Oh no! Am I a reductionist??

On the most informative levels of analysis for any given problem

Fields of study arranged by purity, according to the excellent xkcd.com.

Sometimes I think my career suffers from a kind of split personality disorder. For instance, as a scientist I tend to point out to my colleagues in philosophy that to say “that’s an empirical question” is a good thing, not a conversation stopper. Then again, I also find myself reminding my science colleagues that no matter how hard they try to ignore it, everything they do has philosophical underpinnings, including in terms of metaphysics, epistemology, and ethics.

Or consider the fact that I often argue, again with my science colleagues, that scientism—an attitude of pretty much unquestionable worship of science—is not a good thing. I ever co-edited a book about it! And yet, I turn around and I find myself accused of, you guessed it, scientism!, by a number of my philosophy colleagues (not to mention creationists, mysticists, and other assorted pseudo-thinkers).

I try to console myself that I must be doing something right if I manage to piss off both sides of frequently acrimonious debates, but I don’t know, maybe I’m just engaging in rationalizing wishful thinking.

Lately I’ve been able to add a third issue about which I seem to behave like an epistemic version of Schrödinger’s cat: I may or may not agree with a given position, until someone asks and then I settle down on a specific state. That issue is reductionism.

My dictionary defines reductionism as: “the practice of analyzing and describing a complex phenomenon in terms of phenomena that are held to represent a simpler or more fundamental level, especially when this is said to provide a sufficient explanation.”

Once again, as in the cases described above, sometimes I argue against reductionism, and sometimes in favor. Let me explain.

During much of my career as an evolutionary biologist I have been on record as an anti-reductionist. Specifically, I have been involved for decades in debates against colleagues who thought that the genetic-molecular level of analysis is the fundamental one in biology, and that everything else, from cell behaviors to ecosystem functioning, ultimately reduces to the properties of the molecules of inheritance. One of my major targets throughout that period has been Richard Dawkins, to give you just one name.

To be fair, Dawkins was relatively small potatoes compared with another one of my recurrent targets: Nobel winner physicist Steven Weinberg, who kept insisting that in principle everything in the world could be explained in terms of quantum mechanics. At a conference we attended together (Dawkins was there too!) I kept asking: exactly what principle? I never got an answer.

To my surprise, and definitely not delight, of late I’ve been increasingly accused of being a reductionist! How dare people hurling such an insult to me? Haven’t they read anything I’ve written over the last three decades?? Then again, the accusation came from a number of individuals belonging to the ever more widespread category of fuzzy thinkers, including a number of supporters of pseudoscience and even pseudophilosophy, like the current undisputed champion of panpsychism, Philip Goff.

At this point I better explain myself and try to convince you that one can coherently be both in favor and against reductionism, depending on what we mean by that term and under what circumstances we apply it.

To begin with, there are two fundamentally different kinds of reductionism: methodological and ontological. Roughly speaking, I am a partial and moderate methodological reductionist while at the same time I consider myself to be a thoroughgoing ontological reductionist. Also roughly speaking, the difference between the two is that the first one is a stance about epistemology, the second one about metaphysics. Unfortunately, a lot of people, including many scientists and even the occasional philosopher, confuse the two, thus generating countless misunderstandings.

Methodological reductionism attempts to understand things in terms of underlying components and phenomena. It is, as such, the epistemic foundation of much modern science. It also underlies a lot of common sense reasoning. Let’s say you have never seen a bicycle before and you are trying to understand how it works. It stands to reason that you’ll try to identify its main components and then figure out how they work together in order to make the bicycle go.

Basic components of a bicycle, useful information from a reductionist perspective.

One way to do this might be, for instance, to take out, or otherwise functionally incapacitate, one component at a time and see what happens. You take the handlebars out of commission, for instance, and you’ll discover that the bicycle can no longer be steered. You therefore infer that the handlebars have the function, in a normal bicycle, to steer. Similarly, you take off the brakes and discover that there is no longer any way to slow down the damn thing (oops!). You eliminate the saddle and suddenly find yourself having to stand in order to pedal the vehicle. And so on. This is how reductionism works.

The approach, of course, has limits. It becomes less informative, for instance, when the same component does multiple things, or where a number of components have to work together in order to accomplish a specific function. When faced with such complications you don’t throw away reductionism, you integrate it with an approach known as interactionism, which is basically reductionism applied to systems with functionally integrated parts.

The method I just described, incidentally, is also one of the major ways we have been learning what genes do. You knock one out (causing a mutation) and see what’s missing from the resulting organism. Just as in the case of the bicycle, the approach is complicated by the fact that genes interact with each other and with the environment to yield what biologists call phenotypes (i.e., morphologies, behaviors, and the like). Nevertheless, we have learned a huge amount by using this method ever since the rediscovery of Mendel’s work back in 1900.

What we have seen so far is what I’d call reasonable methodological reductionism. There is also a greedy variety, which is not as helpful. A greedy reductionist, say someone like Steven Weinberg, would not stop at the macroscopic components of the bicycle. He would point out that, at bottom, the bicycle is “made of” quantum states, and that therefore the ultimate description of the system is to be found in the laws of quantum mechanics.

