First published: March 29th, 2019
Status: Putting out ideas

---

I have encountered plenty of ideas within the field of sexual selection that seem completely misguided to me, and I have an urge to retort. In particular, I have found such bonkers things in the wikipedia page of the same name and of which I had expected much better. Hereby are my own thoughts about sexual selection and the evolution of beauty.

---

As so often happens, people elevate concepts to unduly levels of idealization when discussing evolution. It is fine to contrast “natural selection” with “sexual selection”, for example when asking whether a trait X was a result of the former or the latter, as long as one remembers and considers that both of these are but shorthands. Alas, not uncommonly people discuss these as if they were completely different principles or even distinct forces of nature, like gravitation and the electromagnetic forces are.¹

For the sake of order I shall review things from the basics:
Let us say we have different life forms in an environment. They are all capable of reproducing. As they reproduce, some progeny exhibit a variation in traits as a result of mutations in the zygote they developed from, and these variations are heritable, that is, they pass on to their own progeny. As far as these variations have differentiable effect on survivability, individuals with the variation that confers more survivability would tend to survive better, on avereage, all else being equal, and therefore reproduce more, and since the variations are heritable, these variations would increase in their frequency amidst this population. We say these variations are “adaptive” and to the principle that adaptive variations increase over time in frequency in populations we call “natural selection”.
We skip the evolution of sex for the moment and jump to assume that we have in our hands a particular species that reproduces sexually. Consequentially, for a member of the species to reproduce it is not enough for it to merely survive, but it must mate with a conspecific. In other words, it must either find or be found by such a conspecific, copulate with them, and have the offspring reach sexual maturity (and so on and on). We could say that to the degree that a trait is beneficial to bringing about this result of sexually mature offspring via sexual reproduction is (sexually?) adaptive, and that the tendency of such traits to increase in frequency within a population is the principle of “sexual selection”.
It is common to regard “natural selection” and “sexual selection” as distinct forces (originating already in Darwin’s writing?), but really the latter is a special case of the former. The traits whose shaping is attributed to the pressures of “natural selection” are those which facilitate survival at an environment, that is, render effectively the ability to gather food, escape predators, resist pathogens, resist parasites, and so on. We could have just as well had “food selection” and “predator selection”, narrowly referring to the differential survival-promotion of variable traits affecting a member of a species’ abilities to gather food and escape predators, respectively. The reason, of course, why these are lumped together and “sexual selection” stands out is that sexual reproduction adds a complexity of analysis that stems from (species-) self-referentiality. In non-sexually reproducing species the ability to (survive and) multiply is marginally greater than the ability to merely survive. A creature avoids predators, finds and eats food, and when sufficient energy and nutrition surplus had been acquired it either divides or lays eggs. The extra margin of ability necessary for multiplication refers to the ability to successfully divide or lay viable eggs which is distinct from the ability to, for example, gather food. The ability to multiply depends on the environment (for example, different kinds of eggs could be differentially viable at different humidity or ambient temperatures, or require nutrients that might be present in one environment and absent from another) and might be even affected by other conspecifics (for example, if they cannibalize eggs), but it does not require and therefore not dependent (in the strict sense of the word) on other conspecifics. This is in contrast to sexual selection. Not only does the “margin” responsible for multiplication is greater now as it encompasses not just the ability to successfully create a gamete/ zygote but also a behaviour that would lead to a successful copulation, but, more important to the complexity, the traits which affect and promote one’s success of reproducing are not merely present at the one individual, but also ––and increasingly so over time–– in the other conspecifics. To put it differently, (1) in order to sexually reproduce one must have conspecifics in the environment, (2) conspecifics in the environment of one individual are part of the latter’s environment, and, therefore, (3) when a species (that is, its members) evolves, so does that species’ environment, because the species is part of its own environment to which it adapts. And thereby lays the complexity.

To illustrate what I mean through an example, imagine that one member gains a reproductive advantage by somehow taking advantage of their mate. Since this ruse is advantageous the genotype responsible for this phenotype of this member spreads in the population until it becomes a common phenotype. Increasingly this phenotype “meets itself” and since the advantageous was exploitative, it is not possible that both sides reap the advantage. Perhaps their subterfuges “cancel each other”, but is it also possible that their efforts of mutual undermining result in a reproductive success that is lower than whenever two of the “old cooperative phenotypes” copulated. In this case it might be even advantageous to “be exploited” than to (try to) exploit another exploiter.²
This is an example of a negative frequency-depended selection. Were it a positive frequency-depended selection there would not be much complexity, but, On the other hand, one could also imagine more complex examples. Of course, in this example, since the exploitative phenotype could mate both with the non-exploitative phenotype as well as with itself, it means that either the phenotype is not sex-specific, or the members of the species are hermaphrodite.
When a species has two strictly-distinct sexes, the members of each sex would coevolve with those of the other sex. That is, how adaptive a genotype of one sex is would depend on the genotype of the other sex. As with “more general” natural selection, members would be in competition with each other, but in this case the competition would be within the sex (males competing with males, females competing with females) and never with members of another species, such as happens in natural selection when two or more species are “fighting” over the same niche, for example when predating the same prey. That is not to say that there is no conflict between the sexes. While the competition occurs within each sex, the terms of the competitions are set “between the sexes”, so that the females set the environment in which males compete with each other, while the males set the competition for the females. One could imagine that had diverging sexual strategies evolved within one sex which were responded to by divergent sexual strategies of the other sex, speciation could occur on basis of that.

Rendezvous and mutual recognition

Before I touch upon other theories, I’d like to march out my own theory about what’s behind “sexual selection”. First of all, to say the obvious, what drives sexual dimorphism is the different roles that each of the two sexes take in reproduction. More concretely, given an existing dimorphism (and generally, genotype and phenotype of a species), the traits that would be selected for are these that facilitate the reproductive function of a given sex and to the degree that they are are phenotypically restricted to that sex (i.e. appearing only in females when they are adaptive in females) or not maladaptive in the other sex when it exhibits them. The traits which I think sexual selection works on, other than these that are responsible for sexual reproduction proper (production of gametes, copulation and rearing behavior instincts and so on), are those that “solve” one of these two problems: location (finding) and recognition of sexually mature members of the opposite sex (contrasexual conspecifics, or concons, from now on). Solving the problem of location means being able to arrive to the same place at the same time as a concon –– to be able to find it or be found by it. Solving the problem of recognition means evolving a motivation for a behaviour vis-à-vis concons which leads to fertilization/conception, which often necessitates evolving a phenotype (behavioural or morphological) that elicits an analogous behaviour in a concon. The traits that solve either of these are those which, one could say, determine how “attractive” an individual is.
Why these problems? On the one hand, sexually reproducing creatures must be in a certain proximity to each other in order to mate. On the other hand, it is beneficial to spread out in space in order to avoid competition over food. Since they do spread out, in order to mate they have to (1) be able to find each other, that is, arrive at the same place at the same time, and (2) recognize members of their species of the opposite sex. These problems are not trivial, and they are solved through the utilization of auditory, visual and olfactory signals as well as through behaviour, such as the synchronised movement of salmon and of migratory birds to a special location in order to mate.
Both of these problems could be solved to a large degree by making one more conspicuous, whether visually, auditorily, olfactory or through any other signal that is perceived by members of the species. Such traits are otherwise often naturally selected against, as they make one more susceptible to predators and less capable of sneaking on prey.

