1. Introduction: Adaptations, Adaptationism, Evidence, and Humans

The primary focus of the current debates surrounding ‘adaptationism’ seems to be resolving into questions about the sorts of evidence necessary for an adaptive (or a non-adaptive, for that matter) hypothesis to be considered well-supported (see e.g. Griffiths Reference Griffiths1996; Brandon and Rausher Reference Brandon and Rausher1996; Rose and Lauder Reference Rose, Lauder, Rose and Lauder1996 and cites there; Pigliucci and Kaplan 2000 and cites therein); what requirements are accepted as legitimate will obviously have serious implications for the success of evolutionary psychology meeting its stated goal of providing strong empirical support for the adaptive hypotheses put forward (see Lloyd Reference Lloyd1999; Lewontin Reference Lewontin1995). There are a number of different kinds of evidence that can be brought to bear on hypotheses surrounding how and why a particular phenotypic trait (or traits) arose, spread within a population, and has been maintained within that population. These include laboratory and ecological field studies of the consequences of phenotypic manipulations of that trait (or those traits), laboratory evolution experiments designed to test adaptive and non-adaptive hypotheses (such as those involving constraints, or the robustness of proposed pathways), optimization analyses of various sorts, regression analyses of the trait and fitness consequences, and analyses of the phylogeny of the populations involved; in general, more than one type of evidence will need to be gathered for an adaptive (or again, a non-adaptive) hypotheses to be adequately tested against reasonable competitors (see Pigliucci and Kaplan Reference Pigliucci and Kaplan2000).

In this paper, I explore the extent to which the various techniques used to test adaptive hypotheses in non-human populations can be and have been successfully applied to testing adaptive hypotheses in the context of human populations. While evolutionary psychology's main proponents’ focus on complex psychological traits as well as on supposedly universal human adaptations will be revealed to be deeply problematic from the standpoint of putting forward testable hypotheses, it will be suggested that other approaches less focused on these arenas might prove to be promising.

2. Testing Adaptive and Non-Adaptive Hypotheses in Humans: The Impossible and the Difficult

Unfortunately, many of the techniques commonly used to test adaptive hypotheses in other organisms are of little use in testing purported human adaptations. Obviously, experimental manipulations designed to reveal fitness consequences of modifications of putatively adaptive traits, both in laboratory and in natural populations (see Sinervo and Basolo Reference Sinervo, Basolo, Rose and Lauder1996 and cites therein) fall into this category because using these techniques on humans is in general ethically impossible. Partial exceptions are provided by ‘natural’ experiments, in which phenotypic traits are ‘manipulated’ accidentally (either because of e.g. traumatic injury or genetic mutations), but in these cases it is very hard to control for related variables. In the case of hypothesized human psychological adaptations, the manipulations necessary to properly test adaptive hypotheses by phenotypic manipulation might well be beyond the current state-of-the-art in e.g. neuropsychology, even if they were not ethically impossible. Similarly, laboratory evolution experiments designed to, for example, falsify alternate hypotheses (such as those involving constraints and allometry, see Schlichting and Pigliucci Reference Schlichting and Pigliucci1998) or to test the repeatability of adaptive pathways (see Rose, Nusbaum, and Chippindale Reference Rose, Nusbaum, Chippindale, Rose and Lauder1996, and cites therein) can be rejected as all-but impossible in the human case, for both ethical and practical reasons.

‘Transplant’ experiments, a variation of experimental manipulations wherein supposed adaptations to local conditions (e.g. hypotheses about the significance of particular ecotypes) are tested by physically moving the organisms in question to other locales and observing the fitness consequences (see Reznick and Travis Reference Reznick, Travis, Rose and Lauder1996 and cites therein; see Wilson Reference Wilson1994 and cites therein), can only be ‘done’ with humans in a very limited way (e.g. following people or peoples that move), and generally with a self-selecting and likely atypical sample. Again, properly controlling for correlated variables is difficult in these cases. So far, attempts to use these techniques to study human adaptations have been rare.

