The Distasteful Conditional Analysis of Ability

Here is the paper I presented at the 2014 Eastern meeting of the Society of Christian Philosophers, on "Themes from the work of Peter van Inwagen", held at Niagara University, November 7th and 8th.

The Distasteful Conditional Analysis
Peter van Inwagen on determinism, compatibilism, and the ability to do otherwise

Guillaume Bignon

In discussions on the nature of free will and debates on whether it is compatible with determinism—that is, the debate opposing compatibilism and incompatibilism—Peter van Inwagen’s heavyweight contributions are such that one may come to see him as the ‘chief incompatibilist’. Those who like me wish to fearfully maintain (contra van Inwagen) that moral responsibility—and hence free will—is in fact compatible with determinism must therefore deal with van Inwagen’s case and in particular his so-called ‘Consequence Argument’ for incompatibilism.[1] While doing so in detail is not the burden of this paper,[2] it shall concern itself with a closely related question surrounding what Harry Frankfurt has famously named the ‘principle of alternate possibilities’. The principle states that ‘a person is morally responsible for what he has done only if he could have done otherwise’.[3] Indeed, at the heart of the debate on free will, and also fairly central to van Inwagen’s own Consequence Argument, is this idea that free will requires an ‘ability’ to do otherwise than one does: when the agent performs an action freely, the principle says, it must be the case that in a relevant sense, he could have done otherwise. Incompatibilists typically proceed to show that determinism in fact excludes such an ability to do otherwise, and hence conclude that determinism excludes free will and moral responsibility. But as van Inwagen recognizes, much hangs in the balance on just how one understands the word ‘ability’ in the aforementioned condition. What exactly does it mean to have the ‘ability’ to do otherwise?
Traditionally, compatibilists have sought to affirm something like the principle of alternate possibilities, but only by interpreting the ‘ability’ in question, in terms that are compatible with determinism. Theirs is the so-called ‘conditional analysis of ability’. They admit that on determinism, given that all things at the moment of choice are what they are, a person cannot choose otherwise—they are, after all, determined. But they maintain that if, conditionally, contrary to fact, the person had wanted to, or chosen to, or attempted to (or something very much like that), the person would have done otherwise. Compatibilists maintain that free will and moral responsibility require that sort of conditional ability, and do not require the more categorical sense of ability, the only one which admittedly would clash with determinism and hence entail incompatibilism.
On the face of it, this may seem to be a coherent position to take for compatibilists, but Peter van Inwagen and many incompatibilists with him, have had a lot of bad things to say about this apparently distasteful conditional analysis, not to mention William James and Immanuel Kant, respectively calling it a ‘quagmire of evasion’ and a ‘wretched subterfuge’[4]. Beyond the name-calling, I aim in this paper to assess the actual criticisms leveled against the conditional analysis of ability, and show that they in fact don’t hold water, nor undermine any compatibilist claim worth making.

In order to do so, let me begin by highlighting the two rather different principles of alternate possibilities that result from our two different analyses of ability. With the ‘non-conditional’, or ‘categorical’ sense of ability, we have:

Principle 1: ‘a person is morally responsible for what he has done only if, all things inside and outside the person being just as they are at the moment of choice, he could have done otherwise’

And with the ‘conditional’ sense of ability, we have:

Principle 2: ‘a person is morally responsible for what he has done only if he could have done otherwise, had his beliefs and desires inclined him to do so