I’m going assume that, all in all, if any member of a species simply mated with any other mature concon, that is, the first encountered concon that survived into maturity, the species would fare rather well. Surviving to adulthood is a clear proof of being adapt at surviving in the given environment.
In this case, not assuming monogamy and assuming an individual can bear children more than once in a lifetime, it would be adaptive to be able to find a mate as fast as possible once sexually mature and each time the member is ready to reproduce, as that would yield (everything else being equal) more begotten children in one’s lifetime. In this case, sexual selection would drive conspicuousness up, whether by making one more loud, more smelly or more strongly contrasted with one’s visual environment.
When the two sexes have different mating frequency capacities, as is often the case, for example when females have bigger and more expensive gametes, and/or when they need to expend time and energy post-conception in order for the offspring to survive to maturity, the sexual selective pressure to be conspicuous acts differentially on the two sexes. Assuming no paternal care, it is adaptive for males to seek and mate with as many females as possible (assuming cheap sperm and that males mate with less females than energy and time would have allowed them had they had infinite amount of female available to them), while it is adaptive for the females to discriminate mates in order to mate only with the fittest one. If a female became (through evolution) more conspicuous she would get more attention from males, but as any male mating with her beyond the first one would affect very little her reproduction, this conspicuousness wouild add little to her fitness. However, as it would also get the attention of predators, and the attention of any one predator (or prey being sneaked upon) would put her at a great existential risk, conspicuousness would be overall a disadvantage, and indeed in many species the females are inconspicuous.
On the other hand, while the same costs of attention from predators/prey apply to males, they do get an advantage by being more salient to females since mating with any additional female greatly increases their expected number of offspring. This creates a competition: at an environment with males that vary in their saliency, whenever a male is being found by a female, even if he mates with her, the lower a male is in his conspicuousness, the more likely it is that, on average, another male that was more salient than him had been already spotted by her and copulated with her, presumably making it less likely that the late-comer would impregnate her (though here comes in sperm competition).
And thus the females get to become choosy. Not merely because they do not mate indiscriminantly like males, but because being camouflaged (and so protected both from predators but also from “sexual predation” of conspecific males) they are the ones having an easier time finding concons, and therefore probably the sex which searches for the other. Once a female finds a male, she can decide whether to mate with him or not based on strictly “naturally selected” characteristics, that is, traits by which also a male would have judged females had he faced the opportunity to woo more than a single one at a single time.
As these secondary sexual characteristics (promoting conspicuousness) evolve, so would evolve the recognition of concons. Other than finding them, females also must be able to recognize males as concons, that is, their perceptive faculties must initiate a copulatory behaviour upon finding them. Failing to do so would result in no offspring. That recognition does not have to rely on the evolved secondary characteristics and could rely on primary ones, but perhaps as dimorphism grows between the sexes the recognition faculties must adjust in order to recognize concons as conspecifics at all. “Recognition” here does not refer to any cognitive process. Females do not have to “know” anything about the biological relation between themselves and the concons; they merely need to properly respond to them upon an encounter. After a shift has occurred whereby females recognize males by their secondary rather than the primary sexual characteristics the male-male competition intensifies as now these traits are not merely important for being noticed, but to be recognized as males at all.

My theory, therefore, if I may, is as follows. In species in environments where location of conspecifics is challenging and where predation renders a pressure to be inconspicuous (which I think is true for most species in most biomes) “sexual selection”, that is, natural selection that acts through one’s success in mating, would act to increase the conspicuousness of males vis-à-vis conspecifics³ in correlation to the difference in offspring-expenditure costs between females and males and in anticorrelation to the survival risks they confer upon the male such as greater risks of being predated on or greater difficulties in predation. My theory makes the following predictions:

1. The greater the problem of location is, the more conspicuous males would be. When the problem is smaller males would still compete against each other, but the competition is then more likely to be manifested in ways that do not put them in risk of death due to external causes such predation, but rather through confrontation, whether it is intrasexual aggression or sperm competition.

2. In social animals the difference in conspicuousness would be small. This is somewhat of a corollary of (1). When members of a species breed within their group the problem of location is not a problem at all. If they breed across groups then location of a mate is not a problem of finding an individual but a group, which is a. easier to find, the more so the bigger the average group is and b. depends on the conspicuousness of the group as a whole. A conspicuous male (or female) puts his whole group at risk (if it is preyed upon) and would also gather little advantage for himself if other forms of male-male competition is going on such as aggression.

3. The easier predation is, the greater the conspicuousness of males. This is stated somewhat ambiguously on purpose: it is the case both when the species in question is the predator, and when it is the prey. The easier it is to hunt prey (or when conspicuousness does not hinder it) as a predator, and the less predators there are as a potential prey, the more conspicuous males would be. I’ll only note that while this may sound trivial, some theories predict otherwise.

4. The greater the difference in investment between the sexes, the greater the conspicuousness of males. In particular, species that are monogamous or serially monogamous and rear their young together would diverge little in conspicuousness as being noticed by several mates is no longer an advantage to males, and so being noticed by conspecifics is as a big of a challenge to both sexes.
   Further, on the other hand, I would suppose that when the environment exerts a strong pressure against conspicuousness, males would tend to rear their youngs, but only if the problem of location is big enough that a male is more likely to increase his number of offspring this way rather than by finding another female.
   

5. Increased conspicuousness in predators is more likely to be enacted than static, such as an auditory call or sprayed smell as opposed to vivid coloration. Conspicuousness thus being a behaviour they can remain inconspicuous when hunting prey and so interfere little with their fitness in that regard. Of course, if they have no prey which could detect some static conspicuous trait, they could evolve it without bearing costs.

Aposematism

Aposematism (warning-coloration) could evolve from this mating-promoting conspicuousness, I claim. That is, a sexually reproducing species where males are more conspicuous might develop a poison that is aversive to predators. Over time the predators learn (or evolve) to avoid these colored males. At this point the conspicuousness becomes an advantage rather than a disadvantage, driving the males to become even more conspicuousness; while before the evolution of the poison predators rendered a pressure against conspicuousness, now they rendered a pressure in favour of it, since they are more likely to eat the less conspicuous animals. In turn it makes it a disadvantage for the females to be camouflaged which thence evolve to have the same conspicuousness.
I went through the category page of aposematic animals on wikipedia, looking for non-sexually reproducing animals. I didn’t find any. However, there was one species that might if not “contradict” than at least be an exception for the idea above, Acanthodoris lutea, a brightly colored sea slug, and perhaps more generally, its clade Nudibranch. I say an exception because while they do have eyes, it seems like they are only sensitive to light/dark and so they cannot be using color to recognize each other (unless, that is, the their brightness per se, as opposed to their particular color, is a source of mutual attraction). On the other hand, they derive their toxicity from the toxic sponges that they eat. Since these sponges themselves are often vividly colored I wonder if the chemical compounds that give them their colors might not simply end up in the sea slugs (introducing the toxins and the warning coloration concurrently). However, that simply moves the question about the coloration to the sponges. These are, too, sexually reproducing, but they release sperm cells into the water and so they do not even need perception to reproduce, let alone find and recognize each other, so the color is not used for intraspecies communication. However, from a brief review I’ve made, it seems like the sponges that are brightly colored live at abysmal depths of the ocean to which light scarcely penetrates and so color would scarcely be an observable phenomenon. I cannot say much more of this and so I’ll leave it at that.

This here is in contrast to what seems to be the prevailing theories of the evolution of aposematism, or at least as far as I can tell based on the wikipedia page. It seems they all assume that the toxic defences evolved first, followed by coloration, while I suppose that coloration evolved first, followed by the toxic defences (though coloration does intensify and spread to females after the evolution of the defences).
There was one study cited there⁴ which advocated sexual selection as the driver of warning coloration in several populations of Oophaga pumilio, the strawberry poison frog. However, I’d say that they got “sexual selection” wrong, and that made their analysis confused and full of contradictions. In their discussion they expressed their awareness of those contradictions, but they couldn’t square them together. For example, on the one hand they mentioned that bright coloration ––– which females prefer –– would make males more conspicuous to predators, thus treating it as a handicap-principle trait, while on the other hand, on another paragraph, they say that the color deters predators, and so the less bright females are more susceptible to predation. While I agree with the coda of their conclusion section that “[…] sexual and natural selection must be jointly considered in the study of aposematic evolution”, I think their study mostly failed to show the relationship of the two ––– given their orthodox views of sexual selection ––– if only because in 4 out of 5 of the populations they studied the females were as bright or even more so than the males,⁵ making this trait not sexually-dimorphic or even dimorphic in “the wrong direction”.

On Beauty

It has been known for a while that averaged faces, that is, face images produced by “averaging” a multitude of face portraits, are more attractive than most or all of the faces being averaged, even after correcting for “skin smoothening”. The common explanation for that is that an average(d)-looking face is an honest signal of the developmental stability of its owner, that is, of health and resistance to pathogens. However, I think that it is something else.
First of all, as far as I know, these studies always averaged faces of people of the same sex, and I suspect that the result might be somewhat different had the faces being averaged had a sex ratio of 1:1. Would that be the case, this is not to contradict the common explanation per se, but one would have to admit that humans are, then, also sensitive to the proper/healthy development of sex-specific features (that is, sex-hormones-induced). That this is the case is suggested by different studies; there are vartiations in preference of faces with more or less feminine/mascluine features, for example intraindividually over the menstrual cycle of a woman⁶ (and particularly so among females with more feminine faces), and interindividually across people of different constitutions.⁷

