Modal Logic

3.2.1

Conditionals in Sentential Logic

In some cases conditionals in the form of I) do not suciently map what is meant in natural language. As an example assume the following to be true: (a) Not many Americans eat Thai food. Using this statement as an antecedent in a conditional could result in: (c) If many Americans eat Thai food, then the price of gasoline will skyrocket. The consequent of this conditional in no way depends on the antecedent. In natural language this sentence would be regarded as just false or as lacking any truth value. But consider (c) under I): (cI ) [Many Americans eat Thai food] [The price of gasoline will skyrocket] The conditional formed in (cI ) will be rendered as true. That is because row 3 in the following table has to be used for nding the truth value of cI , as [Many Americans eat Thai food] is false due to (a) . X T T F F Y T F T F XY T F T T

Table 3.1: Truth table for conditional in the context of sentential logic

Therefore another form than I) has to be found for (c) .

3.2.2

Conditionals in Modal Logic

Indeed a much better solution is given when using a strict conditional. Such a conditional is a conditional that is true in all possible words2 and is thus dened as: (X Y ) (X Y )
2

cf. D. Bonevac, Deduction: Introductory Symbolic Logic, Wiley, 2003, p. 251.

Modal Logic

Therefore a strict conditional is true when [if X is true in the given possible world, Y is true there as well]3 is true. In other words: ( sc) (X Y ) is true in a world i Y is true in every possible world in which X is true.4 With the formulation of (cII ) (cII ) [Many Americans eat Thai food] [The price of gasoline will skyrocket] one gets what is expected to be the case in natural language. This is so because ( sccII ) has to be true: ( sccII ) [Many Americans eat Thai food] [The price of gasoline will skyrocket] is true in a world i [The price of gasoline will skyrocket] is true in every possible world in which [Many Americans eat Thai food] is true. But ( sccII ) is false, because there is at leat one possible world where [Many Americans eat Thai food] is true while [The price of gasoline will skyrocket] is false, in which case row 2 in Table 3.1 applies. Thus there is at least one counterexample to the assertion that the conditional would be necessary. And if a conditional is not necessarily true, then it cannot count as a strict conditional. Although strict conditionals do a much better job for mapping conditionals of natural language, they dont suce either when it comes to counterfactuals.5 In fact (c) expressed an indicative conditional, not a counterfactual conditional.6 To see the dierence between them, consider rst the following fact ( f ) : ( f ) On November 22, 1963: Somebody shot Kennedy. Now there are at least two conditionals that can be stated and which are based on ( f ) - an indicative conditional (id) and a counterfactual conditional (c f ) : (id) If somebody shot Kennedy and Oswald did not shoot Kennedy, then somebody else shot Kennedy. (c f ) If Oswald had not shot Kennedy, then somebody else would have shot Kennedy.7
3 4

cf. Bonevac, Deduction: Introductory Symbolic Logic cit., p. 399. cf. ivi, p. 254. 5 cf. ivi, p. 395. 6 cf. Moreland and Craig, Philosophical Foundations for a Christian Worldview cit., p. 52.

Modal Logic

These two sentences show how indicative conditionals dier from counterfactual conditionals. While in (id) the fact ( f ) can be added to the antecedent and is in fact usually implied, this cannot be done in (c f ) , because (c f ) would then become nonsense:
(c f ) If somebody shot Kennedy and Oswald had not shot Kennedy, then somebody else would

have shot Kennedy The antecedent speaks of two dierent worlds. While ( f ) is concerned with a world in which somebody shot Kennedy, then second part of the antecedent (Oswald had not shot Kennedy) doesnt guarantee that fact at all. So only the sentence (id) can be stated using a strict conditional: ( scid ) ([Somebody shot Kennedy and Oswald did not shoot Kennedy] [Somebody else shot Kennedy]) is true in a world i [Somebody else shot Kennedy] is true in every possible world in which [Somebody shot Kennedy and Oswald did not shoot Kennedy] is true. That example shows that it suces for indicative conditionals to be formulated in logical terms by using a strict conditional. But this does not apply for counterfactuals, because they are always contrary to fact, hence the name. The antecedent of the statement (c f ) does not assume ( f ) to be true, that somebody shot Kennedy. Using a strict conditional would result in: ( scc f ) ([Oswald did not shoot Kennedy] [Somebody else would have shot Kennedy]) is true in a world i [Somebody else would have shot Kennedy] is true in every possible world in which [Oswald did not shoot Kennedy] is true. But this cannot be correct. The requirement of every possible world where the antecedent holds is too much. It is certainly possible that (i.e. there is at least one possible world where) Oswald really was just a lone gunman that acted on its own and that while he was driving to the place from where he wanted to shoot Kennedy he had an accident. In such a case he would have missed his chance to shoot Kennedy on that day (i.e. the antecedent is true) and since in that possible world he acted alone, it is likely (or at least possible) that nobody else would have shot him on that day either. Therefore evaluating the truth value of counterfactual conditionals has to be done dierently. Not every possible world where the antecedent holds has to be regarded, rather only those worlds are taken into account, which are the most similar or closest to the world in which the counterfactual is stated (which is usually the actual world):

Modal Logic (cc) (X Y ) is true in a world w i Y is true in all the worlds in which X is true that are closest to w.8 Expanding (cc) with (c f ) results in: (ccc f ) ([Oswald did not shoot Kennedy] [Somebody else shot Kennedy]) is true in a world w i

[Somebody else shot Kennedy] is true in all the worlds in which [Oswald did not shoot Kennedy] is true that are closest to w Now the question is: Is (ccc f ) true or not? Well that depends. First note that the question is formulated in the actual world A. In this case the world w mentioned in (ccc f ) is A. Second those worlds that have [Oswald did not shoot Kennedy] as false are the ones from which select the closest to A. But which worlds are closest to A until the day of the assassination? To some degree this stays speculative but still some rough guessing can be done.9 First all worlds are ruled out that dont share the same relevant history with A until that day. So for example those worlds where the Nazis won World War II are excluded. On the other hand this should not be done with worlds where the history only diers with respect to irrelevant facts. Whether or not Jones rubs his nose on February 4th, 1720 seems to be rather unimportant for the worlds history in sum (assuming the rubbing does not have essential consequences). Furthermore, the truth value of (ccc f ) depends on whether in A a conspiracy was going on on the day of the assassination or not. In case of a conspiracy, those worlds are closest to A which also have that conspiracy included. And if in all those worlds someone else shot Kennedy, then (ccc f ) is true. Likewise if in A no conspiracy was going on, then those worlds without a conspiracy are closest to A. And that in all of those worlds someone else than Oswald shot Kennedy even though Oswald was just a lone gunman seems wrong.

3.2.3

Transitivity and Counterfactuals

In general a relation R is transitive if an element X that is related (in the sense of R) to an element Y, and Y is in turn related (in the sense of R) to an element Z, then X is also so related to Z. Both for conditionals in sentential logic and for strict conditionals in modal logic transitivity can be applied:10 But this does not hold for counterfactuals. To see why this is indeed the case, consider rst the following example:
cf. Bonevac, Deduction: Introductory Symbolic Logic cit., p. 401. cf. ivi, p. 400. 10 cf. ivi, p. 397.
9 8