Interventionist Causal Exclusion and Non‐reductive Physicalism

Abstract

The first part of this paper presents an argument showing that the currently most highly acclaimed interventionist theory of causation, i.e. the one advanced by Woodward, excludes supervening macro properties from having a causal influence on effects of their micro supervenience bases. Moreover, this interventionist exclusion argument is demonstrated to rest on weaker premises than classical exclusion arguments. The second part then discusses a weakening of interventionism that Woodward suggests. This weakened version of interventionism turns out either to be inapplicable to cases of downward causation involving supervening macro properties or to render corresponding causal claims meaningless. In sum, the paper argues that, contrary to what many non‐reductive physicalists claim, interventionism does not render non‐reductive physicalism immune to the problem of causal exclusion.

Acknowledgements

I am grateful to Delphine Chapuis, Mehmet Elgin, Derek Turner, and Jim Woodward for very helpful comments and discussions. Moreover, I thank two anonymous referees of this journal for valuable comments on an earlier draft. Finally, I am indebted to the Center for Philosophy of Science of the University of Pittsburgh and to the Swiss National Science Foundation for generous support of this work (grant PBBE1 – 117031).

Notes

[1] For recent versions of the problem cf. Kim (2003) or Kim (2005). Moreover, while the problem was originally seen to only threaten the special case of mental‐to‐physical causation, it has meanwhile been generalized for any kind of macro‐to‐micro causation (cf. Bontly 2002; Kim 2005, 55). Finally, note that a causal and an explanatory version of the exclusion problem can be distinguished (cf. Sabates 2001). This paper will only be concerned with the causal variant. For a persuasive overview over different forms and readings of exclusion arguments cf. Walter (2008).

[2] Cf. e.g. Yablo (1992), Jackson (1996), Horgan (2001), or very recently List and Menzies (n.d.).

[3] In modal systems containing the Barcan Formula (BF) ◊∃xFx is even equivalent to ∃x◊Fx. For details see e.g. Hughes and Cresswell (1996, 246). Irrespective of whether BF is presupposed, I take a de dicto reading of (M) that sidesteps metaphysical questions as to the manner of existence of possibilia to be more in line with the basic non‐metaphysical approach followed in Woodward (2003). The subsequent discussion of the interventionist exclusion argument, however, in no way hinges on this preference of a de dicto reading of (M).

[4] Strictly speaking, of course, (a), (b), and (c) bring about differences in the logical form of the analysis of direct causation contained in (M). For brevity, I subsequently simply speak of the logical form of (M) in this context.

[5] While Woodward, in various passages (e.g. Woodward 2003, 112–113), explicitly characterizes the manipulability of X with respect to Y as a necessary condition of X causing Y, the fixability of the remaining variables is never explicitly identified as necessary condition for causation.

[6] For such ‘neuron diagrams’, which have meanwhile become a classical source of exemplary causal structures, cf. Lewis (1986).

[7] As Woodward (2003, 82–85) shows, his account of token‐level causation is compatible with systematic overdetermination as well.

[8] To be clear, Woodward (2008) does not himself claim to be weakening or even modifying his original theory. Yet, as shall be substantiated below, that is what he seems to be doing nonetheless—very implicitly.

[9] Woodward indicates that, for simplicity, (M ^s ) only accounts for deterministic causal dependencies. This restriction, however, is of no relevance to our current purposes.

[10] To be consistent with the notation used in this paper I change ‘(M)’ to ‘(M ^s )’ in this quotation.

[11] To make this contrast between the original and the weakened version of Woodward's theory most transparent, compare the passage from Woodward (2008) quoted above with Woodward (2003, 112–113), where manipulability is characterized as necessary condition for causation.