αEP: The fundamental failure of the evolutionary psychology premise

This is another addition to my αEP series about evolutionary psychology. Here's the first, and unfortunately there are several more to come.

I have a real problem with evolutionary psychology, and it goes right to the root of the discipline: it's built on a flawed foundation. It relies on a naïve and simplistic understanding of how evolution works (a basic misconception that reminds me of another now-dead discipline, which I'll write about later) — it appeals to many people, though, because that misconception aligns nicely with the cartoon version of evolution in most people's heads, and it also means that every time you criticize evolutionary psychology, you get a swarm of ignorant defenders who assume you're attacking evolution itself.

That misconception is adaptationism.

In a forlorn attempt to forestall the buzzing mob that will immediately accuse me of creationism and of denying natural selection, that does not mean that I think selection is unimportant or not essential. It does not mean that I think other modes of evolution are more important. It means that there is a large collection of mechanisms that all play a significant role in evolution, and that you can't simply pretend that one is all that matters. Not appreciating the importance of these other mechanisms is a bit like being an electrician who thinks that voltage is all that matters, and resistance and current can be ignored.

In particular, random genetic drift, the variation in a population caused by sampling error, is far more significant than most people (including most evolutionary psychologists) assume. Most of the obvious phenotypic variation we see in people, for instance, is not a product of selection: your nose does not have the shape it does, which differs from my nose, which differs from Barack Obama's nose, which differs from George Takei's nose, because we independently descend from populations which had intensely differing patterns of natural and sexual selection for nose shape; no, what we're seeing are chance variations amplified in frequency by drift in different populations.

Why? Because selection is blind to small differences. Chance dominates, unless the selection coefficient is relatively large.

Let me explain why with an analogy to a gambling casino. If you are at all numerate, you know that casinos are a colossal rip-off, great big shiny vacuum cleaners designed to suck all the money out of their clientele's pockets (if you don't know that, stop now, go read a math and statistics book until you understand what's going on). I'll assume you all understand this, and that you also understand that, like selection, it's a consequence of the power of small advantages amplified by repetition. The house has a small advantage (OK, I'm lying a little bit there: the house advantage is typically between 5 and 20%, depending on the game), but every single gambler, every time they play the game, is facing that probability of kicking in a contribution to the house profits.

And it adds up. Every once in a while someone will get lucky and win, but the more they play, the more likely the odds will catch up with them, and eventually, inevitably, they will be busted, and the house will win. This is a powerful statistical consequence, and most people do understand that this is also how evolution can work: small differences played out for generation after generation can eventually result in extinction (if you're at a disadvantage) or fixation (the difference spreads to be expressed in 100% of the population, if advantageous).

But it's not enough. Imagine, for instance, that you discover a loophole in a casino game, and it's enough to shift the odds in your favor. Let's say you identify a set of circumstances in which a roulette wheel turns up red 51% of the time, giving you a 1% advantage if you place a bet on red when those circumstances are favorable. You have a system, and we're going to pretend that it actually works, unlike most gambling systems.

So, equipped with your favorable impression of the power of selection and your 1% gambling advantage, you run out to your local casino and buy one $1 chip, convinced that you can now just gradually run it up into $10 million, just by playing patiently.

I think you can see the flaw right away. You will lose your $1 investment in the first spin of the wheel 49% of the time. Even if you build up a little pile of chips, you can be wiped out by a short run of bad luck. A selective advantage does not represent an inevitable triumph.

Similarly, a selective disadvantage does not represent an inevitable defeat. People still occasionally, very rarely have a net win after a night of playing roulette, despite the hefty house advantage working against them.

So how can you win with your advantage? Gamblers know this one: you need to go in with a substantial stake. Show up with a million dollars that you dole out in $10 bets all night long, so you have a buffer to deal with inevitable vagaries of chance, and the odds are really good that you'll go home in the morning with $1,010,000. (Well, or the casino managers will notice your peculiar betting habits and toss you out — they're rather zealous about protecting their advantage.)

