I suspect that as soon as humans evolved a language capable of assigning nouns to types, they started to make abstractions, and to think that abstractions were real things. Abstraction is a necessary part of knowledge if you want to be able to generalise beyond particulars. But it brings with it dangers, and one of these is what Gould (2002) called the fallacy of abstract reification, and more famously Whitehead (a badly misunderstood thinker, in my opinion) called the fallacy of misplaced concreteness. I will call it the
reification fallacy because it's shorter and snappier and because renaming ideas is half of philosophy.
The reification fallacy is endemic in biology. We reify (literally, make into a thing) all kinds of terms. "Organism" is one of them. So is "gene", "organ" and so on. Some of these denote a class of concrete objects: there really is a heart in mammalian organisms, for example, and the fallacy lies in thinking that the typical form is
the form. In fact, even such a central and gross subsystem as the aortic arch comes in a variety of quite disparate forms. And each individual will have their own variations. But because it is useful to talk about the aortic arch in teaching and so forth, we reify it.
But there are much more abstract objects than organs or
genes (and boy is
that a contested concept in biology). I have in mind here terms like "function", "form" and the recent addition (since about 1940) to the abstract pantheon, "information". I believe, rather controversially, that these abstract properties are properties of ourselves rather than the organisms. Well, that's the headline. In fact some of the things that get referred to are real enough. Genes certainly do the things that get called information transmission, and organs certainly do the things that get called functions. But their status as information and functions, and the extracting of
those processes over all the other processes that the genes do, or the organs do, depend crucially on the interests and focus of the researchers.
So, why do we abstract things? It has to do with the nature of explanation. The standard notion of explanation in the philosophical literature is known as the
deductive-nomological (DN) or Covering Law account, and was formulated by
Carl Hempel. According to this, an explanation has the form
Covering Law[s] C
Initial Conditions I
============
Thing to be explained E
This is the form of a logical argument -
C &
I → E, and so
E is explained. The covering laws are general propositions of the kind ∀
x (F
x) [All
xs are F]. In physics, for example, one might use the law that
PV =
kT (the Ideal Gas Law, where the pressure
P, volume
V, and temperature
T covary according to a constant
k that is specified by the nature of the gas). The explanation would then give values to the variables
P, and
V and
k, and conclude that the temperature
T, which is measured in a gas, is explained. But the "law" here holds true only within a limited range of conditions (when the gas is not a liquid at high
P or a plasma at high
T). That is, it is not exceptionless, and is subsumed under atomic theory.
Now in biology, we want to be able to explain (for example) why a tiger or a zebra has stripes. There is no law about this - it relies on the visual acuity of prey (for tigers) and predators (for zebras) in the appropriate conditions. We set up the conditions
I so that the covering generalisations make sense of the outcome. Something like "stripes are selected for when other organisms cannot sense the edges of the animal and this affects fitness", thus confounding both prey for the tiger and predators for zebras (a current hypothesis is that zebras confuse the targeting of individual animals in herds). To do this we need to abstract away from the particular animal to the general type. "Striped animal" is a type - but it depends very much on the focus of the person doing the explaining. Another explanation of stripes is due to
Alan Turing, who noted that it can form due to
gradients of interacting chemical signals that tell skin cells to express coloration or not. [Apocryphally, Turing was asked if he had explained stripes on a zebra, and he replied "Yes, but the whole animal is harder". Here's a
Java applet you can play with.]
So we might now ask if there is a
natural class of "striped animal". It is surely true that there is, in one sense, for there are animals and they are striped. But whether that class matches the ways evolution generates them - that is, whether "striped animal" is a clearly delineated natural group that can play a role in scientific explanations - it another matter. It might be that we have arbitrarily constructed that class. For instance, is an animal that uses diffraction to confuse its predators or prey a striped animal? How about if the diffraction gratings are linear and aligned in parallel? How far is the class to extend?
The search of suitable abstractions in biology to play a role in explanation has to be balanced against the particularity of biological objects. But we have to do it, and those classes that fail to be general enough and yet true to the particularity of the organisms, that is, fail to be empirically adequate, tend to be eliminated either through neglect by biologists, or by being deliberately abandoned. Such is the dialectic of science.
But sometimes, these abstract objects persist because they are neither obviously false, nor heuristically useless. And yet, they often cause no end of trouble in the thinking of biology. One of these terms, as I noted, is "gene". Griffiths and Stotz [
pdf] have enumerated many kinds of meanings for the term in scientific usage. Some of them are mutually incompatible. Each can be replaced with more descriptive and specific terms, but still people persist in talking about "the" gene. And it has some rather difficult implications, which lead people to say things like "humans are genetically determined in their IQ" and so forth. "Gene" gets reified.
One can multiply examples of this in biology. Is there a solution? Obviously, as a philosopher I think that the sort of work done by philosophers of science plays a role. We can analyse some abstractions the way Griffiths and Stotz have done and situate them in a space of conceptual alternatives to show the confusions. Occasionally we might offer our own formulations. Or we might say that this is a necessary aspect of science, as it refines and forms its ideas, and leave the biologists to sort it out in due time. But since biology has a direct application to social debate and policy, I prefer to think that philosophy can offer a useful method for preventing bad ideas becoming the basis for the application of biology to society.
Explanatory reification is a necessary evil, but it needs to be checked from time to time. It is checked by science, and it can be checked by philosophy, if it takes the science seriously. But there is a broader case of explanatory reification that exceeds the science, and that is the way the terms of science and philosophy are misunderstood by the wider public. Partly this is due to the ways that philosophers misconstrue the generalisations of biology. Mostly it is due to the way that scientists promote their ideas, shorn of nuance and empirical content, to the populace. How often do we read press releases about "the gene" for traits such as homosexuality. Biologists know that this can only be an allele (a gene variant) that is causally implicated in that trait, either by being necessary for it
not to develop (that is, its absence causes the trait if everything else is equal) or by being part of a complex of genes and developmental systems. But that's a lot of words to put in a news report or a popular science book.
There are worse examples. One of my favourite pet hates is "information". This is an abstraction that usually just means that there is a mapping between one sort of structure (like a DNA sequence) and another (like the sequence of amino acids in a protein, which gets its functional properties from the way it is folded into complex shapes, assisted by other molecules). But it is so attractive a notion, so strong in its grip on our mind, that we generalise it to include aspects of our human intentionality. We start to talk about the "meaning" of genes, to continue our example. We think we can "decode" genes and make inferences about the environments of the past, or that there is something
real about the codical nature of genes, such that they are algorithms (another abstraction) that "solve" problems in the world, or represent environments to the organism.
This is a kind of anthropomorphism - we take some abstractions that are, as it were, in the general culture, the folk psychology and categories of ordinary life, and impose them on things for which those categories did not (culturally) evolve. And scientific images are just those sorts of things. It has been known since the beginnings of science that it contradicts "common sense" notions, of how things behave. And it misleads us systematically. Take "function" - the Aristotelian tradition has this (as "final cause") as something that is fundamental to the nature of (living and artifactual) things. More recent versions of functions accounts for them as the outcome of prior selection in the biological or cultural world - a function is whatever it is that selection has selected
for (so-called Proper Functions). So "function" is a post hoc property, not a prior fundamental property.
But I think we can go further and say, that "function" is something that is not only post hoc, but even
more post hoc - it is something about the way the organism has evolved, but which
also depends upon the things being identified and evaluated by the observer (the scientist). That is, a
function is what an observer identifies and explains as being the result of selection for that behaviour or process. But we do not make this kind of caveat, not even implicity, when we ascribe functions. We, to use the terms of critical theory, "privilege" the thing being explained.
A few years back I was talking to a cell biologist working on the role of the actin cytoskeleton in Williams' Syndrome - I believe I've mentioned this before - who bemoaned the fact that in the literature, the "functions" of the actin cytoskeleton (the network of actin filaments that give cells their shapes and structure, among other things) were too many. He wasn't
interested in those functions. They interfered with what he
was interested in. So there is an explanatory relativity to functions. Whether they were important or not depends on what the cognitive interests of the observer are. But the way philosophers of science (and scientists who do philosophy) talk about them, functions are properties of the world. Rather, we should say, they are properties of the ways the world is modelled or not.
This is basically the old problem of nominalism that arose during the middle ages. Are "universals" (general propositions) in the world or are they in the head? Nominalists, including the very influential Locke, say they are in the head. Am I therefore adopting a radical nominalism here? I think I am, partially, but not entirely. These generalisations occur in our best representations of the world. Like physical laws, they are supposed to identify aspects of the world that are common to all the things being explained. There is a two-way process of accommodation going on here. Our representations start, usually with some folk taxonomy or phenomenological salience, as rough first approximations. Evidence forces us to refine them, while we strive to say as much as can be said about the aspects of the world that interest
us. It's not that these general properties don't have any purchase in knowledge of the world, but rather than they are an attempt to shape our representations to express as much as we can about it, as economically as we can, while remaining empirically adequate. One might define science as the process of saying as much as we can as simply as we can based on what we know.
So while the abstractions exist in our heads, and not in the world, they are good representations (or, if you are unscientific, bad ones, though perhaps good representations of social norms) of the world. So where's the fallacy? The fallacy lies not in making inferences from our models and representations, such as making predictions. As I argued above, that is what they are for. The fallacy lies in taking the abstractions at face value; that is, inferring that secondary properties are real because the model
works. Or worse, inferring from our cognitive necessities to the reality of the those necessities in the world. There are too many empirically equivalent models, and too many possible cognitive systems, to think that what is true, for contingent historical reasons, of us, must be true of the world we know.
Consider, as a thought experiment, what might have happened had DNA's structure and processes been discovered before "information" was the go in science, say around 1900. We would not be talking about genetic information now, but rather about something like the "factorial power" or "causal efficacy" of genes. Mendel first called his particles "factors" (from the Latin, of course, for "maker"). Information sounds, when you put it like this, like the
virtus dormativa of Moliére's play. Or take "function" - what would we talk about if Aristotle's works hadn't been rediscovered in the 14th century? Philosophical problems sometimes arise because of cultural contingencies. The purpose ("function"? "goal"? "
telos"?) of science is to eliminate these cultural artifacts in favour of the representations that are as dependent on the world as possible, and on the culture as little as can be realistically achieved.
I think I'll leave it there. A Good Friday sermon, since I'm doing a lot of these lately.
