-- The Determinism and Freedom Philosophy Website --

If you were free in doing something and morally responsible for it, you could have done otherwise. That has seemed a pretty firm proposition among the old, new, clear, unclear and other propositions in the philosophical discussion of freedom and determinism. If you were free in what you did, there was an alternative. It is also at least natural to think that if determinism is true, you can never do otherwise than you do. G. E. Moore, that Cambridge reasoner in whose shadow Wittgenstein ought to be standing, considered the matter. He pointed out that even if determinism is true, there remains a sense in which you can still do otherwise than you do: you will do otherwise if you so choose. That, on reflection, is consistent with determinism. The doctrine of the compatibility of freedom and determinism is saved. Joseph Keim Campbell, strong philosopher at Washington State University, provides the latest thinking on this seemingly unavoidable dispute. You do not have to agree that either compatibilism or incompatibilism must be true in order to appreciate the carefulness of his reasoning in this piece of ongoing American philosophy. It requires and repays attention.

1. Introduction*

This is a defense of strong compatibilism. Roughly, strong compatibilism is the view that (a) free will is essential to moral responsibility, (b) free will requires alternatives, and (c) moral responsibility is compatible with determinism. The expression ‘free will’ is intended to designate the freedom-relevant condition necessary for moral responsibility, whatever that condition may be (cf. McKenna 2003).1 Thus, (a) is true by definition. Strong compatibilism is a version of the alternatives theory of free will. According to the alternatives theory, a person S performs an action a freely only if S has or had alternatives to a. Moreover, S has alternatives to a iff S can perform some contrary action a' instead of a, where one act is contrary to another provided that a person cannot simultaneously perform them both.

Strong compatibilism has long been challenged by incompatibilists who endorse (b) but reject (c). More recently, both semicompatibilists—who accept (c) but reject (b)—and free will nihilists—who hold that no person has or ever had free will—have offered criticisms of (b). This line of attack is motivated by well-known examples offered first by Harry Frankfurt (1969) and then by others. Here is a version of my own Frankfurt example, as re-told by Michael McKenna.

“Desperate for money, Eleanor and her father Roscoe plan to rob a bank. Roscoe fears that Eleanor might change her mind at an inopportune moment. To insure that Eleanor will proceed with the plans, Roscoe secretly implants a mechanism in Eleanor’s brain. Should Eleanor give any indication that she is unwilling to go along with the bank robbery, Roscoe will use the device to render Eleanor unable to do anything other than rob the bank. As it happens, despite a splitting headache, Eleanor willingly robs the bank with her father. The device is never activated.” (McKenna 1998, 259)

Eleanor is apparently morally responsible for her actions even though she could not have done otherwise. Thus, it seems that alternatives are not essential to moral responsibility.

In this essay, I provide a new theory of alternatives, inspired by the work of G.E. Moore (1912) and R. Jay Wallace (1994). I begin with a broad framework for grasping most versions of the alternatives theory: the relevant facts account (§2). The relevant facts account is helpful in understanding both compatibilist and incompatibilist theories but I am especially interested in using it to flesh out two varieties of strong compatibilism: the two-‘cans’ view—according to which ‘can’ is ambiguous like the word ‘odd’—and contextualism—according to which ‘can’ is context sensitive like ‘flat.’ Finally, I offer a version of the two-‘cans’ view and show how my view differs from those of Moore and Wallace (§ 3).

I make several assumptions throughout. Many philosophers hold these assumptions and none of them is particularly relevant to the truth or falsity of strong compatibilism. If one provides a version of the alternatives theory that is compatible with both determinism and the following assumptions, then one has provided a version of the alternatives theory that is compatible with determinism. The assumptions do not beg any substantive questions in the problem of free will and determinism but they allow my presentation to run more smoothly.

First, I assume that there is a freedom-relevant condition that is necessary for moral responsibility (Fischer and Ravizza 1993 and 1998). This claim is to be understood in the following way: a person S is morally responsible for an event e only if either (i) e is an action and S does e freely, or (ii) e is the consequence of an action a, and S does a freely. As I see it, we are primarily responsible for our actions and other events, namely, those events that are the consequences of our actions. Some people believe that we are primarily responsible for our choices but choices are just kinds of actions. We can extend talk of moral responsibility to propositions, as well as events, for some propositions are made true by the occurrence of certain events (van Inwagen 1975 and 1983, Lewis 1981, Perry 2004). Thus, if S were to raise his right hand, he would thereby make it the case that S raises his right hand. Nonetheless, the proposition that S raises his right hand—if it is true—was true long before S made it the case, for propositions are eternally true or false. Propositions do not become true or false when they are made true or false by the occurrence of certain events (Perry 2004).

