Magazine

R623

examined in this task: two were

the same animals providing data
body. We believe that the activity
from the motor and/or sensory
Biological
shown in Figure 1B (cats 1 and
2), and the third (cat 4) was only
systems provides a positive
signal that an obstacle is located
components of
tested in this task. The results between the fore and hind legs, sex differences in
for the three animals are shown
in Figure 2B. With cat 1, the
and thus extends the duration
of the neural representation of color preference
maximum step height remained this obstacle until the hind legs
higher than control values and step over. This simple behavior Anya C. Hurlbert and Yazhu Ling
usually higher than the rod height provides a convincing example of
for all time durations we tested. movement signals being used to The long history of color
With cat 2, the maximum step update the neural representation preference studies has been
height was quite variable during of external obstacles and offers described as “bewildering,
longer pause durations (over a good system to explore the confused and contradictory”
5 seconds) but remained higher neural structures involved. [1]. Although recent studies
than control step height in the [1–3] tend to agree on a universal
majority of trials. With cat 4 there Acknowledgments preference for ‘blue’, the variety
was also some variability, but on Supported from grants from the and lack of control in measurement
most trials the maximum hind leg ­Canadian Institutes of Health Research methods have made it difficult to
step height was also higher than and the Alberta Heritage Foundation extract a systematic, quantitative
the height of the rod and well for Medical Research. description of preference.
above control values. Furthermore, despite abundant
By comparing the step References evidence for sex differences
1. Patla, A.E. (1997). Understanding the
heights when cats pause roles of vision in the control of human
in other visual domains, and
either immediately in front of locomotion. Gait Posture 5, 54–69. specifically in other tasks of
an obstacle or straddling an 2. Patla, A.E., and Vickers, J.N. (1997). color perception [4,5], there is
Where and when do we look as we
obstacle, we have shown that approach and step over an obstacle no conclusive evidence for the
stepping over an obstacle with in the travel path? Neuroreport 17, existence of sex differences in
3661–3665.
the forelegs is necessary to 3. Patla, A.E. (1998). How is human gait
color preference. This fact is
create long-lasting memories to controlled by vision? Ecol. Psychol. 10, perhaps surprising, given the
guide the hind legs (Figure 1). 287–302. prevalence and longevity of the
4. Fowler, G.A., and Sherk, H. (2003). Gaze
We have also shown that visual during visually-guided locomotion in notion that little girls differ from
signals related to the obstacles cats. Behav. Brain Res. 139, 83–96. boys in preferring ‘pink’ [6]. Here
5. Wilkinson, E.J., and Sherk, H.A. (2005).
are not essential for this process The use of visual information for
we report a robust, cross-cultural
(Figure 2). This is strong evidence planning accurate steps in a cluttered sex difference in color preference,
that a neural signal related to environment. Behav. Brain Res. 2, revealed by a rapid paired-
270–274.
the stepping of the forelegs is 6. McVea, D.A., and Pearson, K.G. (2006). comparison task. Individual
responsible for the activation Long-lasting memories of obstacles color preference patterns are
guide leg movements in the walking cat.
or enhancement of additional J. Neurosci. 4, 1175–1178.
summarized by weights on the two
structures of the nervous system 7. Fiset, S., and Dore, F.Y. (2005). Duration fundamental neural dimensions
to produce a long-lasting of cats’ (Felis catus) working memory that underlie color coding in the
for disappearing objects. Anim. Cogn.
memory. There are two possible 1–9. human visual system. We find a
sources of such a signal: the 8. Stein, R.B., Weber, D.J., Aoyagi, Y., consistent sex difference in these
Prochazka, A., Wagenaar, J.B., Shoham,
feedback from cutaneous, S., and Normann, R.A. (2004). Coding
weights, which, we suggest,
muscle, and joint afferents which of position by simultaneously recorded may be linked to the evolution of
can signal the position of the sensory neurones in the cat dorsal root sex-specific behavioral uses of
ganglion. J. Physiol. Pt 3, 883–896.
limb [8]; and the output of motor 9. Drew, T., Jiang, W., Kably, B., and trichromacy.
systems which enhance foreleg Lavoie, S. (1996). Role of the motor We employed a simple,
cortex in the control of visually triggered
flexor activity when stepping over gait modifications. Can. J. Physiol.
forced-choice ‘color-picking’ task
obstacles [9]. We consider the Pharmacol. 4, 426–442. with colorimetrically controlled
latter hypothesis the most likely 10. Wolpert, D.M., and Miall, R.C. (1996). stimuli separating the relative
Forward models for physiological motor
as it is consistent with theories control. Neural Netw. 8, 1265–1279. contributions of hue, saturation
of movement control in which 11. Wolpert, D.M., and Ghahramani, Z. and lightness. Observers used a
(2000). Computational principles of
motor control signals are used to movement neuroscience. Nat. Neurosci.
mouse cursor to select, as rapidly
predict the resulting movement 3 (Suppl.), 1212–1217. as possible, their preferred color
and the associated sensory 12. Flanagan, J.R., Vetter, P., Johansson, from each of a series of pairs of
R.S., and Wolpert, D.M. (2003).
feedback [10–13]. Prediction precedes control in motor small colored rectangles presented
The significance of this finding learning. Curr. Biol. 2, 146–150. sequentially in the center of an
13. Kawato, M. (1999). Internal models for
is that it clearly provides an motor control and trajectory planning.
otherwise neutral CRT display.
example of a signal related to the Curr. Opin. Neurobiol. 6, 718–727. (See Supplemental data available
movement of the limbs (forelegs on-line with this issue for details of
Department of Physiology, University
in this case) being used to of Alberta, Edmonton, Alberta, Canada experimental procedures).
update the representation of the T6G 2H7. We tested 208 observers, aged
location of obstacle close to the E-mail: kpearson@ualberta.ca 20–26. The main population (171)
Current Biology Vol 17 No 16
R624

