Thursday, October 29, 2009

Are we all Middle Easterners now?

Dienekes is arguing that Middle Eastern farmers demographically replaced Europe’s original population between 8,000 and 3,000 years ago. This argument seems to be proven by two recent papers that show no genetic continuity between Europe’s late hunter-gatherers and early farmers. The continent’s current gene pool seems to owe very little to the original Upper Paleolithic and Mesolithic inhabitants. So goes his argument.

This argument raises one obvious problem. It implies that the physical characteristics of Europeans, especially northern Europeans, arose recently and over a short time.

How short? As late as 7500 years ago, hunter-fisher-gatherers still inhabited Europe above a line running from the Netherlands to the Black Sea. The line then gradually moved north, reaching northern Germany about 5500 BP and the eastern and northern agricultural areas of Scandinavia around 4300 BP. This leaves very little time for the evolution of the northern European phenotype, i.e., lightening of the skin to pinkish-white and diversification of hair and eye color into a wide range of hues. This phenotype is attested by historical records going back over two thousand years, so we’re left with a time window of less than five thousand years.

Is that enough time for so much phenotypic change? Perhaps, but the selection pressures would have to be very strong.

Let’s turn to the first of the two papers. Bramanti et al. (2009) compared mtDNA sequences from late hunter-gatherers and early farmers who had lived in northern and central Europe (Lithuania, Poland, Russia, Germany). There was no evidence of genetic continuity between the two populations.

But this paper raises several other points:

1. Modern Europeans are almost as distant genetically from the early farmers as they are from the late hunter-gatherers. To be ancestral to modern Europeans, these farmers and their descendents would need a very low female population size (less than 3,000 individuals). As the authors admit, this figure is well below current archaeological estimates (124,000 individuals).

2. The sample sizes are very small for the early farmers (25 individuals) and the late hunter-gatherers (20 individuals).

3. The sample of late hunter-gatherers covers a much longer time frame (15,400 – 4300 BP) than does the sample of early farmers (7650 - 7400 BP).

In sum, the authors have tried to describe the gene pool of late European hunter-gatherers with data from 20 individuals spread over four countries and over some 11,000 years.

Can such a sample be representative? Doubtful. Besides the smallness of the sample, the late hunter-gatherers were not a homogeneous population. By their time, Europe had completely changed ecologically. Open tundra had given way to forest and it was no longer possible to hunt wandering herds of reindeer. Hunter-gatherers now lived in smaller and more localized groups. Each group would have had its own genetic profile as a result of genetic drift and founder effects.

Even if these 20 individuals fairly represented late hunter-gatherers, the genetic continuity hypothesis is not disproved by genetic differences between them and early farmers. Undoubtedly, some hunter-gatherers adopted farming earlier than others and thus contributed more to the early farmer gene pool. Others never adopted farming and thus contributed nothing. Founder effects would have been considerable.

There are thus two serious problems with Bramanti et al. (2009):

1. The sample of late hunter-gatherers is too small and too scattered over space and time to be representative of the late hunter-gatherer gene pool;

2. The genetic continuity hypothesis does not assume that the early farmer gene pool was a representative cross-section of the late hunter-gatherer gene pool.

Let’s turn to the other paper. Malmström et al. (2009) retrieved mtDNA from 19 late hunter-gatherers and 3 early farmers who lived in southern Scandinavia. The late hunter-gatherers show no genetic continuity with the early farmers or with modern Scandinavians but they do show genetic continuity with modern Baltic populations (i.e., Latvians). This seems consistent with archaeological evidence that the eastern Baltic was a refugium for Europe’s last hunter-gatherers. Indeed, the inland boundaries of Latvia, Lithuania, and Old Prussia may hark back to a time when these people fished and sealed from coastal stations part of the year and then moved some distance inland to hunt game the rest of the year.

This study has the merit of being more narrowly focused in time and space. Like the other study, however, it suffers from very small sample sizes and the likelihood of founder effects. In fact, the early farmer sample is so small that genetic continuity with modern Scandinavians is unsure.

What now?

The challenge now will be to enlarge this sample of late hunter-gatherers. By ‘enlarge’, I don’t simply mean a larger sample. I also mean a larger number of geographic locations to be sampled. Late hunter-gatherers were a heterogeneous bunch. Some contributed a lot to the future gene pool. Others went extinct.

