How to cite

Getting Personal: Pietroski's Dualism

by Paul Noordhof e-mail

Department of Philosophy
University of Nottingham
University Park
Nottingham NG7 2RD - UK

Paul Pietroski’s book Causing Actions is a, at times, detailed, technically accomplished and sophisticated and, at times, programmatic and suggestive defence of a very weak kind of Nonreductive Physicalism. In effect, he is being as Dualist as you can be without being tarred with Descartes’ brush. The book begins with an analysis of action sentences and a claim about actions. This was originally set out in a Mind paper (Pietroski (1998)). According to Pietroski, almost all actions are tryings and tryings are internal states. There is the odd special case he sets aside which I shall mostly ignore (p. 47). He adopts a position very close to that of Jennifer Hornsby (Hornsby (1980)). I found Pietroski’s analysis of action sentences plausible but thought that there was little connection between it and his substantive thesis about action.

                In the second chapter, Pietroski provides an account of the logical form of ascriptions of propositional attitudes. He explains how a believer in Fregean thoughts can, at the same time, endorse semantic innocence. Again, I found that Pietroski’s technical proposal had plausibility but it struck me that this was orthogonal to the position that he worked out on explanation. All he needed for the latter was the idea that there are Fregean thoughts. Their exact role in the semantics of propositional attitudes could be left to one side. For this reason, I will not discuss this element of Pietroski’s book further. Of course, the detail of his presentation may presuppose his views on the semantics of propositional attitude ascriptions but that is a different matter.

                Pietroski presents his main arguments in favour of what he calls Personal Dualism in Chapter 5. As he acknowledges, part of his aim is just to get the possibility of this position into view. He suggests we can subject the arguments to more scrutiny later. I will focus on whether the considerations he offers require us to take seriously the possibility of Personal Dualism. There is the suggestion that they do. His main target is Neuralism: the thesis that mental events are identical or constituted from neural events. He views this doctrine to be inimical to the personal perspective. I shall explain why he has not established this. In particular, I will focus on the argument upon which he places the most weight: Hornsby’s argument from differential vagueness. I find the argument wanting.

                I will finally turn to Pietroski’s defence of the compatibility of Personal Dualism, Psycho-physical causation and causal completeness of the non-mental. Elements of this defence take up Chapters 3, 4, 6 and 7.I argue that, while Pietroski’s account of ceteris paribus laws might be plausible, they are ill-suited to answer worries about efficacy. In order to resolve these difficulties, the distinctive character of Personal Dualism begins to chip away.


1.             All actions lie within

In Chapter 1, Pietroski presents us with two things: first, an analysis of action sentences and, second, a substantive thesis about the nature of actions, namely that they are almost all internal tryings.  Before turning to the connection between these two theses and the plausibility of the substantive thesis, I will set out the details of his analysis.

                Let’s begin with two of Pietroski’s initial action sentences

(1)           Booth shot Lincoln with a pistol

(2)           Booth pulled the trigger with his finger.

Pietroski finds the following three theses compelling.

(A)          Event Analysis: Action sentences should be analysed in terms of quantification over events.

(B)           Action Identity: (1) and (2) are made true by the same action.

(C)           Event Distinctness: the event of Booth shooting Lincoln and the event of Booth pulling the trigger are distinct events (p. 20).

Pietroski suggests the sentences should be provisionally analysed as follows.

(1A)                ($e)[Agent (e, Booth) & ($f)(Terminates-in (e, f) & ShotI (f) & Patient (f, Lincoln) & With-a-pistol (f))].

(2A)                ($e)[Agent (e, Booth) & ($f)(Terminates-in (e, f) & PulledI (f) & Patient (f, the trigger) & With-his-finger (f))].

These need a bit of explanation. ‘($e)’ and ‘($f)’ quantify over events. The formulations take the attribution of actions to be descriptions of events. Hence they respect Pietroski’s intuition that action sentences should be given an Event Analysis. ‘Agent’ and ‘Patient’ capture the thematic roles associated with action-verbs; initiator and affected (p. 26). The decomposition represented by (1A) captures the following inferences:

(I)            From ‘Booth shot Lincoln with a pistol’ it follows that ‘Booth did something with a pistol’ [($e)[Agent (e, Booth) & ($f)(Terminates-in (e, f) & With-a-pistol (f))]. ‘Booth did something with a pistol’ does not follow from ‘Booth pulled the trigger with his finger’ (pp. 18-20).

(II)            From ‘Booth shot Lincoln with a pistol’ it follows that ‘Booth did something’ [($e)[Agent (e, Booth)]] (pp. 26-27).

(III)           From ‘Booth shot Lincoln’ it follows that ‘Something happened to Lincoln’ [($f) [Patient (f, Lincoln)]] (pp. 26-27).

                As Pietroski points out, for a wide class of tensed verbs there are both transitive and intransitive readings (indicated by the subscripts). Thus

(3T)         Booth raisedT his arm.

(3I)          Booth’s arm roseI.

(4T)         Booth shotT Lincoln.

(4I)          Lincoln was shotI.

(5T)         Nora meltedT the chocolate.

(5I)          The chocolate meltedI.

(3I), (4I) and (5I) follow from (3T), (4T) and (5T). Pietroski captures this in his analysis by proposing that

(TI)         VT(e) & Patient (e, y) iff  ($f) (Terminates (e, f) & VI (f) & Patient (f, y)).

In other words, a transitive doing occurs if and only if it terminates in an event which is characterised by the intransitive form of the verb applying to the object to whom the transitive doing was done. If there was a meltingT of the chocolate, then the meltingT terminates in an event which is the chocolate meltingI. Terminates-in (e, f) is defined as, f is a perhaps improper part of e and f is an effect of every proper part of e that is not a part of f (p. 33). The proposed analysis  given of (1) and (2) bears witness to the relations between transitive and intransitive forms just identified. The transitive reading is obtained from the transitive-intransitive conversion principle (TI).