Sexual characteristics of faces

But first, we need to take a step back. As psychometric investigations relying on self-reporting questionnaires, all of these studies suffer from the same problem, namely, that subjects were reporting their opinions about their own attitudes, rather than acting on them. This, first of all, matters, because people don’t always act the way they imagine they would. Furthermore, it makes the results depended on the interpretation of the subjects of what they are being asked. What does “attractive” really mean? Is asking how attractive a face is the same as asking how visually pleasant it is, or is it the same as asking how likely it is that we would appraoch such a stranger? There was only one paper that I have encountered in which subjects were shown faces subsequently, for a total period of time of fixed duration, and in which the subjects could either prolong or decrease the duration of exposure to any one face with a pressing of a key. The results are ostensibely ambiguous –– the researchers don’t know “why” the subjects wished to look longer or less long at any given presented face ––– but, at the same time, they are much more definite, since it was behaviour that was measured, not responses to questions. There were studies that made the question more precise, for example, asking not merely “how attractive a face is” or “which one is more attractive” but querying about how much a person(‘s face) is desireable for a short-term relationship and how much as a long-term relationship. On the one hand it makes things clearer, but on the other hand it casts doubt on the whole premise of the methodology. For example, the former differentiation was used in advnacing a theory about a so-called “mixed mating strategy” and the results were interpreted in that light, namely, that that non-single women in their fertile period of their menstrual cycle have increased interest in “short term relationship” with men with masculine faces is part of a strategy to get impregnated by a man with good genes while continuing to get support (for herself and her child) from the less-masculine and more care-giving partner she has. But that with the masculine man would not be a “short term relationship” but what one would call “a one night stand”. That questionnaires say “short term relationship” and not “hook-up” or “one night stand” or “no strings attached sex” is somewhat understandable; these are prying questions and the researchers want to get sincere responses by avoding affrontery to the defensive sense of decorum of subjects. But this “mild” articulation of the question changes the question itself, so researchers cannot really be sure of what the answers answer. As oppose to euphemisms that are indirect but have a clear reference,⁸ the blurry language used in these studies is simply unclear. What is a “short term relationship”?⁹ On first glance it seems to be referring to a real phenomenon, but I think it doesn’t really hold under scrutiny. Not really a proof as much as a suggestion, but searching for “long term relationship” on Wikipedia directs one to the “Interpersonal relationship” article, but “short term relationship” leads nowhere. Relationships can be “short” (though how long is “short”?), but that is a retrospective and not a prospective view, though in these studies it’s the latter that subjects were asked about (“how much is this person desireable as a short term partner?”). People engage in inconsequential sex with each other (though only 11% of women do so with no emotional involvement whatsoever, according to one study I cannot now find, that is, only 11% engage in sex that is really “inconsequential”), but I don’t think people enter an intimate relationship with a plan for it “not being long lasting”, unless they are aware from the getgo that it is unlikely to go very long e.g. because they are in the country for a circumscribed period of time or because they are terminally ill. Even if for whatever reason (idealogical, mostly, I suspect) they “plan” it to be “short”, I think they really are deluding themselves and that at the end the course the relationship would take is not any more dependent on that plan than the activity of a person at 1:15AM is on the decision to watch “one last youtube video” fifteen minutes earlier.

Nonetheless, I do not hold that these investigations are not at all informative. On this particular topic, however, I’m a little skeptical. One paper by Christine R. Harris¹⁰ that sought to replicate the findings about the changing facial preferences along the menstrual cycle failed to replicate the findings, and given Harris’ seemingly very organized methodology I trust her investigation. Beside her empirical results she also raises several interesting points, for example questioning whether the conditions that would have made such preference relevant existed in the relevant evolutionary times, namely, that an ovulating woman would get together with a “good gened man” in order to concieve a child with him and then the man would simply leave her alone and she would raise the child with a care-giving man, having the “origins of the child” in secret.
Her empirical results not only did not replicate the previous studies of Penton-Voak et al, but they showed the opposite trend, even, that is, that women preferred less masculine men when they were fertile, which somehow goes along with a review of Jones et al¹¹ who generally held the opposite view, but whose introduced orthogonal theory is relevant here. They mention that in the “non-fertile” parts of the menstrual cycle progesterone is elevated, which is also elevated during pregnancy. During pregnancy a woman’s immune system is somewhat repressed, and, in addition, she is responsible via her health for the more sensitive health of her developing child, both of which should make her more avoidant of contact with sickly looking people lest she catches a disease. That is, when she is pregnant (elevated progesterone) it is sensible to be more attracted (approach, seek the company of) towards feminine looking faces (who are associated with more care-giving) and to faces that are similar to hers (a proxy for kin, which, again, are more likely to lend help than strangers), but these tendencies/ “restrictions” are relaxed when her progesterone is low (around ovulation) and she’d be less wary of contact with sick looking people (note that health here is not some sort of “developmental health” but can refer to a transient condition) and more masculine looking men. That being said, since masculine looking men are judged to be more healthy looking, the idea that women are avoidant of sickly looking people when their progesterone level are high goes along the Harris’ result showing a tendency towards preferring masculine looking men during the infertile parts of the cycle. Be as it may, Jones et al’s paper is important in that it points to the fact that “attraction” does not mean “sexual attraction” or, let’s say, “motivation to copulate with”, which seems to had been taking for granted as an assumption in studies where subjects rated faces of the opposite sex.

Just to touch on an additional point, the Jones’ paper cites a study¹² that found that women had a preference for masculine faces on “days of their menstrual cycle when testosterone was elavated”, i.e. around ovulation. While I won’t contest their finding about the association between the elevanted testosterone levels and the preference, I’d contest their association of the elevation and the particular point of the menstrual cycle. First of all, in their own paper, they found that the mean (unsigned! which is suspicious) difference between the highest testosterone test session (these were conducted daily aruond the same time of the day) and the day of ovulation was 7.25 days. Even given the long period of time (~14 days) within a menstrual cycle where the probability of conception is elevated,¹³ 7+ days within an estimated cycle of 28 days is a rather big deviation, where we would expect a synchronisation given the proposed theory. In fact, an unsigned difference of 7 days within a bell-shaped window of 14 days would mark either the very beginning or the very end of the fertile period, so that testosterone either decreases as conception likelihoold increases, or vice versa. In addition, the mean unsigned difference between the session with the lowest testosterone and ovulation was 9.38 days, which according to their paired t-test testing was significantly different. Nonetheless, a 2.1 day difference between polar values of testosterone (min and max) within a 28 day cycle seems very minute. A paper that specifically studied hormonal changes along the menstrual cycle found no regressional model that significanly fitted testosterone levels to the day of the cycle nor changes in its irregularity,¹⁴ though another paper –– which didn’t look for a regression model –– did find a significant rise of testosterone in the pre-ovulation (“fertile”) period.¹⁵

The Immunocompetence-handicap hypothesis

There is the Folstad and Karter’s immunocompetence-handicap hypothesis¹⁶ which has gained a lot of traction. The hypothesis is that, since testosterone inhibits the immune system and is responsible for masculine secondary traits, a masculine face is a honest signal of health: only healthy/ vigurous males could sustain the burden of elevated testosterone. I think the hypothesis is utterly wrong. First, out of the same grounds that I think that the handicap-principle is generally wrong.
Second, it assumes that masculine faces have to be a developmental response to raised testosterone levels, and I think this is false, too. The long-tail of the peacock or the big antlers of the deer are themselves the handicap. In the immunocompetence-handicap hypothesis the actual handicap is the immunosuppression, not the male face. Had the male face signaled vigour, one would have expected an evolution of a development of a male face that is independent of immunosuppression.¹⁷
Third, the immunocompetence-handicap hypothesis postulates that the masculinity of the face is responsive rather than intrinsic. That is, a male would develop a masculine face in response to faring well vis-à-vis parasaties, so that the development would depend on the environement (presence of parasites) and on his immune system. However, it seems like this is not the case. One study¹⁸ took photos of male and fmale siblings (identical and non-identical twins and non-twin siblings) at ages 16 and 18. They develop an algorithm that calculated a metric of “masculinity”. It seems to me that the paper erred about the assumptins that the immunocompetence-handicap hypothesis makes (which might be due to the kind of research that had been conducted in support of the theory, discussed below)) and therefore also on its conclusions, but these interpretations stand beside the study per se. However, regarding the study itself, they made some odd decisions which leads me to take the results with a grain of salt. Their algorithm calculated the metric based only on the shape of the face (texture and so on were not involved), but the number of landmarks on the face that they used was very sparse, 16. While it did successfully separate female and male faces into two distributions, they overlapped considerably and it correctly classified the gender of merely 80% of the faces, which seems rather low to me. On the whole, while I do not think that the algorithm/metric they had used did an excellent job, I do trust that it did what it claimed to do.
What they found was that the correlation of face masculinity was very high for identcal twins (r=0.5), lower for non-identical twins and siblings of the same sex (r=0.23¹⁹), and, most importantly, that it was rather high (r=0.25) for non-identical twins and siblings of the opposite sex. That is, more masculine faced males had more masculine faced sisters, which also correlated with lower attraction ratings (based on human rater) for the sisters. What this suggests is either one of two scenarios, both of which go against the immunocompetence-handicap hypothesis. One scenario is that face-masculinity is an intrinsic feature and not a developmental response, which goes against the premises of the hypothesis. The second scenario is that the masculinity is indeed a developmental response and that the correlation in masculinity of faces is a by product of the correlation of the strength of the immune system. However, if this is the case, it means that females’ faces are likewise responsive, that is, develop masculine features in response to successful coping with parasites, which goes against the whole premise of it being a strategy employed by men, especially since facial masculinity is regarded as less attractive in females. One could perhaps suppose that this strategy is “still evolving” and didn’t reach the optimum point of sexual-segregation, but perhaps that’s giving too much leeway to the hypothesis, and I don’t know whether it is likely to be the case ––– given that support for the theory is given by mentioning other species, as remote as fish and lizard, one can suppose that segregation would have been reached already.
Fourth, the hypothesis would predict that males in societies in environments with low parasitic load (e.g. urban Western societies?) would have on average significantly more masculine faces than those that are in environments with high parasitic load (e.g. modern hunter-gatherer societies, such as the Tsimané; rural communities?). I couldn’t find any papers of studies that looked into this, which leads me to suspect that there isn’t such a difference. Also, looking at footage of hunter-gatherer societies I don’t get the impression that their males are particularly feminine looking. Looking at the correlation between males’ facial masculinity and the parasitic load of the environement they live in could shed some light, though one must keep in mind that it is not the parasitic burden per se that is important, but also how well the males cope with them (hence the “immunocompetence”).
Finally, a meta-analysis of studies where testosterone levels were experimentally manipulated in animals did not find much support for the hypothesis, in particular among mammals.²⁰

