Causation in context: a philosophical and empirical investigation into contextual determinants of causation and causal judgment. Project description Asbjørn Steglich-Petersen, PhD Institute for Philosophy and History of Ideas University of Aarhus [email protected]

Background and purpose of research project This section provides a brief introduction to contextualist theories of causation, identifies some unresolved key issues, and outlines a research project addressing some of these issues. Background Although the role of context in causal judgment and reasoning has been recognised for a long time1, it is only more recently that explicitly contextualist theories of causation have begun to emerge. The recent emergence of such theories is part of a larger contextualist movement in philosophy and linguistics. This section gives a brief introduction to this recent trend in the theory of causation. The basic motivation for contextualist theories of causation can be illustrated by the following example involving a test for counterfactual dependence. Suppose we hypothesize that Suzy’s smoking two packs of cigarettes per day was the cause of Suzy’s lung cancer. The counterfactual test2 says that this is the case if it is true that had Suzy not smoked two packs of cigarettes per day, she wouldn’t have gotten cancer. On the standard semantics of counterfactual statements, this is the case if and only if the scenario closest to reality, in which Suzy doesn’t smoke two packs of cigarettes per day, is a scenario in which Suzy doesn’t have lung cancer. Generally, 1

See in particular Reichenbach 1956; Hart and Honere 1959; Mackie 1965; Dretske 1977; van Fraassen 1980; Unger 1984; Holland 1986. 2 Counterfactual dependence fails as an analysis of causation, but is widely held to be useful as a heuristic test for causal relationships (Schaffer 2005).

1

for any causal claim of the form event c caused event e, the counterfactual test tells us to check whether the scenario closest to reality in which c does not occur, is a scenario in which e occurs or not. But this introduces a source of uncertainty: we are now looking for a scenario in which Suzy doesn’t smoke two packs of cigarettes per day, but we don’t know what she is doing instead. Perhaps she is smoking three packs of cigarettes, or perhaps she is not smoking at all, and naturally, the truth of the counterfactual relationship between her smoking and getting cancer will depend on this. According to the classical semantics of counterfactuals, this question is settled objectively by how much Suzy in fact smokes the scenario closest to reality in which Suzy doesn’t smoke two packs. But it seems odd to say that the truth of the straightforward causal claim that Suzy smoking two packs of cigarettes per day caused her lung cancer depends on whether the most likely alternative scenario is one in which she smokes three packs rather than none. Some have tried to explain this oddity as a matter of pragmatic rather than semantic features of causal discourse (Lewis 1973, 2004), but there are weighty reasons to think that this strategy fails (Menzies 2004). By contrast, contextualist theories of causation resolve this uncertainty on the side of semantics, by letting contextual parameters to fix the truth conditions of causal statements. There are two dominant ways of casting such parameters, one in terms of presupposed contrast classes, and another in terms of presupposed causal models. According to the contrast class approach, causal claims always presuppose a set of relevant contrasts to the causing event (Hitchcock 1995, 1996; Maslen 2004, 2005; Schaffer 2005). In the above example, the contextually relevant contrast might be not smoking at all, thus making the full causal claim this: Suzy smoking two packs of cigarettes per day rather than not smoking at all caused her lung cancer. In order to evaluate the truth-value of this claim, we only need to focus on the counterfactual scenario in which Suzy doesn’t smoke at all, thus resolving the above uncertainty.

2

On the other dominant approach, the relevant parameter should be identified with a causal model, specifying causal relationships between variables in the kind of system that the causal claim is cast within. Such models can either be constructed using structural equations specifying relationships between the values of the relevant variables (Pearl 2000; Hitchcock 2001; Woodward 2003; Menzies 2007), or more informally in terms of an explicit ranking of possible scenarios by closeness to a sphere of ‘normal scenarios’ (Menzies 2004). Again, the result is that the uncertainty in evaluating causal claim will be resolved by the contextual parameters. In the example, we might suppose that the contextually presupposed ‘normal scenario’ is a scenario in which Suzy doesn’t smoke, thus making this possibility the relevant alternative when evaluating the counterfactual relationship smoking and lung cancer. Purpose of research project and working hypothesis There are two main questions about the role of context for causal judgments that remain unresolved in the literature, which I will focus on in the initial phase of the research project. The first is how to correctly model the relevant contextually determined parameters, i.e. should it be in terms of contrast classes, models encoding relationships between causally relevant variables, or something entirely different? The second question is how the context functions to determine the content these parameters – how does context determine which contrast classes or models to count as relevant to particular causal judgments? In addition to these basic theoretical questions, only limited attention has been devoted to how the application of this causal theory might resolve problems in other areas of study where causal concepts play important roles, particularly in the philosophy of mind, and to empirical investigations into the role of contextual parameters in human causal judgment. I will introduce these two in turn. The application of a contextualist account of causation seems particularly promising as a strategy of solving the so-called ‘causal exclusion’ problem in the philosophy of mind. This problem is, in short, that if every physical event has a sufficient physical cause, psychological events will be ‘excluded’ from the causal story of such physical events, including instances of physical behaviour that we would normally explain in terms of psychological events. As suggested by some authors, the contextualist 3

account of causation may offer a way out of this problem, since physical and psychological causal explanations are constrained by different contextual parameters, which may make resolve the competition between the two types of explanation (see e.g. Horgan et al. forthcoming). The second phase of the research project will be devoted to carry out these applications in detail. The empirical investigation into causal judgment has recently been the object of increasing attention3. Hitherto, few studies have been made of the role of contextual factors in causal judgment. Such a study would be important not only for understanding human causal judgment, but also for developing artificial intelligence simulations of such cognitive mechanisms. In addition, philosophers often appeal to conceptual intuitions when developing and evaluating theories of causation, and it would be beneficial to supply these intuitions with empirical data about how causal judgment is actually shaped by contextual factors. A model for such a study would be The Moral Sense Test4, hosted by the Cognitive Evolution Laboratory at Harvard University, and The Intentional Action Test (Knobe 2005). For further details, see the Methods section below. Research plan The proposed research project can be divided into three phases, running from August 2008 to August 2011, as follows: Phase 1 August 2008 – August 2009: Contextual approaches to causation. This phase concerns the prospects and correct form of a general contextualist analysis of causation. It will primarily address the following questions:

•

What are the contextual parameters of causal statements?

