Objectivity over objects: A case study in theory formation (Q5959578)

There has been a considerable change in what could be called Californian philosophy of set theory in the 1990s from platonistic positions as presented by \textit{P. Maddy} [Realism in mathematics (Clarendon Press, Oxford) (1992; Zbl 0762.00001)] to naturalistic positions. The most obvious example of this is Maddy's naturalistic shift in her book [Naturalism in mathematics (Clarendon Press, Oxford) (1997; Zbl 0931.03003)]. One of the many reasons for this shift is -- to quote Tony Martin -- ``a strong intuition that objects do not really matter'' [\textit{D. A.\ Martin}, ``Mathematical evidence'', in: H. G. Dales (ed.) et al., Truth in mathematics (Clarendon Press, Oxford), 215-231 (1998; Zbl 0936.03005)]. Without an ontology for mathematics, the simple correspondence-theoretic answer to the question ``What is mathematical truth?'' isn't available. In a naturalistic approach, you can either (as Maddy does) stop talking about ``truth'' of mathematical statements altogether, and ask about ``acceptance'' vs.\ ``rejection'' instead; or, you can follow authors like Martin, Steel and Hauser and use naturalistic means to determine mathematical truth. In both cases, the naturalist tries to find criteria for acceptability or truth by looking at mathematical (research) practice. Answering such an important question by looking at the practice in the research community yields the danger of transforming philosophy of mathematics into sociology of mathematics, thereby replacing a conceptual analysis of mathematics by a statistical analysis of the attitudes of mathematicians. The mentioned authors, though there are important differences in their approaches, claim that this is not the case: research practice is not irrational, but allows a conceptual analysis; consequently, describing research practice is not just an empirical endeavour of sampling the opinions of mathematicians, but has a genuinely mathematical component that can be expressed in criteria or rules of thumb. (For a detailed discussion of how to describe the mathematical component of research practice, see Chapter III.6 of Maddy's Naturalism in mathematics [loc. cit.].) Hauser goes beyond the purely naturalistic standpoint by claiming that mathematical truth transcends the criteria he gives: ``A naive answer would be that these criteria determine what truth \textit{is}. Clearly, this is utterly inadequate. The application of these criteria allows for the possibility to arrive at mutually inconsistent theories, leaving open which one of them is \textit{true}'' \ (p.~267). Hauser argues ``that the practice of contemporary set theory ...\ is fully compatible with an epistemological realism'' (p.~271). He claims that ``the existence of a mind independent realm of mathematical \textit{objects} plays no role in defending this position. What matters is the \textit{objectivity} discernible in theory formation, i.e., the possibility of a rational evaluation of \textit{evidence} for and against strong hypotheses'' (p.~272). As the subtitle of the paper suggests, Hauser investigates modern developments in higher set theory as a case study to argue for the compatibility of set-theoretic practice with epistemic realism. This has been done in Martin's paper [loc. cit.] for the cases of the cone lemma and the Wadge lemma. Hauser concentrates on the development of an attractive and reasonably complete theory of the projective sets of reals under the Axiom of Projective Determinacy, and \textit{W. H. Woodin}'s new results on the Continuum Hypothesis as discussed in Notices Am. Math. Soc. 48, 567-576 (2001; Zbl 0992.03063) and ibid. 48, 681-690 (2001; Zbl 1047.03041). From these and other examples he isolates and exemplifies criteria for acceptance of axiom candidates, e.g., inevitability, asymmetry, systematize, unify, identity through differences, illumination, prediction, and confirmation, and discusses their rôle in the process of theory formation in higher set theory.