WRIGHT NEELY
TRANSITIVE
ACTION,
INTERVENTION,
AND
PREVENTION
(Received 29 October, 1971) As agents we think of ourselves as having power over the world. We believe that we are, on occasion, capable of molding what happens in the world according to our desires and our convictions about how things ought to be. On these occasions we are not passive victims of circumstances but active beings who shape the course o f events in accordance with our wills. Sometimes we are able to constrain the world to remain the same (in certain respects) when we like the way things are. Sometimes we are able to change the state of things to one which we prefer. Although we know that all too often we are powerless in the face of events, we believe that sometimes, at least, we are not. In this paper I shall be trying to shed some light on this picture which we have of ourselves. I shall argue that when we think of ourselves in this way, we have in mind a very pervasive, but by no means exhaustive, class of actions. I shall argue further that this class of actions cannot be adequately understood apart from a certain point of view or conceptual framework. Finally, I shall argue that an understanding o f how this special point of view is involved in our conception of these actions will help us to solve certain puzzles about the idea of prevention.
There are in the philosophical literature several expressions which serve, at least very roughly, to mark off the class of actions with which we shall be concerned. That great logician of the will, Jeremy Bentham, defined a class of actions which he called transitive acts: Acts may be called transitive, when the motion is communicated from the person of the agent to some foreign body: that is, to such a foreign body on which the effects of it are considered as being material; as where a man runs against you, or throws water in your face.I Philosophical Studies 24 (1973) 65-88. All Rights Reserved Copyright © 1973 by D. Reidel Publishing Company, Dordrecht-Holland
66
WRIGHT NEELY
More recently, Professor Max Black has referred to a ~imilar type of actions as cases of making something happen? Perhaps wisely, Black does not attempt, as Bentham did, to define his expression in a precise way; rather he indicates what he has in mind by giving examples: "closing a window, opening a drawer, turning a door knob, sharpening a pencil" (153). Shifting, in a footnote, to the formal mode, he says that he is investigating "a class of transitive verbs, like 'moving', 'breaking', 'opening', 'upsetting', etc., indicated by the blanket expression 'making something happen' ' ', and adds: When the expression 'making something happen' occurs, the reader may usually imagine the more specific expression 'moving a glass' substituted - with the understanding, however, that the discussion is intended to apply indifferently to an entire class of similar expressions. (153)
This reference to transitive verbs brings Black's characterization of 'making something happen' very close to Bentham's definition of 'transitive acts'. However, the latter's distinction between transitive and intransitive acts cannot be drawn on the basis of grammar alone, for "Jones is moving his arm" has, at least superficially, the same grammatical form as "Jones is moving a glass. ''3 And, as his subsequent discussion makes clear, Black himself does not wish to construe raising one's arm (in the ordinary way) as a case of making something happen. Thus, when he goes on (160) to list some seven features which he takes to be true of any 'perfectly clear case' of making something happen, one of them, the third, appears to rule out raising one's arm. Referring to the agent as 'P' and the object upon which he acts as 'O', he writes: "P made this happen by doing something (moved his hand to O, clasped it, and brought it back to him)" (160). Actions which have this featurethat they are performed by performing some other action - have been called by Professor Arthur Danto 'non-basic actions '4, and although it is not true, as Black's article sometimes seems to imply it is, that all actions correctly describable by expressions involving transitive verbs (e.g., 'moving') have this feature, we shall do greater justice to the spirit of Black's account if we include this feature as a defining property of 'making something happen) Summing up, I can indicate the items with which I shall be concerned in the following way: they are items (A1) which are cases of making something happen and are (also) actions, (A2) in which what is made to
T R A N S I T I V E ACTION, I N T E R V E N T I O N , AND P R E V E N T I O N
67
happen is made to happen by doing something (else), and (A3) in which what is made to happen is something outside the agent. 6 I now wish to argue that when we think of ourselves as having power over the world, the power to change the course of the world in accordance with our wills, it is items of the type delineated by these three conditions that we have in mind. It is, I take it, quite obvious that any item which satisfies the three conditions will be a case of an agent's changing the course of the world; hence my argument will be confined to showing that any such case will satisfy the three conditions. If some event happens in the world, it will be a manifestation of my power only if it is ascribable to me. But we ascribe events to inanimate objects, e.g., billiard balls, and we do not think that they share with us the power to effect changes in the world in accordance with their purposes and desires, for we do not think that they have purposes or desires. The event will be a manifestation of the relevant power only if the event is in accordance with a purpose that the agent has and it is produced by the agent to implement his purpose. But surely this is a sufficient condition of its being the case that the effecting of the event is an action on the part of the agent. Hence, condition (A1) - that the agent makes something happen and that his making this happen is an action which he performs - will apply to any manifestation of our power over the world. It may appear to be equally obvious that condition (A3) will apply since if I succeed in effecting a change in the course of the world, my doing so will be a case of making something happen to (or in) something distinct from myself. However, it does not take much reflection to discover apparent counterexamples to the thesis that this condition will apply to any manifestation of an agent's power over the world. Consider haircutting. If I cut somebody's hair, I effect a change in the world. But I may cut my own hair in exactly the same way that I might cut somebody else's hair. And surely if the former is a case of effecting a change in the world, the latter is also. Now one might simply deny this - for example, on the grounds that an agent who could effect changes only in parts of himself would hardly be thought to be less a victim of circumstances than one who could effect no changes whatever. However, I believe that, since there appear to be no philosophically interesting differences between the acts of cutting someone else's hair and cutting one's own, it is more illuminating to
68
W R I G H T NEELY
grant that both haircutting acts arc manifo8tations of our powor o w r the world and to modify the statement of condition (A3) to read: "in which what is made to happen is made to happen outside the agent qua agent of the particular act in question." On this understanding of the condition we no longer have a counterexample, for when I cut my hair, the hair is no more part of me qua haircutter than is your hair when I cut it. To see this, consider winking. To wink is to close an eye. But if the eye that I close is not my eye, I do not wink. The relevant difference between winking and haircutting may be marked in many ways, but it is quite precisely, if rather mysteriously, marked by saying that the eye is part of the agent qua winker whereas the hair is not part of the agent qua haircutter. At this point it will be objected that I have stacked the cards by the way I have specified the actions. Thus, if I had referred to the haircutting acts as, say, self-barbering and other-barbering, and if I had referred to the winking as, say, one-eye-closing, the desired results would not follow. For, it will be said, my hair is part of me qua self-barber, and my eye is not part of me qua eye-closer. But here the answer is that the hair is not part of me qua agent of the act of self-barbering; it is part of me only qua, so to speak, recipient of that action. And 'one-eye-closing' is just ambiguous. For I can close my eye in the way that I might close your eye. But if I close my eye in the winking way, the eye is part of me qua agent of the act; and if I do not, it is only part of me qua recipient of that act. Thus, on the modified understanding of condition (A3), this objection fails. This idea of an item's being part of an agent qua agent of a particular act is, to be sure, rather too obscure to enable much to hinge on it without further clarification. The following conditions, properly understood, will capture the idea fairly well. An item, i, is part of an agent, a, qua agent of a particular action, A, if and only if (P1) A is a basic action performed by a, (/)2) i is part of a, and (P3) i is something a uses in performing A but is not something he performs that action on (or to). Condition (P1) rules out my hair when I lasso you with my super-long pigtail - unless, in accordance with the complication discussed in note 6, above, this is a basic action. Condition (P2) rules out a piece of equipment I might use in performing a basic action in the way explained in that note. And condition (P3) rules out my hair when I cut it. Condition (P3),
