Sober’s view of evolution

In Discussion on February 1, 2010 by bcnjake

This is admittedly rough thinking in the hopes that I can flesh an idea out via some groupthink.  Since last week’s reading, I’ve been bothered by something that Sober said in his account of evolution.  Specifically, he said that a vague definition of evolution would suffice, since the standard definition (changes in gene frequencies in a population) fails to account for [a] changing genotype frequencies without corresponding gene frequencies changes and [b] evolution taking place outside of the cellular nucleus, which where a gene is located. (“What is Evolutionary Theory”, p. 4-5)  I understand Sober’s criticisms, but fail to see why a more cogent account of evolution cannot be offered.

It seems to me that a better definition would be that evolution is a change in a population’s biological information frequency.  By biological information frequency, I mean (very) roughly any information that is [a] encoded biologically and [b] capable of being passed on through some inheritance mechanism.  Thinking of evolution this way would have some distinct advantages.  First, it accounts for both of Sober’s criticisms of the standard account, in that genotypes and mitochondrial DNA are biological information, even if they are not genetic information in the strictest sense.  Second, this approach would account for not only genetic information, but also epigenetic information and memetic information, two areas that ought to be thought of as evolving.

The obvious objection, at least to me, is that memetic and epigenetic information are either non-biological, lack heritability mechanisms, or both.  The first concern doesn’t strike me as problematic, since epigenetics clearly have a biological component and (besides being encoded biologically) memes could not exist without lower-level biological support.  Biological information must not only exist for memes to propagate, they must be of a certain type of information (i.e. humans can have memes, but earthworms cannot).

As to the second concern, memes are clearly heritable, but this concern might be problematic in the case of epigenetics.  My epigenetic knowledge is limited, but my understanding is that environmental factors trigger latent genetic instructions.  But environmental factors are not heritable the way genetics and memetics are.  How does one get around this?  Would it be fruitful to assert that the potential for epigenetic changes is heritable, or would another avenue be better?  Is there a good reason to not count epigenetic information as evolving?  Is my definition fatally flawed?

As I said, these are all very rough thoughts; more thinking out loud in the hope that someone has a similar inclination and/or helpful ideas.  But I think the basic idea is right.  Evolution can’t only deal with nucleic gene frequencies – there’s more to it.


11 Responses to “Sober’s view of evolution”

  1. Hi Jake,

    I like your post. I do worry, however, that talk of “information” might end up making things worse. Here: is a paper I read a while back by Griffiths concrning the difficulties with thinking about genetic information (not to mention the other types of biological information that are thrown around in the literature). I’m not sure exactly how useful it is for your thoughts above, but it might be worth looking over briefly.

    Still, I think it is certainly a step in the right direction to broaden the definition of evolution to include various forms of inheritance and contribution to development. That is, I certainly agree that there is more to evolution than nucleic gene frequencies. I’m mostly worried that understanding what counts as information – moreover, biologically encoded information – might be rather difficult to work out.

    In the end, though, I think you are on the right track in hoping to expand the definition to include other types of inheritance and biological information (since I lack a better term). It might become difficult to use the definition, however, since we lose the nice quantitative evaluation provided by something like change in gene frequency. That is, unless we can find a way of understanding how to quanitify information and we can understand how it can be inherited. It seems more likely that we could perhaps measure the frequencies of the THINGS that provide/carry infomation – i.e. those things that are actually inherited by the next population. This would perhaps allow us to preserve the nice quantitative test for whether evolution has occured, as well as expand the set of inherited entities beyond just genes.

    Overall, I think it is important to try and understand why we would want to set out looking for a precise definition of evolution (as Sober seems to suggest). I am tempted to think that it, like numerous other theories and disciplines simply does not have a well defined domain without which it will be difficult to provide a precise definition.

  2. Jake and Collin,

    I am interested in the information metaphor. I think the issue that makes it viable is that one must explicitly consider both the genetic information and the other information IN THE SYSTEM. I would include environmental information, especially that which is inherent in inter-specific links and co-evolutionary phenomena. I’d place the Odling-Smee niche construction in this category. Thus, genetic code-as-information is necessary but not sufficient to describe the open system that is evolution.