In short, principle 1 says that moral responsibility requires a categorical ability to do otherwise, and principle 2 says that moral responsibility requires a conditional ability to do otherwise.[5] Now that the ‘ability to do otherwise’ has been disambiguated, we can see that only Principle 1 entails incompatibilism, while Principle 2 is quite compatible with compatibilism (because in the demands of Principle 2, some aspect of the past, namely the agent’s prior desire, is modified in order to make room for alternate possibilities). Given this, I, as a compatibilist, assert that Principle 1 is false, and Principle 2 is true. Of course that is highly controversial, but I shall now argue that incompatibilists tend to beg the question against compatibilism by failing to distinguish between the two senses of ability.
In debates on the matter, incompatibilists tend to offer examples in which an individual is not morally responsible because he lacks the conditional ability to do otherwise, and conclude from it that Principle 1 is true. But that does not follow. The invalid move is rarely challenged because the obvious truth of Principle 2 is found shining in plain sight, and with an equivocation close at hand, its plausibility is improperly transferred to Principle 1 lying in the nearby bushes. But this move is entirely invalid, and Principle 1 does not in fact enjoy the same support.
This invalid move could be exhibited by saying something like this: ‘it would be unreasonable to command someone to do something impossible for them to do. It would be like commanding an armless man to embrace you’.[6] We can easily recognize that an armless man is not reasonably to be held morally responsible for his failure to embrace someone (in this case ‘you’). But what is the exact nature of the inability that so obviously disqualifies him as a proper target of blame? It is one rooted in the fact that without arms, this man could not embrace you even if he wanted to. Even if this man’s beliefs and desires at the moment of choice were such that he wanted to obey the command and embrace you, he would find himself unable to do so; in other words, upon his failure, it is not the case that he could have done otherwise had his beliefs and desires inclined him to do so. Principle 2 thus successfully declares him to be exempt of guilt, and it follows that a compatibilist who affirms Principle 2 need not accept Principle 1 in order to avoid the absurd conclusion that an armless man could be held morally guilty for his failure to embrace someone. Principle 2 is sufficient (and Principle 1 unnecessary) to successfully adjudicate such cases. Intentional or not, this same equivocating maneuver is easily uncovered in incompatibilist writings. Consider the following instances:
David Widerker imagines that Holly had promised to attend Sarah’s big birthday party, and failed to do so. Sarah blamed Holly only later to find out that Holly could not do so due to a failed airline departure. Widerker concludes that this supports the unqualified (and hence incompatibilist) version of the principle of alternate possibilities: ‘it seems rationally incumbent upon Sarah to withdraw her blame and excuse Holly. Holly had no viable alternative to her failure to attend Sarah’s party and, hence, is not morally responsible for her absence’.[7] Yes indeed, barring any responsibility that Holly may bear for possibly making herself dependent on an untrustworthy airline in the first place, we can agree with Widerker that she is not blameworthy; but that verdict is not arrived at on the basis of Principle 1 alone. The much more modest Principle 2 does the job just as well, yielding the correct result: Holly is not morally responsible, because the airline failure made it so that she could not have fulfilled her promise even if she had wanted to. Given the absence of available transportation, Holly failed to keep her promise, and upon her failure, it is not the case that she could have done otherwise had her beliefs and desires inclined her to do so. As a matter of fact, if she is a decent friend, we can further suppose that her beliefs and desires did incline her strongly to do so, though she sorrowfully could not. Here again, Principle 2 is demonstrably sufficient (and Principle 1 unnecessary) to successfully adjudicate Holly and Sarah’s dispute, so that Widerker’s case does not support Principle 1.
David Copp offers another story wherein a boss requires an employee to do something that the employee lacks the ability to do. ‘A supervisor at the post office might demand that a mail carrier cook a soufflĂ© for everyone in the post office in the next five minutes when the mail carrier does not even know what a soufflĂ© is’.[8] Of course that carrier cannot be blamed for the failure, because she could not cook it even if she wanted to. The conditional Principle 2 is used once more to support an unqualified principle of alternate possibilities, and thus falls short of the impossible task.
Our final example comes from the pen of Peter van Inwagen himself, who envisions someone charging you of not speaking up while Jones was being bad-mouthed. This person, van Inwagen says, ‘must withdraw his charge if you can convince him that you were bound and gagged while Jones was being maligned’.[9] And this example is again used to support the unqualified claim that ‘ought implies can’ as an implicit close equivalent of Principle 1. The invalidity of this pattern must be clearly seen by now: you are not responsible if bound and gagged, because you could not have spoken even if you had wanted to. Upon failing to speak, it is not the case that you could have done otherwise had your beliefs and desires inclined you to do so. Principle 2 is sufficient and Principle 1 unnecessary. Hence these cases do not support Principle 1; they only reaffirm our Principle 2 about conditional ability, which I suggest ought to be rather uncontroversial.

With these important distinctions clearly made, we are now ready to hear and assess the criticisms, voiced most importantly by van Inwagen, on the conditional analysis of ability at play in our Principle 2. These criticisms fall broadly under three headings: we are told that the conditional analysis of ability is question begging, that it is not necessary, and that it is not sufficient.

