The evidence Sell et. al. present seems compelling with regards to proposition (i): adults appear to be able to make remarkably accurate estimates of upper-body strength from even degraded cues such as static images of faces. As I noted in Part I, however, the truth of propositions (ii) (that this ability is an adaptation) and (iii) (that upper-body strength determines formidability) are more doubtful. I will assess the evidence for each of these claims, starting with the latter.
Concluding that the truth of (i) implies people can visually estimate fighting ability – the likelihood of an individual prevailing in combat – requires us to assume (iii): that upper-body strength is a good proxy for formidability. Unfortunately, Sell and his colleagues provide only indirect, theoretical, reasons for supposing this is true: namely, the greater sexual dimorphism between upper-body and lower-body strength and the fact that the driving force of certain weapons is largely a function of upper-body strength (p. 576). , These considerations, however, seem far from decisive and while it is certainly plausible that upper-body strength is a very (or the most) important component of fighting ability, rigour clearly requires direct empirical evidence. Other likely components of formidability – speed, stealth, skill, bravery, etc. – are either orthogonal to, or even negatively correlated with, high upper-body strength. There are doubtlessly multiple complex tradeoffs between the different components of fighting ability and thus there are likely multiple local-optima in ‘formidability space’. The point of this argument is that without an empirical determination of the magnitude of the correlation between formidability and strength, Sell et. al.’s conclusion rests on an (admittedly plausible) assumption. More importantly, however, it is at least possible that uncontrolled-for components of formidability may introduce confounds or complications that could influence the correlation between perceived and actual strength in either direction. For example, there may be a semi-independent ability to estimate fighting skill, and, depending on the direction of the correlation between upper-body strength and this skill, it may lead us to under- or overestimate the accuracy of visual assessment of fighting ability. The problems around claim (iii), however, are comparatively minor; the major weakness of Sell et. al.’s paper lies with their claim that the ability to visually estimate formidability evolved by natural selection.
An adaptationist claim like (ii) is significantly more complex than other types of propositions because it entails assertions about the past and about design (Symons, 1992: 140-141). As Richard Burian has explained, when one asserts some trait is an adaptation, “one is claiming not only that the feature was brought about by differential reproduction among alternative forms, but also that the relative advantage of this feature vis-à-vis its alternatives played a significant causal role in its production” (1983: 294). In other words, the assertion that the ability to estimate formidability is an adaptation entails that it evolved over deep time by natural selection, and that the ‘function’ of this psychological trait and its neurological substrate is to detect formidability. To say some feature is an adaptation, then, is a compound claim involving multiple independent propositions, each of which requires substantiation. Sell et. al., it seems to me, fall short of this evidentiary standard, not least because they never mount an explicit defense of (ii), despite the fact that they have the onus and that it is a crucial aspect of their paper. There are, nonetheless, a number of arguments that can be extracted from the paper (or advanced on behalf of the authors). In rough order from least to most persuasive, these are (a) that (i) was previously unknown and that Sell et. al. predicted its existence from evolutionary considerations, (b) that comparative data indicates that such assessments are widespread and perhaps even homologous across taxa, (c) that accurate visual formidability assessment is at a minimum not highly culturally bound, and perhaps universal, and (d) the functional goodness-of-fit between the ‘design problem’ and its ‘solution’.
It seems highly significant, firstly, that Sell and his colleagues predicted the existence of a previously unknown trait – i.e. (i) – from general comparative and evolutionarily psychological considerations. It is important to be careful here, though, because it is entirely possible for (i) to be true but for (ii) to be false (but obviously not vice versa). Some philosophy of science should clarify the situation. Hans Reichenbach (1938) usefully distinguished between the “context of discovery” (the creative process of using background knowledge to invent new hypotheses and theories) and the “context of justification” (the evidence-driven process of testing hypotheses and subjecting them to peer evaluation). For example, Friedrich Kekulé von Stradonitz reportedly first imagined the six-carbon ring structure of benzene after having a dream of an ouroboros (the context of discovery). It does not follow from this, however, that benzene actually had anything to do with snakes, or that testing the idea (the context of justification) involved an ancient symbol. Similarly, even if Sell et. al. predicted the existence of an ability to make formidability estimates from evolutionary theory, it does not necessarily follow that the trait evolved. Propositions (i) and (ii) are logically and epistemically independent, and each needs to be tested against related but different sets of evidence. The fact that (i) was predicted rather than retrodicted from evolution gives as no more than prima facie reason to think it evolved and (a) is thus weak evidence for (ii).