    I am not a fan of the trend to pushing social and cultural information into the process of evolution in the biotic domain. That is, I cringe at memes and the like.

  3. I’m with Randy on memes. I think that if what were talking about is biological evolution then we have to deal with biological heritability, not cultural reproduction.

    I’m interested in looking at the paper Collin referenced, as I’m fairly sympathetic to information talk (indeed, tend to think that it’s not merely a metaphor).

  4. Oops! Now that I look back over the Griffiths paper a little more closely, it probably just says exactly what we all agree on anyway: genetic informatoin should not be separated from other types of biological information. However, it still lays some interesting terrain for the information metaphor (e.g. see Maynard Smith’s view). I still wonder about the definition of information and if using it and a more generic inheritance requirement will really be able to avoid any counterexamples (I’d have to think about it some more). For one thing, I’m not sure that the other causes (sorry Andre) of evolution necessarily require inheritance; e.g. suppose a mutation or recombination occurs. Would we then say that no evolution had occured until the change had been inherited? It seems that the change in gene frequency is sufficient in this case to say that the population has evolved, yet no inheritance has occured (though it likely will later on). Also, my worries about a quantitative test above still trouble me. Anyhow, for those of you (like me) who don’t have time to read the whole Griffiths piece, but are interested, I’ve pasted a short summary I wrote on the paper for Zac’s seminar below (I’ve left off the critique, although it is surely devestating):

    In “Genetic Information: A Metaphor In Search of a Theory,” Paul E. Griffiths argues that the proposal that development is the expression of genetic information is misleading about explanation in molecular biology (Griffiths, 2001). To critique the proposal, Griffiths looks at various definitions of information and argues that all the defensible notions do not imply a difference between genetic and non-genetic developmental causes.

    According to teleosemantic theories of information, things refer to whatever evolution designed them to represent (Millikan, 1984). That is, things derive their meaning from their biological functions that are provided by natural selection. John Maynard Smith proposes to use teleosemantic theories of information to defend the information talk used to distinguish between genetic and non-genetic causes in developmental biology (Maynard Smith, 2000). The basic idea is that genes are intentionally directed onto their effects. Genes’ biological function is to produce the effects for which they were selected. Thus, according to teleosemantic theories of information, genes provide intentional information about their effects. This is supposed to set genes apart from other, non-genetic, causal factors in development. Non-genetic factors, supposedly, do not provide biological information – they are merely material causes. In short, Maynard Smith’s proposal is that teleosemantic theories of intentional information entail that genetic causes provide intentional information and non-genetic causes do not. Therefore, because they are the sole sources of intentional information, the characterization of genes as special types of developmental causes is justified.

    In opposition to Maynard Smith’s proposal, Griffiths defends what he calls the parity thesis: ‘Any defensible definition of information in developmental biology is equally applicable to genetic and non-genetic causal factors in development’ (Griffiths, 2001). In response to the teleosemantic approach, Griffiths argues that it is perfectly consistent with the parity thesis since non-genetic developmental causes will likewise carry intentional information under that approach. This is because several non-genetic causes also have naturally selected for biological functions, which in turn implies that, under teleosemantic theory, they too have intentional information. More precisely, Griffiths’ argument runs as follows. According to teleosemantic theory, things refer to their naturally selected biological functions. Thus, any mechanism that has a naturally selected biological function will have intentional information. Moreover, there are several non-genetic mechanisms in development that have naturally selected biological functions. Therefore, there are several non-genetic mechanisms in development that have intentional information – which implies that having intentional information cannot be what distinguishes genetic from non-genetic causes in development.

