Descartes on the Principle of Alternative Possibilities

C P Ragland. Journal of the History of Philosophy. Volume 44, Issue 3. July 2006.

Descartes certainly believed in free will, but it is far from clear how he understood the nature of freedom. This paper aims to clarify Descartes’s view of freedom to some extent by determining whether he accepted the principle of alternative possibilities (PAP). According to PAP, doing something freely implies being able to do otherwise; freedom consists in a two-way power to do or not do. Commentators do not agree about Descartes’s relation to PAP: some suggest that he never accepted it, others that he did so only in later texts; and still others that he accepted it throughout his career. Commentators also disagree about whether or not PAP conflicts with other things Descartes says about freedom (thus some accuse Descartes of incoherence). Here I will argue that Descartes embraced PAP throughout his career in a way that coheres with his other main claims about freedom.

In what follows, I examine each of Descartes’s main texts on freedom in chronological order: the Meditations (1641), the Principles (1644), and the two letters to Mesland (1644-45). Each of these works, read in isolation, is best interpreted as endorsing PAP. Joined together and read in light of one another, they make it virtually certain that Descartes believed in PAP throughout his career.

But first, a disclaimer. Philosophers who agree that freedom requires alternative possibilities may still disagree about whether such alternatives are compatible with determinism. The incompatibilists among them say that if our every choice were predetermined, we could never choose otherwise and hence would not be free. The compatibilists claim that there is more than one sense of ‘could have done otherwise,’ and that the sense of ‘”could have done otherwise’… denied by determinism is irrelevant to the sense required for freedom.” My goal in this paper is simply to establish that Descartes uniformly endorsed PAP. With respect to exactly how he understood PAP’s relation to determinism, I explore some options, but leave the question open.

Meditations on First Philosophy

In the Fourth Meditation, Descartes defines the will as follows:

… [i] the will, or freedom of choice … simply consists in this: that we are able to do or not do (that is, to affirm or deny, to pursue or avoid); [ii] or rather [vel potius], simply in this: that we are carried in such a way toward what the intellect proposes for affirmation or denial or for pursuit or avoidance, that we feel ourselves determined to it by no external force. (AT 7:57 / CSM 2:40)

This passage contains two important claims. First, the opening seems straightforwardly committed to PAP. Whether we are engaged in an act of judgment (affirming or denying a proposition) or choice (pursuing or avoiding a course of action), our freedom consists in our power to do otherwise. Second, Descartes here uses ‘freedom of choice’ (arbitrii libertas) as a synonym for ‘the will’ (voluntas), suggesting the principle that freedom is essential to the will (FEW): the will is free by its very nature, and every voluntary act is free. FEW combines with PAP to imply that the will can always do otherwise than it does, that we enjoy two-way power with respect to every one of our voluntary actions.

However, a few paragraphs later, Descartes seems to give a counterexample to this conclusion. In what I call the “great light” passage, Descartes reflects on his experience of the cogito argument, and says:

… I could not but judge [non potui … non judicare] something which I understood so clearly to be true; not because I was compelled so to judge by any external force, but because a great light in the intellect was followed by a great inclination in the will, and thus I have believed this more spontaneously and freely as I have been less indifferent to it. (AT 7:58-59 / CSM 1:41; my translation and italics)

Like the earlier passage, this one seems to express two key assumptions. The first is a doctrine of clear and distinct determinism (CDD): in at least some cases where we perceive a proposition P clearly and distinctly, we cannot refrain from assenting to P. CDD implies that we lack two-way power with respect to at least some acts of will. Second, Descartes assumes that judgment is a voluntary act (JVA). These two assumptions combine to imply that the will cannot always do otherwise than it does, that we do not enjoy two-way power with respect to every one of our voluntary actions.

Therefore, the definition of the will and the great light passage seem diametrically opposed, each implying the opposite of the other. More generally, PAP, FEW, CDD, and JVA form an inconsistent quartet: the truth of any three of them logically entails the falsehood of the fourth. There is little wonder that some commentators question the coherence of Descartes’s remarks on freedom.

But Descartes was too smart to miss such an obvious contradiction. Interpretive charity requires us to look for some coherent doctrine beneath the surface of these two passages. Descartes seems clearly committed to FEW and JVA, so I will concentrate on the tension between PAP and CDD. Commentators have tried two basic strategies for resolving it. The first is to deny that Descartes really endorsed PAP. The second is to claim that Descartes uses ‘power’ in two different senses: the two-way power that PAP requires for freedom is not the same as the power ruled out by CDD. I will now explain what I take to be the most plausible version of the first strategy, show why I do not find it promising, and then explore the second strategy in more detail.

