Monday, February 12, 2018

One philosopher's view of random genetic drift

Random genetic drift is the process whereby some allele frequencies change in a population by chance alone. The alleles are not being fixed or eliminated by natural selection. Most of the alleles affected by drift are neutral or nearly neutral with respect to selection. Some are deleterious, in which case they may be accidentally fixed in spite of being selected against. Modern evolutionary theory incorporates random genetic drift as part of population genetics and modern textbooks contain extensive discussions of drift and the influence of population size. The scientific literature has focused recently on the Drift-Barrier Hypothesis, which emphasizes random genetic drift [Learning about modern evolutionary theory: the drift-barrier hypothesis].

Most of the alleles that become fixed in a population are fixed by random genetic drift and not by natural selection. Thus, in a very real sense, drift is the dominant mechanism of evolution. This is especially true in species with large genomes full of junk DNA (like humans) since the majority of alleles occur in junk DNA where they are, by definition, neutral.1 All of the data documenting drift and confirming its importance was discovered by scientists. All of the hypotheses and theories of modern evolution were, and are, developed by scientists.

Nothing in biology makes sense except in the light of population genetics.

Michael Lynch
You might be wondering why I bother to state the obvious; after all, this is the 21st century and everyone who knows about evolution should know about random genetic drift. Well, as it turns out, there are some people who continue to make silly statements about evolution and I need to set the record straight.

One of those people is Massimo Pigliucci, a former scientist who's currently more interested in the philosophy of science. We've encountered him before on Sandwalk [Massimo Pigliucci tries to defend accommodationism (again): result is predictable] [Does Philosophy Generate Knowledge?] [Proponents of the Extended Evolutionary Synthesis (EES) explain their logic using the Central Dogma as an example]. I looks like Pigliucci doesn't have a firm grip on modern evolutionary theory.

His main beef isn't with evolutionary biology. He's mostly upset about the fact that science as a way of knowing is extraordinarily successful whereas philosophy isn't producing many results. He loves to attack any scientist who points out this obvious fact. He accuses them of "scientism" as though that's all it takes to make up for the lack of success of philosophy. His latest rant appears on the Blog of the American Philosophers Association: The Problem with Scientism.

I'm not going to deal with the main part of his article because it's already been covered many times. However, there was one part that caught my eye. That's the part where he lists questions that science (supposedly) can't answer. The list is interesting. Pigliucci says,
Next to last, comes an attitude that seeks to deploy science to answer questions beyond its scope. It seems to me that it is exceedingly easy to come up with questions that either science is wholly unequipped to answer, or for which it can at best provide a (welcome!) degree of relevant background knowledge. I will leave it to colleagues in other disciplines to arrive at their own list, but as far as philosophy is concerned, the following list is just a start:
  • In metaphysics: what is a cause?
  • In logic: is modus ponens a type of valid inference?
  • In epistemology: is knowledge “justified true belief”?
  • In ethics: is abortion permissible once the fetus begins to feel pain?
  • In aesthetics: is there a meaningful difference between Mill’s “low” and “high” pleasures?
  • In philosophy of science: what role does genetic drift play in the logical structure of evolutionary theory?
  • In philosophy of mathematics: what is the ontological status of mathematical objects, such as numbers?
[my emphasis LAM]
Before getting to random genetic drift, I'll just note that my main problem with Pigliucci's argument is that there are other definitions of science that render his discussion meaningless. For example, I prefer the broad definition of science—the one that encompasses several of the Pigliucci's questions [Alan Sokal explains the scientific worldview][Territorial demarcation and the meaning of science]. The second point is that no matter how you define knowledge, philosophers haven't been very successful at adding to our knowledge base. They're good at questions (see above) but not so good at answers. Thus, it's reasonable to claim that science (broad definition) is the only proven method of acquiring knowledge. If that's scientism then I think it's a good working hypothesis.

Now back to random genetic drift. Did you notice that one of the questions that science is "wholly unequiped" to answer is the following: "what role does genetic drift play in the logical structure of evolutionary theory?" Really?

Pigliucci goes on to explain what he means ...
The scientific literature on all the above is basically non-existent, while the philosophical one is huge. None of the above questions admits of answers arising from systematic observations or experiments. While empirical notions may be relevant to some of them (e.g., the one on abortion), it is philosophical arguments that provide the suitable approach.
I hardly know what to say.

