Skip to content

Descartes, Epistemic Principles, Epistemic Circularity, and Scientia

Keith DeRose
At the very start of the Meditations,(1) Descartes writes that his reason for “making a clean sweep” and “beginning again from the very foundations” of knowledge is that by so doing he hopes to “establish some secure and lasting result in science” (AT, 17; AG, 61; emphasis added). In order to attain scientific knowledge (scientia), or what he calls in Meditation V “perfect knowledge” (perfecte scire), Descartes entertains and seeks to remove what he calls a “metaphysical” reason for doubting all that he had previously thought he knew.(2) This reason for doubt is that perhaps he has a nature so radically defective that he is liable to go wrong about anything, even what seems most evident to him.(3) By entertaining so universal a doubt, Descartes seems to have made an alarmingly clean sweep indeed — so clean as to deprive him of any foundations from which to begin. Yet, seemingly oblivious to this problem, he proceeds to build away from the (suspect?) foundation of his clear and distinct perceptions until, as part of his superstructure, he obtains the result that a veracious God exists, which, in turn, is used to prove that Descartes’s clear and distinct perceptions are true after all. By thus using his clear and distinct perceptions to validate those very clear and distinct perceptions, Descartes famously appears to have engaged in questionably circular reasoning.

Since Descartes was an eminently smart person, a desideratum of an interpretation of the Meditations is that it not have Descartes making use of a viciously circular method of validating his clear and distinct perceptions, or, at least, that it not make Descartes’s method out to be obviously circular in an obviously vicious way. In what follows, I defend an interpretation which fulfills this desideratum by means of a certain understanding of what Descartes took scientia to be. By clearly identifying the goal of the Meditations, we will be able to see that Descartes’s progress toward that goal does not involve him in vicious circularity, or, at least, not in any obviously vicious circularity.

The solution to the problem of the alleged Cartesian Circle that I will present (in section B) and defend (in sections C-H) is by no means strikingly novel; it resembles, more or less closely, various other “two level” solutions — as I will call them — that have appeared in the already expansive literature on the Cartesian Circle.(4) My goal, then, is not to present a radically new way of interpreting Descartes. My main task, rather, is the more philosophical than scholarly one of exploring the implications of the view that results when Descartes is interpreted in my way for two intimately related issues central to understanding the Meditations. For the most important attacks to which my and other two-level solutions appear to be vulnerable take the form of claims to the effect that Descartes, interpreted in this two-level way, is up to something silly. By looking carefully at (1) the type of guarantee of truth Descartes is (and the type he is not) seeking and (2) the kind of sceptic Descartes is (and the kind he is not) attempting to defeat, I will show that, interpreted in my way, Descartes may well be up to something rather sensible. Closely related to this main task is a secondary one: to argue that my solution is superior to a very different type of solution put forward by James Van Cleve and described below in section A.

A. Van Cleve’s Solution

Van Cleve handily summarizes the problem of the Cartesian Circle as arising for Descartes because Descartes appeared to commit himself to each of the following propositions:
(1) I can know (be certain) that (P) whatever I perceive clearly and distinctly is true, only if I first know (am certain) that (Q) God exists and is not a deceiver.(2) I can know (be certain) that (Q) God exists and is not a deceiver, only if I first know (am certain) that (P) whatever I perceive clearly and distinctly is true.(5)

Van Cleve’s own solution to the problem is built upon a distinction, which he derives from Anthony Kenny, between two propositions, either of which can be expressed by the sentence, “I am certain of the truth of clear and distinct perceptions”:
(A) For all p, if I clearly and distinctly perceive that p, then I am certain that p.(B) I am certain that (for all p, if I clearly and distinctly perceive that p, then p).

Van Cleve explains:
The difference is that (A) says that whenever I clearly and distinctly perceive any proposition, I will be certain of it (the proposition in question), whereas (B) says that I am certain of a general principle connecting clear and distinct perception with truth. (p. 67)

Van Cleve claims, again following Kenny, that this distinction can make sense of
the notorious fourth paragraph in the Third Meditation, where Descartes appears to oscillate inconsistently between saying, on the one hand, God or no God, I am certain of things when I clearly and distinctly perceive them, and, on the other hand, I can doubt even the truth of clear and distinct perceptions if I do not know that there is a veracious God.(6)

On Van Cleve’s interpretation, Descartes does take (A) to be true of himself even early on in the Meditations, so Descartes is not committed to (2) (what Van Cleve calls the “lower arc” of the circle), but Descartes’s apparent commitment to (2) — in the fourth paragraph of Meditation III, as elsewhere — can be explained by reference to the fact that, on this interpretation, Descartes does not take (B), which is easily confused with (A), to be true of himself until late in the Meditations, after the existence of a veracious God is proved.

On Van Cleve’s interpretation, then, Descartes never considers particular clear and distinct perceptions, like 2+3=5, to be in any way doubtful; he only takes to be doubtful the general proposition that what he clearly and distinctly perceives is true. Since these particular clear and distinct perceptions are never cast into doubt, they are available for legitimate use as premises from which to prove the general principle which is initially cast into doubt.