Though the definition of freedom’s first part seems to endorse PAP, the latter part may show that Descartes does not accept PAP after all. Anthony Kenny offers the most plausible version of this sort of reading. Kenny distinguishes between “liberty of indifference” (two-way power) and “liberty of spontaneity,” which we enjoy with respect to an act “if and only if we do it because we want to do it.” According to Kenny, the first clause defines freedom as liberty of indifference, the second as liberty of spontaneity, and the Or rather’ means that “freewill often does consist in liberty of indifference, but that sometimes it consists only in liberty of spontaneity, and that is all that is essential to it.” In other words, Descartes adds the ‘or rather’ and second clause to make it clear that alternative possibilities are not necessary for freedom.

Though Kenny does not mention this, on his interpretation Cartesian freedom is “asymmetrical” in Susan Wolf’s sense: if we are doing the right thing, freedom does not require that we be able to do otherwise, but if we are doing the wrong thing, it does. On such a view, freedom is ultimately just the ability to avoid error. God “has given me the freedom to assent or not assent in those cases where he did not endow my intellect with a clear and distinct perception” (AT 7:61 / CSM 2:42.) because when obscurity in the intellect makes it possible for us to err, our freedom requires that it is also possible for us to suspend judgment, and hence avoid error (AT 7:59 / CSM 2:41). That is why freedom often consists in liberty of indifference. However, as the great light passage seems to illustrate, when clear perceptions determine us to assent to the truth (and thus avoid error), freedom does not require that we also be able to suspend judgment. That is why spontaneity alone is essential to freedom.

Kenny interprets Descartes via a distinction (indifference vs. spontaneity) from Hume (Treatise [III.2.1]), and thus may seem guilty of anachronism. However, the term spontaneum is Descartes’s own (see AT 4:175 / CSMK 246), and Kenny is correct that the second clause defines spontaneity: in the great light passage, Descartes suggests that he assented to the cogito “spontaneously” because he satisfied the conditions laid down in the second clause.

But does Kenny offer an accurate account of Cartesian spontaneity? “Doing what you want to do” may look more like an account of voluntariness than spontaneity. However, the second clause of Descartes’s definition identifies spontaneity as a way of being “carried,” and the very next sentence reads: “For in order to be free it is not necessary that I can be carried [ferri posse] in both directions, but on the contrary, the more I incline [propendeo] in one direction … the more freely do I choose it” (AT 7:57-58 / CSM 2:40; my translation). This suggests that we can be carried in both directions only if we are inclined in both directions, which in turn suggests that what carries us is an inclination or desire. Furthermore, the great light passage associates spontaneity with “a great inclination in the will.” So Kenny is right to associate spontaneity with doing what we want to do.

However, Kenny’s account ignores Descartes’s reference to external determination. In the second clause, Descartes says that freedom consists in being carried “in such a way toward what the intellect proposes … that we feel ourselves determined to it by no external force” (a nulla vi externa nos ad id determinari sentiamus) (AT 7:57 / CSM 2:40). Vere Chappell glosses this passage as follows: “an action is spontaneous if it is performed by its agent entirely on his own, without being forced … by any external factor.” Descartes’s text itself requires for spontaneity only that we feel undetermined, not that we actually be undetermined. However, there are at least two reasons to think that Chappell’s gloss is correct. First, Descartes is probably using the word ‘feel’ (sentiamus) here to reiterate his frequently stated opinion that we have an inner feeling or experience of freedom. Descartes suggests that this experience of freedom is clear and distinct, and hence (given the divine guarantee) veridical. So if in the experience of freedom we feel undetermined, we really are undetermined. Second, one of the main points of the Fourth Meditation seems to be that because we are certain that we err freely, we can also be certain that God is not causing our errors. But if freedom is merely a feeling, and is thus consistent with behind-the-scenes external control, then how can we be sure that God is not making us err after all?

So Kenny’s account of spontaneity is incomplete. Cartesian spontaneity involves both acting on inclination and being free from external determination. However, Kenny could easily rectify this problem without giving up his basic interpretation of the ‘or rather.’ The real trouble with Kenny’s reading is this: Descartes intends his definition of freedom to explain why there is an analogy between the divine and human will, and Kenny’s reading is hard pressed to deal with this aspect of the definition. A reading on which the definition endorses PAP does a much better job. Before discussing the analogy, I must first lay some groundwork by explaining Descartes’s notion of indifference as it relates to both divine and human freedom.

As Descartes normally uses the term, ‘indifference’ denotes a motivational state that comes in degrees. Perfect indifference is “the state the will is in when it is not impelled more in one direction than in another by any perception of truth or goodness” (AT 4:173 / CSMK 2.45; see also AT 7:58 / CSM 2:40; AT 4:174 / CSMK 245). In a state of perfect indifference, the motivations for and against a given act of will are perfectly balanced. We become progressively less indifferent as the motivations on one side outweigh those on the other, and we lose indifference altogether when we are motivated in only one direction (AT 4:115 / CSMK 2.33; AT 4:174 / CSMK 245-46; AT 4:155 / CSMK 233).”