Optimization arguments that take historical contingencies into account (see for example Sober Reference Sober1996; Orzack and Sober Reference Orzack and Sober1996; Rose and Lauder Reference Rose, Lauder, Rose and Lauder1996; Seger and Stubblefield Reference Seger, Stubblefield, Rose and Lauder1996, and cites therein) seem a more promising avenue for research into human adaptations, as they primarily involve observation rather than manipulation.Footnote ¹ Similarly, techniques making use of broadly historical evidence about the genesis and spread of the trait in question should be one of the more plausible approaches to testing adaptive hypotheses in humans. Indeed, this latter kind of evidence, has been explicitly mentioned as relevant by prominent evolutionary psychologists; Miller, for example, has noted that “examining the distribution of traits across related species with known phylogenies” can be used to “discern when and where evolutionary innovations occurred” (Reference Miller, Crawford and Krebs1998, 117; see also Buss Reference Buss1994, 54–64).

While these techniques are not impossible (or nearly so) in the way that, for example, most phenotypic manipulations of humans would be, there are still important questions about how successful one should expect such techniques to be at developing evidence that can be used to support or contradict adaptive hypotheses in humans. Below I argue that these techniques have been applied with at least moderate success, and adaptive hypotheses have been well-supported, in the case of human adaptations similar to ecotypes. However, despite the claims of some of evolutionary psychology's proponents, I argue here that such evidence is rarely available in the case of purported ‘universal’ human psychological adaptations. The very limited information available on the environments in which key aspects of human evolution took placeFootnote ² make optimization techniques difficult to apply here. Further, while in some cases phylogenetic information about Hominidae may provide evidence relevant to adaptive hypotheses in humans, nature and history have ‘conspired’ to make the task much more difficult with humans than it is in many other species. These difficulties, combined with the above-outlined difficulties with hypotheses in humans, continue to doom many of the hypotheses of interest to evolutionary psychology's most prominent proponents to having to make do with inadequate evidence.

3. Human Adaptations: Discovery and Evidence

In the case of (some) physical adaptations in humans, there is in general agreement regarding, at least, what needs to be explained, and how well supported current explanations are. For example, while human bipedalism seems to most researchers to be a fact in need of explanation, there seems general agreement that no current hypothesis (adaptive or otherwise) is so well supported by the evidence as to exclude all reasonable competitors (see Tattersall Reference Tattersall1995 for discussion). The massive increase in brain size during human evolution is in a similar position—there is agreement that it stands in need of explanation, but no agreement on what that explanation is. While many possible explanations have been proposed, none are generally considered well supported, nor is there agreement on a plausible research program for solving this problem.

Famously, the high prevalence of alleles associated with sickle-cell anemia (the HbS alleles) in certain populations is a fact in need of explanation; unlike the other cases, however, there is agreement on what the correct explanation is (at least in outline). The sickle-cell allele is an adaptation for partial malaria resistance; the alleles associated with sickle-cell anemia in homozygotes are associated with resistance to malaria in heterozygotes. Evidence for this adaptive hypothesis comes from many sources; the alleles associated with sickle-cell anemia are relatively prevalent only in populations recently exposed to environments where malaria is a problem (in addition, work done on the history of populations vis-a-vis migrations, prevalence of HbS allele, and the prevalence of malaria show a clear relationship); finally, reasonable estimates of the fitness of the HbS versus ‘normal’ alleles in various environments correlate fairly well with actual distributions (see Griffiths et al. Reference Griffiths, Miller, Suzuki, Lewontin and Gelbart1996; Das Reference Das1995, and cites therein). There are still any number of questions to be answered in the sickle-cell case, and some fascinating work being done on the extreme heterogeneity of the clinical manifestation of sickle-cell disease,Footnote ³ but the basic adaptive claim seems reasonably well supported. While there has been some research suggesting that other diseases caused by single recessive genes (such as CF) may be the result of alleles with similar adaptive roles (in the case of CF, possibly resistance to typhoid; see Pier et al. Reference Pier1998), as yet none have gathered evidence strong enough to have been generally accepted. Similarly, a promising research avenue involves attempts to explain particular genetic variations through the testing of particular adaptive hypotheses more generally, such as the work being done attempting to link the prevalence of the apparently recent spread of the CCR5?32 ‘HIV resistance allele’ in populations of European decent to outbreaks of Black Death in Europe (see Stephens et al. Reference Stephens1998).