First, then, is the charge that the conditional analysis is question-begging. Peter van Inwagen introduces the conditional analysis as one of the most important ‘argument[s] for the compatibility of free will and determinism’,[10] but remains largely unimpressed, and responds as follows:
What does the [conditional] analysis do for us? How does it affect our understanding of the Compatibility Problem? It does very little for us, so far as I can see, unless we have some reason to think it is correct. Many compatibilists seem to think that they need only present a conditional analysis of ability, defend it against, or modify it in the face of, such counter-examples as may arise, and that they have thereby done what is necessary to defend compatibilism.
            That is not how I see it. The particular analysis of ability that a compatibilist presents is, as I see it, simply one of his premisses; his central premiss, in fact. And premisses need to be defended.[11]
So van Inwagen essentially says that the conditional analysis of ability, taken as an argument for compatibilism, is question begging. I think that is exactly right, and we can join van Inwagen in his castigation: taken as an argument against incompatibilism, it is hopelessly question-begging. But all that means is that compatibilists should not use it like that. Instead, the only compatibilist claim worth making in this neighborhood is much more modest than that: it is simply one that says that a conditional analysis is compatible with determinism, and that it is the only analysis that has been shown to be necessary for moral responsibility and free will. This much I successfully defended above. So the conditional analysis is an utter failure indeed if taken as an attack upon incompatibilism, but it is just fine as a defense of compatibilism against the positive claim that moral responsibility and free will require an ability to do otherwise, a positive claim, which I explained earlier, begs the question of incompatibilism, a circular reasoning camouflaged by the equivocation on the word ‘ability’.

The second criticism of the conditional analysis is the claim that it is not necessary. To show that it’s the case, i.e. that it’s not necessary, van Inwagen first points out in passing that ‘Napoleon could have won at Waterloo’ can hardly mean ‘If Napoleon had chosen to win at Waterloo, Napoleon would have won at Waterloo’.[12] Then, van Inwagen similarly points to the important work by J.L. Austin on that very question.[13] Austin looked at the conditional meaning of ‘can’ in statements of ability and noted that when a person P is said to have the conditional ability to do action A, it must mean something like ‘P would do A if A wanted to’. But Austin found fault with this conditional analysis, because allegedly, ‘that P would do A if A wanted to’ is not necessary for P to be said to have the ability to do A (as was illustrated by van Inwagen’s above case of Napoleon at Waterloo). Robert Kane recounts for us the classic counterexample offered by Austin: the golfer.
For it is sometimes true that a person can (or could) do something, but false that the person would do it (or would have done it), if he or she chose or tried. Austin’s best-known example illustrating this point was one in which he imagined himself a golfer standing over a three-foot putt. It would be perfectly consistent, Austin argued, to say that he could (or had the power to) make the putt, though he might have missed it. For he has made many putts of this length in similar circumstances in the past—and also missed a few. His power to do it, therefore, does not imply that he would do it every time he wanted or tried.[14]
All of this is straightforward and convincing. Unfortunately for the incompatibilist, it fails to undermine any conditional claim really worth making. As a compatibilist, what I contended for above was not at all that ‘any and all talk of ability must be interpreted conditionally, as a counterfactual conditional like the ones employed to define the conditional analysis’. So it is perfectly fine to say that the golfer, in a very real sense ‘has the ability’ to make the shot even if his supposed desire does not ensure success. But this is not the sense of ability that matters for questions of moral responsibility. That kind of ‘guaranteed-success’ conditional ability isn’t (or shouldn’t be!) claimed to be necessary for a proper usage of ability-language. What the conditional version of the principle of alternate possibilities (our principle 2) was demanding a conditional ability to do otherwise for, was moral responsibility. It simply claimed that without a conditional ability to do A, P could not be morally responsible for failing to do A. The golfer example does not refute that. It does not even mention the issue of moral responsibility.
Now alternatively, if the claim of non-necessity at hand is no longer made regarding mere ability language, and actually says that a conditional ability is not necessary for moral responsibility, then the objection becomes wholly self-defeating, coming from incompatibilist libertarians who try with it to enforce the truth of Principle 1. It is absurd to assert that the weaker, conditional ability is not necessary, while maintaining that the stronger, categorical ability is in fact necessary for moral responsibility. As a matter of fact, if one possesses a categorical ability, it follows logically that he also possesses conditional ability; and conversely, if one does not even possess a conditional ability, then it is hopeless to think he has a categorical ability. So if the lack of conditional ability to do otherwise does not exclude moral responsibility, then the lack of categorical ability does so even less. Accordingly, if Principle 2 is false, it follows that Principle 1 is false also, and by attacking the conditional analysis of ability as unnecessary for moral responsibility, the libertarian provides the ammunition to refute his own Principle 1 by modus tollens.
(1) Principle 1 => Principle 2
(2) Principle 2 is false  (as per the objection)
(3) Principle 1 is false
Therefore, incompatibilist advocates of the principle of alternate possibilities cannot possibly find fault with the conditional analysis of the ability to do otherwise on the basis that it is not necessary, without castigating their own non-conditional analysis.
In conclusion, the necessity objection is irrelevant if speaking about necessity for proper descriptions of ability, and it is self-defeating if speaking about necessity for moral responsibility.
The necessity objection being thus a no-starter let us turn to the third of van Inwagen’s objections, the sufficiency question.