•

How does (explanatory, conversational, cognitive) context determine the value of these parameters?

3

See for example Goldvarg et al. 2001; Knobe 2005; Knobe et al. forthcoming; Wolff 2007; Livengood et al. 2007 4 For more on this project, see http://moral.wjh.harvard.edu/

4

•

How can a contextualist analysis of causation avoid relativism about the causal relation?

Projected end product of phase 1: submission of two articles to international peerreviewed journals on the prospects of a contextualism approach to causation. Phase 2 August 2009 – August 2010: Implications of contextualism to the problem of mental causation. This phase concerns the prospects for resolving problems in adjacent areas by the application of a contextualist theory of causation, particularly in the philosophy of mind and cognitive science. It will primarily address the following questions:

•

How do contextual parameters function in psychological causal explanations of speech and behaviour?

•

How might a contextualist approach to causation resolve the problem of mental causation?

Projected end product of phase 2: submission of two articles to international peerreviewed journals on the prospects of a contextualist resolution of the problem of mental causation. Phase 3 August 2010 – August 2011: Empirical study of contextual parameters in ordinary causal judgment and reasoning. This phase will be dedicated to designing and implementing a web based empirical data collection survey with the aim of collecting and analysing data that can help address the following questions:

•

How do contextual parameters function in shaping human causal judgment?

•

Are there cultural or demographic variations in the role and determination of contextual parameters? 5

Projected end product of phase 3: submission of two articles to international peerreviewed journals on the design and results of en empirical study of the influence of contextual parameters in ordinary causal judgment and reasoning.

Bibliography Dretske, F. (1977), ‘Referring to Events’, Midwest Studies in Philosophy 2, 90-9. Goldvarg, E. and Johnson-Laird, P.N. (2001), ‘Naïve Causality: A Mental Model Theory of Causal Meaning and Reasoning’, Cognitive Science 25: 565-610. Hart, H.L.A., and A.M. Honore (1959), Causation in the Law, Oxford: Clarendon Press. Holland, P. (1986), ‘Statistics and Causal Inference’, Journal of American Statistical Association 81, 945-60. Horgan, T. et al. (forthcoming), ‘Mental Causation’ in The Oxford Companion to Causation, Oxford: Oxford University Press. Hitchcock, C. (1995), ‘The Mishap at Reichenbach Fall: Singular vs. General Causation’, Philosophical Studies 78, 257-91. Hitchcock, C. (1996), ‘The Role of Contrast in Causal and Explanatory Claims’, Synthese 107, 395-419. Hitchcock, C. (2001), ‘The Intransitivity of Causation Revealed in Equations and Graphs’, Journal of Philosophy 98, 273-99. Knobe, J. (2005), ‘Cognitive Processes Shaped by Impulse to Blame’, Brooklyn Law Review 71, 929-37. Knobe, J. and Fraser, B. (forthcoming), ‘Causal Judgment and Moral Judgment: Two Experiments’, Moral Psychology, ed. W. Sinnott-Armstrong, Cambridge, Mass.: MIT Press. Lewis, D. (1973b), ‘Causation’, Journal of Philosophy 70, 556-67. Lewis, D. (2004) ‘Causation as Influence’ (expanded version), Causation and Counterfactuals, ed. J. Collins et al., Cambridge, Mass.: MIT Press. Livengood, J. and Edouard, M. (2007), ‘The Folk Probably Don’t Think What You Think They Think: Experiments on Causation by Absence’, Midwest Studies in Philosophy 31, 107-27.

6

Mackie, J.L. (1965), ‘Causes and Conditions’, American Philosophical Quarterly 2/4, 245-55 and 261-4. Maslen, Cei (2004), ‘Causes, Contrasts, and the Nontransitivity of Causation’, Causation and Counterfactuals, ed. J. Collins et al., Cambridge, Mass.: MIT Press. Maslen, Cei (2005), ‘A New Cure for Epiphobia: A Context-sensitive Account of Causal Relevance’, Southern Journal of Philosophy 43, 131-46. Menzies, P. (2004), ‘Difference-making in Context’, Causation and Counterfactuals, ed. J. Collins et al., Cambridge, Mass.: MIT Press. Menzies, P. (2007), ‘Causation in Context’, Causation, Physics, and the Constitution of Reality: Russell’s Republic Revisited, ed. H. Price et al., Oxford: Oxford University Press. Pearl, J. (2000), Causality. Models, Reasoning, and Inference, Cambridge: Cambridge University Press. Reichenbach, H. (1956), The Direction of Time, University of California Press. Schaffer, J. (2005), ‘Contrastive Causation’, The Philosophical Review 114, 297-328. Unger, P (1984), Philosophical Relativity, Oxford: Oxford University Press. Van Fraassen, B. (1980), The Scientific Image, Oxford: Oxford University Press. Wolff, P. (2007), ‘Representing Causation’, Journal of Experimental Psychology: General 136, 82-111. Woodward, J. (2003), Making Things Happen: A Theory of Causal Explanation.

7