that small,
not
large, eyes
are
more attractive
and
sexy
in
women.
It is
unclear
at
this time
why we
obtained this result,
but it
could
reflect
methodological
variation between
our
study
and
other studies
or a
finding
peculiar
to our
sample.
This
is
apparently
the
first
study
to
indicate that measured
facial
symmetry
affects
positive judgments about facial
attractiveness.
FA
showed
a
significant, negative relation-
ship
with
facial attractiveness ratings
in
each
of the
sexes
when
the
effect
of
facial averageness
was
removed statisti-
cally.
The
effect
of
facial symmetry
on
facial
attractiveness
may
be
fruitful
to
study
in
larger samples
and
other popu-
lations.
There
is
increasing
evidence
that body symmetry
plays
a
role
in
sexual selection
in
general
in
animal species
(see Watson
&
Thornhill,
1994).
In
humans,
nonfacial
body
symmetry
was
shown
to
correlate with facial attractiveness
ratings
(Gangestad, Thornhill,
&
Yeo,
in
press).
Our
study,
coupled with Gangestad
et
al.'s
study, suggests that facial
attractiveness, facial symmetry,
and
body symmetry
are
intercorrelated.
Thus, facial beauty
may be a
certification
of
high-quality
development
of the
body
in
general.
As we
discuss
in the
introduction,
a
person's
bilateral
symmetry
may
reflect both high developmental quality
and
individual
heterozygosity, both
of
which
may
have been important
indicators
of
general health
as
well
as
parasite
resistance
in
the
environments
of
human evolutionary history. Preference
of
a
symmetrical mate
may
have evolved because
of
posi-
tive
genetic
effects
on
offspring survival
or
because sym-
metrical individuals
may be
capable
of
greater parental
investment
on
average.
Our
study indicates that
men may
judge averageness
of
facial
features
in
women
as
more attractive,
but
female
facial
symmetry appears
to
have
a
stronger influence than
averageness
on
men's judgments. Composite female faces
are
more attractive
and
more symmetrical
than
individual
female
faces. Composite female faces seem
to be
more
average than individual women's faces only when many
faces
are
used. Also, when facial averageness
was
con-
trolled
in the
partial correlation analysis, facial symmetry
remained significant
as a
predictor
of
women's facial
at-
tractiveness rating. Removing facial symmetry, however,
eliminated
the
significant relationship between facial aver-
ageness
and
ratings
of
attractiveness.
The
previous studies
with
composites that claimed that
facial
attractiveness
is
causally related
to
averageness
of
facial features
did not
control
for
facial
symmetry (e.g., Langlois
&
Roggman,
1990).
The
limited positive
effect
of
facial averageness
on the
attractiveness
of
faces
in our
study must
not be
viewed
as
definitive.
Our
small sample
of
normal faces
may
have
resulted
in
sampling error
in the
facial
features
of our
subjects. Also,
a
large sample
of
faces
may
result
in a
different
and
more accurate population average than
the
population average
in our
study.
There
is
other evidence that both symmetry
and
average-
ness
of
certain facial features
are
important
in the
attrac-
tiveness
of
women's faces. Women
use
makeup
to
make
certain features more average
(e.g.,
interocular
distance
and
nose size). Symmetry
of
features also
is a
major goal
in
making
up the
face. Consistent with
our
findings, women
use
makeup
to
make cheekbones appear larger,
not
average.
These issues
are
discussed
in
each
of the
numerous makeup
books
that
we
examined
(e.g.,
Gold,
1978;
McCrerery,
1986). Also, reconstructive
and
plastic
facial
surgery
is
often
used
to
correct asymmetries
in the
faces
of
both sexes
(e.g.,
Whitaker
&
Pertschuk,
1982; Williamson
&
Varela,
1990).
In
future
research,
it
will
be
important
to
evaluate
the
relative importance
of
averageness
and
symmetry
in
judg-
ments
of
women's
facial
attractiveness. Manipulations
of
facial
features
in
facial
images
may be a
useful
method.
Clearly, nonaverageness
in
certain adult female
facial
fea-
tures
is
maximally
attractive.
This
is the
case
for
chin
size
(smaller than average
is
more attractive)
and lip
size (larger
than
average
is
attractive;
see
Johnston
&
Franklin,
in
press).
Johnston
and
Franklin pointed
out
that
a
small chin
and
large lips
in
women display high estrogen
and
thus
a
hormone profile
of
high female fertility.