                Simplifying a bit, Pietroski captures (C), Event Distinctness, by suggesting that, if the domain of quantification for (1A) and (2A) includes complex events, then there will be a complex event e1 which satisfies (1A) and not (2A) and another, e2, which satisfies (2A) but not (1A).  e1 will be with a pistol, e2 will not. We don’t pull the trigger with a pistol, generally speaking. e2 will be over before e1. e1 will terminate in Lincoln being shot, e2 will terminate in the trigger being pulled. (pp. 21-28).

                This brings us to, in many ways, Pietroski’s key intuition for the development of his account of the nature of action, namely (B): Action Identity. The first step in his attempt to capture this intuition is to analyse (1) and (2) further to explain how the same action could ground both events, that of the shooting and of the trigger pulling. The analysis is not meant to establish that the same action grounds both. It is just meant to be compatible with this possibility. Thus we get

(1AG)                ($a){($e)[grounds (a, e) & action (a, Booth)] & ($f)(Terminates-in (e, f) & ShotI (f) & Patient (f, Lincoln) & With-a-pistol (f))}

(2AG)                ($a){($e)[grounds (a, e) & action (a, Booth)] & ($f)(Terminates-in (e, f) & PulledI (f) & Patient (f, the trigger) & With-his-finger (f))}

Pietroski leaves it open whether (1AG) and (2AG) represent a further articulation of the logical form of action sentences or merely just a deep truth about agency (p. 29). The italicised phrases - my italics - are supposed to provide an account of what it is to be the agent of an event. He claims that the actions picked out by these analyses are identical. Action sentences are ambiguous referring to both the complex events described above and the grounds of the event which are the actions (p. 29). The ambiguity thesis also presents a way round a problem that faces Hornsby’s account of the semantics of action sentences. Hornsby appeals to Causes (e, f) in place of Pietroski’s Terminates (e, f). Hornsby’s analysis appears to have difficulty with

(7)           Nora melted the chocolate with her lens

(8)           Nora moved the lens with her hand.

The adjuncts ‘with her lens’ and ‘with her hand’ do not modify the terminating events. Chocolate does not melt with a lens. Yet they do not modify the causes of the terminating event either if Action Identity is true. The following are false.

(9)           Nora melted the chocolate with her hand.

(10)         Nora moved the lens with her lens.

Pietroski resolves this by taking the adjuncts to modify the complex event e identified above (pp. 37-39).

                I reported above that Pietroski left it open whether (1AG) and (2AG) just represent a deep truth about agency or capture the logical form of action sentences. This suggests that he views his final analysis as neutral over the nature of action. I think this is right. However, sometimes he appears to suggest that his logical analysis plays an essential role in an argument for all actions being internal tryings. He writes at the beginning of his presentation of the analysis in Chapter 1.

When a person (intentionally) moves some part of her body, the motion of her body is an effect of her action, which is a trying to do something. This thesis is defensible; our best semantics of actions sentences requires it, I argue, given some strong intuitions about actions (p. 18, he repeats the claim on p. 46).

But when we came to his introduction of actions as grounds of complex events we saw that, in fact, the analysis presented in (1A) is quite compatible with not taking all actions to be internal tryings. Instead, he explicitly introduces the grounds analysis to enable him to capture Action Identity. He writes

The trick is to see how, given such an account, Booth’s action of pulling the trigger can still be his action of shooting Lincoln.

                This is where appeal to thematic roles comes in. Suppose the following thesis is roughly correct: Agent (e, N) ¬ ($a)[grounds (a,e) & action (a, N)]; N is the agent of event e, just when e is grounded by an action of N (p. 28)

He also conclusively remarks.

the truth of such constructions does not semantically require that actions be inner events. There is conceptual room for the claim that (many) actions are bodily motions (p. 39).

So Pietroski’s analysis is at twice remove from his view of action. First, the further analysis is motivated by Action Identity and not the other way around. Action Identity is a necessary premise in the argument for all actions being internal tryings. Second, the further analysis is merely compatible with his view that actions are internal tryings rather than an essential component in an argument for it.[1] As we shall see below, the positive arguments he provides don’t appeal to the details of his analysis of action sentences. I’m afraid that this rather plays into my prejudices. I believe that rarely, if ever, have semantic analyses shown the nature of worldly entities. Rather, our metaphysical assumptions are fleshed out in semantic analyses. At best, the latter can only make us more explicit and careful about the former. Be that as it may, I think should conclude that the support for Pietroski’s view of action must come from elsewhere. Let us not be bedazzled further by the semantic pyrotechnics and turn to the considerations for taking all actions as internal tryings.

                Pietroski offers roughly five reasons for taking almost all actions to be internal tryings, some of which are familiar from previous writing on the subject. I think they are inadequate. Let me make it clear that I do not resist the idea that internal tryings are actions. What I resist is the idea that, with the odd exception, only internal tryings are actions. I write ‘internal tryings’ because Pietroski supposes that actually grappling with a rock, trying to lift it, is also a trying. It is just that it is an external trying and not an action (p. 45). I consider his reasons for supposing that only internal tryings are actions in turn.

                The first reason is that, intuitively, an agent’s actions are under his or her control in a way that, on reflection, bodily movements fail to be. Bodily events require the body to be cooperating. If actions were also to include events involving the world, then they would require the world to be cooperating too (pp. 23, 42, 44). This makes them even less under the agent’s control. For instance, Booth’s shooting of Lincoln requires Lincoln actually to be shot if the shooting is a world-involving action. So, the thought runs, actions are none of these things. They are just internal tryings.