Second, I assume, without argument, the modified Strawsonian view of moral responsibility: S is morally responsible for an event e (or proposition p) only if S is praiseworthy or blameworthy for bringing about e (or for making it the case that p). Contra P.F. Strawson (1962), that S is praised or blamed is not enough to ground his moral responsibility, even if S is praised or blamed by all of the members of his moral community. In addition, S must be worthy of praise or blame (Fischer and Ravizza 1993, 16-8, and 1998, 6).

2. The Relevant Facts Account2

I begin by restating strong compatibilism in more precise terms, e.g., as the conjunction of the following three claims.

1.        A person is morally responsible for an action only if he does it freely.

2.        A person performs an action freely only if he has or had alternatives to the action.

3.        That someone is morally responsible for some action is consistent with the thesis of determinism.

(1)–(3) correspond, roughly and respectively, to items (a)–(c) in the provisional definition of ‘strong compatibilism’ above.3

The principle of alternative possibilities, or PAP, is a consequence of (1) and (2). It is more succinctly stated as follows:

PAP: A person is morally responsible for an action only if he has or had alternatives to the action.

Note that PAP fails to distinguish between the time of action—the time that an agent performs an action—and the time of ability—the time that an agent had the ability to perform an alternative action (Taylor 1965 and Lehrer and Taylor 1965). Even if we agree that Eleanor could not do otherwise at the time that she robbed the bank there is little reason to suppose that she lacked alternatives to her actions prior to the implantation of the bank-robbing device. Perhaps we should rephrase PAP so that it implies that in order for a person to be morally responsible it is “necessary that one could have done otherwise for some acts in one’s lifetime” (Kane 2002, 697).

Unfortunately the above Frankfurt example can be altered to meet this new version of PAP. Suppose that some god-like creature—call him ‘Harry’—has a full life plan for Eleanor. He implants a device in Eleanor prior to her birth. If Eleanor were about to something that Harry did not intend for her to do, then the device would become activated. As it happens, she does everything that Harry wants her to do. Now it seems that there is nothing that Eleanor could have done otherwise. Similar comments may be made about other responses to the Frankfurt examples (cf. Pereboom 2000). For every new version of PAP, there is another Frankfurt example that apparently undermines it. In order to successfully defend PAP one needs to adopt a new strategy. What we need is not a different version of PAP but an altogether different understanding of alternatives.

The alternatives theory claims that if S does a freely, then S can, or could have, done otherwise. According to this theory, an understanding of the concept of free will rests on a proper understanding of ability sentences. An ability sentence is any sentence of one of three equivalent forms: ‘S can do a,’ or ‘S is able to do a,’ or ‘It is within S’s power to do a’ (van Inwagen 1983, 8; Unger 1984, 55). The terms ‘can,’ ‘able,’ and ‘power’ are ability terms. Most versions of the alternatives theory are also versions of the relevant facts account. According to this account, ‘S can do a’ means ‘S’s doing a is compossible with the relevant facts’ (Lewis 1976, Unger 1984, Sider 1997, and Hawthorne 2001). Particular instances of the relevant facts account differ depending on (a) the facts that are deemed relevant as well as (b) the factors that determine (a). It is debatable whether every version of the alternatives theory is a version of the relevant facts account. Nonetheless, the relevant facts account provides a framework with which we can better understand and evaluate versions of the alternatives theory.

The relevant facts account is well suited for grasping the standard incompatibilist view of alternatives. What does it mean to say that William James has a choice about which way to walk home after one of his lectures? Here is James’s response:

“It means that both Divinity Avenue and Oxford Street are called; but only one, and that one either one, shall be chosen. Now I ask you to seriously suppose that this ambiguity of my choice is real; and then to make the impossible hypothesis that the choice is made twice over, and each time falls on a different street. In other words, imagine that I first walk through Divinity Avenue, and then imagine that the powers that be annihilate ten minutes of time with all that it contained, and set me back at the door of this hall just as I was before the choice was made. Imagine then that, everything else being the same, I now make a different choice and traverse Oxford Street.” (James 1948, 44)

The broad past is “the past together with the laws of nature” (Finch and Warfield 1998, 523). On the most intuitive reading of this passage, James’s suggestion is that having alternatives is having the ability to do otherwise given the broad past. More formally, let t be some time prior to S’s choice—for instance, some time prior to the annihilation of time noted in the above quotation—and let Yt be the set of propositions about the broad past relative to t. According to James, S can (at t) do a only if S’s doing a is compossible with Yt. Call this the ‘incompatibilist criterion.’ Call the associated ability, ‘all-in ability’: the ability to perform an action that is compossible with the entire set of facts about the broad past.

