CHAPTER III.

PRACTICAL RESPONSIBILITY I: NORMATIVE EXPECTATIONS

In Chapter II, I have identified a basic objection to any responsibility-based account of action, the fundamental problem. In a nutshell, the concept of action appears to be prior to the concept of responsibility in the logical order of things. If so, then an account of action in terms of responsibility is impossible. I have also suggested that the challenge thus posed could be met with a concept of practical task-responsibility. The aim of this and the next two chapters is to develop such a concept. I will claim that a person is practically task-responsible for ϕing just in case it would be reasonable (in a special sense I will explain in Chapter V) to expect of her that she ϕ. Two major conceptual tasks lie ahead. First, the concept of expectation involved must be clarified. Second, the concept of reasonableness must be explained. These are the respective tasks of the present and the next two chapters. As we will see, both tasks are rather delicate. In both cases, we will see that the fundamental problem reappears at various junctures in the natural course of explanation of the concepts. I begin the chapter by clarifying the distinction between normative and predictive (or descriptive) expectations (section 1). Sections 2-4 proceed to discuss the concept of normative expectations, since the concept of practical task-responsibility is characterized exclusively in terms of normative expectations. After some preliminary conceptual remarks in section 2, section 3 discusses the question what fulfills normative expectations. This is a delicate topic as this is the first place where the fundamental problem reappears. Section 4 briefly discusses the distinction between practical and moral expectations. Finally, in section 5 I will show how to neutralize the perspectival character of the notion of normative expectations.

64

65

I should note that the aim of this and the next two chapters is primarily to lay the groundwork for the discussion in Chapter VI. As such, the present considerations will not be dialectical. The aim of the chapters is not to defend the concept of practical taskresponsibility but rather to lay down its meaning. Chapter VI will then use thus developed notion to show that it can do some useful philosophical work. 1. Normative vs. Descriptive (Predictive) Expectations I expect of my mother-in-law that she treat me with respect and yet I expect that she will not. That no contradiction is involved is clear. Two different concepts of expectation are involved: the former expectation is normative, the latter predictive or descriptive.1 Here is Patricia Greenspan’s example. “If someone is known to be unusually lazy, say, or simply to dislike a certain kind of action — cleaning up, for instance — it might not be reason for us to ‘expect’ that person to perform it, in the sense of predicting that he will; but it might still be reasonable to think that the person ought to perform it — to expect it of him, in the sense of holding him to a standard which requires it.”2 There are various ways of drawing the distinction between normative and descriptive expectations. As we saw, Greenspan characterizes the distinction in terms of the notion of prediction, on the one hand, and the notion of holding the agent to a demand, on the other. Wallace ties the notion of normative expectation with various reactive emotions we are inclined to feel when the expectation is frustrated (guilt,
1

The distinction has a long standing in sociology, where normative expectations are taken to define social roles (see e.g. Erving Goffman, Stigma. Notes on the Management of Spoiled Identity [New York: Simon & Schuster, 1963]). It has progressively come to occupy a more important place in philosophical literature. For example, Patricia Greenspan has used the notion of reasonable normative expectations to define freedom (“Behavior Control and Freedom of Action,” Philosophical Review 87, 1978, 225-240, and “Unfreedom and Responsibility,” in (ed.) Ferdinand Schoeman, Responsibility, Character, and the Emotions [Cambridge: Cambridge University Press, 1987], pp. 63-80). A similar distinction (though labeled regularity- and rule-engendered expectation) is at work in Steven Lee’s “Omissions,” Southern Journal of Philosophy 16 (1978), 339-354. R.J. Wallace appeals to the notion of normative expectations in giving a compatibilist theory of moral responsibility (Responsibility and the Moral Sentiments [Cambridge, MA: Harvard University Press, 1994]). However, the distinction between normative and descriptive expectations is not always recognized. Susan Sterrett (unpublished manuscript) shows the limitations of D. Lewis’ account of convention due to his failure to take the distinction into account. 2 P. Greenspan, “Unfreedom and Responsibility,” op. cit., p. 72.

