Robert Kurzban

The Evolutionary Psychology Blog

By Robert Kurzban

Robert Kurzban is an Associate Professor at the University of Pennsylvania and author of Why Everyone (Else) Is A Hypocrite. Follow him on Twitter: @rkurzban

Boobies, Blue-footed And Otherwise

Published 5 October, 2011

Watch out, egg!

Suppose someone wrote a paper, which was subsequently covered in the popular press, that proposed a function for some relatively rare but consistently observed behavioral trait in some species. Just because I like blue-footed boobies so much, let’s pick them, and because of its moral overtones, let’s pick something nasty these birds do, siblicide.

So, suppose someone proposed that killing one’s siblings — siblicide — was an adaptation, designed to solve the problem of gathering parental investment at the expense of one’s sibling (a la Trivers), but only when resources were scarce. (By the way, nothing turns on whether this is right or not, and I’ve actually taken liberties here. I’m only interested in how one might react to this line of argument.) That is, the idea is that there is an adaptation among blue footed boobies which is facultative, and that their booby brains are designed in such a way that they try to kill siblings under some conditions but not others. (See this little page aimed at undergrads; or this paper.)

Here is one way some blogger might react to this (bold is my emphasis):

In order for siblicide to have a selective advantage there has to be an important genetic component. Let’s imagine over evolutionary time there were two groups of boobies who differed in their siblicide alleles. One group killed siblings when resources were scarce. The other didn’t. Somehow the siblicidal group managed to have more offspring than the kind group so the allele for not killing kin was eliminated from the population.

Notice the writer has made a mistake. (And I’d like to hold aside, for the moment, the writer’s move here, which implies that the authors think there is one, single, “siblicide gene,” as opposed to a number of genes that have causal effects on the development of the trait.) What gets eliminated from the population is alleles that cause, relative to alternative alleles, not killing kin under particular circumstances. This is a very basic mistake, thinking that the claim of adaptation — surrounding a hypothesized function of a pattern of behavior when it occurs — is a claim that the trait will be seen in every instance of every organism. Many adaptations are, of course, facultative, responding to the environment.

Now suppose our fictitious author continues:

If the scenario is correct then most boobies have to carry the siblicidal allele since it was selected in the past. This seems very strange since most boobies don’t kill their siblings. (The author would then insert a sarcastic footnote, but I’ll just omit that.)

Notice the mistake here. The claim in the original paper is that the function of siblicide has to do with the issue of parental investment, and this adaptation functions by the birds in question being siblicidal under the right circumsatnaces. That is, the claim in the paper is not that all birds will always kill all siblings. (Though this might be the case for the Masked Boobie (pictured above), which might be obligately, rather than facultatively, siblicidal.) The point is that no serious biologist would take the fact that a particular behavior which is posited to have an adaptive function isn’t universal to be evidence against a claim surrounding a facultative adaptation.

Now suppose our hypothetical author said that because this trait, predicted to occur only in certain circumstances was not, in fact, seen all the time, then,well, there is something “seriously wrong” with the field from which the paper is drawn. (In this case, it would be behavioral ecology.)

Now, finally, because the topic at hand is something with moral overtones, let’s say the author punctuated the critique with a pious remark about how siblicide is bad, bad, bad, and that the boobies who engage in it are “assholes,” and added that, hey, we can all overcome our bad, bad traits.

What kind of person would thoroughly botch an argument about a paper, condemn an entire discipline on the basis of the incorrect analysis, and then brandish their moralistic piety by condemning the behavior in question? I don’t know…some kind of Moran?



(Hat tip: Joe P. By the way, obviously I too oppose domestic violence.)

  • anon

    Seems that Moran is mired again in the naturalistic fallacy. The moralistic condemnation at the end gives it away. This outrage would be a complete non sequitur in a purely scientific discussion. But, it makes sense if we assume that Moran falsely interprets theories about function as justifications for wrongdoing. He implies that if wrongful behavior has an evolved function, then the perpetrators can’t change and should get our sympathy. Hence, when perpetrators can change and shouldn’t get sympathy, then their behavior has no evolved function… and there’s something “seriously wrong” with researchers who test functional hypotheses.
    And who could disagree? The world obviously needs more condemnation and less hypothesis testing.

  • Ian

    Dolphins are also known to engage in violence against women because of an unhealthy culture. God I hate those aquatic assholes.

