Thursday, February 04, 2010

Jerry Fodor and natural selection

Philosopher and cognitive scientist, Jerry Fodor has co-written an article with Massimo Piattelli-Palmarini in New Scientist this week, which argues that natural selection "overestimates the contribution the environment makes in shaping the phenotype of a species and correspondingly underestimates the effects of endogenous variables."

Fodor and Piattelli-Palmarini claim that more attention should be devoted to "non-environmental constraints on trait transmission. [These] include constraints imposed 'from below' by physics and chemistry, that is, from molecular interactions upwards, through genes, chromosomes, cells, tissues and organisms. And constraints imposed 'from above' by universal principles of phenotypic form and self-organisation - that is, through the minimum energy expenditure, shortest paths, optimal packing and so on, down to the morphology and structure of organisms."

Unfortunately, Fodor and Piattelli-Palmarini make two crucial conceptual errors:

(i) They conflate natural selection with adaptation to an environment.

(ii) They conflate evolutionary fitness with adaptivity to an environment.

In fact, natural selection and evolutionary fitness can be defined in abstraction from adaptation to an environment. As John Barrow puts it, natural selection "has just three requirements:

(1) The existence of variations among the members of a population. These can be in structure, in function, or in behaviour.

(2) The likelihood of survival, or of reproduction, depends upon those variations.

(3) A means of inheriting characteristics must exist, so that there is some correlation between the nature of parents and their offspring. Those variations that contribute to the likelihood of the parents' survival will thus most probably be inherited.

It should be stressed that under these conditions evolution is not an option. If any population has these properties then it must evolve," (The Artful Universe, p21).

There is no need to refer to the existence of an environment, or adaptation to an environment, in the definition of natural selection. As Lee Smolin points out, "at this formal level, concepts like 'survival of the fittest' or 'competition for resources' play no role...What is responsible for that variation, and what goes into the differential survival rates, is not relevant for how the basic mechanisms of natural selection work," (The Life of the Cosmos, p104).

As a demonstration of this, Smolin's theory of cosmological natural selection requires no environment for the evolution of its hypothetical population of universes.

In fact, the conceptual errors made by Fodor and Piattelli-Palmarini are probably shared by many evolutionary biologists, who probably do conflate natural selection with adaptation to an environment. The endogeneous constraints which Fodor and Piattelli-Palmarini refer to may well play an important role in biological evolution, and they might well demonstrate that biological evolution involves far more than adaptation to an environment. Nevertheless, such endogeneous constraints are consistent with evolution by natural selection.

Equally, this doesn't entail that natural selection is the only process operating within biological evolution. Far from it, it seems that horizontal gene transfer plays a huge role in microbial evolution, and processes such as genetic drift also play a role under specific circumstances. The fact remains, however, that the concept of natural selection is neutral with respect to the importance of the environment.


Jasonee said...

An insertion of DNA via horizontal gene transfer is a bit like a large mutation. It is also like sex in that DNA from two organisms is combined. After a horizontal gene transfer, the organism is still subject to natural selection. So it is not something extra to natural selection.

Jasonee said...

Although you can describe natural selection of a population without explicitly defining what constitutes its environment, you are implicitly admitting the existence of an environment, even if that environment is simply an abstract set of rules. A replicator is just a pattern that replicates, but there must be some framework within which you can meaningfully use the word replication. If this framework is just a computer program or even just a set of axioms and rules of logic then this framework constitutes the environment.
If you have an axiomatic system and then choose to create an instance of this system, then the total state of that system also becomes part of the environment.

Gordon McCabe said...

There's always something fixed, yes, against which the evolution takes place. In the case of a population of universes, evolving by natural selection, the basic laws of physics might be taken as fixed, with the parameters in those laws subject to evolution.

In the case of biological evolution, however, the fixed background might be said to consist of the genetic code, not the ecological environment. Such environments can obviously vary from place to place and from time to time, and against the fixed background of the genetic code, these different environments determine different survival rates for different genomes.

RichardW said...

Hi Gordon. I followed your link in the following discussion:

You say above that you're going to define "natural selection", but then you don't actually do so. Instead, you give three requirements for evolution to occur, which is not the same thing. We need to distinguish between the expression "natural selection" and the theory of evolution by natural selection. I would say that "natural selection" just means "differential reproductive success", or something similar.

