Online research in philosophy

This paper defends the view that the asymmetry of causation can be explained in terms of probabilistic relationships between event types. Papineau first explores three different versions of the "fork asymmetry", namely (i) David Lewis' asymmetry of overdetermination, (ii) the screening-off property of common causes, and (iii) Spirtes', Glymour's and Scheines' analysis of probabilistic graphs. He then argues that this fork asymmetry is both (i) a genuine phenomenon and (ii) a satisfactory metaphysical reduction of causal asymmetry. In his final section he shows how this reduction can account for the relevance of causal direction to human agency, and in particular for the fact that we can manipulate causes to influence their effects, but not vice versa.

One part of the true theory of actual causation is a set of conditions responsible for eliminating all of the non-causes of an effect that can be discerned at the level of counterfactual structure. I defend a proposal for this part of the theory.

This paper deals with Hans Reichenbach's common cause principle. It was propounded by him in (1956, ch. 19), and has been developed and widely applied by Wesley Salmon, e.g. in (1978) and (1984, ch. 8). Thus, it has become one of the focal points of the continuing discussion of causation. The paper addresses five questions. Section 1 asks: What does the principle say? And section 2 asks: What is its philosophical significance? The most important question, of course, is this: Is the principle true? To answer that question, however, one must first consider how one might one argue about it at all. One can do so by way of examples, the subject of section 3, or more theoretically, which is the goal of section 4. Based on an explication of probabilistic causation proposed by me in (1980), (1983), and (1990), section 4 shows that a variant of the principle is provable within a classical framework. The question naturally arises whether the proved variant is adequate, or too weak. This is pursued in section 5. My main conclusion will be that some version of Reichenbach's principle is provably true, and others may be. This may seem overly ambitious, but it is not. The paper does not make any progress on essential worries about the common cause principle arising in the quantum domain; it only establishes more rigorously what has been thought to be plausible at least within a classical framework.

When we philosophers think about causation we are primarily interested in what causation is – what exactly is the relation between cause and effect? Or, more or less equivalently, how and in virtue of what is the cause connected to the effect? But we are also interested in an epistemic issue, viz., the possibility of causal knowledge: how, if at all, can causal knowledge be obtained? The two issues are, of course, conceptually distinct – but to many thinkers, there is a connection between them. A metaphysical account of causation would be useless if it did not make, at least in principle, causal knowledge possible. Conversely, many philosophers, mostly of an empiricist persuasion, have taken the possibility of causal knowledge to act as a constraint on the metaphysics of causation: no feature that cannot in principle become the object of knowledge can be attributed to causation.

When we philosophers think about causation we are primarily interested in what causation is—what exactly is the relation between cause and effect? Or, more or less equivalently, how and in virtue of what is the cause connected to the effect? But we are also interested in an epistemic issue, viz., the possibility of causal knowledge: how, if at all, can causal knowledge be obtained? The two issues are, of course, conceptually distinct—but to many thinkers, there is a connection between them. A metaphysical account of causation would be useless if it did not make, at least in principle, causal knowledge possible. Conversely, many philosophers, mostly of an empiricist persuasion, have taken the possibility of causal knowledge to act as a constraint on the metaphysics of causation: no feature that cannot in principle become the object of knowledge can be attributed to causation.

A probabilistic theory of causation is a theory which holds that the central feature of causation is that causes (usually) raise the probability of their effects. In this dissertation, I defend Hans Reichenbach's original (1953) version of the probabilistic theory of causation, which analyses causal relations in terms of a three place statistical betweenness relation. Unlike most discussions of this theory, I hold that the statistical relation should be taken as a sufficient, but not as necessary, condition for causal betweenness. With this difference in interpretation, Reichenbach's theory is shown to be immune to all of the criticisms which have been raised against it in the last..

A quite popular approach to solving the Causal Exclusion Problem is to adopt a counterfactual theory of causation. In this paper, I distinguish three versions of the Causal Exclusion Argument. I argue that the counterfactualist approach can block the first two exclusion arguments, because the Causal Inheritance Principle and the Upward Causation Principle upon which the two arguments are based respectively are problematic from the perspective of the counterfactual account of causation. However, I attempt to show that the counterfactualist approach is unable to refute a sophisticated version (i.e. the third version) of the exclusion argument in that the Downward Causation Principle, a premise of the third exclusion argument, is actually implied by the counterfactual theory of causation. Therefore, even if other theories of causation might help the non-reductive physicalist to solve the exclusion problem, the counterfactual theory of causation cannot.

The common cause principle states that common causes produce correlations amongst their effects, but that common effects do not produce correlations amongst their causes. I claim that this principle, as explicated in terms of probabilistic relations, is false in classical statistical mechanics. Indeterminism in the form of stationary Markov processes rather than quantum mechanics is found to be a possible saviour of the principle. In addition I argue that if causation is to be explicated in terms of probabilities, then it should be done in terms of probabilistic relations which are invariant under changes of initial distributions. Such relations can also give rise to an asymmetric cause-effect relationship which always runs forwards in time.

The paper criticizes the attempt to account for the direction of causation in terms of objective statistical asymmetries, such as those of the fork asymmetry. Following Ramsey, I argue that the most plausible way to account for causal asymmetry is to regard it as "put in by hand", that is as a feature that agents project onto the world. Its temporal orientation stems from that of ourselves as agents. The crucial statistical asymmetry is an anthropocentric one, namely that we take our actions to be statistically independent of everything except (what we come to call) their effects. I argue that this account explains the intuitive plausibility of Reichenbach's principle of the common cause.

argues that the success of the backward causation hypothesis in quantum mechanics would provide strong support for a version of Reichenbach's account of the direction of causal processes, which takes the direction of causation to rest on the fork asymmetry. He also criticises my perspectival account of the direction of causation, which takes causal asymmetry to be a projection of our own temporal asymmetry as agents. In this reply I take issue with Dowe's argument at three main points: his claim that the backward causation hypothesis in QM is incompatible with my perspectival approach to the direction of causation; his defence of the fork asymmetry approach against a general criticism of mine based on the time-symmetry of microphysics; and his application of his preferred account of the direction of causal processes to the relevant cases in QM.