But this completely confuses ontological and methodological reductionism. I do not question that bicycles, as well as everything else, are ultimately made of fundamental particles, or strings, or fields, or whatever else the most advanced physics tells us the universe is made of. Heck, the strangest suggestion I’ve heard in that department is that there are no “things” at all at the level of the very fabric of reality, only relations among points in space. Sure, okay, whatever. But that doesn’t help me at all if the task is to understand how bicycles work.

How does one distinguish, in principle, between reasonable and greedy reductionism? Other than by the fact that the two labels make it very clear what I think of either, I mean. Reasonable reductionism takes into account the most informative levels of analyses for a given problem and focuses only on those. Greedy reductionism, by contrast, always wants to get to the very bottom, regardless of how useful in practice such attitude turns out to be.

The most informative levels of analyses cannot be established a priori, they are discovered by reasonable guesswork accompanied by good old fashioned trial and error. For instance, if I’m interested in understanding the behavior of a human being under a certain set of circumstances I may include the level of analysis of his brain, muscles, and so on, maybe all the way down to his genetic makeup included. I may even take into consideration levels of analysis above that of the single organism, since the human being in question is probably responding to societal forces and other characteristics of the environment he lives in. Including higher levels means that one, strictly speaking, isn’t even a reductionist. As far as I know there is no specific term to identify the sort of mixed, pragmatic approach I’m describing, so I’ll make one up: levelism and levelist, to indicate respectively the method of picking a small number of most informative levels of analysis and the kind of person who practices such method. I am a levelist, then.

Of course, there is a limit to the levels of analyses that will be useful in any given instance: in the case of human behavior, it would be just as silly to scale up to the level of galaxies as to scale down to the level of quarks. While it is certainly the case that the broadest context of that human’s existence is a universe made of galaxies, and that the same human is made of quarks, electrons, and so forth, neither Einstein’s general relativity nor quantum mechanics will tell me much that is relevant to the issue at hand.

This, incidentally, is not a new problem at all. Listen to Socrates making one of the earliest critiques known concerning what I’ve been calling greedy reductionism:

“I found [Parmenides] altogether forsaking mind or any other principle of order, but having recourse to air, and ether, and water, and other eccentricities [to explain things]. I might compare him to a person who began by maintaining generally that mind is the cause of the actions of Socrates, but who, when he endeavored to explain the causes of my several actions in detail, went on to show that I sit here because my body is made up of bones and muscles; and the bones, as he would say, are hard and have joints which divide them, and the muscles are elastic, and they cover the bones, which have also a covering or environment of flesh and skin which contains them; and as the bones are lifted at their joints by the contraction or relaxation of the muscles, I am able to bend my limbs, and this is why I am sitting here in a curved posture—that is what he would say, and he would have a similar explanation of my talking to you, which he would attribute to sound, and air, and hearing, and he would assign ten thousand other causes of the same sort, forgetting to mention the true cause, which is, that the Athenians have thought fit to condemn me, and accordingly I have thought it better and more right to remain here and undergo my sentence.” (Plato, Phaedo, 98c)

Why was Socrates drinking hemlock?

This is simply beautiful and stunning. Socrates is saying that Parmenides—a Presocratic interested in natural philosophy, i.e., science—would go on and on talking about muscles and bones, and joints in order to explain why Socrates is sitting there, about to drink hemlock. But while it is true that human beings are capable of sitting because of the way their anatomy works, such anatomy is entirely irrelevant to explain the specific situation Socrates found himself in. The answer, the correct level of analysis, has to do with the politics of 4th century BCE Athens and with human psychology, not with limbs and sinews. Parmenides was the first greedy reductionist on record, and Socrates rightly chided him for his mistake.

The opposite of a greedy reductionist is someone who invokes holism, a word indicating that the only proper level of analysis is that of the whole, however defined. The whole human being, or the whole society, or even the whole universe.

Almost unfailingly, holists are fuzzy thinkers, and often purveyors of pseudoscience. They are allergic to reductionist explanations, which they consider anathema. They typically reject serious empirical investigations in favor of anecdotal evidence, favor vague “explanations” and superficial analogies over rigorous theories, especially if mathematically or quantitatively expressed. Think Deepak Chopra, if you need to visualize a clear mental target.

While greedy reductionists are ineffective, holists are downright pernicious, because they don’t just engage in questionable epistemology, they reject the whole notion of epistemic warrant in the first place. Steven Weinberg may have been affected by epistemic hubris when he suggested that, in principle, quantum mechanics could explain everything. But Deepak Chopra is simply talking nonsense when he expounds on “quantum mysticism.” I have no doubt about with whom I’d rather have a beer and engage in discussions about the nature of reality.

At the end of the day we need to remind ourselves of the fact that when human beings try to explain something they are engaging in an exercise of practical epistemology, not theoretical metaphysics. An explanation, according to my faithful dictionary, is “a statement or account that makes something clear,” or at least clearer than it was before. This is accomplished by explaining X, the unknown item, in terms of V, W (levels above X), Y, and Z (levels below X), which are already at least partially known. Reductionism by itself does not do a good job at this, especially when it veers toward its greedy variety. Holism performs even worse. Pragmatic levelism is the way to go.