Maslen’s account of the hard cases

In document The chances of higher-level causation: an investigation into causal exclusion arguments (Page 167-173)

Chapter 3: A problem for counterfactual theories of causation

3.4 Maslen’s account of the hard cases

According to Maslen, general criteria can be given for demarcating the problem cases, but causation is intransitive. The reason why we should have good general criteria differentiating transitive and non-transitive cases is that the validity of explanations by causal intermediaries depends on this. The lack of criteria to differentiate transitive and non-transitive cases would endanger the validity of these widely used causal explanations.

After she develops the basic contrastivist idea and applies it to the easy cases she turns to the hard cases and observes that the criterion already developed for easy cases can’t help us with these (Maslen 2004: 351-354). Let us take a look at a contrastive reformulation of (Ib), I will call it (Ib1*):

Yelling the order (C) rather than giving no order at all (C1*) causes the victim to duck (D) rather than not to duck (to stay upright) (D1*).

The ducking of the victim (D) rather than not ducking (his staying upright) (D2*) causes the survival of the victim (E) rather than the death of the victim (E1*).

Yelling the order (C) rather than giving no order at all (C1*) causes the survival of the victim (E) rather than the death of the victim (E1*).

The conclusion that follows from 1. and 2. by transitivity is obviously false, but there seems to be no way to interpret the contrasts of D as not being identical as it was possible in the case of the easy example. D1*=D2*=‘staying upright’ is a perfectly valid interpretation of the meaning of the default contrast ‘not to duck’ in the context provided. 1. and 2. are true independently of each other, so it seems that the first criterion is insufficient for demarcation. To amend the situation Maslen suggests a further criterion. She claims that there is an interesting trait transitive causal chains have, but short-circuit scenarios seem to lack. It should be true about scenarios involving transitive chains that if the second contrast event had occurred then the first contrast event would have occurred as well (see: Maslen 2004:352). This amounts to a backtracking counterfactual criterion: D*□-›C1*. Her claim is that if this criterion is violated, we have a case of non-transitive causation at hand.

Discussing this second criterion I will put aside methodological questions concerning the evaluation and interpretation of backwards counterfactuals65. I will concentrate on the

problems of its application to the case in question. To check the soundness of the criterion let us apply it to a straightforwardly transitive example first, the case of the provoked bull (Ic1*): The bull being provoked by a stranger (C) rather than having a peaceful environment (C*) caused the bull to become angry (D) rather than to remain tranquil (D*)

The angriness of the bull (D) rather than its being tranquil (D*) caused the bull to attack its owner (E) rather than to behave as usual (E*).

The bull’s being provoked (C) rather than having a peaceful environment (C*) caused the bull to attack its owner (E) rather than to behave as usual (E*).

D*□-›C* seems to be true here. If the bull had been tranquil the bull would have had a tranquil environment earlier. So, it is somewhat plausible to think that the criterion identifies transitive cases in accordance with intuition.

Maslen maintains that her second criterion is violated in the cases of hard counterexamples. At first sight it seems to be true in (Ib1*) that ‘for the victim to stay upright, the master assassin would have not to have given a command’. However, it is not simply the command that alarms the victim, it is the sound produced by the shouting. Concentrating on

65 There is a general worry related to the use of backtracking counterfactuals. Counterfactuals are implemented

by a miracle: introducing the antecedent by a miracle that happens later than the consequent one would break down the directed causal relationship between the antecedent and the consequent. So, it is not easy to see how to implement a backwards counterfactual properly. We should also mention that David Lewis himself was against the use of backtracking counterfactuals in all his versions of the counterfactual theory of causation, although to that worry one can reply that in this case it is not used to evaluate causal status, but to decide whether causal chaining is justified or not, which is a distinct issue. But the first worry concerning the implementation of such counterfactuals holds in any case.

this fact it becomes visible that there are other available contrast events. Let’s call the following reformulation (Ib2*):

Giving a yelled command (C) rather than giving a silent command (C2*) caused the victim to duck (D) rather than to stay upright (D*).

The ducking of the victim (D) rather than his staying upright (D*) caused the survival of the victim (E) rather than his death (E*).

Giving a yelled command (C) rather than giving a silent command (C2*) caused the survival of the victim (E) rather than his death (E*).

Maslen thinks that the availability of contrasts like (C2*) means that D*□-›C1* is not true in (Ib1*) (see: Maslen 2004:352). She generalizes this observation to demarcate scenarios of the same type as failures of transitivity and suggests that there is an analogy with the intransitivity of counterfactuals as it was conceptualized by Lewis (1986). I won’t discuss the question of the transitivity of counterfactuals here66; I only aim to show that the criterion

suggested is insufficient for demarcation.

