我的奮鬥

II.

 At the end of his paper, Sosa concludes that

In sum, Descartes was a foundationalist, and a coherentist, and a reliabilist. Each of these doctrines may be seen to apply, in some illuminating respect, to a system whose architectural elegance survives the erosion of its individual stones.

Most of the time, we are all too well familiar with the foundationalism of Descartes’, but what exactly is the coherentist element in the Cartesian epistemology?

 The key point consists in the relationship between cognitio and scientia. Given the picture delineated in the previous section, the Cartesian epistemology must grant the truth in foundationalism by allowing room for an inference-independent epistemic state of cognitio, no matter it is originated from perception, rational intuition, or anything else. However, cognitio as such can never acquire the highest status of perfect knowledge, scientia, for the simple reason that the former is essentially unreflective. On the other hand, attaining reflective scientia requires a view of ourselves－of our beliefs, our faculties, and our situation－so that we can see the sources of our beliefs are reliable enough. In this sense, scientia is knowledge that enjoys the support of a comprehensively coherent and explanatory world-view. So we may say that it is precisely the very existence of coherence of a whole system of beliefs that renders cognitio and scientia hierarchical, and makes the latter superior. And this recognition of the important role coherence plays in Descartes’ project should lead, quite plausibly, to a tenable coherence theory of empirical knowledge that gives its due to “an inference－independent epistemic state of cognitio”, i.e. the reception of external input outside our system of beliefs.

 In the sixth chapter of his book, The Structure of Empirical Knowledge, Laurence BonJour undertakes to develop a viable conception of “observation” within the general framework of his coherence theory of empirical knowledge. According to him, the urgency of such a conception results from three extremely forceful objections which any acceptable coherence theory of empirical knowledge must meet squarely. First, there is “the alternative coherent systems objection”, which claims that mere appeal to coherence will never be able to help us pick out one uniquely justified system of beliefs, since there will always be many different and incompatible systems of belief which are equally coherent. Second, there is “the input objection”, which claims that if coherence is the sole basis for empirical justification, a system of empirical beliefs might be adequately justified, in spite of being out of contact with the world that it purports to describe. And this is absurd. Third, there is “the problem of truth”, which claims that a coherence theory would fail to show that justification as conceived by the theory is truth-conductive unless it also adopts a coherence theory of truth and the accompanying idealistic metaphysics, while the latter is, if anything, most undesirable when we have empirical knowledge in mind. To this triad, BonJour maintains that the second must be dealt with first, for he believes that the answer to it is essential for the other two objections. Therefore, a cogent concept of “observation” becomes indispensible for a coherence theory of empirical knowledge that remains committed to a proper account for the very possibility for our coherent system of beliefs to “contact” the world.

 In his attempt to the development of the desired conception of observation, BonJour first seeks to refute a straightforward objection against the coherence theory. The objection goes as follows: On the one hand, it is essential to the concept of observation that observational beliefs are noninferential in character, that is to say, they are derived directly or immediately, without any inferential or discursive process; on the other hand, according to the coherence theory, all justifications must be inferential, never direct or immediate. Thus there seems to be a fundamental conflict between the idea of observation and a coherence theory. However, BonJour argues that, there is a conflation of the two senses in the characterization of a belief as “inferential” or “noninferential” in this objection. First, we may be well allowed to ask how a belief is arrived at. Second, we may ask how that belief is justified. And once this distinction is made, it makes perfect sense to maintain that an observational belief must be arrived at noninferentially, while insisting upon its being justified inferentially. The objection no longer holds.

 So, for a coherence theory, an observational belief can have its source directly from the world outside of our system of beliefs, and that is very good news. But how then are we to conceive of the exact way in which an observational belief in a system is justified, if, as coherence theory must claim, it cannot be justified by means of any intrinsically authoritative episode of awareness solely without any reference to other beliefs within that same system?

 In answering this, BonJour simply draws out the amplification of an account already suggested by Wilfrid Sellars in his classic “Empiricism and the Philosophy of Mind”:

An overt or covert token of ‘This is green’ in the presence of a green object…expresses observational knowledge if and only if it is a manifestation of a tendency to produce overt or covert tokens of ‘This is green’－given a certain set－if and only if a green object is being looked at in standard conditions.

Notice that, here, the view is not to be mistaken for externalism of justification, for both Sellars and BonJour would insist that for a belief to constitute knowledge, the basis for its authority must be recognized by the believer. That is to say, the believer must be in a position to infer justifiably from the occurrence of such a belief to its truth with the requisite general knowledge that beliefs of that kind are mostly true. So construed, the justification of an observational will always depend on the general knowledge indicated above, and this general knowledge rests in turn on further observations, and so ad infinitum. In this way, a coherent system that is qualified to contain genuine observational beliefs can be finally constructed.

 To be more specific, how is any observational belief, e.g. there is a red book on the desk, to be justified according to the coherence theory of the kind BonJour is trying to defend? BonJour characterizes this kind of belief as “cognitively spontaneous”, in the sense that it is not inferred. On the contrary, it simply strikes me and is not subjected to my own will. Now, given the characterization, three things are to be met to serve as the premises to justify an observational belief. First, the belief in question is a cognitively spontaneous belief of a certain kind K. Second, the conditions of observation are of a specifiable sort C which can be described as relevantly “normal” with respect to the kind of cognitively spontaneous belief to be justified. Third, it is a true law for us as normal observers that our cognitively spontaneous beliefs of that kind in conditions of the sort specified are highly reliable, that is, very likely to be true. Thus, the justification of the original belief goes as follows:

(I)    I have a cognitively spontaneous belief of kind K that there is a red book on the desk.

