The problem can also be described in terms of freedom. It is because of the reflective character of the mind that we must act, as Kant put it, under the idea of freedom. He says 'we cannot conceive of a reason which consciously responds to a bidding from the outside with respect to its judgments'.³ If the bidding from outside is desire, then the point is that the reflective mind must endorse the desire before it can act on it, it must say to itself that the desire is a reason. As Kant puts it, we must make it our maxim to act on the desire. Then although we may do what desire bids us, we do it freely.

3. Kant, Foundations of the Metaphysics of Morals, p. 448; in Beck's translation, p. 66.

Occasionally one meets the objection that the freedom that we discover in reflection is a delusion. Human actions are causally determined. The philosopher's bugbear, the Scientific World View, threatens once more to deprive us of something we value. When desire calls we think we can take it or leave it, but in fact someone could have predicted exactly what we will do.

But how can this be a problem? The afternoon stretches before me, and I must decide whether to work or to play. Suppose first that you can predict which one I am going to do. That has no effect on me at all: I must still decide what to do. I am tempted to play but worried about work, and I must decide the case on its merits. Suppose next I believe that you can predict which one I'm going to do. You've done it often enough before. What then? I am tempted by play but worried about work, and I must decide the case on its merits.

The worry seems to be that if we were sure we were determined or knew how we were determined then either we could not act or we would not act, or else we would act differently. But why is this supposed to happen? Having discovered that my conduct is predictable, will I now sit quietly in my chair, waiting to see what I will do? Then I will not do anything but sit quietly in my chair. And that had better be what you predicted, or you will have been wrong. But in any case why should I do that, if I think that I ought to be working? Well, suppose that you tell me what you predict I am going to do. If you predict that I am going to work, and I think that I should work, then there is no problem. Or do I now have to do it less freely? If you predict that I am going to play, and I think that I should work, I am glad to have been forewarned. For if I am about to do what I think I have good reason not to do, then a moment of weakness or self-deception must be in the offing, and now I can take precautions against it. And then perhaps I will work after all.

If you are going to tell me what you predict I will do, then your prediction must take into account the effect on me of knowing your prediction, because otherwise it will probably be wrong. Of course it can happen, in a specific kind of case, that knowing the sort of thing I am usually determined to do diminishes my freedom. If I see that I often give in to temptation, I might become discouraged, and fight against it even less hard. But there is no reason to think that this kind of discouragement would be the general result of understanding ourselves better. Or if there is, it must come from some pessimistic philosophy of human nature, not from the Scientific World View If predictions can warn us when our self-control is about to fail, then they are far more likely to increase that self-control than to diminish it. Determinism is no threat to freedom.

Now it will be objected that this is not what philosophers mean when they claim that determinism is a threat to freedom. They aren't talking about a practical problem — that knowledge could somehow take away our freedom — but about a theoretical one — that knowledge would show us we weren't free after all. But how is it supposed to do that? By showing that we could not have done otherwise?

That might show that we aren't responsible.⁴ But it is a different question whether determinism is a threat to responsibility. Freedom is the capacity to do otherwise, not the capacity to have done otherwise. No one has that capacity, because you cannot change the past. That sounds like a joke but I mean it. The freedom discovered in reflection is not a theoretical property which can also be seen by scientists considering the agent's deliberations third-personally and from outside. It is from within the deliberative perspective that we see our desires as providing suggestions which we may take or leave. You will say that this means that our freedom is not 'real' only if you have defined the 'real' as what can he identified by scientists looking at things third-personally and from outside.

4. Actually, I don't think it does. See my 'Creating the Kingdom of Ends: Reciprocity and Responsibility in Personal Relations'.

The point here is the same as the point I made against the argument that reasons are not real because we do not need them for giving scientific explanations of what people think and do. That is not, in the first instance, what we need them for, but that does not show that they are not real. We need them because our reflective nature gives us a choice about what to do. We may need to appeal to the existence of reasons in the course of an explanation of why human beings experience choice in the way that we do, and in particular, of why it seems to us that there are reasons. But that explanation will not take the form 'it seems to us that there are reasons because there really are reasons'. Instead, it will be just the sort of explanation which I am constructing here: reasons exist because we need them, and we need them because of the structure of reflective consciousness, and so on.

In the same way, we do not need the concept of 'freedom' in the first instance because it is required for giving scientific explanations of what people do, but rather to describe the condition in which we find ourselves when we reflect on what to do. But that doesn't mean that I am claiming that our experience of our freedom is scientifically inexplicable. I am claiming that it is to be explained in term of the structure of reflective consciousness, not as the (possibly delusory) perception of a theoretical or metaphysical property of the self.

The Scientific World View is a description of the world which serves the purposes of explanation and prediction. When its concepts are applied correctly it tells us things that are true. But it is not a substitute for human life. And nothing in human life is more real than the fact we must make our decisions and choices 'under the idea of freedom'.⁵ When desire bids, we can indeed take it or leave it. And that is the source of the problem.

