Responsibility and Autonomy.
Some Remarks on Susan Wolf’s Freedom Within Reason.
It is morning. I wake up because my child screams from her bed. I eat breakfast while listening to my deranged neighbour smashing his helmet-covered head against the wall and his mother trying to stop him. My cat jumps from the window sill, scared by the refrigerator compressor starting to run, exhausting warm air.
There is something my child, my deranged neighbour, his mother, my cat, my refrigerator, and myself have in common - we do something, we behave. This is because our movements are internally produced, differentiating my behaviour from what happens to me, or my cat’s behaviour from what happens to him. The screaming of my child reaches my ear (something that happens to me) and I wake up (something I do). The sound of the refregirator compressor starting to run scares my cat (something that happens to him) and he jumps (something he does). There is a difference in the cause of these movements - an internal cause vs. an external one, i.e. the difference between a cause (C) internal to the system (S) producing a (bodily) movement (M), and a cause (C) external to the system (S) causes something (M) to happen to the system (S).
But in both cases above the doings or behaviour of my cat and myself can be seen as triggered or caused by what happened to us. Is then our behaviour not our behaviour, but rather an arbitrary stop in a universal chain of causes? When Dretske is confronted with this he takes a certain pragmatic stance. He says that as «long as the outcome can be classified as the effect of internal causes, the production of this effect can be classified as behavior», but this does not imply that such outcomes must be classified as behaviour, only that it can be. This is to stop speaking of the result (M) that was caused in a certain way and to start speaking of the process (C®M) that brought it about.
If we take this pragmatic stance, it follows that all agents (causes of M) are responsible in some weak or original sense for their behaviour. My cat is responsible for his jumping, because without him there would be no jumping, the same way I am responsible for waking up.
There is, however, another stronger sense of responsibility that is not directly connected to the behaviour of the agent, a sense of responsibility that is implied in the following sentence: “As a deranged person, my neighbour is not responsible for his actions, he is not a responsible person”. So what interests us here is not primarily something about the performance of the agent, but something about the agent as a person. The agent is said not to have had available those personal resources that are necessary to qualify him as accountable for his conduct. It is this that distinguishes me from my fellow agents in the little vignette above. According to this picture, the question of responsibility can be answered by determining the ability of the behaving agent. This suggests that the ability concerned is an ability to reflect on one’s own behaviour, to choose what desire to act on according to an evaluation of it. We see freedom thus as being able to choose and act on what desire we value most. We do not judge an agent responsible if he lacks the ability to value, or lacks the ability to value good things, or is constrained in any way by external or internal forces so that he cannot act on what is valued good, and also he values good; but we do judge an agent responsible if he has the ability to value good things but only does bad things.
So, naturally, we think of an agent as responsible, in part, for what he does, and if he is an evaluator (has an ability to evaluate or has values), we think of him as responsible in part for the degree to which he acts in line with his evaluations. But we are also inclined to think of him as responsible in some sense for these evaluations themselves, or rather for what kind of self he is.
This triple conception of responsibility and ability, i propose, gives rise to a triple conception of autonomy. First, the ability to cause movement (ability1), being an agent (in the wide sense), and thereby being responsible for internally caused movement and direct effect upon environment (responsibility1), gives rise to a certain freedom from the nomic causal chain (autonomy1). Autonomy1 is basic in that the criteria for having it is that the movement is internally produced, and accordingly, all agents (in the wide sense) are included in this class. Secondly, the reflective ability to choose and act according to desires based on evaluation of these desires, distinguishing a better one from the one that presses most strongly (ability2), and thereby being responsible for choosing and acting in line with these evaluations (responsibility2), gives rise to a reflective freedom from own urgent desires (autonomy2). This class includes only those agents who can be evaluators. Thirdly, the ability to choose values, to choose oneself (ability3), and thus the responsibility for oneself (responsibility3), gives rise to a freedom from heredity, environment, and one’s personal history, a freedom to change oneself (autonomy3).
There will be no more talk about the first kind of ability, autonomy and responsibility. I leave the reader to study these arguments for himself in Dretske’s brilliant book Explaining Behavior: Reasons in a World of Causes. Accordingly it is the other two triads this paper will be concerned with, in connection with a recent book on responsibility by Susan Wolf.
Let us first reconnoitre this difficult area of philosophy with its many pitfalls, and point to some. One can distinguish between those who think autonomy (or freedom) is compatible with determinism (optimists or compatibilists) and those who think it is not (pessimists, incompatibilists, or libertarianists). I will not in this paper come down on either of these sides. Wolf is a compatibilist, but is special in the sense that while she holds that responsibility and ability is compatible with determinism, autonomy is not.
Wolf finds herself as a mediator between two stances and two views on responsibility. First, she mediates between the metaphysical stance and the pragmatic stance, between the assumption that there is a fact of the matter as to whether we are free and responsible beings, a fact we might discover and about which we might be wrong; and the assumption that whatever facts are relevant to the justification of the activities connected to responsibility are relevant only because we choose to make them relevant, because we set up the rules that assign these facts a certain weight. She concludes by embracing both stands partly; the pragmatic stance because «the rules governing the complex of practices that are conceptually held together by their connection to the status of responsibility [. . .], have established meanings, however obscure, and it is with these meanings that these practices and their components play the deep and pervasive role in our lives that they do»; and the metaphysical stance «because the concepts of free will and responsibility that are already firmly established are intrinsically metaphysical concepts. That is, it is an essential criterion for the correct application of these concepts that they be subject to the demands of justification by the facts».
It is on this background that Wolf occupies her second role as mediator, a mediator between two views from the metaphysical stance, between the compatibilistic Real Self View, and the incompatibilistic Autonomy View. Proponents of the Real Self View, following her, think that the freedom and power to govern one’s actions according to one’s real self are all the freedom and power it is intelligble or at any rate necessary for a person to hope for. Proponents of the Autonomy View claim that this is not sufficient. To be responsible, they say, an agent must not only be able to govern his actions by his real self, he must also be able to ensure that his real self is not in turn governed by anything else.
«Proponents of autonomy criticize their opponents for being unable to explain why being able to govern one’s actions by one’s real self should make a person responsible given that the person’s real self may itself be the inevitable product of external forces. Proponents of the Real Self View question in return how it can make any difference whether the real self is an inevitable product of external forces or instead an arbitrary existent emerging inexplicably from the void.»
Wolf’s own view, the Reason View, claims that both the criticisms these other views respectively level against each other are correct, and their search for a non-value-laden feature to distinguish responsible agents from non-responsible one’s fails, just because they do not acknowledge the value-laden feature of being able to be in touch with the True and the Good, or Reason.
«The freedom and power necessary for responsibility, then, are the freedom and power to be good, that is, the freedom and power to do the right thing for the right reasons. [. . .] Moreover, one may have this sort of freedom and power without metaphysical independence: One’s ability to be good may arise from one’s experiences; it need not exist despite them.»
It will appear through this paper, I hope, that I have strong sympathies for Wolf’s focus on a value-laden feature as the delimiting concept of responsibility, at least responsibility2 and responsibility3. What I do not support is the way inwhich she rejection autonomy, and her account of especially the Real Self View, a view, or at least some version of it, in which I put my thrust.
One can see my distinction between the second and third triad above as corresponding to, in some sense, Wolf’s distinction between the Real Self View and the Autonomy View. What I will try to do in this paper is to show Wolf’s rejection and thereby conception of autonomy to be misguided on a crucial point, and propose another view (with Charles Taylor) that gives credit to all three views; the Reason View because of its focus on value-laden features of the responsible agent; the Real Self View because of its focus on our freedom to govern our actions according to our real self; and the Autonomy View because it gets the Real Self View to take seriously the geneology of the (real) self, and is challenged to give an account that reassures the autonomy of self. The outcome of this, is a view that per se is compatible with both the compatibilist stance and the incompatibilist stance. In the controvers with Wolf, I have to emphasize that autonomy will not be argued for independently, for this another paper is needed, but it will be argued that the view of autonomy put foreward by Wolf (and Nagel) need not be so strict to avoid being eaten up by determinism. I will propose some future path to follow on this question that is compatible with my view, at the end of the paper.
Wolf claims the following conditions necessary for an agent to be responsible (in the sense of my responsibility2 (and perhaps responsibility3)).
1. Only an agent with will - that is, with desires, goals, purposes and the ability to control his behaviour according to these - can be responsible for anything at all.
2. An agent can be responsible only for events and properties that stand in a relation to him such that his will is or could have been effective in determining the existence of these events or properties.
(These first two conditions are roughly analogous with my ability1.)
3. The context wherein an agent is judged as responsible or not, does not only involve the different attitudes vowen into the practices of praise and blame, but our judgements are made against a background of imagined alternatives that could have but did not occur and against a background of reasons that did or could have influenced the agent in question to do one thing rather than another. This means that it must be possible for the features that might weigh in favor of or against that event in relevant ways to enter into the determination or the content of the agents will, that is, his will must be relevantly intelligent.
But even this does not sum up the sufficient conditions of responsibility, or at least so it seems, because it is possible to come up with situations where an agent must be considered to possess both an effective and an intelligent will, but will nevertheless not be praised or blamed for his actions. Wolf has three examples of such situations: an agent being hypnotized, being threatened, or being a kleptomaniac.
In spite of the differences between these examples, they can all be seen to lead back to the same problem - the source of the agent’s will, i.e. that the agent is not in control of what the content of his or her will will be. This condition of responsibility has throughout the history of philosophy since Kant been termed autonomy, and is a requirement that the agent’s control be ultimate, that his will be determined by himself, and not by anything external.