Perfect indifference comes in two forms. The first is balanced multi-directional motivation, in which “we recognize many reasons pro but as many reasons contra” (AT 4:174/CSMK 2.45). For example, we might be motivated to affirm a proposition, but equally motivated not to affirm it. The second form of perfect indifference is non-motivation. If we recognize no reasons at all, either pro or contra, then we are not inclined “more in one direction than in another.”

In the Sixth Replies, Descartes ascribes the second sort of indifference to God:

It is self-contradictory to suppose that the will of God was not indifferent from eternity with respect to everything which has happened or will ever happen; for it is impossible to imagine that anything is thought of in the divine intellect as good or true, or worthy of belief or action or omission, prior to the decision of the divine will to make it so … Thus the supreme indifference to be found in God is the supreme indication of his omnipotence. (AT 7:431-32 / CSM 1:291)

Because God’s creative decisions are not motivated at all, God has (trivially) a perfect balance with respect to them, and thus enjoys the supreme degree of indifference. As a direct implication of God’s power over standards of truth and goodness, this indifference signifies God’s omnipotence.

Human indifference, on the other hand, is very different from divine indifference:

But as for man, since he finds that the nature of all goodness and truth is already determined by God, and his will cannot tend toward anything else, it is evident that he will embrace what is good and true all the more willingly, and hence more freely, in proportion as he sees it more clearly. He is never indifferent except when he does not know which of the two alternatives is the better or truer, or at least when he does not see this clearly enough to rule out any possibility of doubt. (AT 7:432-33 / CSM 2:292)

Unlike the divine will, the human will cannot act unless the intellect first puts forward some object for its consideration. The intellect must conceive this object as good or true in some respect, so that the will is inclined toward it. Therefore, unlike the divine will, the human will cannot choose from a state of non-motivation; insofar as it bears on choice, human indifference always involves multi-directional motivation. Because we experience indifference only if we fail to see things with perfect clarity, it is a sign of our weakness.

Despite these differences, Descartes insists that “it is principally because of this infinite will within us that we can say we are created in [God’s] image …” (AT 2:628; CSMK 141-42). Descartes expands on this idea in the passage that introduces his definition of freedom:

It is only the will, or freedom of choice, which I experience within me to be so great that the idea of any greater faculty is beyond my grasp; so much so that it is above all in virtue of the will that I understand myself to bear in some way the image and likeness of God. For although God’s will is incomparably greater than mine, both in virtue of the knowledge and power that accompany it and make it more firm and efficacious, and also in virtue of its object, in that it ranges over a greater number of items, nevertheless it does not seem any greater than mine when viewed as will formally and precisely in itself [in se formatter & praedse spectata]. This is because the will simply consists in this: that we are able to do or not do, or rather … (AT 7:56-57 / CSM 2:39-40; my translation)

Note the phrase ‘this is because’ (quid): the definition of freedom is supposed to explain why God’s will is-qua will-no greater than ours. It does so by explaining what will is, as CSM puts it, “in the essential and strict sense.” Descartes might seem to be claiming that the divine and human will share the same essence. But in his reply to an objection about this passage, he states clearly: “no essence can belong univocally to both God and his creatures” (AT 7:433 / CSM 2:2.92). How, then, is our will an image of God’s?

Descartes’s position seems to be that there is an analogy between divine and human will, just as there is an analogy between divine and created substance. In Prinapks 1.51 Descartes defines substance as “a thing which exists in such a way as to depend on no other things for its existence,” and notes that only God is a substance in this sense. Therefore, he concludes that “the term ‘substance’ does not apply univocally … to God and to other things” (AT 83:2.4; CSM 1:210). However, he goes on to define created substances as things that “need only the ordinary concurrence of God in order to exist,” and hence do not depend on any other created things (AT 9b:47; CSM 1:210). The essence of divine substance is different from that of created substance, but there is an analogy between the two because they share a common feature: both involve the general idea of ontological independence. In the same way, Descartes seems to think that there is some point of similarity between the divine will and the human will, and the definition of freedom is supposed to explain what it is.

The first clause of the definition seems to identify two-way power as the point of similarity. This makes sense because Descartes identifies divine freedom with two-way power: God “was free to make it not true that all the radii of the circle are equal-just as free as he was not to create the world” (AT 1:152.; CSMK 2.5). However, on Kenny’s reading, two-way power is not essential to human freedom, so we must look to the second clause to find the point of similarity. As we have seen, the second clause involves two main ideas: doing what we want to do, and being free from external determination. Unfortunately for Kenny, neither of these notions works very well.

