Hume's Defence of Causal Inference (review)

Hume Studies Volume XXVI, Number 2, November 2000, pp. 350-353 FRED WILSON. Hume's Defence of Causal Inference. Toronto: University of Toronto Press, 1997. Pp. xii + 439. ISBN 0-8020-4158-2, cloth, $80.00. The aim of Fred Wilson's detailed treatment of Hume's theory of causal inference is to show that although Hume posed a skeptical challenge concerning the justification of beliefs through causal inference, Hume also presents a solution to that challenge, a solution that includes a justification of causal inference. Wilson's Hume is not a skeptic, but rather a vindicator of the norms of scientific inquiry. This study is wide-ranging, both in terms of the texts Wilson interprets and the connections he draws to the views of other philosophers , both modern and contemporary. Hume's Defence of Causal Inference is divided into three large chapters. In the first chapter, Wilson attacks interpretations of Hume that treat Hume as a Kantian about causal inference and mental activity more generally. Instead of taking the causal principle "same cause, same effect," as an a priori truth, on Wilson's reading, philosophical "practice and discipline" (39) can support its empirical discovery. Wilson cites Hume's rules for judging causes and effects as well as Hume's treatment of identity in support of what he calls the case for the "active mind." Wilson contrasts his view with the interpretations of Lewis White Beck and Robert Paul Wolff, whose Kantian readings of Hume Wilson couples with the claim that the mind is passive. The second chapter bears the book's title, and contains the core interpretation . While Hume provides neither "proof nor guarantee of the principle of causation, the principle is morally certain" (192). Much of this chapter is devoted to showing how Hume's account of probable inference can be understood within the framework of Bayesian probability theory. According to Wilson, Hume's prior probabilities begin with the beliefs of the vulgar. Reflection leads to the refinement of belief, and support for a "wide network of causal processes." Philosophical reflection on the success of the network supports the principle of universal causation itself. Wilson writes, ".. . the evidence that we have from the successful discovery of laws in many areas implies on Bayesian grounds that the principle of universal causation is, as Hume and Mill claim, morally certain" (183, italics Wilson's). In the final chapter Wilson takes up Hume's skeptical argument concerning reason from Book I, part 4, section 1 of the Treatise. Having defended a non-skeptical reading of part 3 of Book I of Hume's Treatise, Wilson addresses the question of whether Hume's skepticism about reason in part iv is real or apparent. The chapter includes a long discussion of Hume's treatment of testimony , particularly Hume's attention to the way reflection on testimony leads to long chains of inference to historical fact. Explicating Hume's views on testimony , Wilson suggests, sheds light on the skeptical argument of "Of scepticism with regard to reason." Hume Studies Book Reviews 351 Although this is a work on causal inference, Wilson fans out from the central concern to a host of related issues. Among the topics that receive treatment are Hume's account of the experimental method, identity, and, as just noted, testimony. These issues are broached through consideration of both historical and contemporary sources. Wilson is as likely to discuss the views of Kuhn or Russell as he is those of Descartes or Locke. Wilson takes great pains to employ the tools of modern logic to explicate Hume's views, with varying degrees of success. One area of concern is the symbolization of causal relations. Wilson uses the arrow of material implication and the double arrow of material equivalence to represent causal relations. For example, Wilson's gloss on 'soluble' means by definition 'if in water then dissolves' is 'x is soluble' means 'x is in water -> χ dissolves.' (32) If the arrow is the symbolization for material implication, then Wilson's formalization suffers from a number of well-known defects, the most important of which is that in material implication a conditional is true whenever the antecedent is false. So any χ which is...