An Infallible Method, Fallible Practitioners:Descartes' Epistemology
W. Zachary Wolff, 2001

Descartes presents a method of inquiry that he claims is infallible. This method is based on two principles, intuition and deduction, which are inherent to any rational mind. However, the method itself is not known a priori, but discovered through careful meditation. The method consists of a set of instructions regarding how to direct one's thought, and what judgments to make regarding the propositions one considers. Gilbert Harman discusses the idea of a reasoned change in view. He shows that reasoned change in view is not identical to deductive logic. Additionally, he shows that if reasoned change in view is based on deduction, it is in a non-obvious manner. These two views are perhaps not as different as they initially seem. Descartes presents a method designed to pursue indubitable truths. Harman seems to have a more practical method in mind. It seems that Descartes' method is a description of reasoned change in view based directly on deduction and valid only for the special case of arriving at indubitable propositions from first principles. Unfortunately, our inability to be certain of correct application of the method renders it useless.

Descartes begins his contemplation of a method by noting that he has, at one time or another, taken as true many propositions he later found to be false (I 117). Later, by considering the possibility that he may be dreaming or being deceived by some demon, he realizes that almost everything he has taken to be true is less than certain (II 13). In light of these considerations, he concludes that it is best to discard all the beliefs he has formerly taken as true. His path to knowledge will begin with a blank page, so as not to be cluttered with any falsehoods (I 117). In this exposition Descartes explicitly states that his goal is to assert only that which is absolutely certain, 'Even if I make but little progress I should at least be sure not to fall,' (I 119).

Gilbert Harman begins his discussion of reasoned change of view with the example of a woman considering what to eat for breakfast (1). This practical example belies his practical goals. Harman attempts to give a much more descriptive account of reasoned change in view, as compared to Descartes' prescriptive account. Or, if Harman's account is to be taken as prescriptive, he sets very different goals than Descartes. Descartes specifically states that he is interested only in the indubitable, 'I should hold back my assent from opinions which are not completely certain and indubitable just as carefully as I do from those which are patently false,' (II 12). Harman, on the other hand, does not state his goal directly, but it is clear in his discussion that he wishes to outline principles of reasoned change in view that allow one to accept uncertain propositions as true (4-5, as well as much of chapter 3).

From the realization that much of what he has believed is false or doubtful, and the desire to believe only that which is indubitable, it follows for Descartes that he should begin with an examination of precisely what can be called into doubt (II 12). He begins by rejecting all sensory information as dubitable through three considerations: First, that the senses are often inaccurate in cases in which the object of perception is distant, faint, or otherwise obscured (II 12). Second, that the perceptions that come to him in dreams often appear real, though they are surely false (II 13). And third, that even when awake he could be deceived by some demon of extraordinary power (II 15). It follows that sense data should not be the basis of his reasoning.

At this point, Descartes comes upon the cogito, the certain knowledge of himself as an existent thinking thing (II 18). This points the way to what he can consider as data for the basis of his reasoning, namely thoughts. It remains to be seen precisely which thoughts are proper objects of inquiry. Based on analogy to sensory perception, in which the most common errors were the result of the item in question being unavailable to clear and distinct perception, Descartes concludes that only thoughts which are subject to clear and distinct intuition by the intellect are proper objects of study (II 24).

By 'intuition' Descartes means the bringing of an idea into close attention by the intellect. By 'clear' he means that the proposition is present and accessible to the attentive mind. By 'distinct' that it is suitably distinguishable from other ideas so as to be made up wholly of what is clear. The use of the term 'clear' in the definition of distinct indicates that perhaps only one concept is really at work here. Descartes' use of 'distinct' seems to mean 'and only clear'. Here, clarity appears to be the primary concept, but I hold that neither of the two concepts is necessarily primary. One can see distinctness as primary by realizing that any idea that passes into attention is clear, in the most immediate sense, but only distinct ideas are valuable, because only these do not drag a host of unclear ideas along with them.

