rationalskepticism.org

Coyne, 2000 wrote:The latest deadweight dragging us (evolutionary biology) closer to phrenology is evolutionary psychology, or the science formerly known as sociobiology. If evolutionary biology is a soft science, then evolutionary psychology is its flabby underbelly.

Given the somewhat controversial title of this essay, it is perhaps necessary for me to preface it with a few disclaimers. Firstly, I am not a creationist and, for all intents and purposes, evolution is TrueTM. Secondly, whenever somebody voices their scepticism over the veracity of evolutionary psychology, they are often met with the retort, “Do you not believe that the brain is a product of evolution?” with the implication that since behaviors are the product of the brain, and the brain is a product of evolution, then behaviors are the product of evolution. This logic, however, is flawed for reasons I will discuss later but I do accept that the brain is an evolved organ with implications for resulting behaviors. And thirdly, this is not a broad scale attack on evolutionary psychology – instead, my focus is on the particular approach to evolutionary psychology known as the “Santa Barbara church of psychology” (Laland and Brown, 2002).

To distinguish between the two approaches, I will follow the nomenclature used by Gray, Heaney and Fairhall (2003) where they refer to this approach as Evolutionary Psychology (EP). This approach (used by popular authors like Steven Pinker in his “How the Mind Works”) attempts to explain a wide range of human behaviors, like whether we have an evolutionary preference for green lawns, with an emphasis on the concept of a modular mind, and utilises a cartoonish view of the Pleistocene – with all considered, we have to wonder whether it should be rebranded as the “Hanna-Barbera church of psychology”.



The standard tool in this area is the explanatory strategy called “reverse engineering” (Pinker, 1997). While ‘normal’ engineering attempts to design solutions to problems, Evolutionary Psychologists argue that current features of the human mind can be explained as solutions to problems presented in our Environment of Evolutionary Adaptedness (Tooby and Cosmides, 1992). For this to be a valid explanatory strategy, Gray et al. (2003) argue that three criteria must be met:

1. all traits are adaptations
2. the traits to be given an adaptive explanation can be easily characterized
3. plausible adaptive explanations are difficult to come by.

As we would expect, these assumptions are frequently violated. Given what we now know about evolutionary processes, the first is perhaps the easiest to refute as it is obvious that not all traits come about as a result of natural selection. Ignoring the more complicated issues of genetic drift, pleiotropy, and epistasis, a perfect example of why we should be sceptical of this claim is Gould and Lewontin’s (1979) concept of the spandrel which, in simple terms, is a byproduct of the selection of another trait. In this sense, asking for the adaptive explanation for some behaviors is akin to asking what the selection pressure was that caused blood to be red.

Gray et al. (2003) discuss the latter two issues in more detail, but essentially the second claim is problematic given the lack of discrete boundaries for certain traits and they use Lewontin’s (1978) example of the “chin” to demonstrate this. They also extensively dissect the third criterion but a successful rebuttal of this is perhaps exemplified by Rosen’s (1982) suggestion that the only two constraints on adaptive explanations are the inventiveness of the author and the gullibility of the audience. It is important to note that I am not suggesting that we should abandon attempts to describe behaviors using adaptive explanations, nor am I saying that all behaviors are spandrels or the result of obscure evolutionary processes, but rather I am highlighting the fact that a plausible story is not evidence in itself. This position is described by Williams (1966) thus:

The ground rule - or perhaps doctrine would be a better term - is that adaptation is a special and onerous concept that should be used only where it is really necessary.

As is clear to most evolutionary biologists, and other interested sceptical parties who are less than enamoured by the efforts of Evolutionary Psychologists, the approach described by Williams above is rarely followed and instead these scientists appear to fire off adaptive explanations with reckless abandon, with their work often consisting of nothing more than folk wisdom with a post hoc just-so story explanation. To attempt to circumvent this, Gray et al. (2003) propose two “common sense” tests: The Grandparent Test, and the Lesser-Spotted Brown Gerbil Test. The first asks us to consider, “Does this work give us any insight into human behavior and cognition beyond popular knowledge?” and the latter asks, “Would this research be publishable in major international journals if the species was a small noncharismatic mammal rather than our own?”. Although these ‘tests’ are only guides and should not be used as definitive tools for ruling out instances of research, it is interesting to note that most examples of EP found in journals fail these basic tests. However, picking examples of this kind to discuss here would be like shooting fish in a barrel, so instead I will look at research behind the evolutionary explanations for cheater detection.



Cosmides (1989) proposed the “social exchange algorithm” which argues that for cooperation to be maintained in a society, we must be able to detect cheaters – with this consistent selection pressure present, humans must have evolved a cognitive mechanism to do this. This idea began with the research using the Wason Card Selection Task which utilises a generalised if P, then Q rule.



The task is straightforward: Given the cards presented in the image above, which cards should you turn over to test the claim that if a card shows an even number on one face, then its reverse side will be red? The correct solution is that you should turn over the “8” and the “brown” card. The other cards are not logically related to the proposition.