66

resentment).3 The distinction can be sharpened by appealing to the metaphor of direction-of-fit introduced by G.E.M. Anscombe.4 Predictive expectations (that p), like beliefs, have a mind-to-world fit: if it is the case that not-p the fault lies with the expectation. Normative expectations (that p), like intentions, desires, etc., have a world-to-mind fit: if it is the case that not-p, the fault is with the world which ought to be changed accordingly. More precisely, we can say that a person predictively expects that p when (among other things) he is disposed to dismiss the expectation as having been wrong if not-p. A person β expects (in the normative sense) of another person α that p when β is disposed to sanction α’s failure to bring about p. β expects (in a normative sense) of α that p when β is disposed to impose a negative sanction on α if α fails to bring it about that p and a positive sanction if α does bring it about that p.5 Correlatively, a person expects of himself that p when he is disposed to negatively sanction his failure to bring about p and positively sanction his success in bringing about that p. Four points deserve a mention. (i) Sanctions are to be understood very liberally. Negative sanctions in particular ought to include the reactive emotions Wallace speaks about. Being susceptible to feeling guilt, resentment or indignation are all forms of being disposed to sanction oneself or others in case of failure to fulfill the expectation.6 But it includes sanctions of a lesser moral

3

This distinction is not crisp, because, as Wallace recognizes, predictive expectations are also often associated with various kinds of emotions. “For example, my expectation about the start of classes may be suffused with a feeling [of] anxiety that has its roots in my childhood experiences of school; the failure of my TV to go on as expected when I activate the remote control may provoke a fit of rage and frustration. But it is not in general the case that expectations of this sort — that is, beliefs about the future — are presumptively associated with any particular attitude” (R.J. Wallace, Responsibility and the Moral Sentiments, op. cit., pp. 20-21). 4 I discuss the distinction in Chapter I, p. 19. 5 Section 3 clarifies what is meant by ‘α brings it about that p’ and ‘α fails to bring it about that p’. 6 Wallace discusses cases of irrational guilt, where one feels guilty without believing that one has frustrated any expectations one accepts. In explaining how this is possible Wallace suggests that we must distinguish between the ends that one values and the ends one is motivated to pursue. In our terms, the distinction is

67

magnitude. Feeling dissatisfied or disappointed by oneself or by another, criticizing oneself or others, etc. are all forms of negative sanctions. But there are also positive sanctions. Various forms of reward or feelings of satisfaction or accomplishment are forms of positive sanction. (ii) Robert Brandom7 argues extensively that to understand normative attitudes in terms of sanctions, one must not attempt to reduce normative attitudes to people’s (or communities’) behavioral dispositions to sanction. Rather, any understanding of normative attitudes must appeal to an already normative notion of sanction. Indeed, it must be the case not only that a person does or tends to sanction non-conforming behavior but that the person ought to sanction it. The above characterization of what it means for one person to expect something of another does not attempt a reduction of the normative attitude of expectation to a mere disposition. When β expects of α that p, β is required to be disposed to negatively sanction α in very specific circumstances, viz. when α fails to bring it about that p.8 In other words, β is required to be correctly disposed to negatively sanction α. Likewise, β is required to be appropriately disposed to place a positive sanction on α, when α does bring it about that p. (iii) It may be worthwhile pointing out that it is not uncharacteristic for philosophers writing on responsibility to focus on negative sanctions. While the availability of the negative side is crucial for an account of action, for it will ultimately allow us to capture negative actions, it is also crucial that the positive side not be left out, for if it were we could not account for positive actions. If there were a reason in principle why the concept of responsibility had to be geared toward the negative side this would constitute a

one between expecting something of oneself and believing that such an expectation is reasonable. Usually these two attitudes go hand in hand, but it is possible for one to expect of oneself what one believes not to be reasonable, in which case one feels guilty (because one is disposed to sanction oneself) but irrationally or unreasonably because one believes that the expectation is unreasonable. 7 Making It Explicit [Cambridge: Harvard University Press, 1994], pp. 34-46. 8 The characterization would not be immune to a charge if the condition of negative sanction were “if β believes that α fails to bring it about that p.”