    • Jesse Marczyk

      And to think people want to “dolphin-safe” our tuna. You know what we should be doing? Dolphin-safeing those female dolphins from the misogynistic culture.

  • Jesse Marczyk

    Since these critiques of EP are some common, how about a response in limerick form?

    When silly critics of evolutionary psych
    Tell the world these studies are like
    Excuses for misogyny
    And evil behavior apology
    Threatening to bring back Third Reich,

    Those critics will proclaim,
    “Those who rape and maim
    Will turn to our field
    For a convenient shield
    In order to avoid any blame”

    When the topic under discussion gets heated
    The misunderstandings are always repeated
    “Genes don’t determine behavior”
    Is always their savior
    Despite this point long being defeated

    Their sense of self-satisfaction
    Persists without any retraction,
    Admission of fallibility,
    Lack of civility,
    Or awareness of any infraction

    It would seem their moral outrage
    Has left them biased and unable to gauge
    Accurately the research they hope to dismiss
    Leaving them only to curse and to hiss
    In a manner unbefitting a sage.

    These critiques are quite the bore,
    and we’ve all heard this shit before.
    We’re left only to shake our fist,
    As they seem persist
    Not unlike an academic cold sore.

  • Larry Moran

    My response is: Boobies and Evolutionary Psychologists.

    It would be nice to see at least one evolutionary psychologist address the real scientific issues I raise. Maybe none of you understand evolution and/or science well enough to recognize the problem?

    • Robert Kurzban

      Your response contains nothing substantive, and contains many outright errors. For instance, you assert that the field makes three assumptions; in all three cases, you have misrepresented the assumptions made in the discipline. Your post raises no issues which have not been addressed repeatedly in many venues. If you actually want to engage with the epistemological underpinnings of the discipline, one paper that might help you is this one:

      Ketelaar, T. & Ellis, B. J. (2000). Are evolutionary explanations unfalsifiable?: Evolutionary psychology and the Lakatosian philosophy of science. Psychological Inquiry, 11, 1-21.

      If you’d like to try to understand the assumptions that actually underlie the discipline, a good place to start is this chapter:

      Tooby, J. & Cosmides, L. (1992). The psychological foundations of culture. In J. Barkow, L. Cosmides, & J. Tooby (Eds.), The adapted mind: Evolutionary psychology and the generation of culture. New York: Oxford University Press.

    • anon

      Seems that the gist of Moran’s issue is that he thinks it’s implausible that complex behaviors such as targeted aggression have a genetic basis, this would require “amazing alleles”, and because evolutionary psychologists don’t always hunt for the underlying alleles, the hypotheses are unscientific.
      Origin of Species has no mention of specific alleles or estimates of selection coefficients, so I guess Darwin didn’t understand evolution or how to think about the functions of adaptations…
      The evolved function of a behavior is not demonstrated by finding a single underlying allele. (Indeed, finding the relevant genetic sequences would only beg the question–What FUNCTION promoted the success of these alleles? A question which requires looking outside of the organism, not inside its genome)
      The evolved function of a behavior is indicated by its articulate complexity, specifically by the mapping of sensory inputs to well-organized behavior. Functional hypotheses are tested by examining the fit between the structure of this input-output mapping and a hypothesized adaptive problem.
      Evolution by natural selection is the only known physical causal process that can generate complex, functional systems such as the eye, kin altruism mechanisms, or kin aggression mechanisms.
      Hence, when a biologist observes complexly organized behavior, candidate explanations come from possible functions which produce that behavior either directly, or as a byproduct of performing some other function.
      *Some* genetic basis can be inferred from the complexity of the trait, but the specifics of those genes need not be of great interest… particularly because our current understanding of how genes contribute to complex traits–whether behavioral or anatomical–is so poor. We know so little about this that it is hardly time to speculate about what is or is not plausible when it comes to the genetic basis of complex traits.
      The quality of thought on Moran’s blog by both the author and commenters is surprisingly shallow… do you really think the way to understand evolved traits is to map them one by one to genes? Again, though this mapping could be interesting, it would only beg the question about the causal processes that selected for the genes/traits. For that question you need to look at adaptive problems.

    • Jesse Marczyk

      When you assume people are looking for genetic determinism, only to find people aren’t testing for genetic determinism, that must make those genetic determinism ideas seem pretty unscientific.