How many of you believe that the following statements are true with respect to random genetic drift and evolutionary theory?
  1. The scientific literature on all the above is basically non-existent.
  2. The philosophical literature is huge.
  3. The question does not admit of answers arising from systematic observations or experiments.
  4. It is philosophical arguments that provide the suitable approach.

1. There are some very rare exceptions where a mutation in junk DNA may have detrimental effects.


  1. Do you understand what he means by the "logical structure of evolutionary theory?" As opposed to just "evolutionary theory"?

    1. No. Do you? Apparently it's something that scientists can't investigate but philosophers can.

    2. I've read him discuss drift before. I doubt he's unaware of the role it plays. It sounds to me like this could just be a misunderstanding

    3. This is likely what he is talking about:

      Is the metaphor of a "force" an appropriate metaphor for describing genetic drift

      The question here seems to be: how best should we describe or think about drift? The question isn't: is drift a significant effect?

    4. How do you think Pigliucci would answer the four questions at the bottom of my post?

      Do you really think it’s important to decide whether the word “force” is appropriate or not? Is that what philosophers worry about? Do they publish papers on whether the force is part of the logical structure of evolutionary theory?

      I’ve read some of his critiques of modern evolutionary theory. He has never convinced me that he understands the importance of drift. He rarely even mentions it but he talks about natural selection a lot.

    5. Well, you know my opinion on the matter. The idea that selection and drift can be meaningfully separated is incorrect and that nobody every calculates what drift looks like in anything but the neutral case. It's not something particularly interesting - though working out things like the probability of fixation given s and N is. As such the idea that drift and selection are forces is misleading, in particular because for forces in physics the principle of superposition works, whereas for selection and drift it doesn't.

      I think this is important, because getting it wrong produces misconceptions and by now we have a particular brand of creationist nonsense that preys upon these misconceptions (Sanford is good at exploiting common misconceptions for instance).

      I mean, I can point to things you put in your post that are nonsense, really. "Most of the alleles that become fixed in a population are fixed by random genetic drift and not by natural selection." doesn't even start to make sense, because models without drift fail to fix anything, allele frequencies go to 1 asymptotically, but never reach 1 in finite time.

      "Most of the alleles affected by drift are neutral or nearly neutral with respect to selection."

      Since most alleles are neutral or nearly neutral, that's hardly surprising.

      "Some are deleterious, in which case they may be accidentally fixed in spite of being selected against."

      At this point I note that alleles that are positively selected are not listed, as if they weren't subject to drift. But as noted above, none of them would get fixed without drift either.

      It's worth noting that in the link Aceofspades, Pigliucci claims that the mathematical models of drift and brownian motion "look very different". That seems ridiculous given the diffusion approximation is a Wiener process aka Brownian motion.

    6. I agree with many of the things Simon has said, particularly with the validity of the analogy between random genetic drift and Brownian motion. I am not at all clear on why Massimo Pigliucci thinks that it is not valid for evolutionary biologists to study random genetic drift while it is valid for physicists to study Brownian motion.

      However I have always thought of both Brownian motion and random genetic drift as "forces". Of course, down at the individual molecule level Brownian motion is not used as a "force", just as at the level of an individual organism, it might die because of a falling log rather than because of random genetic drift.

      Similarly do geologists think of erosion as a force? I suspect that they do and I'm not going to tell them to stop doing so.

    7. I don't think geologist in general think of erosion as a force. But that might be different for phyical geographers, which I didn't have much to do with, so I couldn't tell. But generally most models in geology deal primarily with subsurface structures and the forces there are in fact physical forces - primarily gravity and then you have the effect of convection. I think in a lot of cases you are more likely to find people thinking about energy, when it comes to errosion, because in a lot of cases rock building minerals are balancing internal energy and gravity, with higher pressure leading to denser rock. When you have uplift and exposure, minerals that are further from that equilibrium tend to weather faster.

    8. Joe says, "I have always thought of both Brownian motion and random genetic drift as 'forces'."

      I'm afraid your thoughts don't count because you're a scientist and your views are beyond the scope of science. :-)

    9. @Simon: I was thinking of the kind of geologist you were calling a "physical geographer".