68

prima facie reason for finding responsibility-based approaches to action suspect. Wallace comments on just this point:
It is striking… that the responses of blame and sanction are negative and punitive in character. Of course, there are positive responses to which holding people responsible occasionally disposes us as well: we praise people, for instance, who are outstandingly good and virtuous. But praise does not seem to have the central, defining role that blame and moral sanction occupy in our practice of assigning moral responsibility. [continued in the footnote:] This is not to rule out the very possibility of a system of social reactions organized primarily around the positive responses of praise and reward rather than blame and sanction; such a system might even be superior to our present practice, in some respects.9

Two points are clear from this passage. Wallace rightly or wrongly takes the focus on the negative side to be, first, a characteristic of moral responsibility and, second, of our practices of holding people to be morally responsible. This suggests that there should be no problem in extending his characterization of the basic concept of holding someone (practically) responsible to cover the positive cases. One may be tempted to speculate that the fact that philosophers of responsibility tend to focus on negative cases has a not accidental correlate in the fact that philosophers of action tend to focus on positive cases. One could imagine that the concepts of action and responsibility could be in a kind of equilibrium: the concept of action covering cases of positive as well as of negative actions, and the concept of responsibility (primarily in the sense of answerability) correspondingly attaching to them in the right circumstances. As it happens, the concept of action is focused on the positive cases, while the concept of responsibility tends to be focused on the negative cases. In either case, the focus does not appear to have any solid justification. In what follows, I will be trying to treat both concepts as having an equal bearing on both sides: the positive and the negative. (iv) Another point about the characterization of normative expectations deserves a mention. We have seen in Chapter I how M. Smith used the idea of direction-of-fit to define desires. It will be instructive to consider the difference between these characterizations. Smith understands a desire that p as “that state of a subject that grounds all sorts of his dispositions: like the disposition to ϕ in conditions C, the
9

R.J. Wallace, Responsibility and the Moral Sentiments, op. cit., p. 61.

69

disposition to ϕ in conditions C′, and so on (where, in order for conditions C and C′ to obtain, the subject must have, inter alia, certain beliefs).”10 Thus understood desires are conceived to be intrinsically motivating — they dispose the agent to the desired action. By contrast, expectations are not seen to be intrinsically motivating; they dispose the agent to adopt sanctioning attitudes whether to oneself or to others. That this is not incompatible with normative expectations playing a motivating role will become clear in Chapter VII. But this role is not built in, as it should not be, into the very concept of an expectation. 2. Normative Expectations In what follows, I will be concerned exclusively with normative expectations. The term ‘expect’ is henceforth reserved for normative expectations unless it is explicitly noted otherwise. I will assume that it is possible to formulate all normative expectations in the following canonical form: β expects of α that α ϕ, or: β expects of α that α bring it about that p,

where β is the expector, α is the agent, ‘ϕ’ is an action-verb, ‘p’ is a sentence. Most normative expectations lend themselves to this canonization very well. Thus: Jane expects of Jim that he move his car so that she may drive out of the garage. The teacher expects of his student that she take part in the school play. Jennifer expects of herself that she become another Maria Callas. Other normative expectations may not be explicitly stated in this form, but they can be easily recast. In the simplest case, ‘Sam expects Mary to be here in five minutes’ can be reformulated as ‘Sam expects of Mary that she come here in five minutes’. Likewise, ‘The admiralty expects of the captain that the ship arrive safely’ can be reformulated as ‘The admiralty expects of the captain that he make sure that the ship arrives safely’.

10

Michael Smith, “The Humean Theory of Motivation,” Mind 96 (1987), p. 52.