                In response, we should begin by distinguishing between two senses of control. In one sense, A is under an agent’s control if (i) the full causal explanation of A lies just in the beliefs and desires of the agent, suitably arranged for an exercise of agency and (ii) A is not metaphysically dependent upon the existence of other things in the agent’s environment. For instance, if I turn the door handle, I could not do so unless a door handle were there (arguably, a metaphysical dependence). Equally, I could not open the door without the capacity of the hinges to turn (arguably, a causal dependence). This is my attempt to capture what Pietroski means by writing

It was up to Booth whether or not to act as he did; but it was not up to Booth that Lincoln would die, or even that the bullet would leave the pistol (p. 23)

to capture his sense of control. I concede that, in this sense of control, neither bodily events nor worldly events are under our control. In the second sense of control, A is under our control if we can reliably produce A if we want to. This does not mean that we require control in the first sense. We can arrive at beliefs about the way the world is constructed and how to act upon it in order to produce A. If we are careful in the formation of our beliefs, we can make sure that for a wide range of cases we can produce the effects we want in a reliable fashion.  We minimise the sense in which we are in the hands of the world in attempting to produce A.

                Pietroski has given us little reason to suppose that the sense of control which is most relevant to understanding action is the first one. The second seems just as plausible. In which case, there is nothing to rule out bodily events and some worldly events from being actions. In addition, it is not clear that even internal tryings satisfy the first sense of control. For one thing, it seems that the kind of trying in which an agent is engaged can metaphysically depend upon worldly facts. Hornsby, the main proponent of the position Pietroski defends, gives a nice example of a person trying to learn a Shakespearean sonnet. She remarks that whether or not someone is trying to learn a Shakespearean sonnet depends upon whether it is, in fact, a Shakespearean sonnet that the person is trying to learn (Hornsby (1980),  pp. 35-36).

                It is also plausible that the trying is causally dependent upon states of an agent’s brain. Just as my moving my arm could not take place without the firing of nerves in my arm plus a network of causal relations both in my arm and between my arm and the world, so my trying could not take place if there were not a brain with a certain causally supportive neural-functional architecture. We might be unaware of how our tryings are supported in this way but supported they must be.

                I think the matter is obscured by a cross-cutting issue. The line of thought behind Pietroski’s denial of the status of actions to bodily movings seems to be this. In moving my arm, I am trying to move my arm. The arm moving is a causal consequence of the trying. The causal consequences - the firing of nerves in my arm leading to movement etc. etc. - are extraneous to my action of trying. An agent’s control is undermined by causal dependence on extraneous factors in this sense. I have acted, the thought might run, in trying and after that I am in the hands of my nerves. How could what results be an action? But this seems the wrong way to look at things. When I try, I am in the hands of my brain. I couldn’t try unless my brain were a certain way. This is also partly extraneous to my action. So far, things are on a par. The firing of the nerves after the trying are only extraneous to action in some stronger sense if one is already convinced that bodily movings aren’t actions. Otherwise, the relationship between my arm moving and the firing nerves in my arm is like the relationship between my internal trying and the neural-functional structure of the brain. Both the neural-functional structure of my brain and the structure of nerves-firing, and sundry causal relations in the arms, are partly constitutive of an agent’s actions and partly extraneous causal support.

                The second reason which Pietroski offers for supposing that almost all actions are internal tryings is that actions are things we explain completely in terms of an agent’s beliefs and desires. Actions are the contribution that an agent makes just in virtue of his or her psychology. Bodily events are not like that (p. 23). The points I made with regard to the first reason apply here too. There will be non-psychological factors that are necessary to cite in a full explanation of action whichever way we understand action. So there is no special motivation for denying that movings of one’s body can be actions. Of course, additional non-psychological factors will have to be introduced the more worldly one supposes actions to be. However, it is hard to see why this should be taken as a difference in kind.

                The third reason Pietroski offers is that his understanding of action avoids the problem of deviant causal chains. It is acknowledged on all sides that not all bodily events are actions. Pietroski alleges that, if we are to distinguish between those which are and those which are not, we will have to appeal to their intentional causal history.  However, not just any old intentional causal history will do. Sometimes the connection between an agent’s beliefs and desires and the bodily events which result is deviant. The famous example is of the climber intending to let go of his falling colleague, because of an understandable set of beliefs and desires, with the result that his nervousness at having such an intention makes him let go of the rope rather than the letting go being something he did (Davidson (1973), p. 79).  Pietroski remarks that there is no satisfactory account of when there is deviance and when there is not. Denying that bodily movements count as actions removes this problem (pp. 39-40). We won’t have to identify the right kind of causal path.

                This reasoning strikes me as unsatisfactory in a number of ways. First, we will have to find a solution to the problem of deviant causal chains regardless of what we count as actions. We do not hold people responsible for what is the deviant upshot of their beliefs and desires, we do for what we might otherwise be inclined to call their actions (better, a class of their actions). Similarly, the problem of deviance will require some resolution in explaining the nature of perception. While I am all for avoiding problems if we can, I don’t think there is much merit in avoiding problems in one area at the cost of some intuitiveness if we are still going to have to resolve the problem anyway. Second, the ‘no satisfactory solution has been found’ style of argument is weak because, in general, it can be run to vitiate most philosophical programmes.

                A third problem concerns the matter of successful and unsuccessful actions. What makes S’s trying to do A a successful doing of A? For instance, suppose I try to make somebody laugh by being witty and succeed not because I am particularly witty but because my voice takes on a slightly strangulated staccato type quality that strikes them as amusing. I am amusing them but amusing them is not something I am doing in the sense of being an action of mine. This is different from the case of being genuinely witty and as a result of that amusing someone. Pietroski would seem to lose the distinction between the second type of case and the third. He seems to suppose that successfully doing A is the result of trying to do A and the trying causing A to happen (p. 49). By contrast, Hornsby acknowledges that, in order correctly to describe the trying to do A as a doing of A, there will have to be a non-deviant causal chain (Hornsby (1980), pp. 122-123). By insisting that, with the odd exception, only internal tryings are actions and claiming that, whatever else we suppose to be an action is just an effect of the trying, Pietroski loses the distinction between doing something which, in fact, caused A to happen and doing A. I think that this shows how the notion of action extends further than internal tryings and involves the distinction between deviant and non-deviant causation. Pietroski’s failure to be attentive to this difference is a consequence of his restrictive notion of action.