      I’m confused as to what that has to do with evolutionary psychology, but perhaps I really don’t understand science well enough.

  • The Other Jim

    “Evolution by natural selection is the only known physical causal process that can generate complex, functional systems such as the eye, kin altruism mechanisms, or kin aggression mechanisms.”

    No, it is not.

    A good start on the topic,
    Fernández & Lynch. Nature. (2011) 474(7352):502-5.
    Gray et al. Science. (2010) 330(6006):920-1.
    Koonin. Nucleic Acids Res. (2009) 37(4):1011-34.
    Lynch. Proc Natl Acad Sci U S A. (2007) 104 Suppl 1:8597-604.
    Stoltzfus (1999). Journal of Molecular Evolution, 49 (2), 169-18.

    • anon

      You missed the point. These papers are about causes of “gratuitous complexity” rather than *functional* complexity (think Paley’s watch and the Problem of Design). The processes behind gratuitous complexity lead to worse rather than better designs; they can’t explain the eye, kin altruism, kin aggression, or other well-designed adaptations. To the extent that a trait is functional, natural selection is behind it, to the extent that a trait is non-functional (gratuitous or otherwise) a number of factors could be responsible, including those discussed in these papers.

      • The Other Jim

        I think you are missing the point. Complexity and “gratuitous complexity” look the same. For years people were making up adaptive excuses for things like genome size, introns, RNA editing, etc. It was not until we stopped assuming all things are adaptive that anything in genome architecture started to make sense. Given that the most obvious traits in genome structure are not adaptive, perhaps we should be more careful in all other aspects of biology as well?

        In populations with small effective population sizes (Ne), genetic drift is very powerful. If Ne.s (s = selection coefficient) are <<1 then even in the single allele case, drift is more powerful than selection (some guess ~0.25 is sufficient to hide an adaptive allele). Humans fall into the small Ne category, with our estimated 30,000-8,000 Ne. The selective co-efficient in a single allele case has to be very strong in order for “adaptive evolution” to over-ride drift. It gets worse for complex, multi-allele traits. The same for negative selection. Only really bad alleles are seen and “purged” by negative selection. So what you call “gratuitous complexity” does not make “worse” as you imply. They are neutral difference (or nearly neutral, if you prefer Ohta).

        So this leads into the criticism that started this whole exchange – is a given behaviour so beneficial that it makes Ne.s > ~0.25? If no, the behaviour can be segregating within the population, even at high levels, but it is indeed neutral – non-adaptive. You keep falling back on the eye, so lets look at sight. Does sight give an organism more benefit than some quirky behavoiur? Comparing a base sense to a subtle behaviour difference is a false analogy.

        • The Other Jim

          … and apologies for messing up the italics tag in my post. I was trying to keep the variables in italics, but messed up in paragraph 2…

        • Robert Kurzban

          So, if I might just clarify. Are you saying that the patterns of data of the type that Duntley/Buss describe – men harming women when the probability of sexual infidelity is higher, for instance – might be explained by genetic drift?

          • The Other Jim

            No. I am making no conclusive statement at all. I am saying it is up to each researcher to use neutral evolution as the null hypothesis to reject, before concluding something is probably adaptive. Especially in vertebrates which are renowned for their low Ne.

            In the words of Mike Lynch, “Although those who promote the concept of the adaptive evolution of the above features are by no means intelligent-design advocates, the burden of evidence for invoking an all-powerful guiding hand of natural selection should be no less stringent than one would demand of a creationist. If evolutionary science is to move forward, the standards of the field should be set no lower than in any other area of inquiry. ”
            Lynch. Proc Natl Acad Sci U S A. (2007) 104 Suppl 1:8597-604.

          • Robert Kurzban

            Thanks for the clarification. That does raise another question, if I might. Readers of this blog, I believe, endorse Williams’ view of the relationship between observations (of shape, behavior) and claims of adaptation Biologists do to. To be concrete, I once blogged about an example in seahorses, for instance. Is the inference they want to make (from shape (horse shape) to possible function (prey capture)) incorrect, on your view?