Let’s unpack what the criterion has to say. It seems to be true that starting from a world with an upright victim a world with a silent command is closer than a world where there is no command given at all. In other words, it doesn’t seem to be true that in the closest possible world with an upright victim there is no command either, because there seems to be a world that requires even less departure from actuality which has a silent command instead. If this were true, then according to Maslen the second criterion would demarcate (Ib1*) as a case of non-transitive causation (see Maslen 2004:352–353).

66 In the literature on counterfactuals the issue of transitivity was thoroughly discussed quite early on and people

like Lowe (1990) argued convincingly that all apparent counterexamples involve contextual equivocation, if he was right then the whole analogy between the intransitivity of causation and counterfactuals breaks down.

As far as I know, nobody has tried to criticize this solution directly, not even Jonathan Schaffer who proposed an alternative account which will be investigated in the following section. Schaffer has even said that in case his account fails Maslen’s idea might provide us with a fall-back option, although it is less systematic in dealing with transitivity problems than the new framework he has developed (see Schaffer 2005:311, footnote 24). In what follows, I will show that Maslen’s solution is not only less systematic, it has internal problems. To do so, I will point to the fact that in the context where ‘giving no command’ is the justified, admissible contrast ‘giving a silent command’ is simply not available and therefore the second criterion fails to deliver on its promise.

To spell out this criticism of mine I will draw on resources provided by Northcott (2008) and Reiss (2013a, 2013b) who developed the idea that the contrastivist framework is in need of criteria to decide the admissibility of contrast events. The motivation for this was the realization that contrasts were introduced to disambiguate causal statements. Unfortunately, if there is no principled way of fixing contrasts for particular causal statements than in many cases the method of contrasting does more harm than good as it provides means to change the truth values of causal statements arbitrarily (see Reiss 2013a:1067). The contrastive formulation is only helpful if there is a way to decide which contrast is acceptable for a particular causal statement. In my interpretation, attendance to admissibility criteria is attendance to how a causal statement should be interpreted and evaluated relative to its particular context. According to Reiss, the admissibility of contrasts in cases like (Ib) is determined by what we think to be a possible course of action under the circumstances (see Reiss (2013b:1079–82, 1088). The circumstances we are talking about are nothing else but the facts of the worldly context relevant to the causal statements in question.

Let us suppose that in the actual case the actions described in scenario (Ib) take place in a marketplace that is noisy, hence the need for a shouted command. Under the circumstances the master assassin has only two real choices, either to give a loud command and risk that by doing so he warns the victim as well, or to stay silent and wait for another opportunity. Non-verbal and other ways of signalling are excluded because of the positions the assassins took. The main thing is that we know that a silent command wouldn’t be effective under the circumstances. In other words, the context of our causal statement in (Ib) is such that the contrast in (Ib2*), ‘giving a silent command’, goes against the logic of the situation, no master assassin would try that67. There might be a faraway world where the

master assassin gives a silent command, but in that world many other things are different as well: the marketplace is peaceful, the assassins can easily hear each other, etc. That would be a different world that is far away from the actual where ‘giving no command’ is practically the only alternative for the master assassin.

One could also turn the example around adding contextual features to (Ib) where the marketplace is silent and it is feasible to think that the master assassin would give a silent command. This would select (Ib2*) as the right contrastive interpretation in which case (Ib) is not a counterexample to transitivity and the contrast in (Ib1*) ‘giving no command at all’ takes place in a faraway world.

Knowing the above, we can see that where the first contrastive formulation of the assassin trainee scenario (Ib1*) is justified there the backwards counterfactual criterion

67 It is interesting to highlight here, that examples with a worldly contextual structure isomorphic to (Ib1*),

where the only plausible alternative to a yelled command is no command given at all, are creating problems for transitivity, whereas (Ib2*) is a proper transitive case. So, in our example (Ib) the context of the surface level statements makes a huge difference with respect to the content of the causal statements involved.

required by Maslen is made true by the same contextual features that justify the choice of the first contrast event. If the admission of the contrast event ‘giving no command’ is justified in the actual world and the exclusion of ‘giving silent command’ is grounded in the same facts, then there is no possible way the criterion D*□-›C1* could be false and demarcate the assassin trainee scenario as a case of non-transitive causation. The reader can easily verify that the same logic applies to some other widely discussed hard counterexamples to transitivity. Form this it follows that Maslen’s second criterion cannot serve either as a necessary or as a sufficient condition for identifying hard scenarios as cases of non- transitivity.

In document The chances of higher-level causation: an investigation into causal exclusion arguments (Page 167-173)