There's an important lesson there. Because this is a mechanism built on chance, you need both a selective advantage and a sufficient number of trials for that advantage to work. Few trials: chance dominates. Many trials: selection rules.

In a population of individuals, we have a term for that number of trials: it's called the effective population size, or Ne. Remember, in evolution we care about populations, not individuals, so the variable of interest is not how many times an individual spins that roulette wheel, but how many members of the population descend on that device. Before we can assess the effectiveness of selection for a trait with a coefficient of selection s, we also need to know the population size.

One rule of thumb is that for selection to be effective,

|s| >> 1/Ne

What that means is that selection works best in large populations, while chance dominates in small populations — you need a very large s to make the selective advantage rise above the noise generated by chance variation.

What it also means is that in any population there will be a range of variation that is effectively invisible to selection, a range that will be rather narrow in an immense population of bacteria, but will be relatively wide in small populations…like, for instance, a large, slow-breeding population of Pleistocene primates.

Again, that does not mean that selection did not apply to our Paleolithic ancestors. Being born with a heritable heart defect meant you were strongly selected against, and that trait would be gradually eliminated from the population; being born with testes that produced voluminous robust sperm, or having an immune system that made you more resistant to a common virus, would still give your offspring an edge. But keep in mind that even the most wonderfully advantageous allele that arose by a de novo mutation in you has a good chance of being lost by meiotic segregation (1 over 2 to the number of children you have, to be precise), and even if your children do inherit that one trait, there's a significant probability that one of the multitude of other factors that constrain survival and reproduction can work against it.

I repeat: A selective advantage does not represent an inevitable triumph. A selective disadvantage does not represent an inevitable defeat. Chance is an important element of the game of survival.

Here's a specific example: color blindness. Being unable to discriminate differences in a particular range of wavelengths is a disadvantage -- probably a very small one, but it's clear that having trichromatic vision swept to near-fixation in old world monkeys and apes fairly rapidly, and fairly thoroughly…and not having it is a step backwards. So why hasn't natural selection culled it from our populations? This isn't a recent innovation that hasn't had time to be corrected; trichromacy arose sometime after the split between the old and new world monkeys, 30-40 million years ago. It's X-linked in those other primates, too. Shouldn't this defective allele have been long gone from the primates?

No, and there's a simple explanation: color blindness is a defect that's below the threshold for a strong selection pressure to work against it (all you colorblind readers can heave a sigh of relief—Nature isn't going to come gunning for you).

And if color-blindness is invisible to selection, I'm going to be pretty damned skeptical when an evolutionary psychologist tries to tell me that girls' fondness for pink colors is or was a functional adaptation: a product of a 100,000 years of natural selection. It's not impossible that pink preference could confer a benefit, but the idea that a pink preference was so strongly selected for that we can infer that it must have had a selective advantage is so unlikely that it can be dismissed as totally bogus, in the absence of exceptionally strong evidence for such an improbable circumstance. Furthermore, even if a pink preference existed as a heritable trait (which I doubt), the most likely explanation for its presence in a population is drift, not selection.

I've been assured that this is a decent summary of evolutionary psychology by a credible source. If you understood what I wrote above, you'll immediately see the problem.

*** Is evolutionary psychology legitimate?

The foundation of the approach is very hard to disagree with. Tell me which you believe to be incorrect:

1. Organs are complex functional adaptations, results of selection processes

2. The brain is an organ

3. Therefore, we can understand it in terms of the past, just as we do for every part of the body in humans and in all other life forms on earth. (Note that understanding the history of a feature is not the same as saying any observable trait is adaptive or was. Red blood cells are not red because redness was selected for. Hemoglobin is simply a good oxygen transporter, and happens to be red. We still need to understand all this to explain the redness.)

The foundation of the approach is very easy to disagree with if you have any understanding of modern population genetics. It's blown to flinders at the very first premise.

First of all, let's just dismiss that "complex" quantifier. It's irrelevant and often not true; simple functions can also be the product of selection (or drift!). My next post in this series will deal with the whole complexity canard, which is something I've noted in the past as being one of the favorite buzz words of ID creationists…and complexity usually isn't a product of selection, but primarily of chance. They use it to make their arguments look sciencey when they're not. Evolutionary psychologists like to toss it in for the same reason.