                Pietroski’s fourth reason for rejecting the idea that actions may include bodily movings is that it forces us to distinguish between bodily movements and actions extrinsically by causal history (p. 40). There are two sources of difficulty with this. The first is that it is by no means clear that we have a strong intuition that actions are intrinsically different from mere bodily movings. I think that we can hardly take this as a datum. However, that does not mean that I think we should distinguish between actions and non-actions extrinsically. There are some occasions when it is implausible to ascribe beliefs and desires to explain why something is an action rather than a mere bodily movement. Mindlessly drumming our fingers or scratching are actions but it seems unlikely that a piece of practical deliberation is behind them. Everything else speaks against such ascription. It is just theory-driven. Unfortunately, these cases don’t provide support for Pietroski’s favoured picture since he concedes they are problems for him too. He does not think it appropriate to suppose there are tryings behind the drumming yet he concedes that actions take place (p. 47). These are the exceptions to the claim that all actions are internal tryings.

                The second source of difficulty is that it is by no means clear that we must explain the difference between bodily actions and bodily movements by appeal to causal history, contrary to what Pietroski says. I take it that a bodily action begins with an internal trying and involves a succession of non-deviantly causally related events ending with a bodily movement. This is an intrinsic matter. I differ from Brian O’Shaughnessy just in emphasising the importance of non-deviant causal chains (O’Shaughnessy (1973), pp. 385-386). The only way Pietroski’s objection seems plausible is if you make the mistake of supposing that a bodily moving is a bodily movement with a certain causal history rather than the complex event I have just described.

                This brings me to Pietroski’s fifth reason: Hornsby’s muscle contraction argument (pp. 41-42; Hornsby (1980), pp. 20-32). Suppose that Nora wants to contract the muscles in her forearm. In order to do this, she clenches her fist. The argument then proceeds as follows.

(1)           Nora’s contractingT her muscles is her clenchingT her fist (Action Identity).

(2)           The clenchingI of Nora’s fist was caused by the contractingI of her muscles.

(3)           Nora’s clenchingT her fist is not identical to clenchingI her fist and Nora’s contractingT her muscles is not identical to contractingI her muscles (from (1), (2) by symmetry).

(4)           VT(e) & Agent (e, x) iff  ($f) ((Terminates (e, f) or Causes (e, f)) & V (f) & Patient (f, y).

(5)           Nora’s contractingT her fist either causes or terminates in the contractingI of her muscles (from (1), (3), (4)).

(6)           Nora’s contractingT her fist does not terminate in the contracting of her muscles.


(7)           Nora’s clenchingT her fist causes the contractingI of her muscles (from (1), (5) and (6)).

Bearing in mind my previous remarks about the role of non-deviant causal chains, we need to interpret ‘Terminates (e, f)’ or ‘Causes (e, f)’ to apply only to non-deviant cases if we want ‘e’ in (4) to stand for actions. If it did not stand for actions, it would be inappropriate to appeal to the principle in the context of the argument. I shall make the assumption that (4) is meant to govern actions in what follows. As I noted before, Pietroski takes action sentences to be ambiguous. They either advert to complex events or actions. I agree with Pietroski that action sentences are ambiguous. However, I think that they are ambiguous in the action to which they refer. The soundness of the argument depends upon which reading we take.

                Action sentences might be taken to mean the first action in the chain of non-deviant causes giving rise to a certain action-defining effect E, for instance, clenching the fist. In which case, (4) is true in virtue of the ‘causes’ clause. It is unlikely that muscle contractings are the first cause in the non-deviant chain. Hence (6) is true and we may endorse (7). According to the second reading of action sentences, we don’t focus on the causally basic but, what we might, call the teleologically basic: the action by which an agent aimed to do other things. Nor aimed to contract her muscles by clenching her fist. In which case, the action is anything up to and including the successful achievement of the teleologically basic aim of the agent through which he or she hoped to achieve other things. So if Nora manages to clench her fist, then the action includes the event of clenchingI. If, on the other hand, she is stopped from clenching her fist, then the action runs up to that point and we say she tried to clench her fist. (1) to (3) are true on the second reading. (4) is true of teleologically basic actions. If the agent is stopped from doing something which terminates in VIing then (4) is not falsified because we say tried to VT rather than VT.  On the other hand, (4) is not true of the teleologically non-basic. Nora’s contracting of her muscles terminates in her clenching. To think otherwise is to confuse what we might call one of the non-basic success conditions of the action from the actual extent of the action. In which case, (5) and (7) don’t follow. If this is right, then premise (1) has no tendency to drive the action inwards rather than, as we might say, driving the contracting outwards.

                Pietroski might argue that his theory is simpler and cleaner than the suggestion I have put forward. He claims that almost all actions are internal tryings. Other bodily movings which we might be tempted to call actions are just, by his lights, caused by tryings. Unfortunately, simplicity and cleaniness aren’t everything. His theory ignores an important dimension of action which, although it might give rise to the problem of deviant causal chains, also promises a solution. I have argued that bodily actions are not just those caused by tryings but those partly constituted from tryings plus a non-deviant causal chain leading up to the definitive bodily movement. The non-deviant causal connection between trying to do A and A-ing should be captured in terms of reliable functioning. The thought is that bodily actions (rather than mere movements) are reliably produced by an agent in a way which would allow for the agent to guide his or her action. Part of the reliable production may depend upon proprioceptive and other sensory feedback mechanisms. The guidance would come from the intention or plan behind the action when appropriate (Mele and Moser (1994), pp. 41-52). For simple actions, this guidance need not actually be going on, low level feedback might be enough. In more complex actions, the intention or plan is likely to play a more central role. It is the notion of feedback and monitoring which extends the notion of action beyond internal events of the subject to interactions with the world.