While Van Cleve’s interpretation can make sense of most of the troublesome fourth paragraph of Meditation III, it has severe problems with the end of Meditation I, where Descartes’s deceiving God hypothesis seems to cast doubt upon particular clear and distinct perceptions:
Moreover, I judge that other men sometimes go wrong over what they think they know perfectly well; may not God likewise make me go wrong, whenever I add two and three, or count the sides of a square, or do any simpler thing that might be imagined? (AT, 21; AG, 64)

While the above passage is, I believe, most naturally read as expressing doubts as to whether 2+3=5, whether squares have four sides, and whether Descartes might also be wrong about other (perhaps even simpler) particular clear and distinct perceptions, it can, I suppose, be construed merely as a doubt about the general principle that what he clearly and distinctly perceives is true.(7) (Also, the difficulty in counting the sides of a square could be construed as a possible failure to recognize that a particular figure is a square, rather than as involving a possible mistake concerning the proposition All squares have four sides.) But a few sentences later, Descartes makes it quite clear that all his beliefs are being cast into doubt:
I have no answer to these arguments, but am finally compelled to admit that there is not one of my former beliefs about which a doubt may not properly be raised; and this is not a flippant or ill-considered conclusion, but is based on powerful and well thought-out reasons.(8)

This passage is clear on the point at issue. It is not just a general epistemic principle which is cast into proper doubt: none of Descartes previous beliefs, including, presumably, the just previously considered 2+3=5 and Squares have four sides, escape this doubt. What’s more, even the fourth paragraph of Meditation III, the very paragraph Van Cleve attempts to explain in order to support his interpretation, contains a sentence which Van Cleve cannot account for — in fact, this sentence is the most conclusive piece of evidence against his interpretation to be found in the Meditations. Van Cleve admits:
It must be said, however, that the final sentence of this paragraph — “Without a knowledge of these two truths [God exists and is not a deceiver] I do not see that I can ever be certain of anything” — is an embarrassment for almost any interpretation of Descartes. (p. 67, fn. 30)

But this sentence is only an embarrassment for interpretations which, like Van Cleve’s, construe Descartes as never considering particular clear and distinct perceptions like 2+3=5 as being in any way doubtful and dependent on a proof of God’s veracity to escape this doubt. It is now time for me to present what I think to be the correct interpretation of Descartes on this point — an interpretation for which neither the sentence Van Cleve mentions nor the texts from Meditation I that I cite above are embarrassments.

B. A Two-Level Solution

I have presented and discussed Van Cleve’s solution at some length not only because it is an exemplary case of the type of solution I wish to oppose, but also because much of Van Cleve’s solution can be re-cast into a new framework and will thereby become part of what I think is the correct interpretation.

I start, then, by accepting Van Cleve’s account of how Descartes uses particular clear and distinct perceptions to establish, as a clear and distinct perception, the general truth that Whatever I perceive clearly and distinctly is true.

But we can avoid the drawbacks of Van Cleve’s interpretation (while retaining its insights) by construing Descartes as aiming for something stronger than “mere” clear and distinct perceptions. Descartes’s goal is to attain scientia or perfect knowledge, and clearly and distinctly perceiving that p is true is a necessary, but not a sufficient, condition for having scientia of p. What else is needed? Everything falls into place if we answer this question as follows: One also has to clearly and distinctly perceive to be true the general principle that whatever one perceives clearly and distinctly is true. So, I provisionally put forward the following account of Descartes’s notion of scientia (a refinement will be discussed at the end of section C, below(9)):
S has scientia of p if and only if (1) S clearly and distinctly perceives that p is true and (2) S clearly and distinctly perceives the truth of the general principle that what S clearly and distinctly perceives is true.

The resulting solution to the problem of the circle involves us in two levels of certainty. The lower level is that provided by clear and distinct perception. This is attained in the linear, non-circular manner described by Van Cleve; if we replace “know (be certain) that” in Van Cleve’s (1) and (2) (see the very beginning of Part A, above) with “clearly and distinctly perceive that”, we follow Van Cleve in denying that Descartes accepts (2).

But one’s belief that p remains somewhat doubtful(10) and does not attain the higher level of scientia unless and until one also clearly and distinctly perceives the general principle to be true. Thus, unlike Van Cleve’s solution, this two-level interpretation can explain why Descartes treats particular clear and distinct perceptions as being not completely up to epistemic snuff early on in the Meditations but as attaining the status Descartes seeks (scientia) through Descartes’s theistic labors. Once one comes to clearly and distinctly perceive the general principle to be true, then all of one’s clear and distinct perceptions are raised to the higher level of scientia. If we replace ‘know (be certain) that’ with ‘have scientia of’ in Van Cleve’s (1) and (2), we can deny them both. One needn’t have scientia of P before one can have scientia of Q, nor Q before P. Rather, when one comes to clearly and distinctly perceive P (the general principle) to be true, then it and everything else that one clearly and distinctly perceives (including Q) becomes scientia for one at the same time.

When Descartes explains what he means by scientia, he does not list the conditions I put forward above, but rather describes scientia in terms of a lack of reasons for doubt or invulnerability to sceptical attack. As I now turn to objections to my interpretation, this is the second objection I will address. I will claim that what Descartes explicitly says about scientia is best construed as amounting to the same thing as the account given above. I delay this objection both because the first objection I treat springs to the minds of many in such an urgent way as to demand immediate treatment and also because this first objection is most vividly addressed to the account of scientia I have given above and not to Descartes’s own explanation.

C. The First Objection: Scientia, as I describe it, is not a higher level of knowledge or certainty at all

This objection is made most vivid by considering a made-up parallel notion of Pythientific knowledge, defined as follows:
S Pythientifically knows that p if and only if (1) S clearly and distinctly perceives that p is true and (2) S clearly and distinctly perceives that the Pythagorean theorem is true.

Suppose I can immediately (without proof) clearly and distinctly perceive that 2+3=5, but cannot immediately clearly and distinctly perceive the Pythagorean theorem to be true and know of no proof for its truth. In such a state, I do not Pythientifically know that 2+3=5. How can I come to Pythientifically know it? I will have to construct or understand a proof for the Pythagorean theorem from premises I do clearly and distinctly perceive to be true. Once I thereby come to clearly and distinctly perceive that the Pythagorean theorem is true, my clear and distinct perception that 2+3=5 is “raised” to the level of Pythientific knowledge.

