Reply to Stout
Dept. of Philosophy/Dept. of Linguistics
I agree with many of Rowland Stout's thoughtful remarks. But let me reply by first reviewing the line of thought that led me to regard actions as inner causes of bodily motions, and then saying how I think action is related to agency. For here, Stout and I do evidently disagree about something; although, as I discuss below, I am quite open to the considerations illustrated by his "two gunmen" example. Finally, I'll make a few remarks in reponse to Stout's query about interference and vacuity, with regard to the discussion of ceteris paribus laws.
According to Stout,
Pietroski concludes that actions are inner because they are not bodily motions....Having internalised trying and consequently action, Pietroski must distinguish inner actions from those events which involve the transformation of the world by agents--such events not being inner.
For better or worse, this is not quite how I came to the conclusion. While Causing Actions is primarily about mental causation, I think semantics is often a useful prolegomenon to metaphysics; and the first two chapters are largely about natural language semantics. Like Hornsby and many others, I was impressed by the apparent parallel between 'Nora raised her glass, so her glass rose' and 'Nora raised her hand, so her hand rose'. It has become standard to say that 'raised her glass' means something like 'caused her glass to rise'. But that simple paraphrase is not quite right; and in any case, I think we need an eventish semantics of the sort urged by Davidson. So, borrowing from Thomson and others, I introduced talk of complex events--or processes--that terminate in events like the rising of a glass. And I argued, as a point about the semantics of natural languages, that 'Nora raised her glass' means (roughly) that Nora was the Agent of a complex event that terminated in the rising of her glass; where the Agent of a complex event E participates in E by performing some action A, such that A is a part of E that causes subsequent parts of E. This preserves the important consequence of more traditional causative analyses: if Nora raised her glass, then Nora performed some action that caused her glass to rise.
If this is correct, then at least prima facie, 'Nora raised her hand' means (roughly) that Nora was the Agent of a complex event that terminated in the rising of her hand. So if Nora raised her hand, then Nora performed some action that caused her hand to rise. Seeing no good reasons for resisting this claim--standard arguments to the contrary are easily rebutted--I followed where the semantics seemed to lead. The claim that actions are tryings fit in nicely. And because of semantic considerations stemming from adverb phrases like 'with a pistol' ('with his finger', etc.), I had already distinguished actions from the complex events that "involve the transformation of the world by agents".
This is to treat 'action' (and 'perform') as a quasi-technical term. Other philosophers have used 'basic action' to get at the Agent's causal contribution, abstracting away from its causal consequences (which are due the vicissitudes of nature). But the terminology is irrelevant. The claim is, nearly enough, is as Stout reports: "a person's contribution to the causal order consists in their sorting out their own mind; the rest is up to nature. " But I don't conclude that actions are inner because they aren't bodily motions. I came to think that actions are inner because I came to think--for various converging reasons--that (basic) actions typically cause bodily motions; see also chapter 5, section 3. As I noted in section 3 of chapter 1, semantic considerations alone can't force us to the conclusion that our bodily motions are effects of our personal contributions to the causal order. But I do think that the semantic considerations adduced in chapter one, bolstered by the Cartesian considerations adduced in chapter five (and backed up by the account of mental causation provided in the book), suggest that the best overall account treats actions as inner causes.< 1 >
Stout asks whether Nora's action of raising her arm is (i) a process that results in her arm_rising, or (ii) some event that causes her arm to rise by some other process. For reasons left unclear to me, Stout thinks that once we see this choice, "we must reject the second alternative." But on my view (see pp.29-30, 54), we can refer to either (i) or (ii) by using the ordinary language expression 'action of raising her arm'. Indeed, this is part of my proposed diagnosis of the "one action or many actions" puzzle. Qua theorist, I use 'action', as a quasi-technical term, to pick out (ii). If one finds that terminological choice confusing, one can adopt an alternative.