        • Marco DG

          Actually, it doesn’t look that hard to make Ne*s>0.25 in human populations. With Ne=10,000, it only takes a selection coefficient s>0.000025 (i.e., a 0.0025% increase in fitness) to overcome drift.
          Even with Ne=200 (e.g., a reproductively isolated ancestral human group – 200 is the estimated effective deme size in the Pleistocene; see Eller et al. 2004, Hum. Biol. 76), the required selection coefficient is only s>~0.001 (i.e., a 0.1% increase in fitness – one thousandth of an additional offspring). I wouldn’t call these “very strong” selection coefficients. Am I wrong here?

        • Marco DG

          Just to be clear: the “one thousandth of an offspring” above means that carriers of the favored allele leave 1.001 offspring for each offspring left by carriers of the non-favored allele (i.e., it is a relative advantage, not an absoute one).

          • The Other Jim

            Sorry. I missed this until you pointed this out below…

            You are calculating the point where selection and drift “separate”, so to speak. At and below this, the probability of the allele fixing is (after solving the equation) equal to the mutation rate. Or about ~10^-8 to ~10^-9. Not good odds. The selective co-efficient needs to scale up by many orders of magnitude for the fixation probability to strongly bias for fixation (ie. positive selection).

  • Justin

    It appears that measuring/discovering the alleles for behavior is the only way to do evolutionary research (not just evopsych, but all evolution research) scientifically. So it appears that true science only occurs when things are reduced to alleles. This makes me think that as a psychologist I have never done scientific research because I have never discovered alleles for behavior. Similarly, I have never found neurological evidence (no fMRI, no split brain patients, no neuro-stimulation with primates) to support a major tenet in my research that motivation (effort, persistence, development of new strategies) is higher when the goal is specific and difficult rather than vague and easy (goal difficulty effect). I have a computational model of this effect, I even have a computational model of how individuals strive for two competing goals, but no neurological evidence, no alleles, nothing. I guess I am not a real scientist and this effect must not be real. I mean I would like to think the goal difficulty effect is because of computational mechanisms in the brain, but I just don’t have the evidence for it. I think I will start looking for computational mechanisms in the left thigh to explain this effect.

    • Alex

      Justin your thigh based research program sounds laughable… everyone knows that the major competing hypothesis against the idea that computations actually happen in the brain is that they happen in the toe. Thank god for Fmri work for finally putting that debate to bed.

  • The Other Jim

    @Robert Kurzban
    I can’t reply to you comment, so I’ll start again here. I think we ran into the right-hand wall…

    “Is the inference they want to make (from shape (horse shape) to possible function (prey capture)) incorrect, on your view?”

    The point I am making is that we have no idea. It is a hypothesis. Now it is time to test it.

    • Robert Kurzban

      Thanks for taking the time to comment. This was very useful (to me), and I’m happy to tie it off here. I do want to note that your remark, I think, stands in contrast to what the authors would say. They would argue, I think, that they have more than no idea at all (that this was a “step”) and that their work was a (first, preliminary, etc.) test of the hypothesis. (This is usually where an interlocutor says they still can’t be absolutely certain. Of course, but no one ever makes such claims, so such remarks are unhelpful.) Some biologists think that one can update a prior about a function on the basis of form (shape, behavior), and others, such as yourself, don’t. I’ve mentioned before that Futuyama, in his textbook, endorses this logic, writing: “Several methods are used to infer that a feature is an adaption for some particular function” (p. 261), and he indicates methods that, indeed, people in evolutionary psychology use. If some biologists, including biochemists, reject the logic articulated by Williams, seen in textbooks such as Futuyama’s, and used in papers such as the seahorse paper in Nature Communications, that’s fine. No one is suggesting that there is no room for debate. But note that this practice does not signal incompetence in biology; it fits comfortably within biology, though, again, there are those who disagree about the broader issue. This leaves us with the question of why protests are voiced in the context of adaptationist inferences like those drawn by Buss/Duntley, but not in the context of organisms such as seahorses. Additional research will be needed to address this question.

      Ps: Sorry about the “wall.” That is a side effect of the site layout, but not a design feature to prevent discussion. See what I did there?

      • The Other Jim

        This leaves us with the question of why protests are voiced in the context of adaptationist inferences like those drawn by Buss/Duntley, but not in the context of organisms such as seahorses.

        One word – exposure. You can’t help but notice anything claiming a human behavior – especially a negative one – as it ends up in all of the popular press. Thus it serves as a lightning rod for these discussions. There are many examples in behavioural ecology of excessive belief in adaptive evolution as the only mechanism, but no newspapers cover “behavior of weird fish X” stories.

