Wesley Salmon's Complementarity Thesis: Causalism and Unificationism Reconciled?

Abstract

In his later years, Wesley Salmon believed that the two dominant models of scientific explanation (his own causal‐mechanical model and the unificationist model) were reconcilable. Salmon envisaged a ‘new consensus’ about explanation: he suggested that the two models represent two ‘complementary’ types of explanation, which may ‘peacefully coexist’ because they illuminate different aspects of scientific understanding. This paper traces the development of Salmon's ideas and presents a critical analysis of his complementarity thesis. Salmon's thesis is rejected on the basis of two objections, and an alternative view of the relation between different types of explanation is proposed.

Acknowledgements

Earlier versions of this paper were presented at the International Congress on Causation and Explanation in Natural and Social Sciences in Ghent, Belgium (May 2002), at the 12th International Congress for Logic, Methodology and Philosophy of Science in Oviedo, Spain (August 2003), and in the research group Knowledge, Normativity and Practice (Faculty of Philosophy, Vrije Universiteit Amsterdam). The penultimate version was presented in the EIPE seminar (Department of Philosophy, Erasmus Universiteit Rotterdam) in March 2006. I thank all participants but in particular Caterina Marchionni, for her detailed and valuable commentary. Moreover, I wish to thank Peter Kirschenmann, for suggesting the term ‘superunderstanding’ and for his comments on earlier versions; Daniel Cohnitz, for providing a copy of Peter Railton's thesis; Dennis Dieks, for providing examples from physics; and James W. McAllister and two anonymous referees of this journal, for valuable criticism and suggestions. This research was supported by the Netherlands Organisation for Scientific Research (NWO).

Notes

[1] This definition appears to be generally accepted. Nonetheless, some authors distinguish between theories that ‘merely’ explain and theories that provide understanding (e.g. Cushing 1994, 10–12). For present purposes, this alternative view may be ignored, and understanding will be defined as the product of explanation.

[2] An alternative account of singular causal explanation is given by Woodward (2003, 209–20).

[3] To be sure, Salmon refused to commit himself explicitly to this implication; cf. his statement that there are ‘at least two major aspects’ (Salmon 1998, 76, my italics), and that ‘[p]erhaps there are other [explanatory] virtues that I have not discussed’ (ibid., 78). It remains, however, completely unclear what other aspects or virtues he hinted at—and how these relate to the global/local opposition. Moreover, in the same text, Salmon (1998, 77) referred to an ‘explanatory duality’, which implies the existence of two, and only two, approaches.

[4] Incidentally, Railton's view of causal mechanisms is more liberal, but also more vague, than Salmon's. Railton designates the quantum‐mechanical ‘tunnelling mechanism’ as a causal mechanism, but admits that he is unable to provide a precise definition of what a mechanism is. Thus, he runs the risk of rendering the notion vacuous. Salmon, by contrast, does provide an exact definition (in terms of causal processes and interactions), which however leads to problems with respect to quantum mechanics.

[5] I thank an anonymous referee, for providing this example.

[6] I thank Caterina Marchionni, for bringing this point to my attention.

[7] To be sure, Railton (1981, 243) acknowledges that, when we ask scientists which models ought to be part of the ideal explanatory text, ‘we should be prepared to accept indefiniteness and even disagreement in the answers we receive’. However, this refers to possible controversy about which models count as realistic descriptions, and not to a fundamentally contextual variation in criteria for explanation.

[8] As noted before (footnote 3), Salmon also suggested that more than two types may exist. However, this suggestion conflicts with the idea of complementarity as elaborated in terms of the global–local distinction, which entails that there are only two types.

Additional information

Notes on contributors

Henk W. de Regt

Henk W. de Regt is a University Lecturer at Vrije Universiteit Amsterdam, The Netherlands.