[ print ]

Co-Director of LSE's Centre for Philosophy of Natural and Social Science; Author, The Ant and the Peacock: Altruism and Sexual Selection from Darwin to Today

In The Beginning Is The Theory

Let's eavesdrop on an exchange between Charles Darwin and Karl Popper. Darwin, exasperated at the crass philosophy of science peddled by his critics, exclaims: "How odd it is that anyone should not see that all observation must be for or against some view if it is to be of any service!" And, when the conversation turns to evolution, Popper observes: "All life is problem-solving … from the amoeba to Einstein, the growth of knowledge is always the same".

There is a confluence in their thinking. Though travelling by different pathways, they have arrived at the same insight. It is to do with the primacy and fundamental role of theories—of ideas, hypotheses, perspectives, views, dispositions and the like—in the acquisition and growth of knowledge. Darwin was right to stress that such primacy is needed 'if the observation is to be of any service'. But the role of a 'view' also goes far deeper. As Darwin knew, it is impossible to observe at all without some kind of view. If you are unconvinced, try this demonstration, one that Popper liked to use in lectures. "Observe!" Have you managed that? No. Because, of course, you need to know "Observe what?" All observation is in the light of some theory; all observation must be in the light of some theory. So all observation is theory-laden—not sometimes, not contingently, but always and necessarily.

This is not to depreciate observation, data, facts. On the contrary, it gives them their proper due. Only in the light of a theory, a problem, the quest for a solution, can they speak to us in revealing ways.

Thus the insight is immensely simple. But it has wide relevance and great potency. Hence its elegance and beauty.

Here are two examples, first from Darwin's realm then from Popper's.

• Consider the tedious but tenacious argument 'genes versus environment'. I'll take a well-studied case. Indigo buntings migrate annually over long distances. To solve the problem of navigation, natural selection equipped them with the ability to construct a mental compass by studying the stars in the night sky, boy-scout fashion, during their first few months of life. The fount of this spectacular adaptation is a rich source of information that natural selection, over evolutionary time, has packed into the birds' genes—in particular, information about the rotation of the stellar constellations. Thus buntings that migrate today can use the same instincts and the same environmental regularities to fashion the same precision-built instrument as did their long-dead ancestors.

And all adaptations work in this way. By providing the organism with innate information about the world, they open up resources for the organism to meet its own distinctive adaptive needs; thus natural selection creates the organism's own tailor-made, species-specific environment. And different adaptive problems therefore give rise to different environments; so different species, for example, have different environments.

Thus what constitutes an environment depends on the organism's adaptations. Without innate information, carried by the genes, specifying what constitutes an environment, no environments would exist. And thus environments, far from being separate from biology, autonomous and independent, are themselves in part fashioned by biology. Environment is therefore a biological issue, an issue that necessarily begins with biologically-stored information.

But aren't we anyway all interactionists now? No longer genes versus environment but gene—environment interaction? Yes, of course; interaction is what natural selection designed genes to do. Bunting genes are freighted with information about how to learn from stars because stars are as vital a part of a bunting's environment as is the egg in which it develops or the water that it drinks; buntings without stars are destined to be buntings without descendants. But interaction is not parity; the information must come first. Just try this parity test. Try specifying 'an' environment without first specifying whether it is the environment of a bunting or a human, a male or a female, an adaptation for bird navigation or for human language. The task is of course impossible; the specification must start from the information that is stored in adaptations.And here's another challenge to parity. Genes use environments for a purpose—self-replication. Environments, however, have no purposes; so they do not use genes. Thus bunting-genes are machines for converting stars into more bunting-genes; but stars are not machines for converting bunting-genes into more stars.

• The second example is to do with the notion of objectivity in science. Listen further to Darwin's complaint about misunderstandings over scientific observation: "How profoundly ignorant … [this critic] must be of the very soul of observation! About thirty years ago there was much talk that geologists ought only to observe and not theorize; and I well remember some one saying that at this rate a man might as well go into a gravel-pit and count the pebbles and describe the colours".

Nearly two hundred years later, variants of that thinking still stalk science. Consider the laudable, but now somewhat tarnished, initiative to establish evidence-based policy-making. What went wrong? All too often, objective evidence was taken to be data uncontaminated by the bias of a prior theory. But without 'the very soul' of a theory as guidance, what constitutes evidence? Objectivity isn't to do with stripping out all presuppositions. Indeed, the more that's considered to be possible or desirable, the more the undetected, un-criticised presuppositions and the less the objectivity. At worst, a desired but un-stated goal can be smuggled in at the outset. And the upshot? This well-meant approach is often justifiably derided as 'policy-based evidence-making'.

An egregious example from my own recent experience, which still has me reeling with dismay, was from a researcher on 'gender diversity' whose concern was discrimination against women in the professions. He proudly claimed that his research was absolutely free of any prior assumptions about male/female differences and that it was therefore entirely neutral and unbiased. If any patterns of differences emerged from the data, his neutral, unbiased assumption would be that they were the result of discrimination. So might he accept that evolved sex differences exist? Yes; if it were proven. And what might such a proof look like? Here he fell silent, at a loss—unsurprisingly, given that his 'neutral' hypotheses had comprehensively precluded such differences at the start. What irony that, in the purported interests of scientific objectivity, he ostensibly felt justified in clearing the decks of the entire wealth of current scientific findings.

The Darwin-Popper insight, in spite of its beauty, has yet to attract the admirers it deserves.