Skip to content

More on reductionism

I am presently teaching in a history subject dealing with ideas of nature, and I notice that the historians we are using often refer to a distinction between reductionism and holism. The former is the Bad Old Science (“we murder to dissect”) and the latter is the New Improved Science. This is something often stated as if the issue were obviously resolvable. And it is, in my view, a complete myth.

In biological science, people often suggest that reductionism is more than a mistake: it is in fact a morally dangerous position. Genetic reductionism is supposed to be the explanation of everything including behaviour in terms of genes. Many people have attacked it, including critics of sociobiology. And it must be said that genes function as a magic molecule in many people’s minds. But the problem is not that we give reductionist accounts of traits, but that we do so in terms of single genes. The error here is single cause explanations, not that we look for the properties of causal parts.

To reduce a domain to another, in this case behaviour to biology, is a virtue in science. Nearly all progress in biology has been made by identifying what causes observed properties of organisms and ecologies in terms of the parts of the systems being investigated. Cell theory in the 19th century, genetics in the 20th, and biochemistry throughout have shown us why organisms develop, react and adapt the way they do. The problem is not that we look for explanations in the parts of organisms, but that we do it hamfistedly.

Reductionism is supposed to look only at the parts, according to proponents of “holism”, when according to them we should look at the entire system and how it interacts. But I am hard pressed to find any reductionist who ever denied this. Instead we get methodological decisions to break systems into parts for the purposes of tractability. You simply cannot identify all the variables in a complex system, so scientists in general will attempt to model systems in ways that deal with a few aspects of the system, in order to see how much can be explained that way.

Reductionists believe several things: one of the more important is that the behaviour of a whole is explicable in terms of the properties of the parts that comprise it. Without this we would not have physics, chemistry, cell biology, or any other general science. But focusing on the properties of the parts has never made sense unless we consider how the parts interact. If you know that a cell produces a protein, to understand how it functions in the organism and its environment, you need to consider how that protein is distributed, and how other cells take it up and process it.

Likewise, in general, a reductionist account has to consider the interactions of the parts with each other. The properties of a single electron, for example, can be stated on their own, but how electrons interact with other subatomic particles depends on the properties of those particles, and so a “holistic” approach is implicit in the modelling of the electron itself.

So I get very tired of the general charge that reductionism is unable to understand the system-level properties of the parts. This is simply a rhetorical trick.

I gave my view of “pizza reductionism” before.

8 Comments

  1. I remember reading Koestler’s criticisms of reductionism a number of years back and thinking how transparently obvious it was that his criticism was a strawman. The analogy I came up with is that according to the caracature of reductionism, a man sitting on an elephant is equivalent to an elephant sitting on a man.

  2. physicalist physicalist

    But I am hard pressed to find any reductionist who ever denied this.

    Amen.

  3. Wonderful. I will link this to my students next time I do my theory of science class.

  4. I wish you’d taken more care here to distinguish the different types of reductionism here, as you did in the pizza reductionism post.

    Few holists have ever had a problem with methodological reductionism–the idealizing of phenomenal reality into simplified models to tease out the explanatory role of the parts of the whole. (Even poor old Arthur Koestler never said that properties of parts were not explanatory.) What is objected to (and I’m sure you know this if you follow Rosenberg) is metaphysical reductionism–the claim that these highly simplified, highly idealized models are *more real* than descriptions at the phenomenal level. It’s one thing to reduce our scope on the way down; quite another to fail to expand it on the way up.

    Holists do not disdain part/whole explanation. They do ask us to not mistake the map for the territory, and they question why all explanations need to resolve to the same point of reference. Surely the reason we have sociology or psychology or fantasy baseball is not because the physics would too much time to work out. Even if in theory it’s possible to start from the big bang to explain why Miguel Cabrera is having a batting slump, it would be an odd way to go about it. Even more odd, to my mind, than invoking supernatural causes–even though the big bang is real and gods and demons aren’t–because at least the latter is in the appropriate explanatory realm.

    A final note: single-cause reductionism may not be the method good scientists use, but it’s all too often the way that scientific ideas get communicated to the laity, where they make the leap from methodological to metaphysical. It’s a real problem, not one to be solved by trashing “holists” for their common cause with scientists in making sure simple explanations are put into a proper context.

    • One cannot say everything at once, and I had a particular target. And I do reject the kind of metaphysical realism about models you mention, although I’m not sure that Rosenberg commits that fallacy though – what he seems to say is that the underlying reality is bosons and fermions, but that physics (the models) is incomplete.

      But you seem to make an argument from consequences: one is that we can’t do social (and other complex forms of explanation) science because it is “odd”. Perhaps it is odd because we aren’t Laplacean demons and must do science as fallible and limited epistemic agents (Rosenberg’s instrumentalism) but that we must take an ontic stance anyway. The other is that if we adopt a reductionist account, journalists will get it wrong. Journalists always get it wrong. Perhaps we should reform journalism…

      My point, which I stated explicitly, is that this is about rhetoric. Holism is not an alternative to reductionism, since no reductionism works without “holistic” context. In fact, holism and reductionism are styles of inference. One adopts a broader or narrower scope depending on what one wants to explain. This is not methodological, but a matter of interest.

      • Holism is not an alternative to reductionism, since no reductionism works without “holistic” context”

        True enough when we’re talking about methodological reduction/holism, which is what scientists generally have in mind.

        But when we’re talking about ontological reduction (causal reduction, mereological reduction, or physicalism) then holism is a genuine metaphysical alternative — an alternative that happens to be false, but an alternative nonetheless.

      • Perhaps I could have been more clear. I completely agree that holism is not an alternative to reductionism, nor is reductionism an alternative to holism. Since you don’t cite anyone I can’t be sure who you mean, but most critics of reductionism I have read are referring to the metaphysical variety, and would have no problem concurring that reduction is essential to building explanatory models.

        I think you are wrong about Rosenberg. From what I have read he has put forth that bosions and fermions are *more real* than toothache, the color red, a prison sentence, or an armistice. Perhaps that’s a discussion for another day.

        Again, I don’t know who you are reading. Maybe they are indeed railing against methodological reductionism. That would be foolish, though I find it unlikely.

Comments are closed.

Optimized by Optimole