(II)   Conditions C obtain.

(III)  Cognitively spontaneous beliefs of kind K in conditions C are very likely to be true.

Therefore, my belief that there is a red book in the desk is very likely to be true.

Therefore, (probably) there is a red book on the desk.

The next task would be the justification of the three premises employed in the above argument, if we shall proceed in accordance with the tenet of the coherence theory. For the second premise, it will amount to a perhaps rather lengthy conjunction of subpremises specifying various details of the conditions, and BonJour maintains that it is not necessary to offer a general account of the justification of all the multifarious claims which might play a role in such a premise. This is because these claims must either have the status of background knowledge, and can be justified by appeal to the rest of the cognitive system with the same coherentist account for justification, or they have themselves the status of current observations, and will have to be justified by the very same general sort of argument as indicated above. Moreover, the latter case need not bother us, for observational beliefs may well serve to be premises for each other’s justification. BonJour’s wonderful example for this kind of harmless circularity is this:

For example, I accept various visual beliefs because I believe that the lighting is good and that my eyes are functioning normally, but a considerable part of my reason for the latter beliefs consists in the character and coherence of the visual beliefs which in fact occur.

 For the first premise, it can be further divided into three subpremises: first, that I have the belief whose justification is at issue; second, that it is a belief of a specified kind; third, that it is cognitively spontaneous. So divided, the problem of justifying the first two subpremises is easily dismissed by appeal to what BonJour labels as the “Doxastic Presumption”. A concise formulation of this presumption can be found in the fifth chapter of BonJour’s book, where he claims that

…the Doxastic Presumption is that my representation of my overall system of beliefs is approximately correct.

Here, the point is that this presumption cannot function as a premise to be employed in any justificatory argument in line with the coherence theory, because it is simply impossible to justify this presumption. As Wittgenstein pertinently remarks, justifications come to an end, and this is definitely one of the terminations where skeptics must halt. Rather, what the presumption seeks to affirm is, to repeat BonJour, a basic and unavoidable feature of cognitive practice－that epistemic reflection begins from a representation of oneself as having such and such a specific system of beliefs, and only relative to such a representation can questions of justification be meaningfully asked. Thus, according to the Doxastic Presumption, if my representation of my overall system of beliefs is approximately correct, there is no point to exclude that specific belief whose justification is sought from the region of my correct representation, and the first subpremise can be established. And, in a similar manner, so far as my representation of the belief in question is correct, how can I fail to recognize the kind of that belief as it is? Can I believe that I see a red book on the desk without realizing that this belief is about seeing? If that sounds ridiculous, the second subpremise should leave no room for wavering.

  As for the third subpremise, BonJour’s suggestion is that a claim of cognitive spontaneity is often justifiable by appeal to someone’s grasp of his overall system of beliefs and in particular to the absence of those beliefs serving as premises for derivation of the belief in question and the absence of any reason to think that the belief is discursively arrived at. Furthermore, there is a fact that allows for a simple identification of cognitively spontaneous beliefs－that they are usually extremely nuanced and full of subtleties with which our capacity of derivation is too poor to cope. For example, when we listen to a piece of music, sometimes we immediately believe that it is a variation of a theme that we are already familiar with. In that case, it will not make too much sense to speak of how we could ever derive the belief that that piece of music is a variation. Hence, with the third subpremise justified, the only question that remains is the justification of the third premise which serves to justify an observational belief. To be recalled, that is the premise that cognitively spontaneous beliefs of kind K in conditions C are very likely to be true.

 Here, we have to separate our way from BonJour’s for the sake of understanding Descartes. For BonJour contends, flagrantly, that

Such a premise is a putative empirical law concerning the behavior (in a broad sense) of such observers under such conditions…

And he even goes so far as co claim that

A coherentist account of observation need offer no special account of the justification of such laws.

To be more specific, BonJour believes that within the context of a functioning cognitive system this kind of empirical laws can in principle be defended with enumerative, inductive, and other kinds of argument, although he admits that there is a kind of circularity in this general picture. The problem is, however, if we keep the kind of knowledge Descartes is after in mind, i.e. scientia, we see immediately that no piece of scientia can be gained in this way, if the final justification of that piece depends so heavily upon the contingency of the third premise of BonJour’s schema of justification of empirical knowledge in general. In other words, the truth of this premise must in some sense be a priori, and without this sense, what could possibly be left in the characterization of Descartes as “rationalist”?

 Nor is Sosa’s explication of great help at this point. In the previous quoted paragraph, he simply ignores the actual process through which we are able to eventually come to “a view of ourselves and our place in the universe that is sufficiently comprehensive and coherent to raise us above the level of mere cognitio and into the realm of higher, reflective, enlightened knowledge, or scientia”. And it is clear that, under the illumination of BouJour’s much more elaborate account, this very “view” at issue can be nothing else but something very similar to BonJour’s third premise for the justification of observational belief. However, throughout his paper, Sosa gives us no more information concerning the origin and possibility of this “view” other than a vague promise that

If we persist in such reasoning, nevertheless, enough pieces may eventually come together to afford us to establish such a view. But, that only returns us right back to Descartes’ point of departure－the question Descartes has been asking is precisely one that cries out for a concrete way to reach that view.