Frankfurt's Theory

What is it to be a person? Frankfurt argues that the difference between a person and an animal (human or otherwise) which is not a person is a difference in the structure of will. Frankfurt notes that humans have a characteristic other animals do not: Human persons can form desires for and against particular desires for and against actions. Presumably animals have so-called first order desires, but only human persons can form second order desires.

In order to draw the appropriate distinction, Frankfurt investigates the statement

A wants to x

As Frankfurt points out, "A wants to x" fails to rule out any of (quoting)

a. The prospect of doing x elicits no sensation or intrspectible emotional response in A

b. A is unaware that he wants to x

c. A believes that he does not want to x

d. A wants to refrain from x-ing

e. A wants to y and believes that it is impossible for him both to y and to x

f. A does not "really" want to x

g. A would rather die than x

All of which is to point out that an agent may be unaware of wanting to act, he may be unaware of any feelings about the action in question, he may have conflicting desires, and he may be simply ambivalent. Moreover, the statement that "A wants to x" says nothing of the degree or extent to which A desires to x. Little, then, can be concluded from the statement that "A wants to x".

But there is a distinction of some importance which can be isolated. First, define "first order desire"

Df1: d is one of A's first order desires iff d is expressible by the statement "A wants to x" such that "to x" refers to an action.

Given (a) - (g), it is possible that one of A's first order desires has no bearing on what A actually does. The statement that "A wants to x" in such a case cannot be taken to imply that A's first order desire is motivating A to action. But it may be that "A wants to x" carries with it the implication that A's first order desire is what motivated, or part of what motivated, A to action. Thus,

Df2: d is one of A's first order effective desires iff d is a first order desire and d motivates or will motivate, or is a factor in motivating, A to x.

First order effective desires issue in action. They are the dominant or active desires.

Df3: A's will consists of A's first order effective desires.

Thus A's will is not simply the mereological sum of her desires per se, but rather just those desires which actually motivate her to action. For roughly the same reason, A's will is not coextensive with A's intentions. Moreover, the will is not monolithic: it changes as first order effective desires change. The distinction of interest is that between first order desires which are not effective--i.e., those which do not end up motivating action--and first order desires which are effective--i.e., those which do end up motivating action.

Moving on to second order desires, we have

Df4: D is one of A's second order desires iff D is expressible by the statement "A wants to x" such that "to x" refers to a first order desire.

Corresponding to the distinction between first order desires and first order effective desires, there is a distinction between types of second order desires. In particular

Df5: V is one of A's second-order volitions iff V is one of A's second order desires and "to x" in the statement expressing V "A wants to x" refers to a first order effective desire.

Intuitively, Df5 is the notion that one can form a desire for a (first order) desire without thereby wanting the desire to issue in action. Frankfurt offers the example of a psychologist who treats drug addicts: It might be helpful for her to understand what it is to desire drugs. Thus she might form the second order desire to desire the drugs, without actually wanting to end up taking drugs. On the other hand, one can form second order desires precisely with the idea in mind that the desire wanted be effective: Second order desires of this sort are second-order volitions. Her will is thus partly a function of her second-order volitions.

Based on the distinction between second-order volitions and other second-order desires, Frankfurt distinguishes between persons and wantons.

Df6: If A is a person then, necessarily, A has second-order volitions.

Df7: If A is a wanton then A has no second-order volitions.

"The essential characteristic of a wanton is that he does not care about his will. His desires move him to do certain things, without its being true of him either that he wants to be moved by those desires or that he prefers to be moved by other desires." Extending the notion of a wanton somewhat, an agent A acts wantonly only if they act according to first-order effective desires which are not themselves the result of second-order volitions.

To help make the distinction between person and wanton, Frankfurt provides three cases:

Case 1: The Unwilling Addict

Consider an agent physiologically and psychologically addicted to a drug. The agent struggles desperately and in vain to abstain from taking the drug. In the end, the first order desires which prove effective are those desires which result in her taking the drug. But the agent has second-order volitions to desire to abstain, and indeed has the desire to abstain.

Case 2: The Wanton Addict

Consider an agent just like the first in being physiologically and psychologically addicted to a drug. However, unlike the unwilling addict, the wanton addict has no second-order volitions with respect to desires either for or against taking the drug. The wanton addict is unconcerned one way or the other which first-order desires are effective or 'win out'. Whatever conflict there may be in first order desires is completely up to the first-order desires to sort out.

Case 3: The Willing Addict

Consider an agent just like the first two in being physiologically and psychologically addicted to a drug. Unlike the wanton addict, however, the willing addict forms second-order volitions. But these volitions are to have the desire to take the drug. Were his first order desires to abstain, if he has any, to become effective, the willing addict would do whatever necessary to regain the desire to take the drug.

To the extent that the willing and unwilling addicts form second-order volitions they are persons. But to the extent that the wanton addict fails to form second-order volitions, he is a wanton. More precisely, in not forming second-order volitions with respect to his addiction--either for the desire to abstain or for the desire to take--the wanton addict acts wantonly.

The distinction between person and wanton, based on differences in relevant hiearchies of desires, grounds the difference between having a free will and not. Only persons can be said to have free will, because only persons have second-order volitions. Here's one way to think about it:

A traditional conception of free will has it that a person has free will if they are able to do what they want. But freedom of will is not freedom of action: they are distinct concepts. Freedom of action, presumably, is physically unrestrained or physically uncoerced action. Freedom of the will, on the other hand, is the freedom an agent has to want what he wants to want. That is to say, an agent has a free will to the extent that her will--her first-order effective desires--are a (partial) function of her second-order volitions. "It is in securing the conformity of his will to his second-order volitions, then, that a person exercises freedom of will. And it is in the discrepancy between his will and his second-order volitions, or in his awareness that their coincidence is not his own doing but only a happy chance, that a person who does not have this freedom feels its lack"

The unwilling addict fails to have a free will by virtue of the fact that the first-order effective desires she has (= her will) is not the will she wants. She is powerless to effect a change in her own will.

The wanton addict fails to have a free will vacuously, since he has no desires one way or the other concerning his will.

The willing addict fails to have a free will, despite the fact that his will is precisely what he wants--has the second-order volition for-it to be. For the willing addict it is merely a happy circumstance that his will is as his second-order volition would have it. But his will is a reflection of his second-order volitions: It is not a function of those volitions. The willing addict is like the unwilling addict in having no control over his will.

Df8: An agent A has a free will iff A's first-order effective desires are a (partial) function of her second-order volitions.

Frankfurt's theory of free will has, he thinks, the virtue of being able to explain

a. Why persons value free will, and

b. Why humans ordinarily refuse to ascribe free will to animals.

We value free will for the simple reason that having it implies the satisfaction of many more desires than not having it. We refuse to ascribe free will to animals since we do not find that animals have second-order volitions: they are wantons; unable to want to have a will different from the one they happen to have.

Other theories do not enjoy this explanatory advantage. Chisholm's theory of agent causation, for instance, is unable to provide any justification for the thought that animals fail to have free wills, since it fails to show what is peculiar to persons such that they have free will and others do not. Nor does it explain what's so great about having a free will, since it offers no explanation of the mechanism--to juxtapose the most paradoxical of phrases--of free will by proposing that agent causation is, at root, mysterious.

Lastly, Frankfurt suggests that his theory of free will is neutral with respect to determinism. Nothing in his account rules out determinism, but neither does anything in his account rule out indeterminism. It's possible for the agent to be caused to want the will that he wants, just as it's possible for the agent to want the will that he wants by chance, as it were.