No CrossRef data available.
We use cookies to distinguish you from other users and to provide you with a better experience on our websites. Close this message to accept cookies or find out how to manage your cookie settings.
We use cookies to distinguish you from other users and to provide you with a better experience on our websites. Close this message to accept cookies or find out how to manage your cookie settings.
Published online by Cambridge University Press: 01 January 2020
Suppose that a scientific theory X is well confirmed, and either states or (in conjunction with certain specified systems of logic, mathematics, and semantics) implies a statement like “classes exist”. Suppose, if the statement is not explicit but implied, that the specified systems are ‘accepted’ by two individuals, R and F, as the ‘best available'. Suppose, finally, that both R and F, in some yet to be explained sense of the word, ‘accept’ and use theory X to regulate their experience. Then we have something very like the situation discussed by Putnam and van Fraassen in their debate over ‘fictionalism'. I will argue that, in this situation, there is a great mystery over what would separate a fictionalist F from a realist R. Neither Putnam nor van Fraassen seems to be conscious of the problems involved.
Fictionalism, according to its opponent, Putnam, states
various entities presupposed by scientific and common sense discourse [are] merely “useful fictions”, or that we cannot, at any rate, possibly know that they are more than “useful fictions” (and so we may as well say that that is what they are) (PL 63).
1 Putman, Hilary Philosophy of Logic (New York, 1971)Google Scholar (henceforth PL), and Fraassen, Bas van critical notice of ibid, Canadian Journal of Philosophy 4 (1973-4), pp. 731-44CrossRefGoogle Scholar (henceforth VF).
2 This is not quite accurate. Some theories may turn out to be neither true nor false because they fail of certain presuppositions: for example, in the case of classical mechanics, the notion that all bodies have a determinate observable position. These presuppositions, however, can be regarded as assumptions of the theory, in which case the theory will be false.
3 Some of the ‘paradoxes of confirmation’ may be thought to cast doubt on the assumption that what is evidence for X is evidence for X's logical implications. But I do not see how anything could count as a solution to these paradoxes unless it determined a large class of implications which preserved the confirming power of evidence. There would seem, moreover, to be an overwhelming likelihood that (a) some such implications would be versions of X, and (b) some such versions will remain unasserted, or will be asserted only as parts of some more general theory, and therefore will not be disconfirmed by historical evidence. So objections based on the paradoxes seem to be weak. They would certainly be weak, for example, in the case of Newton's laws. One might even argue that historical experience confirms that the evidence for a theory is also evidence for some of its versions.
4 An apparent alternative is to maintain that the theory is probable only in the sense that it has true test-implications (not in the sense that it is probably true), and that confirmation secures probability. But, in the first place, test implications are often theory-laden and, especially if they involve numbers, full of realist implications. There is, then, some objection, from the fictionalist's point of view, to regarding them as true. Second, this ‘solution’ seems only to defer the question. Why should we care whether a theory has true test-implications? Because this brings us closer to some truth, or only because it means we've found a useful fiction? If probability isn't probability of truth, what is it probability of?
5 A similar problem arises in the case of purely formal theories. One branch of mathematics, for example, may be far more useful than another, yet the two branches may be equally acceptable. If mathematics and logic are acceptable merely as useful fictions, how can this be?
6 I would like to thank Karl and Doug Rautenkranz for their comments on early drafts of this paper.