Doing what we want to do cannot be the point of resemblance because it involves being carried toward what the intellect puts forward for our consideration, and for Descartes such motivation (reasons of truth or goodness grasped prior to choice) cannot possibly apply to God’s will. Spontaneity is essential to the human will, but not the divine will. Self-determination, or freedom from outside constraint, is a more plausible candidate for the point of similarity: God must be essentially free from external determination because there is nothing outside him before he freely creates. However, three considerations make self-determination less plausible than two-way power as the point of similarity.

First, to identify self-determination as the point of similarity, we must gloss over a key detail in the second clause: the phrase ‘in such a way … that we feel ourselves determined by no external force’ modifies ‘that we are carried … toward what the intellect proposes.’ Strictly speaking, the second clause identifies freedom not simply with being undetermined, but with being carried in an undetermined way. As I noted earlier, the second clause describes a kind of spontaneity that involves both being undetermined by external forces and doing what we want to do. Since the latter notion does not apply to God, and infects (as it were) the entire second clause, it seems unlikely that Descartes intended the second clause to describe divine freedom at all. It looks more like a clarification of the nature of human freedom.

Second, and more worrying: if self-determination is the point of similarity, then it is hard to make sense of the first clause. In context, the definition would mean something like: “Divine and human freedom have two-way power in common. Well, actually, no. Rather they have self-determination in common.” If two-way power is not essential to human freedom, then the first clause fails to explain the point of similarity and is extremely misleading. It is hard to see why Descartes would leave it in place.

Finally, the best textual evidence for the idea that self-determination is essential to God’s freedom also suggests that self-determination implies having two-way power. Descartes says that “God cannot have been determined to make it true that contradictories cannot be true together, and therefore … he could have done the opposite” (AT 4:118 / CSMK 235). Descartes’s inference here seems to depend on the following suppressed premise: if an agent is not determined to perform an action, then that agent could have not performed it. This premise seems correct: to be determined by some factor F just is to be unable to do otherwise, given F, so to be not determined by F just is to be able to do otherwise, given F. So if Descartes identifies human freedom with being undetermined by external forces, it seems he would also take this lack of determinism to imply that we are able to do otherwise, given external forces.

I cannot pretend that these considerations completely rule out Kenny’s reading, but I do think they should motivate us to explore alternatives. The obvious alternative is to consider two-way power the point of similarity, in which case it must be essential to human freedom. The ‘or rather’ does not retract PAP, but clarifies Descartes’s understanding of it. Or rather’ means ‘in other words,’ so that the second clause spells out necessary and sufficient conditions for humans to possess the two-way power mentioned in the first clause. For humans, being undetermined by external forces and being able to do otherwise-in the sense necessary for freedom-are two sides of the same coin.

On this alternative reading, the first clause of the definition of freedom explains the essential similarity between divine and human freedom: both involve two-way power. The second clause shifts the focus exclusively to human freedom in order to forestall a potential misunderstanding. Descartes thinks that indifference is necessary for divine freedom: “If some reason for a thing’s being good had preceded [God’s] preordination,” Descartes says, “that reason would have determined him to make that which is best” (AT 7:435 / CSM 2:194; my translation and italics). Were he not indifferent, God would be unable to refrain from making the best, and hence would not be free. Descartes’s readers might conclude that our freedom is like God’s not only in being a two-way power, but also in requiring indifference. Descartes adds the ‘or rather’ and second clause to clarify that human freedom does not require indifference: “For in order to be free it is not necessary that I can be carried [i.e., motivated] in both directions, but on the contrary, the more I incline in one direction … the more freely do I choose it” (AT 7:57-58 / CSM 2:40). So there is a further parallel with the analogy of substance. Just as divine and created substances enjoy two different kinds of ontological independence, the divine and created will enjoy two different kinds of two-way power: God’s kind requires indifference, but our kind does not.

At the same time, the second clause also clarifies the sense in which human freedom requires alternative possibilities. When discussing freedom, Descartes uses the modal terms ‘can’ and ‘could’ in two distinct ways. Sometimes, as in the passage above, being able to go in more than one direction is associated with being motivated in more than one direction: ‘I could have done otherwise’ means ‘I had a reason or motive for doing otherwise.’ Whenever the will is indifferent, it has alternative possibilities in this sense, because it has motives both pro and contra a course of action. So we can call these alternatives of indifference. In other places, Descartes uses ‘I could have done otherwise’ to mean ‘external forces did not determine me to do what I did.’ Call these alternatives of self-determination. Given these two senses of ‘can,’ the first clause is ambiguous in requiring the ability “to do or not do” for freedom. Sensing this ambiguity, Descartes added the second clause to clarify that he means to require alternatives of self-determination. He then followed the second clause with the passage quoted just above, to further clarify that freedom does not require alternatives of indifference.