Could we support a variant of the hypothesis by assuming that the masculinity is intrinsic and not a response? Even disregarding the results mentioned two paragraphs above, I think the answer is no. First, the responsiveness in the original formulation is a critical component that makes a big difference; it being there, and had empirical findings supported it, I’d say that my general reservations about the handicap principle (which postulates predetermined, that is, unresponsive, development) would not apply. Second, it would make very different predictions. From the point of view of the interaction of the scientific community with the theory it seems like it would make little difference, since as it is seems like researchers predict whatever they find (rather than find what they predict), always finding an explanatinos of their results ––– like good conspiracy theorists ––– which doesn’t make for a very good science, as the theory is rendered unfalsifiable.²¹ I’ll demonstrate how the predictions would be different if the morphology of the face is intrinsic and not responsive with one paper. In a paper by Rhodes et at²² they write, The immunocompetence-handicap hypothesis predicts that the expression of secondary sexual traits should be negatively associated with parasite burden, because individuals with good heritable resistance to parasites can afford the immunosuppressive costs of secondary sexual trait expression. This is correct, though I’d have changed the formulation to “[…] because individuals with good heritable resistance to parasites will develop secondary sexual trait expression despite their immunosuppressive costs, affordable due to higher immunocompetence”. However, if the facial morphology was intrinsic and not responsive we would have predicted either no association or the exact opposite; which of the two would have depended on our assumptions in regards to the linkage equilibrium between immunocompetence and facial morphology genes. Nowadays we would have assumed a linkage equalibrium; given modern hygiene and health care, including medication and vaccination, the impact of parasaties and disease on humans in Western societies is greatly diminished, child mortality is very low, and some tradeoff between facial masculinity and immunocompetence would hardly affect one’s survivability. This would mean that also men with masculine faces who wouldn’t have been able “to afford it” at former times now can reach maturity and reproduce, so that the association between masculine faces and immunocompetence would decouple. Nonetheless, one could expect that men with facial masculinity they cannot afford, while not at a real risk of dying, would be less healthy. Assuming, again, linkage equilibrium, that is, that concurrence of facial masculinity and immunocompetence is random, we would expect that, on average, men with more masculine faces would be less healthy than those with less masculine faces, a prediction that is opposite to the one made in the study. What if there was no linkage equilibrium? I think in this case we would expect that on average men with masculine faces would have the same level of health as those with less masculine faces. The environment exerts pressure on humans in the form of parasites and other pathogens so that only individuals with an adequate immune response would survive. This is irrespective of their facial masculinity; the immune system of masculine faced males simply has to be better to reach an adequate level to compensate for the immunosuppression, but at the end they would have on average the smae level on health.
In the study, the pictures whose masculinity was judged by subjects were of adolescents born in California between 1920 and 1929. During that time child mortality was still decreasing²³ and life expectancy increasing, so that communicable diseases (and presumably parasites, too) were a factor that led to the death of people prior to maturity, so one could suppose possiblly a linkage disequilibrium. Nonetheless, this disequilibrium cannot be total. Genetic recombination was still at play and one would expect immunocompetence, even when developed in mascunine faced men, to “leak” to the less masculine faced men. Supposedly an above adequate level of immunocompetence is not harmful. Being below adequate level was significantly more harmful reproduction-wise than the “loss of some children” to extramarital affairs that non-masculine-faced husbands supposedly suffered. So it seems that, on the whole, one would still expect a somewhat diminished health among masculine=faced men ––– again, if face morphology is intrinsic and not a response. But what the study found was a (small, but significant) positive correlation between health and facial masculinity ––– in accordance to the original formulation of the hypothesis.

Intrasexual communication

What I think is much more plausible is that facial masculinity is a part of intrasexual adverserial communication. Let’s first briefly discuss communication. When it goes on between adversaries it often facilitates a reduction of costs. During mating season, deer stags would fight each other. However, before they do so they assay each other, and often one would concede defeat without a fight, presumably knowing that he was very likely to lose. This pre-fighting assessment is a mutual communication where each of the males communicates his vigour to the other, and assesses that of the other.²⁴ When the difference in vigour is discernible and the oonflict is resolved without physical engagement, this mutual-assessment is beneficial to both. Both parties, the obvious would-be-loser and would-be-winner, save themselves the damages and energy expenditure that they would have incured fighting, whether to lose or to win. While the “unchallanged” winner gets a “free win”, the “unchallenging” loser suffers a defeat with no “overhead loses”. Both therefore benefit by making predictions on the outcome and “agreeing on it” by letting the probable loser fleeing unscathed instead of going at it head-to-head.
Facial expressions such as anger work similarly. When two humans interact with each other and one expresses anger at the other, what he communicates is a threat. He communicates that the other has done something that is not to his liking, suggesting that had he done it again it would not be without consequences. Expresssion of anger is used to manipulate the behaviour of the recipient without incurment of much costs in the form of energy or bodily damage. It’s a mutual arrangement; one party agrees to forgo a behavior and the other agrees to forgo the employment of violence. While people can coordinate behaviour in harmonious, eye-to-eye, ways, making consessions and reaching a mutually understood agreement, when someone expresses anger they essentially push some behavioural equilibrium to move in their favour; the concession they make is refraining from being violent, which is hardly a concession at all.

Facial masculinity in young adult males is correlated with testosterone levels.²⁵ Testosterone levels are correlated with aggression²⁶ as well as with dominance.²⁷ Could I suggest that a masculine (dominant) face is akin to a permanent angry-like expression? A masculine face would be dominant in a sense that it communicates (1) strength and (2) a certain relentlessness, an inclination to not cooperate.
Of course, an angry attitude (or facial feature) is not absolutely a winning strategy, and is not without its costs. If it was, we would all look very aggressive and constantly behave angrily. While acting and seeming dominant (tall, muscular, strong, unyielding, relentless) would benefit a person in any interaction –– so long it is not contested –– by letting the person have things their way, it affects negatively others’ willingness to interact with that person, both people who interacted with the dominant person before, and other people who haven’t interacted with him but got a whiff of his dominance from afar. Having a choice and all else being equal, a person might choose to turn to and interact with a cooperative person with whom he could get “a better deal” than to turn to a person he knows from experience or reputation (or appearance) to be domineering. Being aggressive, therefore, is a trade-off between gaining more at a given transaction and being let in on future transactions; by extention, having a dominant look is a general shift of the trade-off ––– on average people would be more yielding to avoid confrontation with the dominant looking person, but they would tend, on average, to be less inclined to interacting with him.
This trade-off is not completely “either-or”. The Frankish Merovingian kings who did not hesitate slaying their siblings and other relatives, as well as misbehaving subordinates, come to mind, but their use of violence alongside them being the beneficiaries of continuous cooperatinon is merely more explicit, or, rather, frank,²⁸ than the violence employed by contemporary authoritarian regimes. The power held by rulers extends the units of agency in the aforementioned “transactions” and alters the nature of the space/timeline along which agents make their decisions regarding interactions:
First, the agency of the leader goes beyond the single person. Even when “authority” has that leader as its locus of power, authority itself is somewhat abstracted and in a way involves the entire population as a component.²⁹ Partially it is a problem of information; even if all or the great majority of the population (including the government) finds the leader to be overly domineering as well as inadequate, unless there is a free flow of information e.g. through free press, the pepole of this majority would not be privy to this information, and would act therefore as if the leader is well supported ––– as it seems to be the case ––– or, in other words, cooperate with him and his authority, thereby rendering in behaviour an undesireable reality for themselves and others.³⁰
Second, while I can avoid interacting with a person I regard to be overly domineering by ceasing to meet up with him, I cannot avoid interacting with the state at which territory I reside, other than by leaving the country or by successfully bringing about the replacement of the regime, both of which are no easy operations.