As we
mention
in
the
discussion
of
women's cheekbones,
it is
possible that
highly
estrogenized
female facial features display immuno-
competence.
Our
approach
attempts
to
understand
why
people
make
judgments
of the
sexual attractiveness
of
human faces
and
what
facial features
are
involved
in
such judgments
and
why.
Said differently,
we
wish
to
know what kind
of
Darwinian selection pressure
led to the
psychological adap-
tation that
processes
information about facial aesthetic value
and
generates
the
different
feelings
and
motivations associ-
ated
with
viewing faces
of
different
aesthetic value.
We
also
wish
to
know what form
of
Darwinian sexual selection
led

FACIAL ATTRACTIVENESS: SYMMETRY
AND
AVERAGENESS
241
to the
evolution
of the
human
facial
features
that
affect
facial
attractiveness.
Aestheticians
commonly
view
beauty
judgments
as
arbitrary
(see
review
in
Kovach,
1974).
Al-
though
there
is a
hypothetical
evolutionary
mechanism
that
may
have
led to the
evolution
of
mate
choice
on the
basis
of
nonfunctional
or
arbitrary
sexual
attractiveness
(Fisher,
1958),
our
results
support
the
view
that
aesthetic
judgments
of
faces
are not
capricious
but
instead
reflect
evolutionary
functional
assessments
and
valuations
of
potential
mates
(Thornhill
&
Gangestad,
1993).
More
specifically,
these
results
are
consistent
at
least
with
the
hypothesis
that
judg-
ments
of
facial
aesthetics
arise
as
outputs
of
psychological
adaptation
that
is
designed
to
assess
a
person's
potential
for
health-related
survival
and
reproduction
in the
environ-
ments
of
human
evolutionary
history
and
that
this
psycho-
logical
adaptation
generates
the
aesthetic
experience
of
highly
attractive
or
beautiful
when
it
encounters
facial
fea-
tures
that
are
certifications
of
developmental
homeostasis
and
of
immunocompetence.
Finally,
these
results
suggest
that
sexual
selection
that
favored
developmentally
healthy
and
immunocompetent
persons
designed
certain
features
of
the
face
that
affect facial
sexual
attractiveness.
Of
course,
more
research
is
needed
to
assess
the
possible
connection
between
infectious
disease,
immunocompetence,
and hu-
man
judgments
of the
beauty
of
human
faces
and
other
features
of
human
bodily
form.
Finally,
we
must
mention
that
there
is
comparative
evi-
dence
that
human
beauty
judgments
are
tied
importantly
to
assessing
parasite
effects.
Gangestad
and
Buss
(in
press)
showed
that
the
importance
that
people
across
societies
place
on
good
looks
in
choice
of a
long-term
mate
correlates
positively
with
the
prevalence
of
parasites
in the
environ-
ment.
The
correlation
is
seen
for
both
men and
women.
References
Alley,
T. R., &
Cunningham,
M. R.
(1991).
Averaged faces
are
attractive,
but
very attractive faces
are not
average. Psycholog-
ical
Science,
2,
123-125.
Benson,
P., &
Perrett,
D.
(1991). Computer averaging
and
manip-
ulations
of
faces.
In P.
Wombell
(Ed.),
Photovideo:
Photogra-
phy in the age of the
computer (pp.
32-38).
London: Rivers
Oram
Press.
Cunningham,
M. R.
(1986).
Measuring
the
physical
in
physical
attractiveness:
Quasi-experiments
on the
sociobiology
of
female
facial
beauty. Journal
of
Personality
and
Social
Psychology,
50,
925-935.
Cunningham,
M.
R.,
Barbee,
A. P., &
Pike,
C. L.
(1990). What
do
women
want? Facial-metric assessment
of
multiple motives
in
the
perception
of
male facial physical attractiveness. Journal
of
Personality
and
Social
Psychology,
59,
61-72.
Darwin,
C.
(1871).
The
descent
of
man and
selection
in
relation
to
sex.
London: Murray.
Enlow,
D. H.
(1990). Facial growth (3rd ed.). Philadelphia:
Harcourt
Brace
Jovanovich.
Fisher,
R. A.
(1958).
The
genetical
theory
of
natural
selection (2nd
ed.).
New
York: Dover.
Folstad,
I.,
&
Karter,
A. J.
(1992). Parasites, bright males
and the
immunocompetence
handicap. American Naturalist, 139, 603-
622.