The ‘losers’ were small inland hunting bands with low population densities. They were less able to integrate agriculture into their nomadic way of life and also more likely to retreat in the face of much larger farming communities.

The ‘winners’ were semi-sedentary coastal groups with relatively high population densities. Because such groups depended more on fishing and sealing than on hunting and gathering, they could more readily integrate farming into their lifestyle, if only as a secondary subsistence activity. They were also more numerous and likelier to withstand encroachment by farming communities.


Bramanti, B., M.G. Thomas, W. Haak, M. Unterlaender, P. Jores, K. Tambets, I. Antanaitis-Jacobs, M.N. Haidle, R. Jankauskas, C.-J. Kind, F. Lueth, T. Terberger, J. Hiller, S. Matsumura, P. Forster, & J. Burger. (2009). Genetic discontinuity between local hunter-gatherers and Central Europe’s first farmers, Science, 326, 137-140

Malmström, H., M.T.P. Gilbert, M.G. Thomas, M. Brandström, J. Storå, P. Molnar, P.K. Andersen, C. Bendixen, G. Holmlund, A. Götherström, & E. Willerslev (2009). Ancient DNA Reveals Lack of Continuity between Neolithic Hunter-Gatherers and Contemporary Scandinavians, Current Biology, doi:10.1016/j.cub.2009.09.017

Thursday, October 22, 2009

Face and gender recognition

It is told that an elder came to Scete with his son who was not yet weaned. The boy was raised in the monastery and did not know there were women. When he became a man, the demons represented images of women to him. He was astonished and informed his father. Now one day the two of them went to Egypt and, seeing some women, the young man told his father, “Father, those are the ones who would come and see me at Scete during the night!”

Sayings of the Fathers (Apophthegmata Patrum) 5th century (Regnault, 1966, p.73)

We seem to be born with the ability to recognize the human face. Even infants as young as 1 month old show a consistent, spontaneous preference for face-like stimuli over nonface-like patterns. Such recognition seems guided by an inborn representation of the main facial features, particularly the eyes and the mouth (Pascalis & Kelly, 2008). Brain-damaged subjects provide further evidence of a mental module that specifically processes facial images:

Associative visual agnosia does not always seem to affect the recognition of all types of stimuli equally. The selectivity in some cases of agnosia lends support to the view that there are specialized systems for recognizing particular types of stimuli. The best known example of this is prosopagnosia, the inability to recognize faces after brain damage. Prosopagnosics cannot recognize familiar people by their faces alone, and must rely on other cues for recognition such as a person’s voice, or distinctive clothing or hairstyles. The disorder can be so severe that even close friends and family members will not be recognized. Although many prosopagnosics have some degree of difficulty recognizing objects other than faces, in some cases the deficit appears strikingly selective for faces. (Farah, 1996)

If this mental representation is inborn, does it come in two forms, one for a female face and another for a male face? Or is it gender-neutral? By studying visual adaptation to facial images, Little et al. (2005) concluded that different neural populations process male and female faces. This difference seems to exist at the level of higher-level neurons that code for the entire face, rather than for specific characteristics (Bestelmeyer et al., 2008). These findings were partially replicated by Jaquet (2007), who found evidence for both common and sex-selective neurons.

Ramsey-Rennels and Langlois (2006) reviewed the literature on male and female face recognition by infants:

First, 3- to 4-month-olds have more difficulty discriminating among male faces and subsequently recognizing them than they do female faces (Quinn et al., 2002). Second, older infants are more skilled at categorizing female faces than they are at categorizing male faces: Whereas 10-month-olds easily recognize that a sex-ambiguous female face does not belong with a group of sex-typical female faces, they have more difficulty excluding a sex-ambiguous male face from a group of sex-typical male faces (data interpretation of Younger & Fearing, 1999, by Ramsey et al., 2005). In addition, there is a lag between when infants recognize that female voices are associated with female faces and when male voices are associated with male faces; infants reliably match female faces and voices at 9 months (Poulin-Dubois, Serbin, Kenyon, & Derbyshire, 1994) but do not reliably match male faces and voices until 18 months. Even at 18 months, infants are more accurate at matching female faces and voices than they are at matching male faces and voices (Poulin-Dubois, Serbin, & Derbyshire, 1998).