Despite the apparent simplicity of this task, Wason (1966) found that only 10% of his subjects answered this correctly. However, the interesting twist on this logical conundrum is that when the context of the problem is framed in a way that is socially relevant, people tend to perform far better. That is, accuracy increases if we change the proposition to: “if you are drinking alcohol then you must be over 18”, and we changed the cards above so that they read “17”, “Beer”, “22” and “Coke”, where the correct cards to turn over are “17” and “beer” (Griggs and Cox, 1982). This means that when we replace the abstract logical notions with a real world example of the same relationship, but with the inclusion of a possible "cheater" (i.e. 17 year olds drinking beer) then we can successfully solve the task by hunting out the cheater. From this, Cosmides predicts that cheater detection is an evolved trait that should only be evoked in social exchange situations, where there is a requirement, benefit and cheater (in accordance with the assumptions of game theory). So here we have a theory that provides novel insight into human cognition and surpasses our folk wisdom, and clearly passes the Grandparent and Lesser-Spotted Brown Gerbil tests. Can we then be confident in our knowledge that this is an evolutionary adaptation? Unfortunately, not yet.

The alternative explanation suggested by Sperber, Cara and Girotto (1995) is that specific properties of the cheater detection scenario employ a more general “exception-testing” rule – this would account for the results in the cheater detection scenarios but it would not support an evolved mechanism that was responsible for cheater detection. If these “properties” could be identified and removed from the cheater detection task, and this resulted in the effect disappearing, then the empirical support for Cosmides’ theory would also disappear. To test this they developed a three-part recipe to ensure correct card selection:

1) the P-and-not-Q case is easier to mentally represent than the P-and-Q case (underage drinkers versus legal-age drinkers);
2) the P-and-not-Q case should be of more importance than P-and-Q case (breakers of the law versus followers of the law)
3) the rule should be clear and unidirectional (there is no implication that legal-age drinkers should be drinking beer).

When looking at Cosmides’ (1989) culture-specific form of the test (where the rule was “If a man eats cassava root, then he must have a tattoo on his face” and the options “Eats cassava root”, “No tattoo”, “Eats molo nuts”, and “Tattoo” were presented - with the "cheater" being the non-tattooed man eating cassava root), Liberman and Klar (1996) noted some inconsistencies between the cheater and non-cheater scenarios. Firstly, in the noncheating scenario there is no specific violating rule (e.g. a man with no tattoo eating a cassava root), secondly, the rule for cheating is strict and exclusive, whereas the non-cheating scenario has reduced importance through the use of qualifiers such as “usually” and “primarily”, and thirdly the non-cheating rule is more easily interpreted as being bidirectional.

To eliminate these confounds, Liberman and Klar reversed the conditions whilst maintaining the basic cheater detection structure and found that detection of non-cheating was at 70%, whereas cheater detection was at 30% - a perfect reversal of the results found in typical cheater detection scenarios. In other words, even though there were still cheaters in the design (non-tattooed men eating cassava root), by removing the biases from the setup so that "cheater detection" was no longer the less difficult task, subjects were less likely to search for them. With the effect completely disappearing under these conditions, it becomes clear that the effect is not a result of social relevance like Cosmides suggested, but is instead simply an effect produced by experimental confounds. In other words, cheater detection is a result of the “saliency” of the cheater in these experiments, and it is this saliency that gives people the correct result, and not the presence of a “cheater”.



This failure to properly adjust variables in an experiment (thus reliably establishing causality) seems to be regular feature of Evolutionary Psychology research, even when they meet the common sense tests suggested above, so why do these factoids spread so quickly and become cemented in popular thought? Do we have an evolved ability to be gullible of EP claims? Most people would reject such an idea, so what is it that separates these “ridiculous” claims from the ridiculous claims made by EP proponents?

It could be the persuasive logic and rhetoric that are often employed as support for their theories, in particular are the two claims that these behavioural traits are; 1) independent of global processes, automatic and often not part of conscious thought, and 2) universal across cultures. On the surface these two arguments appear to give us good reason to believe that a behavior is a result of evolutionary processes, as both points imply that it is instinctual or innate, and even though organisms have the ability to adapt over their lifetime, there is the hidden assumption that such traits are too “complex” to have been learnt. However, these arguments do nothing to support their claims.

The first argument is countered by Gray et al. (2003) with the example of riding a bike; it is clearly a specific process that functions independently from global processes and generally we do not need to consciously operate our bodies in order to successfully ride a bike. As a demonstration of this, ask yourself what you would do if your bike started to tip to one side. Most people reply that they would lean to the opposite side to right themselves but this is incorrect as it would result in the person falling off their bike – instead, when this happens the rider will turn the handlebars which rights their centre of gravity. This meets the requirements of criterion (1), but surely nobody would think that riding a bike is an evolved trait. Part of the reason why we can easily reject such a claim is that the learning period is obvious, but when this learning phase is more subtle (like with language) we are sometimes fooled into reaching the wrong, or premature, conclusions.

