Why the philosophy of scienceactually does matter

Biologists, in general, tend to have relatively little interest in

the philosophy of science. There is a sound basis for this

indifference: the great majority of biologists manage to do

perfectly good science without having had much more than

a brush, usually as undergraduates, with this branch of

philosophy. Furthermore, even well before there was a well-

defined academic discipline called the ``philosophy of

science'', there were outstanding individuals who did superb

scientific work, without benefit of a clearly defined philosophi-

cal framework. Darwin, famously, is on record as saying:

1) that he collected facts wholly without preconceptions

and 2) that he always looked at facts with some provisional

idea of how they might make sense. Clearly, his lack of

consistency, or even understanding, about his own way of

operating did not prevent him from doing ground-breaking

work.

It is no wonder, therefore, that many biologists privately

regard the chief function of the philosophy of science as

something to provide employment to a small number of

academics in philosophy departments. Furthermore, recent

advances in the ways that molecular biologists gather data

seem to threaten a fundamental tenet of the philosophy of

science, further diminishing its relevance to the actual practice

of science. That tenet is that scientific knowledge is advanced

by the testing of hypotheses. The new techniques, which have

created veritable oceans of data to swim through, seem to

have made hypothesis-testing obsolete; the data, it is thought,

can speak for themselves. Indeed, one of the developers of

micro-array techniques has been widely quoted as declaring,

``The era of hypothesis-driven research is over.'' In other

words, Popperism is pooped-out and induction (the pure

gathering of facts and generalization of their meaning without

any preconceived notions) is in.

On p. 104 of this issue, John Allen challenges that notion.

He points out that any assessment or interpretation of

significance involves an hypothesis and that investigation

of any correlation turned up by the new methods involves

hypothesis-testing. Imagine a typical microarray experiment

that is used to track changes in gene expression during the

course of a particular developmental change in a certain

organism. Let us say that you identify 1053 genes whose

mRNA levels go up, 393 that go down, and 4126 that stay the

same. Which ones do you concentrate on? The easy decision

is to ignore the 4126 that stay the same. Yet, that involves an

assumption, in fact an hypothesis; there may be some genes,

even a large number, in that group, that play a role in the

developmental change of interest, although their expression

rates are unaltered. But what about the ones that do change?

Which ones do you investigate first? Those that go up, those

that go down, those that have the largest changes, those that

carry one sequence motif vs. another? Each of those choices

involves an hypothesis about what is important. Rather like

M. Jourdain in Moliere's Le Bourgeois Gentilhomme, who

discovers, to his astonishment, that he has been speaking

prose all his life, the scientist doing a microarray experiment is

continually hypothesis-testing, whether or not he/she thinks

of it that way. This hypothesis-testing may not be hard

Popperian, since many of these experiments seek confirma-

tion rather than falsification, but they are hypotheses, never-

theless. The choice is not whether your research will or will not

be guided by hypotheses but whether the hypotheses being

tested are consciously chosen or whether they are dictated by

hunch and fashion.

Yet, if the detailed work of much (indeed, all) of molecular

biology is based on philosophical tenets, even when not overtly

recognized as such, what about the larger picture? Does the

philosophy of science come into the question of how one

chooses one's scientific goals or major projects in the first

place? Surprisingly, here, too, the answer is ``yes''.

Three years ago, the Novartis Foundation (formerly the

Ciba Foundation) held a symposium on the subject of

reductionism in science. Its title was ``The limits of reduction-

ism'' and the central issue addressed was whether the

properties of complex systems can be deduced or recon-

structed from detailed knowledge of the properties of their

constituents and the direct interactions of those constituents.

This is the sort of philosophical issue that many biologists will

respond to with an upward roll of the eyes and/or a shrug. Yet,

as one of the discussants pointed out succinctly, the issue of

whether reductionist approaches are the sole route to under-

standing complex systems is really a question about what gets

funded, hence about who gets funded. If, for instance, you are

interested in some aspect of the operation of the CNS, do you

take an exclusively hard core reductionist approach, and

concentrate on identifying all the genes and gene products that

might be involved in the relevant areas of the brain? If so, your

tacit assumption is that the properties of the system will

become apparent from their study. Or, do you characterize the

systems properties first, to try to obtain clues as to which

neural processes, regions and (ultimately) molecules, might

be important? Most researchers would probably respond to

this choice with, ``you need both sorts of approach''. But, even

accepting that, how do you apportion the money to them? Do

you give 99% to reductionist strategies and 1% to system

BioEssays 23:1±2, ß 2001 John Wiley & Sons, Inc. BioEssays 23.1 1

Editorial

characterization? Or, should the divide be 90% and 10%,

respectively? Or, do you opt for a 50:50 split?

These decisions about spending are intimately tied to one's

beliefs about: (1) the nature of the entity you are investigating,

(2) how its properties originate, and (3) how those properties

can be most fruitfully investigated. If the funding agency

believes that one can deduce all the properties of the system

from knowing the full inventory of molecules that are

expressed in the cells of interest, and their properties, then it

should put all of its financial eggs in that basket. If it believes,

however, that there are systems properties that cannot be

so deducedÐeven though they ultimately derive from the

physico-chemical properties of the molecules and cells of the

systemÐit should apportion the money differently.

In the end, therefore, questions about the funding of

research are highly contingent upon the nature of the basic

philosophical assumptions made about one's subject and how

it can be investigated most effectively. To ask whether the

philosophy of science should matter to biologists is to ask, in

effect, whether the disbursement of research funds matters to

them. Most bench scientists would agree that the latter

question is of more than mere academic interest.

ASW

Editorial

2 BioEssays 23.1