Wednesday, March 25, 2009

Further thoughts on challenges to the Modern Synthesis

Very good papers this week. Rose and Oakley are actually much more knowledgeable about evolutionary biology sensu stricto than are many modern systems biologists and evo-devo folks who see their fields as challenging the Darwinian orthodoxy. Lots of interesting conceptual issues to chew on here, including:

1. If we want to say whether Darwinism, or the Modern Synthesis, or the possible Postmodern Synthesis, is a 'paradigm shift', we need to define what we mean by 'paradigm shift'. Infamously difficult, due in no small part to ambiguity in the writing of Thomas Kuhn, the physicist/historian/philosopher who came up with the idea of paradigm shifts. Kuhn used 'paradigm' in lots of very different ways, and philosophers have been arguing about his ideas ever since. I do think it's pretty clear that some senses in which Kuhn used the term don't really apply to Darwinism or the Modern Synthesis. As Mike indicated, there was no failure of communication between Darwin and his opponents, or between the developers of the Modern Synthesis and their opponents. People weren't talking past one another. They mostly agreed on what the questions were and simply disagreed about the answers. That means those paradigm shifts, if that's what they were, weren't like 'gestalt switches', contrary to some of Kuhn's claims. (A gestalt switch is like when you look at an ambiguous drawing such as the famous one that can look like either a rabbit or a duck, and switch from seeing it one way to seeing it the other way. The two different perspectives are just that--different. Neither is incompatible with the other, and so the choice between them isn't rational but rather is merely a matter of preference).

2. The distinction between proximate and ultimate explanations, and the relationship between them, seems to me to absolutely indispensable for understanding the relationship between evolutionary biology and molecular biology/genetics/biochemistry. This distinction is, or should be, familiar to biologists. Ethologist Niko Tinbergen wrote a classic 1963 paper on four kinds of questions in ethology, which basically draws this distinction. Unfortunately, a lot of unproductive argument and attempts at synthesis in science basically boil down to arguments over which of these kinds of questions is "best".

3. Following on from 2, it's very interesting to think about how proximate and ultimate explanations relate to one another, both in general and in specific cases. Simple reduction of ultimate to proximate explanations generally isn't possible or desirable. This is in part because of supervenience--properties like "fitness" are not physical properties like "mass" or "charge". So statements about fitness cannot be translated into, and thereby reduced to, statements about physical properties, although an organism's physical properties obviously affect its fitness. This kind of "translation problem" crops up in other fields. It's the basis of the mind-brain problem the prevents psychology from being reduced to neuroscience. States of mind (e.g., "happy", "sad") can be correlated with states of the brain (patterns of neuron firing), but that doesn't mean the mind "is nothing but" the brain.

I actually think there are a lot of really cool scientific questions that are inspired by our new genetic knowledge. But those questions seem to me to fall within the conceptual framework of the Modern Synthesis. For instance, much of the mathematical theory of quantitative genetics developed by Fisher, Haldane, and Wright assumes that a quantitative phenotype of an organism(e.g., its weight or length) is determined in part by its genotype at each of infinitely many loci, each of which has a very small, additive effect on phenotype. This assumption is mathematically convenient, because it turns out to predict that phenotypic variation will be normally distributed (i.e. follow a bell curve), and normal distributions are easy to work with mathematically. This assumption is also empirically supported, in that we do indeed observe that phenotypic variation of quantitative traits like height often is normally distributed, even when all individuals are grown in a shared, controlled environment. But recently, modern molecular biology has falsified the genetic assumptions on which the mathematical theory is based. We can now identify most of the genetic loci that affect, say, height, and we can estimate the additive effect of each genotype at each locus. And it turns out that genetic variation in quantitative traits is mostly due to variation at only a few loci, each of which has a big rather than a small effect on phenotype. But this doesn't mean we should just chuck quantitative genetic theory and start over, since (again) it's a matter of empirical fact that phenotypic variation is normally distributed. So quantitative genetics gives the right answer for the wrong reasons, and it's a very interesting question to try to work out how that's possible. We now know the genome isn't at all like we thought--so how can it be that false assumptions about the genome nevertheless give the right answer?

4. Closely related to ideas about proximate vs. ultimate explanations and supervenience is the idea of "screening off". This is the idea that, if A (an event or state of affairs) causes B and B causes C, B "screens off" C from A. That is, in order to predict C, I only need to know B. Knowing A doesn't help predict C, because A only affects C indirectly, via B. But that doesn't mean one never wants to know anything about A. For instance, once I've explained C by appeal to B, it's natural to ask "What explains B?", which leads me to work backwards along the causal chain. I leave it to you to think about how this idea relates to the ideas discussed in class today.

5. We didn't talk about Gould today, but I hope it's clear that objections to Darwinism as wrong or incomplete tend to run to type. The claim that development imparts some kind of directionality to adaptive evolution (or so strongly constrains it that the constraint is really the main story) resonates with much earlier ideas about orthogenesis and the "bauplan".

6. Objections to Darwinism and the Modern Synthesis also run to type in caricaturing Darwinians as believing that natural selection is all powerful and entirely unconstrained. My experience is that evolutionary biologists focus so much on natural selection not because they think it can produce any adaptation, but because they think it's the only thing that can produce what adaptations exist. And so if you're interested in explaining adaptation, you have no choice but to be interested in natural selection--it's the only game in town. That doesn't mean you don't care about constraints on adaptation. But those constraints aren't interesting in and of themselves, they're only interesting by virtue of their effects on adaptive evolution. Of course, scientists prefer to argue about objective empirical claims rather than subjective claims about what's "interesting". So if you want to argue against the Modern Synthesis, you can't accuse its proponents of making uninteresting claims, you have to accuse them of making false claims.

No comments: