A Critique of Moderate Rationalism
Graduate Institute of Philosophy
National Tsing Hua University, Taiwan
Abstract: The purpose of this paper is to criticize BonJour’s moderate rationalism. First, I try to expose his characterizations of the notion of a priori justification and the relevant notion of rational insight.
Second, I argue that the moderate rationalism is incoherent. Third, I argue, from a Davidsonian point of view, that rational insight, as a direct apprehension of abstract objects involved in a proposition, cannot provide us with a priori justification at all.
1In his In Defense of Pure Reason (DPR), BonJour argues for a version of moderate rationalism,
which holds that there is a priori justification, namely, a mode of epistemic justification that depends only on pure reason or rational insight and not on any experiential or quasi-experiential input. The purpose of this paper is to criticize, from a Davidsonian point of view, that BonJour’s moderate rationalism is incoherent and ill founded.
2According to BonJour, a proposition P is justified a priori (for S at t) if and only if (a) S has
reason R for thinking that P is true. (b) R does not depend on any positive appeal to experience or other causally mediated, quasi-perceptual contact with contingent features of the world. (c) R depends only on pure thought or reason. (d) It is permissible that S’s ability to understand P derives, in whole or in part,
3from experience. Since the idea of a priori justification has been the target of severe and relentless
1 Laurence BonJour, In Defense of Pure Reason, New York: Cambridge University Press, 1998. 2 The term “proposition” is used by BonJour himself. He takes propositions as assertive contents of belief, judgment, or
thought, possessing truth value. Although some philosophers decline to use this term and disavow anything like proposition as typically conceived, BonJour claims that such disagreement does not seem to reflect any serious divergence concerning the objects justification, belief and knowledge. (See DPR pp.26-27.) Thus in this paper, the term “proposition” and “sentence” are
used interchangeably in appropriate contexts. 3 See DPR p.11.
skepticism, BonJour claims that we should take this idea seriously. First, he argues, quite convincingly, that we do need a priori justification even in the case of our ordinary empirical knowledge. We need inferences in terms of which we can obtain beliefs whose contents go beyond direct experience or observation. Those inferences “must seemingly rely on either premises or principles of inference that are at least partially justified a priori.” The requirement of inference for the establishment of knowledge
indicates that the repudiation of all a priori justification is tantamount to intellectual suicide. Thus, in
4order to have knowledge or any sort of intellectual products, we need a priori justification.
Second, BonJour shows us that we do have a priori justifications. His examples are as follows: (1)
Nothing can be red all over and green all over at the same time. (2) If a certain person A is taller than a second person B and the person B is taller than a third person C, then A is taller than person C. (3) There are no round squares. (4) Two plus three equals five. (5) Either David ate the last piece of cake or else
5Jennifer ate it, and Jennifer did not eat it, therefore, David ate the last piece of cake. BonJour thinks that
these examples are extremely obvious and compelling. They altogether provide the prima facie case for
It seems to me that the epistemic status of the sentences (1)-(3) and that of the sentence (5) are quite different, since, unlike the truth of sentence (5), the truth of the former group of the so-called “analytic
sentences” cannot be established by logical means alone. Moreover, mathematical sentences are different from and cannot be completely reduced to logical sentences. However, my complaint is not about the heterogeneity of BonJour’s various examples. Rather, I will place my focus on the propositions of some
4 See DPR pp.-4-5. 5 See DPR pp.100-106.
basic logical rules of inference, such as (5), law of excluded middle, and so on, for if there indeed are propositions justified a priori at all, then any basic logical rule of inference must be one of them.
Nevertheless, if BonJour’s account of a priori justification fails in the case of those logical sentences, so
will it likely be in the cases of other sorts of sentences. Now, my question is: are the basic logical rules of inference justified in an a priori manner suggested by BonJour? In other words, do their justifications by no means depend on any positive appeal to experience or causal contact with the external world but only on pure thought or rational insight? If the answer is “yes”, then it does provide case for rationalism. But I am afraid that the case is not that obvious and compelling as BonJour thinks that it is. Conspicuously,
6BonJour’s argument assumes that justification of logical inference is a priori in character. At first sight,
regarding the rules of logical inference as being justified a priori seems to be innocent. After all, they are
the paradigm cases of a priori justified propositions, if such there are. However, given BonJour’s own
notion of a priori justification, the issue of whether the rules of logical inference are a priori justified in
that particular sense is not that clear as it first seems to be.
According to the condition (a) mentioned above, for P to be justified a priori for S at t, S must have
reason R for thinking the P is true. However, what is R? To determine BonJour’s answer to this question, we bad better to see his further characterization of a priori justification: “According to rationalism, a
priori justification occurs when the mind directly or intuitively sees or grasps or apprehends (or perhaps merely seems to itself to see or grasp or apprehend) a necessary fact about the nature or structure of reality. Such an apprehension may of course be discursively mediated by a series of steps of the same
6 For BonJour, inferences required for empirical knowledge include logical and non-logical ones. This is the reason why he characterizes the premises or the principles of inference, on which the inferences in question rely, as being “at least partially”
justified a priori. However, since ultimately speaking, the logical rules of inference must be involved in our acquisition of empirical knowledge, it does not seem to affect too much if we simply concentrate on the logical inference in our discussion of
kind, as in a deductive argument. But in the simplest case it is allegedly direct and unmediated, incapable of being reduced to or explained by any rational or cognitive process of a more basic sort – since any such
7explanation would tacitly presuppose apprehensions of this very same kind.” Let us suppose that the
justification of P is the simplest case, then the passage claims that the a priori justification of P is based
upon S’s apprehension of a necessary fact about the nature or structure of reality. Does it indicate that the reason R for which S thinks that P is true is such an apprehension? I am not sure, but if it does, then BonJour’s statements are inconsistent. On the one hand, R as apprehension is said to be incapable of
being reduced to or explained by any rational or cognitive process of a more basic sort. Such an intuitive grasp is regarded as the most basic rational process, which depends on no further thought. On the other hand, according to condition (c), R is characterized as depending on pure thought or reason. Hence, there seems to be an incoherent view regarding the independence of R on further thought.