TRANSITIVE ACTION, INTERVENTION, AND PREVENTION
69
however, must be understook in such a way as to include my eye (lid) when I wink and my arm when I raise my arm as a basic action; i.e., we must regard my eye (lid) as something I use in winking and my arm as what I use in performing an act of arm-raising. Obviously, more needs to be said, but this must suffice for the present. There is a more general objection to the applicability of condition (A3) which we must now consider. According to it, any action which I perform can be viewed as a manifestation of my power over the world. The argument is this: If I perform any action (possibly excepting some omissions), I effect a change. But, since the world is all that is, any change must occur in the world. Whenever a change occurs in the world, the world changes. Hence, whenever I effect a change, I change the world. It follows that whenever I perform any action (except omissions), I manifest my power over the world. When, for example, I wink or raise my arm - actions which according to the argument of this paper are to be excluded from the class of actions in which I manifest my power over the world - I change the world from being in the state in which my eye is open or my arm is down to being in the state in which my eye is closed or my arm is up. This objection is cleaflfly a powerful one, but it is important to see precisely what it shows - and what it does not. I shall not attempt to show that it is in any way internally incoherent; rather I shall argue that at best it provides an alternative to the way of looking at actions which we actually use and which is embodied in the statement of the picture we have o f ourselves that I sketched in the opening paragraph of this paper. That is, I shall argue that it provides a way in which we could look at actions but not an objection to the way in which we actually do look at them. I shall also show that adopting it at the expense of our ordinary viewpoint would require changes in other concepts that we actually have, notably, the concept of prevention. There are three features which define the point of view of the objection. First, the world is thought of as all that is. Second, all actions are construed as cases of making something happen (i.e., of effecting a change). And finally, changes in the world are thought of as alterations of state. Each of these is essential to the objection: the argument of the objection will fail if we refuse to accept any one of them. I think it is clear that we do not always look at things in this way. For
70
W R I G H T NEELY
example, it is clearly a manifestation of my power over the world when I prevent some event from happening which, apart from my action, was going to occur. We can, (and, at least very often, do) think of this sort of case in the following way: what the agent does is to constrain the world to remain the same in a certain respect, for example to keep the door from opening. Notice that such actions are quite different from omissions: the agent does not merely fail to open the door, does not merely let the door stay shut; he keeps the door from opening, makes the door stay shut. When we think of things in this way, we cannot both think of a change in the world as a change of state and think of the action as one of effecting a change in the world, although we may do either. If we do the latter, then we must construe the effected change not as an alteration o f the state of the world but as a change in its behavior. To be sure, we may give up our ordinary way of looking at these cases and adopt the one recommended by the objection. Then we should think o f what the agent does as, for example, to lean against the door with great force, to place his body against the door and tense his muscles, the nonopening of the door being merely a consequence o f what he actually does. We might think of such cases in this way, but we actually do not. And it is surely as artificial to say that we must so think of them as it is to say that we must think of what we really see as being the immediate impact of light rays on our retinas. Once it is realized that we do not think o f all actions in which we manifest our power over the world in the way embodied in the objection, it becomes clear, I think, that even in cases in which we think of what the agent does as effecting an alteration of the state of the world, we (at least very often) think of the change he effects as being, in addition, a change in the behavior of the world, a deviation from the normal or natural way of the world's development, this being defined by such notions as active and passive powers, laws and state descriptions, or natural ends. (Indeed, as we shall see, there is a very good epistemological explanation of our looking at things in this way as far as possible since it is our knowledge that a particular change is, in addition to being an alteration of the state of the world, a deviation from the normal that enables us to know that an agent has in fact exercised his power over the world. One might say, if he wished, that it is not so much that our concept of an agent's making something happen in the world involves
TRANSITIVE ACTION, INTERVENTION, AND PREVENTION
71
as part of its meaning that a change has occurred which is, in the relevant sense, abnormal, but rather that it is the fact that such changes are deviations from the normal that enables us rationally to apply the concept of an agent's making something happen in the world in a particular case.) Insofar as we do think of things in this way, the world in which the agent effects the change is thought of as having its own nature and, hence, as being, in this sense, independent of his will - independent of him as an agent. (If the behavior were wholly a function of his will, he would have no need to change it.) On this way of conceiving the situation, the agent is thought of as standing over against a world whose behavior he may (on occasion) change by means of his action. The changes - whether or not they are also alterations of the state of the world - are thought of as changes in the behavior of the world. And only those actions which influence the development of the world outside the agent are construed as effecting changes in the world. Further confirmation of the extent to which this point of view is deeply embedded in our ordinary ways of thinking about actions may be obtained by again considering the notion of prevention. For the concept of prevention, like its companion concept of making something happen, has application only to those cases in which an agent affects something which is not part of himself qua agent of his action. There are, to be sure, eases in which I can be said to have prevented myself from raising my arm (for example, by preventing my arm from rising), but this is true only when I do so by doing something other than simply not raising my arm. If all I do is not raise my arm, then I do not prevent myself from raising my arm; I simply do not raise my arm. Thus, we ordinarily view actions in a way incompatible with the one proposed by the objection; and our ordinary way has its own rationale and coherence. This does not, of course, mean that we could not or should not abandbn this way of looking at things in favor of the other one. And I am prepared to allow that for certain purposes it might be philosophically useful - even necessary - to do so. If we retain the ordinary approach, however, the objection fails since this approach involves the denial of one or more of its central theses. In particular, as long as we stick with the ordinary approach, we shall have to think of condition (A3) as applying to all actions in which an agent manifests his power over the world.