Note that the incompatibilist criterion is not an analysis since it only specifies a necessary condition for having alternatives. Nonetheless, it does identify a condition that is central to most versions of the argument for incompatibilism. As Peter van Inwagen writes: “it seems that our freedom can only be the freedom to add to the actual past; it seems that our freedom can only be the freedom to act in accordance with the laws of nature” (2000, 167; cf. Ginet 1990, 102-3). Hence, the incompatibilist claims that, given determinism, we lack the all-in ability to do otherwise. Of course, the incompatibilist believes that all-in ability is the only kind of ability, at least when considering the ‘can’ of free will.

For this reason, the incompatibilist adopts an invariantist theory about assertions of ability sentences. Invariantism is the denial of contextualism and contextualism is the view that the truth conditions of ability sentences vary according to the context in which those sentences are uttered (Unger 1984; cf. DeRose 1992, 492). Contextualist theories are currently in vogue in epistemology, particularly in the analysis of ‘know’ as it occurs in sentences like ‘I know that I have a head.’ But contextual analyses of ‘can,’ ‘free action,’ and other terms relevant to the free will debate have also been put forth.4

David Lewis offers a contextualist analysis of ‘can’ in response to a paradox about time travel: the grandfather paradox.5 In Lewis’s example, Tim travels back in time in an attempt to kill Grandfather prior to the birth of Tim’s parents. He fails to kill Grandfather (since doing so would lead to a contradiction) but we may ask, could Tim have done otherwise? Lewis writes:

“We have this seeming contradiction: ‘Tim doesn’t, but can, because he has what it takes’ versus ‘Tim doesn’t, and can’t, because it’s logically impossible to change the past.’ I reply that there is no contradiction. Both conclusions are true, and for the reasons given. They are compatible because ‘can’ is equivocal.” (1976, 77)

Here is another helpful quotation from Ted Sider.

“Lewis’s idea is that a statement attributing ability, like ‘Tim can kill Grandfather,’ is ambiguous. The statement means ‘Tim’s killing Grandfather is compossible with a certain set of facts,’ but the relevant set of facts may vary from one context of utterance to another. When we say that Tim can kill Grandfather because he has what it takes, we mean that his killing Grandfather is compossible with a certain set of facts that includes only relatively ‘local’ facts about the killing situation; when we say that Tim can’t kill Grandfather because Grandfather is Tim’s grandfather, we mean that Tim’s killing Grandfather isn’t compossible with a more inclusive set of facts that includes the fact that Grandfather survived his youth and helped produce Tim.” (1997, 143)

According to both Lewis and Sider, the truth conditions of ability sentences vary according to the context in which those sentences are uttered. Thus, an utterance of the sentence ‘Tim can kill Grandfather’ may be true in one context yet false in another.

In the above quotations, Lewis and Sider suggest that the variance in the truth conditions of ability sentences is due to a change in meaning. Lewis claims that “‘can’ is equivocal” (1976, 77) and Sider claims that “a statement attributing ability ¼ is ambiguous” (1997, 143). Both assertions are questionable and unnecessary. If contextualism is understood in terms of the relevant facts account, then it seems that ‘S can do a’ always means the same thing: that S’s doing a is compossible with the relevant facts. What varies from context to context—if contextualism is true—is the set of facts that is counted as relevant, not the meaning of ‘can.’ Contextualism is committed to a variation in the content of assertions of ability sentences—to the “proposition that embodies [their] truth-conditions”—but not to a variation in the meaning of such assertions—to “what is fixed by the conventions for the use of expressions that we learn when we learn a language” (Perry 1997).

