Humean Supervenience and Possible Worlds

The position defended is compatible with a coherent development of a reductive account of modality. It is less significant for the principle of recombination than previously thought and the principle of recombination is unlikely to play successfully its role as a principle of plenitude in any event. The distinct existences principle is defended against an argument that it is necessarily false. In fact, by comparison, we have more reason to believe in the possibility of worlds in which Humean supervenience is true and there is causation, than in physicalism about phenomenal consciousness. The counterfactual theory of causation, and surrounding framework, explains why this is the appropriate verdict at which to arrive. Some aspects of the variety of causation may be understood as a determinate-determinable relation but the different vertically fundamental bases are better understood as partial realizations. Causation is one horizontally fundamental metaphysical category but there may be others.

The Ontology of Chance

Just as laws are variably realized so are objective chances: in the patterns identified by the best system analysis and in propensities. Theories of chance face two significant problems: the problem of undermining that is alleged to afflict Humean accounts of chance and, second, the relationship of chance to frequencies and, thus, to successful action. Although some propensity accounts can avoid undermining, they do so at the expense of the second relationship. More concessive propensity theories make some headway with regard to the second problem but start to suffer from the first problem. The perceived advantage for agents in conforming their beliefs to chances, understood as propensities, is rooted in the same mistake about induction identified in Chapter 14. So the successful treatment of chance does not tell in favour of one theory of the laws that support them than another.

Causal Non-Symmetry

Causation is a non-symmetric rather than asymmetric relation. Different bases of causal non-symmetry include an asymmetry of overdetermination, the independence condition, and agency. Causal non-symmetry can be rooted in one or more of these three while also recognizing a fourth non-symmetry appealing to a primitive non-symmetric chance-raising. Each counts as an appropriate basis for causal non-symmetry because it is a (partial) realization of non-symmetric chance-raising. Key moves involve a refinement of how to understand the way in which the asymmetry of overdetermination works, and how it interacts with the revised similarity weighting, the contribution of the independence condition to a proper understanding of the transition period, the role that appeals to primitive non-symmetric chance-raising should play in the treatment of problem cases, the circumstances in which an appeal to an interlevel non-symmetry of agency may be appropriate, and the priority ordering of these various realizations of causal non-symmetry.

Processes and Prevention

The proposed analysis of causation is compatible with allowing that there are ways to distinguish the variety that falls under it. The same characteristics as those who take causation to involve substantial causal processes characterize kinds of causation without these characteristics themselves serve to characterize causation in general. This is an advantage because the theories that make an appeal to substantial processes in understanding causation face considerable difficulties. The attempt to tie causation to the presence of substantial causal processes between cause and effect fails to be justified by appeal to responsibility, or by its capacity to make sense of causal locality and the intrinsic character of causal processes. Some claim that a counterfactual theory closes off certain options with regard to the property understanding of Bell inequalities. This is not the case.

Causal Circumstances

The proposed analysis of causation has two immediate consequences regarding causal relata. The first is that the analysis will be satisfied by both an element in some causal circumstances for an effect and also by the causal circumstances taken in toto for that event, sometimes called the total cause of the effect. The second is that the analysis does not distinguish between elements in the causal circumstances, characterizing some as causes, others as enabling conditions, to take one alternative. These consequences are defended against those inclined to reject them or propose supplementation to the analysis. The idea of causes as contra-normal conditions either derives from certain biases at work in our attributions of causes or can be the basis for a pragmatic explanation of why we favour elements in causal circumstances as causes over others. It is a mistake to take contra-normality to be part of the nature of causation.

The Non-Transitivity of Causation

There are clear cases in which causation is not transitive and this drops out of the analysis developed in which causation involves a certain kind of chance-raising involved after subtracting competitor processes. Attempts to explain away these cases to secure the transitivity of causation are a mistake. Alternative ways of capturing the non-transitivity of causation involve fixing the competitor processes in order to detect chance-raising dependencies between the target cause and effect. This alternative manoeuvre engenders problems. The non-transitivity of causation is better understood in the terms of my analysis rather than by appealing to the idea that causes are difference-makers (in a specified sense) or, in the kind of cases considered, switchers by interaction with a process.

A Counterfactual Analysis of Causation

A counterfactual analysis of causation is developed by distinctive notion of chance-raising characterized by probabilistic Σ‎-dependence, when a causal chain is complete appealing to chance-raising at a time just before the time of the effect, and a requirement that the causal chain is made up of actual events to avoid the standard problems with conditional analyses, due to potential changes in the circumstances when the antecedents are true. Although the development takes the form of a consideration of difficult cases of causation (especially probabilistic cases of pre-emption), the resulting idea has independent motivation and simplicity. It is that causes of a target event are those which (independently of its competitors) both make the mean chance of an effect very much greater than its mean background chance, and actually influence the probability of the effect in this way, at the time at which the effect occurred via a complete causal chain.

Causation and Laws

Laws are potential patterns of causation characterized by the counterfactual analysis. The various accounts of law in the literature—or proposed refinements of them—are not properly thought of as accounts of law but rather ways in which laws may be realized. The various potential counterexamples to Humean accounts of laws are best seen as appealing to worlds in which laws are not realized in the Humean way, rather than showing something about the nature of law. This point is particularly salient bearing in mind the intellectual difficulties of non-Humean accounts of law—such as necessitation between properties accounts or the powers ontology—and the fact that their virtues have been rather overstated. In particular, this is outlined regarding counterfactual support and their alleged superiority in providing a basis for the rationality of induction.

Non-Causal Counterfactual Dependence and Intrinsicality

When counterfactuals hold concerning entities with properties that stand in some kind of loose existential dependency relation, counterfactual dependence only indicates a causal relationship if part of their corresponding minimal supervenience bases satisfies the analysis of causation. The idea has application even if properties are understood in ways proponents of a powers ontology recommend. An analysis of intrinsic properties in this chapter appeals to three features—External Independence, Duplication Characterization, and Maximizing Recombination—each of which, by itself, doesn’t quite work to demarcate what we have in mind. This provides a second way of approaching the issue as well as assisting with later analysis of varieties of Humean supervenience.

Negative Causation and the Relationality of Causation

Negative causation and dispositions as mutual manifestation partners don’t threaten the relationality of causation. Separating the supervenience of truth on being from explanation and the distinctive relevance of truth-makers enables us to avoid arguments from parsimony against positive causal surrogates drawing upon the idea of truth-maker necessitation. If truth-maker necessitation were the proper characterization of how to explain the truth of all statements, and negative causal statements in particular, then a fact of totality, or the essentiality of properties of the world, would be the most parsimonious explanation of their truth albeit at the expense of identifying something that conveys their distinctive subject matter. Within the recommended framework, the existence of true negative causal statements provided motivation for recognizing the existence of positive causal surrogates as part of the explanation of their truth. The truth of negative causal statements is the basis for an argument against the truth-making picture.