Other phenotypic traits in humans for which adaptive explanations have been offered and seem reasonably well-supported include skin color (fair skin for diets poor in vitamin D in locales with low levels of sun-light), and some variations in body size and shape between populations in radically different climates (see for example Cavalli-Sforza and Cavalli-Sforza 1995; Lewontin Reference Lewontin1995). There are undoubtedly others. What is startling, though, is that even basic phenotypic features (such as our large brains, bipedalism, opposable thumbs, relative-hairlessness, etc.) that are generally agreed to have arisen after the linage that gave rise to humans diverged from those linages that gave rise to the great apes have in general not been explained by hypotheses (adaptive or otherwise) that have gained wide acceptance (see Tattersall Reference Tattersall1995 and Reference Tattersall2000 for a review). This is certainly not a feature of adaptive explanations shared with other species; for many species with species-typical phenotypic traits there are adequate adaptive explanations for these traits. Why then have acceptable explanations even for such banal human traits as our bipedal stance been so hard to come by in the human case?

4. Phylogenetic Histories and Bad Luck

The basic difficulty with testing adaptive hypotheses for widely-shared human traits stems from an unfortunate feature of our phylogenetic history. Our closest living relatives are the great apes; however, as these things go, they aren't really all that close. It is widely agreed that the most recent common ancestor we share with the great apes was at the very least 6 m.y.a., and possibly rather longer (see Tattersall Reference Tattersall1995, 218; Goodman et al. Reference Goodman1998). Nor do those apes themselves represent a particularly diverse range of species; only the two extant species of chimpanzee share a relatively recent common ancestor with another extant species (Tattersall Reference Tattersall1995, 218; Goodman et al. Reference Goodman1998); although some have suggested that DNA evidence implies that orangutans may be best thought of as representing two distinct species, having diverged perhaps 1.7 m.y.a. (see Gragneux et al. 1999). There are, then, perhaps 7 or so species in the ‘Hominini’ tribe, that (perhaps) share a common ancestor on the order of 10–14 m.y.a. or so, and perhaps another few species of ‘lesser apes’ in the Hominidae family, sharing a common ancestor with the great apes on the order of 18 m.y.a. or so (see Goodman et al. Reference Goodman1998; Gragneux et al. 1999).

The problem with such a sparsely populated clade is that the basic techniques that permit one to figure out how putative adaptations relate to selective regimes are difficult if not impossible to assess without reasonably large numbers of reasonably close relatives. To get good evidence that a trait was subject to strong selective forces, and figure out what those forces were (and hence what the trait might be an adaptation for), requires comparing phylogenetic histories of the populations and traits in question to ecological field studies (or work through experimental manipulations) of the fitness consequences of the traits in questions (see for example Griffiths Reference Griffiths1996; Larson and Losos Reference Larson, Losos, Rose and Lauder1996; Sinervo and Basolo Reference Sinervo, Basolo, Rose and Lauder1996; Leroi et al. Reference Leroi, Rose and Lauder1994). These techniques are only plausible if there are available variations on the traits to test against the known histories of the lineages.

This evidence is not at all impossible to come by in other species, even for ‘species typical’ adaptive traits. For example, Sinervo and Basolo (Reference Sinervo, Basolo, Rose and Lauder1996) describe the research that was done to test the hypothesis that the ‘sword’ of the sword-tailed fish (Xiphophorus) is an adaptation to female preference for male fish with swords. Phenotypic manipulations, mainly the addition or subtraction of swords, were used both to measure the strength of the female preference for swords (and hence the fitness consequences of being ‘sword-ed’) in current populations, and, in combination with phylogenetic information about Xiphorphorus and related genuses (such as the closely related Pirapella), to try to unravel the history of the trait and preference. The studies suggest that the female preference for swords arose prior to the swords, that is, that the primitive trait is a female preference for swords but sword-less males. This, they suggest, supports the hypothesis that the swords are an adaptation to female preference (Sinervo and Basolo Reference Sinervo, Basolo, Rose and Lauder1996, 173–175). These tests were possible because a reasonable phylogeny exists for fish of these types; the genus Xiphophorus itself comprises several dozen species, the males of which have sword-tails (of varying lengths), and the females all have a preference for swords. The females of some closely related genera (such as Pirapella) share the preference for swords, even where none of the males of their species are sworded; since evidence suggests that sword-less-ness is the primitive trait, the adaptive hypothesis is well-supported (see Sinervo and Basolo Reference Sinervo, Basolo, Rose and Lauder1996, 172). While Sinervo and Basolo point out that many questions still remain to be answered, this sort of research into putative adaptations certainly addresses many of the problems with ‘adaptationism’ that, for example, Gould and Lewontin pointed towards (Gould and Lewontin Reference Gould and Lewontin1979).