    Griffiths provides several examples in support of his claim that non-genetic mechanisms have biological functions. First, the DNA methylation inheritance system attaches additional chemical groups to future generations’ DNA. These additional chemical groups block the transcription of the genes to which they are attached and thereby influence the development of organisms. They are, however, inherited via an inheritance system that runs parallel to DNA inheritance mechanisms (what Griffiths refers to as an epigenetic inheritance system). A second set of examples comes from developmental systems thinking. Various microorganisms, bee colony cultural structures, and parasite-host associations are passed on to future generations via non-genetic inheritance mechanisms. All of these examples cite non-genetic developmental causes, and yet, these mechanisms are likely to have adaptive explanations. In other words, because these non-genetic factors are inherited and persist long enough for natural selection to act on them, they are likely to have naturally selected biological functions. Consequently, under the teleosemantic theory they will have intentional information. Therefore, teleosemantic theories of information obey Griffiths’ parity thesis since both genetic and non-genetic causes will have intentional information.

  5. All,

    Thanks for the helpful commentary. I’m in the process of formulating a response, but I wanted to note one think that surprised me. I didn’t expect such a harsh reaction to the idea that memes should included in an account of evolution. Including memes seems reasonable to me (and possibly helpful – rather than say that economics evolve or physics evolves or fashion evolves, we can reduce those ideas to evolving memes). What are your collective arguments against their inclusion? Anyone else reading this who thinks memes ought to be included have an argument as to why?

  6. “For one thing, I’m not sure that the other causes (sorry Andre) of evolution necessarily require inheritance; e.g. suppose a mutation or recombination occurs. Would we then say that no evolution had occured until the change had been inherited? ”

    I say, “Yes”, Collin, because mutation is a form of variation. Until it has been selected (for) and retained in the gene pool, there is no impact on the population — future generations, that is.

  7. Hi Randy. Thanks for the response. Its just that thought that makes me wonder. There certainly is an impact on the population before selection occurs. It is true that selection hasn’t impacted what “shows up” in future generations, but the gene frequency has already changed – i.e. the mutation has already had an impact on the make up of the population, then selection changes things still further. If we restrict the term evolution merely to things that have been selected for, then it seems like natural selection is the only cause of evolution, which is incorrect (selection isnt necessary for evolution, only for evolution by natural selection). It is true that natural selection requires variation, but mutation etc. I don’t think require selection to say that they have caused the population to evolve (e.g. not if we talk about mere changes in gene frequency). The population is certainly different even without selection (but I surely don’t want to understate the importance of NSs impact). What do you think?

  8. Collin, I am in agreement with your points, but I always come back to the retention issue. For organisms that have mutations (small numbers), there is a small change in the distribution of genotypes and, perhaps, phenotypes in the extant population. But, if that mutation causes reproductive isolation for that organism or the mutation disappears by chance through sexual reproduction, then the mutation was effectively a non-event in evolution. I would change my argument if the organism didn’t reproduce sexually, of course.

    I think this is an element of the Wright-Fisher debate that lasted 3 decades.

  9. A generalized informational evolution would be bad because you’d have to nail down what you mean by “information.” Remember Sober’s example of fast parents cause better nutrition for offspring causes fast offspring? There is inheritance here, but no genetic inheritance. If you want to talk about informational inheritance, you’d have to figure out why no information is being inherited here, or grant that this is a type of evolution we should be interested in. I’m sympathetic to non-genetic evolution, though, but it’s best to talk about each type of evolution individually rather than some generalized informational evolution.

    As for epigenetics, there is certainly some inheritance, but it’s not clear to me that epigenes(?) are stable enough to be acted on by natural selection. I would think they would tend to “mutate” after just a few generations.

    I’m sympathetic to the notion of memetic evolution, but it’s certainly problematic. One main problem is identifying memes. Two people are self-identified as Christian – does that mean they have the same meme, even if one is a Southern Baptist and the other a Roman Catholic? (And even finer grained differences in beliefs than that.) Memes are not as atomic as genes or even phenotypic traits. I’m also fairly confident that humans tend to do more selection of memes than nature does, although that’s an empirical question. That’s also not enough to rule out memetic evolution, since human filtering would be a selection process. I’m sure there are more objections than that to memetics, so I’m curious what objections Josh and Randy have.