        Your reply comment once again is mis-representing my opinion a bit. I am not denying “one can update a prior about a function on the basis of form (shape, behavior)”. The bee in my bonnet is that the priors are often defined as “my adaptive story” versus “their adaptive story”. The first test should be a question of whether there is any evidence of adaptation versus drift. Eliminate boring old drift, then get on to the interesting stuff! Especially in large vertebrates with small Ne, that are prone to drift. It is possible that a trait can fix that has no benefit or may even be slightly deleterious. This is known, but is continually ignored.

        • Jesse Marczyk

          I don’t see how drift would help to explain why men seem more inclined towards violence against their mates – among other behaviors – when paternity uncertainty is higher.

          I also don’t see how exposure is the simple, one-word explanation you suggest it is. It raises the question – if it’s true – as to why certain types of papers tend to get more exposure and act as lightning rods for debate. Holding the exposure constant would probably leave you with a great deal more debate about adaptationist papers dealing with humans than non-humans. Also, exposure probably has little to do with the voiced concerns over the paper (sample size, generalizability, , the ability to make those inferences in the first place, etc)

  • The Other Jim

    I don’t see how drift would help to explain why men seem more inclined towards violence against their mates – among other behaviors – when paternity uncertainty is higher.

    No one doubts the existence of the behavior. It is the adaptive hypothesis that is doubted (violence against a mate in a social species increases your reproductive success). See my population-genetic arguments above. This had better be VERY beneficial to convey an selective co-efficeint great enough to be selected for in a low Ne population like humans. And if it is so highly selected for, the alleles will carry the signature of a selective sweep. Test for that.

    I think your point is just boiling this down to a simple correlation versus causation debate. Do you believe penis size correlates to a nation’s GDP? Some do… ( Are you skeptical of a causative effect here? Even though the researcher are very convinced of their stats?

    To use the analogy trick, what is the purpose of the adipose fin in a fish? (second figure on the right at Highly conserved. Prominent in large predatory fish. Insert adaptive story here. But it is removed by fisheries conservation officers to mark hatchery raised fish (often in trout and salmon), so that fishermen / conservation biologists can distinguish wild-born ones from the hatchery ones. Removing this highly conserved organ has no fitness effect on the fish studied so far (until the conservation laws say you can eat the ones with no fins, but must release those with a fin ;-) ). What does this say about the adaptive story we could have made above?

    Also, exposure probably has little to do with the voiced concerns over the paper (sample size, generalizability, , the ability to make those inferences in the first place, etc)

    As someone who routinely gets into these debates with my behavioral ecology colleagues over these same issues, I disagree completely with this statement. The debates go on at seminars or conferences at departmental meetings, and not on blogs in the public eye. But the debate gets just as lively.

    • Jesse Marczyk

      I don’t really know much about a GDP/Penis size correlation or adipose tissue in fish, so I wouldn’t comment on them

      Just so I’m clear: you say certain violent mating tactics would need to have a great fitness benefit to be selected for in a small population and we should look for a certain type of genetic signature because of that?

      If I’m wrong there, feel free to ignore the corresponding parts of this response.

      I would say four things on that point: (1) I don’t (and I presume others don’t as well) know what alleles are involved in the behavior, so it’s hard to really test for anything about the alleles at this point.

      (2) It makes certain assumptions about the population size during which the behavior began to make it’s appearance.

      (3) Precisely how large of a selection co-efficient do you estimate it would need? If such tactics result in a male siring one additional offspring of his own X% of the time, rather than investing in the offspring of another male while foregoing about 3 years of his mate’s reproductive life, how large would X need to be? I suspect not very, given the rather large fitness cost that could be avoided.

      (4) It would seem the behavior itself would need to exist before it could be subject to drift in the first place, no? If that is the case, drift might help to explain it’s prevalence but not it’s origin or maintenance.

      • The Other Jim

        I think at this point I should just point you to a population genetics textbook and leave it at that…

        • Jesse Marczyk

          You could, but it’s kind of a non-answer.

  • The Other Jim

    It’s not a non-answer. Sometimes a blog comment line is insufficient. By that comment, it seems you have no grounding in the discipline at all. And the wiki pages are ridiculously brief on the topic.