But, our objector will very plausibly insist, my belief that 2+3=5 hasn’t been raised to a higher level at all: it is none the better for my having come to clearly and distinctly perceive that the Pythagorean theorem is true. I have expanded the range of what I clearly and distinctly perceive to be true, but I haven’t thereby raised the epistemic status of what I already clearly and distinctly perceived to be true before the process of constructing the proof.

Likewise, the objection continues, even if Descartes succeeds in proving the general principle that whatever he clearly and distinctly perceives is true, this only expands the range of his clear and distinct perceptions, it does not raise the epistemic status of his belief that 2+3=5.(11) If you already clearly and distinctly perceive something to be true, you cannot raise the status of that belief to something higher than that of clear and distinct perception just by coming to clearly and distinctly something else to be true. Just as Pythientifically knowing that 2+3=5 is no better than clearly and distinctly perceiving that 2+3=5 is true, so also is scientifically knowing no better.

My response is that while the Pythagorean theorem is not related to one’s belief that 2+3=5 in any way that makes it plausible to suppose that knowing the former puts one in any better epistemic position with respect to the latter, the same cannot be said about the general proposition that what one clearly and distinctly perceives is true. This general proposition is intimately related to one’s belief that 2+3=5: It is the epistemic principle behind that belief.

An epistemic principle is a statement about the reliability of a particular faculty, belief-forming mechanism, or way of forming beliefs. Thus, an epistemic principle will state that the beliefs formed in a particular way are all true or, alternatively, that they are mostly true. Typically, epistemic principles are not considered by the subjects of the beliefs in question. Thus, when I come to believe that there is a book in front of me by looking at it, I typically entertain no thoughts about the reliability of my faculty of sense perception, or, to divide things more finely, of my faculty of vision. But, interestingly, even though a particular belief was formed with no thought about any relevant epistemic principle, the belief can be undermined and thereby be cast into doubt by casting doubt on the principle behind it.(12) If, for example, I am given a strong reason to believe that my senses are thoroughly unreliable, the beliefs I form through the senses all seem to suffer an epistemic demotion.

Still more interesting is the fact that Descartes’s own sceptical arguments of Meditation I all seek to thus undermine, rather than to oppose, their target beliefs. This difference can best be made clear by means of an example. Suppose a teacher, while giving a final exam, is grading the course papers of the students taking the final. Frank approaches the teacher to ask a question about the final and notices that it is his paper that the teacher is presently reading. For the next several minutes, Frank glances up at his teacher from time to time, and notices that the teacher is scowling as he reads, causing Frank to believe that the teacher strongly dislikes his paper. Later that evening, two friends each comfort Frank by giving him reason to doubt his distressing belief. One friend relates to Frank that he has just come from the philosophy department where he overheard Frank’s teacher praising his paper to a colleague. This friend opposes Frank’s belief that his teacher dislikes his paper by giving him reason to believe the opposite. A second friend, who knows Frank’s teacher very well, comforts Frank by informing him that his teacher always scowls as he reads papers, even papers he likes very much. The teacher’s scowling is not an indication of disapproval (although it is often mistaken for such), but rather of concentration. This friend gives Frank no reason to think that his teacher likes his paper, but he does cast considerable doubt upon Frank’s belief nonetheless. By giving Frank reason to think that the method by which he came to have the belief is unreliable, this friend has undermined, rather than opposed, Frank’s belief.

Descartes’s sceptical arguments likewise cast beliefs into doubt by targeting the principles behind them — by calling into question the reliability of the processes by which the beliefs are formed rather than by giving reason to believe the opposite. Given the fact that (as Descartes was well aware) an attack upon the principle behind a belief can constitute a successful undermining of that belief itself, we can see that the status of a belief is intimately tied to the status the epistemic principle behind the belief has for one. Thus, it is plausible to suppose that shoring up the epistemic principle will result in some significant epistemic promotion for the beliefs that result from the relevant faculty.

Thus, if one already clearly and distinctly perceives that 2+3=5, while coming to clearly and distinctly perceive that the Pythagorean theorem is true obviously bestows no significant further epistemic virtue upon one’s belief that 2+3=5, coming to clearly and distinctly perceive the epistemic principle behind the belief that 2+3=5 is not so clearly impotent for this task. One epistemic virtue that one’s belief that 2+3=5 may thereby gain is that it becomes less vulnerable to undermining sceptical attacks. And one might well consider this a significant epistemic virtue. Furthermore, it is clear that Descartes thought this to be a very significant epistemic virtue indeed, as we will see in the following section.

One could in principle satisfy the two conditions for scientia I propose in section B while being completely oblivious to there being any connection between the general epistemic principle and one’s particular clear and distinct perception that, say, 2+3=5. Here, it seems, one’s clear and distinct perception that 2+3=5 would not enjoy much (if any) added epistemic virtue for one’s having come to clearly and distinctly perceive the general principle to be true, and Descartes, I thus strongly suspect, would not want to count someone in such a state as having scientia of the fact that 2+3=5. My account of scientia in section B is then incomplete. In order to have scientia of p, one must also meet some further requirement to the effect that one recognizes the connection between the general principle that what one clearly and distinctly perceives is true and one’s belief that p. That one realize that the general principle is the epistemic principle behind one’s belief that p and therefore be disposed to fend off certain undermining sceptical attacks directed at one’s belief that p by appealing to one’s knowledge of the general principle, or something close to this, should, no doubt, be part of this further requirement. But, having already gone far beyond anything to be found in any Descartes text, I will leave open the precise nature of this further requirement.