70

We need to say a little bit about the actors involved. Then, we will consider the that-clause in some detail, for it is there that the fundamental problem of responsibilitybased approaches, discussed in Chapter II, resurfaces. Three preliminary points first. (i) I will primarily speak of individual people as holding each other to expectations. However, there is no barrier to thinking that other social agents can hold each other to expectations. One state may expect of another state that it not be invaded. A firm may expect loyalty of its employees. A group of people may expect another group to play fair. And so on. (ii) It is also possible to expect something of oneself, in fact many expectations are reflexive or self-directed. The concept of a reflexive expectation is indeed a very close kin to the concept of intention. Usually, when I have a prior intention to do something I expect of myself that I do it.11 (iii) The that-clause in the expectation specifies the description under which the act is expected of α by β. Its occurrence is thus not transparent. What is expected is never a concrete particular performance but rather a type. So, it does not follow that if Jane expects Jim to greet her friend by waving hello, and if his waving hello happens to be identical to his voting for a challenger, then she expects him to vote for the challenger. 3. Fulfilling Normative Expectations: Actions and Performances Normative expectations can be fulfilled, frustrated, or neither fulfilled nor frustrated by events. Suppose that Mary expects of John that he bring her his homework by 5pm. The event of John handing over the homework to Mary at 3pm fulfills her expectation. So would the event of John’s sending his homework by mail if it gets there by 5pm.12 However, if John sits in the library at 5pm (perhaps intending to bring the

11 12

This is not meant to imply that the concepts are identical. One might read the original expectation in a more strict way. One might insist that the only way for the expectation to be fulfilled would be if John physically brings the homework to Mary. I shall not worry about this issue at all. I doubt that there is a way of settling the question by looking at the language used. At the same time, nothing in our account will depend on the issue being resolved in one way or another. It will suffice for my purposes to assume that the content of the expectation is settled by its fulfillment and frustration conditions.

71

homework at 6pm), Mary’s expectation would be frustrated. Finally, there are events that will neither fulfill nor frustrate the expectation, e.g.: Vesuvius exploding a century ago, the Congress passing a bill at 5pm that day, John procrastinating on the steps of the library at 4pm, and so on. We need to consider the reappearance of the fundamental problem alluded to at the beginning. Here is the problem in a nutshell. The that-clause appears to be an agentive statement. If so, then in order to make sense of normative expectations, we would have to have a firm grasp on the notion of action. But this seems to render the very project at hand circular, for I have recommended that the concept of action is to be illuminated by the concept of normative expectation. Once again, the order of the concepts of action and normative expectation appears to be the reverse of that needed by the project. The first step toward a solution consists in noting that the occurrence of ‘α ϕ’ in the that-clause does not yet prejudge the issue. What matters is how we think about the performances that fulfill and frustrate the expectations. There are two ways of thinking about the fulfillment and frustration conditions. Either, one may think that expectations are fulfilled (frustrated) only by actions — we may say that the expectations are agentively fulfilled (frustrated). Or, one may think that expectations are fulfilled (frustrated) by a more liberal class of performances which includes not only actions but mere happenings. In this case, we may say that the expectations are only prima facie fulfilled (frustrated). Consider an example. A guest at a party becomes annoyed by the hostess’s bragging about her authentic Persian rug not just a little too much. He thinks to himself that she would probably be annoyed if something damaged it and immediately thinks of his cup of tea. This is how he comes to expect himself to knock the cup of tea when he reaches forward for some sugar. Just as he is about to do that a muscle spasm shakes his arm thus making it bend in such a way that he knocks the cup of tea from the table where it spills over the bragged about rug. The guest’s expectation of himself is prima facie fulfilled, but it is not agentively fulfilled. Given our task to construct an account of action in terms of fulfillment and frustration of normative expectations, the project would indeed be circular if we took

72

normative expectations to be fulfilled only by actions (i.e. to be fulfilled agentively). We are accordingly committed to taking expectations to be prima facie fulfilled (frustrated). In other words, normative expectations must be construed as being fulfilled by performances (which includes actions and mere happenings). The distinction of agentive fulfillment and frustration conditions will then be made in terms of the standard of reasonableness (in a special sense to be explained in Chapters IV and V).