Sell et. al. cite a large and growing body of literature that documents parallels between human and non-human conflict, including, importantly, evidence that non-human animals can visually detect formidability. If this trait is homologous across species, including humans – and that is a gargantuan if – we can be confident (ii) is true since homology suggests that the emergence and persistence of the trait is due to natural selection. It should be clear, however, that building a convincing phylogenetic case for such a widespread homology would be a mammoth undertaking, and no one, as far as I know, has yet done so. The fact that there seems to be a preliminary case for homology is at best suggestive, no conclusions can reasonably be drawn until much more science is done. In other words, were (b) true we could reasonably infer (ii), but we simply do not have enough evidence to conclude (b) is in fact true so, on current evidence, it provides minimal support.
The logic of argument (c) is the following: given that traits that reliably emerge in developmentally normal individuals are likely (though not necessarily) adaptations, demonstrating that the ability to visual estimate formidability is a human universal would strongly favor proposition (ii). As we saw above, Sell et. al. did demonstrate convincingly that the trait is not highly culturally contingent: if US subjects can estimate the upper-body strength of culturally distant men, the ability must be fairly general. The usual criterion for being admitted to the list of human universals (Brown, 1991), however, is having been documented in a large number of (or ideally, all) cultures, which is obviously not true of visual strength assessments. More significantly, the data are compatible with a wide range of scenarios, from near-parochialism to universality. For one thing, only US subjects ever made strength estimates so the evidence is compatible with the hypothesis that the ability is culturally-insensitive (it works on everyone) but that only US subjects have it. Similarly, it is possible US subjects would not be able to make accurate formidability estimates of men from even more distant cultures – the Hadza, say, or Mongolians. While I think neither of these scenarios likely, we do not have the empirical evidence to reject them (or many other hypotheses), so (c) also provides somewhat equivocal support for (ii).
By my reckoning, the authors' most convincing (though entirely implicit) argument for (ii) is the functional or specialized ‘fit’ between the survival problem and the structure evolved to ‘solve’ it. In their classic “The Psychological Foundations of Culture” (pdf) Leda Cosmides and John Tooby set out the general logic of this argument as follows:
The conspicuously distinctive cumulative impacts of chance and selection allowed the development of rigorous standards of evidence for recognizing and establishing the existence of adaptations and distinguishing them from the non-adaptive aspects of organisms... Complex adaptations are usually species typical; moreover, they are so well-organized and such good engineering solutions to adaptive problems that chance coordination between problem and solution is effectively ruled out as a plausible explanation. Adaptations are recognizable by “evidence of special design” (Williams, 1966); that is, by recognizing certain features of the evolved species-typical design of an organism as “components of some special problem-solving machinery” (Williams, 1985: 1) that solve an evolutionarily long standing-problem. Standards for recognizing special design include factors such as economy, efficiency, complexity, precision, and reliability, which, like a key fitting a lock, render the design too good a solution to a defined adaptive problem to be coincidence. (Cosmides & Tooby, 1992: 62 – 63; references other than to quotations removed).The canonical example of such a ‘special problem-solving machine’ is the human visual system (the eye and the attached neural architecture): the system is far too intricate to have come about by chance. Given that natural selection is the only known natural explanation for functional complexity (Dawkins, 1986), invoking the former is our only (non-Skyhook invoking) option when confronted by a sufficiently improbable instance of the latter. The extent to which argument (d) supports proposition (ii), then, rests on whether visual formidability detection exhibits functional complexity not explicable without reference to natural selection. I will argue that, while the evidence is somewhat equivocal and uncertainties remain, on balance (d) warrants tentative acceptance of (ii).
