[Same disclaimer as before: some of the following gets pretty technical, sorry guys, but we are talking professional philosophy here...]
Just two speakers for this symposium: Justin Remhof (University of Illinois-Urbana), on “Nietzsche’s reconception of science: overcoming nihilism,” and Robert Northcott (University of Missouri-St. Louis) on “Rethinking genetic drift.”
We start with Remhof. There are apparently two views of Nietzsche’s understanding of science: either in terms of scientific naturalism and its power to discover things about the world, or in terms of science as inherently nihilistic (jee, you think? Nietzsche as a nihilist?).
Remhof is going to argue that for Nietzsche science is actually life-affirming, as opposed to Christian-inspired nihilism. The basic idea is that determinate truths (coming from religion) are life-denying because they cannot be satisfied (i.e., verified). Science, on the other hand, is life-affirming because it is comfortable with the indeterminacy of our truths about the world.
I am not a Nietzsche scholar by any means, but I find it uncomfortable to put “science” in the same paragraph as words like “life-affirming reconception of ontology,” “will to power ontology,” or “extensionally individuated.” But that’s just me and my instinctive skepticism about anything that smells too much of continental philosophy...
Ok, an example straight from Remhof’s outline, seemingly quoting Nietzsche:
“If I make up the definition of a mammal, and then, after inspecting a camel, declare ‘look, a mammal,’ I have indeed brought a truth to light in this way. ... “[This] contains not a single point which would be ‘true in itself’ or really and universally valid apart from man.”
Hmm, really? It seems to me that a mammal remains a mammal whether or not there are men around to verify that fact. It is trivially true that without humans around nobody would conceptualize what a mammal is, or that the term “mammal” is linguistically arbitrary, but that’s not what Nietzsche is saying. “Mammals” are in fact an objective truth about the world, specifically they are a particular phylogenetic lineage of organisms on planet earth, with a given history and a number of non-arbitrarily distinctive features (despite the existence of Platypi).
More explicitly, again Remhof presumably quoting or paraphrasing Nietzsche:
“‘Truth’ is therefore not something there, that might be found or discovered — but something that must be created and gives name to a process ... [it is an] active determining — not a becoming conscious of something that is in itself firm and determined.”
No, no, no. Truth is something “out there,” and the only sensible discussion is about the epistemic limitations of human beings, which make it so that we can rarely be certain of the truths we think we discover. We don’t create truth, we discover it (partially, under certain conditions, using some methods, of which science is certainly a primary one).
One more perhaps obvious comment: it seems to me that “life affirming” and “life negating” are not attributes that are properly applied to science. Science’s object, unlike religion, is not to help us figure a way out of nihilism — that’s the job of philosophy!
[Incidentally, the talk referred to above is “philosophy of science” in a fairly non-standard sense of the term, from my experience.]
And now to Northcott and drift, a surprisingly controversial topic in both evolutionary genetics and philosophy of biology, about which I have written in collaboration with Jonathan Kaplan. So, generally speaking, drift is a sort of population-level “sampling error” that creates random fluctuations in gene frequencies as a result of the finiteness of the size of biological populations. (This is directly analogous to the observation that short runs of coin-flipping do not typically result in exactly 50-50 outcomes of tail and head, again because the run is short and subject to stochastic outcomes.)
According to Northcott “selection probabilities leave out many idiosyncratic [causal, as opposed to systematic] factors. The ‘sampling error’ presumed to generate drift is the result of these latter factors.”
So, we can think of drift as an outcome (instead of a process), but outcomes cannot explain themselves, so we still need a causal explanation of the outcome of drift.
Another thought is that drift is a sort of causal dustbin, something that captures the action of all factors the effects of which are not captured by selection probabilities. The problem here is that drift then becomes a placeholder for unknown causes.
A third possibility is to conceptualize drift as indiscriminate sampling (as opposed to the discriminate sampling of natural selection). This is hardly better than the dustbin scenario, since whatever factors are responsible for such indiscriminate sampling remain unknown.
