Reductionism Reconsidered: A Dialogue between Ursula Goodenough and George Ellis

From: Ursula Goodenough

I lack the training to respond to Geoge Ellis's anti-reductionism arguments with respect to physics, but I don't believe his arguments with respect to biology should stand unchallenged. He makes two:

"(c) The central theme of evolutionary biology is the development of particular DNA codings through an evolutionary process, in order to adapt an organism to its ecological niche. This is a classical case of top down action from the environment to detailed biological microstructure - in essence the DNA coding incorporates an image of the ecosystem environment in which the organism functions. There is no way you could ever predict this coding on the basis of biochemistry or microphysics alone. The environment acts in a top-down way in the biological hierarchical structuring and so [along with other causal factors] fixes that coding."

The central theme of evolutionary biology is not the development of particular "DNA codings" *in order to* adapt an organism to its ecological niche. The "DNA codings" mutate randomly, generating instructions for making new protein shapes and hence organisms with novel phenotypes. Organisms with phenotypes that give them a margin to better survive/produce viable offspring in the ecological niche then transmit the "codings" responsible for those phenotypes into future generations in the lineage. The DNA coding doesn't incorporate an image of the environment; the environment selects a coding that endows the organism with a fitness advantage in that environment. No one that I'm aware of argues that you can predict the coding on the basis of biochemistry or microphysics alone; Darwin understood the role of the environment in the evolutionary process without any understanding of either. If selection for biological adaptation qualifies as a "top-down action," then isn't any selective act a "top-down" action? When the walls of a container dictate how much water can be poured into the container, do they not also perform a top-down action? If so, then I don't understand how this is a useful concept.

"(d) The reading of DNA by an organism in the developmental process is not a mechanistic process but is context dependent. For example recent research on genes and various hereditary diseases shows that existence of the gene for such diseases in the organism is not a sufficient cause for the disease to in fact occur."

I don't know of any biologist who would disagree with the statement that the reading of DNA is context-dependent -- this was the central insight of Jacob and Monod for the bacterial lac operon in the 1950s. Development, in particular, is context-dependent all the way down, with what happens before having everything to do with what happens next (for an excellent for-the-layperson book describing this, I highly recommend Enrico Coen's "The Art of Genes", Oxford, 1999). But this doesn't mean that the reading is not a mechanistic process. It is *also* a mechanistic process, one that is regulated by context. The fact that I can operate my car at different speeds in different contexts doesn't mean that my car doesn't operate by a mechanistic process.

The disease-gene argument depends on the nature of the gene and on the rest of the genome and on the environment. A person may carry additional genes that suppress/modify the effects of the original "disease gene" such that it doesn't lead to disease, whereas another person may not carry such suppressors and display the disease. Or, the "disease gene" may only participate in the chain of events leading to disease under certain environmental conditions. Or, the "disease gene" may increase the likelihood of incurring disease via some other agency. For example, many of the mutant genes implicated in cancer encode defective versions of enzymes that normally repair lesions in DNA. With enhanced levels of misrepair, the carrier is far more likely to incur mutations in genes that suppress the development of malignancy, but this is a statistical event, and some folks get lucky.


From: George Ellis

Hi Ursula

Thanks for your comments and questions on my posting. In response, please see the interpolations below.

George

> The central theme of evolutionary biology is not the development

> of particular "DNA codings" *in order to* adapt an organism to its
> ecological niche. The "DNA codings" mutate randomly, generating
> instructions for making new protein shapes and hence organisms with
> novel phenotypes. Organisms with phenotypes that give them a margin
> to better survive/produce viable offspring in the ecological niche
> then transmit the "codings" responsible for those phenotypes into
> future generations in the lineage. The DNA coding doesn't
> incorporate an image of the environment; the environment selects a
> coding that endows the organism with a fitness advantage in that
> environment.

In my view this is just semantics: we are talking about the same mechanism, using somewhat different language. In my view the outcome is as I stated: you can call it poetic language if you like [and it's the kind of language you use repeatedly in your book], but the DNA coding is what it is inter alia because of the nature of the environment in which the organism lives. If the environment were different the coding would be different. In that sense it incorporates an image of the environment. It is an active image, in the sense that the organism is adapted to the way the environment actually is and responds to that situation by its behaviour. If you don't want to agree that it is legitimate to say that this means the organism has in some sense internalised an image of the environment, so be it

> No one that I'm aware of argues that you can predict the coding on
> the basis of biochemistry or microphysics alone; Darwin understood
> the role of the environment in the evolutionary process without
> any understanding of either.

I agree. I'm stating that this is significant in terms of understanding the meaning of the concept of reductionism. It is not always emphasized in the way done here.

> If selection for biological adaptation qualifies as a "top-down
> action," then isn't any selective act a "top-down" action? When
> the walls of a container dictate how much water can be poured into
> the container, do they not also perform a top-down action?

yes it is. I agree. What this means is that the idea that you can tell by microphysics alone what will happen is not true. You also need to define the macro environment, which plays a major role in what happens at the micro level and hence at the macro level. In the case of the water in the container this is simple enough but important to recognise; in the case of the human brain it may be -
indeed almost certainly is - a dominant feature in relating what happens in regard to higher level descriptions and concepts, and hence also at the micro level..

> If so, then I don't understand how this is a useful concept.

It means you must be very careful in talking about determinism on the basis of bottom-up (mechanistic) action. If the macro situation dominates what happens, then macro levels of meaning come into play and control what happens [in a way consistent with the micro physics].

> I don't know of any biologist who would disagree with the statement that
> the reading of DNA is context-dependent -- this was the central insight of
> Jacob and Monod for the bacterial lac operon in the 1950s. Development, in
> particular, is context-dependent all the way down, with what happens before
> having everything to do with what happens next (for an excellent
> for-the-layperson book describing this, I highly recommend Enrico Coen's
> "The Art of Genes", Oxford, 1999). But this doesn't mean that the reading
> is not a mechanistic process. It is *also* a mechanistic process, one that
> is regulated by context. The fact that I can operate my car at different
> speeds in different contexts doesn't mean that my car doesn't operate by a
> mechanistic process.

I have not denied any of this, or that it is mechanistic at the micro level - insofar as any quantum system is mechanistic, which is to a rather limited extent. What the context dependence means is that you can't in any serious sense predict what will happen if you insist on a micro description alone. There are multiple levels of description and meaning in action. The higher levels of meaning cannot be *reduced* to a description solely in terms of lower level concepts. That is the key point. They are compatible with the lower levels, of course. But they in a serious sense determine what actually happens. The lower levels place remarkably little constraint on what happens at the higher levels [e.g. I can type any text I like on this machine; the electrons in the machine allow this], where independent levels of causality and phenomenology come into play.

An example is that my computer is coding this whole conversation in binary form. You cannot understand the meaning of those binary digits without translating to the higher level of meaning in which they become words on the screen. Yes the binary digits in one sense give a complete description of what is going on; however in fact to give someone that list of digits would totally obscure what they represent and how the computer will deal with them. They would not know if it represented a spread-sheet, a picture, or movie, or whatever. The software loaded to read the binary digits and act on them constitutes the internal environment that gives the micro description meaning and defines the actions that will be performed on it by the computer [such as displaying it on a screen to give an image of text]. The lower level description is congruent with this higher level of meaning but completely fails to disclose it.

> The disease-gene argument depends on the nature of the gene and on the rest
> of the genome and on the environment. A person may carry additional genes
> that suppress/modify the effects of the original "disease gene" such that
> it doesn't lead to disease, whereas another person may not carry such
> suppressors and display the disease. Or, the "disease gene" may only
> participate in the chain of events leading to disease under certain
> environmental conditions. Or, the "disease gene" may increase the
> likelihood of incurring disease via some other agency. For example, many of
> the mutant genes implicated in cancer encode defective versions of enzymes
> that normally repair lesions in DNA. With enhanced levels of misrepair,
> the carrier is far more likely to incur mutations in genes that suppress
> the development of malignancy, but this is a statistical event, and some
> folks get lucky.

Right. So 'mechanistic in principle' works out to 'not mechanistic in practice'. That is the point. The macro situations you discuss determine what happens, not specific micro features by themselves [which do work mechanistically but in a context of larger meaning that largely determines the outcome]. You cannot easily say that the result is mechanistically determined once you allow this, inter alia because the macro enviroment includes the result of conscious decisions [I will or will not seek medical treatment for the condition].

