DID PERRIN PROVIDE AN EXPERIMENTAL ARGUMENT FOR SCIENTIFIC REALISM?
Or did he merely furnish an answer to what Carnap calls an internal question? That is, Perrin's argument more than simply an empirical argument for the existence of a particular natural kind, namely, molecules—entities which had been previously postulated by scientists but not shown to exist? It is this and a good deal more.
It is an argument to the existence of something that was regarded as dubious or objectionable on philosophical or methodological grounds. These grounds included the idea that from what is observed, one can make valid inferences only to what is observable, not to what is unobservable. Moreover, they are grounds for rejecting unobservables generally, not just molecules. (Duhem, for example, rejected a range of physical theories postulating unobservables.)Perrin did not respond to these critics by making molecules observable. Rather, he claimed that there is a reasonable method or mode of reasoning (the intuitive method) that starts with observed experimental results and can be used to infer the existence of things that are unobservable (or “invisible,” or “hidden,” to use Perrin's terms). He showed in detail how this method could be used to infer the existence of invisible molecules from experimental observations on Brownian motion. And he clearly re-
39. “The use of the intuitive method has not, of course, been used only in the study of atoms, any more than the inductive method has found its sole application in energetics”— Atoms, p. vii.
40.Statement (2) is close to what Leplin calls “minimal epistemic realism,” a doctrine he defends in A Novel Defense of Scientific Realism (New York: Oxford, 1997). garded this intuitive method as a general one, not restricted simply to molecules.39 This is not just scientific business as usual. It is not simply an argument establishing (or rejecting) the existence of just one more type of physical entity.
Nor did Perrin or his antirealist opponents take this as scientific business as usual.The scientific realism implicit in Perrin's arguments can be put like this:
(1) There are unobservables (for example, molecules).
(2) Their existence and their properties can be inferred (only) on empirical grounds, in some cases from experiments, so that a claim to know they exist and have these properties is justified.40
(3) A legitimate mode of reasoning that can be used for this purpose involves two important components:
(a) causal-eliminative reasoning to the existence of the postulated entity, and to certain claims about its properties, from other experimental results;
(b) an argument to the conclusion that the particular experimental results obtained are very probable given the existence of the postulated entity and properties.
Part (a) in (3) is intended to establish the high probability for the claim that the entity exists and has certain properties. Part (b) can be shown to sustain and possibly increase that probability on the basis of the new experiments. Reasoning involving (a) and (b) can be used to infer the existence of entities whether observable or unobservable.
The scientific realism reflected in points (1 )—(3) does not say that theories about unobservables in mature sciences are (approximately) true and that the unobservables postulated exist, or that the aim of science is to obtain true theories about unobservables as well as observables. Nor does it adopt either some metaphysical viewpoint external to science, saying that there is a mind- and theory-independent way the world is and/ or emphatically endorsing the adequacy of a scientific representation of this world, or a pragmatic viewpoint stressing the usefulness of scientific theorizing. It is not scientific realism in these senses.
Nor is it a restricted internal realism that says simply that molecules (electrons, or whatever) exist. It is much more general than this, and it has a methodological as well as an ontological component.
It claims that there are unobservables (Perrin's realm of “invisible simplicity”), and that valid arguments can be used to infer their existence and properties; these are entities which, on philosophical and/or methodological grounds, antirealists reject or understand in some nonrealistic way. It says, however, the arguments for these entities and claims about their properties are empirical ones, so that what entities exist in this realm and what properties they have are empirical questions. The particular empirical argument for a given unobservable will depend on, and vary with, the unobservable postulated. No general empirical argument can be given for all unobservables postulated. Nor can the issue of the existence of unobservables be settled a priori. Nevertheless, there is an a priori assumption that is essential to scientific realism so understood, namely, that a valid mode of reasoning can be employed (such as (a) and (b) in point (3)) which can justify a belief in the truth of propositions about the unobservable entities inferred.Does this deserve the name “scientific realism”? Fine may answer “no,” claiming that it does not go beyond what his neutral NOA permits. Fine says that NOA, and hence realism and antirealism, “accept(s) the results of scientific investigations as ‘true' on par with more homely truths.”[142] Hence, he will insist, since the results of scientific investigation include empirically based inferences to the existence of unobservables such as molecules, NOA, together with realism and antirealism, accepts both (1) and (2). Moreover, he may add that NOA, and hence both realism and antirealism, accepts point (3), since, he says, NOA sanctions “ordinary relations of confirmation and evidential support, subject to the usual scientific canons” (p. 98) (although he does not formulate any such canons, or indicate whether they include the type of inference in point (3)). But if NOA does endorse points (1)-(3), then, I suggest, it endorses a form of scientific realism rejected by scientists of the sort Perrin was opposing who are generally classified as antirealists.[143] These scientists, including Duhem, Mach, Ostwald, and Poincare, claim either that point (1) is false or else that, if it is true, it is unknowable by empirical means.
