1. Introduction
In two recent papers, Michael Della Rocca locates a distinct instance of circular reasoning in Descartes’ Fourth Meditation – i.e. distinct from the traditional ‘Cartesian Circle’ of the Third Meditation (Reference Della Rocca2011, Reference Della Rocca and Gaukroger2006).Footnote 1 This new circle, Della Rocca alleges, is an even more intractable fallacy: it is a bona fide instance of premise circularity in which the ‘truth rule’ (‘everything I clearly and distinctly perceive is true’) functions as a sub-premise in its own proof. While Della Rocca’s provocative reading has garnered substantial attention in the literature (see, e.g. Carriero Reference Carriero2009, 278–9, 462–3; Christofidou Reference Christofidou2013; 245 n. 10; Naaman-Zauderer Reference Naaman-Zauderer2010, 98–100), it has not yet received the refutation it deserves. We supply that refutation here. We show that even granting Della Rocca’s interpretation of the Fourth Meditation, his charge of circularity is unwarranted.
2. The problem of meditation IV
Della Rocca begins his interpretation of the Fourth Meditation by noticing that it is structurally analogous to the Third Meditation. He writes,
After [Descartes] lays ‘it down as a general rule’ near the beginning of Meditation III ‘that whatever I perceive very clearly and distinctly is true’ [AT VII 35, CSM II 24]Footnote 2, Descartes immediately casts this rule into doubt by raising the possibility that God is a deceiver. So, too, after finally offering an argument later in Meditation III for the claim that God is not a deceiver, Descartes immediately goes on in Meditation IV to raise a doubt about this claim. (Reference Della Rocca and Gaukroger2006, 142–3, Reference Della Rocca2011, 94)
Descartes raises this worry about God’s non-deceptiveness in the following passage:
[S]ince God does not wish to deceive me, he surely did not give me the kind of faculty which would ever enable me to go wrong while using it correctly. There would be no further doubt [dubium] on this issue were it not that what I have just said appears to imply that I am incapable of ever going wrong. For if everything that is in me comes from God, and he did not endow me with a faculty for making mistakes, it appears that I can never go wrong. (AT VII 54, CSM II 37–8)
Della Rocca glosses the problem here as follows:
But obviously, as Descartes recognizes, I do sometimes make mistakes, and so the argument for the claim that God is not a deceiver and that clear and distinct ideas are true is called into doubt. Descartes will not have removed this doubt and beaten back the skeptic about clear and distinct ideas until he has shown how it is compatible with God’s nondeceptiveness that we sometimes err. (Reference Della Rocca and Gaukroger2006, 143, Reference Della Rocca2011, 94)
To resolve this problem, Descartes must ‘answer the following question: How, if God is not a deceiver and thus does not allow clear and distinct ideas to be false, can God allow that I ever make mistakes, in particular with regard to ideas that are not clear and distinct?’ (Reference Della Rocca and Gaukroger2006, 143, Reference Della Rocca2011, 94). In other words, Descartes needs to show that while God’s veracity (‘God is not a deceiver’) entails the truth rule (‘clear and distinct ideas are true’), it does not entail human infallibility.Footnote 3
Descartes blocks the inference to human infallibility with a ‘freewill defense’ that appeals to a distinction between ‘user-error’ and ‘manufacturer-error’Footnote 4 along with a norm of correct assent. Descartes reasons that if we judge correctly and go wrong, then our faculty of judgment must have a design or manufacturing problem, for which God would be culpable. But if we ignore our manufacturer’s instructions for correct use – if the cause of error lies not in the faculties themselves but in our misuse of those faculties – then God is not to blame. Descartes thinks God has given us the following clear instruction for correctly using our faculty of judgment:
[I]t is clear by the natural light that the perception of the intellect should always precede the determination of the will. (AT VII 60, CSM II 41)
Della Rocca paraphrases this deliverance of the natural light as follows:
Norm of Assent (NA): “[W]e should assent only to clear and distinct ideas”. (Reference Della Rocca and Gaukroger2006, 157)Footnote 5
We are free to follow NA or to flout it. When we flout it, we place ourselves outside of the divine guarantee by using our faculties contrary to the manner God intends. The result is a ‘user-error’ for which the ‘manufacturer’ of our faculties bears no responsibility. Thus, God’s goodness does not entail total infallibility, but only infallibility when we judge correctly.
3. The case for the new circle
Della Rocca’s circularity charge begins by asking Descartes to justify the norm of assent. Della Rocca asks: ‘Why, for Descartes, should we assent only to clear and distinct ideas? Without a good reason for this claim, Descartes will lack an effective way of putting to rest his doubt about clear and distinct ideas’ (Reference Della Rocca and Gaukroger2006, 157). Della Rocca then contends that Descartes realized this need and attempted to justify NA in the following passage:
But if in such [non-clear and distinct] cases I either affirm or deny, then I am not using my free will correctly. If I go for the alternative which is false, then obviously I shall be in error; if I take the other side, then it is by pure chance that I arrive at the truth. (AT VII 59–60, CSM II 41)
According to Della Rocca, Descartes here affirms that we should not assent to non-clear-and-distinct perceptions because such judgments would be either false or ‘at best accidentally true’ (Reference Della Rocca and Gaukroger2006, 157). Conversely, clear and distinct ideas are neither false nor accidently true. From which it followsFootnote 6 that clear and distinct ideas are true and guaranteed to be so. Thus, Della Rocca concludes:
The reason Descartes offers [to justify the norm of assent] seems to be ... [that] clear and distinct ideas are guaranteed to be true and non-clear and distinct are not. (Reference Della Rocca and Gaukroger2006, 157)
But offering this justification for NA is a fatal error. The truth rule, recall, says that all clear and distinct ideas are true. But here Descartes is appealing to an even stronger notion – not only are all clear and distinct ideas true, but they are guaranteed to be so and only they are so guaranteed. We will refer to this strengthened version of the truth rule as the
Super Truth Rule (STR): “[O]nly clear and distinct ideas ... are guaranteed to be true”. (Della Rocca Reference Della Rocca and Gaukroger2006, 157)
According to Della Rocca, Descartes derives NA from STR. But given the purported aim of Meditation IV, grounding the norm of assent in this way begs the question. As Della Rocca explains,
One of the premises needed for arguing that clear and distinct ideas are true is ... that we should assent only to clear and distinct ideas. This claim in turn requires argument, and the argument for it seems to be based on the claim that clear and distinct ideas are guaranteed to be true. So, one of the premises for the claim that clear and distinct ideas are true is that clear and distinct ideas are guaranteed to be true. Here the conclusion – indeed, a strengthened version of the conclusion – is itself a premise in the argument. (Reference Della Rocca and Gaukroger2006, 158–9)
And therefore, Della Rocca concludes, Descartes is guilty of a particularly vicious form of circular reasoning in the Fourth Meditation: here, the truth rule is not just prematurely trusted – it is used as a premise to prove itself.
4. Our initial refutation
We begin with a fairly simple argument. Notice the structure of the problem Descartes faces: the goodness of God and the proof of the truth rule are called into question because God’s veracity (G) seems to entail the outlandish claim that humans are totally infallible (I). To solve this problem, Descartes merely needs to show that (I) does not follow from (G). And since (I) is inconsistent with the actuality of human error (E), showing that (G) does not entail (I) is equivalent to showing that the following set is consistent: {(G), (E)}.
Furthermore, demonstrating the consistency of a set of propositions only requires appeal to the logical relations that hold among the members of that set. It does not require any additional factual claims. That is to say, the entire discourse of such a project is non-committal about the truth of the notions it employs. Descartes need only invoke elements of his ‘freewill defense’ – including NA and whatever grounds it – as hypotheses for the purposes of showing that (E) is consistent with (G). Even if Descartes does appeal to STR to ground NA, such an appeal does not commit him to the truth of STR. But, if Descartes does not need to endorse the truth of STR, then his use of STR does not presuppose the final conclusion of his argument – the truth rule. Thus, there is no circle.