5. Kant himself says that 'People who are accustomed merely to explanations by natural sciences' refuse to acknowledge the existence of freedom and its imperatives because 'they are stirred by the proud claims of speculative reason, which makes its power so strongly felt in other fields, to band together in a general call to arms, as it were, to defend the omnipotence of theoretical reason.' Kant, The Metaphysics of Morals, p. 378; in Gregor's translation, pp. i83-184.

'Reason' means reflective success. So if I decide that my desire is a reason to act, I must decide that on reflection I endorse that desire. And here we run into the problem. For how do I decide that? Is the claim that I look at the desire, and see that it is intrinsically normative, or that its object is? Then all of the arguments against realism await us. Does the desire or its object inherit its normativity from something else? Then we must ask what makes that other thing normative, what makes it the source of a reason. And now of course the usual regress threatens. What brings such a course of reflection to a successful end?

Kant, as I mentioned, described this problem in terms of freedom. He defines a free will as a rational causality which is effective without being determined by any alien cause. Anything outside of the will counts as an alien cause, including the desires and inclinations of the person. The free will must be entirely self-determining. Yet, because the will is a causality, it must act according to some law or other. Kant says: 'Since the concept of a causality entails that of laws . . . it follows that freedom is by no means lawless . ..'⁶ Alternatively, we may say that since the will is practical reason, it cannot be conceived as acting and choosing for no reason. Since reasons are derived from principles, the free will must have a principle. But because the will is free, no law or principle can be imposed on it from outside. Kant concludes that the will must be autonomous: that is, it must have its own law or principle. And here again we arrive at the problem. For where is this law to come from? If it is imposed on the will from outside then the will is not free. So the will must make the law for itself. But until the will has a law or principle, there is nothing from which it can derive a reason. So how can it have any reason for making one law rather than another?

6. Kant, Foundations of the Metaphysics of Morals, p. 446; in Beck's translation, p. 65.

Well, here is Kant's answer. The categorical imperative, as represented by the Formula of Universal Law, tells us to act only on a maxim which we could will to be a law. And this, according to Kant, is the law of a free will. To see why, we need only compare the problem faced by the free will with the content of the categorical imperative. The problem faced by the free will is this: the will must have a law, but because the will is free, it must be its own law And nothing determines what that law must be. All that it has to be is a law. Now consider the content of the categorical imperative, as represented by the Formula of Universal Law. The categorical imperative merely tells us to choose a law Its only constraint on our choice is that it has the form of a law. And nothing determines what the law must be. All that it has to be is a law.

Therefore the categorical imperative is the law of a free will. It does not impose any external constraint on the free will's activities, but simply arises from the nature of the will. It describes what a free will must do in order to be what it is. It must choose a maxim it can regard as a law.⁷

7. This is a reading of the argument Kant gives in Foundations of the Metaphysics of Morals, pp. 446-448; in Beck's translation, pp. 64-67; and in Critique of Practical Reason under the heading 'Problem n, p. 29; in Beck's translation, pp. 28-29. It is defended in greater detail in my 'Morality as Freedom'.

(The Sources of Normativity, pp.94-98)

Korsgaard began with a search for moral entities (and values) that might provide the basis for normativity. She concludes that all the accounts of normativity she discusses are "true."

• Normativity must spring from the commands of a legislative authority.
  (Hobbes)

• Reasons are intrinsically moral entities. What we should do when a desire presents itself is to reflect on it, to look at it more objectively to see whether it is such an entity.
  (Clarke, Moore, Nagel)

• Reflection has the power to compel obedience, and to punish us for disobedience. It in turn is bound to govern us by laws that are good. Reflection makes us authorities over ourselves.
  (Hume, Mill, Williams)

• In so far as we have authority over ourselves, we can make laws for ourselves, and those laws will be normative.
  Autonomy is the source of moral obligation.
  (Kant, Rawls, Korsgaard)

Finally, Korsgaard ends with a brilliant criticism of John Mackie's famous 'argument from queerness'.

According to Mackie, it is fantastic to think that the world contains objective values, or intrinsically normative entities. For in order to do what values do, they would have to meet certain impossible criteria. They would have to be entities of a very strange sort. utterly unlike anything else in the universe. The way that we know them would have to be different from the way that we know other sorts of facts. Knowledge of them, Mackie said, would have to provide the knower with both a direction and a motive. For when you met an objective value, according to Mackie, it would have to be — and I'm nearly quoting now — able both to tell you what to do and make you do it. And nothing is like that.

But Mackie is wrong and realism is right. Of course there are entities that meet these criteria. It's true that they are queer sorts of entities, and that knowing them isn't like anything else. But that doesn't mean that they don't exist. John Mackie must have been alone in his room with the Scientific World View when he wrote those words. For it is the most familiar fact of human life that the world contains entities that can tell us what to do and make us do it. They are people, and the other animals.

(The Sources of Normativity, p.166)