Mere Happenings Actions

Performances

Figure 1. The relation between the class of performances, actions and mere happenings.

I should note a certain delicacy in trying to characterize the category of performances. I do not want to offer any theoretical characterization of what a performance is beyond saying that it includes the category of actions and of mere happenings. This will mean that bodily movements count as performances. However, saying anything beyond that is controversial for it hinges on highly controversial questions in the ontology of action,13 which I will not address in the dissertation. I will try to circumvent the issue by focussing the category of performances on bodily
Some of the important voices in the debate include: G.E.M. Anscombe, Intention (Ithaca: Cornell University Press, 1957); “Under a Description,” Nous 13 (1979), 219-233; Annette C. Baier, “Ways and Means,” Canadian Journal of Philosophy 1 (1972), 275-293; Donald Davidson, “Agency,” in Essays on Actions and Events (Oxford: Clarendon Press, 1980), pp. 43-61; Lawrence H. Davis, Theory of Action (Englewood Cliffs: Prentice-Hall, 1979); Carl Ginet, On Action (Cambridge: Cambridge University Press, 1990); Alvin I. Goldman, A Theory of Human Action (Englewood Cliffs, NJ: Prentice-Hall, 1970); Jennifer Hornsby, Actions (London: Routledge & Kegan Paul, 1980).; Hugh McCann, “Volition and Basic Action,”
13

73

movements. Most theorists struggling with the question of the proper way of carving up ontological space for actions agree at least on the fact that some bodily movements (raising an arm, reaching out, walking) count as actions.14 I should point out, however, that this focus is contrary to the spirit of the account here developed. For there is nothing in the account that would dictate the thought that actions must be thought of paradigmatically in terms of bodily movements. Even if this were true for individual actions (which I do not believe), it would hardly be so for collective, institutional, or more generally, social actions. However, it is less clear that we can give as intuitively appealing a characterization of the category of performances for those kinds of actions without getting involved in the very complex issues surrounding the proper way of constructing the ontology for a theory of action. I shall therefore simplify the account here by restricting the category of performances to the category of bodily movements. 4. Moral vs. Practical Normative Expectations One of the tasks of a responsibility-based account of action is to make sure that the concept of responsibility is broader than the more familiar concepts of moral or legal responsibility. The concept of practical responsibility can be delineated by means of the concept of practical normative expectations. It may be helpful to follow R.J. Wallace’s attempt to develop the concept of moral responsibility in terms of moral normative expectations. Wallace finds the concept of responsibility based on normative expectations too inclusive for his purposes. He consequently restricts the expectations relevant to his task to ones that have a specifically moral justification, i.e. to moral expectations. Since our task is to offer a responsibilitybased account of action not of moral action, we ought to use a more inclusive concept of (practical) responsibility. It is natural for us simply not to restrict the class of normative expectations to just those that have a moral justification and instead to include all

Philosophical Review 83 (1974), 451-473; Judith Jarvis Thomson, “The Time of a Killing,” Journal of Philosophy 68 (1971), 115-132. 14 This is a somewhat deceitful depiction because there is a considerable difference in the way in which bodily movements are conceived. Jennifer Hornsby (Actions, op. cit.) in particular offers a very esoteric interpretation of the bodily movements that are actions.

