Ian Hacking has written an interesting, confusing, fast-reading, slow-digesting, exasperating, idiosyncratic book which is must reading for anyone interested in the philosophy of science. The introduction is alarming indeed. After describing Feyerabend’s position (‘There are many rationalities, many styles of reason, and also many good modes of life where nothing worth calling reason matters very much’), Hacking adds: ‘My own attitude to rationality is too much like that of Feyerabend to discuss it further.’ Fortunately, this professed deconstructionism turns out to be so much hype: Hacking thinks well of both Feyeraband and Austin, but it’s on Austin’s side that he finds himself when the chips are down. In any case, as he himself tells us, ‘what follows is about scientific realism, not rationality.’
In a charming sub-section of the first chapter titled ‘If you can spray them, then they are real’, Hacking explains that he ‘never thought twice’ about scientific realism until a friend told him about an experiment in which the experimenter altered the charge on a tiny niobium ball to determine whether the passage from positive to negative charge occurs at zero or at plus or minus one-third the charge of the electron (the purpose being to try to detect free quarks). How did the experimenter alter the charge on the niobium ball? ‘Well, at that stage,’ said Hacking’s friend, ‘we spray it with positrons to increase the charge or with electrons to decrease the charge.’ ‘From that day forth I’ve been a scientific realist,’ Hacking writes. ‘So far as I’m concerned, if you can spray them then they are real.’ Samuel Johnson long ago made the same argument. (‘If you can kick them, then they are real.’) Of course, Johnson didn’t actually say anything like this, but when he ‘refuted’ Berkeley by kicking a stone he was making the same point. Still, as all philosophers know, there are things wrong with such short ways with old and deep questions.
The trouble is that Berkeley didn’t doubt the existence of stones. It is hard to say what a stone or a table is in, for example, the language of elementary particle physics. It isn’t just a structured aggregate of atoms because (as Saul Kripke has emphasised) the identity conditions for the stone aren’t the same as the identity conditions for the aggregate of atoms (it would be the same stone but not the same aggregate of atoms if one atom were removed). Deep and sophisticated theories about ‘time-slices’ and ‘space-time regions’ and ‘continuity’ have been proposed to reconcile the scientific conceptual scheme with common sense. But any such reconciliation will allow one to believe both that the world ‘ultimately’ consists of atoms and electrons and protons and the like, and that it makes sense to speak of certain heaps (or, better, certain space-time rivers of matter) as ‘stones’ and ‘tables’ and ‘chairs’. An exact translation of ordinary language into scientific language need not be possible. In the same way, a Bishop Berkeley can believe both that the world ‘ultimately’ consists of minds and their sense-impressions (‘spirits and their ideas’) and that we can speak of certain rivers of sense-impressions as ‘stones’ and ‘tables’ and ‘chairs’, at least ‘speaking with the vulgar’. If Berkeley had doubted that there were stones or that one could kick them, Johnson’s act would have been a striking reminder of the wrongness of Berkeley’s view; as it is, it merely begged the question. Why isn’t Hacking simply repeating Johnson’s mistake?
Well, I don’t know why Hacking isn’t. (That’s why I called the book ‘exasperating’.) Hacking may not have an argument for scientific realism at all, as opposed to a psychological observation about what convinces us. (He may even reject this distinction.) But to understand the book one has to turn to the other side of the ‘scientific realism’ question, to realism about theories. Here Hacking parts company with most ‘scientific realists’. His realism is a robust belief in the independent existence of certain objects (‘if you can spray them, then they are real’), not a belief in the objective truth of the schemes of representation and explanation we call scientific theories.
Hacking’s doubts about the objectivity of scientific theory are connected with a problem which has fascinated many philosophers (including the logical positivists, contrary to a claim of Hacking’s). This is the problem of Equivalent Descriptions – the problem posed by the fact that, as Hacking puts it (page 143), ‘there might be several ways to represent the same facts.’ To see why this is a problem, consider the history of science as it might have been written before the awareness of the problem of Equivalent Descriptions. (Hacking traces this awareness to Hertz’s posthumous book, The Principles of Mechanics of 1894.)