                Obviously this picture needs fleshing out. However, I think we have enough to explain why flukes and spasms don’t count as actions even if they actually satisfy what an agent was intending to do. They depart from the means by which an agent may guide his or action. If I seize the door handle and turn it, there is a world of difference between doing this as a result of a lucky spasm and doing it partly as a result of feedback from my arm movement to eventual clenching. A non-deviant causal chain is one which allows for the normal play of feedback and adjustment at all relevant points in the causal chain. A deviant causal chain breaks the potential cycle of feedback and adjustment at some point prior to the termination of the action. The agent’s intention controls and modifies the action as a result of the feedback. This is but a brief sketch. It is just to give a feel for the kind of position that might contrast with one favoured by Pietroski.


2.                Personal Dualism and the Argument from Differential Vagueness

Pietroski argues for a form of Personal Dualism. According to Pietroski, persons are primitive although not ontologically basic. Part and parcel of Personal Dualism seems to be the following theses.

(a)           Persons have a corporeal nature (are essentially embodied) (pp. 164-168).

(b)           Persons are the locus of freedom and they operate in the space of reasons (p. 150).

(c)           Mental events cannot be described without bringing a thinking subject onto the scene i.e. they cannot be described impersonally. By contrast, neural events can be described impersonally (p. 154)

(d)           Neuralism is false (pp. 5-6).

Much of Pietroski’s discussion is influenced, as he acknowledges, by P. F. Strawson’s views on persons (Strawson (1958); Strawson (1959), pp. 87-116). It is suggestive rather than demonstrative. I thought it might be helpful to focus on the connection between Pietroski’s views of persons and Neuralism. He suggests that Neuralism is ‘a capitulation to the idea that human actions are just one more species of impersonal occurrence’ (p. 178) or again ‘neuralism actually threatens the idea that reasons are causes by identifying mental events with events characterizable without reference to persons. Identifying reasons with impersonally characterised events threatens our view of ourselves as agents whose actions are free’ (p. 12).

                We might get a clearer idea of the force of Pietroski’s own position if we contrast it with two other options that seem available to those who are convinced by (a) to (c) above: Emergent Property Dualism and Anomalous Monism. According to the Emergent Property Dualist, there  are personal properties over and above neural properties. They are related to neural properties by emergent psycho-physical laws.  If events are coarsely individuated, Neuralism could still be true. Neural events would just, also, possess these personal properties. The neural events I have in mind would take the form: the activation of S’s brain at location L. Specifying neural events in this way suggests that (c) should be given a more substantial reading. The neural events so specified are essentially characterised by reference to a person, S. I take it that this would not be enough for Pietroski though. For the sake of argument, I will assume that the activation of S’s brain at location L is an impersonal description in the relevant sense and use ‘impersonal’ in this sense hereafter. Then in order to capture the force of Pietroski’s emphasis on personal properties, the Emergent Property Dualist would attribute (additional) personal properties to the neural events. These personal properties could be characterised in no other way but in terms of the perspective of S and subjects like S. Although Emergent Property Dualism would allow that mental events could be described impersonally by mentioning their neural properties.  They could not be so described if they are to be described in their mental aspects.

                If events are finely individuated, then it is plausible that personal events are distinct from neural events according to the Emergent Property Dualist. In which case, Neuralism would be false. However, and this is a point whose relevance will be clearer shortly, we could still pick out the neural events with which the personal events are correlated as a result of psychophysical law. If Personal Dualism were understood as a kind of Emergent Property Dualism, then the truth or falsity of Neuralism would seem far less interesting. All that matters is whether there are personal properties. I think it is clear that this is not what Pietroski has in mind.

                Pietroski’s Personal Dualism is not a version of Davidson’s Anomalous Monism either (Davidson (1970)). Davidson, like Pietroski, claims that the personal point of view cannot be abandoned without changing the subject and no longer talking about minds. Davidson does not quite put the emphasis on persons in the way that Pietroski does though. Instead the key idea is that we adopt a scheme of interpretation by which we make sense of people’s behaviour through ascribing beliefs and desires to them. Nevertheless, he need not reject Pietroski’s stress on the importance of persons being taken as primitive. There is no radical disagreement here. The disagreement lies in Davidson’s espousal of Neuralism. Davidson would not view the identification of mental events with neural events as giving up on the personal perspective or only playing lip service to it. The perspective we adopt in interpreting people’s behaviour is just another way of identifying neural events. Keeping the personal in view is refusing to give up on this perspective.

                As far as I can see, Pietroski’s reason for rejecting either of these two positions lies with the argument from differential vagueness. He may suspect that the other two positions fail to capture the personal perspective appropriately. Nevertheless the official justification and rationale lies with the argument he finds in Hornsby’s early work (Hornsby (1981)). In Pietroski’s hands, the argument runs as follows.

(1)           If an event E is identical with an event F, then they have the same vagueness of spatiotemporal boundaries.

(2)           Mental events have different vagueness of spatiotemporal boundaries from any neural events.


(3)           Mental events are not identical with or even constituted from neural events (p. 173).

I will not challenge premise (1) which seems to be an uncontentious application of Leibniz’s law. Pietroski recognises that (2) is open to challenge. We might just be ignorant of the precise boundaries of mental events. Pietroski suggests that for the Neuralist to appeal to this would be question begging (p. 175). However, this seems a little quick. Our understanding of mental events has grown up from two sources. First, our introspective appreciation of our own mental goings on. Second, our observation and attempt to make sense of the behaviour of others. Neither requires obtaining precise information about the spatiotemporal location of mental events. In the case of introspection, there is a fairly clear sense in which it does not provide a spatial presentation of the mental events about which it is meant to provide information. So the Neuralist has an explanation for why it is not presently possible to say anything more precise. This ready explanation of the vagueness which attaches to mental ascriptions can be cashed out into either an Epistemic or a Semantic account of the vagueness of mental concepts. I would not wish to proclaim between them here.

                In response to this anxiety, Pietroski makes a number of moves. He writes ‘I see no independent reason for supposing that tryings have much sharper boundaries than initial reasons suggest’ (p. 175). This suggests that either he has considerations to show that non-ontological accounts of vagueness are uniformly inadequate or that there are special considerations to show that the vagueness which attaches to mental ascriptions cannot be explained in non-ontological fashion.

