«Of Ghostly and Mechanical Events* JONATHAN SCHAFFER University of Massachusetts, Amherst Two of the assumptions that drive most contemporary ...»
Here is Pietroski’s sufficient condition for causation: C causes E if (i) C and E are actual, distinct events, and (ii) there are singular event thoughts TI and T2 such that (a) T1 is about C, (b) T2 is about E, and (c) T1 P-explains T2 (2000, ch. 3, &?.2).5It remains to gloss the notion of P-explanation, which Pietroski thinks of as explanation via a style of deductive-nomological argument that allows for the use of (non-vacuous) ceteris paribus laws.
What I find problematic about Pietroski’s account of mental causation is (i) his sufficient condition is not extensionally adequate, (ii) it is unclear what if any underlying conception of causation is in play, and (iii) his acceptance of psychological laws is in tension with his anti-naturalism.
Extensional adequacy: Here are three counterexamples to Pietroski’s sufficient condition. The first counterexample concerns extrinsic descriptions.
Take any three events C1, C2, and E, such that C1 causes E but C2 does not.
Consider the following singular event thoughts about C2: the event that occurred 100 miles from the occurrence of C1, the event that occurred 200 miles from the occurrence of E, the event whose occurrence did not cause the occurrence of E. Each of these singular event thoughts will generate its own DN-argument to the conclusion that E occurred. So by the sufficient condition C2 causes E, when by stipulation it does not. Here the natural solution would be to limit the allowable singular event thoughts to the intrinsic ones.
But since Pietroski is interested in establishing the causal efficacy of contentful mental states such as tryings, this natural solution would require taking an intrinsic (harrow) view of content, which Pietroski himself opposes (2000, p.
The second counterexample concerns the causal asymmetry. Suppose that C causes E, and that the subsuming law is a biconditional: [(Vx)(Fxo (3y)Gy)l. Then there will be a DN-argument from the thought that E occurred to the thought that C occurred. So by the sufficient condition E causes C, when by stipulation it is the other way around: C causes E.6 Here the correct 5 I’ve reformulated Pietroski’s proposal by adding clause (i) (Pietroski informally recognizes the need for something like this on p. 103), and dropping the transitive closure part (none of the examples discussed in the main text require chaining).
6 Pietroski’s provides a sufficient condition for being a non-vacuous ceferrs paribus law (2000, ch.4, $4.2), which is supposed to build in an asymmetry. But (i) this simply does not apply when the law is strict, as in the example in the main text, and (ii) since Pietroski is only offering a sufficient condition for non-vacuity, the most he could show would be that the time-reversed version of the law fails to meet his sufficient condition, when what he needs to show is that the time-reversed version of the law fails to meet some necessuty condition.
The third counterexample concerns preempted backups. Suppose that (i) C 1 and E occur, (ii) they are subsumed under a ceteris paribus law, (iii) C 1 is cut-off on route to E, and (iv) E is produced by an independent causal route
For example, suppose that (i) I try to raise my arm (Cl) and my arm rises (E), (ii) there is a ceterisparibus law linking tryings to results (Pietroski 2000, ch.
3, 93.3), (iii I am struck by lightning (C2) in such a way as to short-circuit the electrical signal from my brain on route to my deltoid muscles (D), and (iv) the lightning strike causes my arm to rise anyway. Then there will be the right sort of DN-argument from C1 to E but obviously no causal link.’ This is the worst of the counterexamples vis-a-vis event dualism, since the neuralist may well charge that this is exactly how to describe ‘mental causation’ on event dualism. That is, the neuralist may well claim that the mentalbehavioral link would be preempted by the neural-behavioral link (Kim 1989), which would then reinstate the very behavior that the mental event aimed unsuccessfully to produce. One natural solution to preemption is to require the existence of a connecting process. But it is not at all obvious that this solution would still allow mental causation, since one natural way of glossing a connecting process is via a physical mechanism.’ ’ Pietroski is aware of this problem (2000, p. 109, pp, 134-5), and suggests that the ceteris paribus condition on the relevant law would be violated. I agree that this is intuitively the right reply, but I don’t see how Pietroski’s official conditions respect this intuition.