Unfortunately, the concept of autonomy has not been lessened in obscurity since Kant’s account of the noumenal self outside time and causality. Because a closer look at ourselves and our actions may also make us doubt the autonomy of agents not so obviously unfree as in the examples above. If we look to the source of will in completely everyday actions, autonomy seems to exclude me as a responsible agent the moment I grab a fire extinguisher to extinguish the fire in my kitchen, or less dramatically the moment I turn up the heat because I feel cold. It seems that I do not do these things because of me wanting to do them, rather I do them because there is something or someone demanding me to.
The following is what Thomas Nagel in his book The View From Nowhere takes to be our ordinary conception or belief of autonomy.
«It presents itself initially as the belief that antecedent circumstances, including the condition of the agent, leave some of the things undetermined: they are determined only by our choices, which are motivationally explicable but not themselves causally determined. Although many of the external and internal conditions of choice are inevitably fixed by the world and not under my control, some range of open possibilities is generally presented to me on an occasion of action - and when by acting I make one of these possibilities actual, the final explanation of this (once the background which defines the possibilities has been taken into account) is given by the intentional explanation of my action, which is comprehensible only through my point of view. My reason for doing it is the whole reason why it happened, and no further explanation is either necessary or possible. (My doing it for no particular reason is a limiting case of this kind of explanation.)»
Through his whole book Nagel stresses the distinction between two points of view, the internal or subjective and the external or objective. From the internal point of view or perspective our actions are primed with alternative possibilities, and one of them is made actual by what we do. From the external point of view, not only the circumstances of action as they present themselves to the agent are taken in, but also the conditions and influences lying behind the action, including the complete nature of the agent himself. By this perspective we cease to face the world and instead become parts of it, products and manifestations of it as a whole; merely a happening. But in spite of this, our feeling or rather belief that we are the authors of our own actions, are as strong as ever. But, asks Nagel, what belief is this? His general answer is that it is no intelligible belief at all, and tries to show this.
He begins with the description of our ordinary conception of autonomy cited above and continues it by again pointing to the fact that the external view seems to contradict such autonomy, because it admits only one kind of explanation, the nomical causal type. The defence of freedom on the other hand requires the acknowledgement of a different kind of explantion essentially connected to the agent’s point of view.
Nagel’s complaint against such a conception is that «while it may give a correct surface description of our prereflective sense of our own autonomy, when we look at the idea closely, it collapses. The alternative form of explanation doesn’t really explain the action at all». The intuitive idea of autonomy includes conflicting elements, it both is and is not a way of explaining why an action was made. When A makes an autonomous choice such as accepting whether to play a game of chess, and there are reasons on both sides of the issue, we are supposed to be able to explain what A did by pointing to his reasons for accepting. A free action should not be determined by antecedent conditions and should be fully explainable intentionally by way of reasons and purposes, etc. But, says Nagel, in the case of A accepting a game of chess or not, we could equally have explained his refusing to play, if he did not want to play, «by referring to the reason on the other side - and he could have refused for those other reasons: that is the essential claim of autonomy».
«Intentional explanation, if there is such thing, can explain either choice in terms of the appropriate reasons, since either choice would be intelligible if it occured. But for this very reason it cannot explain why [A accepted the game of chess] for the reasons in favor instead of refusing it for the reasons against. It cannot explain on grounds of intelligibility why one of two intelligible courses of action, both of which were possible, occurred. And even where it can account for this in terms of further reasons, there will be a point at which the explaining gives out. We say that someone’s character and the values are revealed by the choices he makes in such circumstances, but if these are indeed independent conditions, they too must either have or lack an explanation.»
By not taking us outside the point of view of the agent, intentional explanations must come to an end when all available reasons have been given, that is, before it has done its explanational job, namely answering «why I did what I did rather than the alternative that was causally open to me».
Either this question has no answer or it has one that takes us «outside the domain of subjective normative reasons and into the domain of formative causes of my character or personality». But this last answer leaves no hope either, for by increasing our objectivity (externality) we seem to acquire increased control over what will influence our actions, but this means, and the logical meaning of being really free, indicates that we would have to act from a standpoint completely outside ourselves, choosing everything about ourselves, including all our principles of choice - creating ourselves from nothing, so to speak (Neurath’s ship). But this is of course self-contradictory. Nagel is therefore at loss to account for what we believe in believing that we are autonomous.
When we broke off Wolf’s account of autonomy above, she had been led to doubt the concept of autonomy because of the source of our desires. We shall now see how she develops her argument, and how much it resembles Nagel’s.
The above observations suggested a picture where our desires result from a combination of heredity and environment, and where we try to satisfy these as well as we can by acting as the situation demands. If this is so, then the source of our desires is an external one, demanded by situation, and determined by heredity and environment. But how can then the concept of autonomy be the mark of responsibility, distinguishing between normal cases and the exeptional cases given as examples above? This seems to rule out autonomy and accordingly responsibility, as Nagel also argued. There is reason, however, to doubt both the plausability and accuracy of this account, and its explanational force, and instead argue that even though we in some situations are determined by external forces, we nevertheless typically act as autonomous agents. But unfortunately this, Wolf says (and following Nagel), is «likely to make the condition of autonomy seem even more puzzling than before». For if we remove any external control, the agent’s control seems to be controlled by nothing at all, and having one’s will must be explained as a result of random events or be a matter of sheer inexplicable fact.
«Autonomy, then, requires that the content of an agent’s will (which, we may assume, determines the agent’s behavior) be up to the agent herself, and this is opposed not only to its being up to anything else, but also to its not being up to anything at all. But now the concept of an autonomous agent may seem to be an impossible one.»
We still have the option of internal explanation, but, says Wolf, this is equally hopeless, because for any agent and act we can still ask for a further explanation of why that agent performed that act. So answering this question in terms of intentions will only postpone the problem, new internal features will be uncovered, but only until we «reach a set of features that must be explained by facts external to the agent, or our explanation will simply come to an end, with the understanding that the agent’s possessing these features is either a random occurence or a brute, inexplicable fact». This is of course exactly Nagel’s point, in fact, nearly verbatimely.
For instance, if I decide to go over to my deranged neighbour to help his mother stop him from smashing his head against the wall, I do it because I want to help. But why do I want to help? Perhaps because it fits my set of values and what person I want to be, a helpful, concerning one. But why do I have this set of values and not another, the set of a satanist for instance? We could explain this by my heredity and upbringing, recent experiences, etc., but then I fit Wolf’s description of a non-autonomous agent again. So the internal explanation seems to end up explaining very little.
«In order for an agent to be autonomous, it seems, not only must the agent’s behaviour be governable by her self - her deeper self, if you like - and this must in turn be governable by her (still deeper?) self, ad infinitum. If there are forces behind the agent, making the agent what she is, then her control of her behavior is only intermediate, and therefore superficial. But if there are no forces behind the agent making the agent what she is, then her identity seems to be arbitrary. [. . .] The idea of an autonomous agent appears to be the idea of a prime mover unmoved whose self can endlessly account for itself and for the behavior that it intentionally exhibits or allows. But this idea seems incoherent or, at any rate, logically impossible.»
The view that Wolf calls the Real Self View, resembles mostly a view defended by Harry G. Frankfurt and others. This view tries, according to her, to avoid the concept of autonomy by, instead of setting up autonomy as a requirement of responsibility, attributes an agent’s behaviour to his real self which thereby constitutes a sufficient and necessary condition of responsibility. This view does not require that an agent be endlessly accountable to himself, i.e. it does not require that one’s self be governable by one’s self ad infinitum. What is required is that an agent have a real self, and that he be able to govern his behaviour in accordance with it. It does not matter, however, where his real self comes from, whether it comes from outside or inside him, or from nowhere at all.
Frankfurt differentiates persons from other creatures by referring to «the structure of a person’s will». Humans are not alone in having desires and motives, or in making choices, what he calls first-order desires, but in our ability to form what he call second-order desires. It is however not having second-order desires generally that he regards essensially to being a responsible person, it is rather having second-order volitions, which is a specific type of second-order desires, the want of some desire to govern our behaviour. The difference between second-order desire in general and in particular (second-order volitions) comes to the fore in the following two sentences:
1) Peter wants to suckle from his mother’s breast.
2) John want’s his desire to stop smoking (to shape up) to be his will.
Peter, being a baby, qualifies to what Frankfurt calls a wanton. The term ‘wanton’ refers not only, like in the case of Peter, to creatures with first- and second-order desires and no second-order volitions, but also to creatures with first-order desires and no second-order desires (and therefore no second-order volitions either). «The essential characteristic of a wanton», he says, «is that he does not care about his will.» What Wolf refers to as a real self then, seems to be the self which is developed or which we develop through our second-order desires or second-order volitions.
Wolf refers to Frankfurt’s distinction between first-order desires and second-order volitions as the distinction between desires and values.
«The difference between values and other desires [. . .] is a difference not in content but in source. The same object of motivation may be merely desired by one person and valued by another, and what is valued at one time may be merely desired (by the same person) at another.»
The relation between valuing and second-order volitions in Frankfurt’s view is that we fail to be free when the desires which are effective in action fail to be the desires we want to be effective in action, and the desires we want to be effective in action are those expressive of what we value. An agent values something if he thinks it good or if he thinks there is some reason to want it. Values are motivations we care about having, they matter to us, and arise out of or are supported by an agent’s real self.
The picture that can be seen to grow out of these considerations is one which sets our valuing selves above our more brute, desiring selves - a map of the structure of a person’s will. Our desiring selves may well on this picture seem to fit what Nagel calls the external or objective point of view, that is, they may be seen as either the inevitable product of the influences of our heredity and environment or arbitrary collections of random dispositions and other properties. If we were nothing more than our desiring selves we would not be free and responsible beings. But just because we are able to form and act on values (second-order volitions), we are able to transcend this causal chain. It is tempting then, in the light of the contrast between our valuing and our merely desiring selves, to identify our valuing selves with autonomous selves and thus adopt the Autonomy View. For in some sense in which we cannot help having the desires we have, we can help what values we have.