72
WRIGHT
NEELY
Let us now suppose that anything which enters into my actions at all enters into them in the way that my eyelid does when I wink. If my argument so far is correct, it follows that in this situation I would not have the kind of power over the world which we are here exploring. But to see this is to see that condition (A2) will apply to any manifestation of my power over the world, for the envisioned situation would be precisely a world in which I never make anything happen by doing something else. I conclude, therefore, that it is a truth of descriptive metaphysics that an item is a manifestation of an agent's power over the world if and only if it meets conditions (A1), (A2), and (A3). II
It is the main thesis of this paper that our thinking about actions of the making-something-happen type involves a special point of view or conceptual framework. I shall now try to spell out more fully what this framework is. In performing an action of this type, the agent acts on something. Hence, in thinking of myself as making something happen, I must think that there is something distinct from myself (distinct from myself, that is, qua agent of this particular action), namely that upon which I think of myself as acting. In order that I should be able to think of myself as acting on this thing, I must have some way of distinguishing between those changes which it undergoes of itself and those which I effect: I cannot think of myself as having made the drawer close if I think that the drawer closed of itself. I can only make this distinction if I think of the thing or things on which I act as having their own characteristic ways of behaving: I must think of the thing or things as forming a kind of system which evolves according to its own laws, a system in the development of which I may intervene or not, as suits my purposes. I cannot credit myself with having made something happen in the system with which I interfere unless I have some idea of how that system would have developed apart from my intervention. Of course, I could be mistaken in a particular case about the natural evolution of the system and, hence, about whether I have indeed made something happen in it, and I may in other cases be unsure whether I have made something happen, as when I wonder whether a sound I heard was something I made happen by turning over in bed. I shall call this point of view, in which I think
TRANSITIVE ACTION, INTERVENTION, AND PREVENTION
73
of myself as standing over against a system with its own characteristic ways of behaving, the intervener framework. It is the framework in which I think of myself as intervening in the development of a system of things independent of myself and as, by my intervention, changing the outcome of that development from what it would have been had the system been left to evolve in its own way. In speaking here of a system evolving according to laws, I do not wish to exaggerate the sophistication or determinateness of the conception which I must have of that upon which I act. The conception which I attribute to myself would have been available to anyone - an Aristotelian, say - and is not one which had to await the rise of the New Science or the Kantian revolution in philosophy. The crucial idea is of things being independent of me and having their own natures; it need not be expressed in terms of laws. Nevertheless, it is convenient to speak in this way, and I shall continue to do so. Furthermore, it should not be thought that we have here an argument, analogous to one of Kant's arguments in the Second Analogy, for the thesis that I must think of any system of things on which I am capable of acting as being deterministic, as evolving according to perfectly definite deterministic laws. But I must have some idea of how it behaves if I am to be able nonarbitrarily to think of myself as changing it, i.e., to have reason to think of myself as changing it. Suppose I come across an object of an unfamiliar sort. I have no idea how it will behave. I poke it, and it goes into state 0. Every time I come across such an object (i.e., one which has the same visual appearance, say), I poke it, and it goes into state 0. Do I not here have very good evidence that my poking one of these objects made it go into state 0 without having any idea how it behaves on its own? No. For I at least know that such objects do not go into state 0 on their own. And I know that you do not have to keep nudging them to get them to stay in state 0 (or, perhaps, that you do). In the very act of coming to discover that I can affect their behavior, I come to discover something of what their nature is. But what if they capriciously and, as far as I can see, unpredictably go into state 0 (and, perhaps, back out again)? Well, is there perhaps available some vague statistical information about their behavior? Does it take them, on the average, a few minutes or hours or years to go into state 0? If so, then I can determine whether my poking them makes it more likely that they
74
WRIGHT
NEELY
will go into state O. Perhaps they always and immediately do when I poke them, but they do only on the average every few minutes (or whatever) if I do not. Or perhaps they do every few minutes when I poke them but only every few years if I do not. The point is that I must have s o m e information about how they behave if left to themselves in order to have r e a s o n to believe that I can affect their behavior. It is not at all necessary that they behave in a perfectly deterministic way, but I must have some knowledge about how they behave if left to themselves if I am to be able with reason to think that I have the power to change them. It should be noted that this conception o f the situation which I have when I think of myself as making something happen involves as part o f its own logic the possibility that I should also be thought of as being part of a more encompassing causal system which includes at least myself and the system in the development of which I intervene. For the very possibility of my intervention carries with it the idea that I am capable of causal interaction with the elements in it and, hence, that - in this sense - I am a part of such an encompassing system. But, of course, I cannot conceive of the situation in both ways at the same time. I have brought out the features of the intervener framework from the point of view of the agent himself, but it also operates from the point o f view of one who is observing the agent. In thinking of another person as making something happen, I must, in the same way and for the same reasons, think of him as in one sense standing outside a system in the development of which he intervenes. But it is also true that, as noted in the first person case, it must be possible to think o f the agent as being part o f a larger causal system which includes himself and that upon which he acts. Here we see an important part - though only a part - of what is involved in saying, as many philosophers have, that genuine agency cannot appear to the external observer; if the observer is to think of there being an agent making something happen, he must 'carve up' the situation into an agent and a system outside of which the agent stands. In this way agency c a n appear to the observer. But, as we have seen, it is always possible for the observer to think of the person as part of the more embracing system, and it may be very natural - even inevitable - for him to do so in certain contexts, for example, if he is a psychologist thinking in terms of the forces acting on the agent as well as the effects which the agent himself has in the total situation, or if he is
TRANSITIVE ACTION, INTERVENTION, AND PREVENTION
75