(They do not regard propositions about unobservables on a “par with more homely truths.”) And they reject points (2) and (3), since they raise general methodological objections to inferences from what is observed to what is unobservable.Moreover, if an antirealist says either that unobservables do not exist or that, if they do, they are empirically unknowable, and if he is an empiricist about science, then, as in the case of Duhem and van Fraassen, he is likely to deny that the aim of science is to provide true theories about unobservables. This is because he holds that such an aim cannot be satisfied at all or cannot be satisfied by empirical means. Although this position does not follow deductively from a denial of point (1 )-(3), it is a natural claim to make if one wants to retain empiricism in science and hold that the aim of science can be achieved. Similarly, if one is a realist and states that unobservables do exist and that claims to know this in particular cases can be empirically justified, then it is natural to assert that the aim of science, or at least one of its aims, is to provide true theories about such unobservables.
The most important reason that antirealists such as Duhem and van Fraassen have for saying that the aim of science is (a) to provide theories that save the phenomena, rather than (b) to provide theories that are true, is that they regard (a) but not (b) as doable and empirically justifiable. By contrast, realists such as Perrin and Newton regard both (a) and (b) as doable and empirically justifiable. Since they do, and since they also regard unobservables (such as molecules and universal gravity, which Newton considered a force extended to bodies “beyond the range of the senses”) as causally responsible for observable phenomena (such as Brownian motion and motions of the planets), and since they regard particular causal claims of the latter sort as justifiable empirically, it would be natural for them to hold that:
(4) One of the aims of science is to provide (approximately) true theories of what the world is like.
where truth applies to unobservables as well as observables. Point (4) is indeed strongly suggested by the methodological remarks of Newton (for example, in his Rules 1 and 4)[144] and Perrin (in his intuitive method). Accordingly, since it is not entailed by realist points (1 )-(3), it is reasonable to consider it part of the realist position. Its truth is not demonstrated by Perrin's empirical argument for molecules, but it is plausible to say that Perrin presupposed its truth in conducting his investigation into the cause of Brownian motion.
The same cannot be said, however, for the doctrine of realism defended by Boyd and Psillos and attacked by Laudan. Points (1 )-(4) make no claims about whether, in general, scientific theories in the mature sciences are true, contain terms that refer to objects that exist, and describe a mind- and theory-independent world; such a realism is much stronger than (1)-(4). Nor were any such general claims about theories in the mature sciences made or presupposed by Perrin in his investigation (or by Newton in his argument for the existence of a universal gravitational force).
For scientific realism of the sort that was supported by scientists such as Perrin and rejected by his opponents, points (1 )-(4) suffice. These scientists claimed (or denied) that there is an empirically knowable realm of unobservables responsible for observable effects, and since there is, one of the aims of science is to provide true theories about this realm. Whether any particular theory is true is to be determined by empirical considerations specific to that theory. From the fact that the theory is part of a mature science one cannot infer that it is true or probable. Indeed, as Laudan has argued, historically many such theories have been empirically refuted. For scientific realism of the kind under attack by Duhem, Mach, and other scientists (as well as by van Fraassen), and defended by Perrin (and by Salmon), what is important is the idea that there is a realm of unobservables, claims about which can be empirically justified as true. Whether unobservables do exist, and if so which ones, and what properties they have, are issues to be determined by empirical arguments of the sort Perrin provided for molecules. Accordingly, I regard Salmon's conclusion as justified. In a historically and conceptually important sense of “scientific realism” (though not in every sense assigned to that term), Perrin's experimental argument for molecules provides and empirical basis for scientific realism.