Northcott puts forth the view that the real causal factor is in fact population size itself, which needs to be distinguished by other non-selective factors. He points out, though, that if in a given population we start with two alleles (A and B) at equal frequencies, and there is no selection going on, and A drifts to fixation, this is not fully explained by drift-as-finite population size because that level of population size is also compatible with B going to fixation instead.
Generally speaking, Northcott takes the currently standard view that explanations are contrastive (X rather than Y explains Z) and admits of degrees of explanation. One can in fact use probability theory to calculate an explanatory score, or degree of explanatory strength for each proposed contrast (equations provided in the paper by the author).
Applying this approach to the example of the A,B alleles above (where the explanatory contrast is between finite and infinite population size), Northcott concludes that finite population size only weakly explains why A goes to fixation rather than B, but it fully explains why either A or B will go to fixation without selection.
I actually like Robert’s approach, but some fundamental questions remain, it seems to me: in what sense, precisely, is population size a causal factor in (as opposed to just being a correlate of) changing gene frequencies? And, as the calculation above clearly shows: what other factors do explain why a particular allele rather than another goes to fixation?
This comment has been removed by the author.
ReplyDeleteI am starting to lean towards Gillespie's position that genetic draft not genetic drift is the dominant stochastic force in population genetics. Since genetic draft depends much less on population size, this makes the discussion moot. How can the population genetics term of effective population size be considered a cause anyway, when it bears only a weak relationship to census population size and when we really do not understand what determines it?
ReplyDeleteJoanna,
ReplyDeleteNorthcott was careful to specify that the theory is expressed in terms of Ne, but the concept is the same.
As for genetic draft, it seems to me to be an entirely different kind of stochastic phenomenon, so I don't see why that concept would replace the discussion on drift.
The title of Gillespie's main paper on this is "Is the population size of a species relevant to its evolution?" He does mean census population size here, I believe. If genetic draft is important, and I believe it is, the answer seems to be no. Saying that what is meant in discussions of genetic drift is effective population size, without having any coherent theory of whether effective population size makes sense as a concept and if so what it means, doesn't seem to help the debate.
ReplyDeleteQuoting Massimo: "..but some fundamental questions remain...in what sense, precisely, is population size a causal factor in (as opposed to just being a correlate of) changing gene frequencies? And...what other factors do explain why a particular allele rather than another goes to fixation?"
ReplyDeleteI should admit that I have not read all of your relevant papers, essays, or books about this topic (although I have ordered a book from amazon:)), but those two 'fundamental' questions that you left us with seem to me to be less opaque then you made them seem (but perhaps I don't understand exactly what you are asking).
For the first question, the answer seems to be that it depends on the situation you are talking about. At mutation-drift equilibrium for example, where the mutation rate is equal to the substitution rate, the effective population size cancels out, the rate of gene frequency change independent of Ne,and hence Ne is not a causal factor. (of course that result only follows from the assumption of neutrality and a constant mutation rate).
In other cases, Ne is a causal factor in gene frequency change over time for all of the familiar population genetic reasons. The average time between consecutive mutation events is dependent on the effective population size (since larger population sizes have more genomes, and hence more opportunities for mutation events, the mean time between such events will be less than it is for smaller populations), the magnitude of genetic drift in a population will be determined by Ne (proportional to the inverse of Ne), and the effectiveness of selection depends on Ne (more effective with larger Ne).
But of course you are familiar with all of this, making me think that I don't really understand what it is you are asking.
However, it is more obvious that in your second question you are referring to Northcotts' conclusion that the fixation of A in the case of a population with alleles A and B with equal frequencies and no selection, is not fully explained by drift-as-finite population size because that level of population size is also compatible with B going to fixation instead.
Why do you need 'other factors' to explain that? Isn't it just simply a matter of probability? There are two alleles with equal initial frequencies, the fates of which will be determined by drift. So, as a long term average, A will be fixed 50% of the time, and B 50% of the time (if you iterated the scenario). The fact that in one particular case, 'A' drifts to fixation is not predictable in the short term, and does not require additional 'factors' to explain why it was 'A' and not 'B'. That seems clear, but maybe I am missing something...?