    Here’s a free, but a little old, textbook on-line

  • Jesse Marczyk

    When asked whether you thought the data about the mating violence “might be explained by drift” you said, essentially, “No, but yes”. On top of that, you seem to be suggesting “Trait X is the result of drift” be something of a default assumption until shown otherwise (i.e. “eliminate boring drift first, then get to the interesting stuff”).
    The only test you suggested was to look at the alleles, which we can’t do because we don’t know what alleles to even look at.

    When asked whether the inference from form to function could be made, there was another non-answer. You seemed to answer as if the specific hypothesis itself about prey-capture was the question, rather than the possible adaptive inference. If you don’t know whether something’s an adaptation or not, that’s fine, but it’s a different question from whether you feel such an inference could, in principle, be made.

    I don’t think this is a simple disagreement of correlation/causation. I feel the disagreement may run a bit deeper. For the same reason, I don’t think the disagreements that occur in the popular press are due to mere exposure.

    • The Other Jim

      We are talking past each other.

      I made 2 points. 1) Evo-psych gets picked on because they get a lot of press with controversial topics, like spouse abuse . This makes the paper more well known than a similarly framed paper on some species that journalists do not care about. The same criticisms of what often happens in these papers are also valid criticisms of other fields.

      Completely different point 2) The criticism of all work claiming to show an adaptive (positively selected) trait of any kind should be “did they rule out the more mundane models of the trait being present in the population?”. The good old null hypothesis. Neutral evolution is a fact as much as positive selection. So the standby explanation that “it’s there so it is an adaptive trait” no longer cuts it. Yes, this can be hard. But that is an insufficient excuse to not do it.

      In this specific case, drift or a spandrel could also very easily explain mate violence. Which option I find most intellectually pleasing is of no consequence, as none have any empirical evidence to back it.


      • Jesse Marczyk

        I think we are talking past each other. So I’ll be happy to call it off myself.

        We might just disagree on the starting point. I see no reason to favor the “trait X was/is reproductively neutral” option as the default. “It’s neutral” is a hypothesis that needs to be supported as much as the adaptive one.
        This isn’t about taking the easy or the hard way through things.

        As for what we’d each consider evidence either in support of one hypothesis or another, we may just differ there as well.

      • Marco DG

        The prior probability of the “null hypothesis”/drift explanation being true is often very small to begin with. See my calculations above – a very small fitness advantage (in the order of a 0.1% increase in fitness) is sufficient to overcome drift even in very small populations. With an ancestral human-like effective population size, even a fitness effect in the order of 0.001% would have been enough.
        Thus, if a given trait can plausibly have (or have had) a nontrivial fitness effect (not a “very large” one), drift explanations should be considered unlikely from the outset. Of course, they might be true, but should not be taken as the default: if the trait plausibly has (or had) a nontrivial fitness effect, the prior probability of an adaptive explanation being true is much larger than that of the drift explanation. This means that (a) in such cases, the burden of proof actually falls on those advancing a drift explanation; and (b) the custom of looking for adaptive explanations first is inferentially justified whenever the trait can plausibly have a nontrivial fitness effect; it doesn’t take evidence of very strong selection to justify adaptive reasoning.

        • The Other Jim

          Again, sorry for missing your post above, until you commented here.

          See above for why this number does not ensure fixation. The selective co-efficient must be much higher than in your calculation above. What you calculated was the point where the fixation probability of neutral versus positive selection begin to differ. The selection co-efficient needs to increase by a lot before the single allele for the trait has a high probability of fixation. At that point, it is still effectively neutral (and interestingly, a trait equally bad will also behave the same… so a neutral allele, a deleterious allele that reduced reproductive success or positively selected allele that increased it would all have the same probability of fixation at the Ne.s=0.25 boundary – in the ballpark of ~10^-8 or less.

          I would argue that positive selection is a specific claim. Especially when a mechanism is proposed. A neutral processes (ie drift) is the null to disprove before concluding there is positive selection. Then the hard work begins of fining the “why”.