  10. Dan, you have a lot of interesting points. When I have time tomorrow, I’ll respond to them more fully, but one thing that jumps out at me as interesting (especially in light of the Dawkins piece and the debate that it stirred). You mentioned that humans do more selection for memes than nature does. Does this mean that human selection is somehow occurring “outside” of nature? If so, why would you think that? Are you suggesting that humans are intentionally selecting memes (which is not an unfair definition of artificial selection)? If so, I think that you’re largely wrong. Think of everything from language, which is largely memetic and not intentionally selected for (at least your mother tongue), to a tune in your head that you can’t get out and, given the chance, would excise as quickly as possible. If that’s not what you mean, what do you mean? Like you, I’m waiting for Josh and Randy’s elucidations with baited breath.

    I agree that it’s a very difficult (and interesting) task to identify memes, but I don’t think it’s impossible. My understanding of memetics is that larger structures like Christianity or Buddhism aren’t memes, because they’re made of up so many different ideas (I believe the term for these structures is “memeplex”), but I’m not as versed in the literature as I’d like to be.

    But to your largest point about “informational evolution”, I again agree that “information” needs to be nailed down (hence part of the motivation for this post), but the advantage of doing so is enormous. I don’t think, though, that we should treat types of evolution as separate things. since I’m inclined to argue that biological evolution IS informational evolution.

    Well, I suppose that I responded now (albeit in a stream of consciousness sort of way), so strike my first sentence. But still, plenty to discuss.

  11. I’ve thought of other possible reasons why memetics is not being accepted. I think the main one is not that it is false, but that it’s a solution in search of a problem. What can memetics do that isn’t already being done by theories in the relevant fields?

    Jake said:

    “You mentioned that humans do more selection for memes than nature does. Does this mean that human selection is somehow occurring “outside” of nature? If so, why would you think that? Are you suggesting that humans are intentionally selecting memes (which is not an unfair definition of artificial selection)?”

    Sorry for speaking so imprecisely. I’ll flesh this out more. I think one of the main reasons memetics is so widely objected to is that Dawkins seemed to present memes as “mind viruses.” But most memes promote human survival, otherwise we would never have evolved the capacity for culture in the first place. (Memes are like plasmids, not viruses.) Furthermore, the main barrier to the spread of memes is not the death of their carriers, but human cognitive faculties – our bullshit detectors. Some memes are beliefs. Of those memes that are beliefs, most of them are not just plausible nonsense (or even plausible nonsense that just happens to promote the survival of their carriers or promote the spread of the meme) but are actually *knowledge*. So a big chunk of the field memetics attempts to explain is already occupied by epistemology.

    For non-doxastic memes, we humans probably have other ways of screening out memes that don’t do what we want. This doesn’t rule out evolution, of course, it just imposes some selection forces. But I think this is a point that “memeticists” tend to underplay.

    You bring up an example of a contagious song. I think this highlights one of the biggest problems with memetics – just what do you mean by “meme”? They have to be 1) culturally inherited, 2) stable enough to be acted on by natural selection, and 3) identifiable by field scientists. (How do you know if two individuals have the *same* meme?) It would probably also help if memes constitute a natural kind. There’s enough differences between beliefs and other culturally transmitted objects (like contagious songs) that you might not want to use the same theory to deal with them all.

    I think that is my biggest objection to a generalized informational theory of evolution – “information” is not a natural kind. It would probably be better to develop a separate theory for each kind of inheritable trait you want to deal with. There would be some core of theory that would analogically apply to all the various forms of evolution, but you’d need a lot of auxiliary hypotheses to deal with the differences between the fields. (You might even want to separate mDNA evolution from nuclear DNA evolution. Our interests and our mitochondria’s interests might not always coincide – we’re just their environment.) This is the approach “Universal Darwinism” is taking, according to this week’s reading, but it doesn’t look like it’s having too much success.

    I rather like the neo-Schumpeterian approach. It has a rigorously defined gene-analogue (the routines used in large firms and other institutions). It has a generalized core of evolutionary theory, but a lot of auxiliary hypothesis to tie it to the field. It is also not a “solution in search of a problem,” and looks like it’s had some empirical success. This is probably a better approach than a more general “memetics.”

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: