I've posted an excerpt from section 3.5.4 of Christine Korsgaard's Locke Lectures after the jump.
In Kantian moral psychology, the starting point for action is what Kant calls an incentive (Triebfeder). An incentive is a motivationally loaded representation of an object. I am using the term “object” broadly here to include not only substances but also states of affairs and activities. The object may be actually perceived, or conceived as a possible item in the environment, a way that things might be. You are subject to an incentive when you are aware of the features of some object that make the object attractive or appealing to you. Perhaps the object satisfies one of your needs; or perhaps because of the nature of your species or your own particular nature the object is one you are capable of enjoying. It interests you, it arouses the exercise of your faculties, it excites your natural curiosity, or it provides some sort of emotional comfort or satisfaction. It doesn’t matter what – something about you makes you conceive this object as appealing or welcome in a particular way. The object answers to something in you or to the condition you are in. Incentives can also be negative. You may represent an object to yourself as painful or threatening or disgusting, or in some other way unwelcome.
Incentives operate on animals causally but they do not directly cause the animal’s movements. If an incentive directly caused the animal’s movement, it would be something within the animal, not the animal as a whole, that determined the movement, and then as we have seen it would not be a case of action. A desire for food, after all, can cause you to salivate. If it also could cause you to go to the refrigerator, then salivating and going to the refrigerator would equally be actions. If we are to count a movement as an action, the movement must be caused by the animal itself, not by its representations or perceptions. Instead, an incentive works on an animal by making some movement or response seem appropriate to it, by presenting it as a thing to do. According to Kant, incentives work in conjunction with principles, which determine (or perhaps I should say describe) the agent’s responses to those incentives, responses which are guided by the agent’s conception of the world. The principle represents what Aristotle calls the agent’s “contribution” to the action, the thing needed to make it voluntary. Every action must involve both an incentive and a principle: that is, something is presented to the animal’s consciousness, on which it then acts.
In the human case, in the case of a person, it is easy to say what makes action different from mere response, for human beings act on reasons. A person’s principles determine what the person counts as a reason. To the extent that the person determines himself to intentional movement, he takes his desire for food to provide him with a reason for going to the refrigerator; and that is not the same as its directly causing him to go to the refrigerator. We may represent this fact – his own causality or self-determination – by saying that it is his principle to get something to eat when he feels hungry, at least absent some reason why not.