Mathematics as a quasi-empirical science (Q850486)

In his paper, Gianluigi Oliveri discusses whether Imre Lakatos' Methodology of Scientific Research Programmes (MSRP) can be applied to mathematics (Lakatos presented his MSRPs in [\textit{I. Lakatos}, The methodology of scientific research programmes. Philosophical papers. Vol. 1. Edited by John Worrall and Gregory Currie. (Cambridge) etc.: Cambridge University Press. (1978; Zbl 0373.02002)]). Lakatos' MSRP describes a research programme as consisting of a ``hard core'' of assumptions that are held to be irrefutable facts, a ``protective belt'' of auxiliary hypotheses, a ``positive heuristic'' that prescribes which questions are to be investigated and apparent anomalies should be accommodated and turned into examples, and a ``negative heuristic'' that includes methods for explaining away anomalies. A series of theories originating from a common hard core and belonging to a scientific research programme is then, according to Lakatos, defined to be ``progressive'' if it successfully extends the empirical content, and ``degenerative'' if its empirical content is stagnating or decreasing. Lakatos' MSRP was a refinement on the theories of Thomas Kuhn and Karl Popper describing change and progress in the natural sciences. In fact, Lakatos did not explicitly have mathematics in mind, but the author argues that such an application is indeed possible provided that a few minor changes are made to Lakatos' central concepts such as ``hard core'', ``protective belt'', ``positive and negative heuristics'', and ``theoretical and empirical progress'' of scientific research programmes. As an illustration, he then analyzes the development of set theory from Georg Cantor's original ``naïve set theory'' (NST) around 1880 through Ernst Zermelo's reformulation of set theory (Z) in 1908 culminating in the axiom system known as Zermelo-Frankel set theory (ZFC) widely used today. An essential part of Lakatos' MSRP concerns the characterization of ``progressive programmes'' by rational and cognitive means, i.e. without recourse to sociological explanations. In particular, Lakatos devised the criteria of ``theoretical and empirical progress'' to capture the ability of a new theory to predict new theoretical knowledge (theoretical progress) that could be corroborated (empirical progress). As these concepts of progress lend themselves most naturally to experimental science, some authors have argued that Lakatos' MSRP is unfit for mathematics (for instance, the historian and philosopher of mathematics Teun Koetsier has argued this point of view in his analysis of Lakatos' philosophy of mathematics [\textit{T. Koetsier}, Lakatos' philosophy of mathematics. A historical approach. Stud. Hist. Philos. Math., 3, North-Holland, Amsterdam etc. (1991; Zbl 0743.00017)]. Oliveri disagrees with Koetsier's views and translates Lakatos' two concepts of progress into ``theoretical progress'' (there are results provable in each new theory belonging to the programme which are not provable within its predecessor) and ``factual progress'' (each new theory has a model). In his argument, Oliveri reconstructs the history of set theory accordingly as a sequence of three theories belonging to a mathematical research programme. As a Lakatosian rational reconstruction, Oliveri's account of the history of set theory focuses on changes in the hard core, the heuristics, and the protective belt of set theory. In particular, Oliveri argues that ``Zermelo's concept of set in Z is not the product of a [\dots] conceptual refinement [\dots] whereby, as a consequence of being confronted with a large number of exceptions, Cantor's definition is eventually brought to relate only to a smaller range of objects than it was originally supposed to relate to and is replaced by a more general definition of set'' (p.~64). Such conceptual refinements had been central to Lakatos' thought about concept formation in the \textit{Proofs and Refutations} [\textit{I. Lakatos}, Proofs and refutations. The logic of mathematical discovery. Cambridge University Press, (Cambridge) (1976; Zbl 0334.00022)]. Conceptual refinement in mathematics was also important to Koetsier's analysis against which Oliveri argues. The \textit{Proofs and Refutations} is, altogether, ignored in Oliveri's analyses which, instead, provide a different account of a non-cumulative aspect of mathematics by claiming discontinuous changes in the hard core of a research programme. However, Oliveri does not explicitly address how the central concepts of successive theories relate to each other when such changes in the hard core are accepted. Oliveri's rational reconstruction is based on readings of primary literature and draws on a similar project undertaken by \textit{M. Hallett} in the 1980s [Br. J. Philos. Sci. 30, 1--25 (1979; Zbl 0438.03004) and 135--159 (1979; Zbl 0438.03005); Cantorian set theory and limitation of size. Oxf. Logic Guides, 10, Clarendon Press, (Oxford) (1986; Zbl 0656.03030)]. However, nothing of the historical and philosophical literature published in the last decade is taken into account, although for instance Cantor's set theory has recently been the subject of insightful research (for instance by José Ferreirós [\textit{J. Ferreirós Domínguez}, Labyrinth of thought. A history of set theory and its role in modern mathematics. Sci. Networks, Hist. Stud., 23, Birkhäuser, (Basel) (1999; Zbl 0934.03058)] and others). This leaves the impression that the historical account is, indeed, a rational reconstruction and the reader may be left with some doubts as to its degree of historical correctness in the details. Lakatos, himself, would not have minded some historical inaccuracies. In fact, in the \textit{Proofs and Refutations}, Lakatos also presented a philosophically reconstructed history with the remark that ``the real history will chime in in the footnotes'' (\textit{Proofs and Refutations}, p.~5). As for Oliveri and notes, he did not apply these for providing the ``real history'' but put much material in them that could profitably have been incorporated in the main text. As may be evident from these comments, Oliveri's ambition is a philosophical one using a case study from the history of mathematics. The project of showing how Lakatos' MSRP needs only minor adjustments to fit the successive stages in the development of set theory is over-all convincing. However, the title ``mathematics as an quasi-empirical science'' is only implicitly addressed, and the paper could perhaps have been more appropriately entitled ``How Lakatos' MSRP can be applied to set theory''. The reviewer's main objection is that there are deep philosophical questions concerning the purported non-cumulative nature of concept-change that deserve further elaboration. All-in-all, the paper is a nice exposition, although an updated bibliography and small revisions to the structure of the article could have been beneficial.