A pre-Hertzian historian of science might have said that Carnot’s theory of the ideal steam engine ‘turned out to be approximately correct, except that the reference to “caloric” as a mysterious fluid is an unnecessary element, and that it was also necessary to unify Carnot’s ideas with the principle of the conservation of energy.’ This is the traditional picture of science ‘converging’ to The Truth. In the same spirit, one might today write a history in which one said that what Newton did was to discover that there is a causal contribution that the presence of one mass (say, the sun) makes to the acceleration of another (say, the earth). The magnitude of this acceleration is described to a high degree of accuracy by Newton’s inverse square law. Today (thanks to Einstein) we have discovered the mechanism of this causal contribution (space-time curvature) and eliminated the reference to Absolute Space as an unnecessary element.
Now, a famous textbook of General Relativity by Weinberg presents the theory in a very un-Einsteinian way: in the Weinberg version there is no curved space-time! Suppose the Weinberg theory to be fully inter-translatable with the Einstein theory (as it is for most applications). Then, if we ‘spoke the language’ of the Weinberg theory we would simply say that we had discovered (thanks to Weinberg) the mechanism of Newton’s ‘gravitational attraction’ (a field with a massless quantum of spin two): but this is a very different answer from that yielded by the Einsteinian version of the theory.
Is this so bad? I consider myself a ‘realist’ with a small r (an ‘internal’ realist, I like to say), and I would not be troubled by such a situation. I would be prepared to treat the Weinberg theory and the Einstein theory as mere notational variants if a full inter-translatability could be established. ‘But then we couldn’t give an answer to the question Is space-time curved or not? that was independent of our own conceptual choices!’ But why should we be able to do that? Why should it trouble us so that we cannot divide up ‘what makes a statement true’ into a part that is The Way Things Are in Themselves and another which is Our Conceptual Contribution? Why shouldn’t we say that both versions are correct, if either is?
This is not Hacking’s reaction, however. Hacking wants truth to be totally non-perspectival, and this leads him to what I find the least persuasive distinction in the book: he attempts to distinguish between ‘simple, non-representational assertions’ (My typewriter is on the desk), concerning which there is a ‘truth of the matter’, and ‘a barrage of more or less instructive representations’, which is all we get in physical theory.
Where Hacking’s book is strongest is in its documentation of the extent to which an experimenter’s confidence in the reality of what he is manipulating can be independent of commitments to a serious explanatory theory. Hacking does convince me that philosophers of science are woefully ignorant of the facts of life about experimentation, and the second part of this book is an excellent remedy for this woeful ignorance. But what I find less convincing is the supposed independence of experimental reality from representation. I can describe a measurement or a manipulation as a measurement or a manipulation of electromagnetic fields: but I can also describe the same measurement or manipulation as a measurement or manipulation of forces acting at a distance (non-instantaneously) on bits of matter. What we manipulate does not carry its own description, its Aristotelian essence, on its face. This is a pervasive phenomenon. A Hopi describes a situation as ‘producing a forked pattern in a bush by hand action’ where we speak of ‘pulling a branch aside’. A Hopi Hacking might say that ‘if you can produce them by hand action then they are real,’ and thus establish the ‘reality’ of forked patterns!
If I cannot buy Hacking’s arguments, I do appreciate his numerous careful discussions of various views (including two chapters on my own views), as well as his rich account of actual science and past science. But I would draw a different conclusion. Instead of trying to save metaphysical realism (‘externalism’) by restricting it to ‘simple, non-representational assertions’ (kicking Samuel Johnson’s stone), I would insist that there is no philosophically interesting class of assertions that is innocent of our own conceptual contribution (and could thus serve as our ‘representation-independent’ hold on reality). The search for that conceptually uncontaminated Reality ‘out there’ still looks like a mug’s game.