Now we need to consider the second argument – that if something is universal across cultures, then it cannot possibly be learnt. Is this true? Of course not. Whilst it is necessarily true that an evolutionary behavior would be universal across cultures, it is not true that a universal behavior is an evolutionary behavior. This is because species-specific behaviors can either be a result of an innate trait, or the result of shared species-specific patterns of experience. In other words, if the environmental factor that produces a particular learning experience is present across all individuals of the species, then we would expect them all to learn the same behavior. Again I turn to Gray et al. (2003) for an example, where they point out that all humans, no matter what culture you look at, will eat soup from a bowl and not a plate. The common environmental variable here is gravity and it gives us a universal behavior – however, the literature on the evolved “eating soup from a bowl” behavior is relatively scarce in the EP journals.



At this point it might not be entirely clear what the popular misconception is and, arguably, I should have outlined this in the introduction but some background on the topic was necessary. The popular misconception is that we have any significant understanding of evolved behaviors in humans. This belief is pushed out year after year in books by Pinker, Buss, Tooby and others, and it has now become more of an exercise in politics rather than attracting interest in science and rational thinking. Consistently these EP journals print articles discussing how women prefer the colour pink because it reminds them of red berries from the hunter-gatherer times of our ancestors (e.g. Hurlbert and Ling, 2007), ignoring the fact that the preference for pink in women is an extremely recent trend from the last few centuries (traditionally baby boys were dressed in pink and girls in blue), and ignoring the fact that hunter-gatherer roles were not separated by sex; or articles about how men are attracted to red lipstick because they look like vaginas (e.g. Elliot and Niesta, 2008). Even the more credible claims like cheater detection, or men being attracted to women with low weight-to-hip ratios (Singh, 1993), are plagued by poorly thought out methodological designs and an over-eagerness to ignore the relevant literature on possible learning mechanisms that could account for the data – so much so that they earn themselves the reputation of being ‘behavioral creationists’.

I started this essay with the disclaimer that this is not a broad attack on evolutionary psychology, and it is not a denial of the fact that the brain is an evolved organ, and I want to reiterate those points. The intention of this essay is to highlight the flaws and inconsistencies in the field, not to convince people to reject it wholesale, but instead to increase the scepticism surrounding this field. If a claim is made to the effect of “We evolved to do X/ prefer Y/ etc” then the question we should ask is “what research experimentally separated the learnt effects from evolved processes?”. The misconception is not that behaviors can, or have, developed in organisms as the result of evolutionary processes, but rather the belief that we can prematurely accept these conclusions based on faulty logic and an overreliance on (and misapplication of) evolutionary principles.

REFERENCES

Cosmides, L. (1989). The logic of social exchange: Has natural selection shaped how humans reason? Studies with the Wason selection task. Cognition, 31, 187-276.

Coyne, J.A. (2000). The fairy tales of evolutionary psychology: Of vice and men. The New Republic, 3 April, pp. 27-34.

Elliot, A. J., & Niesta, D. (2008). Romantic red: Red enhances men's attraction to women. Journal of Personality and Social Psychology, 95, 1150-1164.

Gould, S.J., & Lewontin, R.C. (1979). The spandrels of San Marco and the Panglossian paradigm: a critique of the adaptationist progam. Proceedings of the Royal Society of London B, 205, 581-598.

Gray, R. D., Heaney, M., & Fairhall, S. (2003). Evolutionary psychology and the challenge of adaptive explanation. In J. Fitness & K. Sterelny (Eds.), From Mating to Mentality (pp. 247-268).

Griggs, R. & Cox, R. (1982). The elusive thematic material effect in Wason’s selection task. British Journal of Psychology, 73, 407-420.

Hurlbert, A. & Ling, Y. (2007). Biological components of sex differences in color preference. Current Biology, 17, 623-625.

Laland, K.N., & Brown, G.R. (2002). Sense and nonsense: Evolutionary perspectives on human behavior. Oxford, UK: Oxford University Press.

Lewontin, R.C. (1978). Adaptation. Scientific American, 293, 212-228.

Liberman, N., & Klar, Y. (1996). Hypothesis testing in Wason’s selection task: social exchange cheating detection or task understanding. Cognition, 58, 127-156.

Pinker, S. (1997). How the mind works. London: Allen Lane.

Rosen, D.E. (1982). Teleostean interrelationships, morphological function and evolutionary inference. American Zoologist, 22, 261-273.

Singh, D. (1993). Adaptive significance of female physical attractiveness: Role of waist-to-hip ratio. Journal of Personality and Social Psychology, 65(2), 293-307.

Sperber, D., Cara, F., & Girotto, V. (1995). Relevance theory explains the selection task. Cognition, 57, 31-95.

Tooby, J., & Cosmides, L. (Eds.). (1992). The psychological foundations of culture. In, J.H. Barkow, L., Cosmides, & J. Tooby, (Eds.), The Adapted Mind: Evolutionary Psychology and the Generation of Culture, (pp. 19-136). Oxford: Oxford University Press.

Wason, P. (1966). Reasoning. In, B.M. Foss (Ed.), New horizons in psychology. Harmondsworth, UK: Penguin.

Williams, G.C. (1966). Adaptation and Natural Selection. Princeton University Press, Princeton, N.J.