5. An analogical rejoinder
In Della Rocca’s second paper (Reference Della Rocca2011), he responds to something like this objection: ‘[o]ne might argue,’ he notes, that ‘all Descartes needs to do is to show that God’s veracity, and thus the guaranteed truth of clear and distinct ideas, is consistent with our occasional errors with regard to ideas that are not clear and distinct’ (99). However, he quickly retorts: ‘But again, the point is: How does Descartes show this consistency?’ (99). He does so, Della Rocca insists, by illegitimately and prematurely relying on the claim that clear and distinct ideas are guaranteed to be true. Della Rocca makes his point by way of the following analogy:
Imagine trying to reconcile one’s theory that Joe is the murderer with apparently conflicting DNA evidence that points to someone else, Susie. You reconcile these by claiming that the DNA evidence is misleading. But, let us say, your only basis for claiming that the DNA evidence is misleading is that Joe (and not Susie) is the murderer … In this case, you have not really defended your theory in light of the DNA evidence. What you need is a way of discounting the apparent counterevidence, a way that is independent of the claim that Joe is the murderer … To avoid circularity, Descartes needs an independent way of reconciling occasional errors, on the one hand, with the guaranteed truth of clear and distinct ideas and God’s veracity, on the other. (Reference Della Rocca2011, 99)
Despite its intuitive pull, Della Rocca’s Joe and Susie analogy misrepresents Descartes’ dialectical situation. The analogy is based on the inductive, probabilistic reasoning of courtrooms. If the attorney appeals to ‘Joe is the murderer’ in response to his opponent, he must assume the actual truth of that claim and so beg the question. But the same does not hold for Descartes since the problem Descartes faces is not inductive. What Descartes must overcome is that the reasoning that leads to the truth rule seems also to entail human infallibility (I). That is to say, there is a purported logically necessary connection between God’s veracity (G) and (I).Footnote 7 Hence, Descartes’ dialectical situation differs considerably from the attorney seeking to prosecute Joe.
Once one sees the switch of dialectical contexts, the analogy loses its force. Given that Descartes must only demonstrate the logical consistency of the set {(G), (E)}, he must do no more than show (semantically speaking) that there is at least one logically possible truth value assignment that would make all statements of the set true. Descartes need not say anything about the actual truth values of the statements he invokes. And since the super truth rule is a logical possibility, there is nothing illegitimate about assuming it. Therefore, even if Descartes grounds NA on STR for the purposes of blocking the inference to (I), he does not beg the question thereby.
6. A modal rejoinder
Della Rocca has responded to our last line of reasoning by noting that on the most commonly used modal logic – System S5, which he believes Descartes would accept – the possibility of the super truth rule entails the actuality of the truth rule (2014).Footnote 8 And thus even given our defense above, Descartes would still be guilty of circularity.
To see Della Rocca’s rejoinder, consider that the truth rule (TR) says that all clear and distinct ideas are true, and the super truth rule (STR) says that all (and only) clear and distinct ideas are guaranteed to be true (N.B. guaranteed = necessary, for Della Rocca). Therefore, STR asserts the necessity of TR (□TR). So, if Descartes assumes the mere possibility of the super truth rule, he thereby assumes:
◊□TR
This means that at some accessible possible world – call it world 2 – the truth rule is necessarily true, i.e. (□TR, w2) = T. Now, System S5 has Euclidean frame conditions, which entails that accessibility relations are symmetric. Thus, since world 1 (the actual world) accesses world 2, it follows that world 2 must also access world 1. And thus, since □TR is true at world 2 it would follow by the valuation rule for ‘□’ that TR is true at the actual world. So on System S5,
◊□TR ⊢ TR
Hence, Della Rocca concludes, even if Descartes merely assumes the super truth rule as a possibility, he thereby assumes the actual truth of the truth rule, and thus begs the question.