There are here again a couple of distinct worries related to the conditional analysis of ability’s not being sufficient. As was the case right above with the question of necessity, when assessing whether the conditional analysis is sufficient, not all incompatibilists actually complain about it being not sufficient for the same thing.
Peter van Inwagen (like many others) argues that the conditional analysis ‘person P would do action A if P wanted to’ is not sufficient to declare that P has the ability to do A. To show that this is the case, a counterexample is offered wherein one has this kind of conditional ability to perform an action, but can still properly be said not to have the ability to perform it. The counterexample that van Inwagen uses to that effect is one called ‘red candy’,[15] devised by Keith Lehrer, and retold in the following words by Robert Kane:
Suppose someone presents you with a tray of red candies. Nothing would prevent you from eating one of the candies, if you chose to. But you have a pathological fear of blood and of eating anything the color of blood, so you cannot choose to eat the candies; and so you cannot eat them.[16]
My compatibilist response to such examples is the same as I gave above to the similar complaints about necessity. Just as conditional analyses are not meant to be necessary to encapsulate any and all statements of ability, they are not meant to be sufficient to encapsulate any and all such statements either. It is perfectly fine to say that in a real sense the haemophobic person ‘cannot’ eat the red candy. It is just not the sense of ability that Principle 2 is concerned about. Principle 2 does not tell us what can sufficiently be the case in order to declare a person ‘able’ in all senses. It merely tells us of one specific sort of inability that excludes moral responsibility. Therefore this first sufficiency objection is simply irrelevant to the truth of Principle 2 and its conditional analysis of ability.
The second and more serious worry of incompatibilists regarding the sufficiency of the conditional analysis of ability is that it fails to deliver the right conclusion in some important moral cases. They fear that in the absence of the stronger Principle 1, Principle 2 is inadequate to make proper judgments of moral responsibility as it allegedly fails to exculpate agents who are nevertheless obviously not responsible for their action. ‘One standard objection maintains that the mere fact that an agent acts on his preference or does “as he pleases” is insufficient to sustain attributions of moral responsibility’.[17] Hugh McCann thus objects,
‘compulsives, addicts, people operating under duress—virtually everyone whose freedom to will differently we ordinarily view as compromised—would count by this criterion as free. Surely, if determinism is true, they would have willed differently had their strongest motives been different. Yet these are the people whose responsibility for decisions we would question, precisely because we think their strongest motive was too influential’.[18]
These counter-examples brought forward by incompatibilists also include cases of pathological fear,[19] and cases of what I call ‘overriding manipulation’[20] where a skillful hypnotizer or a mad-scientist could use a laser to manipulate impulses in a patient’s brain, and successfully control the patient’s actions. If the mad scientist or hypnotizer used his power to have the patient commit a moral wrong, Principle 2 would be useless to exculpate the patient. Indeed, he ‘could have done otherwise if his beliefs and desires had inclined him to do so’, and of course for these desires to be such, it would have been necessary that the scientist not be present, but it remains that the patient has the conditional ability, and thus Principle 2 would fail to annul the patient’s moral responsibility, whereas we all realize that he is neither free nor guilty because he is under the controlling scheme of the mad scientist or hypnotizer; such manipulation does indeed exclude moral responsibility.
The problem with this objection, however, is that it is both true and irrelevant. Why should one worry here about this kind of sufficiency? What claim or principle of the compatibilist are incompatibilist writers contending against? Principle 2 maintains that it is necessary to have the conditional ability to do otherwise if wanted, but it never said that this was sufficient. Daniel Speak objects that the ‘truth of the relevant conditional does not actually suffice for ability’.[21] But this was never part of the claim. Principle 2 tells us only that moral responsibility entails the conditional ability to do otherwise; it does not conversely tell us that the conditional ability to do otherwise alone entails moral responsibility. It tells us that if a person could not do otherwise even if his beliefs and desires had inclined him to do so, then he lacks moral responsibility; but Principle 2 says nothing about anything else that may jointly be required for securing moral responsibility. And here, the compatibilist is free to consider and accept any refined, convincing analysis he pleases, and to list any number of possibly necessary items, which together do constitute a sufficient criterion to positively affirm moral responsibility. Compatibilists are perfectly free to include many such items in that list;[22] they only maintain (and the above arguments have defended) that ‘a categorical, libertarian ability to do otherwise’ is not one of them.