«In some sense in which we are stuck with our desires, it seems, we have a choice about our values. And when we focus on those occations in which we actively think about and decide what values to embrace, it seems that prior to our thought we can take one path or another [. . .], for our choice arises out of and is explained by our reflections and deliberations.»
But, says Wolf, it is doubtful that the sense in which we seem to have a choice about our values is properly characterized as an experience of autonomous choice, because what we choose to care about may yet be interpreted in a way that suggests their determination by something external. If we think of our desires as given to us (through whatever set of conditions), and our values as formed by us, the claim that our valuing selves transcend our merely desiring selves, suggests along further Kantian principles, following Wolf, that this ability to form values is something we have «in virtue of having something else - namely, a faculty or method for making judgements and decitions». And this faculty, it is natural, «in the light of the fact that our ability to form values is in addition associated at least hypothetically with the activities of reflection and deliberation», to refer to as Reason. Our values, then, can be seen as being controlled or chosen in accordance with Reason, or (what amounts to the same thing) with our reasons.
The connection between Reason (reasons) and values as Wolf indicates I take to be something like the following. Reason is essentially a normative term, leading us to true beliefs and good values. When I act according to a good value, I have good reasons for doing it. When I decide to go over to my deranged neighbour to help his mother stop him from smashing his head against the wall, I do it because I want to help. Helping other people is something I do because it is good, and when doing it I am judged a good, helpful, concerning human being. Reasons are part of what it means to value something. Value something without giving or having a reason for it seems like a contradiction in terms.
«Our valuing selves may then be identified with our rational selves.» But with this identification, the problem of autonomy reappears through the fact that, if Reason as a faculty is like the given desires in that an agent cannot help having or lacking it, and if it generates motives that an agent cannot help acting upon, then an agent who acts in accordance with Reason is likewise not in ultimate control of his actions. If this is so, Nagel’s challenge is as strong as ever, that is, how can we explain my choice, if autonomy is true? How can an internal explanation explain my choice as free (or autonomous) when I am bound to follow my Reason or (best) reasons? While Nagel closes the case of autonomy with the sentence: «I have to conclude that what we want is something impossible»; Wolf concludes that it is something we don’t particularily want to want.
Being released from the chains of one external determining force just to end up caught in another does not make an agent autonomous. Of course one could postulate another higher faculty to free him from the chains of Reason, but autonomy would also require that agent be free from this further faculty, and so on, ad infinitum. In the light of this, then:
«We may as well understand “Reason” to refer to the highest faculty or set of faculties there are, the faculty or set of faculties, that is, that are most likely to lead us to form true beliefs and good values. If we understand Reason in this way, the autonomous agent must be one who is able to act in accordance with Reason or not.»
Autonomy, then, on this account, is the ability to make radical choices, choices made on no basis and the exercise of no faculty. Since a radical choice must be made on no basis, there can be no explanation of how and why the agent chooses the alternative he does.
«[W]e occationally find ourselves with the choices that no amount of deliberation can settle, choices in which the reason for two or more alternatives are equally strong or incommensurable, or in which our wills simply fail to be engaged by the consideration that a certain alternative is more rational. From the inside, it would seem that we make these choices on no basis».
And in more common situations where we, instead of making radical choices, make rational ones, we can still think that we had the ability to choose an irrational alternative. Our conception of autonomous agents, then, is as follows:
«They must be agents who not only do make choices on no basis when there is no basis on which to make them, but also can make choices on no basis even when some basis is available.»
I think it is worthwhile to have a closer look at this conception of radical choice, because its role in Wolf’s rejection of autonomy is central, and since I think it is confused, it cannot stand unchallenged. But first we have to have a look at Wolf’s solution, the Reason View.
Two requirements support the Reason View, following Wolf; that is, a person’s status as a responsible agent rests on, one, his ability to make his behaviour conform to his deepest values, and two, his ability to form, assess, and revise those values on the basis of a recognition and appreciation of the True and the Good. It is the second requirement that is original about this view, she says.
«If one has the ability to act in accordance with Reason [. . .] it seems that one may be responsible even without being autonomous [. . .][and] if one lacks the ability to act in accordance with Reason, one cannot be responsible even if one is autonomous».
It is this ability to act in accordance with Reason that is supposed to be opposed to the ability to act according to one’s real self, and the ability to act autonomously.
This ability, according to Wolf, is «the ability to be in touch with the True and the Good». What makes responsible beings special is their ability to recognize good values as opposed to bad ones and to act in a way that expresses appreciation of this recognition, i.e. to be capable of recognizing things one might be responsible for.
What does it mean to act according to Reason? It means to do what is True and Good. The True and the Good are objective entities that refer to what is judged to be true and good in our society; compare the pragmatic stance. To be able to do something irrational (non-reasonable) would be to do what is not true or good. It is this ability that we do not want to have, according to Wolf.
What is the meaning of ‘not want’? I go over to help my deranged neighbours mother stop him from smashing his head against the wall. When I do this I do what is considered right or good. But I would not think of myself as non-autonomous even though I act according to Reason, that is, the will of the society. I feel that it is my freedom to do just these things. And I sometimes can’t help myself doing them, because they are what I want to do, and nothing else.
Of course, I could at any time have done something else, for instance something irrational - I am able to do it, so why not? What Wolf means by ‘not want’ can be formulated in the question: If I follow Reason and accordingly am non-autonomous, would I have any reason to believe myself unfree? So if freedom and also responsibility can be accounted for without autonomy, why would I want autonomy?
The Reason View is a solution to Nagel’s challenge because it answers two questions derived from this challenge. First, it is an answer to the question of how we can be free agents even if we act according to something determined by formative causes of our character or personality. Second, it answers the question of how we can be responsible beings without being autonomous. The answer to the first is, as we have seen, that we do not want the opportunity or ability to choose what is not reasonable, an opportunity that would undermine rather than support our feeling of freedom. In other words, freedom is not to be able to choose between Reason and irrationality, but rather to be able to act according to Reason. The answer to the second question then, is that having the ability to act according to Reason (the True and the Good) is a necessary and sufficient condition for someone being a responsible agent. If this ability is missing or the agent does not have the ability to act on his ability, then we do not judge him to be responsible for his actions.
As we remember, Nagel’s challenge was to give an account of autonomy that would make it plausible in spite of its obvious difficulties. How can one explain why an agent chose one possible path instead of another, if autonomy is true? Wolf followed up this point and led us to the belief that ultimately this must be explained by a radical choice of which values to pursue, or we will have to give up the concept of autonomy. If we can dismiss, however, the talk of a radical choice between Reason and irrationality on the ground of its impossibility, then things might look a bit different for the case of autonomy.
Charles Taylor, in his article “What is Human Agency?”, refers to a story Sartre gives as an example in his book L’Existentialisme est un Humanisme, of one of his students who is torn between remaining with his ailing mother and going off to join the Resistance. Sartre’s point is that there is no way of solving this moral dilemma by way of reason or some over-reaching considerations. He has to settle the question by radical choice. But, says Taylor, what makes this account of a moral dilemma plausible, is precisely what undermines Sartre’s position.
«We see a grievous moral dilemma because the young man is faced here with two powerful moral claims. [. . .] But it is a dilemma only because the claims themselves are not created by radical choice. [. . .] Indeed, if serious moral claims were created by radical choice, the young man could have a grievous dilemma about whether to go and get an ice cream cone, and then again he could decide not to.»
The absurdity of Sartre’s position comes to the fore when we apply it to a case where I have an ailing mother and no rival obligation. Do I stay, or do I go for a holiday on the Riviera? There is no doubt I should stay, but of course, I might not. So in this sense there is also a radical choice open, whether to do what we ought or not, or rather to follow reason or not. So far so good for Wolf, but let us ask the following: Can we construe the determination of what we ought to do in a given situation as issuing from a radical choice? For what would this be like? We would be faced with the two alternatives, an sich, so to say, for on the level of radical choice «these alternatives have as yet no contrastive characterization, that is, one is not the path of duty, while the other is that of selfish indulgence, or whatever». It is this contrastive description that will be created by radical choice. So, asks Taylor, what does this choice consist of? He answers as follows:
«Well, I might ponder the two possibilities, and then I might just find myself doing one rather than another. But this brings us to the limit where choice fades into non-choice. Do I really choose if I just start doing one of the alternatives? And above all, this kind of resolution has no place for the judgement ‘I owe it to my mother to stay’, which is supposed to issue from the choice.»
To have this judgement issue from radical choice is not a growing certainty that this judgement is right, for this would not be an account of radical choice, but instead our coming to see that our obligation lies here, and then, our choice would be grounded. In order for there to be a choice, we cannot just find ourselves in one of the alternatives. We have in some sense to feel the pull of each alternative, but given the radical choice situation, my sense that I owe it to my mother to stay instead of going to the Riviera cannot be what pulls me, because this has to issue from the choice. Then, following Taylor, an agent of radical choice has to choose, if in fact he chooses, like a simple weigher or weak evaluator, and not as a strong evaluator.
The concepts of simple and strong evaluator are two kinds of evaluation of desires, which is a development of Frankfurt’s concept of second-order desires, to encompass evaluations which are concerned with outcomes (weak) and evaluations which are concerned with the quality of our desires (strong). What is important here is to see that by quality of desire is meant the qualitative worth of desires. It is this that is missing in the typical cases where, for example, I choose a holiday in the north rather than in the south, and where the favoured alternative is selected not because of the worth of the underlying desire. But this is neither to say that in a weak evaluation the motivations are homogeneous, nor that they are simply quantitative. A weak evaluation can include terms like beauty, well-being, joy, disgust, exhileration, relaxion, etc., but what it misses is a distinction between the desires as to worth, and that is why it is no strong evaluation. I ultimately opt for the north over the south not because there is something more worthy about being exhilerated than relaxed, but just because “I feel like it”. Weak evaluations are only quantitative in the weak sense of not involving qualitative distinctions of worth. Being more fun or of better value are just cover terms for preferred. Nor can we say that weak evaluations are only concerned with outcomes, and never with desires, i.e. that all cases of second-order desires are strong evaluations. For both second-order desires and volitions can be had on the basis of weak evaluations. I want my addiction to rich desserts to abate so that I can control my weight, or I want my desire to stop smoking to be preponent, because I know I will have a better time all things considered if I stop. Again what is missing in these examples is the distinction between the alternative desires in terms of unworthyness or base, alienating or trivial, dishonourable, etc.
Taylor sets out two interlocking criteria on whether desires are distinguished as to worth, the second following from the first:
1) «In weak evaluation, for something to be judged good it is sufficient that it be desired, whereas in strong evaluation there is also a use of ‘good’ or some other evaluative term for which being desired is not sufficient; indeed some desires or desired consummations can be judged as bad, base, ignoble, trivial, superficial, unworthy, and so on.»
2) In weak evaluation, when «one desired alternative is set aside, it is only on grounds of its contingent incompatibility with a more desired alternative». But with strong evaluation this is not necessarily the case. I refrain from committing some cowardly act, not because this act at this moment would make any other desired act impossible, like fighting to the bitter end (of course I cannot do both these things at the same time, but it is not because of this that I make my choice), but because I value courageous action as a part of a mode of life, and I aspire to be such a person.
We need to qualify these criteria further. The term ‘to be judged good’ need not imply that the good is desired. To see this we might invoke a distinction between valuing something and to judge something valuable. While valuing something as good implies that the good is desired, to judge something good valuable does not imply this. One can easily think of someone who judges something good valuable, indicating that he would want to be such a person, without desiring it. So if for something to be judged good, that it be desired is not necessary, how can it then be sufficient in the strong sense implied by this criterion? I think we must understand Taylor’s concept of evaluation exclusively in the sense of valuing something. Because of this then, he can assess that for something to be valued in weak evaluation it is a necessary and sufficient criteria that it be desired, while in strong evaluation for something to be valued it is necessary that it be desired, but not sufficient. Some considerations above the level of desire have to be accomplished to have this sort of evaluation. I think these considerations are connected to what I, in ftn. 51, call valuing most.
Another qualification concerns the second criteria and the distinction between the contingent incompatibility of weak evaluation and the non-contingent incompatibility of strong evaluation. One could think of a situation in which a strong evaluation brings forth two contingent incompatible desires, for instance my want to be heroic or brave on the left flank and on the right flank. Another objection comes forth if we look at another situation from two points of view, one strong evaluated, the other weak. At a battlefield, I have two desires, to flee and to fight. As a weak evaluator I would do what I most desired (according to the first criteria), let us say I wanted to flee. As a strong evaluator I would judge not only according to my strongest desire but also invoke concepts like courage and cowardice to describe my desires. In neither case would it seem as though these desires were contingent. Fleeing and fighting would in no possible world be said to happen to the same individual at the same time, so they logically exclude each other. So it seems that contingency is not decisive in upholding this distinction, even though it fits Taylor’s own examples very well. These objections can be dealt with if we get to understand Taylors claim of contingency better, and to do this we shall first look at the concept of contrastive language that he develops just from this concept of contingency.
While the concept of fighting can be understood without the concept of fleeing, the concept of courage cannot be understood without its grammatical opposite cowardice. This means that while the alternative desires in weak evaluation can either be described contrastively or non-contrastively, the alternative desires in strong evaluation must be described contrastively, i.e., the contrastive element is built into these concepts. In other words, strong evaluations are based on a language of evaluative distinctions, a language «in which different desires are described as noble or base, integrating or fragmenting, courageous or cowardly, clairvoyant or blind, and so on». This means that while a desire to flee, described as an action of cowardice when seen from the point of view of a strong evaluator, grammatically invokes the concept of courage (and the other way around), the desire to flee does not, when seen from the point of view of a weak evaluator, invoke grammatically the concept of fighting. Fighting and fleeing are not contingently incompatible in the way that they could both logically be possible at the same time (for the same person), but they are contingently incompatible in the sense that brushing teeth or playing chess is contingently incompatible with both. And it is in this last sense that courage and cowardice are not contingently incompatible. Their incompatibility is based on a contrastive language, a language in which the terms have a certain internal or grammatic dependency.
The distinction between weak and strong evaluations opens up for a certain sense of choice of interpretation, between a non-qualitative interpretation and a qualitative one. If I base my case of not over-eating on my sense of well-being and ability to move, I choose a non-qualitative self-interpretation of my problem. This can help me solve my problem, but it does not have the same depth as when I interpret my problem as one of dignity and myself as a self-diciplined, autonomous agent. The question, however, of which alternative is more valid or true to reality, can rise in this connection. But to be in error here is not just to make a misdescription, as when I describe an animal as a cat when it is really a lynx, rather it is a certain destortion of reality. Another point growing out of the distinction between strong and weak evaluation is the recognition of the different kinds of self each involves. The weak evaluator has the necessary features of being a person or a self: reflection, evaluation and will; but in contrast to the strong evaluator he lacks something else we often refers to with the metaphor of ‘depth’.
The depth of the strong evaluator is revealed in his use of a richer language. While the weak evaluator evaluates desires according to what he desires plus a calculation of consequences, and a registering of the conclusion that desire A is more desirable than desire B, the strong evaluator is not similarily inarticulate. His language is one in which the expression of superiority of desire A is made in terms of worth, that is, of contrastive characterization. The sense of depth, however, goes deeper than articulation, it also involves the concept of a person, or the different possible modes of being of the agent.
«Motivations or desires do not only count in virtue of the attraction of the consummation but also in virtue of the kind of life and kind of subject that these desires properly belong to.»
This also opens a new possibility, a possibility of a plurality of visions which there was not before, so that my choice of desire may not just be between what is clearly the higher and the lower, but also between two incommensurable ways of looking at my choice.
So it seems that while the agent of radical choice wants his choice to be based on strong evaluations, it is in fact based on weak ones, because radical choice excludes quality-concerns ex hypothesi. For again by hypothesis, «what leads him to call one alternative higher or more worthy is not that in his experience it appears to be so, for then his evaluations would be judgements, not choices; but rather that he is led to plump for one rather than the other after considering the attractiveness [outcome] of both alternatives».
There is a sense of responsibility for oneself that goes along with the notion of the agent as a strong evaluator.
«Naturally we think of the agent as responsible, in part for what he does; and since he is an evaluator, we think of him as responsible in part for the degree to which he acts in line with his evaluations. But we are also inclined to think of him as responsible in some sense for these evaluations themselves.»
Accordingly we can distinguish between what I have called responsibility2 and responsibility3:
1) Responsibility2 is already implicit in the notion of will. A being capable of evaluating desires may find that the upshot of such evaluation is in conflict with the most urgent desire. Indeed, we might think of it as a necessary feature of the capacity to evaluate desires that one is able to distinguish a better one from the one that presses most strongly. Thus we think of an agent as partly responsible for what he does and for the degree to which he acts in line with his evaluations.
2) Responsibility3 builds on our ‘modern’ notion of the self and is stronger. We think of the agent not only as partly responsible for what he does, for the degree to which he acts in line with his evaluations, but also as responsible in some sense for these evaluations. 
As I have already proposed, two senses of autonomy follow this distinction. Autonomy2 is one in which, although one’s desires are given (but nevertheless autonomous in the sense of autonomy1), one’s will or choice to act according to these desires is free, that is, one is free to pick the desire one wants to act on. Autonomy3 on the other hand builds on the recognition that not only are we responsible for what desire we act on according to our evaluation, but we are also responsible for our evaluations, that is, we are autonomous in the sense that we are free to choose what evaluation we want to use in picking a then accordingly wanted desire.
It is ultimately the question of how to understand responsibility3 and accordingly autonomy3, that brings forth a Wolfean understanding of it as a radical choice between evaluations or values. As Taylor says:
«The Nietzschean term ‘value’, suggested by our ‘evaluation’, carries this idea that our ‘values’ are our creations, that they ultimately repose on our espousing them. But to say that they ultimately repose on our espousing them is to say that they issue from a radical choice, that is, a choice which is not grounded in any reasons.»
It is on the background of these considerations that we have to understand Wolf’s rejection of autonomy and Taylor’s rejection of her rejection. On Wolf’s account, a radical choice is in fact made up against reasons (Reason), that is, what is seen as true and good. Following her conception of what an autonomous agents must be like, she says:
«In other words, they must be agents for whom no basis for choice is necessitating. If the balance of reasons supports one alternative over all the others, it is still open to them to choose whether to act in accordance with the balance of reasons or not.»
This is however not the claim of radical choice, for as we saw with Taylor, what distinguishes a radical choice is not that it does not necessitate, but that it is made on the basis of no strong evaluations, i.e, on the basis of no qualitative considerations, whatsoever. «The agent of radical choice would», in Taylor’s words, «at the moment of choice have ex hypothesi no horizon of evaluation.»
But will not this make the case for autonomy and accordingly responsibility just worse, showing radical choice to be not only something we do not want, but also impossible? What is the sense we can give to responsibility3 and autonomy3, if we are not to understand it in terms of radical choice? There is in fact another sense in which we are radically responsible, that is, autonomous. If we look a little closer at the grammar of evaluations, we see that evaluations are not chosen.
«On the contrary they are articulations of our sense of what is worthy, or higher, or more integrated, or more fulfilling, and so forth. But this sense can never be fully or satisfactorily articulated. And moreover it touches on matters where there is so much room for self-deception, for distortion, for blindness and insensitivity, that the question can always arise whether one is sure, and the injunction is always in place to look again.»
Articulations, for Taylor, are closely connected to other concepts he uses, especially in his philosophy of language, concepts like: expression and manifestation, that stand in sharp contrast to designation. «What expression manifests can only be manifested in expression.»
An articulation is not some description of a fully independent object, that our values or evaluations are canalizations of clear-cut choices made by some faculty. The whole Kantian philosophical tradition is soaked through by thinking that the product of something needs some faculty, more generalized than its product, but after all, entirely similar in form. Of course the form of a bronze statue is governed, or rather determined, by its mould, but there is no reason to think of values and other human articulations as governed, or again rather determined, by some value-moulder. Or to say it differently, even though values govern, in some sense, a sphere of desires, there is no reason to think that some other faculty governs a sphere of values. The problem we are confronted with, and which in fact the whole Kantian tradition is faced with, is our conformity in thought, a conformity that is evidently manifest in our thinking of human action - thinking of it in terms of nomic causality, or in terms of foundry.
«To give a certain articulation», says Taylor, «is to shape our sense of what we desire or what we hold important in a certain way.» Our self-interpretations - descriptions or redescriptions of desires - are partly constitutive of our experiences, that is, an alterered description of our motivations can be inseparable from a change in this motivation. By this we are not putting forward a hypothesis of a causal connection between the altering of a description and our experience of our situation, that we alter our descriptions and then as a result our experience alters, but rather «it is that certain modes of experience are not possible without certain self-descriptions».
Taylor gives as an example a story of a person that is addicted to over-eating, a person that finds it hard to resist treating himself to rich dessert. There are two main kinds of ways this person can evaluate his desire, following the distinction between weak and strong evaluations. Strongly evaluated, or evaluated by way of a language of qualitative contrast, his over-eating can be described in terms of dignity and degradation; not being capable of freeing himself from his addiction and instead ruins his health is not very admirable, it deprives him of his dignity and degrades him to a person run by his desires. Weakly evaluated, however, his problem may be seen as a question of quantity of satisfaction. Eating too much dessert increases the cholesterol in his blood, makes him fat, ruins his health, and so on. If we say that this man has been talked into seeing his problem as a merely quantitative question rather than qualitative, then his inner struggle has been transformed and is now quite another experience. Through this new interpretation of his desire, his experience has altered, it is no longer understood as a seeking for dignity and self-respect but as a question of more satisfaction. The opposed desires which are the ‘objects’ undergoing redescription here - the craving for rich desserts and his dissatisfaction with himself at such indulgence - are not independent in the sense outline here. When he comes to accept the new description of his desire to control himself, the desire itself has altered. It is difficult to avoid the ordinary descriptive model that prescribes a picture where we might think of the change in terms of some sense of shame and degradation being detached from our desire to resist over-indulgence, which has now simply the rational goal of increasing over-all satisfaction (utilitarianism). This picture tries to maintain the impression that the elements are just rearranged while remaining the same. But on a closer inspection we see that on this picture too the sense of shame does not remain the same. It has become something quite different. It is in this way that our self-interpretations, our altered descriptions of motivations, are partly constitutive of our experience.
We have seen that a strong evaluator is distinguished from a weak one through the following features. A strong evaluator is concerned with the quality of a desire, not primarily with its outcome, that is, quality in the sense of worth for the agent. To value something as a strong evaluator it is necessary, but not sufficient that it be desired, something in addition is needed. This something in addition is found in the use of a richer language, a language of depth. The richer language makes the incompatibility between values non-contingent, not contingent like in weak evaluations. The richer language also involves that the evaluation of our desires are articulations of what we find worthy or base, etc. This means further that these articulations are constitutive of our experience of our desires, and are therefore self-interpretations. So we are, as strong evaluators, as human beings and as responsible agents, self-interpreting.
Taylor (1977b) offer a picture of man as a self-interpreting being through five claims. These are as follows.
1. Some of our experienced motivations (feelings, emotions, desires) involve import-ascriptions.
2. Some of these imports are subject-referring.
3. Our subject-referring feelings are the basis of our understanding of what it is to be human.
4. These feelings are constituted by the articulations we come to accept of them.
5. These articulations, which we can think of as interpretations, require language.
I will not here discuss the impact of all these claims, but I will look closer into the concept of import-ascriptions, since it seems to help us understand better the concept of strong evaluation.
Emotions are essentially related to certain objects, or at least, so it is often said to be. To experience fear is to experience some object as terrifying or dangerous, and to experience shame is to experience some object or situation as shameful or humiliating. The natural objection to this is that some emotions are objectless, as nameless dread or unfocused anxiety. But what remains is my sense of the siuation. I still have a sense of threat, of that something harmful will happen. So perhaps it is better not to put the point in terms of an essensial relation to objects, but speak rather of these emotions as essentially involving a sense of our situation. Obviously the same is valid for desires. The strong evaluated experience of the person above, addicted to over-eating, that finds it hard to resist the desire of treating himself with rich dessert, experience a loss of self-control and accordingly shame because of this. Experiencing a given desire involves experiencing our situation as being of a certain kind or having a certain property. But this property cannot be neutral, cannot be something to which we are indifferent, or else we would not be moved. Rather, experiencing a desire is to be aware of one’s situation as humiliating, or shameful, or outrageous, or dismaying, or exhilerating, or wonderful, etc. Each of these adjectives defines what Taylor calls an import.
«By ‘import’ I mean a way in which something can be relevant or of importance to the desires or purposes or aspirations or feelings of a subject; or otherwise put, a property of something whereby it is a matter of non-indifference to a subject.»
‘Whereby’ is meant in a strong sense, because in identifying the import of a given situation we are picking out what in the given situation gives the grounds or basis of our feelings, or what would give such grounds, or perhaps should give such grounds, if we feel nothing or have inappropriate feelings. We are not just stating that we experience a certain feeling in this situation, like we do when we say that we feel pain because the situation is painful, or we say we desire cake because we like it. This last is what weak evaluations are all about. We now have another concept distinguishing strong evaluations from weak ones, strong evaluations involve imports.
«[E]xperiencing a given emotion [or desire] involves experiencing our situation as bearing a certain import, where for the ascription of the import it is not sufficient just that I feel this way, but rather the import gives the grounds or basis for the feeling. And this is why saying what an emotion [or desire] is like involves making explicit the sense of the situation it incorporates, or, in our present terms, the import of the situation as we experience it.»
The fact of self-interpretations being constitutive of experience says nothing whatsoever about how changes in both descriptions and experience are brought about. Taylor holds that there are two different ways that change can be brought about; (i) we are in some circumstances led to reflect on our experience of our situation, and can through this gain new insight into it, and hence change our experience. In other circumstances, and more fundamentally, (ii) we see that certain descriptions of experience are unacceptable or incomprehensible to us because of the nature of our experience - our former description may seem to avoid the deeply rooted character the experience has for us. The way descriptions and experiences are bound together in this constitutive relation, admits then of influences in two directions; (i) experience of our situation may be altered by way of our description of a motivation, because we are coming to fresh insight, but also (ii) our description will alter because of the nature of the experience that it is deeply embedded in us.
«[ii] Our attempts to formulate what we hold important must, like descriptions, strive to be faithful to something. But what they strive to be faithful to is not an independent object with a fixed degree and a manner of evidence, but rather a largely inarticulate sense of what is of decisive importance. An articulation of this ‘object’ tends to make it something different from what it was before.
[i] And by the same token a new articulation does not leave its ‘object’ evident or obscure to us in the same manner or degree as before. In the fact of shaping it, it makes it accessible and/or inaccessible in new ways.»
Since articulations partly shape their objects in these two ways, they are intrinsically open to challenge in a way that simple descriptions are not. Evaluation is such that there is always room for re-evaluation.
Wolf’s radical choice-description of autonomy3, then, can be redescribed as a radical re-evaluation. This radicallity, however, is not radical in the sense of a choice without criteria, but rather in the sense that our looking again at our values (evaluations) can be so undertaken that in principle no formulations are considered unrevisable.
«The question can always be posed: ought I to re-evaluate my most basic evaluations? Have I really understood what is essential to my identity? Have I truly determined what I sense to be the highest mode of life?»
A radical re-evaluation questions the most basic terms, those in which other evaluations are carried on. It is always possible for a reflective human being to question what is considered right and good at any time. This is what it is to have ability3, and thus be autonomous3.
But how can such re-evaluations be carried out? There is no metalanguage or higher faculty in which I can assess rival self-interpretations or govern these according to. If it were, there would not be radical re-evaluation, and probably not autonomy3. The re-evaluation is carried on in the language available, but with a stance of attention to what the articulations are meant to indicate and «with a readiness to receive any Gestalt shift in our view of the situation, any quite innovative set of categories in which to see our [situation], that might come our way in inspiration».
Having reached bedrock in our explanation of autonomy with the concept of articulation, is there still any power left in Nagel’s arguments for the impossibility of autonomy3? To the libertarian claim that «anyone who does not accept an account of what I was up to as a basic explanation of action is the victim of a very limited conception of what an explanation is - a conception locked into the objective standpoint which therefore begs the question against the concept of autonomy» - which may be seen as compatible with the view argued for above - Nagel raises the following objection:
«[A]ren’t these autonomous subjective explanations really just descriptions of how it seemed to the agent [. . .] to what he did; why are they something more than impressions? Of course they are at least impressions, but we take them to be impressions of something, something whose reality is not guaranteed by the impression.»
And, he says, not being able to say what this something is he is at a dead end. He is at a dead end, a dead Kantian end, because he does not realize the grammar of values or evaluations, as articulations or avowals. They are not impressions, they are expressions.
Articulations, that is, strong evaluator descriptions, just because they partly shape their objects, engage our responsibility in a way that simple descriptions do not, following Taylor. And this happens in two related ways which corresponds to the two directions of influence mentioned above.
Let us first concern ourselves with situations where our responsibility is engaged according to altering of descriptions because of the nature of the deeply embedded shape of experience (ii). Our insights into what is important and of value (the True and the Good, following Wolf) are often limited by our experiences, and we may fail to understand a certain insight, or to see the point of some moral advice given. An insensitive person, or fanatic, cannot see what he does to others, that he for instance by his act affronts someone’s sense of honour. Such a person cannot listen to us because he has closed off all sensitivity to questions of honour, etc., due, perhaps, to earlier experiences in his life, shaping his present experience, making these questions of no account for him. The current shape of his experience makes it impossible for him to allow the insights we are pressing on him, he cannot admit them without his whole stance towards these matters crumbling, and this stance may be motivationally of deep importance to him. What this person fails to see, because of the nature of his shape of experience, is a certain insight, an insight we feel is right. He has accordingly another description of experience than us, one that has been altered because of what he has become. Thus we take the limits of that person’s insight as a judgement on him. Because of what he has become, «he cannot see certain things, cannot understand the point of certain descriptions of experience». On this account, Taylor divides between three senses of responsibility2:
1) We attribute responsibility to people in relations to outcomes that they can presently encompass or avoid.
2) We attribute responsibility to what the agent could have done differently in the past.
3) We attribute responsibility in that we judge people morally on the basis of what insights they see or do not see.
We remember Wolf’s example of a woman who walks past a shop window, seeing a book that she knows her friend has been searching for for ages, but does not stop to buy it, because «she is so unfamiliar with examples of sincere, noninstrumental friendships that the thought “I should buy this book, just to make my friend happy” cannot help appearing irrational to her». This woman acts in accordance with her Real Self, a self incapable of real friendship; her reaction to such situations (or description of them) are what they are because of what she is. Because of this we would not judge her responsible in the first sense, and neither in the second, at least if what she has become can be attributed to things happening to her. We could however judge her responsible for not seeing that her evaluation goes against what we call friendship, if she is capable of it at all. In this way we avoid Wolf’s judgement that «according to the Real Self View, this person is responsible for her failure in friendship». Instead, she is responsible for failing to see a certain insight, the sense in which we think people responsible for their evaluations that has nothing to do with radical choice.
In situations, however, where our responsibility is engaged according to our alteration of ourselves and our experience by fresh insight, i.e. (i), we think of responsibility of evaluations in another sense, a sense that has more of the radical choice situation inherent.
«Responsibility falls to us in the sense that it is always possible that fresh insight might alter my evaluations and hence even myself for the better. So that within the limits of my capacity to change myself be fresh insight, within the limits of the first direction of [. . .] influence, I am responsible [. . .] for my evaluations.»
This means that we are responsible3 in some sense for the persons, or Real Selves, we are. We must be open to challenge even our deepest and fundamental evaluation, those which provide the terms in which other less basic ones are made. Of course, this is among those things that makes being a person difficult, and existentially challenging.
We have seen that Wolf in her rejection of autonomy, rejects both what I have called autonomy2 and autonomy3. But although she really gives no account for the freedom of self that I have called autonomy3 and accordingly responsibility3 and ability3, she tries to give an account for responsibility2 and ability2 without autonomy2. Let us look at her arguments.
According to the Reason View, the freedom necessary for responsibility consists in the ability to do the right thing for the right reasons, i.e., the ability to choose and to act according to Reason (the True and the Good).
«For it seems compatible with a person’s being able to choose and to act in accordance with Reason that she be determined to do so, and so the answer to the question “Is determinism compatible with responsibility?” would seem at least sometimes to be yes.»
Wolf gives a characterization of what is involved in attributing to someone the ability to X, which consists of two claims, one positive and one negative:
1) «[T]he individual to whom the ability is attributed possesses whatever capacities, skills, talents, knowledge, and so on are necessary for X-ing.»
2) «[N]othing interferes with or prevents the exercise of the relevant capacities, skills, talents, and so on.»
Accordingly, the claim that A is unable to X can be understood as the claim that either (L): A lacks at least one of the capacities, talents, etc., that are necessary for X-ing; or (P): something prevents or interferes with A’s exercise of these capacities, talents, etc. Wanting to know whether (D): being determined not to X implies (U): being unable to X, that is, whether D implies U, is a matter of wanting to know whether D implies either L or P.
It is hard to see why the truth of D always should imply the truth of L. If I am determined not to go over to my neighbour, it does not mean that I am not able to. That there is no connection between D and the truth or falsety of P, however, is not so obvious. What we need to know is whether D implies that something prevents one from X-ing, that is, we need to know whether, if I was determined not to go over to my neighbour, something prevented me from exercising my ability to do so. This question clearly presses in front a question of what kind of determinism is at stake here. It is doubtful that the kind of determinism, or the kind of determinant of action that von Wright calls a nomic cause, operates in this way.
Wolf comes down, through a history of God, possible worlds, and Rose, on the compatibilist assumption that a «psychological ability to do otherwise is all the ability it makes sense for someone to care about». If it is determined on the physical level of explanation that I will choose to remain seated at my breakfast table, then can I not choose to go over to my neighbour after all, even if I have the ability to do so? This depends on the answer to the following question, following Wolf: Do the physical facts make me choose (or explain why I choose) to remain seated, or does my choice to remain seated make (or explain why) the physical facts come out as they do? Obviously, the latter seem more reasonable than the former.
My situation when deciding what to do, then, includes the following two salient features: first, I have all the skills, talents, and opportunities that are necessary for me to go over to my neighbour to help his mother. Second, nothing interferes with or prevents me from exercising these skills, talents, and capacities. My child is safely secured in her playpen, I have no relevant aversion that would make me want to avoid the situation, and no other physical constraints, from the building for instance, stop me from doing it. So if one holds that the psychological level of explanation is as basic or more basic than any other level of explanation with respect to my action, then, the ability to do otherwise relevant to an assessment of my freedom and responsibility seems to be discernible at the psychological level of explanation alone.
The abilities necessary for responsibility, then, are not as such incompatible with physical determinism, following Wolf.
«It being divinely or physically impossible that A do X does not imply that A lacks the ability to do X, that A cannot do X, in any sense relevant to the assessment of A’s responsibility. It is not, then part of the meaning of “ability” or “can” used in contexts of responsibility-assessment that they imply physical (or divine) possibility.»
Following this, arguments that move from “It is determined that A do X” or “It is impossible that A do X” to “A cannot do other than X, in a sense relevant to freedom and responsibility” are inadequate. We are perhaps physically determined, but not psychologically determined, not determined at the psychological level of explanation. We are faced with choices for which we have skills, talents, and capacities necessary for choosing one way or the other, and nothing prevents us from exercising our skills, talents, etc. This means that we are faced with choices for which it is compatible with our psychological histories up to now, in conjunction with all the psychological and psychophysical laws that are true for us, that we choose one way, and it is also compatible with all this that we choose another.
Wolf’s compatibilist solution shows us why our very natural assumption is unwarranted, that is, the assumption that if it is physically determined that A do X, then it must be the case that something prevents A from doing anything other than X, or that somethings interfere with A’s ability not to do X. This tendency, says Wolf, «to think that physical determinism must interfere with our psychological freedom comes, I believe, from the thought, “Given physical conditions and laws just as they are, my behavior is determined regardless of what I want, think, and choose”.»
It is difficult I think to understand Wolf’s position on this point, partly because she argues with a story of divine determinism, long forgotten (luckily) in philosophy, and partly because she does not properly distinguish between physical and psychological explanations. Intuitively, however, I think her arguments right in some sense, at least compatible with the view I have exposed above, for I do not see any reason why Wolf cannot acknowledge at least autonomy2. The reason for this, I think, is either that what Wolf calls ability is what I call autonomy, or what I call autonomy is an additional feature not properly taken care of by her theory of responsibility. Some explanation is also to be found in our different thinking about determinism. I acknowledge a weak sense of autonomy already at a low level of complexity of agency (autonomy1), an autonomy connected to the ability of a system to respond to the external world, that is, represent it and act on such a representation (belief). Since representation always involves the possibility of misrepresentation, it is disconnected, although in a weak sense, by way of meaning, from the nomic causal chain. Representation is not a blind act, rather it is dependent upon internal features or abilities of the system in question. If we construe this autonomy in an ability of the agent, however, there might not be so much disagreement between Wolf and myself.
The inner nature of Wolf’s case against autonomy2 and thus autonomy3 is that she thinks of it like a choice without any guidelines, a radical choice. At the same time she sets up this choice as a choice we do not want, because it would be a choice between Reason and irrationality. As a solution to this she puts up the possibly externally determined ability to act according to Reason as the mark of responsibility. I have shown, I hope, that in choosing what to do in a way that could make us responsible, we act as strong evaluators, as agents with knowledge to what is true and good. In this, there is the sense of responsibility that includes the self of the agent, not only his actions. Neither of these can be construed as radical choices. The possibility of radical choices counted out, we see that autonomy cannot be as radical as indicated by Wolf. In what sense can we then talk of autonomy? I tried to construe this along lines of agents as creators of meaning, as able to articulate reasons for choices, and as able to articulate themselves, making radical re-evaluations. But if this is an ability, is it one we have according to something else? Something external? Yes, I do think so, but not in any simple way. In constructing our lives and articulating our choices, we do something other than just following a determined path, or at least it needs other explanations than nomic causality or determinism can offer. On these grounds, I think Wolf’s concept of ability has to be investigated, to see its real nature, and its geneology. And until this investiagtion has been completed, we need at least autonomy as a guide, or perhaps as a reminder of what we are looking for. No doubt, meaning and the concept of self is at stake her.
Thus, I see two possibilities for further investigations based on my critique of Wolf, either to acknowledge her concept of ability and letting my concepts of autonomy (autonomy2 & 3) name this ability, or to argue that my sense of autonomy names something missing in Wolf’s account, and thus reassure it as a necessary condition of responsibility distinguishable from ability. Neither of the possibilities will be pursued here, however, because this paper is running out of space. But before I end it, I will investigate tentatively, what kinds of explanations are compatible with my view.
