Page 277 - Encyclopedia of Philosophy of Language

that...,Couldadecisioneverberegrettedifonehad no reason to believe that it would indeed have made a difference if it had been decided otherwise? And would it ever be justified to call someone to account for his deeds, if everything would be exactly the same as it is now no matter what had been done by whom? Clearly, one does often reason about unactualized possibilities, thereby employing so-called counter- factual conditionals:
If she had asked me, I would have danced with her. (1)
Counterfactuals are typically uttered in contexts where the antecedent is known to be false. Therefore, unlike indicative conditionals, they cannot possibly be analyzed as material implications. Material impli- cations with a false antecedent are true no matter what the truth value of the consequent is. But one does not want to be forced to call both sentences (1) and (2) true:
Theobviousnextquestioniswhichaccidental truths can serve as further premises and which cannot. Not everything goes. For one thing, we wouldn't want to allow the negation of the antecedent as a further premise even though the antecedent, together with its negation, implies the consequent. But where do we have to draw the line? The only natural answer to this question seems to be this: those accidental truths B for which the sentence If A had been the case, B still would have been the case is true. But as Goodman acknowledged, this answer turns the analysis into a circular one.
2. Minimal ChangeTheories
In 1968Robert Stalnaker proposed an account of Counterfactuals in which no attempt was made to explain away the reference to unactualized possi- bilities. Starting point for his analysis was a test for evaluating the acceptability of conditionals originally devised by Frank Ramsey. It can be summed up as follows:
First, hypothetically, make the minimal revision of your stock of beliefs required to assume the antecedent. Then, evaluate the acceptability of the consequent on the basis of this revised body of beliefs.
Ramsey's original suggestion only covered the case in which the antecedent is consistent with the agent's stock of beliefs. In that case, which is typical of indica- tive conditionals, no adjustments are required. Fol- lowing an idea of Rescher 1964, Stalnaker generalizes this to the case in which the antecedent cannot simply be added to the agent's stock of beliefs without intro- ducing a contradiction. In this case, which is typical of Counterfactuals, adjustments are required.
In effect what Stalnaker does is reconstruct the above belief conditions as truth conditions. He thinks thattruthmaynotbeallowedtodependonbeliefs, that you have to appeal to the facts. So, in his rebuilt version the actual world plays the role that the agent's stock of beliefs plays in Ramsey's. And the minimal revision of the agent's stock of beliefs required to assume the antecedent is taken up as that possible world at which, (a) the antecedent is true and which (b) in all other respects differs minimally from the actual world.
This proposal raises an immediate question: Which of the conceivably many possible worlds at which the antecedent is true will be the world most similar to the actual world? According to Stalnaker, this is in essence a pragmatic question which has little to do with the semantic problem he is concerned with. He is ready to admit that contextual features may make a differ- ence to the particular world which has the property concerned. But how these contextual features make that difference is less important, the only thing that matters is that there is an outcome.
Note that Stalnaker assumes that there will always 255
If she had asked me, I would not have danced with her.
1. The Metalinguistic Approach
(2)
Taken at face value, Counterfactuals refer to unac- tualized possibilities—a kind of entities that phil- osophers, in particular those standing in the empiricist tradition, look upon with suspicion. Accordingly, sev- eral attempts have been made to show that Counter- factuals are only apparently about unactualized possibilities, and to give a logical analysis of their meaning in which no recourse to such entities is made. The locus classicus here is Goodman 1947. On Good- man's account a counterfactual If A had been the case, C would have been the case can best be thought of as a metalinguistic statement expressing that the ante- cedent A together with some suitable further premises /?/,..., Bn, logically implies the consequent C.
WhichfurtherpremisesB1,...,Earesuitabletobe used with a given antecedent? Obvious candidates to consider are sentences that express causal connections or other lawlike relationships between matters spoken of in the antecedent and matters spoken of in the consequent. We believe that John would have fallen to the ground, if he had jumped out of the window, and we appeal to a very simple form of the laws of motion to prove our point.
Natural laws are not the only further premises one needs. There are no natural laws establishing a con- nection between the antecedent and consequent of the next sentence:
If I had looked in my wallet, I would have (3) found a penny.
This statement may very well be true just because there happened to be a penny in my wallet at the occasion I am referring to. Hence, in addition to natu- ral laws also accidental truths have to be allowed as further premises.
Counterfactuals