Descartes' reasoning above is not necessarily an application of the method, but the means by which he comes to it. Every piece of the method has been carefully justified save one: the validity of deduction. Descartes takes this as so basic as to be unquestionable. It is inherent in any mind. 'The deduction or pure inference of one thing from another can never be performed wrongly by an intellect which is in the least degree rational,' (I 12). The method, now, seems to be the following set of instructions: Start from scratch. Consider only those propositions that can be clearly and distinctly intuited. Believe only that which is absolutely certain. Make use of deduction to proceed from certain truths to further certain truths. Believe these deduced truths.

The method, then, takes as its input all those propositions that are known to be true at any given time. From these, it generates as its output new propositions to be taken as true. I see this as the best way to divide the labor done by the method. Because we are considering the method as a set of instructions including, 'Believe these deduced truths,' propositions are an appropriate output. Alternatively, the method could be taken to have instructions as its output, each instruction being of the form, 'Believe P,' where P is a proposition that the employer of the method has generated through application of the method. The last line of the method might then be, 'Output the instruction, 'Believe these deduced truths.' However, I think the above explication is more parsimonious.

The method relies approximately equally on intuition and deduction, both of which are absolutely necessary for its functioning. Without our power of intuition, the ability to hold a clear and distinct thought in the attentive mind for the purpose of studying it, our powers of deduction would have no thoughts with which to work. Without our powers of deduction, we could never increase our knowledge based on that which we already know.

Gilbert Harman presents a powerful argument that a reasoned change in view cannot be brought about by the principles of logic, and, in fact, that principles of logic are of a different kind from principles of belief revision. To illustrate this, he gives the example of Modus Ponens: P and if P then Q taken together imply Q (3). It is clear, as he says, that, 'Such a rule by itself says nothing at all in particular about belief revision,' (3). Harman does admit, 'It may be that some principles of belief revisions refer to such principles of argument, that is, to principles of implication,' (3). This is exactly the sort of system Descartes seems to have created. His 'method' corresponds to Harman's 'rules of reasoning'. Descartes takes the principles of implication to be obvious and inherent to any mind (I 12). The method states that any proposition implied by those propositions known with certainty to be true is also known with certainty to be true.

Another difference between argument and reasoning that Harman points out is that argument is cumulative while reasoning is not. Descartes' particular brand of reasoning is in fact cumulative, after the first step of the method, that of discarding all beliefs, has been executed. Where Harman argues that believing both P and if P then Q simultaneously does not always give one reason to believe Q (11), he is not speaking of belief in the same sense as Descartes. Harman suggests that in the above situation, it is sometimes best to reject P or if P then Q (11), but clearly nothing which we believe with absolute certainty should ever be rejected.

When properly applied, Descartes' method must be infallible, if we believe that human understanding of the rules of implication is correct. However, even the tiniest misconceptions can destroy the validity of the conclusions. I feel that this is illustrated in Descartes' argument for the existence of God, which I hold fails due to a subtle error in his conception of causation whereby causes must contain their effects in some way (II 28). It remains to be asked then, whether one can ever know with certainty that one is applying the method correctly. It seems that the answer here is no. There may be cases of egregious error, in which it becomes clear that one is applying the method incorrectly, such as if one used the method to deduce two contradictory statements. Lack of this positive evidence for error in the application is not positive evidence for soundness of the application. Once this is accepted, it seems that nothing beyond the cogito can be know with certainty. Whatever conclusion I reach may be the result of my incorrect application of the method. Still, it is I who reached a conclusion, and so I must exist.

So, Descartes has presented us with a deductive method of reasoning that is infallible in producing new certain truths from prior certain truths in an abstract sense, but is useless to us in practice. The method, properly applied, produces indubitable truths, but for any given application of the method there is some doubt as to the correctness of the application, so there is some doubt in the truth produced. Gilbert Harman presents a view of reasoning that is separate from deduction. His form of reasoning is not designed to produce certain truths, but to guide us in how we ought to think about the world around us. While potentially useful, it lacks some of the intellectual satisfaction of Descartes' attempt.

References:
Descartes, Rene. Meditations
Harman, Gilbert. Reasoned Change in View