On the other hand, one does not always have the freedom to choose or avoid interactions with another concrete individual. Undoubtedly many of us heard of stories of abusive people, such as employers or spouses, who would terrorize others to the extent that not only the victims feel compelled to cooperate, but they are afraid to seek help, for example by reporting the terrorizer to the police. I’m not referring here to an interaction that is circumscribed in time and place, but where one person would have control over others over the span of many years or even decades. That is, the person, by employing aggression, does not merely gets things his own way at the present, but dictates the pattern of future interactions. He manages to have it “both ways”; to eat the cake and have it too.³¹
How come? One way to subsume this case under the “trade-off rule” is to extend it by having it refer to non-atomic interactions, that is, with interactions that do not have to be localized in space in time. Indeed, many times the decision whether to interact again with a person is not availabe between “atomic interactions”,³² if only because of the way people associate with one another, whether it is shaped by compliance with social norms or innate “social instincts”. We could now look backwards and even use these concepts in order to define what an abuser is: a person who employs tactics in order to force another to come in interaction with them, and uses coercion in order to define the terms of any transaction. I suppose that often those people get things their way and die peacefully, but other times when the (non-atomic) interaction ends the abused party not only would cease to interact with the abusing party, but take measures in order to retribute.
This extention of the theory would subsume the example above of tyrranies, where the interaction between the leader and the led is non-atomic. And while some tyrants get away with it, some eventually get their due, such as Nero, Maximilien Robespierre or Nicolae Ceaușescu. This is true not only among humans, but also at least among chimpanzees. Alpha males rise to power not merely through strengh and the application of sheer force, but also through their ability to form “alliances” with others. Undoubtedly the alphaness of one chimpanzee is variably good for the rest of the group, and there have been accounts of former alpha male chimps being murdered after they had fallen from their mighty positions.³³

Honest signal

The biggest flaw of the immunocompetence-handicap hypothesis is its assumption of a necessary causal link between testosterone levels and facial musculinity. Had facial masculinity been absolutely attracive to the opposite sex and sexually selected for, human males would have evolved to develop such facial features independently of their immunosuppressive testosterone levels. However, I claim, the association between facial masculinity and testosterone level is adaptive when we regard the former as communicating dominance.

Assuming that testosterone levels correlate with dominant behaviour, and that facial masculinity correlates with the perception (by others) of being dominant, what I’d like to claim is that it is advantageous to seem dominant if you act dominant, to seem cooperative if you act cooperatively, or, in other words, that the development of facial masculinity in response to testosterone levels is adaptive. I’ll treat the behaviour and appearnace as binary features, which I think does not lead to a loss of generality.

Seeming dominant when dominant is advantgeous over seeming dominant when cooperative

Seeming dominant would generally speaking, on average, decrease other people’s willingness to approach and engage with the person. A person who just looks dominant incurs the costs of the appearance without utilizing the advantage their looks could confer if they used aggression/intimidation to get things his own way. If you seem dominant, its advantageous to act dominant.

Seeming dominant when dominant is advantgeous over seeming cooperative when dominant

Seeming dominant would generally speaking, on average, increase other people’s willingness to yield to unfair propositions made by the person. A person who acts domineering without seeming the part would not be taken as seriously (not being perceived as dangerous/ strong). However, by acting domineering he would decrease the willingness of those he tried to dominate in the past to engage with him again in the future. He asks for more than people would have been ready to give him, which possibly makes him seem even more unfair of a person than someone who demands the same but also looks domineerig. Possibly. Maybe becuase people have a general sense of what is right/ just as far as allocation of resources is concerned, and with a person who doesn’t seem domineering the gap between “what he demands” and “what he deserves” seem wider. Or, perhaps more likely, the dominant person, seeming more powerful (and able?), seems like a better person to appease, in line with his potential to hurt/contribute. That is, like a person who is more important to reckon with than the submissive looking person. If you act dominant, its advantageous to seem dominant.

Seeming cooperative when cooperative is advantgeous over seeming dominant when cooperative

Seeming cooperative would generally speaking, on average, increase other people’s willingness to approach and engage with the person. A person that seems cooperative would attract to himself the engagement of others, thus forming cooperative relationships faster and easier than the cooperative person who looks domineering. If you act cooperatively, its advantageous to seem cooperative.

Seeming cooperative when cooperative is advantgeous over seeming cooperative when dominant

Seeming cooperative would generally speaking, on average, decrease other people’s willingness to yield to unfair propositions made by the person. A cooperative appearance would attract others to engage with a person, but if he acts domineering in response to their approach they’d feel like they had gotten the wrong address and gradually disengage. The domineering person, on the other hand, would not get what he asked for, and, worse still, would get less than what he would have gotten had he made a cooperative approach, since the others reject him. If you seem cooperative, its advantageous to act cooperatively.

Cooperativity vs. Dominance

This goes beyond the question of the association between testosterone (dominance) and masculine faces, but the above calls for the missing comparison in order to complete the picture, namely, a comparison between being and seeming cooperative and being and seeming dominant. However, while the former comparsions were between a “stable” and an “unstable” strategy, here we want to compare between two “stable” strategies, and as for the answer which is advantageous we must reply that “it depends”. For one thing, it depends on the strategy employed by others, in a manner that is akin to the negative frequncy-depended selection that was mentioned above. In tribal societies, as well as among kin and to an extent friends in societies with a central authority, if everybody else employs a “cooperative strategy” it would be advantageous to be dominating. If everyone else is a pushover, much is to be gained by being a bully, as the latter will never face resistence and will thus reap much benefit for himself. On the other hand, if everyone else is non-cooperative, it might be advantageous to be cooperative; it is better to get the shorter end than getting no stick at all. While the rest of (non-)society keeps on fighting with each other, the submissive person gets a little bit here and there by yielding instead of promoting conflict to the end result that ––– all else being equal ––– he has a little more than average.

In the modern world, however, we interact with many people who are complete strangers, and our interaction is affected to a very large extent by existing institutions, such as the government, police, law, economic system and so on, which altogether shape how adaptive any one strategy (cooperation or dominance) is. It would be futile for a person to try and squeeze a lower price for cucumbers out of a cashier in a chain-supermarket. Any kind of aggression that would have any effect is a robbery, and that would provoke a response from the police and therehby very likely a costly outcome that would greatly exceed what had been gained with violence. In general, law and the police together discourage the utilization of dominance for the purprose of taking advantage of others in interpersonal interactions, and even “discourage the appearance” of being dominant as this attracts the attention of the arm of the law. On the other hand, a lassez-faire market economy encourages dominance in vertical relationships from above down; the system leads all economic activity towards an ever increasing productive efficiency which gets the cooperative non-squeezers out of competition.

Attractiong to dominance

I used the expression “intrasexual communication” above as a reference to the existing hypothesis that facial masculinity is part of an intrasexual competition. However, the dominance of a face is of course also communicated to members of the other sex. Moreover, a dominant face could be selected for, that is, it could be affected by a process of sexual selection, though here this terminology becomes unhelpful.
As mentioned above, whether being cooperative or dominant is more adaptive depends on various environmental (social and otherwise) variables. The appearnace and behaviour on the cooperaton/submission-aggression/dominance spectrum affects interaction both with potential mates and potential competitors, e.g. other males, though this competition is not exclusively over mates but also over other resources, which makes it perhaps more accurate to say that this trait undergoes social selection. Given a degree of monogamy in humans, the “choice of strategy” of a mate is relevant when mating (1) to the degree that it is heritable, (2) in the way that it affects resource acquistion and (3) distribution between partners and their children.

Some evidence for the above can be seen in one study that found a strong correlation between a preference of masculine faces in mates by women and the gini coefficient of the country the women were from.³⁴ That is, in countries with greater economic inequality women have a stronger preference for masculine-faced mates.
This makes sense from a strategic point of view. The stronger the pattern of inequality in a society is, the more there is to gain by being aggressive and the more there is to lose by being complacently agreeable; cooperating is unfortunately remunerated by being taken advantage of. It therefore would benefit a woman to couple herself with a man who knows how to advance himself in society and thereby provide resources, for her and her children’s sake. On the other hand, in a country with greater economic equality there would be less variance in how much a husband earns, and so what determines how much resources reach the wife and children is his degree of cooperativity.³⁵ Moreover, it might be the case that an increased dominance would not be viewed positively in an egalitarian society, thus decreasing rather than increasing how much a man can gain in resources. In either socioeconomic scenario the assumption is that women more or less correctly perceive the degree of economic inequality in the country and adjust their behaviour accordingly.³⁶ It is therefore that the intrasexual male-male (signaling) competition becomes an aspect, via monogamy, of intersexual male-female (male to female signaling) competition. Without monogamy males would be solely selected for their (presumably heritable) ability to survive in the social environment, such as their ability to seize resources at the expense of other conspecifics, independently of how they might have shared their resources had they stayed, and without the male-male competition males would have been selected only on the amount of investment they were expected to bestow upon the family (which does not necessarily mean selecting the one who maximises investment; see here). In either case, of course, the mate is also judged based on “general fitness”, the ability to survive in the non-social environement.