But that first statement has to be revised to make it fit reality. How about…

Some features of organs are functional adaptations, the result of selection processes. Others are not.

Now you can see that the first problem the evolutionary psychologists have to confront is whether the feature they are examining is actually a functional adaptation; they can't simply assume that it is, as they often do, and then proceed on their merry way, building hypotheses to explain an assertion that they haven't yet established as true.

Wait, check that. Actually, the first thing they have to do is show that the feature they're examining is a direct product of a genetic variant in the first place, and shows some pattern of inheritance. They often skip this step, too.

Now this is not to say that every single researcher and paper in evolutionary psychology sucks. I've read a few that were decent (and lately people have sent me some others), but I've noticed something interesting: the farther the paper gets away from the "psychology" part, the more it looks at wider variation in populations, the better it is at narrowing the discussion to traits that actually exhibit demonstrable patterns of inheritance, and the farther it moves away from this Pleistocene nonsense, the stronger it is. The best of the work is more about quantitative genetics and comparative ethology; the more it fits under this banner of the Pleistocene hypothesis, and worse, the EEA and this kind of awful human centered crap, the farther I want to throw the paper across the room. Don't even get me started on papers that assign deep evolutionary significance to the results of surveys in Psych 101 classes; those need to be pissed on at length.

Too often what I think of when I see these bad human psychology papers is a comparison to my favorite research animal, the zebrafish. Zebrafish are great as a model system for studying developmental mechanisms, they are abysmally deficient for studying evolutionary processes…but I imagine their evolution could be studied, with a lot of hard work looking at native populations and closely related species. Same with people: they are dismal specimens for studying evolution of behavior, although you might be able to do it with a wide enough and deep enough reach. EP is a hypothesis that tries to short-circuit the requirements for good evolutionary biology research.

The only salvation for the field is to get away from the BS I quoted above.


More like this

Thank you, that was a very interesting post. I'm currently studying for a degree in Neuroscience and one of the topics I'm considering pursuing for my PhD is the parallel evolution of communication structures (and the resulting social structures) in the brains of mammals and birds. I had already come to some similar conclusions about EP, but it is good to have them crystallised into a more defined argument. Hopefully I will avoid making these kinds of mistakes.

I think that these problems go beyond EP though to other fields of evolutionary 'research' (and also in ascribing psychological phenomena to simplified neurological models). There seems to be a broadly felt temptation to try to force every aspect of life into the evolutionary paradigm, especially in pop science and it seems that everywhere I look scientists are trying to describe simple, almost incidental traits of biology according to the principles of selective adaptation. For me the big lesson here is that you can't take short-cuts in science. When you appeal to a broad paradigm like evolution you have to be extremely careful in describing the complete modelling process from start to finish, using every tool that science provides. And, as always rigorous, quantitative methodologies provide the most compelling arguments.

By Matt Hutton (not verified) on 12 Dec 2012 #permalink

Oh boy, can I have the popcorn franchise for this thread?

Sorry, the preceding is not a substantial contribution to the discussion. I'm probably not even able to make a truly substantial contribution. It's just a prediction. I've been eager to see a discussion of this subject. Can't wait for more comments.

By Carl Wasson (not verified) on 12 Dec 2012 #permalink

The argument doen't make a lot of sense. Color blindness is prevailing at a low percentage in the population presuamably like gayness. Weak preference for pink presumably applies to a majority of girls? I think the main problem with the article is that while it is true 'A selective advantage does not represent an inevitable triumph' it is false to say 'A selective disadvantage does not represent an inevitable defeat.' In the long run it would. Darwin regettted not calling natural selection natural elimination. So if a 'bad' trait has been around for a vey long time and not been eliminated it must be assoicatiated with something good otherwise it really wouldn't persist.

