Professor Van Fraassen’s book is a recent addition to the Clarendon Library of Logic and Philosophy which Mr Jonathan Cohen is editing for the Oxford University Press. Its aim, as expressed in the blurb, is ‘to develop an empiricist alternative to both logical positivism and scientific realism’. In fact, Van Fraassen has very little to say about logical positivism, which he regards as philosophically outmoded, and devotes nearly all his energy to confronting scientific realism with what he calls a constructive alternative. This consists of three interlocking theories, one of them concerning the relation of scientific theories to the world, which Van Fraassen identifies with their ‘empirical import’, the second a theory of scientific explanation, and the third an account of probability as ‘it occurs within scientific theories’. Van Fraassen claims to have avoided technicalities throughout, but he presumes that his readers have a fair command of logic and mathematics, and I doubt if his treatment of probability, in particular, would mean very much to anyone who was not already well versed in quantum physics. Van Fraassen’s own command both of the history of physics and of the contemporary literature bearing on its philosophy is not in doubt.
Scientific realism, according to Van Fraassen, consists in the view that scientific theories are to be construed literally, and that to accept them is to believe them to be true. Most philosophers who contest this view deny the first of these propositions; they treat the more abstract entities to which the theories ostensibly refer as convenient fictions which are introduced for the purpose of linking observable phenomena. Van Fraassen takes the much less common and indeed rather surprising course of accepting the first proposition but denying the second. He parts company with positivism and instrumentalism just because of their denial that scientific theories mean what they appear to say. His brand of anti-realism commits him to a literal construal of scientific terms, but allows him to maintain that acceptance of a scientific theory need not imply belief in its truth. All that is required is the belief that the theory is ‘empirically adequate’, which means that it has at least one model into which all the phenomena that come within its range can be made to fit. This position is ingeniously designed to give Van Fraassen the best of both worlds. He is spared the impossible task of ‘reducing’ statements about atomic particles to reports of actual or possible observations, and at the same time he need only consider the bearing of these statements upon observable phenomena.
Van Fraassen makes it clear that when he speaks about observable phenomena he is referring to physical objects and not to anything of the order of sense-data. He waves aside the suggestion that ‘the observable objects and processes we recognise in our world are [themselves] postulated entities, believed in because they best explain and systematise [our] sense-experiences,’ on the ground that ‘philosophers spent the first five decades of this century refuting the presuppositions that lie behind it.’ Here, for once, his scholarship forsakes him. In the first half of this century the best philosophers such as Russell, Moore, C.I. Lewis and Moritz Schlick accepted the theory of sense-data. It is only in the last thirty years that it has fallen out of fashion. This point, however, is not important for Van Fraassen’s argument. A more serious charge against him is that no clear line can be drawn between theory and observation. He rebuts this by admitting that ‘observable’ is a vague predicate, but arguing that when it comes to particular cases there are differences of degree which are sufficient for his purpose. For instance, ‘a look through a telescope at the moons of Jupiter’ is reckoned by him to be a case of observation, whereas ‘purported observation of micro-particles in a cloud chamber’ does not qualify. The particles are detected by observable means but they are not themselves observable. I think that in this case and many others like it the distinction is justified.
A point on which Van Fraassen insists is that this distinction between theoretical and observational terms is not syntactical, a mere matter of where we choose to draw the line within our terminology, nor is it a distinction that is to be made on purely philosophical grounds. ‘If,’ he says, ‘there are limits to observation, these are a subject for empirical science, and not for philosophical analysis.’ Here, I think, he draws too sharp a distinction between philosophy and science, besides underestimating the part that convention plays in fixing the boundaries of observability. No doubt there will be good scientific reasons at any given stage for deciding how these boundaries are to be drawn, but it may be accounted a philosophical task to bring these reasons to light.
Having concluded that a theory is ‘a family of structures’, reserved the title of ‘appearances’ for ‘the structures which can be described in experimental and measurement reports’, and declared a theory to be ‘empirically adequate if it has some model such that all appearances are isomorphic to empirical substructures of that model’, Van Fraassen turns to the topic of scientific explanation. His treatment of this topic is, to my mind, the most interesting part of his book. He agrees with other theorists that ‘explanation of why an event happens consists (typically) in an exhibition of salient factors in the part of the causal net formed by lines “leading up” to that event,’ but then argues convincingly that the choice of salient factors varies with the context, including the aims and interest of the person who is offering the explanation. Thus the meaning of sentences of the form ‘A causes B’ is not univocal. In interpreting them, we have always to allow for the contexts in which they are uttered. Neither do we seek for explanations in the void. We want to know why a given event happened rather than some other that might have occurred in its place. These alternatives form what Van Fraassen calls a ‘contrast-class’, and he maintains, I think rightly, that this and what he calls ‘a relevance relation’ are both factors for which the scientific theories to which we resort for our explanations must leave room. A great merit of this account is that it dispenses with ‘inextricably modal’ or counterfactual elements.
Maintaining as he does that there is no modality in the scientific description of the world, and acknowledging that science includes irreducibly probabilistic theories, Van Fraassen is troubled by the fact that probability, as it figures in these theories, functions as a modality, the probability of an event being regarded as equal to ‘the relative frequency with which it would occur, were a suitably designed experiment performed often enough under suitable conditions’. He tries to extricate himself from this difficulty by tying modality to scientific language but confining it to the part of scientific language that is not used for the description of the actual world. In short, we need to show that our scientific talk does not commit us ‘to some sort of metaphysical beliefs such as that alternative possible words are real.’ I think that this is a promising line of escape, but it needs to be developed in more detail.
Van Fraassen concludes his book with a short satirical chapter in which he draws an analogy between the arguments commonly used by scientific realists and the five ways in which St Thomas Aquinas sought to prove the existence of God. That the humour here is somewhat laboured should not be allowed to detract from the overall merit of this interesting and enterprising book.