Face Averageness

What I’d claim is that the averaged face is more attractive because it fits better one’s internal schema of what a concon looks like. When selecting a mate or even being merely attracted to another concon one relies on more than a face, but when only a face is available one is attracted to the degree that the face looks like a member of the opposite sex (or of the same sex when one is attracted to these). In other words, one is attracted to a face to the degree that one recognizes that face as belonging to the opposite sex. Perhaps, even, to the degree that it does not look like the face of a member of the same sex while still looking like a face of a human. And to put it in the terms formerly laid, to the degree that the face “solved” the problem of recognition.
Why humans rely (also) on faces to distinguish the sex of other humans is a separate question. I can hypothesize different reasons, such as the loss of perception acuity of cues such as olfactory ones, or females evolving to have more infantile features³⁷ in order to look younger and therefore more fertile, altogether evolving a dimorphism of the face that one could rely on. It could also have to do with humans being social creatures with an elaborate interpersonal communication that utilizes the face as a central medium: the eyes are an important source of information, and while body movements are too, the signals of body language are large, and one can follow their expressiveness without directing attention away from the eyes. Attention on the face (via the eyes?) allowed for a communication to develop which relied on minute changes of the face, further making the face an attention-worthy locus of a conspecific body.³⁸ This necessitated the development of a perceptive faculty that could detect and distinguish these minute dynamic differences, which had to rely on the evolution of a precise perception of static facial features, which allowed us humans to, among other things, recognize each other through this medium.³⁹ This acuity could have been then leveraged for extraction of information about conspecifics through their face.⁴⁰ That information could have been related to fecundity and sexual relation, in turn creating a sexually selective pressure that shaped the faces of both sexes in their respective stereotypical directions, leading to dimorphism.

What I hypothesis the case to be is that humans have an innate (“geneticlaly encoded”) but rough visual schema of the human face, which allows them to correctly direct their attention to face-like looking things shortly after birth. I think that that is the case is rather well supported by the fact of the human tendency to “see faces” where they are not. Running an internet search for “things that look like faces” says it all; we merey require two dots above a horizontal line or blob to “see a face”. However, what faces “really” look like is learned through exposure throughout a person’s life. That is, a basic schema orients a person’s attention, and the attention leads to the learning of what a person’s face looks like. This learning, through some averaging, creates a detailed schema that allows a person to differentiate humans from non-humans, kin from non-kin, and males from females.⁴¹ It’s simpler to have this kind of information learned rather than being “hard coded”: it’s more efficient, since morphologically-evolved changes (of others) do not require a repsonsive evolution of the perceptual faculties, and it’s more effective because a species does not need to go through the failed experimentations of persons being born with “the wrong facial detailed schemas”, that is, detailed schemas that do not match the “facial environment” she is born into. Had such detailed schemas innate and not learned, it would have been less likely that the simple schema (two dots and a line) would persevere. Interestingly enough, there were studies that found that chimpanzees are able to learn, through exposure, to differentiate individual human faces.⁴² Moreover, there’s some research that suggests that humans are not the only species that can read emotions off human faces. For example, one paper showed rather convincingly that dogs (mainly Border Collies) can distinguish between happy and angry human faces.⁴³ Conversely, humans rely on the same auditory features of vocal expressions to assess the intent of the calls of humans and dogs.⁴⁴

This idea that preference for regular –– averaged, symmetrical – faces is due to the perceptional faculties’ design rather than an evolution responding to an adaptive advantage accrued through mating with developmentally-stable concons is not new; studies have suggested animals would prefer a novel averaged stimulus over non-novel non-averaged stimuli (as well as other novel stimuli).⁴⁵ It is important to note that as far as there is a preference to faces with low “flactuating symmetry”, it is a learned rather than an innate preference. That is, the symmetry to which there is an emergent preference is a result of averaging of faces one has been exposed to. Therefore the paper, for example, that found no preference for symmetrical over non-symmetrical upturned faces,⁴⁶ constitutes no evidence against this hypothesis (as the upturned faces constitute a set of unfamiliar stimuli). It is also important to note that for this all to work, individuals need to be able to correctly differentiate conspecifics from non-conspecifics, as to not average the latter into the image of the former. Other than relying on cues from other faculties (auditory, olfactory, when aveaging visual stimuli), supposedly one might not consider individuals whose appaerance is far enough from the existing image of what a conspecific looks like.
How could one disassociate the percetpual hypothesis from the developmental stability one? On the one hand, one could design an experiment where an animal, which recognizes conspecifics visually, is exposed only to one-sided asymmetrical conspecifics, and see how it affects its mate selection. It seems rather difficult to pull off. Perhaps one could apply colored stickers on one side of others’ bodies, but for this experient one would have to have the assumption that these colorations are attended to. One the other hand, one could look whether there really is a correlation between health and face averageness. Had there been one it would not have provided evidence for or against either of these two alternatives, since the preference might still be a product of the perceptional faculties rather than of selection,⁴⁷ but a lack of such a correlation would have been a strong evidence against the developmental stability hypothesis.

First, let’s consider a paper that sought to show a correlation between health and face normality. A 2000 paper by Rhodes et al.²² used black and white photographs of 17 year old individuals born between 1920 and 1929 in the USA whose health records were available and who were judged for asymmetry and non-avergeness (“distinctiveness”, in the study described to the raters as “the ease with which a face could be picked out of a crowd”) by contemporary subjects, as well as measured for asymmetry based on geometrical feautures of the pictures. While their report mentions some evidence for a correlation between ill-health and rated non-avergeness, their findings are not so convincing. First, their statistics are rather sloppy. They mention a “marginal correlation between midadult health and measured facial asymmetry at 17 in females”. They add that this should be interpreted with caution given the small sample size, but they ignore the glaring fact that the p value for the correlation was p < 0.1, found in two statistics out of 24 statistics derived for “measured asymmetry”. That is, at a rate of 2/24 = 0.083 < 0.1. Furthermore, given that in total they looked at 72 statistics, the three p < 0.01 and one p < 0.001 statistics they are found are somewhat unremarkable.
These significant statistics were found for a correlation between childhood ill-health and current (17 years old) rated non-averageness in males, and between adolescent and current ill-health and current rated non-averageness in females. The fact that there are gender differences in these correlations over developmental times is very suspicious. Perhaps something in development would give rise to different correations between face avergeness at 17 and health at different periods in life, but that it is variable between the two genders is extremely peculiar. Could it be explained?
Let us assume that the face is generally planned to develop symmetrically (further discussed below). We could make an assumption but this is almost a corollary of a definition: on average faces are planned to develop in a typical, that is, average, non-distinct, kind of way. We’ll assume that a failure to develop a averageness is a failure to develop “as planned”, and that being ill is a disturbance of the proper functioning of the body which might then lead to disturbances in proper development and therefore in developed averageness. How do we explain correlation between childhood ill-health and non-averageness at 17 (as seen in males)? The skull and face are formed in childhood so perhaps the “scars of development” at that time are strongly engraved into the shape of the face such that deviations from the plan “stick” onto adulthood, while disturbances later in life are less critical since the face had already formed. But, then, why would the females not exhibit such a correlation too? One could have explained this difference if, for example, the faces of females underwent a strong development later in life (adolescence and late-adolescence), that is, a stronger development than that which male faces undergo, but the reverse is true: female faces experience a decline in growth at 13 and stop growing at 15, while males exepience an extended facial development that starts at pubetry and continues until at least 18.³⁷
On the other hand, a paper by Pound et al.⁴⁸ with a large sample size (n=4732) of South West England individuals and a well conducted statistical analysis, found no correlation between fluctuating symmetry and a myriad of health conditions.⁴⁹ They suggested that the attraction to low-FA faces might be an overgeneralization of an adaptation that avoided mating with individuals with gross facial asymmetries. That is, small fluctuations are irrelevant to health but great fluctuations are, and the avoidance of the latter led to the avoidance of the former. I’d in turn say that the avoidance of gross asymmetries in mates is an overgeneralization of the avoidance of mating with individuals of others species.

• the recognition was earlier suggested
• However, it concentrated on the symmetry of stimulus, however:
• Can one decouple these?
• symmetry not associated with health.

• brain development is not symmetrical, and brain developmetn affects the face. (fluctuatings vs directional assymetry). THe autistic faces

• for fluctuating symmetry to indicate bad development, symmetry (as a whole) had to first evolve anyway Why do we have symmetry? symmetrical functionality But the shape/morphology of the brain, like other oragns, is not interfacting the outside (though the constraints on it ARE, such as the room it has). Brain affects face morphology, so FA/DA that is adaptive could evolve (autistic faces). Assmetry of external organs (hands, legs) could evolve becuase the brain is asymmetrical and employs them differently (perhaps the asymmetry is “phenotypical only”; are lefties asymmetrical in the same way?).