We may contrast contextualism with the two-‘cans’ view endorsed by Moore (1912). Moore distinguishes between the hypothetical sense and the categorical sense of ‘can.’ The latter is picked out by the concept of all-in ability noted above; the former is usually identified with the standard hypothetical analysis of ‘could have done otherwise’: if S had wanted (or tried, etc.) to do otherwise, then S would have done otherwise.6 Perhaps the standard hypothetical analysis is really a version of the relevant facts account. If it is, then Moore may be wrong in his claim that the word ‘can’ is ambiguous. Nonetheless, Moore holds that ability terms and sentences have two distinct meanings and in doing so he endorses the two-‘can’ view and distinguishes his position from the one held by the contextualist.7

The contextualist about ability terms adds to the relevant facts account the claim that the relevant facts vary with the context of utterance. There are many different ways that one might explain how and why this is so. According to a popular version of contextualism—call it ‘naïve contextualism’—these variations result due to differences in our sphere of attention. John Hawthorne writes:

“When ordinary speakers utter English claims of the form ‘S did x freely’ (and their synonyms), they frequently speak the truth. But when our sphere of attention is widened by philosophical inquiry, we are rarely in a position to truly utter the English words ‘S did x freely’. Accordingly, the English words ‘S did x freely’ (and ‘It is up to S whether or not he does x’ and ‘S did x of his own free will’ etc.) must have a meaning that somehow allows its truth conditions to vary according to the sphere of attention.” (2001, 68)

Naïve contextualism is a kind of compatibilism, for it claims that in ordinary contexts our standards for determining what facts are relevant are less restrictive since we attend to only “relatively ‘local’ facts.” Yet in philosophical contexts our attention is draw elsewhere—to facts about the neurological springs of our action, or to the entire set of facts about the broad past—and we subsequently deny that our actions are free.

It is important to realize that naïve contextualism is a compatibilist theory even though it endorses a thesis similar to the incompatibilist criterion. Suppose that S does a. Could S have done otherwise? According to the naïve contextualist, in ordinary contexts the answer is ‘Yes’ since in those contexts we are only attentive to the local facts and S’s doing otherwise is compossible with that set of facts. In this case, even if determinism is true and S’s doing otherwise is not compossible with the set of facts about the broad past, it is still the case that S has or had alternatives. Since it concedes that the same act may be both free and determined, naïve contextualism is a compatibilist theory. That his actions are determined is not enough to render an agent unfree but if our attention shifts to a wider set of facts—e.g., to one that includes the entire broad past—then the context changes, as well, and in this new context an utterance of the same ability sentence comes out false. Hence, the naïve contextualist accepts a version of the incompatibilist criterion: If S’s doing a is not compossible with the broad past and these facts are not properly ignored, then S cannot do a (cf. Hawthorne 2001, 74).

3. Compatibilist Alternatives

In this section, I sketch out a compatibilist theory of alternatives that is a variant of the two-‘cans’ view. My version of the alternatives theory is opposed to naïve contextualism but my criticisms of naïve contextualism leave room for other contextualist theories as well as other non-contextualist theories. This is all to the good. I am not certain whether ability sentences are genuinely ambiguous or whether they suffer from some more complex linguistic malady. Since it is relatively easy to talk in terms of different senses of ability sentences, I make a provisional case for the claim that the two-‘cans’ view is correct. If it turns out that ‘can’ is more like ‘flat’ than it is like ‘odd,’ or even that the relevant difference involves the contents of assertions of ability sentences rather than the meanings of those sentences (cf. Stainton forthcoming), then my comments may be amended without much loss to the strong compatibilist position.
Central to my presentation here is the distinction between two different senses of ability terms and sentences: the all-in ability sense and the general ability sense. It might be better to talk about a distinction between all-in abilities and general abilities, where S has an all-in ability to do something iff the appropriate ability sentence is true in the all-in ability sense, and S has a general ability to do something iff the appropriate ability sentence is true in the general ability sense. 

Often when we say ‘S can doing otherwise’ we mean that S’s doing otherwise is compossible with all of the facts about the broad past. It is our all-in ability that withers upon reflection, for in this sense any fact about the broad past is relevant to whether or not one has alternatives. As we discover and contemplate more information about the world, our all-in ability to do otherwise appears to correspondingly evaporate. All-in ability is exemplified in both the James thought experiment and the incompatibilist criterion, noted above.

Ability sentences may also express a more general sense. In order to understand the difference, consider first the following joke. (It is a philosophical joke, so it’s not very funny.) A man has a bird in a cage. A friend asks him, “Can your bird fly?” The man looks at the bird in the cage and responds, “No. Not at the moment.” Whereupon the friend retorts, “If it can’t fly, then why is it in a cage?”  Peter Unger (1984, 55) offers a related example: “while riding in a train with a pianist friend, a person might ask the musician, ‘Can you play “One O’clock Jump”?’ The pianist may reply, ‘Yes, I can.’” Unger notes that “the lack of any piano on the train will not falsify the musician’s claims” but that “in the hotel two months later, matters of truth will be evaluated differently: The absence of piano might then falsify.”