                I will pass over general considerations (as indeed does Pietroski) and consider whether there is something more to be said in favour of his reading of the vagueness drawn from his reflections about persons. A preliminary point is that, if Emergent Property Dualism properly captures the personal perspective, there is further reason to suppose that a non-ontological explanation of vagueness must be appropriate in this case. Personal properties would be related by psychophysical law to neural properties. There would be no vagueness about which properties they were related to and, hence, no vagueness about identifying the neural events which were either identical to or correlated with mental events by psychophysical law. So it is not at all obvious that emphasis on the personal perspective alone will provide the kind of considerations Pietroski needs. It strikes me that Pietroski’s lack of attention to this possibility reveals his tacit commitment to a style of Non-reductive Physicalism which preserves a sense in which mental properties are nothing over and above physical properties, hard as this is to specify.

                There is one pretty straightforward reason why mental events are unlikely to be identical to neural events which I would like to get out of the way so we can get to the heart of the issue for Pietroski. Patterns recognised at the personal level need not be captured at the neural level. Neurophysiologists might be largely uninterested in the kind of things that are singled out at the personal level. If neural events are those which are categorised by the neurophysiologist, then mental events need not be neural events. Nevertheless it might be thought that they are constituted from neural events. That’s still enough for Neuralism to be true. Pietroski denies that mental events are even constituted from neural events. So I will focus on this claim.

                One option which Pietroski dismisses is that neural events may constitute mental events even if it is vague which neural events are part of the constituting set. In a similar way, mountains may be constituted from rocks but it might be vague quite which rocks are part of a mountain at the boundary. Pietroski cites two reasons for being unhappy with this option. The first is that Gareth Evans and David Lewis, in his note on Evan’s article, have shown that the suggestion is of dubious coherence. The second is that it could not answer our concerns about mental causation. The first reason is a peculiar one for Pietroski to cite. Evans presents an argument to show that there could not be vague objects (Evans (1978), pp. 176-177; Lewis (1988)). Vagueness, according to this view, would be an artifact of our ignorance or semantic indeterminacy. But if this is right, then Pietroski’s argument from differential vagueness does not work. For all that he has said, mental events are identical with or constituted from neural events. They cannot differ in vagueness of spatiotemporal boundaries because vagueness is not a real feature of the world. It is just our terms for them which are vague. So my first concern is that Pietroski faces a dilemma. He either agrees that vague constitution is an option or he loses the argument from differential vagueness.

                The point about mental causation will receive slightly more sustained treatment when we turn to Pietroski’s own proposal. However, for the moment, we may note the following. Taking mental events to be vaguely constituted from neural events puts them on a par with mountains and their rocklike constituents. It seems to me that there is considerable prima facie plausibility in holding that mountains are efficacious. If so, then the comparison with mountains would show that there was no greater problem with the efficacy of mental events than there was with the efficacy of mountains. This would be some step forward.

                The other option that Pietroski considers is that we might arbitrarily sharpen our mental discourse so that we could identify the neural event which is identical with a mental event or a precisely delineated set of neural events which constitute the mental event. He argues that this  option is not open to us. He also rejects a supervaluationist approach to vagueness in which one could assert that every mental event is a neural event because every sharpening of our concept of a mental event would pick out a precisely specified neural event (p. 177). His reason is that arbitrary precisification in either sense would detract from the goal of personal ascriptions. He writes

Action descriptions, however, purport to be descriptions of a person’s contributions to history. And it cannot be a matter of mere decision which event is Nora’s action of raising her hand. To say otherwise - that we can simply stipulate (from a third person perspective) which neural event is Nora’s trying - is tantamount to saying that no event is Nora’s action.... Our actions are tryings: and if these events are not biochemical, we should aim to understand this fact about ourselves (p. 178).

This seems to contain his main reason for endorsing the argument from differential vagueness, his other reasons discussed a moment ago are mentioned more briefly in passing. Unfortunately, it is just as hard to see the force of this final consideration.The fact that there a number of different arbitrary precisifications we might adopt each of which would be a good candidate for being the mental event does not imply that, in fact, there are no mental events. The existence of a number of different equally good candidates does not imply that there are no candidates. Embarrassments of riches don’t become poverty of existents overnight.

                It seems that Pietroski feels that somehow, if we identify mental events with neural events, we are forced to abandon the personal perspective. But it is hard to see the basis for this. Either the personal is compatible with the impersonal or it is not. If it is, then an arbitrary precisification will still capture the generalisations and patterns we perceive at the personal level. There are just a number of ways in which this might have been done. If it is not compatible, then it doesn’t matter whether things are precise or vague. There is still no way of fitting personal reasons in an impersonal world. Either way, it seems that there is nothing to lose. In this respect, it seems to me that Donald Davidson is far clearer in his spelling out of the commitments of the personal perspective (Davidson (1991), pp. 160-166).


3.             The Resolution of Personal Dualism, the Completeness of Physics and Mind-Brain Interaction

A preliminary sketch of Pietroski’s resolution of Personal Dualism, the completeness of physics and Mind-Brain interaction would run as follows (pp. 109-111). The existence of ceteris paribus laws holding between mental events and bodily behaviour show that mental events are explanatory of bodily behaviour. Causation is the transitive extensionalisation of explanation. Hence the existence of ceteris paribus laws shows that mental events cause behaviour. There are various ways in which the claim that physics is complete might be thought to undermine the reasoning just given but these don’t stand up. I agree with Pietroski in much of his discussion of these latter points (Noordhof (1999a), Noordhof (1999b)). I don’t agree with Pietroski that we should to appeal to ceteris paribus laws in order to defend the efficacy of mental events understood in his sense (Noordhof (1999a), pp. 307-308). However, since he is merely offering a sufficient condition on explanation and thereby causation, our positions are not in direct conflict. So I will not bog the discussion down by wheeling out my preferred account. Instead, I want to challenge Pietroski’s central idea that ceteris paribus laws can do the job. First, let me set his position out in a bit more detail.