Still, whatever my route to the conclusion that Nora's (basic) action caused her bodily motion, Stout is right to think that I would balk at the following idea: "transforming the world in line with one's intentions should be at the heart of a philosophical account of agency." It's not that I have no place for Agents transforming the world. On the contrary, my view is explicitly that we are the Agents of complex events that include (as parts) changes in our environments--changes effected by us. And depending on what a "philosophical" account is, I might even be able to accept Stout's claim. But while actions typically affect the world, I don't see any good reason for thinking that my actions somehow include effects external to me--at least not if actions are to track the contributions of persons in a world that can be described in impersonal terms. To be sure, one must guard against "going internal" too quickly. But one must also guard against the tendency--which I do not ascribe to Stout--to "go external" simply because we interact with our environment. And while I would not pretend to have established anything, Causing Actions is indeed an attempt to provide an account of mental causation congenial to the idea that (basic) actions are volitions--once this intuitive view is purged of the bad idea that every volition is caused by a prior action.< 2 >
Stout rightly says that the kinds of cases I discussed were highly simplified. As I noted in the book (pp. 45, 112ff, 147), any serious account of action has to deal with actions like going to the airport, which involve sequences of simpler actions organized by intentions and plans (of the sort discussed by Davis, Bratman and others). I abstracted away from such complexity in order to get a handle on how even the simplest cases of mental causation are possible without the assumption that mental events are biochemical events. And where Stout sees "artificial examples" and an "unhealthy interest in mere bits of behavior," others might see bets about how to usefully attack the really hard problems by starting with simpler ones. There is a fine line between useful idealization and distracting distortions. But it's very hard to tell in advance which simplifications are distorting. (Do physicists--biologists, linguists, etc.--exhibit an unhealthy interest in simplified problems?) I certainly agree with the need to accommodate and (with help) account for more interesting cases of action; though I don't yet see reason for thinking that my proposed ways of dealing with the simple cases will make it harder to do so.
In fact, I quite like Stout's distinction between gunman-1 and gunman-2: the former was merely "having a go," which turned out to be successful, at shooting the president; while the latter was intent on shooting the president, and so would have fired again had he missed. Since both gunmen satisfy the predicates 'shot the president' and 'tried to shoot the president', one should not assume that everyone who satisfies these predicates is the same from an intentional perspective that classifies Agents according to the kinds of personal causal contributions they make. Exactly right.< 3 >
Nonetheless, from my point of view, both gunmen performed an action such that: it initiated a pulling of the trigger, and it initiated a shooting of the president. So in each case, there was an action that was both an action of shooting and an action of pulling. But gunman-2 did not merely perform an action that turned out to be an action of shooting; gunman-2 performed this action in the context of a broader plan. For gunman-2, the action of pulling the trigger the first (and as it happpened, the only) time was part of a plan to get the president shot; and that plan included contingency plans--shoot again if the first shot misses, and so on. This is, as Stout notes, an important difference between the two gunmen; and it would be very nice to have more to say about this. But I was focussed on the puzzles concerning mental causation that attend even the less interesting action of gunman-1. And I don't see how reflection on gunman-2 shows that the action by gunman-1 was not causally upstream of his finger motion. Nor do I see how the existence of a broader intentional plan shows that the actual action by gunman-2 (of pulling the trigger exactly once, as it turned out) was not causally upstream of his finger motion. I agree that the underlying causal processes are different in the two gunmen, and that the action of shooting the president is not identical with the action of making sure that the president is shot. But the latter complex action, involving a plan and the action of assessing the situation after the first shot was fired, presumably involved the simpler action of pulling the trigger (before checking to see if further action was needed). And I see no reason for denying that this simpler action, which caused the discharge a bullet which entered the president, was both an action of pulling and an action of shooting.
Stout graciously remarks that even if I am wrong about all this, the heart of Causing Actions is untouched. But while I like the idea of getting the same results with less work, I suspect that "neuralists" would be (even more) unimpressed with arguments for not identifying actions with biochemcial events, if actions are not inner tryings that cause bodily motions. One of the prime "arguments" for neuralism is that the inner causes of (human) bodily motions are biochemical causes; the only alternative, it is standardly alleged, is some kind of Cartesian dualism. Part of my aim in Causing Actions is to reply as follows: our actions themselves are inner causes of our bodily motions (cf. Armstrong, Hornsby); but these causes are not biochemical; so we should be prepared to find out that the mental causes of actions are not biochemical either.
Finally, Stout asks me to respond to a concern about the issue of "independent interference" as it arises in my discussion of ceteris paribus laws. This is, in my view, a good place to press. But let me stress that I was not offering any analysis of ceteris paribus laws (or 'ceteris paribus laws'); see p. 123. My aim was to make some suggestions about how to understand talk of idealizations, which we clearly need in science, in a way that shows why "hedged" laws need not be (and typically are not) vacuous. The (familiar) idea was that, modulo worries about vacuity, appeals to ceteris paribus laws are legitimate in accounts of mental causation. In short, if we can make mental causation no more mysterious than ceteris paribus laws, I'd regard that as progress.