D. The Second Objection: Reasons for Doubt, Scientia, and the Atheist Geometer

In order to meet the objection registered and fulfill the promise made in the last paragraph of section B, it is now time to connect the conditions for scientia I have proposed to what Descartes himself has to say about scientia. Any treatment of the circle should take account of the “atheist geometer” passage in Descartes’s Reply to Objections II:
That an atheist can know clearly that the three angles of a triangle are equal to two right angles, I do not deny, I merely affirm that, on the other hand, such knowledge on his part cannot constitute true science, because no knowledge that can be rendered doubtful should be called science. Since he is, as supposed, an Atheist, he cannot be sure that he is not deceived in the things that seem most evident to him, as has been sufficiently shown; and though perchance the doubt does not occur to him, nevertheless it may come up, if he examine the matter, or if another suggests it; he can never be safe from it unless he first recognizes the existence of a God.(13)

In no other passage does Descartes give us as much insight into the relation between belief in God’s existence and the epistemic status of one’s beliefs in particular clear and distinct perceptions. Note in particular three features of this important passage. First, Descartes’s evaluation of the atheist geometer’s belief in the geometrical theorem (which, presumably, the geometer has seen proven) is largely positive: Descartes concedes to his objector that this atheist can “know clearly” that it is true. Second, the status this atheist’s belief lacks is clearly identified by Descartes as scientia. Thus, it is very plausible to suppose that this is the status Descartes seeks for his own beliefs in the Meditations — the status that depends upon his knowing that God exists. Third, the reason Descartes cites for why the atheist’s belief cannot be considered a piece of scientia is that it is vulnerable to sceptical attack: it “can be rendered doubtful.” The sceptical attack to which the atheist’s belief is vulnerable seems to be an undermining attack aimed at the principle that what’s most evident to him is true: This atheist “cannot be sure that he is not deceived in the things that seem most evident to him.” In short, the atheist is vulnerable to Descartes’s metaphysical reason for doubt.

Elsewhere, in a letter to Regius (May 24, 1640), immediately after having cited the doubt that “perhaps our nature is such that we go wrong even in the most evident matters,” Descartes requires, for a belief to count as scientia, that there be no “reason which might lead us to doubt” (K, 74).

But what will Descartes count as a reason for doubt? And just how safe from sceptical attack must a belief be (and in exactly what way must it be safe) for it to qualify as scientia? Well, why would Descartes think that the atheist, but not himself, is subject to the doubt that his nature is so radically defective that he is liable to go wrong about anything, even about what he clearly and distinctly perceives? Because, apparently, Descartes has, by means of a proof going through the existence of God, come to clearly and distinctly perceive that his nature is not deceptive in the way the above reason for doubt alleges. Thus, it seems, Descartes takes it as a necessary condition for some possibility r (e.g., My nature is so defective. . .), to provide a subject S with a reason for doubting p, that S not clearly and distinctly perceive that r is false. Since Descartes does clearly and distinctly perceive that the deceiving nature hypothesis is false, it does not count, for him, as a reason for doubt. The atheist, on the other hand, cannot clearly and distinctly see that this hypothesis is false. Furthermore, a proposition, r, need not be incompatible with another proposition, p, in order to provide S with a reason for doubting that p; r need not oppose p, it can also undermine p by being incompatible with the epistemic principle behind p. The deceiving nature hypothesis is incompatible with the epistemic principle behind the atheist’s clear and distinct perceptions, and thus provides the atheist with a reason for doubting them.

Given the lack of specifics in the texts, I think it would be a mistake here to try to specify a set of individually necessary and jointly sufficient conditions for when, on Descartes’s view, r provides S with a reason for doubting p,(14) and then to make explicit the relation between not having a reason for doubting p and having scientia that p. Things will get very messy, and we will very quickly go too far beyond anything Descartes ever said for the project to be worthwhile, since what is ultimately of interest are the conditions for when p is scientia for S.

Descartes never gives explicit conditions for when S has scientia that p, either. But the conditions I cite above in Part B (and as refined in the direction discussed at the end of Part C) provide an attractive account of Descartes’s procedure in the Meditations, and can now also be seen to hook up with the hints, in terms of lack of reasons for doubt, that Descartes does give about what it takes to have scientia. The advantage Descartes thinks he has over the atheist geometer is that he is, while the atheist geometer is not, able to, in a certain way, fend off the metaphysical reason for doubt. If the doubt were to occur to the atheist, he would just have to admit that, as far as what he clearly and distinctly perceives goes, he may have such a deceptive nature. Descartes, on the other hand, thinks he can clearly and distinctly perceive that his nature is not faulty in the way alleged, because he can, while the atheist cannot, prove this about himself.

E. Epistemic Circularity

Question: But what kind of “proof” could this be? Even if Descartes could successfully argue from clear and distinct perceptions that his faculty of clear and distinct perception is reliable,(15) wouldn’t this argument “beg the question”?

Answer: This is what William Alston has called an epistemically circular argument (see Alston, 1986). But this is just to give a label to a type of argument for the reliability of a way of forming beliefs that takes as its premises the deliverences of that very way of forming beliefs. What is of real interest is that it is not at all clear that such an argument is powerless to improve one’s epistemic position vis-a-vis its conclusion. Alston, in fact, argues that one can use an epistemically circular argument to justify one’s belief in its conclusion (pp. 11-14) and that one could use such an argument to successfully show or establish its conclusion (p. 15).

While Alston’s arguments are plausible, these results (especially the latter concerning showing or establishing) are at the same time questionable, I believe. Still, a brief discussion will suffice to show that it is very far from clear that Descartes’s epistemically circular argument would be useless for the purpose of improving Descartes’s epistemic position vis-a-vis its conclusion.

First, recall the results of Part C above. Unlike the type of circularity described by Van Cleve (see the very beginning of Part A, above), I am not ascribing to Descartes a view according to which the following three mutually inconsistent claims hold for any two propositions p and q and any single epistemic status J:
1. My belief that p can have status J only if my belief that q first has status J.2. My belief that q can have status J only if my belief that p first has status J.3. My belief that p and/or my belief that q is of status J.