                Pietroski claims that causation is a relation between events, explanation a relation between facts (p. 89). Facts are true thoughts. They are abstract entities (pp. 90, 94-95). Facts are individuated by the way subjects think of things (p. 90). Pietroski provides the following sufficient condition on explanation.

F1 explains F2 if F1 is an instance of the fact that an event of type T1 occurred; F2 is an instance of the fact that an event of type T2 occurred; and ceteris paribus if a T1 event occurs, then a T2 event occurs (p. 10)

He characterises ceteris paribus laws as follows.

    ‘cp["x (Fx è $yGy)]’ is non-vacuous if

    (i) ‘F’ and ‘G’ are otherwise nomological; and

    (ii) "x (Fx è either $yGy or there is an interferer INT such that INT explains why ¬$yGy despite Fx) and

    (iii) either some instance of F explains some instance of G or some instance of F in conjunction with an interferer INT explains some fact from which it follows that ¬$yGy (p. 126)

The point of the first clause is to avoid counterexamples involving gerrymandered predicates. The heart of the proposal is the second clause. When things aren’t equal there is an interferer which explains why there is no G. The interferer must be independently explanatory in other contexts. It cannot be tailor-made to explain why an otherwise faltering law does not face a counter instance (p. 125). The third clause just insists that the law does some explanatory work. There must be instances which obey it (p. 127).

                The standard worry about such an account of explanation is that merely appealing to laws of any kind plus matters of particular fact can’t capture the full force of the explanatory relation. In addition, we need to identify something as a cause. Without this, things will be classified as explanations which intuitively are not. As Pietroski acknowledges, this would undermine his approach (pp. 142-143).

                One case, which Pietroski discusses, is the pendulum law.

(PL)         P = 2pÖ (L/g)

where P is the period of the pendulum, L is the length of the string and g a gravitational constant. Intuitively, the length of the string explains the period of the pendulum but not vice versa. Yet the law backs inferences in either direction (pp. 141-142). Pietroski draws a distinction between the law as stated and it being used as what he, following Cummins, calls a transition law. Transition laws relate changes in period to changes in length and vice versa (p. 142; Cummins (1983)). Pietroski notes the transition laws are ceteris paribus. They would not govern changes in length, or in period, if, when the change is made, an additional source of drag is introduced (p. 143). Here, he argues, we find the required explanatory asymmetry. Introducing the source of drag explains why the period does not change in the predicted way as the length changes. However it does not explain why the length does not change in the predicted way when the period is changed (p. 144). So the candidate transition law which relates changes in period to changes in length fails clause (ii) of his account of non-vacuous ceteris paribus laws.

                One concern I have about this suggestion is over how Pietroski cashes out ‘explanation’ in the claims he has just made about the interfering drag. What stops there being a ceteris paribus law relating introductions of drag with failure of the length to be different? We can imagine a scientist legitimately inferring as follows. The period has changed. So one would expect the length of the pendulum string to be different. It is not. So there must be some undetected source of drag which would explain why the pendulum string need not be different in length. If there were a ceteris paribus law of the appropriate sort, then the interfering drag would count as explanatory in the case of the transition law relating changes in period with changes in length.

                Pietroski’s answer will be to identify another interferer which putatively shows how the candidate ceteris paribus law involving the relationship between drag and changes of length fails clause (ii) of his account of non-vacuous ceteris paribus laws. But then the question can arise all over again. To avoid a vicious regress, Pietroski must deny that we should understand the explanatory character of at least one relevant interferer in terms of ceteris paribus laws. He more or less acknowledges this feature of his account on page 146 although not precisely concerning the issue I have raised. He also appears to be committed to claiming that ceteris paribus laws cannot be developed into strict laws otherwise his method of characterising the asymmetry would have limited application.

                Pietroski’s proposal also seems to have difficulties with a case of Segal and Sober’s (Segal and Sober (1991), pp. 3-5). Segal and Sober claim that it is a ceteris paribus law that, if blue-eyed humans successfully mate, their off-spring will be blue-eyed. Yet being blue-eyed is not causally relevant to producing blue-eyed children. It is something to do with genetics.  Pietroski’s fix is to claim that parents with eyes cosmetically altered to the colour blue will provide counterinstances to this law which cannot be explained away by interfering factors (p. 144). Unfortunately, I don’t think this works. First, we could reformulate the ceteris paribus law to mention humans with congenitally blue eyes rather than merely blue eyes. Then it seems we won’t have a counterinstance which cannot be explained away by an interfering factor. I guess Pietroski might insist that he could make a similar manoeuvre to deal with this case. This brings me to the second point. Pietroski relies upon the fact that there may be more than one way in which to cause an event such as a human being blue-eyed. The manoeuvre will not work if there is only one way to cause a particular antecedent of a ceteris paribus law to be realised. Yet, this should not settle that the event mentioned is a cause.

                There is one more problem. The reason why humans with eyes cosmetically altered to blue often fail to give birth to blue-eyed babies is that they have the wrong genetic material. Pietroski needs to explain why this wouldn’t count as an interfering factor. If it could, then Segal and Sober’s original ceteris paribus law would be counted as non-vacuously true by Pietroski’s lights.

                Of course it is possible that Pietroski may be able to add to his account of ceteris paribus laws so that he can deal with these problems. I question whether this is the right way to go. We seem happy to allow ceteris paribus laws without causality between antecedent and consequent. There is good reason for this. We are interested in noting certain patterns either for the purposes of rough prediction or because we want to investigate them further. Segal and Sober’s initial claim that there is a ceteris paribus law holding between blue-eyed parents and blue-eyed children seems good in spite of Pietroski’s point about cosmetic alteration. It is tempting to think that this is another circumstance in which all things aren’t equal. If Pietroski wants to depart from our ordinary usage then that’s fine. But I think it would be better to be honest about this and state that one is interested in patterns of causal relevance rather than appeal to the notion of ceteris paribus laws. Appeal to ceteris paribus laws has the appearance of being respectful of scientific practice while showing how it appeases philosophical concerns. But the appearance is misleading. Philosophical concerns are forcing us to distort scientific practice by providing an account of ceteris paribus laws that has little plausibility outside of its capacity to assuage philosophical concerns.