Josh,
ReplyDeletegood comment, but my first question had to do with in what sense population size - which is a descriptive demographic parameter - has causal effects on gene frequencies, as opposed to being a proxy for other physical causes that are correlated to Ne.
The second question also remains open, because probabilities don't explain anything, they just describe, right?
Thanks again, Massimo!
ReplyDeleteThe mid-period Nietzsche (esp. in The Gay Science) is somewhat enthusiastic about natural science insofar as it can deliver a realistic psychology to us. In a basically Humean or proto-Freudian mood, he becomes excited about the possibilities for such a psychology to liberate us from overly moralized accounts of who we are and how we act. This he sees as a precondition for an overcoming of modern morality.
However, I simply cannot imagine him arguing that "science" per se (especially in the extremely inclusive 21st century sense) has any intrinsic properties at all, let alone the property of being "life-affirming". Such a proclamation would be extremely obtuse for a man with such extraordinary historical sense, the man who believed that "only something without a history can be defined".
I see. So, if you are asking how population size 'itself' causally effects gene frequencies, as opposed to being a proxy for other physical causes, then I am totally at a loss, and the question (unfortunately) seems more mysterious than before. Population size certainly does have causal effects on gene frequency change, but it is only through its effect on the dynamics of the mechanisms of change (selection, drift, mutation, migration). Right? Surely you are not asking if population size itself is a mechanism for gene frequency change?
ReplyDeleteFor the second question; it's true I think, that probabilities don't 'explain' anything, only describe. The only explanation i can think of though (pardon the argument from lack of imagination), for why A is fixed rather than B, would sound something like 'The set of physical processes (indiscriminate 'random' sampling processes) were such that the accumulation of sampling errors over generations, lead to the fixation of the A allele'. - but that just simply says it was an unpredictable, chance event, and doesn't seem like it is the kind of answer you are after.
I mean, certainly there will be a series of physical events that causally lead to the fixation of A in some particular case, but you can't give a general answer to why it was A and not B. All you can say as a generality is that, as a long term average, each will be fixed 50% of the time (given the model Northcott described).
I agree with Josh. As an engineer, I'm very comfortable with the fact that some things, especially complex systems, are stochastic. If you're at the metastable peak of a two-dimensional hill in state space, the gradients in all directions drag the system in whichever direction it is first perturbed. That first perturbation is likely to be an unmodeled microfeature or even an unmodelable sum/product of many microfeatures. There is always interest in improving models to include more features, but the perturbations are likely too small for us to measure, so likely too small for us to confirm the model.
ReplyDeleteWhat, exactly, would you like to model, reveal, or comprehend from the system? I would like to think of an analogy with a simpler dynamical system, say a mechanical system, but I don't understand what you're looking for enough to do that, yet.
What Nietzsche really wanted;
ReplyDelete"Elimination of the weak and defective, the first principle of our philosophy. And we should help them to do it!"
Nietzsche, The AntiChrist, sec. 2
Keep in mind that Nietzsche died insane, and his writings were clearly affected by his insanity in the decade before his breakdown.
Atheists should be cautious about being linked with him.
Winston,
ReplyDeleteI think your quote is out of context. Through that passage and most of the book, Nietzsche refers to Christianity. Even though Nietzsche is harsh in his critique, the translation you chose to use seems to be misleading.
The Antichrist was writen before his nervous breakdown though it was publised later. Obviuosly he wasn't insane before his breakdown that's why it's called a breakdown.
Truth is some 'thing' 'out there' only metaphorically speaking; and metaphors are certainly constructed, not 'objective facts'. Methinks you are bewitched by language. Scientific models are not the final say on all things worldly. Indeed, various different scientific models can be brought to bare on the same phenomenon.
ReplyDeleteTake the wave/particle nature of light for example. Is light 'objectively' a wave or particle? Answer: depends on the scientifc model the subject chooses to bring to bare.
If you use a scientific model that is only capable of "discovering" (as you say) waves, guess what you're going to "find"? Waves. The whole activity is more contrived than you imagine.
Scientific models of reality are constructed, not discovered, which should be obvious because a method (itself constructed) was consciously chosen by a subject to derive them.
I believe Nietzsche would agree.