      But OK, let's say a "process" rather than a "force". Brownian motion is a process in the same sense that Mendelian segregation is a process (or genetic drift is a process, or erosion is a process, or natural selection is a process). And there simply is no reason to have one of these be out-of-bounds for evolutionary biologists.

      Important early work on the effects of genetic drift was done by Sewall Wright, R.A. Fisher, J.B.S. Haldane, and Motoo Kimura. None of whom were philosophers.

    10. I don't think of geomorphologists as geologists, but that's probably a cultural issue. In German universities you generally have geography departments, which are internally subdivided into phsical geography and anthropogeography. In many US university the geography departments are generally focused on the social science aspect and earth science departments contain the research groups that do surface processes.

      The reason I'm so opposed to thinking of selection and drift as forces, is that as a concept in physics, the main selling point of forces is superposition. There are formalisms in physics that entirely disperse with the notion of force and for some problems they are easier to use. So if something is analogous to a force, superposition should apply and then s shouldn't appear in the variance term of the diffusion approximation.

      As far as the other point goes, I'm not sure what Pigliucci means by "the logical structure of evolutioanry theory". But throwing "logical structure of evolutionary theory" into scholar google, led me to a dissertation (Jun, 2014), which states: "Although the founding works by Fisher, Wright, Haldane and the subsequent theorists established the rich field of mathematical genetics that flourished in the 20th century, philosophers have long questioned its theoretical nature. In particular, they are concerned whether the equations used in mathematical genetics describe causal processes underlying evolution, or something else."

    11. @Simon: If the philosophers don't want to call evolutionary processes "causal", that is fine, but I don't think that it would change how you and I do our math. We'll just have to wait and see what new set of tools they come up with. and whether they can show that the conclusions we come to from using our existing tools are invalid.

    12. I don't think their goal is for us to do anything different. They aren't doing science, they are doing philosophy and when "What is a cause?" is a metaphysical question, I don't think the question whether evolutionary processes are causal can be subject to empirical study. It's also not a question we need to have an answer to to do biology.

    13. A better treatment of Pigliucci's views on drift can be found in his book with Jonathan Kaplan (Pigliucci's advisor at Tennessee) "Making Sense of Evolution". But they essentially take one definition of drift and call it unfulfilling. I think they overlook other definitions and practicality.

  2. Or the 40-year-old distinction between a theory and a research program, integral to Pigliucci's thought?

    1. Yes. I know the difference between evolutionary theory and a research program. Random genetic drift is an integral part of modern evolutionary theory. You can have a research program that's devoted to detecting instances of drift in evolving populations. You can also have a research program that attempts to refute the role of drift.

  3. Philosophy is as useful to physicists (and biologists) as ornithology is to birds.

    1. Not original of course and wrong of course. The protection and even survival of many species of bird threatened by humans is also assisted by humans drawing on the insights from - you guessed it (didn't you?) Ornithology

    2. It is attributed to Richard Feynman.

  4. I find the notion of Genetic Drift & Neutral Theory as major forces in Evolution particularly challenging topics in the high school classroom.

    I made an attempt to wrap my students heads around the notion that Evolution is not always about Natural Selection for adaptive advantage – with a case study worksheet

    I am not sure I succeeded. The relevant bit starts at page 5. The problem here is that I am venturing into uncharted territory as far as introductory Bio is concerned.

    I would appreciate any corrections or suggestions for improvement from any corner.

  5. Add me to the list of people who do not understand the question. I would appreciate if you would explain what you think Pigliucci means by it, because I don't understand that either. What is a "role in the logical structure"?

  6. "The second point is that no matter how you define knowledge, philosophers haven't been very successful at adding to our knowledge base. They're good at questions (see above) but not so good at answers."

    This is exactly the purpose of philosophy, to be able to get at knowledge by understanding how to ask meaningful questions.

    1. Which of the seven questions listed above are “meaningful”?

      Can anyone else ask meaningful questions? Scientists? Historians? Engineers?

  7. I don’t expect this response will be greeted charitably (the principle of charity being a philosophical concept, after all) but I’ll respond anyway.