Perhaps 1 am missing something here. But of this much I am confident: Pietroski ultimately needs a necessary condition somewhere to rule that things like preempted backups are not causes.
Thus David Fair says: “The first [preempting] hoodlum is the source of the energy that the baseball transfers to the window; the second [backup hoodlum] is not. And the first causes the window to shatter; the second does not.” (1 979, p. 230; see also Wesley Salmon 1997, p. 475) Tropes to the rescue? -Douglas Ehring (1997) has argued that causal processes can be understood as the worldlines of tropes.
236 JONATHAN SCHAFFER The concept o causation: It is unclear what underlying conception of f causation is in play throughout Pietroski’s discussion. Does Pietroski believe that his explanation-based sufficient condition holds (i) because causation is to be analyzed in terms of explanation, (ii) because explanation is to be analyzed in terms of causation, or (iii) something else? His official line is that he is merely providing a sufficient condition for causation, not an analysis or reduction (2000, p. 104; also p. 142). Yet his less official glosses are in terms of conceiving of causation in terms of explanation (2000, p. 4; also p. 89), and his applications sometimes presuppose the existence of an unarticulated necessary condition (2000, p. 141; also p. 144).9 Pietroski then returns to this issue in his final chapter (2000, ch. 7), and there endorses the Strawsonian view that causation is a primitive (and at times directly observable) relation.
If causation is to be analyzed in terms of explanation, then, given a DN view of explanation, the resulting concept of causation will be some variant of a lawful sufficiency theory in explanatory disguise. Pietroski ultimately rejects this option (2000, ch. 7, $1.1) because he thinks it requires a Humean regularity theory, which he takes to be in conflict with the intuition that causal relations a re intrinsic t o their r elata (2000, p. 2 19). This is not quite right, since the notion of lawhood in play need not be read in a Humean way.” But in any case I think that the dualist ought to be a Humean. If causal relations are just lav&l regularities, then mental causation will become completely unproblematic.” More on this below.
If explanation is to be analyzed in terms of causation though, then (i) Pietroski’s sufficient condition is trivial, and (ii) the neuralist will deny that the mental event really explains the resulting behavior, since she denies that the mental event really causes the resulting behavior. Nor should we expect her to be swayed by the existence of a DN argument in this case, since she will likely regard DN arguments as merely d efeasible indicators for c ausal relations, and she will likely maintain that one of the defeating circumstances, namely preemption, is in play here.
And if causation is taken to be a primitive (and at times directly observable) relation, then (i) it is hard to resist the idea of ‘causal oomph!’ with its Moreover if causation is prior to explanation it is hard to understand why Pietroski should bother attempting to account for the explanatory asymmetry in the relatively convoluted terms in which he does it, rather than just simply assimilating the explanatory asymmetry to an antecedently given causal asymmetry.
Thus David Armstrong (1999) has defended an identification of causal relations with instantiations of second-order Necessitation relations between universals, yielding a nonHumean lawful sufficiency view. Armstrong notes the ‘welcome consequence’ that, “[Elach instantiation of a universal is complete in itself, so the law will be present completely in each instantiation. So where singular causation is the instantiation of such a law it will be a completely intrinsic relation.” (1 999, p. 184) While I admit that the inhinsicness argument against Humeanism has some intuitive force, I think there are independent and overriding reasons for rejecting intrinsicness, involving cases of causation by disconnection (Schaffer 2000, $3).
REVIEW ESSAYdisastrous implications that overdetermination should generate excess ‘oomph!’ (-the person hits twice as hard, or jumps twice as far). Pietroski rightly rejects the association of causation with ‘oomph!’ (2000, pp. 238-9), but his primitivism is based on the idea that we observe “exerting physical force on physical things or having force exerted on us by physical things” (2000, p. 220) And this feeling of exerting force just is what is meant by ‘oomph!’. Moreover, (ii) it is hard to see why the theoretical posit of additional overdetermining factors wouldn’t be superfluous. I worry that Pietroski has implicitly slid from a reductionist conception of causation, to a primitivist oomph-style conception, to none of the above.
