Rule Following, Error Theory and Eliminativism

Abstract

In this paper, I argue for three main claims. First, that there are two broad sorts of error theory about a particular region of thought and talk, eliminativist error theories and non-eliminativist error theories. Second, that an error theory about rule following can only be an eliminativist view of rule following, and therefore an eliminativist view of meaning and content on a par with Paul Churchland’s prima facie implausible eliminativism about the propositional attitudes. Third, that despite some superficial appearances to the contrary, non-eliminativist error theory does not provide a plausible vehicle for understanding the ‘sceptical solution’ to the sceptical paradox about rule-following developed in Saul Kripke’s Wittgenstein on Rules and Private Language.

Keywords:

• rule-following
• meaning
• intentional content
• error theory
• eliminativism
• Kripke’s Wittgenstein

Acknowledgements

This paper was written while I was a visiting scholar at the University of Vienna in June 2014: I’m grateful to Martin Kusch and his colleagues for their hospitality. For helpful comments and discussion I’m grateful to Jiri Benovsky, Ulf Hlobil, Martin Kusch, Charles Pigden, Ali Saboohi, Olivia Soltanescu, Kirk Surgener, Joss Walker, Zach Weber and Daniel Wee. I’m also grateful to two anonymous referees for this journal, and to a seminar audience at the University of Otago.

Notes

1 Crispin Wright (1984) argues that both Quinean eliminativism about meaning and non-factualism about meaning are potentially incoherent. Wright’s arguments strike me as very powerful, but as far as I can see he does not (in Wright, 1984) explicitly consider error theories of the sort we are primarily concerned with in this paper. In Wright, 2002, he uses ‘eliminativism’ to mean ‘error theory’ and doesn’t distinguish between what I am calling eliminativist and non-eliminativist variants of error theory. See also section 5 below.

2 Paul Boghossian (1990, p. 159) draws a similar distinction between ‘eliminativist’ and ‘instrumentalist’ error theories. I prefer to avoid the ‘instrumentalist’ label in this context, however, to avoid confusing error theories with forms of non-cognitivism or non-factualism (according to which the relevant class of judgements are not truth-apt [as opposed to false]).

3 As Field (1980, p. xi) puts it, ‘mathematical entities are not theoretically indispensable’.

4 It might be objected that Kripke’s Wittgenstein’s sceptic only denies that there are facts about which norm an agent is following, not that there are facts about norm compliance: after all, doesn’t he grant that there is such a thing as compliance with the addition rule and compliance with the quaddition rule en route to concluding that there is no fact of the matter as to whether an agent is following one rather than the other? However, this is to forget that the notion of compliance was only granted pro tem in order to allow the sceptical argument to get underway: having established that there are no facts about rule following, Kripke (1982, p. 55) concludes (adverting to Wittgenstein, 1958, § 201) that ‘there can be neither accord, nor conflict’. The idea of a rule with determinate compliance conditions that cannot (even in principle) be followed is clearly regarded by Kripke as paradoxical.

5 McGinn interprets Field 1980 as proposing a kind of non-cognitivist view on which mathematical sentences don’t have genuine truth-conditions. It seems to me, though, that Field is actually proposing a view that is a form of cognitivist non-eliminativist error theory.

6 Wright 1984 and Hale 1997 also construe the sceptical solution as proposing a form of non-factualism or non-cognitivism about meaning.

7 Of course, it could be that Kripke intended the sceptical solution as a kind of non-eliminativist error theory, but simply failed to see that an error theory about rule following cannot be non-eliminativist. A charitable reading of Kripke, though, would avoid viewing him as proposing a positive position in WRPL chapter 3 that is untenable in the light of the main conclusions of WRPL chapter 2.

8 For some discussion of whether the sceptical solution might be viewed as proposing a plausible form of non-factualism about meaning, see Boghossian (1990), Wright (1984), Miller (2007) chapter 5, and Miller (2011). Note that Wright (1993) discusses a position that Wright calls ‘concessive eliminative materialism’, but that position is actually a form of non-factualism according to which ascriptions of content have a non-fact-stating role. The eliminative materialism we have been concerned with by contrast assigns ascriptions of content a fact-stating or descriptive role, so that they are viewed as false rather than as failing of truth-aptitude. Some philosophers (e.g., Wilson, 1994) have argued that the sceptical solution should be interpreted as a form of non error-theoretic factualism. For a critique of such readings, see Miller, 2010. It’s an interesting question how the idea of a non-eliminativist error theory, as we have been concerned with it, relates to the ‘conservationist’ position described in Olson, 2014.