thinking of himself as an agent capable of intervening in the development of the more embracing system. Of course, there may be other reasons that the observer cannot think of the agent and his actions as part of a more embracing system, but these do not concern us here. As far as anything which we have seen is concerned, if either way of looking at the situation is possible, the other is possible as well; but, again, the observer cannot simultaneously adopt both points of view. Adoption o f the intervener point of view carries with it almost no metaphysical commitments. It does, to be sure, require that the agent be conceived as capable of causal interaction with that on which he acts, but it says nothing about what must be the case in order that such causal interaction be possible. Thus, the agent may be conceived dualistically, as consisting of a conscious mental substance and an extended physical substance, for as Broad 7 and others have pointed out, dualism does not, by itself, rule out the possibility of interaction; it was Descartes's physics which created for him the problem of interaction. I am certainly not concerned here to defend Cartesian dualism - indeed, I know o f no good reason for thinking it to be true - I am only pointing out that the intervener framework itself does not rule it out. Again, there is nothing in the intervener point of view itself which dictates any decision as to what is to count as belonging to the agent and what as belonging to the world on which he acts, whatever sorts of entities we may take to be involved in the total picture. The agent need not be thought of as coextensive with his body or with any proper part thereof nor with any part of his mind. Furthermore, nothing in the intervener framework itself precludes the possibility of a sort of fluctuation in what things are assigned to each side. Thus, in certain contexts and for certain purposes features may be taken to be parts or states or attributes of the agent which in other contexts and for other purposes would be relegated to the status of features of the world on which he acts, and it is important to the ways in which we actually employ the distinction that this be the case. The intervener framework itself entails no restrictions as to how these lines may be drawn; it is a form of experience and not a metaphysical thesis. Finally, it should be noted that the intervener framework itself, though it is involved in our thinking about actions of the making-something-happen type, transcends the concept of action. It is often invoked in contexts which do not involve actions at all. Indeed, it is involved in
76
WRIGHT
NEELY
what is perhaps the most pervasive concept of causation, s There appears, as John Stuart Mill pointed out 9, to be no reason that any lawfully evolving system may not be conceptually 'carved up' into a subsystem and intervening elements. Thus, just as we can think of a mischievous boy's playful interference with the system of a number of billiard balls in motion on a billiard table by occasionally pushing one ball or picking up another and rep!acing it 1~ so we might think of one of the balls as interfering with the regular development of the others. Here we should be applying a truncated version of the full intervener framework, for there is in this case no question of the intervener's deliberating whether to intervene, no purpose it might thereby fulfill, and no agency in the full human sense. But this truncated version of the intervener framework is perfectly dear, and we do employ it very often. This point is, as we shall see, important in understanding the notion of prevention. An understanding of the rule played by the intervener framework in our concept of transitive acts is, I believe, capable of shedding light on a number of philosophical problems. Although it is by no means the whole story, it is necessary for adequate comprehension of the fact that we think of ourselves as both a part of nature and as transcending nature in the sense that we can, if we choose and if we are sufficiently ingenious, change nature, to our purposes, n It is capable of illuminating the fact, often noted by historians and philosophers of history, that sometimes men make events and sometimes they are made by them - and that sometimes it is hard to say which. And it is necessary for an adequate understanding of such concepts as effort, freedom, and power. In this paper, however, I shall examine only one philosophical application of the framework adumbrated above: I shall show that it is capable of resolving certain philosophical puzzles, recently brought to the attention of philosophers by Professor Richard Taylor, relating to the concept of prevention. III
Taylor presents 12 some puzzles relating to the notion of prevention and argues that they can only be solved by seeing that the concept of prevention is dependent on the concept of purpose: "... this notion [of purpose] is ... essential to the description of anything whatever as preventive" (197). "... the idea of prevention is unintelligible apart from the idea of
T R A N S I T I V E ACTION, I N T E R V E N T I O N , AND P R E V E N T I O N
77
purpose or goal" (197). I shall argue that this thesis is false (and, hence, that it does not solve Taylor's puzzles) and that the analysis of making something happen which I have presented enables us to solve the puzzles. In the course of so doing, I shall also try to bring out the various ways in which the concept of purpose is related to the concept of prevention. Taylor raises the question, "Why do we have no corollary to the concept of prevention? Or taking the word 'postvention' to be the verbal complement of 'prevention', we may ask: "Why does 'postvention' not even occur in our language?" (185). He proposes as an analysis of 'preventing': "To have prevented some event is, precisely, to have done something that was, under the circumstances then and thereafter prevailing, both sufficient and necessary for though logically, independent of, the subsequent non-occurrence of that event" (186). After giving some examples of preventive actions and arguing that each of the elements in the proposed analysis is necessary, he offers a parallel definition of 'postventing': "... to have postvented some event is, precisely, to have done something which was, under the conditions then and theretofore prevailing, both sufficient and necessary for, though logically independent of, the antecedent non-occurrence of that event" (189). He then suggests that people often do postvent things in this sense "though for some reason this is never referred to as postvention" (ibid.). By breakfasting with her husband on a certain morning, a wife may well, in the sense defined, have postvented his oversleeping on that morning, and by breakfasting alone, another wife may have postvented her husband's having forgone a trip. (See 189-190.) After considering and rejecting four candidates for answers to the question why the notion of postvention is useless - why the word for it does not even occur in our language - Taylor proceeds to give his own. But his answer is, to me at least, rather obscure. He points out that in the typical examples of preventive actions the agent does what he does as a means of (or for the purpose of) preventing the occurrence of the event which is said to have been prevented but that in the cases of postvention which he presented the agent does not do what he does as a means (or for the purpose) of postventing the event that is postvented. (See 193-194). A preventive action, then, is something m o r e than an action that is antecedently necessary and sufficient for whatever it prevents. ~8 It is easy to supply examples of
78
WRIGHT NEELY