«also: recognizing faces of other animals. Recognizing expressinos of other animals: apes(?), dogs. Dogs recognizing our expressions.»```` «allowing us to quickly discern and decode the facial expresions of groups we had never encountered, even of creatuers we havn’t seen before?» «how sexual orientation arises? The notion of the gender to be attracted to»

«mention the other studies that postulated the recognition aspect, but criticize their assumption that it is due to symmetrical input being easier to process. And, indeed, that paper about the upside down faces»

« average because of developmental stability or because of a schema of an idea? –– decouple. Maybe seek some inspiration from the idea of the Platonic idelas. Is average really always better? no. Things about objects, tools, averageness there. Mention how extremes can be selected for (dimorphism) while averageness is selected for at the saime time. (one answer: the extreme is a product of seeking averageness elesewhere. For example, one can have an idea of the averagped child, having a schema of youth. Then females are selected for for looking young, which creates an “extreme selection” because they are not necessarily young. Nonetheless, the “direction” in which there’s a selectino is within the repertoire of human morphology.»

Dimorphism in the body might stem from exaggeration of functional differences and from intrasexual competition. A wider hip might be necessary for women especially in order to accommodate the big head of human newborns (the bottleneck of delivery as anyone who ever tried to move through a school fence even as a child or adolescent –– with smaller head to body ratio –– knows), but sex-specific accumulation of body fat in that area is presumably not “functional” but aimed to increase their perception (by others) as female, or at least avoiding decreasing that perception. Stronger skeletal muscles and, perhaps?, wider-shoulders might assist males if they often engage in physical fights between each other, but it also makes them, in turn, be recognized as such, males, to other humans. (voices): https://www.sciencedirect.com/science/article/pii/S0960982209020491

Alternative Sexual Selection theories

I’d like to hereby criticize a few existing theories. I quote some definitions from a 2008 paper:

Sexual conflict: Sexual conflict refers to intersexual conflict. It describes the diverging interests in males and females to optimize their fitness. From the point of view of one sex, this is done best if the mate cares for the young thus freeing resources (e.g., time and energy) that can be invested in additional offspring. Regarding to polygynous mating systems, sexual conflict means the optimization of male reproductive success by having mated with multiple females although reproductive success of a polygynously mated female is reduced.
Good genes: Good genes refer to the assumption of an additive indirect genetic benefit that increases offspring quality. Offspring quality, that is, viability, leads to a compensation of any inferior direct reproductive success, that is, fewer offspring.
Fisherian runaway selection: Fisherian runaway selection assumes indirect genetic benefits in mating decisions. This process assumes that sexual selection for a specific trait was at first correlated with a small increase in viability. During evolution, the female preference for this trait becomes exaggerated while the good genes effect loses its importance. At the end, sexual selection is driven only by the attractiveness of this trait and not by any additive genetic effect that is relevant for natural selection.
Sexy son hypothesis: The sexy son hypothesis (SSH) is closely related to the good genes assumption and the Fisherian runaway selection process. SSH, too, assumes indirect genetic benefits that are able to compensate any inferior direct reproductive success, that is, fewer offspring. The main difference between good genes and SSH is that SSH assumes an indirect effect due to the attractiveness of the sons whereas good genes focus on viability in sons and daughters. However, attractiveness is not specified and can be every trait that increases a male’s probability to become polygynous.
Compatible genes: Compatible genes also belong to indirect genetic benefits that increase the genetic quality of offspring. Whereas good genes generally lead to positive effects, compatible genes are beneficial only for specific individuals. That is, a positive effect depends on the interplay of the individual genetic architecture of both mates.⁵⁰

Regarding the Fisherian runaway selection. As stated above, my own theory is that traits which are sexually dimorphic and seem maladaptive from a natural selection point of view are traits that aim to solve the problem of mate location. But what about Fisher’s idea? The idea is that first there was the appearance of a trait T which was adaptive. At some point it got to be recognized by females, a recognition that was adaptive because they could choose mates that would lead to more adaptive offspring (all else being equal; presumably how fit a male is depends on more than a single trait). Because the trait was now “recognized as adaptive”, that is, it became sexually attractive, there was evolutionary pressure to exaggerate it in order to attract mates, only this exaggeration loses the benefit of the original trait.
I’d expect a stabilization of a trait. That is, the “runaway” part seem to me unsound. Imagine a population where virtually all males exhibit trait T, and that many or all have had this trait “run away”, that is, it surpassed the point in which it is adaptive and either lost its adaptive power or even became maladaptive, that is, worst than not having the trait T at all. We could expect –– by the very same theory –– that a male that exhibited a diminished version of the trait (T-), or even not exhibit the trait at all (T0), would start the same process but “in the opposite direction”. His offspring would be more adaptive than those of the T males, and those females who would recognize that would have their offspring better adapted, and so on until the trait would “run away” back. Or, perhaps more likely, the trait would oscillate until its optimal degree becomes the average phenotype of the population. Both “more than” and “less than” are really qualitatively symmetrical around the optimum. The Fisherian runaway principle is not consistent with itself.

The sexy son hypothesis is much more problematic, partially because it seems like the concept has evolved in meaning since its inception. The term first appeared on a 1979 paper by Partick J. Weatherhead and Raleigh J. Robertson, Offspring quality and the polygyny threshold: “the sexy son hypothesis”. They discuss bird populations and put forward a mathematical model. The premise is that there are males who are monogamous and males who are polygamous. The behaviour of the males is heritable, i.e. sons of polygamous males are polygamous, likewise for monogamous fathers and sons. Females who mate with monogamous males have more offspring than those who mate with polygamous males (thanks to paternal parental support), but polygamous males have more offspring than monogamous males. Doing some algebra they conclude that even though on the short-term the mates of polygamous males suffer in terms of lower offspring number, in the long term this coupling is advantageous to them because their sons are polygamous and so will have more offspring –– the mate’s grandchildren –– hence the “sexy sons”. The problem with the model is that it doesn’t hold water. The math is wrong.⁵¹

However, the sexy son hypothesis is often evoked by saying that sometimes females would prefer an attractive male that would not invest in the offspring over a less attractive male who would, because their offspring would likely to have sexy sons which in turn would have, being promiscuous, a large number of offspring that would compensate for the loss. That is, they’d have less children but more grandchildren. But as Huk and Winkel block-quoted above mention, often it is not specified what attractiveness is, and it really makes the model rather hollow. You might as well say “magic”. Why are they attractive? The words of the evokers often make it sound like the trait is arbitrary, that is, it is desirable and attractive because it’s attractive, it’s attractive because others find it attractive so the sons would be attractive and sought after. As if all females colluded and decided on a standard by which to judge males. But this is, of course, nonsense. One could imagine some evolutionary glitch that would make females be attracted to an “arbitrary” trait, but that would be the evolutionary equivalent of an economic-bubble, and we know what happens to economic-bubbles in the long run, which is the only run that evolution cares about.⁵²

I’d like to propose a variation on the hypothesis. I could have called it the “selfish son hypothesis” had it not had the same acronnymm, but then again “selfish” is also misleading. It could be called promiscuous son in times of plenty hypothesis (PSTPH), though it doesn’t quite rolls on one’s tongue. To use the same register as in the original, I’d call it the douchy son hypothesis (DSH). First, it assumes that females are strongly limited in the number of offspring they can produce at any given time period. That is, we are concerned here with animals that gestate or lay eggs that are of a size comperable to the size of the female, as oppose to animals that spawn in the hunders or thousands. Second, it assumes that males’ rearing style (monogamous, non-monogamous) is heritable. The hypothesis is that if the environment is “easy”, that is, not harsh, with plenty of availabe food and little predative danger, females would prefer as mates males who are promiscuous because, again, they would then have more grandchildren. If the mother can bring her offspring on her own, the father’s lingerilng presence confers no advantage. It could confer a surplus of time, but the mother has little to do with it. It could have spared her energy, but since energy is freely avaiable, nothing is really spared. Since the reasing style is heritable, mother of sons whose father is monogamous would at most mate with one female. However, if the female mates with a male who is promiscous, that is, one that would go and seek other females to mate with after he mates with her, her sons (which would reach maturity at the same rate as if the father stuck around, given the giving environment) would likewise be promiscuous and would mate, on average, with more females than the sons she could have had with a monogamous father, and therefore she would have more grandchildren. Therefore, in such conditions, if females could at all discern “monogamous” males from “promiscuous” males, it would benefit them to reject monogamous males and mate only with males who are promiscuos.