By axel jencks (not verified) on 13 Dec 2012 #permalink

Axel, deleterious traits are often found in small portions of the population. Some deleterious traits can be found in a large portion of the population if the trait itself is not horribly debilitating, nearsightedness would be an example, and some deleterious traits may persist simply because it's the best compromise, which you may appreciate if you experience lower-back pain.

Then there's stuff like mental illness: some conditions don't reach full severity until after one is likely to have reproduced, such as dementia or schizophrenia. Other mental illnesses may be deleterious to the individual, but also carry an increased risk of unplanned pregnancies (ADHD and Bipolar Disorder, due to the impulsive behavior and risk-seeking). Evolution doesn't particularly care about whether your offspring are legitimate: bastard children will do just fine. Still, only someone who is completely clueless woukd argue that these conditions are not detrimental to the individual.

Ultimately, background levels of various deleterious, neutral, or otherwise nn-optimal traits are an essential part of evolution. Consider a classic case like the London moths during the Industrial Revolution: it's intuitively obvious how the change in tree color due to soot altered the selective pressure on moths, black moths became easier for predators to spot and white moths could blend in, the reverse of the prior conditions. Shortly thereafter, most moths were white instead of black. What is not immediately intuitively obvious is that for this to happen, there must have been at least some white moths present beforehand, when it was a deleterious trait.

Let's define deleterious traits as those (with a genetic basis) that harm reproductive success relative to other traits that do better. Then in this case, surely they can exist for only 3 reasons:

1) They are new. Natural selection has not had enough time to eliminate (or fixate) them.
2) They are not deleterious under all conditions. Maybe they become beneficial in some climates or at some gene frequencies etc.
3) They are linked to other beneficial traits e.g. pleiotropy.

Note that if such a trait reaches 'fixation' by pure luck it is no longer deleterious under this definition as the alternatives that do better have been removed from the pool.

Now consider variability in nose length as P.Z. does. First let's assume nose length has some bearing on reproductive success.

Now the first thing to say is that presumably nose length is not specified to the nearest millimeter genetically. In other words there is room for variation that is not heritable, part of reason why everyone's nose it not a fixed size.

But let's assume there is a heritable component. By looking at my dad's nose or my genes I can infer my nose is likely to be a least 3 inches long, much larger than average.

Then the reason for this must be one of the 3 above. It is a new mutation (on an evolutionary timescale), it is a beneficial trait in some environments or it is linked to another trait that confers reproductive advantage.

Now of course if the long nose confers only a weak reproductive disadvantage then natural selection needs much more time to do its work. But there is no arbitrary threshold beneath which a deleterious trait is invisible to selection.

By axel jencks (not verified) on 14 Dec 2012 #permalink

Axel, the weak preference for pink applies to cultures where pink is considered a feminine color. This is not true in the Middle East, where it is considered a male color, and in fact pink was not considered a feminine color in the Western world until around the late 1800's/early 1900's. Prior to that, it was considered a childish red (which was definitely the male color).

A couple things:

1. I would say that the premise behind evolutionary psychology is that evolution occurs in the brain as well as outside the brain. The entire concept of discrediting all of EvoPsych because of genetic variability is, in itself, flawed.

2. I think that good EvoPsych eliminates or provides evidence against the idea that a specific mental trait comes from socialization. If certain behavior cannot come from socialization, then it must come from our biology. If it comes from our biology, then it must be due to evolution.

3. Good evopsych also compares certain behaviors traits among various species, and compares whether certain behaviors come from the frontal lobe- which can be trained just like Pavlov's dog, or the cerebellum/Basal Ganglia/inner brain, which is the portion of our brain that is older from an evolutionary standpoint and much more prone to create a link between genetics and behavior.

As an example, let me introduce Roy Baumeister's essay, "Is there anything good about men?" http://www.psy.fsu.edu/~baumeistertice/goodaboutmen.htm

His argument boils down to three components:

1. In virtually all physical and mental traits, men show more variability than women. Intelligence, height, weight, aggression, desire to work hard, etc. Also, men are much more prone to risk-taking than women.