actions which fit that description but which are not preventive, and also easy to supply examples of preventive actions which do not fit it. To say, however, that one's action is postventive is not to say anything more than that it is subsequently necessary and sufficient for the non-occurrence of what it postvents. (195). This is plausible enough, but it does not appear to deal directly with the notions of prevention and postvention at all. Perhaps the easiest way to make the point is this: for Taylor a preventive action is more than an action which prevents some event; it is an action which is done for the purpose of preventing some event. But there are, as Taylor himself notes (see 199), actions which prevent things but which are not in his sense preventive actions. Hence, we are left with the notion o f preventing. The fact that the concept of preventive action, as introduced by Taylor, involves the concept of purpose does not show that the notion o f preventing itself involves the idea of purpose. Furthermore, there is another problem. Taylor's original question was, " W h y do we have no use for the concept of postvention?" He attempted to answer this question by pointing out a contrast between preventive actions and postventive actions. This procedure transforms the original question into a different one: " W h y do we have no use for the concept of postventive action, i.e., action performed for the purpose ofpostventing some event?" But as far as anything we have seen so far is concerned, Taylor does not appear to have answered this new question either. He has said, in effect, " W e just do not do things for the purpose of postventing some event although we do perform actions for the purpose of preventing some event." But we still have the question as to why this is the case. To this question his answer appears to be that the notion of purpose (or of means and ends) is essentially forward looking and, thus, cannot be intelligibly fitted onto the notion of postvention: ... it is plainly untrue that the first woman breakfasted with her husband as a means of getting him up on time, or that the second woman breakfasted alone as a means of getting her husband off on a trip. Even if we suppose these two women to have had such goals or purposes, we cannot represent these actions as suitable or even intelligible means to their accomplishment. (193-I94) Although Taylor is surely right about this, it is not clear on his account why we should not so represent these actions. After all, on his account they were successful (were they not?) in accomplishing these ends. Or, to put it another way, if certain events do in fact postvent others, why
TRANSITIVE ACTION, INTERVENTION, AND PREVENTION
79
should it be unintelligible that we should exploit this fact by producing events of the first type in order that events of the second type should not have occurred just as we exploit, in a similar way, the fact that certain events prevent others? It is not, of course, (nor does Taylor suggest that it is) unintelligible that a person should have, and should be represented as having, the purpose of postventing some event. A dealer in a stud poker game may snap his fingers in order that the next card he deals to his opponent not be the club which the opponent needs to complete his flush. But, since the opponent will - as the dealer knows, if he is an honest man and one who does not believe in a sufficiently dramatic form of telekinesis in fact get the next card, this action amounts to an attempt to postvent a club's having been shuffled into that position in the pack. There is, then, no conceptual absurdity in the idea of someone's having the purpose of postventing some event. And this is true even if there is a conceptual absurdity in the idea of postvention itself: People can and do have absurd, even self-contradictory, purposes: numerous mathematicians tried to square the circle. However, in all cases like the one cited, the action is regarded, by rational men, as superstitious. Thus, the answer to our question is that we have no use for the concept of postventive action because all (almost all?) applications of the concept are superstitious, and we are rational men. The question now becomes, "Why is it not rational to set out to postvent some event?" That is, granted that we can intelligibly represent a man's having as his purpose the postventing of some event, is it intelligible that a man perform an action which in fact postvents some event? Consider first the case of making something to have happened. Should we say that the wife in Taylor's first example makes it to have happened that her husband got up at t - A t by breakfasting with him at t? Clearly not, but why not? We are granted to know, in this example, that her breakfasting with him at t is, under the circumstances, physically but not logically necessary and sufficient for his getting up a t - - A t . (This is equivalent to saying that under the circumstances his getting up at t-At is physically but not logically necessary and sufficient for her breakfasting with him at t.) Why are we not entitled to conclude on the basis of this information that she made it happen that he got up at t - - A t ? The answer is that there is something more involved in saying
80
W R I G H T NEELY
that one event makes another event happen than the fact that we are entitled to infer the occurrence of the one from the occurrence of the other, however difficult it may be to specify what this more is. Taylor might be expected to reply that it is quite easy to specify the additional requirement: it is simply temporal priority. But that we should only say that the first event makes the second happen and not the other way around is, according to Taylor, only a grammatical fact, a point about the usage of words. If we neglect this fact about our ordinary concept of making something happen, then we find, Taylor holds, that there is no further difference between one event's being a necessary and sufficient condition (in the prevailing circumstances) for another and one event's making the other happen. This, however, is false. As William Dray has pointed out 14, it may well be that the fall of a barometer is, in a given set of circumstances, a necessary and sufficient condition of subsequent rain. If this is the case, we may justifiably infer from the fact that it is raining that the barometer fell shortly before and vice versa. But we should not conclude that the fall of the barometer made it happen that it rained even though its fall is an antecedently necessary and sufficient condition, in the circumstances, for the subsequent rain stormlL That this feature of the concepts of making something happen, causality, and explanation is deeply embedded in our thinking is revealed by the differences in the counterfactuals authorized by the several situations. If I make it happen that a drawer doses by pushing it, we may infer that the drawer would not have dosed had I not pushed it. But in the barometer case we should not say that if the barometer pointer had not moved counterclockwise, it would not have rained although we can say that, given the circumstances (including that the pointer was not constrained, etc.), if it had also been the case that the barometer pointer had not moved counterclockwise, it could not have been the case that it subsequently did not rain. The 'c' here is a mark of the episternic character of the corresponding generalization.16 Taylor tells us that the cases of postvention "warrant exactly the same familiar type of counterfactual inference that statements about prevention permit" (190). This is false. In discussing his examples of prevention, Taylor formulates the counterfactuals in the usual way: "... if it is true that a physician has prevented someone's death by a timely operation, then we can say that had he (contrary to fact) not performed the opera-