It is important to emphasize at the outset, however, that it is possible for strength assessment to be a by-product of other visual adaptations. It is uncontroversial that the visual system and its various components are adaptations, so only natural selection could explain the very existence of vision. But with (ii) Sell et. al. claim that in addition to the other visual adaptations – edge detectors, facial recognition, retinotopic maps, etc. – there is an information processing mechanism in the brain specifically adapted to analyze the incoming visual data in order to assess formidability. In other words, the human brain contains a ‘device’, coupled to pre-existing visual and cognitive adaptations, to extract and process information from human bodies and faces and it exists just because without it accurate strength and formidability estimates would not be possible (at least not without extensive learning). The latter clause is important. Nobody posits the existence of a “tattoo detection module” since the ability to detect tattoos ‘comes free with’ – is a by-product of – the general visual system. Sell et. al. are saying, in effect, that formidability detection is not like tattoo detection: it does not come free with vision because it requires additional specialized processing. It seems to me that the case for such specialized processing when estimating upper-body strength is rather weak when the body is visible. As Daniel Dennett has pointed out:
since in the case of humans (and only humans) there is always another possible source [for] adaptations… – namely culture – one cannot so readily infer that there has been genetic evolution for the trait in question. Even in the case of nonhuman animals, the inference from adaptation to genetic basis is risky when the adaptation in question is not an anatomical feature but a behavioral pattern which is an obvious Good Trick. For then there is another possible explanation: the general nonstupidity of the species… [T]he more obvious the move, the less secure the inference that it has to have been copied from predecessors – specifically carried by the genes. (1995: 485).Assuming the prior existence of a complex visual system, determining upper-body strength when the body can be seen plausibly only requires the obvious Good Trick of adding up muscle mass. In other words, when the body is visible accurate strength estimation arguably does not require special problem-solving machinery. The case is quite different, however, with regards to strength estimates made when only the face is visible. Recall that Sell et. al. demonstrated raters made strength judgments from face-only photographs that are highly correlated with measured strength: the independent effect of upper-body strength was γ=0.31 (p=10-11) in the US group, γ=0.18 (p=0.0003) for the Tsimane subjects, and γ=0.43 (p=10-5) for the Andean participants. Importantly, there does not seem to be an obvious Good Trick that could account for these findings. To see what I mean, compare the following two photographs:
On the left is Kimbo Slice, a martial arts fighter; on the right is South African rugby player Bakkies Botha. Concluding that Kimbo Slice is a formidable man from the above photograph seems straightforward and rather unimpressive: you can just see how big he is and that he was powerful muscles. Determining that Bakkies Botha is formidable from the above photograph (and he is - he's one of the toughest men in world rugby and South Africa's enforcer), however, would be far more impressive: doing so requires making non-obvious inferences. The most plausible mechanism that links strength and facial morphology is that both traits covary with long term exposure to testosterone (p. 581). A facial structure indicative of a history of high levels of circulating testosterone is therefore ipso facto a reliable marker of upper-body strength. Clearly, the ability to determine which faces show signs of high levels of testosterone is a non-trivial engineering problem, and thus the ability is very unlikely to ‘come free with’ a visual system. Accurate strength estimates from faces therefore strongly implicates specialized and functionally complex cognitive machinery. Given how credible it is that formidability detection is fitness enhancing – by informing decisions on whether to fight or flee, thus avoiding confrontations with more powerful opponents – it is likely that natural selection for such an ability played a ‘significant causal role in its production’. In other words, (d) provides good (though not overwhelming) reason to think (ii) is indeed true: formidability detection is an adaptation that evolved over deep time by natural selection.
Sell, A., Cosmides, L., Tooby, J., Sznycer, D., von Rueden, C., & Gurven, M. (2009). Human adaptations for the visual assessment of strength and fighting ability from the body and face Proceedings of the Royal Society B: Biological Sciences, 276 (1656), 575-584 DOI: 10.1098/rspb.2008.1177