    All the questions are meaningful. Are you suggesting that none of them are meaningful, except maybe the bit about genetic drift? Or is it that, with respect to all the questions except the one about genetic drift, science really can provide no clear answer to them (which is the author’s point), but that because of this fact alone, the questions are not meaningful? IOW, only science can ask meaningful questions. This is scientism, not science.

    What research program will establish the ontological status of numbers, or answer any of the other questions apart from drift?

    Even with respect to drift vs. selection, I notice that biologists disagree. I distinctly recall a few years ago on this blog, you debated Richard Dawkins on this point — the subject being horns, iirc correctly. It was a pretty heated exchange.

    In another thread you wrote this:

    “My view is that science is a way of knowing that relies on evidence. This encompasses all kinds of approaches that are outside of the traditional boundaries of biology, geology, physics, chemistry etc. There are no subjects that are immune to investigation using science (broad definition) and that includes the supernatural world if it exists.

    Most of those investigations do not involve conducting experiments in the way that you seem to envisage.

    Most of the activities of Intelligent Design Creationists count as science, in my opinion. They are invariably examples of BAD science but that's not limited to ID. Lots of traditional scientists also do bad science but nobody questions whether they are doing science.”

    But what you wrote above is pre-eminently a philosophical stance, and NOT a scientific one! It goes right to the heart of the demarcation problem, which is philosophy. It is also EXACTLY the position adopted by the philosopher Brad Monton, in his 2006 philosophy paper critiquing the Kitzmiller/Dover decision. That paper is here:

    All this said, it seems evident that the writer oversteps when he says, with respect to drift: “The scientific literature on all the above is basically non-existent, while the philosophical one is huge.”

    But as Einstein recognized, science is inevitably shot through with philosophy.

    1. Question #3 is not meaningful because we already know that there’s no simple answer. No definition of “knowledge” satisfies everyone.

      Question #4 is not meaningful because for the vast majority of women the issue of whether to have an abortion or not doesn’t hinge on whether a fetus can feel pain. The question is especially not meaningful for old men.

      I know that philosophers love epistemology. They’ve been struggling for over a hundred years to come up with a definition of “science” and “knowledge.” I’m a scientist. I thought about it and picked a few working definitions that help me decide whether ID is science and whether someone is doing good science or bad science. The fact that philosophers can’t do this and move on doesn’t mean they have exclusive rights to the problem. Just the reverse. The fact they can almost never ANSWER a meaningful question means they are irrelevant.

  8. Questions without simple answers aren't meaningful?

    1. Questions with simple answers are mostly boring.

      Questions that can't be answered are usually not meaningful.

      Questions with complicated answers may or may not be meaningful.

  9. So you decide what science is but do not acknowledge that this is a philosophical stance?

    1. That’s correct. I do not acknowledge that philosophers have a monopoly on defining science. And I do not acknowledge that choosing a definition of something automatically puts it in the domain of philosophy.

  10. Also, nobody said philosophers have "exclusive rights" to the demarcation problem. I also notice you evaded the bulk of my post.

    1. What part of your bulk would you like me to address?

  11. Somehow my post the the one just previously publish vanished.

  12. The post that vanished was: You claim to decide what science is, but do not acknowledge that this is a philosophical stance. Is that right?

  13. A philosopher would notice the difference between evolutionary theory and natural history.

  14. Our Philosophy Department recently announced a day-long symposium on "Philosophy of Probability". One of the speakers, Marshall Abrams of University of Alaska - Birmingham is to speak on "Imprecise Chance in Evolution". With a heavy heart I looked him up on the web to see what his argument would be. To my surprise, he argues that the way population geneticists think about factors like random genetic drift is appropriate. At least that's what I think he says, based on very little actual reading of his works. Here is for his most recent
    an abstract of his most recent paper
    and here is an earlier paper.
    I'll be attending his talk. Should I go with a less heavy heart?

    1. What sciences would survive if they were required to provide evidence for their clams? Would population genetics? How about evolutionary biology?
      I don't have to motion the origin of life "sciences" do I?

      It is in the best interests of universities to promote the exclusive access to "scientific knowledge" whether it is scientific or not. It means control and in many cases money...

      This set up of the education system promotes an illusion of exclusive access to superior knowledge and the benefits that suppose to follow that are just as allusive as the supped scientific knowledge that in majority of cases is without any merit.

  15. Oops, he's at University of Alabama - Birmingham.