Claims (1) and (2) above set up a circularity between one’s belief in p and one’s belief in q. This circularity is clearly vicious: It obviously cuts one off from attaining status J for either belief. In contrast, on my interpretation, as I pointed out in Part C, where p and q are Van Cleve’s P and Q, then if J is clearly and distinctly perceiving, Descartes can deny (2), and if J is having scientia, he can deny both (1) and (2). Thus, the epistemic circularity I ascribe to Descartes is not so clearly vicious. His claims to having attained scientia do not generate any easily seen contradiction with any other commitments I ascribe to him.

Next, as Alston has pointed out (Alston, 1985, p. 449), we should remember that the use of a faculty could result in one’s coming to learn that the faculty is unreliable. Alternatively, it might issue no result regarding its own reliability. In light of this, if Descartes were right that his faculty of clear and distinct perception is self-verifying (rather than self-undermining), this would by no means be an obviously worthless result. Such an epistemically circular verification may well be of some significant value.

It is worth noting that, as several people have pointed out to me, when a faculty and its resulting beliefs are otherwise highly suspect, an epistemically circular self-verification of the faculty seems to be of little to no value.* (The reader is left to construct her own examples to show this.) It is when a faculty and its resulting beliefs are seen as already having something epistemically going for them prior to such self-verification that this process plausibly seems to bestow further epistemic virtue on them. As will become most evident in the next section, on my reading of him, Descartes does take his clear and distinct perceptions as having a good deal of epistemic merit even before his faculty of clear and distinct perceptions has verified itself.

*Don’t over-react to this fact, and rashly conclude from it that all epistemically circular arguments are worthless.  This is a perfectly general phenomenon: all arguments whose premises come from highly suspect sources — whether these arguments are epistemically circular or not — are pretty worthless.

F. But Is it Enough? Answer to Van Cleve

On my reading, then, Descartes sought clear and distinct perceptions which were buttressed by a “proof” of the epistemic principle behind them. This “proof”, as I’ve admitted, would have to be an epistemically circular one. In the previous section, I pointed out that this epistemic circularity does not clearly render the argument worthless. But even if this argument were of some epistemic value, how could it possibly constitute a proof of its conclusion? How could it serve the purposes of Descartes, who seems to have wanted some kind of guarantee of truth?

My first response will be specific to the disagreement between Van Cleve and me. Van Cleve raised this question regarding a lack of a guarantee of truth against the solutions of Alan Gewirth and Fred Feldman. These solutions were, like mine, two-level solutions according to which, roughly, one reaches the higher level of certainty by attaining lower level certainty of a fact that renders one safe from undermining sceptical hypotheses. Van Cleve charges Gewirth and Feldman with thereby setting Descartes’s standards for certainty far too low. Since my solution resembles those attacked by Van Cleve, one might wonder whether it may also fall prey to this criticism.

Gewirth’s and Feldman’s mistakes lie in making the lower level of certainty too low. Gewirth’s lower level is that of “psychological certainty,” which seems to amount to nothing more than being psychologically incapable of doubting the proposition in question,(16) while Feldman identifies the status as “practical certainty,” which is nothing more than the level of certainty needed to live up to ordinary, everyday requirements for knowledge.(17) Van Cleve is probably right that having one of these kinds of low-level certainty of p plus having the same low-level certainty of some other proposition cannot possibly amount to the very high level of certainty Descartes sought.

Regardless of how well Gewirth’s and Feldman’s readings stand up to this criticism, Van Cleve’s interpretation can gain no advantage at all over mine by means of it. For, if anything, I set Descartes’s standards for certainty higher than does Van Cleve. On Van Cleve’s interpretation, one’s belief that p attains some very high epistemic status in virtue of one’s having clearly and distinctly perceived that p. I follow Van Cleve here. That is, my lower level is already very high — whatever very high epistemic status Descartes thinks a belief attains in virtue of one’s having clearly and distinctly perceived it to be true.(18) (Recall Descartes’s largely positive epistemic evaluation of the atheist geometer’s belief in the theorem.) But on Van Cleve’s interpretation, that’s the end of the matter. On my interpretation, Descartes seeks an even higher epistemic status (scientia) which he obtains by also clearly and distinctly perceiving to be true the epistemic principle behind his clear and distinct perceptions. While there is room to wonder whether so proving the epistemic principle really does raise the epistemic status of one’s clear and distinct perceptions, it certainly does not lower it. Whatever “guarantee of truth” Descartes’s clear and distinct perceptions afford him on Van Cleve’s interpretation, he has at least that much of a guarantee of truth on mine.

G. But Is it Enough? The Absolute Sceptic and the Restrictive Foundationalist Sceptic

But is it an absolute guarantee? Is it enough to satisfy the requirements of an absolute sceptic? It seems not.

But no one — not even Descartes — should endeavor to satisfy such an absolute sceptic, at least insofar as I understand the type of sceptic in question. This absolute sceptic, I take it, is fairly close to what Thomas Reid called a “thorough and consistent sceptic.”(19) She will accept the hypothesis that her faculties are unreliable as a reason for doubt until it is proven that this hypothesis is false. She therefore will not accept any belief that does not live up to the requirement that the faculty from which it is derived has been proven to be reliable. But what will she count as such a “proof”? No argument whose premises are not antecedently acceptable by the above requirement will so count. Interestingly, Reid, in one place at least, interprets Descartes as attempting to refute just such a sceptic (Reid, 1895b, p. 447). But, as Reid realized, since nothing can be proven if there are no premises available from which to argue, this sceptic’s requirements are so obviously insatiable that only a fool would try to satisfy them. Such a sceptic, as Reid remarks, must “be left to enjoy his scepticism” (Reid, 1895b, p. 447).