If, as I have suggested, Descartes distinguished implicitly between two senses of ‘could have done otherwise,’ this would explain why he was not fazed by the apparent contradiction between the first clause of the definition (which endorses PAP), and the great light passage (which seems to deny PAP). When Descartes says that he could not but judge the cogito true, he means that he had no reason to do otherwise, that all of his inclinations were on the side of assent. But he states explicitly that he was not determined to assent by any external force. He lacked alternatives of indifference, but still enjoyed alternatives of self-determination. And the first clause requires only alternatives of self-determination for freedom.

Thus far I have explained a general strategy of interpretation that makes two key claims. First, the second part of Descartes’s definition clarifies rather than retracts his commitment to PAP, and second, the great light passage does not conflict with the definition because each employs a different sense of ‘can.’ All things considered, I believe this reading fits the texts more snugly than Kenny’s reading. My general strategy can develop into at least three more specific interpretations, depending on how we answer two key questions about the second clause’s crucial phrase, ‘determined to it by no external force.’

On an incompatibilist reading, we are determined to an action just in case an external force sufficiently causes the act of will either directly or indirectly. So in the situation just diagrammed, the act of will would be “externally determined.” On this reading, freedom requires both that the act of will not be causally determined by an external force at the time of action, and that if the act is determined by internal forces, those internal forces cannot themselves be causally determined by prior external forces.

Now for the second key question: what does Descartes count as an “external force”? It seems God would be such a force, and that is perhaps why Descartes insists, in Principles 1.41, that God’s providence “leaves the free actions of men undetermined” (AT 8a:20 / CSM 1:2.06). The body, too, seems external (see Passions of the Soul, esp. AT 11:328 / CSM 1:328). But Descartes’s position regarding the intellect is ambiguous: does ‘external’ mean ‘external to the self (thinking substance),’ so that the intellect is internal, or does it mean ‘external to the will,’ so that the intellect is external?

If we assume an incompatibilist interpretation of ‘determined’ and further suppose that the intellect is an external force, then we will be led to what I call the “radical freedom” interpretation. Consider again the great light passage, where a clear perception in Descartes’s intellect causes him to be inclined in only one direction (to assent to the cogito). If that inclination in turn determined the will’s act, then Descartes’s assent would be indirectly determined by an external force. Since Descartes was not thus determined, it must have been psychologically possible for him to withhold assent in the face of this one-way motivation. In scholastic terms, he must retain freedom with regard to the exercise of his will (to act or not act) even if (due to his motives) he lacks freedom with regard to the specification of his act (say, to affirm or deny). On this view, the claim that Descartes lacked alternatives of indifference-that he had no reason or motive for doing otherwise-means that for him to do otherwise would be immoral or irrational, but not psychologically impossible.

The two remaining interpretations agree that for Descartes it is psychologically impossible to withhold assent unless we have a reason for holding back. Therefore, if Descartes had no motive to do otherwise than assent to the cogito, then he was (at that moment) psychologically unable to do otherwise. The two interpretations differ over why this psychological inability to do otherwise is compatible with freedom.

The first adopts the compatibilist reading of ‘determined.’ On this view, the clear perception in the intellect determines the will to be inclined in only one way, which in turn determines the will to act in that way. However, even if the intellect is “external,” this indirect determinism is no threat to freedom. If the will’s inclinations had been different, the will would have done something else, and so the agent could have done otherwise in the only sense necessary for freedom. On this reading, Descartes endorses what Joseph K. Campbell calls “two-way” compatibilism, which endorses PAP, but then gives a hypothetical account of alternative possibilities, rendering them compatible with determinism.

The final interpretation adopts the incompatibilist reading of ‘determined,’ but insists that in the great light passage (and all relevantly similar situations), the intellect is not an external force. On this view, ‘external’ really means ‘external to the will or its influence.” In the great light passage, Descartes’s clear perceptions are “internal” because they were brought about by an earlier act of will regarding how to focus attention (and this earlier act was not determined by the intellect’s contents). On this reading, the will can be both free and determined by the intellect in specific cases, but it could not be free if always determined by the intellect. So on this reading, Descartes does not endorse what Laura Eckstrom calls a “narrowly construed” version of PAP, according to which an act of will is free at a given time only if the agent is psychologically able to do otherwise at that time. Rather he endorses a broader version of PAP that allows for inability to do otherwise at the time of action, provided that the agent enjoyed the relevant sort of alternatives at some earlier time.

There is much to be said both for and against each of these interpretations. However, this is not the place to determine how exactly Descartes understood PAP. We must move on to examine whether he also affirms PAP in his later writings.

Principles of Philosophy

In the Part One of the Principles, Descartes’s discussion of freedom is less detailed than in the Meditations. He claims neither that divine and human freedoms are essentially similar, nor that indifference is unnecessary for human freedom. He is also silent concerning liberty of spontaneous self-determination. However, he says nothing to contradict any of those earlier claims, and he clearly remains committed to PAP, FEW, CDD, and JVA.

