Constructive Empiricism and the Vices of Voluntarism

Abstract

Constructive empiricism – as formulated by Bas van Fraassen – makes no epistemological claims about the nature of science. Rather, it is a view about the aim of science, to be situated within van Fraassen’s broader voluntarist epistemology. Yet while this epistemically minimalist framework may have various advantages in defending the epistemic relevance of constructive empiricism, I show how it also has various disadvantages in maintaining its internal coherence.

Keywords:

• constructive empiricism
• epistemic voluntarism

Notes

This paper was originally given to the Philosophy Workshop at the Department of History and Philosophy of Science, University of Cambridge; my thanks to those who gave comments and suggestions, especially Alex Broadbent, Elly Kingma, Steve John and Mark Sprevak. I would also like to thank the Master and Fellows of Churchill College, Cambridge, where I completed this work as a Research Fellow.

1 It should be noted, of course, that van Fraassen’s epistemic voluntarism consists of more than the justificatory negative elements sketched above; it also consists of a positive element concerning the role of values and pragmatics within our epistemic judgments, and the sort of commitments one undertakes in making such judgments – these are particularly well illustrated in the context of van Fraassen’s defence of the Reflection Principle (van Fraassen, 1984; 1995) and in his articulation of an epistemic stance (van Fraassen, 2002). However, since these positive elements will have no bearing on my subsequent argument – which concerns the internal coherence of the constructive empiricist’s position – I shall be concerned exclusively with the negative elements of the position in what follows.

2 See, for example, Hacking, 1985 and Churchland, 1985 respectively.

3 For recent discussion of the pros and cons of van Fraassen’s epistemology, see Monton, 2007: in particular, Ladyman, 2007 argues that such a minimalist epistemic framework collapses any distinction between realism, empiricism and scepticism; Chakravartty, 2007 argues that such a framework leads to some unfortunate forms of relativism; and Psillos, 2007 raises some doubts as to whether such a framework captures any intuitive notion of rationality. See also van Fraassen’s (2007) response in the same volume.

4 For van Fraassen’s original response to Musgrave, see van Fraassen, 1985: p. 256. For various interpretations of van Fraassen’s response, see Kukla, 1998: pp. 138–9), Muller, 2004 and Dicken and Lipton, 2006.

5 See Dicken, 2009.

6 An anonymous referee for this journal has suggested to me that the constructive empiricist could simply ignore this difficulty, arguing instead that science itself only licenses beliefs about the unobservability of actual phenomena, and that therefore it is not required of the constructive empiricist to make up this alleged doxastic deficit. Such a response has some merit, although I think it’s fair to say that this would be both a descriptively implausible account of scientific practice, and straightforwardly at odds with van Fraassen’s own understanding of constructive empiricism (it would after all lead to the unattractive position of having an epistemic policy that forbade belief in the consequences of our scientific theories that were about actual electrons, but which could manage no opinion whatsoever about what we should believe about an additional electron, had it existed!). In any case, in order to defend such a view – which simply states that the constructive empiricist need not accommodate a range of beliefs usually considered as a basic desideratum of a philosophical theory – the constructive empiricist would be forced to appeal to the minimal epistemic constraints licensed by a voluntarist epistemology, avoidance of which (as we shall see) is the primary motivation for such a move.

7 The point, of course, is that while ‘being a sphere with a diameter greater than 10 miles’ is clearly not a modal property in the same way that ‘being observable’ is, Muller and van Fraassen appear to have no grounds upon which to make this distinction. The constructive empiricist must show how he can accommodate the range of beliefs necessary for the internal coherence of his position, given that such beliefs are not entailed by his belief in the empirical adequacy of his scientific theories – and if that challenge requires him to stipulate the modal scope of his beliefs, he must justify why it is that this stipulation only covers some beliefs and not others. To claim that some properties are modal and others not is merely to state the problem, not to solve it.

8 An anonymous referee for this journal has suggested to me that the constructive empiricist could justify his amended epistemic policy on the grounds that since constructive empiricism aims to capture scientific practice, and since scientists themselves infer their beliefs to be about more than the actual observable phenomena, such an amendment simply offers a more accurate description of science and its standards. However, the issue of course is not about the accuracy of the constructive empiricist’s description so much as it is about the internal coherence of the constructive empiricist’s position – with how he can be entitled to this amendment to his epistemic policy when it clearly goes beyond that which is entailed by the constructive empiricist’s belief in the empirical adequacy of his scientific theories.

9 Monton and van Fraassen, 2003: p. 406; Ladyman, 2004: pp. 763–4. As Ladyman points out, however, there may well be something incompatible between modal realism and van Fraassen’s conception of empiricism, of which his constructive empiricism is an integral component: not only does van Fraassen (2002) characterize empiricism as a sceptical stance towards metaphysics, but in some cases he even seems to suggest that the denial of objective modality is a definitive component of empiricism (e.g. van Fraassen, 1977).

10 This of course raises another worry, this time concerning the status of these models, which according to van Fraassen (1980: pp. 64–9; 1989: pp. 217–32) are abstract, mathematical objects. Indeed, one can raise here an analogous problem about mathematics for the constructive empiricist to those Musgrave and Ladyman raise for unobservables and counterfactuals respectively. For the objection that the constructive empiricist must adopt an attitude towards abstract, mathematical objects that is inconsistent with his view concerning the aim of science, see Rosen, 1994; for an attempt to secure the constructive empiricist’s commitments along more parsimonious (mathematical fictionalist) lines, see Bueno, 1999; for the argument that such a strategy falls short of satisfying the constructive empiricist’s (meta‐logical) needs, see Dicken, 2006.

11 Moreover, there may be something of an ad hominem lurking here, since Ladyman (Ladyman and Ross, 2007) pursues his philosophy of science against the backdrop of a highly naturalized metaphysics. There is thus a methodological tension to be found between his view that we should essentially allow contemporary physics to settle our metaphysical disputes, and the metaphysically loaded criteria he brings to bear in his criticism of Monton and van Fraassen’s deflationary account of counterfactuals. The extent to which one finds the meta‐linguistic approach unsatisfactory will be determined by the sorts of pre‐scientific cost‐benefit analysis of ontological commitment that one brings to the debate, the sort of thing championed by David Lewis and supposedly renounced by the naturalistic metaphysician. Indeed, a purely philosophical debate over the attractive features of a theory of modality is something that floats quite freely of any naturalistic constraint; and no debate over the appropriate semantics for counterfactuals is going to make any headway in a unified, naturalistic metaphysics. For more, see Dicken, 2008.