Descartes was no fool. He must have realized that he needed some facts which are available for legitimate use as premises before he could begin to prove anything.

To avoid construing Descartes as foolishly trying to refute the absolute sceptic, while still trying to read him as targeting as radical a sceptic as he could intelligently hope to defeat, one naturally tries to limit the premises Descartes will help himself to without argumentation to only the most evident facts — in Descartes’s case, these will be those facts he can immediately intuit to be true in a clear and distinct fashion. Since Descartes thought that this class of facts was extremely restricted — it seems he thought it to be exhausted by necessary truths self-evident to one and self-evident facts about one’s own mental states — he can still be read as targeting a very radical sceptic and thereby seeking a very high level of certainty. The certainty he seeks can only be attained for those facts which fit into the very restrictive class described above or which can be deductively proven on their basis. The sceptic he tries to defeat — the “restrictive foundationalist sceptic,” as I will call her — will not accept, for example, that there is an external world until it can be so proven — no small feat, despite Descartes’s claim to have accomplished it.

Continuing on this line of interpretation, then, one will naturally, like Van Cleve, interpret Descartes as thinking that his clear and distinct perceptions are, from the outset, already at the level of certainty Descartes ultimately seeks for his beliefs. But then one will also, like Van Cleve, have to face the problem that Descartes seems to think all of his beliefs — even his clear and distinct perceptions — depend upon the proof for God’s existence in order to attain the status Descartes desires.

H. The Middle Ground

As long as we are thinking only in terms of one level of certainty, we are faced with a dilemma: Either Descartes’s clear and distinct intuitions (clear and distinct perceptions not based on argument but immediately intuited) are already at that level from the outset, or they are not. There’s a single initial test that a belief either passes or fails. Descartes either retains the belief, putting it among his stock of certainties from which he can attempt to prove other things, or he throws it out unless and until it can be proven on the basis of those beliefs that do pass the initial test. If Descartes’s clear and distinct intuitions initially fail the test, they then must be thrown out, and will be unavailable for use as premises from which to prove anything. But if they pass, then they are already at the desired level from the outset and they obviously don’t need a proof of God’s existence to attain that status.

A two-level reading avoids this dilemma by opening up a middle ground of initial epistemic appraisal. Descartes’s clear and distinct intuitions initially fail to attain the higher level of scientia. But they’re not thereby thrown out. They are rather taken to be at the already very high lower level, and that is enough, on my two-level reading, to make them available for use as premises in Descartes’s proof of the general principle that his clear and distinct perceptions are true.

On this two-level reading, Descartes does not have to drop his sights all the way down to the level of certainty afforded simply by clearly and distinctly perceiving a proposition to be true in order to avoid the quixotic task of trying to refute the absolute sceptic. Again, a middle ground is opened up. The sceptic Descartes targets is not so impossibly demanding as the absolute sceptic, but neither is she as easily satisfied as the restrictive foundationalist sceptic. Simply clearly and distinctly perceiving a proposition to be true, although this affords a high degree of certainty, is not enough to satisfy her, because she will count the deceiving nature hypothesis as still providing her with a reason for doubting such propositions. She is bothered by the fact that, as far as she can clearly and distinctly perceive, this undermining hypothesis may be true. As far as her clear and distinct perceptions go, she has nothing to say against it. But, unlike the absolute sceptic, she will allow as a proof of its falsity a (n epistemically circular) demonstration that it is false from particular clear and distinct intuitions.

We are left with two questions, which are really just two sides of the same coin. (1) Is this “middle” sceptic that Descartes targets more demanding than the restrictive foundationalist sceptic in any interesting or important way? (2) Is scientia in any interesting or important way a higher epistemic status than is that provided by simple clear and distinct perception? The answers to these questions will turn on how valuable Descartes’s epistemically circular validation of his faculty of clear and distinct perception could be.(20)

1. When citing the Meditations, as well as the Objections and Replies, I will first give the page reference to Vol. VII of AT and then to the translation being used.

2. Descartes calls this a “metaphysical” reason for doubt in the fourth paragraph of Meditation III (AT, 36; AG, 77-78), where he also makes it clear that this reason initially casts a shadow on all of his beliefs, but that he will, nonetheless, endeavor to “remove it as soon as possible.” 

3. Since Descartes feels he received his nature from God, this doubt often takes the form of a deceiving God hypothesis — the hypothesis that God (or a god) has given him such a radically defective nature. Thus, in Meditation I, Descartes writes, “Moreover, I judge that other men sometimes go wrong over what they think they know perfectly well; may not God likewise make me go wrong, whenever I add two and three, or count the sides of a square, or do any simpler thing that might be imagined?” (AT, 21; AG, 64). And in Meditation III we find, “When I was considering some very simple and easy point in arithmetic of geometry, e.g. that two and three together make five, did I not perceive this clearly enough to assert its truth? My only reason for afterwards doubting such things was that it occurred to me that perhaps some God might have given me such a nature that I was deceived even about what seems most obvious” (AT, 35-36; AG, 77). But we find out in the paragraph immediately following the Meditation I passage quoted above that Descartes’ worry is a general worry about his having a deceptive nature, a worry that would apply regardless of what one thinks is the source of one’s nature; in fact, ascribing one’s nature to some cause other than God only intensifies the worry. Descartes writes that whether one ascribes “my attaining my present condition to fate, or to chance, or to a continuous series of events, or to any other cause, delusion and error certainly seem to be imperfections, and so this ascription of less power to the source of my being will mean that I am more likely to be so imperfect that I always go wrong” (AT, 21; AG, 64). The metaphysical reason for doubt is given the more general formulation in Meditation V, where Descartes writes of his earlier worry that “I am so constituted as to go wrong sometimes about what I think I perceive most evidently” (AT, 70; AG, 107); and in Descartes’s letter to Regius of May 24, 1640 and the “atheist geometer” passage in the Reply to Objections II, both quoted below in Part D. 

