Causation, for the purpose of this essay, is taken to be a relation between events. (One might alternatively define causation where the relata are framed as facts, or states of affairs, or an agent and an event. That analysis then would differ, although there are strong arguments against the use of some of these, e.g. Davidson’s slingshot argument against fact causation). Counterfactuals, to counterfactual conditionals, are statements of the form ‘If it had been the case that p, then it would have been the case that q’ - where the antecedent expresses something that is ‘contrary to fact’, or what is actually the case, and the auxiliary verbs appearing in the antecedent and the consequent are subjunctive in mood. An analyses of causation in terms of counterfactuals would trace the meanings of the statements of causation, like ‘Event c was a cause of event e’, to counterfactual conditionals of the form ‘If c had not happened, then e would not have happened’ (Here, c and e are shorthands for cause and effect).
The simplest analysis analysis through counterfactuals would be :
Event c was a cause of event e if and only if c occurred and e occurred and if c had not occurred, then e would not have occurred.
However, such a simple analysis would allow an event to be its own cause, since it might be the case, by way of such an analysis, that c is a cause of e AND e is a cause of c. For if c and e are such events that always go together, like moving my arm and moving my elbow for example, then it would always be the case that e would not have occurred had c not occurred and vice versa. Even if we refine the analysis to include only wholly distinct events, there would be other problems. Some of them are described below:
I will grant that the above problems do not prove that a counterfactual analysis of causation is not possible - but only that the simple analysis given above wouldn’t do the job. It can be argued that although the above analysis is mistaken, there might be some analysis similarly based on counterfactuals which is so refined that all problematic cases are excluded and taken care of. But the above example at least show that, even if there is the possibility of a counterfactual analyses of causation, such an account would have to be at least a bit involved and sophisticated.
However, my contention here is stronger: I believe that any analysis of causation based on counterfactuals is off the rail from the start. This is because the evaluation of truth and falsity of counterfactual conditionals does not satisfy at least two constraints which we might reasonably require of any analyses of causal relations: (1) it should not be circular, i.e. the analyses should not rely on causal considerations themselves and (2) it should represent causality as an objective relation which obtains between its relate independent of how humans choose to describe or frame those entities. Both these criticisms are founded on the way we interpret and evaluate counterfactual conditionals. Let me take them one by one.
For counterfactuals to provide a valid basis for the meaning of causation, it is imperative that the counterfactuals, in turn, do not derive their meaning from causation itself. The meaning of a statement of the form ‘If c had not happened, then e would not have happened’ is not as straightforward as a statement like, say, ‘Everyime c happens, e happens’, because while the latter is a statement of observation about the world, the former seems to point out to an alternative reality.
The most common way of interpreting a counterfactual is by an appeal to the notion of possible worlds. A counterfactual conditional of the form ‘If it had been the case that p, then it would have been the case that q’ is interpreted as a statement about a world which is fairly similar to the actual world except where p is true. However, whether the statement is true or not depends on which causal propositions are true in this alternative but close-to-actual world. For example, whether the statement ‘If I had let go off the stone, it would have fallen to the ground’ is true or not will depend, in part, on what causal matters of fact or causal laws are true in this alternative world - whether the stone is supported by a thread, whether there is person ready to catch the stone, whether the gravitational law holds, and so forth. To assess the truth or falsity of counterfactual conditionals, then, we need a notion of causation in the first place.
It might be argued that we need causation to evaluate a counterfactual proposition as true of false does not imply we need causation to interpret its meaning via the notion of possible worlds. Yet can we really understand a proposition yet not have any idea of what it means for it to be true or false? - which would be true for a person who does not grasp the concept of causation in the first place. Unless we get an alternative account for the meaning of counterfactual conditions which does not require the concept of causation, the circularity objection would hold.
There are other reasons to suspect that an analysis of causation based on counterfactuals is misguided. This is because the truth value of a counterfactuals is highly context dependent, while causal propositions are deemed to be, at least in part, independent of human conversational contexts. The same speaker might ascribe different truth values to the same counterfactual statement in different contexts. For example, I might maintain that ‘If I had let go off the stone, it would have fallen to the ground’ is true in a lecture on gravitational laws, and false in a conversational context in a where I am suggesting that no person in their right mind would have let the stone fall - like in a conversation about semi-precisious stones, for example. Thus, in different conversational contexts, the same counterfactual statement might have different propositional content, unlike causal propositions, whose propositional content is supposed to represent purely objective matter of fact.
One might try to rid the counterfactual analyses of its subjectiveness by stipulating that counterfactual conditionals are to be interpreted in such a way which is suited to conversations in which causal matters are the subject of discourse. But if this manouvre rids the counterfactual analysis of the problem of subjectivity, it brings in circularity. Instead of the counterfactual analysis throwing light on the meaning of causal statements, we would already need to have a firm grasp of the notion of causality in order to interpret counterfactual statements in ways that are appropriate to the conversational contexts in which they are used.
Thus, on grounds of its context-dependence, and implicit dependence on the meaning of causation, it seems that any counterfactual analyses of causation is misguided.
19 February 2015