Causation, Projection, Inference, and Agency [chapter]

Helen Beebee
2015 Passions and Projections  
The world, or rather that part of it with which we are acquainted, exhibits as we must all agree a good deal of regularity of succession. I contend that over and above that it exhibits no feature called causal necessity, but that we make sentences called causal laws from which (i.e. having made which) we proceed to actions and propositions connected with them in a certain way, and say that a fact asserted in a proposition which is an instance of causal law is a case of causal necessity. F. P.
more » ... msey, 'General Propositions and Causality' (1929, 160) is the thought that what is projected is a habit of inference. But inference runs in both directions: we infer causes from effects just as happily as we infer effects from causes. Hume himself solves this problem by stipulation: causes precede their effects. But in the absence of a satisfactory account of why causes precede their effects -and evidential relations on their own would appear to provide no such account -Hume's stipulation appears at best ad hoc; and at worst it stands a fighting chance of simply being empirically false. A second strand of projectivist thinking about causationwith its roots in the work of Frank Ramsey, and given a much fuller articulation by Huw Price -aims to deliver asymmetry by conceiving causation as a projection not of our habits of inference but of our perspective as deliberating agents. 3 Conceiving the question of a viable projectivist account of causation as a straight choice between inference and agency, however, masks two broader commonalities. First, the Ramsey-Price view, like the Hume-Blackburn view, takes causation to be an evidential relation. What distinguishes the former from the latter is not the eschewal of evidential relations as the way to understand causation, but the claim that it is evidential relations of a certain sort -viz, those that stem from the perspective of the agent -that characterise causation. Second, there is a deep connection between both views on the one hand and a long-running theme in the literature on conditionals on the other. Causation is, of course, a conditional relationship in a sense: effects generally depend on their causes, and there is a long-standing tradition within Humeanism of understanding causal laws ('As cause Bs') as equivalent to, or at least very closely related to, generalised conditionals ('if something is an A, it will be followed by a B'). Projectivist accounts of causation take the 'truth' of causal claims to be an expression or projection of our 3 Price's global expressivism perhaps makes 'projectivist' an unhappy term to use to describe his account of causation, since (unlike Blackburn) Price does not believe in the kind of representational language with which projectivist talk might be contrasted. On the other hand, Price describes his view of causation as 'perspectival', and that, presumably, is supposed to draw a contrast with something that is non-perspectival (the laws of physics, say), even if the non-perspectival also fails to be representational. (It may be that non-perspectival claims, but not the perspectival ones, are -in Price's sense of the term -'e-representations' (see Price, this volume, §5).) At any rate, supposing that local projectivism is a viable option in principle, the local projectivist about causation can in principle appropriate Price's view on causation to suit her own ends; there is nothing in Price's account of causation that depends on global expressivism. In a sense, then, the Price of this chapter is this fictional character and not the real Price; but given my purposes this should not matter very much. attitudes: the conditional structure of reality is in some sense no more than the conditional structure of our thought ('if I see an A, I shall regard it as grounds for expecting a B', or perhaps 'if I do a, b will follow'). But there is also a long tradition, tracing back to Ramsey himself, of holding that indicative (and counterfactual) conditionals themselves lack truth conditions: to be confident that if X, then Y is to be confident that Y on the supposition that X. One's degree of confidence in the latter can be represented as Pr(Y/X), but this degree of confidence does not amount to a degree of confidence that some proposition is true. We might, then, attempt to extract a projectivist account of causation from a 'projectivist' account of conditionals: we conceive of causes as grounds for inferring effects because the conditionals we sign up to in conceiving causes in this way are themselves expressions of our inferential commitments (they are, to use Ryle's expression, 'inference tickets') rather than statements of objective conditional facts. (This is not a new proposal; indeed, I take it to be pretty much Ramsey's view.) One virtue of proceeding in this manner is that it delivers independent motivation for what I suspect many philosophers regard as a deeply unattractive aspect of projectivism about causation: the fact that it renders our causal talk nontruth-apt or non-fact-stating (at least on a realist, as opposed to a quasi-realist, understanding of the notions of 'truth' and 'fact'). Projectivists themselves, of course, consider this cost, if they regard it as a cost at all, to be worth the benefit, where the benefit is broadly metaphysical. We can have our cake, constructed only from the delectably sparse ingredients of the Humean mosaic, and eat it too -that is, we can agree with the realists that causal talk and thought are crucially important to our cognitive and practical lives and that to advocate the eschewal of causal vocabulary, or to analyse it away, would be a mistake. The benefit of regarding conditional discourse as non-fact-stating, by contrast, lies not in its metaphysics but in its logic and epistemology -relatively neutral territory, in other words, from the point of view of the hotly contested battle lines that are fought over by projectivists about causation and their 'objectivist' opponents (though it may be that objectivists about causation will want to resist the claim that conditionals are non-fact-stating too -indeed, they better had if they agree that causation should be analysed in terms of conditionals, as subscribers to the counterfactual analysis do). Having sketched the Hume-Blackburn and Ramsey-Price versions of projectivism about causation and explained the latter's solution to the worry about causal asymmetry in §2, in §3 I discuss accounts of conditionals that regard them as 'rules' or 'inference tickets' rather than propositions, and explain the connection between degrees of confidence in conditionals on the one hand and conditional probabilities on the other. I then flesh out the suggestion made above, that a projectivist account of causation can in principle be premised on the kind of account of conditionals just described, and explain how, in that light, the fundamental difference between the Hume-Blackburn and Ramsey-Price positions can be seen as a difference in attitude to conditional probabilities. Since the Ramsey-Price version of this view can, and the Hume-Blackburn version cannot, account for the asymmetry of causation, the Ramsey-Price version is preferable. In §4, I say something about what divides the kind of conditional projectivist account of causation discussed earlier and the kind of conditional objectivist account held by defenders of the counterfactual analysis of causation -both of which agree on (or at least can be used to serve the cause of) a broadly Humean view of what there is. In §5, I briefly sum up. Two versions of projectivism, and their ancestry Hume locates the impression-source of the idea of necessary connection in the inference we draw from causes to effects -an inference that is itself a matter of 'Custom or Habit' (1748/51, 43): having observed Cs being followed by Es on sufficiently many past occasions, on observing a new C I come to expect an E to follow. The central question for interpreters of Hume has been what the implications of this discovery are supposed to be for our idea of causation. Hume appears to hold that, given its impression-source, the idea of necessary connection cannot represent any real, mind-independent relation between causes and effects. But he shows no inclination to retract his earlier claim that the idea of necessary connection is a component of the idea of causation. Nor does he hold that causal claims are claims about our own inferences: in saying that c caused e, I am not merely reporting that on observing c I came to expect e. Nor does he reject our causal talk as unintelligible or false: he provides us with 'rules by which to judge of causes and effects ' (1739-40, 173) and makes ample use of causal claims himself. How, then, can Hume simultaneously hold, as he appears to, that the idea of necessary connection represents neither a feature of us -the 'determination of the mind to pass from one object to its usual attendant' (1739-40, 165) -nor a feature of the external world; that that idea is an essential component of our idea of causation; use of the term 7 ) expresses a 'cognitive attitude', viz, one's reliance on the inference from As to Bs: a commitment, that is, to regarding future As that I meet as Bs.
doi:10.1093/acprof:oso/9780198723172.003.0002 fatcat:ygs7qy57bredzkvox23huraovy