He asserts CDD in Principle 43: “the minds of all of us have been so molded by nature that whenever we perceive something clearly, we spontaneously give our assent to it and are quite unable to doubt its truth” (AT 8a:21 / CSM 1:207). As before, clear and distinct perceptions determine the will.

Principle 39 could be taken to suggest that the when the will is thus determined, it does not enjoy two-way power. Descartes says that “we have power in many cases to give or withhold assent at will,” and that in the past when we imagined ourselves the victims of an omnipotent deceiver, our freedom was “nonetheless so great as to enable us to abstain from believing whatever was not quite certain or fully examined” (AT 83:19-10 / CSM 1:205-06; my italics) These remarks could be taken to mean that we lack any sort of two-way power with respect to what is fully certain.

However, consider the following passage from Principle 37:

it is a supreme perfection in man that he acts voluntarily, that is, freely; this makes him in a special way the author of his actions and deserving of praise for what he does. We do not praise automatons for accurately producing all the movements they were designed to perform, because the production of these movements occurs necessarily. It is the designer who is praised … for in constructing [automatons] he acted not out of necessity but freely. By the same principle, when we embrace the truth, our doing so voluntarily is much more to our credit than would be the case if we could not do otherwise [quam si non possemus non amplecti]. (AT 83:19 / CSM 1:105)

The opening of this passage might seem to work against my reading, for Descartes equates freedom with voluntariness (FEW again), and it seems that we can do something voluntarily even if we could not have done otherwise (as Frankfurt-Style Counterexamples to PAP show). However, the rest of the passage indicates that in Descartes’s opinion, we act voluntarily only if we could have done otherwise. The example of the automatons and their designer suggests that for Descartes, a good action is worthy of praise (i.e., to our credit) only if we perform it freely (i.e., voluntarily), and we perform an action freely only if we do not do so necessarily.

In this context, the locution ‘Doing so voluntarily is more to our credit than would be the case’ leads us to expect the completion ‘if we did so necessarily.’ But instead, Descartes says: “if we could not do otherwise.” This suggests that for Descartes the phrases ‘we did so necessarily’ and ‘we could not have done otherwise’ are equivalent. And if these phrases are equivalent, then Descartes must think that we perform an action freely or voluntarily only if we could have done otherwise.

Furthermore, Descartes here imagines a case where we deserve credit for embracing-i.e., believing-the truth. But for Descartes, we deserve credit for believing something only if we perceive it clearly and distinctly. We deserve blame for believing what we do not clearly perceive: “when we give our assent to something which is not clearly perceived, this is always a misuse of our judgment, even if by chance we stumble on the truth” (AT 8a:21 / CSM 1:207). So in the passage above, Descartes must be imagining a case in which we assent to a clearly perceived truth. In such a case, he maintains, we assent voluntarily (JVA) and freely, and therefore could have done otherwise (PAP).

Principle 37 thus commits Descartes to JVA, FEW, and PAP; Principle 43 commits him to CDD. His apparent problem is the same-he seems to maintain both that we can and that we cannot withhold assent from clear perceptions-and there is nothing to suggest that he has altered his solution to this problem since the Meditations.

The Letters to Mesland

In the mid-1640s the young Jesuit Denis Mesland wrote to Descartes with some questions about the Meditations. One of these concerned free will, particularly our ability to suspend judgment. In a letter dated 2 May 1644, Descartes responded to Mesland as follows:

I agree with you when you say that we can suspend our judgment; but I tried to explain in what manner this can be done. For it seems to me certain that “a great light in the intellect is followed by a great inclination in the will”; so that upon seeing very clearly that a thing is good for us, it is very difficult, and even, as I believe, impossible, while one remains in this thought, to stop the course of our desire. But the nature of the soul is such that it hardly attends for more than a moment to a single thing; hence, as soon as our attention turns from the reasons which show us that the thing is good for us, and we merely keep in our memory the thought that it appeared desirable to us, we can call up before our mind some other reason to make us doubt it, and so suspend our judgment, and perhaps even form a contrary judgment. (AT 4:116 / CSMK 133-34)

This passage implies an analogue of CDD, affirming with respect to the Good what earlier texts said about the True: if we see it clearly, it is impossible to hold back from it.

However, Descartes’s commitment to PAP is less clear in this text. He says that to suspend judgment, we must turn our attention away from the reasons that make a thing’s goodness clear and distinct to us. This seems to imply that we can suspend judgment only after our perceptions are no longer clear. Descartes may mean that at the time of clear perception, we are unable to refrain from pursuing the good. Nevertheless, our pursuit is voluntary, and Descartes goes on to say (re-asserting FEW): “I call free in the general sense whatever is voluntary” (AT 4:116 / CSMK 2.34). So this 1644 letter leaves it open that we can be free at a time without having two-way power at that time.