4. There are important differences among various two-level solutions, and, in particular, between the solution I present and various of its predecessors. But I will not, for the most part, seek to explain or justify the departures I take from these predecessors — although I will in section F explain and justify one very important departure I take from the two-level solutions of Fred Feldman and Alan Gewirth. My neglect of most of my predecessors is not due to my thinking their work is not worth discussing. On the contrary, I believe the work done on the Cartesian Circle has been, for the most part, of high quality. The neglect is rather due to the character of the present paper, which seeks to defend and explore the implications of a particular interpretation rather than to explain its relation to various other interpretations. 

5. Van Cleve, 1979, p. 55. Throughout this paper I will use capital P and Q to stand for the specific propositions identified in this quotation, leaving lower-case letters for use as variables over propositions or as dummy sentence letters. To accommodate this practice, I have in this quotation changed Van Cleve’s lower case letters to capitals. 

6. Van Cleve, 1979, p. 67. Descartes’s positive statement in this paragraph, “But whenever I turn to the things themselves which I think I perceive very clearly, I am quite convinced by them so that I spontaneously exclaim: ‘Let who will deceive me, he can never bring it about that I should be a nonentity at the time of thinking I am something; nor that it should ever be true that I have never existed, since it is now true that I exist; nor even that two and three together should be more or less than five; or other such things in which I see a manifest contradiction'” (AT, 36; AG, 77), is, I believe, best read as a psychological statement: Descartes is registering the fact that he is psychologically incapable of doubting clear and distinct perceptions while he is having them. The main word, ‘convinced’ (Latin: persuadeor) certainly suggests this reading. But Descartes immediately goes on to make it clear that he has a reason for doubting his clear and distinct perceptions — slight and metaphysical though it may be — even though he is psychologically incapable of doubting for that reason while in the grip of a clear and distinct perception. And the presence of this reason, whether or not one is presently psychologically capable of doubting on its basis, is enough on my interpretation, to disqualify the belief from being a piece of scientia, even while one is clearly and distinctly perceiving it

7. Descartes certainly does consider this general principle to be doubtful, but as I read this passage, he also considers doubtful the particular clear and distinct perceptions. In fact, he takes the latter to be doubtful because he considers the former doubtful. The reading I oppose is the one Van Cleve would have to give this passage: That doubt is cast only upon the general principle and that no epistemic shadow at all is thought to be thereby cast upon the particular clear and distinct perceptions. 

8. AT 21; CSM, vol. II, 14-15. I deviate here from my usual practice of using the AG translation because AG has Descartes admitting that “none of my former ideas are beyond legitimate doubt” (AG, 64; emphasis added): ‘beliefs’ seems better than ‘ideas’ here, although no noun corresponding to either occurs in the Latin, which is “nihil . . . ex iis quae olim vera putabam” — “none of those things that I used to think true.” Incidentally, the reason I do not use the Cottingham translation throughout is that I do not like his translating of scientia (see note 13, below) — which is, of course, very important for the present paper. 

9. An additional refinement is suggested by a terminological remark Descartes makes in his Reply to Objections II. Descartes there proposes to limit the term scientia so as to apply only to conclusions of demonstrations and not to things immediately intuited to be true, because, he claims, “knowledge of first principles is not usually called science by dialecticians” (AT, 140; HR, vol. II, 38). Willis Doney points out that Descartes does not consistently follow this proposal in the body of the Meditations, but that he, in any case, introduces the term perfectissima certitudo, which applies to first principles as well as to conclusions, later in his reply to his critics (Doney, 1970, p. 388, fn. 4). In what follows, I ignore Descartes’s proposal to follow the usage of the dialecticians, and will use scientia in such a way as to apply to first principles as well as to conclusions, although the truth conditions I give may be, in strict Cartesian terminology, those of perfectissima certitudo rather than of scientia. In this strict terminology, in order for S to have scientia of p, the following additional condition must be met: S must not have immediately intuited the truth of p, but must rather have come to clearly and distinctly perceive that p by means of a proof. 

10. This should not be construed as a psychological remark. As remarked in note 6, Descartes does hold that one is psychologically incapable of doubting what one clearly and distinctly perceives — at least while one is clearly and distinctly perceiving it. But one still has a reason for doubting these things — the metaphysical reason — and they are in this evaluative, rather than psychological, sense doubtful. 

11. This is very close to an objection Van Cleve raises against Alan Gewirth’s interpretation (Van Cleve, 1979, pp. 60-61). 

12. It will often be difficult to specify which principle is “behind” any given belief, since belief-forming mechanisms can be individuated in many ways (widely or narrowly). In fact, the sceptic/anti-sceptic dialectic early on in the First Meditation can be read as a dispute over the specificity of relevant epistemic principles. The sceptic asserts, “I have sometimes caught the senses deceiving me; and a wise man never entirely trusts those who have once cheated him,” and the anti-sceptic replies, “But although the senses may sometimes deceive us about some minute or remote objects, yet there are many other facts as to which doubt is plainly impossible, although they are gathered from the same source: e.g. that I am here, sitting by the fire, wearing a winter cloak, holding this paper in my hands, and so on” (AT, 18; AG, 61-62). Here, the sceptic uses the history of actual error to show the falsity of the principle that The beliefs I form through the senses are all true. The anti-sceptic’s initial reply can be read as an insistence that his sensory beliefs obtained under certain optimal conditions are not based on that very general principle, but rather on a more specific one concerning beliefs formed under these optimal conditions. Thus, actual errors made under worse conditions do not undermine the judgments made under these better conditions. This problem of the individuation of belief-forming mechanisms and the specificity of epistemic principles should not detain us here, for Descartes clearly thought that What I clearly and distinctly perceive is true is the epistemic principle “behind” all his particular clear and distinct perceptions: If this principle is cast into doubt, his particular clear and distinct perceptions are thereby all undermined. 