Another suggested principle is the handicap theory of Amotz Zahavi. To put it in his own words,

My interpretation of sexual selection implies that sexual selection is effective only by selecting a character which lowers the survival of the individual. [An argument along this line was given by Emlen (1973) among several other arguments.] It is possible to consider the handicap as a kind of a test imposed on the individual. An individual with a well developed sexually selected character, is an individual which has survived a test. A female which could discriminate a male possessing a sexually selected character, from one without it, can discriminate between a male which has passed a test and one which has not been tested.⁵³

That is, a handicapped animal, usually male, proves through his displayed handicap (bright and large plumage, antlers &c) that it had managed to survive despite the handicap it has. The handicap is a sort of compromise; on the part of the females because they have to choose a mate who is disadvantaged (by the handicap) in order to be sure that he is otherwise “fit”, and on the part of the males as they have to handicap themselves in order to advertise their fitness. I must say that the theory is rather attractive, and that the paper presenting it is generally well put. However, it has some problems.
Shortly after the Zahavi paper was published, a paper by John Maynard Smith⁵⁴ was published which set to make a mathematical model out of Zahavi’s model, and found it to be implausible.⁵⁵ Though I think Smith might have incorporated erroneous assumptions into the model which left me less than convinced by it, it’s a good step in the direction of investigating the plausibility of the hypothesis, and I’d like to rerun such a model with the corrections I think it deserves.
That being said, I don’t imagine the handicap principle would emerge triumphant. As opposed to Smith’s model where you had a genotype that led to females mating only with handicapped males and another which led females mate randomly, imagine the genotype separated females to those who mated only with handicapped males and to those who mated only with non-handicapped males. Then we’d have two separate populations who only interbreed, but with some trickling of offspring from one group to the other (though I suppose that over time this crossing would go to zero). All else being equal, it seems clear that the non-handicapped population would grow faster ––– assuming there’s no “population size cap” (rendered by scarcity of resources or predation, or both). That is, on average, members of the non-handicap population beget more offspring that reach maturity than those of the handicapped population. That is, the former population is fitter. If we introduce a new female that can freely choose which population to belong to, she’d choose the non-handicapped one. Same with a newly introduced male. Which brings me to the one graph presented in the Zahavi paper:

We have here an abstraction, a quantification of traits of animals, that is, a mapping from an animal wits its set of attributes to a number that represents its quality, its “fitness”. That quality is on the y-axis, while the degree of the handicapped trait runs on the x-axis from the left where it’s at its ideal natural-selection-wise. We have two curves, representing respectively the “worst” and “best” members of the species as a function of handicap degree, and an additional curve that is simply the averaged quality of these two curves. There are a few problems with this graph.
Presumably the quality measured on the y-axis is an “exclusive quality”, that is, the quality of the animals without including the handicap, which, as Zahavi mentions, is a compromise and which makes the animals worse off in terms of surviving. In this case the lower curve should indeed go up: animals that otherwise have already a not such a great genotype who in addition grow the handicapping trait would simply not survive, leaving the “last place” to animals of higher quality. The top curve of “highest quality”, however, should be flat. Presumably an animal can both have a very fit genotype AND grow the handicap. Unless, that is, we think of it in terms of the phenotype of the creature and only include part of the handicap on the y-axis. For example, while an animal of the highest-quality can grow or not grow a handicapping trait (long tail, large antlers), there’s an energy cost that makes it impossible for it to be as fit as it could be without the handicap. Perhaps one cannot both grow and maintain strong legs and bones, efficient lungs and heart, keen perception and so on AND put the resources into growing the handicap. Notice, however, that the decline of the top curve would affect the slope of the “average quality” curve”. Either way, a flat top curve would still keep the average curve rising as a function of the handicap degree, which is Zahavi’s main point in presenting the graph: females that choose more handicapped males would be going for a higher quality mate on average. However, it’s important to put this in the right context; as Smith briefly mentioned in the paper above, the inheritance of the traits is important, and Zahavi doesn’t speak about it.
That the averaged curve in this graph goes up would be relevant only if (1) the traits responsible for the high quality of the male are heritable, and, in particular, heritable by the daughters, and (2) the daughters do not inherit the handicap. Both of these are very likely, but then we have to add that the incline of the average curve is only relevant as far as the daughters are concerned. Presumably the sons would inherit the handicap as well, meaning that the female is choosing “worse excepted sons”. She might have as well chosen a male with crooked legs and inborn blindness.⁵⁶ The graph we would care about when we consider the sons is with an “inclusive quality” on the y-axis, that is, a measure that captures the total natural-selection-relevance, that is, a measure of the fitness that takes into consideration the handicap. In that case, the curves would look different. The top curve would indeed decline–– the otherwise best animals get hindered by the additional handicap, making it more difficult for them to survive. They lose fitness. The bottom curve, however, would be flat, assuming a “survival fitness-threshold”; if your fitness score is below that threshold you are very likely to not survive into maturity. Animals of low quality who had also grown horns would simply not survive, but animals that are just fit enough to make it with their handicap would be just as good at survival as the animals of lower quality without the horn, and hence the flat curve of “worst quality”. In this case, of course, the “average quality” would decline as a function of handicap degree, in which case the females would get the best mates, on average, by mating with males without the handicap.

At the end, what the handicap-principle suggests is that given female choice, as far as discernible traits go (whether visible, smelled, audiable..), males would tend to evolve them as suboptimally as possible without killing themselves, thus advertising how fit they are otherwise (in non-discernible ways). But we don’t see dimorphism that leads males to have crooked legs, asymmetrical wings, maladaptive mouths, smell that attracts predators and so on. Instead, I claim, these traits that puzzled Darwin and many after him are functional, not dysfunctional. Plumage, as I wrote before, is used to solve the location and recognition problem. I want hereby to address some points from Zahavi’s paper:

It is probably more difficult to discriminate quality among mates with cryptic plumage.

I don’t see it as the case. A plumage is “cryptic” as far as it blends in with the environment, but is essentially not any different from other plumage. A bird of paradise might be cryptic inside a busy neon-light district.

Species which have environmental conditions allowing for prolonged periods of pair formation, or others which cannot put males to a high predation risk probably, choose such a strategy [of cryptic plumage]. This may explain Lein’s (1973) finding that monomorphic species of warblers take longer to pair than sexually dimorphic species with colourful male plumage. This may be the reason why northern duck species, whose short breeding season demands a short period of pair formation, evolved striking male plumage. Where time for breeding is short and where males do not participate as much in parental care, males evolve the colourful plumage as a result of female preference. The male’s colourful plumage then serves as a mark of quality and helps females choose good quality males.

My answer to this, of course, is that the shorter the breeding time, the greater the urgency of the solution of the location problem is. If everyone is cryptic then some might not find a mate within a proscribed amount of time. With female choice, if you are a male, by making yourself colorful you guarantee that you would be found. Zahavi tries here a sleight of hand and lump several things together. He wrote “Where time for breeding is short and where males do not participate as much in parental care”. It is the latter, the lack of paternal care, that gives rise to female choice and therefore to the colorful plumage of males. Had this plumage followed the handicap-principle the “time for breeding” should have no effect whatsoever on the dimorphism.

Zahavi also mentions antlers. While not the first to attest the usefulness of antlers, one 2018 paper by Metz et al⁵⁷ exhibits several points nicely. First, antlers are clearly serving a protective function against predators. As opposed to horns, antlers are temporary structures and are shed and regrown. Among elks, antlered males (as opposed to those who had shed them) were less likely to be attacked by wolves, and less likely to be killed by them. Second, big strong antlers are evidence for their carrier’s fitness, but in a way that is completely opposite to Zahavi’s model. Elks shed their antlers annually between breeding seasons, but not immediately after one –– a further evidence that they are not only as good as female-magnets; they retain them for up to 6 months after their rut. Those who shed their antlers earlier increase their risk of predation, but, on the other hand, since they have more time to regrow their antlers, they would have bigger antlers by the time of the next breeding season. Metz et al found out that early antler shedders were more healthy and fit than those who did so later on. This makes sense: being fit these males have a chance to survive despite being without antlers; presumably the not so fit males, had they shed their antlers early on, would not have survived confrontation with wolves, while the fit ones are able to make up for their lack of weapons with their vigour. It then follows that the bigger antlers of the next mating season are a direct and honest evidence of the fitness of their carriers. But it is so not because the antlers are a handicap as Zahavi suggested, but, on the very contrary, because they are an asset. The bigger antlers are evidence that their carriers got rid of their previous set earlier, attracting a disproportional attention from predators. And, third, elks do use the antlers during male-male competition (for females), and so the securing of females comes about through a functional utilization of the antlers, rather than them standing for a conspicuous handicap. If elks could shoot down rivals with eye-lasers they would have done it and gained all the ladies, even if the latter couldn’t detect their weaponry. In short, Zahavi’s handicap-principle certainly does not apply to elks, and probably not to other antler or horn wearing animals.

---