Gallon,
F.
(1879).
Composite portraits, made
by
combining those
of
many
different
persons
in a
single resultant
figure.
Journal
of
the
Anthropological Institute,
8,
132-144.
Gangestad,
S. W.,
Thornhill,
R.,
&
Yeo,
R. A. (in
press).
Facial
attractiveness, developmental stability
and
fluctuating asymme-
try. Ethology
and
Sociobiology.
Gangestad,
S.
W.,
&
Buss,
D. M. (in
press). Pathogen prevalence
and
human mate preferences. Ethology
and
Sociobiology.
Gold,
S.
(1978).
How to do
your face.
New
York: Dell.
Hamilton,
W. D.
(1982). Pathogens
as
causes
of
genetic diversity
in
their host populations.
In R. M.
Anderson
& R. M. May
(Eds.),
Population biology
of
infectious
diseases (pp.
269-296).
New
York:
Springer-Verlag.
Hamilton,
W.
D.,
&
Zuk,
M.
(1982).
Heritable true fitness
and
bright birds:
A
role
for
parasites? Science, 218,
384-387.
Hausfater,
G., &
Thornhill,
R.
(Eds.).
(1990).
Parasites
and
sexual
selection [Special
issue].
American Zoologist,
30(2).
Jones,
D.,
&
Hill,
K.
(1993).
Criteria
of
facial attractiveness
in five
populations. Human Nature,
4,
271-296.
Johnston,
V.
S.,
&
Franklin,
M. (in
press).
Is
beauty
in the eye of
the
beholder? Ethology
and
Sociobiology.
Keating,
C. F.
(1985).
Gender
and the
physiognomy
of
dominance
and
attractiveness. Social
Psychology
Quarterly,
48,
61-70.
Keating,
C.
F.,
Mazur,
A.,
&
Segall,
M. H.
(1981).
A
cross-cultural
exploration
of
physiognomic traits
of
dominance
and
happiness.
Ethology
and
Sociobiology,
2,
41-48.
Kovach,
F. J.
(1974).
Philosophy
of
beauty.
Norman: University
of
Oklahoma Press.
Langlois,
J.
H.,
&
Roggman,
L. A.
(1990).
Attractive faces
are
only
average. Psychological Science,
1,
115-121.
Leary,
R.
F.,
&
Allendorf,
F. W.
(1989).
Fluctuating asymmetry
as
an
indicator
of
stress: Implications
for
conservation biology.
Trends
in
Ecology
and
Evolution,
4,
214-217.
Manning,
J.
T.,
&
Chamberlain,
A. T.
(1993). Fluctuating asym-
metry,
sexual selection
and
canine teeth
in
primates. Proceed-
ings
of the
Royal
Society
of
London,
B,
251,
83-87.
Mazur,
A.,
Mazur,
J.,
&
Keating,
C. F.
(1984). Military rank
attainment
of a
West Point class:
Effects
of
cadet's physical
features.
American
Journal
of
Sociology,
90,
125-150.
McCrerery,
L.
(1986). Redhead: Fabulous
makeovers
for
ultimate
beauty.
New
York: Collier Books.
M011er,
A. P.
(1992). Parasites
differentially
increases
the
degree
of
fluctuating
asymmetry
in
secondary sexual characters. Jour-
nal
of
Evolutionary
Biology,
5,
691-699.
M011er,
A. P., &
Pomiankowski,
A.
(1993). Fluctuating asymmetry
and
sexual selection.
Genetica,
89,
267-279.
Parson,
P. A.
(1990).
Fluctuating asymmetry:
An
epigenetic
mea-
sure
of
stress. Biological Review,
65,
131-145.
Parson,
P. A.
(1992).
Fluctuating asymmetry:
A
biological monitor
of
environmental
and
genomic
stress.
Heredity,
68,
361-364.
Soul6,
M.,
&
Cuzin-Roudy,
J.
(1982).
Allometric
variation:
2.
Developmental
instability
of
extreme phenotypes. American
Naturalist,
120,
765-786.
Symons,
D.
(1979).
The
evolution
of
human
sexuality.
Oxford:
Oxford
University Press.
Tanner,
J. M.
(1978).
Foetus into man: Physical growth
from
conception
to
maturity. London: Open Books.
Thornhill,
R.,
&
Gangestad,
S.