Or perhaps we should take R, not as S’s apprehension, but as S’s beliefs, thought, judgments or their
contents. If R is S’s belief, then a threat from the infinite regress problem of epistemic justification may arise. BonJour never claims that R is self-justified. Rather, he holds that R, being in the simplest case, depends only on pure thought or reason. Consider the above example (5), which consists of three propositions. BonJour writes: “if I understand the three propositions involved, I will be able to see or grasp or apprehend directly and immediately that the indicated conclusion follows from the indicated premises, that is, that there is no way for the premises to be true without the conclusion being true as
8well.” Here the direct and immediate grasp is what BonJour called “rational insight”, which “does not
7 DPR pp.15-16. 8 DPR p.105.
9depend on any particular sort of criterion or any further discursive or retiocinative process.” When one
carefully and reflectively considers the proposition in question, according to BonJour, she is “able simply
10to see or grasp or apprehend that the proposition is necessary.” Thus he claims that as long as such an
insight occurs, proposition (5) is justified a priori and it does not require our appeal to the rule of
Although S’s a priori justification of a proposition depends on her rational insight, such an insight is a consequence of her understanding of the proposition. However, if her understanding is mistaken, then the a priori justification will be ruined. Therefore, the correctness of understanding plays a crucial role in a
priori justification. Ironically, such a rationalist position does not seem to be too much different from an empiricist one, since in dealing with the problem of apriority of justification or knowledge, the basic strategy of some version of empiricism is to ground it on meaning or understanding. However, being a rationalist, BonJour has a peculiar, though sketchy and tentative, view of understanding (comprehension) and mental content, which roughly holds that, at least in the case of simple a priori justified proposition,
(i) the content of thought is internally accessible to the mind and (ii) understanding of such a simple proposition is merely a matter of apprehension of abstract objects involved in the proposition, without the
12need of any perceptual or quasi-perceptual relation and (iii) the justification is merely based upon one’s
correct understanding of the proposition.
BonJour calls his version of rationalism “moderate” because he allows one’s a priori justification of a
proposition to be fallible. However, if the a priori justification of P is based up S’s correct understanding
9 DPR pp.106-107. 10 DPR p.106. BonJour’s emphasis. 11 DPR p.105. 12 See DPR p.180 and p.183.
of P, as he claims, then the moderate rationalism is incoherent. For BonJour, a correct understanding must see the “necessary fact” about the nature or structure of reality or that the proposition is necessary.
Nevertheless, since the fallibility of justification means that S is justified in believing that P and yet P is false and, moreover, since correct understanding implies the existence of what BonJour calls “necessary fact” or the necessity of the proposition, it is difficult to admit the possibility of the fallibility of the a
priori justification in the above sense. That we see that a proposition is necessary and yet it is false is merely incoherent. Thus BonJour’s version of rationalism cannot be moderate.
Furthermore, even if we agree with BonJour that a priori justification of P is based on correct
understanding of P, still it is ill founded to hold that correct understanding is a matter of direct apprehension of abstract objects involved in P, because such a view totally ignores the holistic character of beliefs, in the sense that beliefs are identified, partially, by their relations to one another. For instance, when S believes that “(A ？ ;A)” is not true, it is too rush to conclude that S is irrational. Perhaps she does,
but it is also possible that S simply adopts a different version of logic. To determine which is the case, we have to check S’s other beliefs. For instance, we may find that S also believes that “ ; (P ； Q)? P ” and
“ (P ？ Q) ； S? (P ； S) ？ (Q ； S) ” are both false and, finally, understand that what S uses is intuitionist logic. Simply considering S’s belief that P, without relating it to S’s other beliefs, we cannot tell whether S’s understanding of P is correct or not. We even cannot tell what P means. Thus rational insight, as a
direct apprehension, cannot provide us with justification, a priori or not, of any proposition at all.
BonJour is correct in arguing that we need basic logical principles to establish our knowledge. What he does not see is the deep root of those principles in our thought. They are not abstract objects of apprehension but constitutive of beliefs. From a Davidsonian point of view, if we take ourselves as
creatures that have propositional attitudes or act intentionally, as we do, then the basic principles of logic, among others, must be shared by all of us. It is not because we all have rational insight about those principles but because they as are norm of rationality and, more importantly, such a norm is a condition of
13having thoughts. Of course, one may try to think of the possibility of someone’s violating of all logical
principles. If it was possible, then, according to Davidson, we could no longer attribute thought/belief to that person at all. Therefore, being a creature having propositional attitudes, one must be rational in
general. However, this by no means implies that there is a definite list of “basic principles of rationality”. There is no such a list. Davidson writes: “The kinds and degrees of deviation from the norm of rationality
that we can understand or explain are not settled in advance. We make sense of aberrations when they are seen against a background of rationality; but the background can be constituted in various ways to make
14various forms of battiness comprehensible.” Here space does not allow me to elaborate Daivdson’s
view of logic, but I think that if we take his line seriously and scrutinize it deeply enough, then the sharp dichotomy between a priori and a posteriori on the one hand and that between rationalism and
empiricism on the other will finally be abandoned.
13 For example, see Donald Davidson, “Incoherence and Irrationality,” Dialectica, 39, (1985) 14 Op. cit., p.352.