T R A N S I T I V E ACTION, I N T E R V E N T I O N ,
AND P R E V E N T I O N
81
tion, the man would have soon died" (187). But in the case of the example which we have been discussing, Taylor (correctly) does not formulate the counterfactuals in the usual way. He does not say, "had the woman not breakfasted with her husband, he would have overslept" - something which no one would say - but rather, "had she (contrary to fact) not breakfasted with him, we would be warranted in concluding that he had overslept", and he points out that this "is but another way of saying that his not oversleeping was, under the totality of conditions assumed, sufficient for her breakfasting with him" (190). And this is but another way o f saying that. But since we should say this but should not endorse the usual counterfactual (stated above), we may conclude that the ordinary counterfactual is not but another way of saying that his not oversleeping was necessary and sufficient for her breakfasting with him. This discussion has shown that the cases of prevention (as we actually employ the concept) and postvention (as defined by Taylor) are not parallel, but it has done so in a way which raises further questions, e.g., what else is there in a case in which one event makes another event happen over and above the mutual inferability of the propositions that each occurs? Other than to point out a difference in the counterfactuals authorized by the two cases - in the present state of things, not a very helpful point - our discussion has not provided us with an answer to this question. But rather than attempting to answer this very difficult question here, I shall turn to the task of showing how the intervener analysis enables us to resolve Taylor's puzzle. We can imagine a perfectly rational woman deliberating as follows: If I tell my husband the news, then he will forgo the trip he has planned for tomorrow. If I do not tell him, he will go away tomorrow as planned. Should I tell him or not? I should like him to be here tomorrow, and the trip can wait. So, I shall tell him and thus prevent his going. We cannot imagine a perfectly rational woman deliberating as follows: If I have breakfast with my husband this morning, then he will have forgone the trip he was going to go on yesterday. If I do not, then he will have gone on the trip yesterday. Should I have breakfast with him or not? I should like him to be home today, and the trip can wait (could have waited?). So, I shall have breakfast with him and thus postvent his having gone on the trip. What is the essential difference between the two cases? Well, there are
82
WRIGHT
NEELY
lots of differences, and any one of them might very well account for the absurdity of the second case. But if we remember the role of the intervener framework, we can spot one crucial difference. In the first case, the system in the development of which the woman contemplates intervention (i.e., her husband, etc.) is, if left to itself, going to develop in a certain way: her husband is going to go on his trip. And this is just what makes her preventing his going possible: if he were not going to go anyway, then she could not prevent his going. But in the other case, the situation is not going to develop in the way she does not desire (if she can have breakfast with him at all). The development of the system has a unique temporal direction, and it is this, rather than the supposed unique temporal direction of the concept of purpose, which accounts for the presence and usefulness of the concept of prevention and the absence and uselessness of the concept of postvention. One can only prevent what is going to happen. And, of course, it is no good to reply that if one succeeds in preventing it, then it was not really going to happen since he was going to do what he did. For to say that it was going to happen is to make a reference to the natural course of events in the system in the development of which the agent intervened, not to the system which includes him and the system in the development of which he intervened. (Note that even after the prevention has been pulled off, it is still true to say that the prevented event was going to happen.) This unique temporal direction of the developing system is an ineradicable fact. No analysis of the situation in essentially static terms (with temporal indices) like inferability is adequate to this fact. For even in a forward-looking case it cannot be eliminated. Suppose there is a man who can sometimes make a very odd but easily identifiable sound - a peculiar yodel, say. And suppose that we have good inductive evidence that even when the clouds are looking like rain, when he succeeds in making this sound, it never rains. Should we say in such a case that he is sometimes able to prevent an imminent rainstorm? Rather we should search around, I suspect, for some such explanation as this: there is some factor, as yet unknown, which enables him to produce the sound and which also prevents rain. Suppose that we find such a factor, for example, a 0.01~ concentration of a substance X in the atmosphere, introduced by a nearby factory on certain days, substance X being such that when it is present, it affects the man's vocal cords in
T R A N S I T I V E ACTION, I N T E R V E N T I O N , AND P R E V E N T I O N
83
a way which understandably enables him to make the sound and such that meteorological theory enables us to understand why the presence of the substance prevents rain. Here we should say that, given that he wants to prevent the rain and that he believes that he can do so by yodeling and that he is not prevented from yodeling, if it had also been the case that he did not succeed in yodeling, it could not but have been the case that it rained. But we should not say that if he had not succeeded in yodeling, it would have rained. The static analysis fails here just as it did in the case of making something happen, discussed earlier. We may conclude that insofar as Taylor's accounts of prevention and postvention are parallel, the account he presents of the former is inadequate. And we may conclude that the proper analysis of actions which prevent something from happening is this: given a system of things, S, such that an event, E, is going to happen if S develops in its natural way, to prevent E is to do something constituting an intervention in the natural development of S, such that, it having been done, Ewill not happen. We have seen that the idea of purpose is involved in the idea of preventive action only because it is involved in the notion of action and not because it is involved in the notion of prevention. And we have seen that the explanation of the uselessness of the idea of postventive action lies not in the way in which the notion of purpose is involved but in the way in which the intervener framework is involved. Let us now turn to Taylor's arguments for his general thesis that the idea of prevention involves the idea of purpose. He points out that "many things that are preventive are not actions at all" (196). Thus, we have a threefold classification (though Taylor himself tends to play down the importance of the second member): "preventive actions" (193ff.), "actions which prevent, but which are not as such preventive" (199), and "preventive states and events which are not actions" (196). We have presented an account of the first two. Let us turn to Taylor's examples of the third: ... a man can be prevented from leaving a burning building by smoke, even though no one is making the smoke; a man can be prevented from reaching a mountaintop by a landslide, even though no one starts the landslide; and so on. Examples are easy to multiply... (196).