          • Marco DG

            Yes, but the conceptual point at stake here is not the fate of an individual allele. The same mutation can arise repeatedly in a population (across individuals and/or time); also, different mutations with similar phenotypic effects may arise at different times. The point I wanted to stress is that, in human-like populations, drift dominates selection only for trivially small selection pressures. For example, if type B allele is selected over A allele with s=.01 (not a large effect), this is already three orders of magnitude larger than the “separation point” if Ne=10,000. This may not ensure fixation of any individual type B allele, but selection will dominate drift in the long run.
            Finally, I don’t agree that drift must be actively disproved before entertaining an adaptive explanation. In fact, the null-hypothesis approach is not the best way to do statistical inference, and this is why you have Bayesian and information-based approaches that directly compare alternative hypotheses with one another. In many cases, the null (or “nil”) is extremely unlikely to begin with, and it makes little sense to treat it as the default that must be disproved. For example, when dealing with mating-related traits, whose impact on fitness is almost certainly non-trivial, adaptive explanations are prima facie more likely to be true than drift explanations.
            I am aware that we can still legitimately disagree at this point (e.g., about the philosophy of null-hypothesis testing and other technicalities). However, the claim that you need evidence of very strong selection on a trait in order to make a plausible case for adaptation is false – or, at any rate, much more controversial than you implied in previous posts.

  • The Other Jim

    @ Marco DG
    s=0.01 or more is considered strong selection. Bacterial antibiotic resistance from plasmids (unable to grow versus able to grow) is experimentally measured in the range of 0.42-0.30 ( s=0.003, there is apprciable selection, but it is slow on our evolutionary tiomescale. From a paper on human genome selection “For selection coefficients of about 0.3% or less, the average time to fixation of a new favored allele is considerably longer than the ~70,000 years since the split of the African and non-African HapMap populations.” (

    However, the claim that you need evidence of very strong selection on a trait in order to make a plausible case for adaptation is false –

    Apologies for being unclear. This is not what I am saying. I am saying that from pop-gen modeling, the selective co-efficeint must be relatively strong to escape drift, and fix in the population with Ne ~ what humans experience. From you numbers, an “upward trend” can be expected at >0.000025, but this is 120x lower than the 0.003 discussed above, where we would expect a slow (but measurable!) fixation event. I’ve not seen models for 0.000025, but for s=0.0001 in the PLoS Genetics paper, things are starting to boarderline our ability to distinguish from drift (s= -0.0001 could not push the deleterious allele out of a population – model results are in the supplementals).

    We will agree to disagree on drift as a null. But neutral evolution is measurably abundant. I do think it is the best null we have (or a drift-selection prior starts very close to 0.95 in favour of drift). Again to lift a quote from the PLoS Genetics paper cited above, “We have argued here that strong, sustained selection that drives alleles from low frequency to near fixation has been relatively rare during the past ~70 KY of human evolution.”

    Thanks for the discussions all, but I’ve invested too much time here already, and have a few other things to attend to, so will have to leave the party now.

    • Jesse Marczyk

      Perhaps one more quick question, even if it won’t be answered: what kind of estimates – in terms of time – are we talking about for drift to drive something to fixation?

    • Marco DG

      @ The Other Jim

      Thanks for the interesting discussion. Of course, the extent of positive selection in the human genome is still a matter of debate; also, the methods used to infer past selection have limited detection and resolution power. An emphasis on ovewhelming neutrality may be premature at this point. In the PLoS Genetics paper you cite, the authors say:

      “We have argued here that strong, sustained selection that drives alleles from low frequency to near fixation has been relatively rare during the past ~70 KY of human evolution.”

      But then go on:

      “Is this conclusion compatible with recent work on haplotype-based signals reporting an abundance of partial sweeps with selection coefficients of ≥1%? One possible explanation for the apparent discrepancy is that there might be many more partial sweeps than completed sweeps. This could occur if selection pressures tend to be highly variable so that favored alleles often rise to intermediate frequency and then start to drift as a result of fluctuating selection pressures or polygenic adaptation. Alternatively, it is possible that recent studies have substantially overestimated the number and strength of partial sweeps.”

      In other words, the jury is still out on this issue. There are many alternative scenarios out there, and the role of neutral evolution may be smaller than expected, especially for traits such as social behavior, cognition, and so on. We’ll see.

  • Pingback: Could Evolutionary Psychology’s Critics Pass Evolutionary Psychology’s Midterms? | Evolutionary Psychology

  • Pingback: Hardcore Porn-etry | Pop Psychology

Copyright 2011 Robert Kurzban, all rights reserved.

Opinions expressed in this blog do not reflect the opinions of the editorial staff of the journal.

Evolutionary Psychology - An open access peer-reviewed journal - ISSN 1474-7049 © Ian Pitchford and Robert M. Young; individual articles © the author(s)

You're in!