In conclusion, we see that these various misunderstandings of the real claim made by compatibilists with their modest Principle 2 have led incompatibilists to place misguided expectations on the so-called ‘conditional analysis’ of ability. We saw above that they wanted it to deliver many things it was never intended to produce: an argument for compatibilism, a necessary criterion for all descriptions of ability, a sufficient criterion for all descriptions of ability, and a sufficient criterion for all attributions of moral responsibility; when all it really is, is a necessary criterion for attributions of moral responsibility. So of course these misplaced expectations are bound to be frustrated, but since I showed that they do nothing to undermine compatibilism, that loss will not be grieved by anyone, and compatibilists will happily attend the funeral.

[1] See its thorough articulation in Peter van Inwagen, An Essay On Free Will (Oxford: Clarendon Press, 1983), 55-105.
[2] It is one of the central burdens of my PhD dissertation.
[3] Harry G. Frankfurt, ‘Alternate Possibilities and Moral Responsibilities’ in Harry G. Frankfurt, The Importance of What We Care About (New York, NY: Cambridge University Press, 1998), 1.
[4] Both quoted in Daniel C. Dennett, Elbow Room: The Varieties of Free Will Worth Wanting (Cambridge, MA: Bradford Books, 1984), 131.
[5] These two senses of ability are similarly distinguished by Ishtiyaque Haji who names them ‘the hypothetical sense of “can”’ and ‘the categorical sense of “can”’. Ishtiyaque Haji, Incompatibilism’s Allure: Principal Arguments for Incompatibilism (Peterborough, Ontario: Broadview Press, 2009), 34-36. In the theological world, we find the same distinction in Jonathan Edwards and A.W. Pink. Edwards writes ‘of the distinction of natural and moral necessity, and inability’ in Jonathan Edwards, Freedom of the Will (New York: Cosimo Classics, 2007), 23-31 (Part I, section IV). As to A.W. Pink, all references to this distinction in the 1945 edition of The Sovereignty of God were unacceptably removed by the editors of the 1961 edition. See explanation and valid criticism of this move by Paul Helm at
[6] This preliminary, illustrative case is actually taken from a real-world example of a popular source: Jerry Vines, ‘Sermon on John 3:16’ in Whosoever Will: A Biblical-Theological Critique of Five-Point Calvinism, ed. David L. Allen & Steve W. Lemke (Nashville, TN: B&H Publishing Group, 2010), 26.
[7] Michael McKenna and David Widerker, ‘Introduction’ in Moral Responsibility and Alternative Possibilities: Essays on the Importance of Alternative Possibilities, ed. David Widerker & Michael McKenna (Burlington, VT: Ashgate Publishing Company, 2006), 3.
[8] David Copp, ‘“Ought” Implies “Can”, Blameworthiness, and the Principle of Alternate Possibilities’ in Moral Responsibility and Alternative Possibilities: Essays on the Importance of Alternative Possibilities, ed. David Widerker & Michael McKenna (Burlington, VT: Ashgate Publishing Company, 2006), 271.
[9] Peter van Inwagen, An Essay On Free Will (Oxford: Clarendon Press, 1983), 161.
[10] Ibid., 114.
[11] Ibid., 121.
[12] Ibid., 115.
[13] Ibid., 235, note 7.
[14] Robert Kane, The Significance of Free Will (New York, NY: Oxford University Press, 1998), 54.
[15] van Inwagen, Essay, 115.
[16] Kane, Significance, 57.
[17] David M. Ciocchi, ‘Reconciling Divine Sovereignty and Human Freedom,’ in JETS 37/3 (September 1994), 409.
[18] Hugh J. McCann, The Works of Agency : On Human Action, Will, and Freedom (Ithaca, NY : Cornell University Press, 1998), 177.
[19] van Inwagen, Essay, 115.
[20] I deal in details with the manipulation argument for incompatibilism in my upcoming PhD dissertation.
[21] Daniel Speak, “The Consequence Argument Revisited” in Robert Kane, ed., The Oxford Handbook of Free Will 2nd ed. (New York, NY: Oxford University Press, 2011), 123.
[22] Though this exhaustive list is notoriously difficult to provide, whether one is a compatibilist or an incompatibilist.