In summary, the idea many have of reductionism as implying macro events are completely causally determined by the laws acting at the micro level is simply not true, because it is also correct to say that what happens at the micro level is to a large degree determined by macro structure and what happens at the macro level. This makes a key difference in terms of one's understanding of what is going on. Yes the micro level is mechanistic - but it is mechanistic within the context of boundary conditions, macro constraints, and macro influences which in fact largely decide what the actual outcome is. This is what opens the way for the levels of meaning and order at higher levels to have a reality and causal efficacy despite their workings being based on largely deterministic mechanisms operating at the microlevel ('largely' because of quantum physics - it is my view that quantum indeterminism, whose ontological nature is a disputable issue, may play a significant role in the operation of the brain -
an unpopular position. What is indisputable is that neuroscientists who ignore quantum physics are discounting the essential foundations of the subject in which they work, and by doing so they may be missing essential elements in the way the brain functions.)

I hope this helps. It is an interesting theme.


From: Ursula Goodenough

And responses to responses!

On Fri, 28 Jan 2000, Ellis, GFR, George, Prof. wrote:

>
> in my view this is just semantics: we are talking about the same
> mechanism, using somewhat different language. In my view the outcome
> is as I stated: you can call it poetic language if you like [and it's
> the kind of language you use repeatedly in your book], but the DNA
> coding is what it is inter alia because of the nature of the
> environment in which the organism lives.If the environment were
> different the coding would be different. In that sense it
> incorporates an image of the environment. It is an active image, in
> the sense that the organism is adapted to the way the environment
> actually is and responds to that situation by its behaviour. If you
> don't want to agree that it is legitimate to say that this means the
> organism has in some sense internalised an image of the environment,
> so be it [I believe systems theorists will be happy with this
> language]. That sentence can be omitted.
>

It amounts to the same thing in terms of the final result -- although to my mind a more appropriate framing would be to state that the genome encodes pathways etc. that *anticipate* a particular environment. Where I had problem with your "semantics" is that it reads, to my ear, as Lamarckian, that the organism somehow perceives the environment and modifies its genome accordingly, rather than mutating randomly and then the environment selects the more adapted creature.

> > No one that I'm aware of argues that you can predict the coding on
> > the basis of biochemistry or microphysics alone; Darwin understood
> > the role of the environment in the evolutionary process without
> > any understanding of either.
>
> I agree. I'm stating that this is significant in terms of
> understanding the meaning of the concept of reductionism. It is not
> always emphasized in the way done here.

But this isn't fair to reductionism. As I try to emphasize in Sacred Depths, we "evil" reductionists are not in fact oblivious for a second that the reduced understanding of what's going on doesn't have higher-order contexts in which it's embedded. I'm making the claim that you are unfairly pillorizing reductionism. Sure anybody can go into Dawkins and pull out a sentence that makes it sound like he thinks that the reduced explanation carries the day, but that's no more fair than saying that someone describing the wonders of quarks is saying that quarks are the whole picture.

I need to know who you are reading, George, who claims that you can tell by microphysics alone what will happen. Who is arguing determinism? I'm stumped.

Molecular biologists etc. are totally aware of this [the relationship between macroevents and micro manifestations of genes].

>
> [*Addition*: In summary, the idea many have of reductionism as
> implying macro events re completely causally determined by the laws
> acting at the micro level is simply not true, because it is also
> correct to say that what happens at the micro level is to a large
> degree determined by macro structure and what happens at the macro
> level. This makes a key difference in terms of one's understanding
> of what is going on. Yes the micro level is mechanistic - but it is
> mechanistic within the context of boundary conditions, macro
> constraints, and macro influences which in fact largely decide what
> the actual outcome is. This is what opens the way for the levels of
> meaning and order at higher levels to have a reality and causal
> efficacy despite their workings being based on largely deterministic
> mechanisms operating at the microlevel (`largely' because of quantum
> physics - it is my view that quantum indeterminism, whose
> ontological nature is a disputable issue, may play a significant
> role in the operation of the brain - an unpopular position.

That's an assessment I can agree with!

> What is
> indisputable is that neuroscientists who ignore quantum physics are
> discounting the essential foundations of the subject in which they
> work, and by doing so they may be missing essential elements in the
> way the brain functions.) ].
>

Well it's sure not obvious to me how this would work, but you're the physicist and not me!

Join Metanexus Today

Metanexus fosters a growing international network of individuals and groups exploring the dynamic interface between cosmos, nature and culture. Membership is open to all. Join Now!