This evidence could be interpreted in two ways: a) infants better recognize female faces because they have more experience with mothers than with fathers; or b) female face recognition develops earlier than male face recognition because humans have evolved to recognize a female caregiver at an early age. To date, there has been no attempt to replicate the above findings with mother-absent/father-present infants. Quinn et al. (2002) found that such infants show a weak preference for male faces (59%) but there is no indication that they are better at recognizing male faces than female ones.


Bestelmeyer, P.E.G., B.C. Jones, L.M. DeBruine, A.C. Little, D.I. Perrett, A. Schneider, L.L.M. Welling, & C.A. Conway. (2008). Sex-contingent face aftereffects depend on perceptual category rather than structural encoding, Cognition, 107, 353-365.

Duchaine, B.C., G. Yovel, E.J. Butterworth, & K. Nakayama. (2006). Prosopagnosia as an impairment to face-specific mechanisms: Elimination of the alternative hypotheses in a developmental case, Cognitive Neuropsychology,

Farah, M.J. (1996). Is face recognition ‘special’? Evidence from neuropsychology, Behavioural Brain Research, 76, 181-189.

Jaquet, E. (2007). Perceptual aftereffects reveal dissociable adaptive coding of faces of different races and sexes, PhD thesis, School of Psychology, University of Western Australia.

Little, A.C., L.M. DeBruine, & B.C. Jones. (2005). Sex-contingent face aftereffects suggest distinct neural populations code male and female faces, Proceedings of the Royal Society of London, Series B, 272, 2283-2287.

Pascalis, O., & D.J. Kelly. (2008). Face processing, in M. Haith & J. Benson (eds.) Encyclopedia of Infant and Early Childhood Development, pp. 471-478, Elsevier.

Quinn, P.C., Yahr, J., Kuhn, A., Slater, A.M., & Pascalis, O. (2002). Representation of the gender of human faces by infants: A preference for female. Perception, 31, 1109–1121.

Ramsey-Rennels, J.L., & J.H. Langlois. (2006). Infants’ differential processing of female and male faces, Current Directions in Psychological Science, 15, 59-62.

Regnault, D.L. (1966). Les sentences des pères du désert. Les Apophtegmes des pères. Sarthe: Abbaye Saint-Pierre de Solesmes.

Thursday, October 15, 2009

Sexual selection and ancestral Europeans

I have argued that sexual selection has varied within our species in both intensity and direction (men selecting women or women selecting men) (Frost, 2006; Frost, 2008). In particular, it seems to have varied along a north-south gradient with men being more strongly selected in the tropical zone and women in the temperate and arctic zones. Women appear to have been most strongly selected among humans inhabiting ‘continental steppe-tundra’. This kind of environment creates the highest ratio of females to males among individuals willing to mate—by making it too costly for men to provision additional wives and by greatly raising male mortality over female mortality through long hunting distances.

Today, tundra is generally limited to discontinuous patches of land: arctic islands and coastlines, alpine areas above the tree line, etc. Yet it is only when tundra covers large land areas that it can support large herds of migrating herbivores. Such herds can in turn support a relatively large human population, but at the cost of high male mortality—because the men have to cover long distances to seek out and follow the wandering herds.

As late as 10,000 years ago, continental steppe-tundra covered an extensive land mass, particularly in Eurasia. It was thus one of the main adaptive landscapes of modern humans during their evolution outside Africa. In particular, it might explain the unusual physical appearance of Europeans, i.e., their feminized face shape and their complex of highly visible color traits (diverse palette of hair and eye colors, depigmentation of skin color to pinkish-white).

At this point, people ask: “But why would this sexual selection play out only in ice-age Europe? What about northern Asia? There must have been lots of steppe-tundra there as well.”

There was, but it lay much further north than in Europe and was less hospitable to humans. It was all the more inhospitable because it stretched further into the heart of Eurasia and away from the warming and moistening influence of the Atlantic. Thus, the Asian steppe-tundra never supported as many humans as did the European steppe-tundra. Indeed, it seems to have been devoid of human life at the height of the last ice age (Goebel, 1999, pp. 218, 222-223).

On a map of ice-age Eurasia, the steppe-tundra belt would look like a large blotch covering the plains of northern and eastern Europe plus a narrower strip running farther north across Asia. By a geographic accident—a large mass of ice covering Scandinavia—it had been pushed much further south in Europe than elsewhere. This was where the steppe-tundra could support substantial and continuous human settlement.

When making this argument, I usually stress the word ‘continuous.’ But the word ‘substantial’ is probably more important. The larger the population, the greater the chance that interesting variants will appear through mutation:

Small populations have limited variability at any one time and low absolute incidence of mutation, and they may be subject to genetic drift. They are also likely to be narrowly localized and so more subject to rapid extinction by a regional catastrophe. … Other things being equal, the larger the population the more potential variability, at least, it is likely to have and the larger its absolute rate of mutation will be. (Simpson, 1953, p. 297)


Frost, P. (2008). Sexual selection and human geographic variation, Special Issue: Proceedings of the 2nd Annual Meeting of the NorthEastern Evolutionary Psychology Society. Journal of Social, Evolutionary, and Cultural Psychology, 2(4), pp. 169-191.

Frost, P. (2006). European hair and eye color - A case of frequency-dependent sexual selection? Evolution and Human Behavior, 27, 85-103.

Goebel, T. (1999). Pleistocene human colonization of Siberia and peopling of the Americas: An ecological approach. Evolutionary Anthropology, 8, 208‑227.

Simpson, G.G. (1953). The Major Features of Evolution, New York: Columbia University Press.

Thursday, October 8, 2009

Facial skin color and sexual selection

The human mind seems to use facial color to determine whether a person is male or female. A man has a relatively dark facial color that contrasts poorly with his lip and eye color. Conversely, a woman has a relatively light facial color that contrasts sharply with her lip and eye color (Russell, 2003; Russell, 2009; Russell, in press).

This kind of sex-recognition algorithm has been a channel for sexual selection in many species. When selecting a mate, an animal tends to choose the ones most easily recognizable as the opposite sex. Over many generations, such selection will cause the relevant sex-specific cues to be accentuated (Manning, 1972, pp. 47-49).

The degree of accentuation will depend on the intensity of sexual selection and on whether males have been selecting females or females selecting males. Among ancestral humans, sexual selection seems to have varied in both intensity and direction along a north-south gradient (Frost, 2006; Frost, 2008). In the tropical zone, women gathered food year-round, so a second wife would cost little in terms of food provisioning. With more men becoming polygynous, fewer women were left unmated. The pressure of sexual selection was thus on men, with women being the ones who could pick and choose mates.

This situation reversed outside the tropical zone. First, polygyny was costlier because women could not gather food in winter. Second, male mortality exceeded female mortality because men had to hunt over longer distances. Together, these two trends resulted in too few men competing for too many women. This was particularly so on continental steppe-tundra, where women had almost no opportunities for food gathering and where men had to hunt wandering herds of herbivores over long distances. The pressure of sexual selection was thus on women, with men being the ones who could pick and choose mates.

Sexual selection and lightening of skin color

If light skin is perceived as a sign of femininity, sexual selection of women should tend to lighten female skin. This kind of selection became possible once ancestral humans had left the tropical zone. On the one hand, there was less natural selection for dark skin as a barrier to intense UV radiation. On the other, as explained above, there was stronger sexual selection of women because they outnumbered men on the mate market. Women should thus be increasingly lighter-skinned than men with increasing distance from the tropical zone, this sex difference being greatest among those humans that once inhabited the large expanses of continental steppe-tundra in northern and eastern Europe. Since most skin pigmentation genes are not sex-linked, selection for lighter-skinned women would also lighten mean skin color (i.e., both males and females). Thus, mean skin color should likewise lighten along the same north-south gradient.

How do these predictions stack up against reality? They accurately describe geographic variation in mean skin color (Frost, 2008). But they poorly describe geographic variation in female depigmentation relative to male skin color. In fact, female skin reflectance exceeds male skin reflectance the most among humans at medium latitudes with medium skin color (Frost, 2007; Madrigal & Kelly, 2006). This may be a ‘ceiling effect’. Northern and eastern Europeans are close to the physiological limit of skin depigmentation. Their women cannot be much whiter than the mean skin color because they have, so to speak, very little headroom left—the mean skin color is already scrunched up against the ceiling of maximum skin whiteness.

Sexual selection and increase in facial color contrast

There seems to be similar geographic variation in the contrast between facial color and lip/eye color. This contrast is weakest among tropical humans. It is strongest, however, not among northern/eastern Europeans but among East Asians (Russell, in press). This is largely because East Asians have dark eyes and relatively light facial skin. The contrast effect is even stronger if we factor in their jet-black hair, which further sets off the lightness of the female face. Nonetheless, facial color contrast is no more sexually dimorphic among East Asians than it is among Europeans (Russell, in press).

Why would Europeans score lower than East Asians on facial color contrast? It may be that sexual selection for dark eyes and dark hair relaxed among ancestral Europeans once their facial skin had lightened to the point of becoming pinkish-white. At that point, the color contrast was more than sufficient. This, in turn, may have allowed rare color preference to generate sexual selection for diverse hair and eye colors. This process may have then acquired a dynamic of its own that competed with the older dynamic of facial color contrast. Alternately, rare color preference may have always been a weak selection pressure that manifests itself only under conditions of intense sexual selection (Frost, 2006; Frost, 2008).


In sum, if we examine geographic variation in skin color and in facial color contrast, the pattern is largely consistent with increasingly intense sexual selection of women along a north-south gradient. This selection would have been minimal among tropical humans and maximal among arctic humans, particularly those that once lived on continental steppe-tundra—where polygyny was constrained the most and where male mortality exceeded female mortality the most. There are, however, deviations from the expected pattern that may be due to ceiling effects and release of sexual selection for rare hair and eye colors.

Among ancestral Europeans, this process of sexual selection seems to have been a multi-stage process. It likely began c. 30,000 BP with the first penetration by modern humans of the steppe-tundra belt (southwestern France). This initial stage would correspond to certain physical changes that are common to Europeans and East Asians. Stage I ended with the onset of the glacial maximum (c. 20,000 BP), which blocked East-West gene flow by merging the Fenno-Scandian and Ural icecaps and by forming large glacial lakes along the Ob (Rogers, 1986; Crawford et al, 1997). Stages II and III would correspond to later physical changes that are specific to Europeans.

Stage I – head hair lengthens, face shape feminizes, skin lightens (30,000–20,000 BP ?)
Stage II – skin lightens to pinkish-white (20,000–15,000 BP ?)
Stage III – hair and eye color diversifies (15,000–10,000 BP ?)


Crawford, M.H., Williams, J.T., & Duggirala, R. (1997). Genetic structure of the indigenous populations of Siberia. American Journal of Physical Anthropology, 104, 177-192.

Frost, P. (2008). Sexual selection and human geographic variation, Special Issue: Proceedings of the 2nd Annual Meeting of the NorthEastern Evolutionary Psychology Society. Journal of Social, Evolutionary, and Cultural Psychology, 2(4), pp. 169-191.

Frost, P. (2007). Comment on Human skin-color sexual dimorphism: A test of the sexual selection hypothesis, American Journal of Physical Anthropology, 133, 779-781.

Frost, P. (2006). European hair and eye color - A case of frequency-dependent sexual selection? Evolution and Human Behavior, 27, 85-103.

Madrigal, L., & W. Kelly. (2006). Human skin-color sexual dimorphism: A test of the sexual selection hypothesis, American Journal of Physical Anthropology, 132, 470-482.

Manning, A. (1972). An Introduction to Animal Behaviour, 2nd edition, London: Edward Arnold.

Rogers, R.A. (1986). Language, human subspeciation, and Ice Age barriers in Northern Siberia. Canadian Journal of Anthropology, 5, 11‑22.

Russell, R. (in press). Why cosmetics work. In Adams, R., Ambady, N., Nakayama, K., & Shimojo, S. (Eds.) The Science of Social Vision. New York: Oxford University Press.

Russell, R. (2009). A sex difference in facial contrast and its exaggeration by cosmetics, Perception, 38, 1211-1219.

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

Thursday, October 1, 2009

Facial color and sex recognition

Upper left: average of 22 Caucasian female faces
Upper right: average of 22 Caucasian male faces
Lower left: white pixels are where the female average is lighter than the male average
Lower right: white pixels are where the male average is lighter than the female average

To a large degree, we do not learn to recognize whether a human face is male or female (Bestelmeyer et al., 2008; Little et al., 2005). This mental task is mainly performed by a hardwired algorithm that uses certain visual cues, one of them being facial color (Frost, 1994). Men are more reddish-brown in complexion because their skin has more melanin and hemoglobin (Edwards & Duntley, 1939). Women are paler and show greater contrast between the color of their face and that of their lips and eyes (Russell, 2009). This algorithm is used not only for visual recognition but also for tasks apparently related to sexual attraction and social dominance (Feinman & Gill, 1978; Ioan et al., 2007).

Richard Russell has investigated the way we use facial color to identify male and female human faces. In one experiment, he morphed together 22 photos of Caucasian female faces and then 22 photos of Caucasian male faces. The participants were clean-shaven and did not wear make-up. As we can see from the above composites, the visually average face is noticeably lighter when it is female than when it is male. There is also greater contrast between facial color and lip/eye color on the female face than on the male one.

Russell (in press) argues that the human mind uses lip and eye color as a benchmark for visual processing of facial color:
If female skin is lighter than male skin, but female eyes and lips are not lighter than male eyes and lips, there should be greater luminance contrast surrounding female eyes and lips than male eyes and lips. This would be important, because the visual system is sensitive to contrast rather than to absolute luminance differences. Indeed, luminance contrast is the cue to which most neurons in the early visual cortex respond. Moreover, contrast internal to the face would be robust to changes in illumination. The black ink of this text under direct mid-day sun reflects more light than does the white page under dim indoor lighting, yet in both contexts the text appears black and the page appears white because the contrast between the two is constant. In the same way, a sex difference in contrast could be a particularly robust cue for sex classification. If there is a sex difference in contrast it would also mean that the femaleness of the face could be increased by lightening the skin or by darkening the eyes and lips—either change would increase the contrast. (Russell, in press)

He goes on to argue that this sex difference in facial color is reflected in the development of women’s cosmetics.

The received style of cosmetics involves darkening the eyes and lips while leaving the rest of the face largely unchanged. This is one of two patterns of cosmetic application that could increase facial contrast (the other being to significantly lighten the entire face, except for the eyes and lips). (Russell, in press)

This same pattern has appeared in a wide range of culture areas (ancient Egypt, Mesopotamia, South Asia, East Asia, Mesoamerica), in some cases independently of influence from other culture areas.


Bestelmeyer, P.E.G., B.C. Jones, L.M. DeBruine, A.C. Little, D.I. Perrett, A. Schneider, L.L.M. Welling, & C.A. Conway. (2008). Sex-contingent face aftereffects depend on perceptual category rather than structural encoding, Cognition, 107, 353-365.

Edwards, E.A., & Duntley, S.Q. (1939). The pigments and color of living human skin. American Journal of Anatomy, 65, 1-33.

Feinman, S. & G.W. Gill. (1978). Sex differences in physical attractiveness preferences, Journal of Social Psychology, 105, 43-52.

Frost, P. (1994). Preference for darker faces in photographs at different phases of the menstrual cycle: Preliminary assessment of evidence for a hormonal relationship, Perceptual and Motor Skills, 79, 507-514.

Ioan, S., Sandulache, M., Avramescu, S., Ilie, A., & Neacsu, A. (2007). Red is a distractor for men in competition. Evolution and Human Behavior, 28, 285-293.

Little, A.C., L.M. DeBruine, & B.C. Jones. (2005). Sex-contingent face aftereffects suggest distinct neural populations code male and female faces, Proceedings of the Royal Society of London, Series B, 272, 2283-2287.

Russell, R. (in press) Why cosmetics work. In Adams, R., Ambady, N., Nakayama, K., & Shimojo, S. (Eds.) The Science of Social Vision. New York: Oxford University Press

Russell, R.( 2009). A sex difference in facial contrast and its exaggeration by cosmetics, Perception, 38, 1211-1219

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

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

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