2. In almost all other species, men show more genetic variability and risk-taking behavior than women.

3. Biological evidence shows that we have twice as many female ancestors as male ancestors. i.e. much of our evolutionary history involves single men impregnating many women, instead of simple monogamous behavior.

The conclusion that Roy reaches is that variability and risk-taking behavior is genetically selected for in men, which makes sense since men produce and carry millions of gametes and don't need to take a 9 month break after a successful insemination; and since risk-taking introduces new, hostile environments, variability gives more men the opportunity for the Christopher Columbuses and the Magellans to survive new atmospheres.

Genetic variation and the "natural selection casino" does not negate this hypothesis (I could make the argument that it bolsters the hypothesis), and "socialization theory" would have trouble explaining why mental traits are as variable as physical traits, and why these patterns are found in other species as well.

Another interesting application of evopsych is explaining why men have a refractory period after ejaculation. It is nearly universal, but I don't see many examples of parents teaching their children not to maintain an erection after ejaculation. The evopsych explanation is that knocking semen out of a vagina is not beneficial to reproduction.

Another example is a book published by Thornhill and Palmer about the evolutionary psychology behind rape. I haven't read it yet, but I read an article about it:

The basic premise is that the impulse to rape is not something that is socially taught; insects and other species have those same impulses. Sexual desire is not regulated by the cerebral cortex, so the desire to rape is likely not something that is taught by society. The authors researched how different factors affected the psychological trauma of rape and the likelihood/probability of a rape occurring; this gave them a perspective of the selection pressures that surround rape. The authors' conclusion was that to prevent rape, it would be wise to educate young men about their sexual impulses, and to focus on the criminal penalties of rape in the education rather than focusing on how it hurts a woman.

In conclusion, it is a soft science; you won't see many conclusions with p<0.05. However, it is often the only way to study many areas. Good evolutionary psychology incorporates microbiology, neuroscience, and cross-species examination. In my opinion, great evolutionary psychology would include Monte Carlo methods (similar to how we derived the fact that we have twice as many female ancestors as male ancestors from our knowledge of humanity's youngest common male ancestor and humanity's youngest common female ancestor), but I have yet to see that; evopsych is a relatively new field that could use much more improvement.

Evopsych, once it is refined further, can be used to figure out how closely two traits are linked together, and can ultimately be used as a tool to help match certain mental traits to certain gene sequences. We already know that genes affect our behavior: http://www.pnas.org/content/early/2008/09/02/0803081105. Why not try to gather information from as many sources as possible to determine exactly how evolution has affected our behavior?

I also want to point out that while genetics and microbiology can explain why colorblindness has not been eliminated yet (i.e. dominant vs recessive traits, the fact that reproductive success depends on the whole gamut of a person's traits instead of on just one particular trait, etc.), evolutionary psychology, when done properly, certainly can explain why color blindness is so rare (as long as you're willing to subtract the "psychology" portion of evolutionary psychology").

In fact, I'd be willing to venture a guess that if you can identify the risk factors that decrease a colorblind person's probability of success, and if you can identify areas in the world where those risk factors are higher, then you would be able to predict the relative prevalence of colorblindness around the world. Professor Myers, I think that your article pointed out the *difficulties* that evolutionary psychologists face when doing their work, you did not necessarily point out the *flaw in their premise*.

@axel, your comment is pretty illustrative of the problems Meyers is talking about. If evolutionary psychology can't explain the prevalence of a deleterious trait (like colorblindness), then turn that frown upside down and say it must have some positive value we just haven't figured out yet. It's preserving the theory at the expense of the facts.

Likewise, your comment about "gayness" shows that some evopsych adherents are unwilling or unable to ask themselves how they are classifying traits in the first place. Does "gayness" mean exclusive homosexuality? How does situational homosexuality fall under that? What about men who have sex with men but don't identify as "gay"? You can't nail down a trait to investigate if you don't know what the trait means. Colorblindness is straightforward; behavioral characteristics like "gayness" or "liking" pink (how much like? do your college freshmen really exemplify that preference throughout humanity?) are not.