(1993). Human facial beauty:
Averageness, symmetry
and
parasite resistance. Human Nature,
4,
237-269.
Tooby,
J.
(1982).
Pathogens, polymorphism
and the
evolution
of
sex. Journal
of
Theoretical
Biology,
97,
557-576.
Treu,
G.
(1914).
Durchschnittsbild
und
Schonheit
[Composite
pictures
and
beauty].
Zeitschrift
fur
Aesthetik
und
Allgemeine
Kunstwissenschaft,
9,
433-448.

242
KARL
GRAMMER
AND
RANDY
THORNHILL
Watson,
P. W., &
Thornhill,
R.
(1994).
Fluctuating
asymmetry
and
sexual
selection.
Trends
in
Ecology
and
Evolution,
9,
21-25.
Wedekind,
C.
(1992).
Detailed
information
about
parasites
revealed
by
sexual
ornamentation.
Proceedings
of the
Royal
Society
of
London,
B,
247,
169-174.
Whitaker,
L.
A.,
&
Pertschuk,
M.
(1982).
Facial
skeletal
contour-
ing for
aesthetic
purposes.
Plastic
and
Reconstructive
Surgery,
69,
145-253.
Williamson,
E. H., &
Varela,
J. G.
(1990).
Correction
of
mandib-
ular
asymmetry
with
the
ligated
anterior
repositioning
splint.
Journal
of
Craniomandibular
Practice,
8,
30-34.
Zuk,
M.
(1992).
The
role
of
parasites
in
sexual
selection:
Current
evidence
and
future
directions.
Advances
in the
Study
of
Behav-
ior,
21,
39-67.
Received June
30,
1993
Revision received December
5,
1993
Accepted December
6,
1993
•
Articles
to Be
Published
in the
Journal
of
Comparative
Psychology
December
1994, Vol. 108,
No. 4
Constraints
on
Great
Apes'
Imitation: Model
and
Action Selectivity
in
Rehabilitant
Orangutan Imitation (Pongo pygmaeus) Anne
E.
Russon
and
Birute
M.
F.
Galdikas
Frequency Ratio Discrimination
by
Zebra
Finches
(Taeniopygia
guttata)
and
Humans
(Homo
sapiens) Ronald
Weisman,
Milan Njegovan,
and
Stephen
Ito
Marmoset
(Callithrix
jacchus
jacchus)
Spatial
Memory
in a
Foraging
Task:
Win-
Stay
vs.
Win-Shift Strategies Suzanne
E.
MacDonald, John
C.
Pang,
and
Stephanie
Gibeault
Automated Recording
of
Individual Performance
and
Hand Preference During
Joystick-Task
Acquisition
in
Group-Living
Macaca
radiata
Michael
W.
Andrews
and
Leonard
A.
Rosenblum
Stimulus Control
of
Defensive Behaviors
of
Garter Snakes
(Thamnophis
sirtalis):
Effects
of Eye
Spots
and
Movement Chris Bern
and
Harold
A.
Herzog,
Jr.
Evidence
for
Spatial Working Memory
in
Honey Bees
(Apis
melliferas)
Michael
F.
Brown
and
Gregory
E.
Demos
The
Production
and
Comprehension
of
Referential Pointing
by
Orangutans (Pongo
Pygmaeus)
Josep Call
and
Michael
Tomasello
Density Dependent Foraging Strategy
of a
Large Carnivorous Lizard
(Varanus
albigularis)
Jennifer
D.
Kaufman,
Gordon
M.
Burghardt,
and
John
A.
Phillips
Sequential Structure
in
Behavioral Components
of
Object Investigation
by
Long-Evans Rats Michael
J.
Renner
and
Charles
P.
Seltzer
Object
Permanence
in
Rhesus Monkeys
(Macaca
mulatto)
and
Melinda
A.
Novak
Sandra
T. de
Blois
Twenty-Five Years
of
Behavioral Research
on
Great
Apes:
Trends
Between
1967
and
1991
Patricia
L.
Hebert
and
Martin
Courtois
Performance
of
Four Seed-Caching
Corvid
Species
in the
Radial-Arm Maze
Analog Alan
C.
Kamil,
Russell
P.
Balda,
and
Deborah
J.
Olson
Comparison
of
Predatory Behaviors
of
Deer Mice
(Peromyscus
maniculatus)
and
Grasshopper Mice
(Onychomys
leucogaster) William
M.
Langley