13. AT, 141; HR, vol. II, 39. AG does not contain this passage. CSM translates scientia (‘science’, in HR) simply as ‘knowledge’ (vol. II, 101), which seems too weak. CSM has Descartes admitting that the atheist geometer can be “clearly aware” of the fact that HR translates Descartes as admitting that he can “know clearly”; the Latin here is “clare cognoscere” (AT, 141). 

14. For an attempt to make explicit such conditions for when one has a valid ground for doubting a proposition, see Curley, 1978 (Curley’s final formulation appears on p. 122). While Curley’s account seems, for the most part, to be on the right track and roughly in line with my present interpretation, I have problems with some details of Curley’s account. Most importantly, Curley seems to me to go wrong in giving too much of a role to our clear and distinct perceptions’ property of being assent-compelling. While I will not here argue against Curley’s account, my reasons for thinking it a mistake to give this property such a large role will become evident in section F (and especially in note 16) below. 

15. In this paper I am not concerned with the substantive question of whether Descartes is successful in constructing such a “proof”, but rather in the methodological issue of whether any such “proof”, even if successful, could accomplish what Descartes needs. 

16. Descartes certainly did think it a mark of a clear and distinct perceptions that they compel assent: While one is clearly and distinctly perceiving that p, one is unable to refrain from believing that p. And this feature of clear and distinct perceptions plays the vital role in Descartes’s Meditation IV argument that his clear and distinct perceptions are true (since God would be a deceiver if they weren’t). But Descartes wisely did not want to define clear and distinct perceptions solely in terms of compelling assent. One can imagine a creature who is compelled to believe certain propositions that it is in no position whatsoever to know. Descartes’s clear and distinct perceptions not only compel assent, but also are supposed to put one in an excellent epistemic position to know that which one believes. 

17. Feldman writes, “Practical certainty is certainty of a degree no greater than is required for ordinary knowledge. . . .Anyone who knows anything (in the ordinary sense of ‘knows’) has practical certainty with respect to whatever it is he knows” (Feldman, 1975, p. 41). 

18. Here it may be objected that Descartes does assign a higher status to his clear and distinct perceptions on Van Cleve’s interpretation than on mine, since on Van Cleve’s interpretation they are already (before being verified by God) guaranteed to be true, while they cannot be so on mine, since Descartes seeks a further guarantee when read in my way. But the issue here is what Descartes is willing to count as a “guarantee of truth.” My Descartes is more demanding as to what he will consider a guarantee of truth than is Van Cleve’s Descartes. 

19. See Reid, 1895a, p. 130. For discussion of Reid’s response to the sceptic, Alston, 1985, pp. 444-446 and DeRose, 1989, pp. 326-331. 

20. Thanks to Robert M. Adams, John Carroll, John Richardson, Marleen Rozemond, Roy Sorenson, Peter Unger, and James Van Cleve for comments on a previous draft of this paper and to Rogers Albritton and Alan Nelson for helpful discussions and criticisms of a previous attempt to get Descartes straight on these matters.


Descartes, Rene

AT: C. Adam, P. Tannery, ed., Oeuvres de Descartes (Paris: L. Cerf, 1897-1913), Vol. VII.

AG: E. Anscombe, P. Geach, tr., Descartes: Philosophical Writings (Indianapolis: Bobbs-Merrill, 1971).

CSM: J. Cottingham, R. Stoothoff, D. Murdoch, tr., Philosophical Writings of Descartes (Cambridge University Press, 1985).

HR: E.S. Haldane, G.R.T. Ross, tr., The Philosophical Works of Descartes (Cambridge University Press, 1984 reprint).

K: A. Kenny, tr., Descartes: Philosophical Letters (Minneapolis: University of Minnesota Press, 1970).


Alston, William P.: 1985, “Thomas Reid on Epistemic Principles,” History of Philosophy Quarterly 2: 435-452.

Alston, William P.: 1986, “Epistemic Circularity,” Philosophy and Phenomenological Research 47: 1-30.

Curley, E.M.: 1978, Descartes against the Skeptics (Harvard UP), Chapter 5.

DeRose, Keith: 1989, “Reid’s Anti-Sensationalism and His Realism,” The Philosophical Review 98: 313-348.

Doney, Willis: 1970, “Descartes’s Conception of Perfect Knowledge,” Journal of the History of Philosophy 8: 387-403.

Feldman, Fred: 1975, “Epistemic Appraisal and the Cartesian Circle,” Philosophical Studies 27: 37-55.

Gewirth, Alan: 1941, “The Cartesian Circle,” The Philosophical Review 50: 368-395.

Gewirth, Alan: 1970, “The Cartesian Circle Reconsidered,” Journal of Philosophy 67: 685-700.

Gewirth, Alan: 1971, “Descartes: Two Disputed Questions,” Journal of Philosophy 68: 288-296.

Kenny, Anthony: 1968, Descartes (New York: Random House).

Kenny, Anthony: 1970, “The Cartesian Circle and the Eternal Truths,” Journal of Philosophy 67: 685-700.

Reid, Thomas: 1895a, Inquiry into the Human Mind on the Principles of Common Sense, from W. Hamilton, ed., The Works of Thomas Reid (Edinburgh, Scotland: James Thin), eighth edition, vol. I.

Reid, Thomas: 1895b, Essays on the Intellectual Powers of the Human Mind, also from The Works of Thomas Reid, vol. I.

Van Cleve, James: 1979, “Foundationalism, Epistemic Principles, and the Cartesian Circle,” The Philosophical Review 88: 55-91.

Skip to toolbar