We saw that Wolf distinguished between two compatible explanations, one physical and the other psychological. Dennett (1978) distinguishes between different explanational stances that one can take against a system, all compatible with determinism (or mechanism), and Dretske (1988) indicates that his reason-descriptions are compatible with physical deterministic explanations. Pears (1978) and Davidson (1973) argues respectively that intentional explanations are compatible and have several similarities to psychological determinism, and that reasons are causes in some sense.
Davidson (1973) proposes a causal theory of freedom, in which causal power is «a property of an object such that a change of a certain sort in the object causes an event of another sort», but it is necessary in this case to investigate what ‘change of a certain sort’ is. Dretske points to a way with his concept of a representational system, but this meets per se only the requests on a low level of autonomy (autonomy1), and explains little (at least as his theory stands) about higher states. This indicates that the power of this kind of explanations may decrease as the complexity of the system and its actions increases. Certainly this corrupts the possibility of prediction.
Dennett (1978), distinguishes between three stances one can take against a system to explain and predict its behaviour. First, the design stance, that says that if one can know exactly how a system is designed, one can predict the response to external impulses. Second, the physical stance, from which our predictions are based on the actual state of the particular system, and are worked out by applying whatever knowledge we have of the laws of nature. Third, the intentional stance, he says, tends to be the most appropriate when the system one is dealing with is too complex to be dealt with effectively from the other stances. Intentional explanations explain a bit of behaviour, an action, by making it reasonable in the light of certain beliefs, intentions, desires ascribed to the agent. Dretske (1988) can be seen to combine this laste stance with the first, in that he tries to explain the intentionality from the design of the system, the systems ability to be intentional. There is also a similarity with Wolf on this matter, in her focus on the ability of the agent to behave according to Reason. In her assessment of the scope of her theory, she says that the Reason View shall provide us with what we would need to know if we wanted to discover whether a person is responsible for an act or not. And knowing this, she says, is to know whether someone have the ability to act according to the True and the Good, ie., Reason. This we could call a dispositional stance, a mix between the design and the intentional stance. The explanation that comes out of this is one in which prediction is based first, on the ability of the agent to express or be in touch with Reason, and just secondly on the agent’s rationale for the action. That this is so can be seen in what Wolf calls the asymmetry of the Reason View.
So what about my view, what explanations supports this view, or at least are compatible with it? It can be argued, I think, that all the explanations mentioned above are compatible with my view, so the question becomes rather what explanation of autonomy or freedom secures my view in the best way, or what explanation attends in the best possible way to the features I have focused on as important? It has to take care of the following features, among others: articulation, freedom to radically re-evaluate, capacity for strong evaluation.
Let us here just opt for a psychological explanation of freedom in line with what Wolf takes a such to be. She says it must mean that agents are faced with choices for which it is compatible with their psychological histories up to the point of these choices, in conjunction with all the psychological and psychophysical laws that are true of them, that they choose one way, and it is also compatible with all this that they choose another. This is the opposite of psychological determinism - that is, the thesis that all psychological events are uniquely and wholly determined by conjunction of laws and states of affairs that are capable of description at the psychological level of explanation - and is rejected by the characterization of what it means to attribute ability to someone, given above.
But what does it mean that a choice is compatible with an agents psychological history? It means that whatever action the agent does, as long as acting on a choice compatible with his psychological history, it is compatible with his psychological story. Let us say that an agent’s psychological history (A) has enabled him with an ability both to do what is right and good (g), and what is wrong and bad (b). This means that A is compatible with both doing g and b. We will judge him responsible for both actions, praise him for the first and blame him for the second. The question now is what makes him choose? An ability to be in touch with the True and the Good is not a state of mind, or some disposition. It is not so that we can compare the ability to do g with the doing of g. Rather being able to do g and to do b is being able to evaluate these actions as good or bad. And to evaluate something as good or bad means to understand what good or bad means. Choosing between g and b means being a strong evaluator, and being a strong evaluator means articulating one’s self. So it seems to me that looking at a choice of action as a sort of stimulus-respons process, like: A ® (g Ú b), is too simple. The process of choice seems more like a looping process in that the objects of choice needs not remain the same through the process of choice. Therefore it seems that an explanation of freedom based on psychological facts about the agent cannot really give an account of what I have stressed as important for assigning responsibility2 and especially responsibility3.
Let me argue from another angle. If we explain the doing of g from the ability of g, and the doing of b from the ability of b, we can say that if the ability to g is acted on, the ability to b is not acted on. Seen from a psychological point of view, the explanation of why the agent did g or b is completely like, since both is compatible with A. So this form of explanation does not tell us anything at all about why the agent acted, it is completely backward-looking, and backward looking in a blindfolded way. Telling why the agent did what he did, we need forward-looking explanations, explanations that are teleological and intentional, that is, descriptions of behaviour in form of purpose, bent, desire, etc., and descriptions of behaviour that takes account of the meaning of things, environment and self, for the agent. If we can do this properly, we are closer to give an account of autonomy and responsibility that takes care of the features that I have argued are important. Then we might also be able to answer Nagel’s challenge better, that is, why the agent did what he did rather than the alternative that was causally open to him.
· Davidson (1971) - Davidson, Donald: “Agency”, Essays on Actions and Events, New York 1980.
· Davidson (1973) - Davidson, Donald: “Freedom to Act”, Essays on Actions and Events, New York 1980.
· Dennett (1978) - Dennett, Daniel C.: “Mechanism and Responsibility”, Essays on Freedom of Action, (ed.) Ted Honderich, London 1978.
· Dretske (1988) - Dretske, Fred: Explaining Behavior: Reasons in a World of Causes, London 1995.
· Frankfurt (1971) - Frankfurt, Harry G.: “Freedom of the Will and the Concept of a Person”.
· Frankfurt (1991) - Frankfurt, Harry G.: “The Faintest Passion”.
· Johnston (1989) - Johnston, Paul: Wittgenstein and Moral Philosophy, London 1989.
· Nagel (1974) - Nagel, Thomas: “Subjective and Objective”, Mortal Questions, New York 1996.
· Nagel (1986) - Nagel, Thomas: The View From Nowhere, New York 1986.
· Pears (1978) - Pears, David: “Rational Explanation of Actions and Psychological Determinism”, Essays on Freedom of Action, (ed.) Ted Honderich, London 1978.
· Strawson (1962) - Strawson, Peter: “Freedom and Resentment”.
· Taylor (1976) - Taylor, Charles: “Responsibility for Self”.
· Taylor (1977) - Taylor, Charles: “What is Human Agency”, Human Agency and Language: Philosophical Papers 1, New York 1992.
· Taylor (1977b) - Taylor, Charles: “Self-interpreting Animals”, Human Agency and Language: Philosophical Papers 1, New York 1992.
· Taylor (1978) - Taylor, Charles: “Language and Human Nature”, Human Agency and Language: Philosophical Papers 1, New York 1992.
· Wittgenstein (1953) - Wittgenstein, Ludvig: Philosophical Investigations, Oxford 1992.
· Wolf (1990) - Wolf, Susan: Freedom Within Reason, New York 1993.
· Von Wright (1980) - von Wright, G. H.: “Freedom and Determination”, Acta Philosophica Fennica, Vol. 31 (1980), No. 1.
 A paper written for: Deleksamen til Filosofi hovedfag, område 3. Vår 1997.
 I follow Fred Dretske’s arguments in Explaining Behavior (Dretske (1988)) that actions are a particular species of behaviour. To qualify as an action it seems that it is either something done voluntarily or deliberately, or a direct consequence, whether intended or foreseen or not, of such a voluntary act. Boarders are blurred here, however, so there is not easy to put down what qualify some behaviour as action.
 By internal is not here meant intentional, because that would deprive far to many beings the status of doers.
 This distinction between happening to and doing is prone to break down or collapse when we try to use it on objects like stones: what does a stone do when you put it in water? It sinks. There is however no reason to look at this behaviour as internally produced, so our distinction is really applicable only to systems that exhibit enough structural complexity and internal articulation to make the internal-external difference clear and well motivated. Vide Dretske (1988), pp. 1-11.
 Not to be confuced with the pragmatic stance mentioned later on.
 Dretske (1988), p. 27.
 This term is from von Wright (1980). Nomic causes (or Humean causes) are «causes related to their effect by a law which is an inductive generalization» (Von Wright (1980), p. 27.)
 The distinction is from Strawson (1962). Actually he doesn’t use the concept of autonomy, rather responsibility as opposed to determinism. Since I take autonomy in the right sense (the sense that refers to the sense our belief in it must refer to) to be a necessary ingredient of responsibility, I think that if determinism is incompatible with responsibility, it is also incompatible with autonomy, and if responsibility in some sense is compatible with determinism, then autonomy is also compatible with determinsim.
 Wolf (1990), pp. 17-18.
 Ibid., p. 21.
 Ibid., pp. 76-77.
 Ibid., p. 77.
 This to avoid the unwanted side effect that agents with wills are responsible for the colour of their skin (exceptions made by certain modern developments in medicine counted out), their height, size of shoes, or their dribbling in sleep, mumbling when gagged, etc.
 Nagel (1986), pp. 114-115.
 A distinction he first invokes in his paper “Subjective and Objective”, Nagel (1974).
 Nagel (1986), p. 115.
 Ibid., pp. 115-116.
 Ibid., p. 116.
 Ibid., p. 116.
 Ibid., p. 117.
 Wolf (1990), p. 13.
 Ibid., p. 13.
 Ibid., p. 13.
 The sentencefragment: «reach a set of features that must be explained by facts external to the agent, or our explanation will simply come to an end» is pretty close to what Nagel (1986) says on p. 116: «intentional explanations must simply come to an end when all available reasons has been given». This is not very strange given the fact that Wolf tells us that her greatest dept in her work is to Thomas Nagel (vide Preface, p. Ix, in Wolf (1990)).