He goes on to remark: What is still left, however, even in those cases, is the notion of an end or purpose, and I want to insist that this notion is still essential to the description o f anything whatever
84
WRIGHT NEELY
as preventive. Thus, the smoke could not be described as preventing a man from leaving a burning building except on the supposition that it was his purpose, goal, or intention to leave, and the landslide can be considered as preventing a man from reaching a mountaintop only on the supposition that he was trying to reach it. Had the first man, for example, intended to stay in the building all the while, had this been his purpose perhaps in order to test some fireproof suit - then he could not be said to have been prevented by the smoke from leaving. The most one could say is that he would have been prevented, in case he had tried. 17(197) N o t e that there are some examples o f prevention which a p p e a r to be incompatible with the thesis o f the passage - which do not, that is, appear to involve the n o t i o n o f purpose at all - examples which T a y l o r himself had introduced earlier in his discussion. Accordingly, he tries to show that they are not, in fact, counterexamples to his thesis. One such example is " t h a t a m a n is sometimes prevented f r o m dying by the automatic p r o d u c t i o n o f antibodies in his b l o o d " (198). Taylor explains a w a y the force o f this as a counterexample by saying".., this.., statement, I think, makes sense only on the supposition that it is the m a n ' s elementary purpose to remain alive" (198). A p a r t f r o m the problematic status o f an 'elementary purpose' here, this is surely just false, for if the m a n were trying to die - perhaps he exposed himself to the disease deliberately the antibodies would still be correctly said to have prevented his dying. But o f course, this would be a purpose, too. But what if he did not in fact care one w a y or the other - surely an imaginable, n o t to say frequent, case? W o u l d this m e a n that the antibodies did n o t prevent his death? No. W e m a y imagine that a d o c t o r examines a m a n in the hospital. H e determines that the reason the m a n recovered was that the p r o d u c t i o n o f the antibodies in his b l o o d h a d occurred. W o u l d he then have to determine anything a b o u t the m a n ' s purposes before he could correctly say that the p r o d u c t i o n o f the antibodies prevented the patient's death? A n o t h e r o f T a y l o r ' s examples is " t h a t trees are sometimes prevented f r o m blossoming b y a natural contamination in the earth "(198). Taylor's w a y with this example is just to eliminate it: With respect to the tree, I believe the statement makes sense only on the supposition, doubtless erroneous, that it is the tree's purpose to blossom. If, as I believe, this is metaphorical or poetical and not strictly true, then I believe it is also not strictly true that the tree was prevented from blossoming. All that we can say is that it normally would have blossomed, but that conditions in this case were not normal. (ibid.) But this is n o t an a r g u m e n t at all. T a y l o r asserts that to say that blossoming is the tree's purpose is metaphorical is and concludes f r o m this
TRANSITIVE ACTION, INTERVENTION, AND PREVENTION
85
alone that it is metaphorical to say that the tree was prevented from blossoming. But, of course, whether this does follow is just the point at issue. It may be noted in this connection that a tree may be prevented from falling as well as from blossoming. And no one, I think, would wish to say that falling is, even metaphorically, a purpose of a tree. Again, an object, for example another billiard ball, may prevent a given billiard ball from, say, colliding with a third billiard ball whether or not it was anybody's purpose that the collisions should or should not occur. Why, then, is it correct to invoke the notion of prevention in all these cases? As I see it, it is correct as long as we are thinking of the situation in terms of the intervener framework, The man, apart from the antibodies in his blood, was going to die, but, the antibodies intervened, and he did not. The tree, apart from the contamination, was going to blossom, but "conditions in this case were not normal". And so on. It is important to realize that the intervener framework plays a role in cases of prevention over and above the fact of necessary and sufficient conditions. It is this fact which enables us to resolve another of Taylor's puzzles about prevention. He emphasizes (197), quite correctly, that of the vast number of events which do not occur, we consider only a small portion as having been prevented from occurring although in most cases, and if determinism is true, in all, there are necessary and sufficient conditions for their non-occurrence. Our theory enables us to account for this important fact by holding that the concept of prevention goes with the intervener conception, and in any given situation, we cannot consider all of the dements as interveners although we may perhaps be able to consider a n y on e (or any group) of them as intervening. In spite of the fact that it is, as we have seen, oversimplified and, strictly speaking, incorrect, Taylor's account of prevention and purpose is illuminating. As I see it, the truth which underlies his discussion may be summed up in the following points: (1) We seldom have any need to speak of prevention except in cases where a purpose is frustrated (or luckily fulfilled in spite of some obstacles). (2) We can only speak of prevention when we are viewing the situation in the way embodied in the intervener framework. (3) That we do take the intervener point of view and how we take it O.e., which things are regarded as interveners) depends at least very often on our purposes. (This alone, of course, is not sufficient
86
WRIGHT NEELY
to m a k e it very illuminating to say that the notion o f prevention involves the notion o f purpose in any special way: our application o f any concept depends in part o n o u r interests.) (4) The intervener f r a m e w o r k itself m a y have in fact (and, in any case, is illuminatingly regarded as having) arisen naturally out o f purposive contexts in the way in which CoUingw o o d suggested in connection with the closely related concept o f causality. 19 (This is the truth behind Taylor's contention that o u r invocation o f the concept o f prevention in the case o f the tree is metaphorical.) (5) Finally, as we have seen (note 17, above), there m a y be a sense o f 'prevent' according to which a m a n m a y be said to have been prevented f r o m doing something n o t just in that some factor keeps him f r o m developing in the way in which he would normally have developed but in that some factor keeps him (or his world) f r o m developing in the way he intends. This sense o f 'prevent' (if it exists) does presuppose the notion o f purpose. University o f lllinois at Urbana-Champaign