74

normative expectations. They will have to be subjected to normative appraisal (to exclude arbitrary expectations, for instance) but such normative appraisal will show them to be inappropriate or appropriate in particular situations, it will not show them not to be practical. In what follows, any normative expectation is considered to be a practical normative expectation with one exception. Those expectations that have either an empty fulfillment or an empty frustration set are not considered practical. Any expectation of the form ‘β expects of α that α bring it about that p’, where p is either logically or physically non-contingent, is not practical. If one were to expect of someone that 2+2=4, such an expectation could not be frustrated; similarly, if one were to expect of someone that 2+2=5, such an expectation could not be fulfilled. The intuitive reason behind the refusal to classify such expectations as practical should be clear: there is nothing anyone can do to make it the case that what is expected is the case or is not the case. Expectations with a non-empty fulfillment and a non-empty frustration set are thus considered to be practical expectations. This is to say something about their content, rather than about their propriety. Practical expectations may be appropriate (reasonable) or inappropriate (unreasonable). In the next two chapters, we will see the complex conditions that are responsible for expectations being reasonable in various situations. 5. ‘It is (would be) reasonable to expect of α that α ϕ’ So far we have spoken of one person expecting something of another person. For reasons that will become clearer, we need to introduce another concept that does not explicitly mention the person who holds another to the expectation. Since it is not clear that there is a settled intuitive meaning of the phrase “it would be reasonable to expect something of a person,” I will distinguish two readings and use one of them consistently throughout the dissertation. I should emphasize that my aim in this section is to understand what it means to say that it would be reasonable to expect something of α, not to understand what the reasonableness of an expectation consists in. For the rest of this section, I shall assume that the notion expressed by ‘β’s expectation of α that α ϕ is reasonable’ is clear.

75

It may be helpful to begin with a slightly simpler notion. Sometimes, we may say that it would be reasonable for a person to expect something of another. For example, it may be reasonable for me to expect of my husband that he does dishes from time to time. The first point to note is that to say so is not yet to say that I in fact do expect him to do the dishes. The claim that is made is rather conditional: if I were to expect this of him, my expectation would be reasonable. Thus, more generally, ‘it would be reasonable for β to expect of α that α ϕ’ means: ‘if β were to expect of α that α ϕ, such an expectation of β’s would be reasonable’. Occasionally, we might want to say not only that it would be reasonable for β to expect something of α but that it is reasonable for β to expect it of α. In some cases, to say that it is reasonable for a person to expect something of another presupposes that that person does hold the other to the expectation. The claim then merely assesses the expectation as reasonable. Thus, knowing that a coach expects of his athletes that they abstain from drinking, we may judge that it is reasonable for him to expect it of them. In other cases, one may say that it is reasonable for a person to expect something of another rather than saying that it would be reasonable, in order to emphasize that one believes the person ought to hold the other to the expectation. So, one might say that it not only would be reasonable for me to expect of my husband to help out with the dishes, but that it is reasonable for me to hold him to the expectation, i.e. that I ought to expect it of him. I will use the phrase thus: ‘it is reasonable for β to expect of α that α ϕ’ means: ‘if β were to expect of α that α ϕ, such an expectation of β’s would be reasonable, and β either holds or ought to hold α to the expectation that α ϕ’.15

15

For our purposes, the difference between the claim that it is reasonable to expect something of someone and the claim that it would be reasonable to expect it of her is negligible. It will have no bearing on any substantive commitments.

76

Given that we understand what it means to say that it would be reasonable for a person to expect something of another, we can ask what it means to say that it would be reasonable to expect something of another. There are two interpretations one could give. On one reading, to say that it would be reasonable to expect of α that α ϕ is to say that it would be reasonable for everybody to expect of α that α ϕ. (In other words, for any person ξ, were ξ to expect of α that α ϕ, ξ’s expectation of α that α ϕ would be reasonable). This is probably what we mean when we say that it is reasonable to expect of a person that she not kill another. Such an expectation would be reasonable no matter who expected this of the person. We might say that it would be universally reasonable to expect of α that α ϕ. But there is a weaker reading, according to which to say that it would be reasonable to expect of α that α ϕ is to say that it would be reasonable for someone to expect of α that α ϕ (or: for some person ξ, were ξ to expect of α that α ϕ, ξ’s expectation of α that α ϕ would be reasonable). To see this as a plausible interpretation, imagine that α has a certain position in a hierarchical organization. Let us suppose that α is a computer programmer and it is part of his job to produce a certain amount of code within a specified amount of time. When we say that it is part of his job (which he accepted of his own will, etc.), we believe that ceteris paribus it is reasonable to expect of him, among other things, to produce this amount of code in the specified amount of time. To believe that it would be reasonable to expect this of him is now no longer to believe that it would be reasonable for everybody to expect it of him. Rather, it is to believe that it would be reasonable for some person (e.g. his supervisor, his firm, his coworkers) to expect of him that he produce the code. Henceforth: ‘it would be reasonable to expect of α that α ϕ’ means ‘For some person ξ, were ξ to expect of α that α ϕ, ξ’s expectation of α that α ϕ would be reasonable’ while