The perils o anti-naturalism: It is crucial to Pietroski’s account of mental f causation that there are psychological laws of nature, and yet Pietroski denies that psychology is a special science and denies that the mental is part of the natural order (2000, p. 8; also p. 165). This is an awkward conjunction.
Pietroski is welcome to stipulate that there are generalizations appropriate to the mental that, while not scientific, still count as ‘laws’ (laws of hermeneutics, perhaps?) But then he is not free to continue to suppose that subsumption under such ‘laws’ is indicative of explanation or ca~sation.l*’~~ This conflict between lawfulness and anti-naturalism comes to a head when one considers the possibility of a ceteris paribus law failing due to breakdowns from below. Pietroski recognizes that laws can fail due to the presence of ‘interfering factors’, but does not discuss whether those interfering factors must be at the same level as the foreground factors (all his examples have this feature) or may also be breakdowns ‘from below’ in which the abNormal circumstances cannot be described in psychological vocabulary, but rather represent failures in lower level mechanisms. Fodor’s (1974) picture of ceteris paribus laws is more attuned to the possibility of
breakdowns from below then to the possibility of interference from beyond:
238 JONATHAN SCHAFFER I take it as obvious that there can be breakdowns from below (indeed my above a rm-rising e xample, with a breakdown i n the C NS, had this feature;
such a breakdown is simply not capturable in the vocabulary of intentional psychology). But if mental events are not macro-events in the hierarchy of the levels of nature, then there is no ‘below’, and there is no explanation for how the breakdown of a physical mechanism could imperil mental causation. I 4 I should like to conclude this section, though, by suggesting that the event (/trope) dualist can provide a plausible account of mental causation if she accepts both naturalism and Humeanism.
By accepting naturalism, the event dualist gains four interrelated advantages. First, she locates mental events in the realm of scientific laws and explanations. Second, she locates mental events in the natural hierarchy with reference to which ‘breakdowns from below’ become explicable. Third, she exorcises what Pietroski himself recognizes to be “the last bastion of the spooky” (2000, p. 234). Fourth, she integrates the mental within an empirically plausible, structurally elegant, and laudibly non-anthropocentric layered world vie^.'^ (At this point the view should no longer be called ‘dualism’, since what is being postulated is a multiplicity of levels, including the physical, the chemical, the biological, and the psychological.) Why then does Pietroski object to naturalism? Pietroski’s primary objection, as far as I can see, is his concern that “our concept of action-what a person does, what she contributes to history-seems to exclude anything that merely happens due to nature.” (2000, p. 153) His concern is that naturalism leaves us no room to distinguish actions such as throwing a rock, from mere happenings such as getting a bruise. But it seems to me that the naturalistic levels-theorist has exactly as much room to make this distinction as Pietroski does: in both cases the distinction is to be drawn in terms of whether there is l4 Noordhof (2001, pp. 27-8) raises this issue. Pietroski responds that (i) his embrace of supervenience allows him to explain this, and (ii) that this objection in some way illicitly assumes that “mentalistic cp-laws are just one more species of cp-law that we discover within the Scientific Image” (2001a, p. 9). As far as supervenience (if Pietroski is even entitled to supervenience: see 93), all supervenience says is that there can be no differences in M-respects without differences in N-respects. But this does not explain why M occurrences that are subvened by N1 occurrences lead to B, while M occurrences that are subvened by N2 occurrences do not lead to B. What needs to be explained is how the subvening differences can bear on the M-B connection, and supervenience alone does not touch this. As far as whether naturalism has been illicitly assumed, 1 don’t understand what portion of the problem Pietroski thinks he is challenging here. Perhaps I am missing the point. Is Pietroski denying that there can b e b reakdows from below i n the mental case?
I5 Kim is perhaps the leading exponent of this picture: “The Cartesian model of a bifurcnred world has been replaced by that of a lnyered world, a hierarchically stratified structure of ‘‘levels’’ or “orders” of entities and their characteristic properties.” ( I 993, p. 337) 239