 Wolf (1990), p. 14. Vide also my account of Nagel above, and p. 118 in Nagel (1986).
 Frankfurt (1971), p. 82.
 Ibid., p. 86.
 In this she is following Gary Watson’s suggestion in “Free agency”, Journal of Philosophy 72 (April 1975), p. 205-220.
 Wolf (1990), p. 49. As Gjelsvik (1995) claims, there are two main answers to the question of what it is to value something; either it is a form of belief, or a form of desire. There are problems connected to both answers. «The problem with the first reply is that it makes it very hard to see how our valuings can motivate us in action. Beliefs are representations of how things are: believing something to be the case does not necessarily amount to having a favourable attitude towards its being the case. The problem with the second reply is that it seems as if we may fail to value what we desire. To think of valuing simply as desiring is therefore seen as difficult or impossible.» (Gjelsvik (1995), p. 1.) Despite this, Gjelsvik takes the view that valuing is a form of desire, although he argues against views on valuing (like Frankfurt’s) which sees it as second-order desire.
Gjelsvik puts up what he calls a filter that shall set valuing apart from other pro-attitudes (like wanting, whishing, desiring, preferring), without an appeal to second-order desires. Most pro-attitudes allow second-level attitudes of the form: if you PRO x, you can PRO that you do not PRO x. This however does not work for values: for, if you PRO x, you cannot PRO that you do not PRO x for a value x. If you value x you may wish or desire that you did not value it, but you cannot value that you did not value x, as you may desire that you did not desire x. This means that if there is any second-order valuing at all, it cannot conflict with first-order valuing. Accordingly we cannot distance ourselves from our values as we can with our desires.
I will not comment on his critique, but just note two things. First, that even though Charles Taylor (who’s views I will later explain) at times can seem to indicate that he thinks of valuing or evaluation as a form of belief (that we believe something to be of value), I think this is in fact not so. It can seem so because his focus is on the cognitive elements of values. Rather he says that something being of value means more than being just desired, it needs something in addition, and this addition is something upheld by language. Secondly, that even though Gjelsvik’s critique seems plausible and devastating for features of Frankfurt’s and especially Lewis’ views, I do not think it interfere with any central arguments against Wolf. The debate seems prima facie periphere also to Taylor’s distinction between a strong and a weak evaluator, but it could be interesting to see what implications this has. One possible point of contact would be where Taylor talks about articulation of evaluations or values. This indicates a close but not a one-to-one connection between desires and values. Gjelsvik’s views on this matter, I think, is not enough developed to give an account of any possible outcome.
 I owe this formulation to Gjelsvik (1995), p. 21.
 Wolf (1990), p. 51.
 Ibid., p. 51.
 Ibid., p. 51.
 Doing something good is necessarily connected grammatically to doing something bad. We will later see that such a judgement needs a strong evaluator, that is, reasons with words like ‘good’ can be given just because the reason-giver knows the distinction between good and bad, and what it means to be a good human being rather than a bad one.
 Ibid., p. 51.
 Nagel (1986), p. 117.
 Vide Wolf (1990), p. 48.
 Wolf (1990), p. 54.
 Ibid., p. 54.
 Ibid., p. 55.
 Ibid., p. 67.
 Ibid., p. 77.
 Vide Wolf (1990), pp. 117-147.
 The conditions of responsibility central to Strawson (1962), of the attitudes and responses upholding a society without there would be no question of responsibility at all, and that the agent possess a potentially effective and intelligent will, are intrinsic to this conception of a responsible agent being able to act according to the True and the Good.
 Taylor (1977), pp. 29-30.
 Ibid., p. 30.
 Ibid., pp. 30-31.
 Taylor (1977), p. 18.
 Ibid., pp. 18-19.
 Gjelsvik (1995) has argued this. «Believing something valuable is not the same thing as valuing it.» (p. 10).
 Gjelsvik (1995) has a distinction that settles his view on valuing. This distinction is between simple desires and preference-desires. Simple desires can be incompatible in that they are just the options we are attracted to, while a preference-desire is the desire for the thing we value most. Accordingly, when we have a first-order simple desire for something, we normally value that thing. Since the logic of valuing (vide ftn. 29) prohibits incompatible valuings of different orders, and since we often have diverging second-order simple desires, they cannot be valuings. In the cases where we have diverging second-order preferences, they express what we value most. The first-order preference in those cases express what we desire most. This is because valuing most can be found on only one of the levels, for if we thought of both levels as valuings, we could value that we do not value x. Therefore what we value most is what we value.
It is difficult to assess what this means for Taylor’s view, but it seems to me that he mostly thinks of valuing as valuing most. In this case where I have to deside what to do, either to be couragous on the right or the left flank, there seems to be a situation with conflicting desires, but actually with no conflicting values. I value courage over cowardice, that is, I desire most to fight because I value most courage. So there seems to be no incompatibility between values at all, just between desires, in this case.
 One could think of a situation where someone were fighting while fleeing, but then the fleeing would be from another fight - one cannot flee from the fight fighting while fleeing.
 On p. 19, Taylor (1977), Taylor gives an example of a weak evaluator that chooses to go to lunch later, although hungry now, because then he shall be able to lunch and swim. He could however be happy to have both worlds: if the pool were open now, he could aussage his immediate hunger as well as enjoying a swim at lunch-time.
 Taylor (1977), p. 19.
 Ibid., p. 25.
 Ibid., p. 32. There is nevertheless something true about radical choice, which Taylor acknowledges. He says: «The real force of the theory of radical choice comes from the sense that there are different moral perspectives, that there is a plurality of moral visions, [. . .] between which it seems very hard to adjudicate» (p. 33). We can find the same point expressed, among others, in Kierkegaard and in Wittgenstein. For Wittgenstein the accepting or rejecting of moral judgements involves taking up a substantive position, and hence no logical or conceptual consideration can but leave the question entirely open. We are facing a radical choice of what moral system to follow, in this meaning, because there are no other criterias outside these systems that can settle the dispute (vide chapter 6 in Johnston (1989)). For Kierkegaard the choice between living an ethical life and living as a christian is not based on external criteria; what can settle the matter is a leap of faith, a substantial choice, or radical choice. But of course both authors describes choices made on strong evaluations.
 Taylor (1976), p. 118.
 Taylor does not argue much for the distinction between these two sense of responsibility, and neither does he argue much here for his naming of this second sense ‘modern’. He develops this view more in his book Sources of the Self: The Making of the Modern Identity, Cambridge 1989.
 Taylor (1977), p. 29.
 Wolf (1990), p. 55.
 Taylor (1977), p. 35.
 Taylor (1976), p. 122. My italics.
 Taylor (1978), p. 219. Articulations are akin to what Dretske refers to as theory-loaded descriptions of behaviour, descriptions that have special implications about the character of the internal cause. Theory-loaded descriptions figure in many disguises, not only as verbs, like Davidson’s verbs implying intentional actions - asserting, cheating, taking a square root, lying (Vide Davidson (1971), p. 45) - and other verbs (among them descriptions of animal behaviour) like stalking, hiding, pretending, asking a question, hunting, avoiding, protecting, threatening; but also as nouns, like a wound - a surgeon does not wound you (at least if he does his job properly) like your worst enemy do, even though the result may look the same. (This last example is from R. Norwood Hanson, Patterns of Discovery, 1958, p. 55; referred to in Dretske (1988), p. 8.)
 Taylor (1977), p. 36.
 Taylor uses the terms ‘reformulation’, ‘redescription’, and ‘interpretation’, in the present sense of the terms, interchangeably.
 By motivation, Taylor means: desires, aspirations and evaluation. Vide Taylor (1977), p. 36. This may indicate that Taylor has not read Watson’s 1975 paper “Free Agency”, inwhere he distinguishes between a valuational system and a motivational system (system of desires), or at least that he does not acknowledge the distinction (in terms). For Taylor I think, motivation is a certain force, naturally connected to what we value most or desire most.
 The constitutive aspect of articulations and the non-descriptive, non-thruth functional aspect of it, reminds me of Wittgenstein’s assessment of the grammar of Äusserungen or avowals in Philosophische Untersuchungen, where he stresses the nature of our interest in them. «The criteria for the truth of the confession that I thought such-and-such are not the criteria for a true description of a process. And the importance of the true confession does not reside in its being a correct and certain report of a process. It resides rather in the special consequences which can be drawn from a confession whose truth is guaranteed by the special criteria of truthfulness.» (Wittgenstein (1953), p. 222.) The critera of truthfulness or sincerity of cource is completely different from the criteria of truth and falsety, and underlines the special roles reasons and other avowals play in our lives, as expressions of the individual’s inner life.
 Taylor (1977), p. 37.
 Vide Taylor (1977b), pp. 75-76.
 Taylor (1977b), p. 48. My italics.
 Taylor (1977b), p. 49.
 Taylor (1977), p. 38.
 Taylor (1976), p. 124.
 Taylor (1976), pp. 124-125. Taylor ties this to our struggle with philosophical problems. Obviously Wittgenstein’s view on philosophy can be cited in support for this connection.
 Nagel (1986), p. 117.
 Ibid., p. 117.
 Taylor (1977), p. 39.
 Wolf (1990), p. 85.
 Ibid., p. 85.
 Taylor (1977), p. 39.
 Wolf (1990), p. 96.
 Ibid., p. 101.
 Ibid., p. 101.
 Ibid., p. 112.
 Ibid., pp. 113-114.
 Ibid., p. 115.
 Vide Dretske (1988), especially the chapters number 3, 4, and 5.
 Davidson (1973), p. 64.
 Vide Wolf (1990), p. 87.
 Vide Wolf (1990), pp. 79-81.