77

‘it would be universally reasonable to expect of α that α ϕ’ means ‘For every person ξ, were ξ to expect of α that α ϕ, ξ’s expectation of α that α ϕ would be reasonable’. The concept of it being reasonable to expect something of an agent will be of central importance for the account of action I shall offer. It has the advantage that it is applicable even in cases where no-one actually holds the agent to the expectation. Let us note that under the second reading, the concept of it being reasonable to expect something of an agent leaves the possibility of conflicting expectations open. It may be reasonable to expect of α that α ϕ (because it is reasonable for β to expect of α that α ϕ) but it may also be reasonable to expect of α that α not-ϕ (because it is reasonable for γ to expect of α that α not-ϕ). It may be that one of the claims (β’s or γ’s) is actually stronger, but it may also be that there is no way of deciding on their strength. Such a possibility ought not to be excluded by fiat. It would be decided by fiat if the phrase were to be used in its universal sense. ••• Before going on to discuss the difficult topic of what makes expectations reasonable, it may be worthwhile assembling all the preliminary ingredients into an account of practical task-responsibility, and summarizing how some of the objections discussed in Chapter II are met. α is practically (task-)responsible for ϕing if and only if it would be reasonable to hold α to the practical normative expectation that α ϕ. Three points ought to be emphasized. First, the concept at stake is one of practical (rather than legal or moral) responsibility. As explained in section 4, all non-empty expectations are considered to be practical. Second, it is a forward-looking, not a backward-looking, concept of responsibility. To expect something of a person is to hold her responsible (in a forward-looking sense) for the carrying out of a certain task, i.e. to hold her task-responsible. These two features are important in allaying the fundamental problem. We will remember that the doubts arise in view of the fact that it is natural to think that responsibility ascriptions presuppose the knowledge whether an action has been

78

performed. It thus seems impossible to try to construe a concept of action in terms of the (apparently later, in the logical order of things) concept of responsibility. The developed concept of practical task-responsibility resolves the problem in two ways. First, the concept of practical responsibility is broader than the concepts of moral or legal responsibility. It is quite intelligible to claim that the concept of action logically precedes the concepts of moral and legal responsibility, while it depends on the concept of practical responsibility. Second, the concept of practical task-responsibility, unlike the concepts of moral or legal responsibility that give rise to the objection, is a forwardlooking concept. We seem to be compelled to think that backward-looking concepts of responsibility presuppose the concept of action, for it is most natural for us to think that we are morally and legally responsible for our actions. It is not equally compelling to think that the forward-looking concept of task-responsibility presupposes the concept of action. As we saw (section 3), a case could be made that it does after all. One could argue that what fulfills normative expectations are actions, in which case the fundamental problem reappears. I have, however, argued that we can also understand normative expectations as being fulfilled by performances (comprising actions as well as nonactions, mere happenings). We will see in the next chapters that more conceptual work will need to be done before the fundamental problem is held at bay. The third, and final, point about the characterization of practical task-responsibility is that it involves an appeal to the standard of reasonableness. This is intended to eliminate arbitrary or otherwise inappropriate expectations of a person as counting toward her being practically responsible for something. Chapters IV-V will be devoted to clarifying what it means to say that normative expectations are reasonable.