Given the above examples, it is natural to conclude that ability sentences are ambiguous. A bird is in a cage and someone asks, ‘Can the bird fly?’ The answer is ‘Yes’ if we are talking about the bird’s general abilities. The bird would not be in a cage unless it had the general ability to fly. But the bird does not have the all-in ability to fly, for it cannot fly in the all-in sense given that it is in a cage. Similarly, a man might have the general ability to play ‘One O’clock Jump’ even if there is no piano available. When there is no piano available it is not this general ability that is lost, though indeed another ability may be absent. For there is certainly a sense in which one cannot play a song on a piano if there is no piano available to play the song.

My distinction between all-in and general abilities is influenced by the work of Moore (1912) and Wallace (1994) but it is not the same as either of their distinctions. All-in abilities correspond well with Moore’s categorical sense of ‘can’ but general abilities are quite different from the hypothetical sense of ‘can.’ With Wallace it is just the opposite. My notion of general abilities is adopted from Wallace’s own view. But all-in abilities are different from Wallace’s particular abilities. Wallace’s distinction between particular and general abilities is a lot like the difference between one’s general ability to play the piano—in my sense—versus one’s ability to play the piano at a particular moment of time, for instance, one’s ability to play the piano now. But when I say that a person has the general ability to play ‘One O’clock Jump’ I mean that there is a sense in which he can play ‘One O’clock Jump’ now, even if there is no piano available. At this very moment—regardless of the presence or absence of a piano—there are lots of folks who can play ‘One O’clock Jump’ in the general sense though I cannot. Nonetheless, I am willing to admit that there might be another, more restrictive sense in which one cannot play ‘One O’clock Jump’ if there is no piano available to play the song.

In order for Wallace’s view to be of use to the strong compatibilist it must be amended and combined with the insights of Moore. Moore is correct: ‘can’ is ambiguous and, thus, so are ability sentences. However, Moore was wrong about the details of this ambiguity. Wallace’s concept of general ability is preferable to Moore’s hypothetical sense of ‘can.’ The new proposal combines the theories of Moore and Wallace: ability sentences are ambiguous between the all-in sense and the general sense.

Given my version of the two-‘cans’ view, how should the strong compatibilist respond to the Frankfurt example noted above? Recall that in this example Eleanor is supposed to be morally blameworthy for her action. Suppose that we slightly alter the example: Eleanor decides not to rob the bank and Roscoe simply flips a switch causing her to do so anyway. In this new case Eleanor is no longer morally blameworthy for robbing the bank. What is the difference between the two cases, the original case where the device is not activated and the new case where it is activated? One cannot describe the difference in terms of Eleanor’s all-in abilities since in both situations she lacks the all-in ability to do otherwise. The difference lies in the things that she cannot do in a more general sense. Though the Frankfurt examples show that all-in abilities are not essential to moral responsibility, it would be wrong to conclude from those examples that general abilities are not essential to moral responsibility.

General abilities are more fundamental than all-in abilities. If a bird lacks the general ability to fly, then it also lacks the all-in ability to fly. I cannot play the piano in a general sense, so I am not able to play it in the all-in sense. But as the above examples indicate, a bird may have a general ability to fly yet lack the all-in ability to do so. Similar comments apply to the pianist example. It is the fundamental nature of our general abilities that make them appropriate candidates for underlying the freedom-relevant condition necessary for moral responsibility. As the Frankfurt examples illustrate, one may be praiseworthy or blameworthy for his actions even if he lacks the all-in ability to do otherwise but few would say the same were one to lack general abilities like “the power to grasp and apply the principles that support the moral obligations we accept, and to control one’s behavior by the light of such principles” (Wallace 1994, 188). That Eleanor cannot do otherwise in the all-in sense is irrelevant to our judgment that she is morally blameworthy but if she could not do otherwise in a more general sense—as is the case when the bank-robbing device is activated, for instance—our judgment would rightfully be different.8

One might think that Unger’s pianist example is supportive of naïve contextualism but that is not so. There is a general sense in which any pianist can play the piano even if he is riding in a train and there is no piano available to play. This is why it is appropriate to call one a ‘pianist’ or a ‘musician’ whether or not there is an instrument available to play. A pianist’s general ability to play ‘One O’clock Jump’ has to do with his training and “his knowledge of the jazz repertoire” (Unger 1984, 55) not with incidental facts like the availability of a piano. Even if one believes that there is no piano available to play he might still say that he is able to play ‘One O’clock Jump’ in this general sense. In one and the same context he might admit both that he can play ‘One O’clock Jump,’ for he has what it takes, and that he cannot play the piece, since there is no piano available. This is why naïve contextualism is wrong.