The problem is that this sort of evidence just isn't available in the case of putative universal human adaptive traits; we have no extant relatives which are suspected of sharing similar selective regimes and that can therefore be used to test the fitness consequences of the supposed adaptations. If all or most of the estimated dozen or so extinct hominid species (comprising, perhaps two or three genera) still existed, phylogenetic studies would certainly be easier, and might well be useful for distinguishing between competing hypotheses about the spread and maintenance of phenotypic traits of interest. Unfortunately for testing adaptive hypotheses in humans, all the other hominids are extinct, and so comparisons between the groups, with special attention to the fitness consequences of differences in key traits, are impossible.Footnote ⁴

The best place to look for testable human adaptations, then, is either for traits widely shared in the hominoid (or larger) clade (hence, not uniquely human traits) or for traits that are of local adaptive significance (such as the HbS mutation or the CCR5Δ32 mutation). However, there is very little known systematic diversity within the genus Homo, either at the level of meaningful phenotypes or at the genetic level. Indeed, current DNA evidence points to humans having far less genetic diversity even than the other great apes, despite the relatively old age of our clade (Gragneux et al. 1999, 5081).Footnote ⁵ This may mean that there will be relatively few significant local variations to be found; however, more work needs to be done looking for such variation before any such claims could be supported.

5. Giving Up on Universal and Unique Features: What's Left for Evolutionary Psychology?

Insofar as it is the goal of many of the most vocal proponents of evolutionary psychology to explain ‘human nature’ as a collection of adaptive traits unique to humans (see Cosmides, Tooby, and Barkow Reference Cosmides, Tooby, Barkow, Cosmides and Tooby1992, 5), the unfortunate structure of our clade, combined with the ethical constraints on human research, seems very problematic.Footnote ⁶ Evidence relating to other related goals, however, is more plausibly obtainable.

Wilson, for example, has noted that evolutionary psychology has not dealt seriously with the possibility of studying adaptive phenotypic variation between human populations; he argues that there are both theoretical and empirical reasons to suspect that such variation might exist (Wilson Reference Wilson1994). And some of the critical difficulties in gathering evidence relevant to claims regarding universal uniquely human adaptions don't apply to gathering evidence relevant to more local, population-level adaptations. For example, when population-level psychological adaptations are suspected, adoption studies (if properly executed) might provide ‘natural’ transplant experiments. It is possible that studies of immigrants to new locales could also shed some light on hypothesized population-level adaptations. And regression analyses done on populations where the selective pressures are known (or suspected) could provide quantitative assessments of the strength of particular selective pressures and perhaps even ‘catch’ adaptation in action (see Lande and Arnold Reference Lande and Arnold1983; see also Schlichting and Pigliucci Reference Schlichting and Pigliucci1998, 166ff. and cites therein for more recent developments). Of course, gathering such evidence would likely be difficult. Not only would it require cross-cultural field work, but it would also require careful work on adoption and immigration practices in order to attempt to deal with the large number of confounding (cultural, etc.) factors.Footnote ⁷

Another approach is to deal not with hypothesized adaptive traits that are uniquely human, but rather with adaptive traits shared by at least some of our close relatives. For example, Sterelny and Griffiths point out that it seems like it should be possible to do good work on the origin of those emotions widely shared within the Hominidae linage (Reference Sterelny and Griffiths1999, chapter 14). While little of that work has been done, hypotheses about the origins and adaptive significance of some aspects of our emotions might be tested by comparisons to related species which share problems (and solutions?) of varying similarity. Similarly, the work done testing theories regarding sperm competition show at least a good (albeit still very problematic) start (see Futuyma Reference Futuyma1998, 359, 588–589 for an introduction to some of the higher quality research that has been done on this subject).Footnote ⁸ While shared psychological traits will likely be less complex than uniquely human psychological traits, testing their adaptive significance may be possible.

Evolutionary psychology has not yet developed the tools necessary to uncover our “shared human nature” (if such there is—see Dupré Reference Dupré1998) any more than physical anthropology has been able to uncover the specifics even of such clear human adaptations as our bipedalism. It is obvious that our brains were subject to selective pressures during our evolutionary history; it is not at all obvious what those pressures were. Rather than over-reaching by attempting to uncover the historical causes of psychological features shared by all and only humans, evolutionary psychology might be better off attempting to develop ways of adequately testing possible adaptive traits of a less ambitious nature, either because they are widely shared outside our species, or shared only by specific ‘eco-types’ within our species.