NOTES 1 An Introduction to thePrinciples o f Morals andLegislation, Chapter VII, Section XIIL Bentham defines 'transitive' and 'intransitive' acts as species of what he calls 'positive external acts'. Thus, for him, omissions (negative acts) and mental (internal) acts are neither transitive nor intransitive. Although I believe that mental actions - by no means all of what philosophers call 'mental acts' are actions - may be divided into two classes analogous to transitive and intransitive acts, in this paper I shall confine my attention to what Bentham calls 'external acts'. 2 'Making Something Happen', in Sidney Hook (ed.), Determinism and Freedom, New York 1958, reprinted in Black, Models and Metaphors, Ithaca 1962, pp. 153-169. Page references to Black's article will be from the latter and will be inserted in parentheses in the text. Bentham thought that his distinction could be drawn on the basis of grammar, indeed that the distinction 'took its rise' from 'the latter grammarians'. See lee. cir., footnote. 4 See especially his papers, 'Basic Actions' The American Philosophical Quarterly 2 (1965), 141-148, and 'What We Can Do', The Journal of Philosophy (1963), 443--445. 5 Indeed, Black calls the conjunction of the seven features a 'criterion' of making something happen. Professor David Shwayder uses the expression 'effective acts' to mark off a similar class of actions. See his book, The Stratification of Behaviour, London and New York 1965, pp. 159-164. 6 Although it would not disturb me to discover that any of these conditions is redundant, I shall pause to point out that condition (A2) is not (on the most obvious concep-
T R A N S I T I V E A C T I O N , I N T E R V E N T I O N , AND P R E V E N T I O N
87
tion of what is part of an agent), i.e., that condition (A2) does not follow from the other two conditions. If it be granted that when I perform the action of raising my arm in the ordinary way, my contracting my muscles is not an action that I perform, then I think it must be granted that there is no apriori reason that I cannot perform actions involving things which would not ordinarily be considered parts of me (or of my body) unless the bodily movements involved in such actions were themselves actions that I perform. Furthermore, I believe that investigation reveals that people do perform actions which essentially involve extra-bodily materials but which are not such that the bodily movements which they involve are actions. Thus, a man may learn to operate a piece of equipment (e.g., a prosthesis, a baseball bat, or a musical instrumen0 in such a way that what might in the first instance have been the bodily actions by which he operated the equipment become, with practice and training, no more actions that the man performs than are the muscle contractions when he raises his arm. At the rather crude level of analysis on which we shall be operating in this paper, however, this complication can usually be neglected without doing too much violence to the truth, 7 See his The MindandltsPlaee in Nature, London 1925, Chapter III, especially p. 98. s On this point see H. L. A. Hart and A. M. Honore, Causation in the Law, Oxford 1959, especially pp. 2-30. 9 System of Logic, Book III, Chapter V, Section HI. 10 The example (which he uses for quite different purposes) is Herbert Feigl's. See his 'The "Mental" and the "Physical"' in Minnesota Studies in the Philosophy of Science, Vol. II (ed. by H. Feigl, M. Striven, and G. Maxwell), Minneapolis 1958, p. 78. n Of course, when we take into account that we have purposes - and other facts about ourselves - we may find that we are, as agents, in important respects irreducible to the rest of the natural world and, thus, that we transcend it in a more absolute way. But we also may not. 12 'Prevention, Postvention, and the Will' in Keith Lehrer (ed.), Freedom andDeterminism, New York 1966, pp. 65-85. This paper is reprinted with minor changes as Chapter 13 of Taylor's book, Action and Purpose, Englewood Cliffs, New Jersey, 1966, pp. 185-202. Page references will be to the latter and will be inserted in parentheses in the text. 18 Although the passage appears as I have quoted it in both the article and the book, it surely contains an unintentional oversight on the part of Taylor or his proofreaders. The first sentence should read: " A preventive action, then, is something more than an action that is antecedently necessary and su~cient for the non-occurrence of whatever it prevents". 14 'Taylor and Chisholm on Making Things to Have Happened', Analysis 20 (1959-60), 79-82. 15 One way to put this point is this: I f p may be justifiably inferred from q and vice versa, then we may explain how we know that-p by citing our knowledge that-q as our reason and vice versa. But it does not follow that we may explain the fact that-p by citing that-q as our reason and vice versa. Indeed, in the barometer case, the fact that the barometer falls is not the explanation of the fact that it subsequently rains, and the fact that it subsequently rains is not the explanation of the fact of the prior falling of the barometer. The point is a familiar one in the philosophy of science and is prominently featured in discussions of the asymmetry of prediction and explanation. 16 Cf. W. S. Sellars, 'Cotmterfaetuals, Dispositions, and the Causal Modalities', in
88
W R I G H T NEELY
Minnesota Studies in the Philosophy o f Science, Vol. H (ed. by H. Feigl, M. Striven, and G. Maxwell), Minneapolis 1958, pp. 225-308, especially pp. 240--248. lr These examples show that the truth of (1) If a were to try to bring about e, factor f would prevent him from succeeding does not imply (by itself) the truth of (2) fprevents a from bringing about e although (1) together with (3) a has the purpose (intention, goal) of bringing about e would imply (2). If this is correct, then there is a sense of 'prevent' which is assimilable to our model only on the assumption that it is a conceptual truth that someone's having the bringing about of e as his purpose normally or naturally develops into his setting out to bring about e (in the absence of intervening factors). I believe that this assumption is correct (although time considerations complicate the matter somewhat I may now have as my purpose (intention, goal) to climb the mountain next year without now showing any tendency to climb the mountain), but even ff it is not, the resulting existence of this special sense of 'prevent' (which is applicable only to cases in which what is prevented is an action and which is definable in terms of the other notion of prevention and the notion of purpose) does not affect any of my conclusions. 18 Whether this is correct is not at issue here; we may simply assume that it is correct. 19 R. G. Collingwood, 'On the So-Called Idea of Causation', Proceedings o f the Aristotelian Society 38 (1937-38), 85-108, and An Essay on Metaphysics, Oxford 1940,