But what you've given is not a even a basic principle of evolutionary theory, because not all of your three requirements are necessary for evolution. If you suspend number 2 (differential reproductive success) you can still have neutral evolution.
So you haven't actually said what these are requirements for. You might say that they are requirements for "adaptive" evolution, but then you've brought the concept of adaptation into it, which you don't want to do.

By the way, I don't think I've made a good job of defining "adaptation" in that other discussion. It's a tricky problem, and I'm still thinking about it. If you can find a way of explaining the significance of differential reproductive success in evolution without using the words adaptation or adaptive, I'd be pleased to hear it.

RichardW said...

On reflection, I don't think you can explain evolution properly without introducing the idea of adaptation or functionality (the two are closely related). I'll quote part of my latest post in that other discussion. (I'd rather continue the discussion here, as the standard of thinking seems higher!)

>>[Consider] a specific example. Suppose we have a mixed population of brown and white bears living in a snow-covered environment, the colour being a heritable trait. Suppose that the white bears survive and reproduce better than the brown because they are better camouflaged against the snow, so the proportion of white bears increases over time. Then it seems reasonable to say that the population is adapting to its environment.

Why does this seem reasonable? It's because the population is becoming better at being camouflaged in the white environment. More generally, the population is acquiring a functional trait. It's the existence of such functional traits that needs explaining. People don't ask "how did organisms get to have such reproductive success?". They ask "how did organisms get to have such functional traits as camouflaged fur and eyes?".

So I would now define adaptation of a population (to an environment) as the development and spreading of traits that are functional (in that environment). The white colour of fur in the example above is for camouflage (i.e. has the function of camouflage). Similarly, eyes are for seeing. In an ultimate evolutionary sense, eyes are important because they contribute to reproductive success. But we don't say that eyes are for reproducing. It's more meaningful to say that eyes are for seeing.<<

The significance of natural selection is that it explains the existence of such traits as camouflaged fur and eyes. But what is the property that makes these traits interesting? I would say that it's their functionality or adaptedness, and particularly in more complex cases, their functional (adaptive) complexity.

Gordon McCabe said...

Nicely argued Richard.

You say, "If you can find a way of explaining the significance of differential reproductive success in evolution without using the words adaptation or adaptive, I'd be pleased to hear it."

Well, suppose that a mutation occurs in a sub-population of some species, which increases the efficiency with which free radicals, the by-products of metabolism, are scavenged from cells. Suppose that by so doing, the average lifetime of these animals is lengthened.

All other things being equal, this sub-population will come to statistically dominate the species, yet there is no adaptation to the environment involved.

RichardW said...

Hi Gordon. Thanks for replying.

In my second post I defined adaptation in terms of "functional traits", and I'll assume you agree that the population in your example is developing a functional trait. We can think about functionality without necessarily mentioning an environment, which is why in my definition I only mentioned the environment parenthetically. But I would say that generally (if not always) functionality is relative to an environment. What's functional in one environment is not necessarily functional in a different environment. So developing a functional trait is adapting to an environment where that trait is functional.

This is more obvious in some cases than others. In my example above, white fur only functions as camouflage in a white environment, so it seems very natural to say that the population is adapting to the white environment. I'm unfamiliar with the physics and biology of your example, but I assume the role of the environment is much less clear there. Perhaps it's difficult to imagine an alternative environment in which that specific trait isn't functional. But I expect such an environment is at least theoretically possible.

Even if one can think of cases where it's hard to see functionality as depending on the environment, I think it makes sense to use the word adaptation, since (a) most of the time the role of the environment is clear, and (b) even when it isn't we can still think of adaptation as meaning just development of functionality (without regard to the environment). If you think that cases of (b) are significant and that this is an unreasonable abuse of the word "adaptation", then by all means talk about functionality instead. I suggested in my last post that you could use either concept, since they're closely related.

I notice that your post didn't mention functionality either. But then you haven't really addressed the challenge that you quoted. What you've given is a description or account of an evolutionary scenario, rather than an explanation. If it's an explanation, what does it explain?

RichardW said...

By the way, I think my definition of "adaptation" needs further work. The mere spread of a functional trait is not sufficient to be considered adaptation, since it may have spread by neutral evolution, i.e. independently of its functionality. So I would now define adaptation of a population (to an environment) as the development and spreading of traits because they are functional (in that environment).

Note that this isn't a standard definition. I haven't seen another definition that I like. But all the definitions I've seen involve some property which is likely to depend on the environment, be it functionality, fitness, reproductive success, etc.