The Omniscience and Omnipotence of Science

I have changed my mind about the omniscience and omnipotence of science. I now realize that science is strictly limited, and that it is extremely dangerous not to appreciate this.

Science proceeds in general by being reductionist. This term is used in different ways in different contexts but here I take it to mean that scientists begin by observing a world that seems infinitely complex and inchoate, and in order to make sense of it they first "reduce" it to a series of bite-sized problems, each of which can then be made the subject of testable hypotheses which, as far as possible, take mathematical form.

Fair enough. The approach is obviously powerful, and it is hard to see how solid progress of a factual kind could be made in any other way. It produces answers of the kind known as "robust". "Robust" does not of course mean "unequivocally true" and still less does it meet the lawyers' criteria — "the whole truth, and nothing but the truth". But robustness is pretty good; certainly good enough to be going on with.

The limitation is obvious, however. Scientists produce robust answers only because they take great care to tailor the questions. As Sir Peter Medawar said, "Science is the art of the soluble" (within the time and with the tools available). 

Clearly it is a huge mistake to assume that what is soluble is all there is — but some scientists make this mistake routinely.

Or to put the matter another way: they tend conveniently to forget that they arrived at their "robust" conclusions by ignoring as a matter of strategy all the complexities of a kind that seemed inconvenient. But all too often, scientists then are apt to extrapolate from the conclusions they have drawn from their strategically simplified view of the world, to the whole, real world.

Two examples of a quite different kind will suffice: 

1: In the 19th century the study of animal psychology was a mess. On the one hand we had some studies of nerve function by a few physiologists, and on the other we had reams of wondrous but intractable natural history which George Romanes in particular tried to put into some kind of order. But there was nothing much in between. The behaviourists of the 20th century did much to sort out the mess by focusing on the one manifestation of animal psychology that is directly observable and measurable — their behaviour.

Fair enough. But when I was at university in the early 1960s behaviourism ruled everything. Concepts such as "mind" and "consciousness" were banished. B F Skinner even tried to explain the human acquisition of language in terms of his "operant conditioning".

Since then the behaviourist agenda has largely been put in its place. Its methods are still useful (still helping to provide "robust" results) but discussions now are far broader. "Consciousness", "feeling", even "mind" are back on the agenda.

Of course you can argue that in this instance science proved itself to be self-correcting — although this historically is not quite true. Noam Chomsky, not generally recognized as a scientist, did much to dent behaviourist confidence through his own analysis of language.

But for decades the confident assertions of the behaviourists ruled and, I reckon, they were in many ways immensely damaging. In particular they reinforced the Cartesian notion that animals are mere machines, and can be treated as such. Animals such as chimpanzees were routinely regarded simply as useful physiological "models" of human beings who could be more readily abused than humans can. Jane Goodall in particular provided the corrective to this — but she had difficulty getting published at first precisely because she refused to toe the hard-nosed Cartesian (behaviourist-inspired) line. The causes of animal welfare and conservation are still bedeviled by the attitude that animals are simply "machines" and by the crude belief that modern science has "proved" that this is so.

2: In the matter of GMOs we are seeing the crude simplifications still in their uncorrected form. By genetic engineering it is possible (sometimes) to increase crop yield. Other things being equal, high yields are better than low yields. Ergo (the argument goes) GMOs must be good and anyone who says differently must be a fool (unable to understand the science) or wicked (some kind of elitist, trying to hold the peasants back).

But anyone who knows anything about farming in the real world (as opposed to the cosseted experimental fields of the English home counties and of California) knows that yield is by no means the be-all and end-all. Inter alia, high yields require high inputs of resources and capital — the very things that are often lacking. Yield typically matters far less than long-term security — acceptable yields in bad years rather than bumper yields in the best conditions. Security requires individual toughness and variety — neither of which necessarily correlate with super-crop status. In a time of climate change, resilience is obviously of paramount importance — but this is not, alas, obvious to the people who make policy. Bumper crops in good years cause glut — unless the market is regulated;  and glut in the current economic climate (though not necessarily in the real world of the US and the EU) depresses prices and put farmers out of work.

Eventually the penny may drop — that the benison of the trial plot over a few years cannot necessarily be transferred to real farms in the world as a whole. But by that time the traditional crops that could have carried humanity through will be gone, and the people who know how to farm them will be living and dying in urban slums (which, says the UN, are now home to a billion people).

Behind all this nonsense and horror lies the simplistic belief, of a lot of scientists (though by no means all, to be fair) and politicians and captains of industry, that science understands all (ie is omniscient, or soon will be) and that its high technologies can dig us out of any hole we may dig ourselves into (ie is omnipotent).

Absolutely not.