We wonder about attributing symmetric frame conditions to Descartes. With his creation doctrine of eternal truths, Descartes seems to posit two levels of modal truths – absolute and restricted to creation – that do not access each other (see, e.g. [AT I 152, CSMK 25]; [AT V 224, CSMK 358–9]; and [AT IV 118, CSMK 235]). And this would imply non-symmetric frame conditions.Footnote 9 But even if we set that thorny issue aside, Della Rocca’s objection is still flawed in a more straightforward sense. With or without S5, the way in which Descartes must regard STR, for purposes of showing set consistency, is not accurately represented as a full-blooded modal commitment to the possibility of STR in the actual world.
Notice that whether or not a set is consistent is wholly a matter of the terms that are already members of the set. Sets can be made inconsistent by the addition of new members, but inconsistent sets cannot be made consistent via addition. The set {p, ~p, q} will always remain inconsistent no matter how many terms are added. Thus, when Descartes demonstrates to the reader that error is consistent with his premises, he does not, and cannot, make the set consistent, if it is not already so. For Descartes to succeed, the set must already be consistent.
But if the set is already consistent, then what is Descartes doing with his story about our ability to flout the norm of assent? At least in terms of the project to ‘block’ the apparent entailment to (I), all such talk can only be pedagogical. Descartes is merely painting a conceptual backdrop that makes salient to the reader those aspects of the set {(G), (E)} that render it consistent.Footnote 10 All features of such a hypothetical story can be fully disposed of once they achieve their purpose – i.e. once the reader sees that the set is consistent.
Moreover, the story Della Rocca attributes to Descartes (including STR) is not the only one available for this purpose. A story without STR could work just as well. For example, Descartes could say that humans are unable to withhold assent to clear and distinct perceptions, but are able to withhold it from non-clear and distinct perceptions.Footnote 11 Given that ‘ought implies can,’ this difference in our abilities would explain why NA obliges us to withhold assent from non-clear and distinct ideas but not from clear and distinct ones. This alternative grounding for NA, along with other features of Descartes’ freewill defense, would show that (G) and (E) can both be simultaneously true.Footnote 12
Hence, if Descartes chooses to use STR as part of his hypothetical story, this is merely a rhetorical choice.Footnote 13 By using such a pedagogical aid, Descartes does not commit himself to any metaphysical truths at all, not even to the truth of any possibility claims. Therefore, even if symmetric frame conditions were appropriate for Cartesian modalities (see n. 9), there would still be nothing viciously circular about including STR in his hypothetical story.
7. Conclusion
Michael Della Rocca’s ‘new circle’ is based on several interpretive claims: (1) while attempting to prove the truth rule, Descartes realizes that his premises seem to lead to a patently false result – namely that we are infallible. (2) Descartes must first solve this puzzle in order to continue his proof of the truth rule. (3) In the course of solving this puzzle, Descartes relies on a norm of correct assent (NA). And, (4) Descartes grounds NA on the super truth rule (STR) – all and only clear and distinct ideas are guaranteed to be true. But, (5) STR is just a strengthened version of the truth rule. Thus, Della Rocca infers that by grounding NA in this way, Descartes is guilty of premise circularity – he literally uses the truth rule to prove itself.
In the refutation above we have intentionally avoided contesting any of Della Rocca’s core interpretive claims (1) through (5). Instead, we have shown that even granting Della Rocca his reading of Meditation IV, it does not follow that Descartes is guilty of premise circularity. The reason, in short, is that the project to show that an entailment does not hold is merely the project to show a given set is consistent. And this project only requires true claims about the logical relations that hold among the members of the set. It requires no commitment to truths about reality beyond. Hence, both Della Rocca’s Joe and Susie analogy and his modal rejoinder are misplaced, and, Della Rocca has failed to find a new instance of circularity in Meditation IV.