Figure 1. Mean hue prefer- S–(L+M) (‘blue–yellow’) and L–M

A 1.0 ence curves.
UK Female (n=92)
(‘red–green’) neuronal mechanisms
(A) British subjects. (B) Chi-
UK Male (n=79) which encode colors. We therefore
nese subjects (±s.e.m.).
0.8 Hue values are obtained decomposed the hue preference
Mean proportion preferred

from CIE-LUV coordinates, curves in terms of fixed basis

using the background color functions which explicitly match
0.6 as reference white. The the two cone-opponent contrast
horizontal bar indicates only components (see Supplemental
approximately the tested
0.4 data). These account for 70%
hues and is not an accu-
rate reproduction of the of the population variance. For
spectrum. sub-populations by sex and
0.2 nationality, the fixed components
account for between 64% (Chinese
0 females) and 72% (British females)
1 2 3 4 5 6 of the variance.
Hue angle (radians) Each individual hue preference
B 1.0 curve is thereby reduced to
China Female (n=18) two physiologically meaningful
China Male (n=19)
weights. While the ‘blue–yellow’
0.8 contrast component accounts for
Mean proportion preferred

the greatest variance across the

0.6 population (44.5% S−(L+M); 25.5%
L−M), the ‘red–green’ contrast
component accounts for the
0.4 greater variance within the male
population alone (41% L−M; 28%
0.2 S–(L+M)). Only the ‘red–green’
weights show a consistent sex
difference across all populations.
0
1 2 3 4 5 6
On average, all males give large
Hue angle (radians) negative weight to the L−M axis,
Current Biology whereas all females weight it
slightly positively (sex difference
p < 0.00001). That is, females
were British Caucasian (79 male). A The average female preference prefer colors with ‘reddish’ contrast
sub-population (37) were mainland rises steeply to a sustained peak against the background, whereas
Han Chinese (19 male), the majority in the reddish-purple region, and males prefer the opposite. On
having left China for the UK within falls rapidly in the greenish-yellow average, all subjects give positive
the past year (range 0.5–3 years). region, whereas the male weight to the S–(L+M) contrast
Observers were tested in three preference is shifted towards component (‘bluish’ contrasts),
different experiments, each of blue-green and less pronounced. with British females weighting it
which included the same pair-wise Although there is a significant significantly higher than British
comparisons for a standard main effect of hue for both sexes males (p < 0.00001) (Figure 2).
group of eight colors (varying independently (p < 0.000001 Although male reaction times
hue; saturation 0.5; lightness 80). males; p < 0.000001 females) are significantly faster on average
A subgroup of 90 subjects (28 and together (p < 0.000001), the (1.26 seconds) than female (1.33
British females, 25 British males, variance in preference over all hues seconds) (p < 0.00001), both
and the Chinese sub-population), is significantly greater for females females and males respond faster
performed the standard versus males (p < 0.00001). to ‘bluish’ versus ‘yellowish’
experiment twice, with a two-week Individual female preference contrasts (reaction times correlate
interval. Here we report results for curves are also more stable over negatively with S-cone-contrast
the standard color group common time, for the subgroup of 90 increments of the preferred hue;
to all experiments. subjects tested twice (p < 0.002). female r = −0.1061, p < 0.00001;
We obtained hue preference The predictability of the male r = −0.0348, p < 0.01).
curves by plotting for each of the individual hue preference curves Thus, while both males and
eight standard hues the proportion prompted us to seek more concise females share a natural preference
of trials on which it was preferred descriptors. Principal component for ‘bluish’ contrasts, the female
(Figure 1). We found that hue analysis reveals that three factors preference for ‘reddish’ contrasts
preference curves do not vary alone explain 79% of the variance further shifts her peak towards
significantly for different lightness across the entire population. the reddish region of the hue
and saturation levels (Figure S1 in The first two factors strongly circle: girls’ preference for pink
the Supplemental data). The mean resemble the cone-opponent may have evolved on top of a
hue preference curves for males contrast components of the natural, universal preference
and females differ significantly. stimuli — the fundamental for blue. We speculate that this
Magazine
R625

preference, which are consistent

A
0.3 0.06 with the evolution of sex-specific
S-(L+M) contrast L-M contrast behavioral uses of trichromacy.
Yet while these differences may
0 be innate, they may also be
0
modulated by cultural context or
−0.2 2 4 6 −0.06 individual experience. In China,
red is the color of ‘good luck’,
B C and our Chinese subpopulation
2.0 3.0 gives stronger weighting for
* p<0.05
*** *** p<0.00001
* reddish colors than the British.
*** 2.0
1.6 Color preference patterns are
1.0 *** *** nonetheless systematically
1.2 governed by sensory encoding,
0.0 and, to a significant extent,
0.8 predictable.
–1.0