Similar comments apply to the bird example. By merely asking ‘Can the bird in the cage fly?’ one automatically draws attention to the fact that the bird is in a cage. This would make this fact a relevant one if naïve contextualism were true. If naïve contextualism were true, it would be impossible to correctly answer ‘Yes’ to the question ‘Can the bird in the cage fly?,’ for the bird cannot fly in the all-in sense given that it is in a cage. Suppose that someone wants to take the bird out of the cage and hold it in his hands. He might ask, ‘Can the bird in the cage fly?’ before doing so. On some occasions the correct answer to this question is going to be ‘Yes.’ It is not clear how the naïve contextualist may admit this.

For these reasons, I am inclined to believe that naïve contextualism is false: the content of ability sentences does not change simply because of a change in our sphere of attention. This alone, though, does not entail either that contextualism is false or that the two-‘cans’ view is true. But it does suggest that merely attending to a fact does not in and of itself make the fact a relevant one. As I indicated above, we should separate the semantic issues—what to say about the precise meaning and content of ability sentences—from the compatibility issue. I think that I have given reason to believe that (a) on at least some occasions there are truthful utterances of ability sentences, (b) often the truthfulness of those utterances is essential to the moral praiseworthiness or blameworthiness of an individual’s action, and (c) all of this is independent of the truth or falsity of determinism. This is enough to motivate strong compatibilism even though it is consistent with the two-‘can’ view, contextualism, and a few other theories about the semantics of ability sentences (cf. Stainton forthcoming).

In a preface to his own reprinted article supporting free will nihilism, Derk Pereboom writes: “Even if we are never morally responsible for our actions, and even if we adopt this position, it still makes sense to pursue projects that give our lives meaning, we can still have adequate reason to do what is right, and we can still participate in good interpersonal relationships” (1994, 242; my emphasis). Elsewhere, Pereboom (2000) argues against the alternatives theory by claiming that in Frankfurt examples persons cannot do otherwise yet are, intuitively, morally responsible for their actions. Ted Honderich argues in favor of determinism and against compatibilism yet claims that even if determinism is true “we can persist in certain responses to the desires and intentions of others” (1988, 533; my emphasis). In support of free will nihilism, Galen Strawson writes: “whenever we are able to give a true full rational explanation of an action, that action, at least, is not free” (1986, 26; my emphasis). In each case there a reference to things that we are able to do even though we may be unable to do anything in the all-in sense. I maintain that the reference is to a general ability that is more significant and more fundamental than the all-in ability that these writers believe is either incompatible with determinism or irrelevant to issues of moral responsibility.



Campbell, Joseph Keim. 1997. “A Compatibilist Theory of Alternative Possibilities.” Philosophical Studies 88: 319-30.

———. 1999. “Descartes on Spontaneity, Indifference, and Alternatives.” New Essays on the Rationalists, edited by Rocco J. Gennaro and Charles Huenemann. New York and Oxford: Oxford University Press.

———. Forthcoming. “On the Impossibility of Time Travel.” Paradoxes of Time Travel: A Philosophical Reader, edited by Joseph Keim Campbell. Calgary: Broadview Press.

Campbell, Joseph Keim, Michael O’Rourke, and David Shier, eds. 2004a. Freedom and Determinism. Cambridge, Mass.: The MIT Press.

———. 2004b. “Freedom and Determinism: A Framework.” In Campbell, O’Rourke, and Shier 2004a.

DeRose, Keith. 1992. “Contextualism and Knowledge Attributions.” Philosophy and Phenomenological Research 52: 913-29.

———. 1999. “Introduction: Responding to Skepticism.” Skepticism a Contemporary Reader, edited by Keith DeRose and Ted A. Warfield. Oxford: Oxford University Press.

Feldman, Richard. 2004. “Freedom and Contextualism.” In Campbell, O’Rourke, and Shier 2004a.

Finch, Alicia, and Ted Warfield. 1998. “The Mind Argument and Libertarianism.” Mind 107, 515-28.