Let me also note that the appeal to "independent" intereferers was introduced to handle a different kind of concern from the one Stout raises. Suppose someone states an alleged (hedged) law of telekinesis, and they "explain" the nonmovement of spoons by citing local "ectoplasmic disturbances" that interfere with their capacity to move the spoons. I think such a person is committed to there being some effect of the hypothesized "ectoplasmic disturbances" apart from the convenient interference of telekinetic powers; otherwise, their alleged telekinetic law would be vacuous. As "a (surely imperfect) attempt to capture this intuition (p.125)," I offered the constraint referring to logical/analytic/casual consequences of apparent exceptions to laws. And this was all in the context of proposing a sufficient condition for nonvacuity that can be satisfied by intentional ceteris paribus laws. So the dialectic gets mildly complicated.
(And there are issues, discussed briefly in the book, about whether basic physical laws present special complications.) But in any case, Stout wants to know why the fact that "the spoon is stationary and subject to no physical forces" is not itself an interfering factor.
I'm not entirely sure what to say about this. But my inclination is not to take a hard line and declare that being stationary just can't be an interfering factor. If the absence of oxygen can explain why a struck match doesn't light, why shouldn't staying still be able to explain something? (Especially if one has action in mind.) I see no reason, however, to think that being stationary can be cited as an interfering factor that explains why the spoons don't move. That seems too close to "explaining" P by saying P.< 4 >
But what about being subject to no physical forces? Could that be an "interfering" factor that explains why the alleged telekinetic law faces (what I call in the book) an abNormal instance in which the spoons don't move? Well, only if the alleged telekinetic power is supposed to be some kind of nonphysical force; but then the absence of physical forces would, prima facie, appear to be an ideal case for the manifestation of telekinetic powers. So we would need to hear more about why the absence of physical forces interferes with telekinesis. (See the related discussion at pp. 133-5). And similarly if the alleged interferer was being subject to no (other) physical forces. It's possible, I suppose, that someone could tell a story according to which the telekinetic power is manifested only in the presence of other forces. But then the question would be whether the alleged telekinetic power explains anything--as required by clause (iii) of the condition proposed on p. 126--not already explained by the other forces.
Unsurprisingly, the surrounding issues quickly get very complex; see section 2 of chapter 4. But luckily, I don't need a "philosopher's stone" that distinguishes good science from pseudoscience. (And Stout does not suggest otherwise.) For my purposes, it is enough if there are nonvacuous (though hedged) intentional laws of the sort described: generalizations such that one can explain why the generalizations face the counterinstances they do by citing factors that one ignores when stating (in an idealized way) the effects of certain antecedent conditions.
1_ Thus, while I agree that "Booth's personal causal contribution would have been the same if he had merely tried to pull the trigger but his muscles had been temporarily paralyzed," I would not expect any philosopher not antecedently convinced to accept this premise and immediately conclude that "Booth's personal causal contribution--his action--is causally upstream even from the motion of his finger." And for the record, I would be happy to say that the (complex) event of Booth pulling the trigger was distinct from the (complex) event of Booth shooting Lincoln. But it does sound wrong (to my ears) to say that the causing of Nora's arm to rise caused Nora's arm to rise. I wouldn't, however, put much weight on such judgments.
2_ Indeed, I find myself growing more internalist day by day. My current work on semantics leads me to think that while speakers obviously use language to talk about things around them, the meanings of linguistic expressions have little if anything to do with connections between those expressions and the things speakers talk about--at least not if meaning is something we can study scientifically in the way that linguists apparently do. (For discussion, see my "Character of Natural Language Semantics" forthcoming in Epistemology of Language , edited by Alex Barber: Oxford, OUP.) If this is correct, as Chomsky has been urging, perhaps we should undertake a wholesale reevaluation of externalist trends initiated by reflection on meaning. This leaves open the possibility that rational relations hold between entities that are at least often individuated by reference to the things we think about. But if the meanings of natural language expressions can be individuated without regard to rational relations, as Chomsky has been urging, matters are very complex indeed.
3_ I am fully open to this point. See section 4.2 of chapter 1, and especially p. 45, where I stress that not all tryings are (basic) actions--if only because one can try to do one thing by doing something else. My view is that paradigmatic actions are tryings. I readily grant that 'trying' may be satisfied by a diverse range of events.
4_ There is also a fair amount of discussion, towards the end of chapter 4, about how hedged-law explanation works by "partially" accounting for the explanandum via the hedged-law and "taking up the slack" by appeal to factors that the law idealizes away from.