Sunday, December 12, 2010

Gender/face recognition: hue and luminosity

Averaged female face (left) and averaged male face (right). The key facial regions for gender recognition, in terms of either response time or accuracy, seem to be where facial skin borders the lips or the eyes.

The human face is a special visual object. We do not learn to recognize it. Instead, it is processed by the brain via a hardwired mechanism. There seems to be an evolutionary tendency to hardwire recognition of objects that appear often enough while being significant enough to human existence.

One task of this mechanism is to tell a female face from a male face. To this end, we unconsciously focus on several visual cues, including skin tone. It is well established that skin pigmentation visibly differs between men and women. This sexual dimorphism reflects differences in both constitutive pigmentation (untanned skin) and facultative pigmentation (tanning capacity). In comparison to women, men have higher concentrations of melanin and hemoglobin in their skin and lower concentrations of carotene. Men also tan more readily than women do for equal exposure times.

The psychologist Richard Russell argues that this visual cue has two components: (1) the absolute luminosity of facial skin and (2) the contrast between this luminosity and that of the lips and the eyes. Now a study from the Université de Montréal points to a third component: differences in hue, i.e., the degree of ruddiness and yellowness. Hue assists gender recognition particularly where facial skin borders the mouth region. In contrast, luminosity is most helpful where facial skin borders the eye/eyebrow region.

This gender recognition works faster with hue than with luminosity. If the observer is too distant or the lighting too dim, the brain falls back on the “slow channel” of luminosity:

This suggests that humans do use chromatic cues for discriminating face gender: When it’s informative, they use it and respond rapidly (for evidence that color is perceived faster than shape, see Holcombe & Cavanagh, 2001; Moutoussis & Zeki, 1997a, 1997b); when it’s not, they have to rely on the more robust and more sluggish luminance cues.
(Dupuis-Roy et al., 2009)

Interestingly, the authors conclude that this mechanism may be located in the infero-temporal cortex, since this brain region is involved in both face perception and color perception.

H/T to Eugene


Dupuis-Roy, N., I. Fortin, D. Fiset, and F. Gosselin. (2009). Uncovering gender discrimination cues in a realistic setting. Journal of Vision, 9(2), 10, 1–8., doi:10.1167/9.2.10.

Russell, R. (2003). Sex, beauty, and the relative luminance of facial features, Perception, 32, 1093-1107.

Russell, R., B. Duchaine, and K. Nakayama. (2009). Super-recognizers: People with extraordinary face recognition ability. Psychonomic Bulletin & Review, 16(2), 252-257.

Russell, R. and P. Sinha. (2007). Real-world face recognition: The importance of surface reflectance properties, Perception, 36, 1368-1374.

Russell, R., P. Sinha, I. Biederman, and M. Nederhouser. (2006). Is pigmentation important for face recognition? Evidence from contrast negation, Perception, 35, 749-759.


Tod said...

Very much in line with your ideas as far as I can see. No offense, but as I understood it the post 2 weeks ago was suggesting that this kind of research was not being done in North America.

Peter Frost said...


Apparently, I was wrong. They're a group of graduate students who have taken an interest in this field of research. I should have more to say about them next week ...

Tod said...

For most of North America you weren't wrong, there is obviously something about Québec!

(Very sophisticated looking study anyway, that kind of research expertise has quite a few commercial applications too I should think)

Anonymous said...

Just a layperson here, but wow, this is interesting stuff you've been posting. Thanks.

It's so interesting that the second I clicked on this post, I recognized in a split second,
the female from the male.

OT for sure, but this facial recognition subject makes me wonder if it's at all possible that people who are gay might have had as infants difficulty recogizing or processing the male facial image from the female one.

Tod said...

If contrast is a secondary mechanism it's less likely Europeans evolved blue eyes before white skin.

I've read that women become a shade permanently darker with each pregnancy. That seems like an anomaly if lighter skin induces care and provisioning

Peter Frost said...


It would be interesting to see whether a lower capacity for gender recognition would correlate with more indiscriminate sexual behavior, i.e., increased bisexuality.


Point well taken. Sexual selection for lighter-colored hair and eyes may have been inhibited as long as skin color remained below a certain level of skin reflectance.

If we compare human populations, we see many examples of light skin co-existing with dark hair and eyes. But there are no examples of the reverse. Skin depigmentation seems to be a prerequisite for changes to hair and eye color.

Kiran said...


I am Asha, working as assistant professor in one of the colleges inIindia .I am keep seeing your blog from past 1 year its very informative. I would like to my phd related to face or object recognition could you help me out in this regard.