In a subsequent letter to Mesland from 1645, Descartes expands on the relation of free will to time. He says: “freedom considered in the acts of the will at the moment when they are elicited” does not involve two-way power, “for what is done cannot remain undone as long as it is being done” (AT 4:174 / CSMK 246). This shows that Descartes would reject a version or interpretation of PAP requiring that the will be able to do otherwise both before and during its act. However, in the same letter Descartes explicitly accepts a more modest version of PAP, according to which being free implies having alternative possibilities during the interval after the intellect puts forward an object for deliberation but before the will elicits its act of choice or judgment. “Considered with respect to the time before [acts of will] are elicited,” Descartes says, freedom entails “a positive faculty of determining oneself to one or other of two contraries, that is to say, to pursue or avoid, to affirm or deny” (my italics; AT 4:173 / CSMK 245).

This last quotation reaffirms both JVA and PAP. It also suggests that Descartes was committed to PAP in his earlier letter to Mesland. The quotation identifies the will’s essential positive power of self-determination as a two-way power. Thus Descartes was probably hinting at the will’s two-way power in 1644, when he told Mesland: “since you regard freedom not simply as indifference but rather as a real and positive power to determine oneself, the difference between us is merely a verbal one-for I agree that the will has such a power” (AT 4:116/ CSMK 2.34; my italics). It seems that in his correspondence with Mesland, Descartes is still committed to FEW, JVA, PAP, and CDD. His apparent problem is still the same.

In what I call the “two senses” passage from the 1645 letter, Descartes tries to solve his problem by further explaining our positive two-way power:

I do not deny that the will has this positive faculty. Indeed, I think it has it not only with respect to those actions to which it is not pushed by any evident reasons on one side rather than on the other, but also with respect to all other actions; so that when a very evident reason moves us in one direction, although morally speaking we can hardly move in the contrary direction, absolutely speaking we can. For it is always open to us to hold back from pursuing a clearly known good, or from admitting a clearly perceived truth, provided we consider it a good thing to demonstrate the freedom of our will by so doing. (AT 4:173; CSMK 2.45)

Descartes reiterates that that two-way power is essential to the will (PAP). He reconciles PAP with CDD by distinguishing two different senses of ‘can,’ explicitly employing the sort of strategy that (I claim) he used implicitly in the Fourth Meditation. Georges Moyal is correct in saying that this passage “provides the key to the unity of Descartes’ thoughts on freedom.”

But before declaring the two senses passage a smoking gun, we should pause to consider Kenny’s reading of it. Kenny takes the Meditations, Prinaples, and 1644 letter all to deny that two-way power is essential to freedom, and he thinks that this 1645 letter is no exception. He builds his case on the last sentence of the two senses passage. According to Kenny,

when Descartes says … that it is always open to us to hold back from pursuing a clearly known good, or from admitting a clearly perceived truth, he need not mean that we can do this at the very moment of perceiving the good and the true. Rather, we must distract our attention, as he said in the 1644 letter. One way of doing this would be to dwell on the thought that it would be a good thing to demonstrate our free will… this would render the perception of truth and goodness unclear. (“Descartes on the Will,” 18-19)

As Kenny reads them, neither letter to Mesland claims that we enjoy two-way power during clear perception. We enjoy it only after we have distracted our attention.

Kenny reads the last sentence of the “two senses” passage as a gloss on the procedure for suspending judgment outlined in the 1644 letter. This reading of the last sentence might be correct, but it does not support Kenny’s overall interpretation of the passage. Kenny ignores the distinction between moral and absolute senses of ‘can,’ but Descartes clearly intends the last sentence of the passage to elucidate that distinction. It explains why, absolutely speaking, the will has two-way power with respect to all its acts, including those that occur “when a very evident reason moves us in one direction” (and thus determines the will “morally speaking”). Therefore, the last sentence cannot be denying that two-way power is in some sense essential to freedom.

Furthermore, the last sentence is probably not merely a gloss on the 1644 remarks about suspending judgment. In the second Replies, Descartes distinguishes two classes of clear and distinct truths: some “are so transparently clear and at the same time so simple that we cannot ever think of them without believing them to be true” (these are clear and distinct per se), while others “are perceived very clearly by our intellect so long as we attend to the arguments on which our knowledge of them depends” (these are clear and distinct peraliud) (AT 7:145-46 / CSM 2:104). The 1644 passage explains how to suspend judgment with respect to pursuing an object that is good per aliud (by turning our attention “from the reasons which show us that the thing is good for us”), but does not show how we could suspend judgment with respect to objects that are per se clear and distinct. The “two-senses” passage, however, concerns an ability to hold back that we have with respect to both per se and per aliud clear truths. Thus it is unwise to use the 1644 passage as our primary lens for understanding the “two-senses” passage.