Fischer, John Martin, ed. 1986. Moral Responsibility. Ithaca: Cornell University Press.

———. 1994. The Metaphysics of Free Will: An Essay on Control. Oxford: Blackwell Publishing.

Fischer, John Martin, and Mark Ravizza, 1993. “Introduction.” Perspectives on Moral Responsibility, edited by John Martin Fischer and Mark Ravizza. Ithaca: Cornell University Press.

———. 1998. Responsibility and Control: A Theory of Moral Responsibility. Cambridge, U.K.: Cambridge University Press.

Frankfurt, Harry G. 1969. “Alternate Possibilities and Moral Responsibility.” Journal of Philosophy 45, 829-39. Reprinted in Pereboom 1997 and Fischer 1986.

Ginet, Carl. 1990. On Action. Cambridge, U.K.: Cambridge University Press.

Hawthorne, John. 2001. “Freedom in Context.” Philosophical Studies 104, 63-79.

Honderich, Ted. 1988. A Theory of Determinism. Oxford: Clarendon Press.

James, William. 1948. “The Dilemma of Determinism.” Essays in Pragmaticism. New York: Hafner Publishing Company.

Kane, Robert. 2002. “Responsibility, Reactive Attitudes and Free Will: Reflections on Wallace’s Theory.” Philosophy and Phenomenological Research 64, 693-698.

Lehrer, Keith and Richard Taylor. 1965. “Time, Truth and Modalities.” Mind 74, 390-98.

Lewis, David. 1976. “The Paradoxes of Time Travel.” American Philosophical Quarterly 13, 145-52. Reprinted in Lewis 1986. (Page references are to the latter version.)

———. 1979. “Score Keeping in a Language Game.” Journal of Philosophical Logic 8: 339-59. Reprinted in Lewis 1983.

———. 1981. “Are We Free to Break the Laws?” Theoria 47: 113-21. Reprinted in Lewis 1986.

———. 1983. Philosophical Papers, Volume I. Oxford: Oxford University Press.

———. 1986. Philosophical Papers, Volume II. Oxford: Oxford University Press.

Markosian, Ned. 1999. “A Compatibilist Version of the Theory of Agent Causation.” Pacific Philosophical Quarterly 80: 257-77.

McKenna, Michael S. 1998. “Does Strong Compatibilism Survive Frankfurt Counter-Examples?” Philosophical Studies 91, 259-64.

———. 2003. “Neo’s Freedom . . . Whoa!” Matrix Reloaded Website:

Moore, G.E. 1912. Ethics. New York: Oxford University Press.

O’Connor, Timothy, ed. 1995. Agents, Causes, and Events: Essays on Indeterminism and Free Will. Oxford: Oxford University Press.

Pereboom, Derk. 1994. “Determinism al Dente.” Nous 29: 21-45. Reprinted in Pereboom 1997. (Page references are to the latter version.)

———, ed. 1997. Free Will. Indianapolis: Hackett Publishing Company.

———. 2000. “The Explanatory Irrelevance of Alternative Possibilities.” In Tomberlin 2000.

Perry, John. 1997. “Indexicals and Demonstratives.” A Companion to the Philosophy of Language, edited by Bob Hale and Crispen Wright. Oxford: Blackwell Publishing.

———. 2004. “Compatibilist Options.” In Campbell, O’Rourke, and Shier 2004a.

Reid, Thomas. 1983. Inquiry and Essays, edited by Ronald E. Beanblossom and Keith Lehrer. Indianapolis: Hackett Publishing Company.

Sider, Ted. 1997. “A New Grandfather Paradox?” Philosophy and Phenomenological Research 57, 139-44.

Stainton, Robert J. Forthcoming. “Contextualism in Epistemology and the Context Sensitivity of ‘Knows’.”

Strawson, Galen. 1986. Freedom and Belief. Oxford: Clarendon Press. Chapter 2 is excerpted in O’Connor 1995. (Page references are to the later version.)

Strawson, P.F. 1962. “Freedom and Resentment.” Proceedings of the British Academy 48: 1-25. Reprinted in Pereboom 1997.

Taylor, Richard [aka Diodorus Cronus]. 1965. “Time, Truth and Ability.” Analysis 25, 137-41.

Tomberlin, James E., ed. 2000. Action and Freedom: Philosophical Perspectives 14. Boston and Oxford: Blackwell Publishing.