0.4 Supplemental data

–2.0
Supplemental data, including experimen-
0.0 –3.0 tal procedures, are available at http://
All UK China All UK China
www.current-biology.com/cgi/content/
Male Female full/17/16/R623/DC1
Current Biology
Acknowledgements
Figure 2. Decomposition of individual hue preference curves into two cone-­contrast
We thank Lucy J. Robinson for her
components, the weights on which account for differences between sexes and
­assistance in collecting data for the main
­cultures.
sub-group of subjects, and Thomas
(A) Cone-contrast components of the eight standard colors as a function of CIE-
Pollet and Rachel Sore for additional
LUV hue angle (radians), used as basis functions to extract individual hue preference
weights. Left: S−(L+M) contrast. Right: L−M contrast. (B) Mean weights for S−(L+M) ­assistance in collecting data.
contrast component. (C) Mean weights for L−M contrast component. Means are for
the entire population (‘All’; 98 males; 110 females) and constituent sub-populations References
separately (‘UK’; 79 males; 92 females) (‘China’; 19 males; 18 females). Significant dif- 1. McManus, I.C., Jones, A.L., and Cottrell,
ferences based on independent two-sample t-tests are marked with asterisks; error J. (1981). The aesthetics of colour.
bars are s.e.m. Perception 10, 651–666.
2. Granger, G.W. (1952). Objectivity of colour
preference. Nature 170, 778–780.
sex difference arose from sex- for the evolution of trichromacy 3. Ou, L.C., Luo, M.R., Woodcock, A.,
and Wright, A. (2004). A study of colour
specific functional specializations is the need to discriminate subtle emotion and colour preference. Part III:
in the evolutionary division of changes in skin color due to Colour preference modeling. Color Res.
Appl. 29, 381–389.
labour. The hunter-gatherer emotional states and social-sexual 4. Greene, K., and Gynther, M. (1995). Blue
theory proposes that female signals [9]; again, females may versus periwinkle: color identification and
brains should be specialized for have honed these adaptations gender. Percep. Mot. Skills 80, 27–32.
5. Bimler, D., Kirkland, J., and Jameson, K.A.
gathering-related tasks and is for their roles as care-givers and (2004). Quantifying variations in personal
supported by studies of visual ‘empathizers’ [10]. color spaces: Are there sex differences in
color vision? Color Res. Appl. 29, 128–134.
spatial abilities [7]. Trichromacy As further support for the 6. Alexander, G.M. (2003). An evolutionary
and the L–M opponent channel ‘female brain’ hypotheses, we find perspective of sex-typed toy preferences:
are ‘modern’ adaptations in that observers’ femininity scores pink, blue, and the brain. Arch. Sex.
Behav. 32, 7–17.
primate evolution thought to on the Bem Sex Role inventory 7. Silverman, I., and Eals, M. (1992).
have evolved to facilitate the correlate significantly with Sex Differences in Spatial Abilities:
Evolutionary Theory and Data (New York,
identification of ripe, yellow fruit L−M cone-contrast component NY: Oxford Press).
or edible red leaves embedded weights for all subjects (rho = 8. Regan, B.C., Julliot, C., Simmen, B.,
in green foliage [8]. It is therefore 0.333; p < 0.002), but not with S Vienot, F., Charles-Dominique, P., and
Mollon, J.D. (2001). Fruits, foliage and the
plausible that, in specializing for cone-contrast weights, for the evolution of primate colour vision. Phil.
gathering, the female brain honed tested subgroup of 90 subjects. Trans. R. Soc. Lond. B 356, 229–283.
9. Changizi, M.A., Zhang, Q., and Shimojo, S.
the trichromatic adaptations, Within cultures, for the Chinese (2006). Bare skin, blood and the evolution
and these underpin the female sub-population, masculinity of primate colour vision. Biol. Lett. 2,
preference for objects ‘redder’ correlates negatively with L−M 217–221.
10. Baron-Cohen, S. (2002). The extreme male
than the background. As a cone-contrast component brain theory of autism. Trends Cog. Sci. 6,
gatherer, the female would also (r = −0.531; p = 0.019) whereas 248–254.
need to be more aware of color for the British sub-population,
information than the hunter. This femininity correlates positively
Institute of Neuroscience and School
requirement would emerge as with L−M cone-contrast weight
of Biology and Psychology, Newcastle
greater certainty and more stability (r = 0.437; p = 0.002). University, Newcastle upon Tyne NE2
in female color preference, which Our results demonstrate 4HH, UK.
we find. An alternative explanation robust sex differences in color E-mail: anya.hurlbert@ncl.ac.uk