Contrary to Fact: Cause, Chance, Natural Law, and their relation to Counterfactuals.

Counterfactuals are conditionals in which we trace out how things would, might or could have been if reality had been different. If kangaroos had no tails, they would topple over. If things had been different yesterday, they would have been different today. If this sugar had been put in water, it would have dissolved. Such thought experiments are linked to our concepts of causality, natural necessity, disposition, chance, and the direction of time. The fact that kangaroos would have toppled over if they had tails seems to imply that their having tails is part of what causes them to stay upright, that the both not toppling over and having tails is a non-accidental facts. The fact that things would have been different today, if they had been different yesterday if things had been different today, seems to imply that the past has the present as its effect, and not vice versa. That the sugar would have dissolved if it had been place in water indicates that the sugar has a disposition to dissolve in water. The enigma of counterfactuals is the dual fact that they are judgements about what is contrary to fact, the non-actual, but are, at the same time, intimately linked to our deepest modal and temporal commitments about the actual.

To crack the enigma we need to know what guides our formation of counterfactual judgements, and what makes them true. The currently dominant view about how to answer these questions is that associated with David Lewis. Lewis's conception of counterfactuals is a child of the possible-worlds semantics that came to prominence in the 70s. For Lewis, a basic would-counterfactual is analysed in these terms:
If P had been the case Q would have been is true iff in the nearest possible worlds in which P is true, Q is true.
Here nearness in worlds is fixed by similarity of fact and natural law. Natural law is anterior to counterfactuals. Laws in themselves make no reference to causation or intrinsic modal features of the actual world. With the semantics of counterfactuals configured in this way, theorists have deployed them as analytical tools. Causation is then analysed in terms of counterfactual dependency, and dispositions.

I criticize this cluster of views and then: (a) I propose an alternative view of counterfactuals/a constructive, inferential approach/in which relations between fact-sized things (events) determine the outcomes of counterfactual judgements, rather than worlds and similarity relations between them. (b) I outline the implications of this view for how we understand causation and natural necessity, developing a chance view of causation, and an expressivist treatment of law that is neither Humean nor Necessitarian. (c) I offer a theory of chance that draws on expressivism and interventionist theories (of cause), which allows for a middle path between non-Humean and Humean truthmakers for chance statements. But there is a twist. Counterfactuals re-appear again in the explication of chance. Thus counterfactuals are both at the top of the analytic chain and at the bottom. However, I show how there is no circularity since the concepts of analysis in play differ between truth-conditional reduction and expressive reduction.