Unger, Peter. 1984. Philosophical Relativity. Minneapolis: University of Minnesota Press.

van Inwagen, Peter. 1975. “The Incompatibility of Free Will and Determinism.” Philosophical Studies 27: 185-199. Reprinted in Pereboom 1997.

———. 1983. An Essay on Free Will. Oxford: Clarendon Press.

———. 2000. “Free Will Remains a Mystery.” In Tomberlin 2000.

Wallace, R. Jay. 1994. Responsibility and the Moral Sentiments. Cambridge, Mass.: Harvard University Press.



* Versions of this paper were presented at the Bellingham Summer Philosophy Conference as well as the Western Canadian Philosophical Association Conference. I thank my commentators—David Sosa and Peter Murphy, respectively—as well as other members of the audience—especially Hud Hudson and David Zimmerman—for helpful criticisms and feed-back. I also thank John Martin Fischer, Michael O’Rourke, and an anonymous referee from the Canadian Journal of Philosophy for written comments on an earlier draft. Lastly, I thank David Shier and Harry Silverstein for useful discussions on issues related to this paper.

1. Following Peter van Inwagen (1983, 8), I use the term ‘free will’ out of respect for tradition. By saying that one has free will I do not mean to imply that the person has some faculty, e.g., the will, that has the property of being free. Perhaps the expression ‘moral freedom’ (cf. Markosian 1999, 258) is preferable to ‘free will,’ for it is common to identify having free will with having alternatives, and it is equally common to believe that alternatives are not necessary for moral responsibility and, hence, neither is free will. But philosophers who think along these lines—for instance, semicompatibilists (Fischer 1994)—generally accept that there is a freedom-relevant condition that is necessary for moral responsibility. Thus, when they claim that free will is not essential to moral responsibility they are using the term ‘free will’ in a way that is different than the way that I am using it here.

2. The structure of this section is borrowed from Keith DeRose (1999).

3. Determinism is the conjunction of the following two theses: “For every instant of time, there is a proposition that expresses the state of the world at that instant” and “If p and q are any propositions that express the state of the world at some instants, then the conjunction of p with the laws of nature entails q” (van Inwagen 1983, 65). For other definitions of the key terms of the free will debate, see van Inwagen (1983) and Campbell, O’Rourke, and Shier (2004b).

For the purposes of this essay, (3) expresses the thesis of compatibilism whereas incompatibilism is the denial of (3). Some philosophers use the term ‘compatibilism’ to designate the view that the free will thesis—the claim that some persons have free will—is compatible with determinism (van Inwagen 1983). But since free will just is the freedom relevant condition necessary for moral responsibility, this usage is not significantly different from my own.

4. Contextualist theories of ability terms are developed by Lewis (1976 and 1979) and Sider (1997). A contextualist theory of ‘free action’ is presented by John Hawthorne (2001). Both kinds of theories are discussed by Peter Unger (1984, 54-8). Richard Feldman (2004) offers several compelling criticisms of contextualism. In this section, I’m presenting broad strong compatibilist strategies, so many of the important details of contextualism are left out. For a more complete understanding of contextualist theories of freedom, see Feldman (2004).

5. For a general introduction to the grandfather paradox, and other paradoxes of time travel, see Campbell (forthcoming).

6. Moore’s own analysis of the hypothetical sense of ‘could have done otherwise’ is more elaborate and more interesting. According to Moore, we could have done otherwise—in the hypothetical sense—iff three conditions hold: “(1) that we ... should have acted differently, if we had chosen to; (2) that similarly we ...  should have chosen differently, if we had chosen so to choose; and (3) that it was ... possible that we should have chosen differently, in the sense that no man could know for certain that we should not so choose” (Moore 1912, 94). (1) is equivalent to the standard hypothetical analysis, so Moore’s analysis is clearly more detailed than the standard one.

7. Unger (1983) suggests that the distinction between contextualism and invariantism maps on neatly, and respectively, to the distinction between compatibilism and incompatibilism. But the two-‘cans’ view shows that this is not so. The two-‘cans’ view is a kind of invariantism, for invariantism is just the denial of contextualism and the two-‘cans’ view is not a kind of contextualism. What we need is a distinction that lines up contextualism and the two-‘cans’ view on the same side and incompatibilist versions of the alternatives theory on the other.

8. These comments fit well with my previous claim that free will is a collection of active powers and cognitive capacities essential to moral responsibility (Campbell 1997 and 1999). The term ‘active power’ originally comes from Thomas Reid (1983).

HOME to Det & Free website
HOME to T.H. website