As I understand it, the “two senses” passage maintains that with respect to one and the same act of will, and at the same time, we can be both morally unable to hold back and absolutely able to hold back. Much about this moral/absolute distinction is puzzling, but I suspect that it maps onto the earlier distinction between two senses of ‘can’ in the Fourth Meditation: “moral” alternatives are alternatives of indifference, and “absolute” alternatives are alternative of self-determination. If I am right, then the moral/absolute distinction could mean at least three different things, corresponding to the three interpretations discussed above.

First, moral necessity may be a kind of deontic necessity. If so, Descartes’s claim that we are morally unable to hold back means that morality (or rationality) does not permit us to hold back, that we ought to act in accord with a very evident reason. Though this normative reading of moral necessity is compatible with the other two interpretations below, it has been stressed mainly by advocates of the “radical freedom” interpretation, who insist that for Descartes it is always psychologically possible for us to flout the rules of reason or morality (hence we can hold back “absolutely speaking”). On this reading, the phrase ‘we can hardly move in the contrary direction’ is making only a normative point, and does not mean that clear perceptions can psychologically determine assent. The last sentence of the passage (‘for it is always open to us to hold back … provided we consider it a good thing to demonstrate the freedom of our will’) seems to cause a problem for this reading: if we can hold back when our motives drive us in only one direction, then why does Descartes seem to make our ability to hold back depend on the presence of a countervailing motive? Perhaps Descartes means that such a motive is needed not to render holding back psychologically possible, but to render it rational.

When Descartes says that it is morally impossible to hold back from an evident reason, he may mean not only that we ought to act in accord with the reason, but also that it is psychologically impossible for us not to. On the “two-way compatibilist” reading, what is morally possible is what is psychologically possible in the actual circumstances of choice. Given my awareness of a very evident reason to do A, I cannot but do A. What is absolutely possible is what would (or could) have happened in a relevantly similar choice situation: if, in addition to being aware of A, I were also experiencing a countervailing motive (such as the desire to prove my freedom), I would be psychologically able to hold back from A.

The distinction is similar on the less radical incompatibilist interpretation, but with an important twist. On that reading, moral possibility would be psychological possibility in the actual choice situation. But to say that an action is “absolutely” possible for me would be to say both that I could have performed it in an alternative situation, and that in the past it was in my power-morally speaking-to bring about that alternative situation, or not. On this view, it is not enough that I could have held back if there had been a countervailing motive: I must also have been able to determine whether or not such a motive would be present.


I have argued that over time Descartes consistently believed in PAP, FEW, JVA, and CDD. I have also tried to show that the language of the Fourth Meditation suggests a distinction between two different kinds of alternatives: to say that we “can do or not do” means either that we are motivated in alternative directions, or that no external force determines our action. This distinction allows Descartes to claim that PAP and CDD can be true of one and the same action, because PAP requires alternatives in the latter sense, and CDD removes them only in the former sense. I have argued, further, that with his appeal to the difference between moral and absolute possibility in the 1645 letter to Mesland, Descartes attempts to make basically the same distinction. So Descartes not only consistently affirmed both PAP and CDD, but also consistently employed the same general strategy for reconciling them with each other.

Because it is not immediately obvious how Descartes understood the nature of “moral” and “absolute” alternatives, I have explored three different ways of cashing out the distinction between them. In doing so, I do not mean to suggest that Descartes had no clear or precise view about the nature of these alternatives (though that is possible, and it is also possible that his view of these alternatives developed across time). On the contrary, I think it likely that Descartes did have a clear and consistent view about the nature of moral and absolute alternatives. But I cannot argue for my opinion here, because doing so is beyond the scope of this paper. It would require making an involved textual case for one of the three interpretations discussed above, and that effort is best left for another time.

Some might object that my interpretation makes too much of the final letter to Mesland. Although in that letter Descartes purports to clarify the Meditations account of freedom, there is some reason to doubt his sincerity. Mesland was a Jesuit, and Jesuits were fierce defenders of PAP-like principles. Descartes may have simply been telling Mesland what he wanted to hear. For it seems that at that stage of his career, Descartes was trying to curry favor with Jesuits.

Though I am inclined to take Descartes at his word, I agree that it would be a dubious procedure to simply (as Vere Chappell says) “read the qualification expressed by the phrase ‘morally speaking’ back into Descartes’ earlier statement.” But that is not what I have done. I have argued that the Fourth Meditation itself suggests a distinction between two different kinds of alternatives. Indeed, I think considerations about the analogy between divine and human freedom would make this the most plausible reading of the Fourth Meditation even if we did not have the 1645 letter to Mesland. If I am correct, there is good reason to think that in the 1645 letter Descartes was sincere after all: he had accepted PAP all along.