                In the last chapter of his book, Pietroski suggests that we perceive causal relations. One of the causal relations we perceive is that tryings give rise to bodily movements. He believes that our perception of causal relations allows us to arrive at an explanation by citing a cause without regard to ceteris paribus law. This raises the question (as Pietroski remarks) of why we should appeal to ceteris paribus laws at all in trying to establish that mental events are efficacious (pp. 225-230). His answer is that theorising can lead us to adopt a more restrictive notion of what are legitimate explanatory tools. For instance, we might insist on appealing only to strict laws or impersonally characterised laws. His defence of the legitimacy of ceteris paribus laws, and explanations based upon them, is supposed to answer this concern (pp. 230-232). If my reservations about his approach have foundation, then he has acknowledged the legitimacy of a challenge that his preferred approach is ill suited to answer.

                One diagnosis of the problem is that Pietroski’s approach ignores the vital question of the relationship between the events cited in the antecedent of a ceteris paribus law and the events mentioned in lower level laws governing the micro level. This is revealed in, at least, two of the worries I raised about the Segal and Sober case. I claimed that the fact that there is more than one way in which the parents may be caused to have blue eyes should not be part of the basis for a decision about whether the parents having blue eyes is a cause of their children having blue eyes. More important is the relationship to the genetic factor in the parents. Is the relationship close enough for the parents being blue eyed to be causally relevant? The fact that being blue eyed is some causal distance from the genetic material of the parents having an influence on the genetic make up of the children suggests not. The importance of this matter is even clearer in my last worry, namely that we could view having the wrong kind of genetic material as an interfering factor for the ceteris paribus law between blue-eyed parents and blue-eyed children. The obvious answer is that this must be wrong because the mechanism behind children having blue eyes is genetic. We shouldn’t think of such factors as just interfering with other laws. Unfortunately, Pietroski cannot afford to give this answer without undermining Personal Dualism.  Appealing to considerations about the mechanism involved in causal relations is likely to favour impersonal over personal causation. Nor can Pietroski appeal to the fact that the ceteris paribus law, relating blue eyes in parents to blue eyes in children, is derivative from more exceptionless laws. Presumably laws at the personal level will have more exceptions than those at the neural level and be derivable from them.

                The challenge for Personal Dualists is to explain how the mental bears the right kind of relations to the neural given that the neural does not even vaguely constitute the mental. Pietroski’s emphasis on the global supervenience of the mental on the physical rather than a more local supervenience makes his position awkwardly placed to assuage this kind of anxiety (pp. 179-215).


<1> Things may be a little worse than that. When Pietroski considers the sentences

The rock broke the window

Nora broke the window

Pietroski argues that they should receive the same analysis even though the first does not involve agency. In fact, he also notes that the second need not involve agency. The breaking might have occurred as a result of a deviant causal chain (p. 50). What this suggests is that the logical form does not need the grounds of the action sentence to be an action. Indeed, for a univocal application of the logical form, it is better if there is no action but merely the idea of, as Pietroski puts it, the salient initiator (p. 51).


Robert Cummins (1983), The Nature of Psychological Explanation (Cambridge, Massachusetts, The MIT Press).

Donald Davidson (1970), ‘Mental Events’, in L. Foster and J. W. Swanson (eds.), Experience and Theory, and in his (1980) Essays on Actions and Events (Oxford, Oxford University Press), pp. 207-225.

Donald Davidson (1973), ‘Freedom to Act’, Ted Honderich (ed.), Essays on Freedom of Action (London, Routledge and Kegan Paul), pp. 137-156, and in his (1984) Essays on Actions and Events (Oxford, Oxford University Press), pp. 63-81 [page references in text to latter].

Donald Davidson (1991), ‘Three Varieties of Knowledge’, A. Phillips Griffiths (ed.), A. J. Ayer Memorial Essays (Cambridge, Cambridge University Press), pp. 153-166

Gareth Evans (1978), ‘Can There Be Vague Objects?’, Analysis, 38, 4, p. 208, and in his (1985), Collected Papers (Oxford, Oxford University Press), pp. 176-177 [page references in text to latter].

Jennifer Hornsby (1980), Actions (London, Routledge and Kegan Paul).

Jennifer Hornsby (1981), ‘Which Physical Events are Mental Events’, Proceedings of the Aristotelian Society, 81, pp. 73-92.

David Lewis (1988), ‘Vague Identity: Evans Misunderstood’, Analysis, 48, 128-130.

Alfred R. Mele and Paul K. Moser (1994), ‘Intentional Action’, Noûs, 28, no. 1, pp. 39-68.

Paul Noordhof (1999a), ‘Causation by Content’, Mind and Language, 14, no. 3, September 1999, pp. 291-320.

Paul Noordhof (1999b), ‘The Overdetermination Argument versus the Cause-and-Essence Principle - No Contest’, Mind, April, 108, no. 430, pp. 367-375.

Brian O’Shaughnessy (1973), ‘Trying (as the Mental «Pineal Gland»)’ The Journal of Philosophy, 70, no. 133, pp. 365-386.

Paul M. Pietroski (1998), ‘Actions, Adjuncts, and Agency’, Mind, 107, pp. 73-111.

Paul M. Pietroski (2000), Causing Actions (Oxford, Oxford University Press).

Gabriel Segal and Elliot Sober (1991), ‘The Causal Efficacy of Content’, Philosophical Studies, 63, pp. 1-30.

P. F. Strawson (1958), ‘Persons’, H. Feigl and M. Scriven (eds.), Minnesota Studies in Philosophy of Science, 1 (Minneapolis, University of Minnesota Press).

P. F. Strawson (1959), Individuals (London, Methuen).

top | back to symposium index | Pietroski's reply