Lectures on Bonjour’s The Structure of Empirical Knowledge Chapter 8


     Copyright © 2013 Bruce W. Hauptli


Chapter 8. Coherence and Truth [Reply to the Truth Objection]


8.1 The Problem: Justification and Truth:


157 “As we argued at length in Chapter 1, an essential part of the task of an adequate epistemological theory, in addition to providing an account of the standards of epistemological justification, is to provide an argument or rationale of some sort for thinking that an inquirer who accepts beliefs which are justified according to that account (and rejects ones which are not) is thereby at least likely to arrive at truth, that adhering to those standards is truth-conducive.” 


158 A coherence theory of the nature of truth which identifies truth with long-run coherence would have a ready answer to the “problem” here!  This is not the sort of view which BonJour wishes to champion.  He wishes to defend the traditional, common-sensical view which he will call metaphysical realism which is centered upon a correspondence theory of the nature of truth (rather than a coherence theory of its nature). 


8.2 Realism and the Correspondence Theory of Truth:


159-160 The correspondence theory of the nature of truth is characterized as the view which “...holds that a belief or statement is true if it corresponds to or agrees with the appropriate independent reality.” 


-160 Given this characterization, of course, it is clear that the correspondence theory presupposes “...that there is an independent reality with which our beliefs and statements may correspond or fail to correspond.”  According to BonJour, this is the central tenet of metaphysical realism. 


-161 To deny this would be to adhere to metaphysical idealism: “...the view that things of some specified kind, or perhaps, even things in general, exist only as objects of thought, as depicted or represented by minds.” 


--An undiscussed alternative:metaphysical pragmatism”—agrees with metaphysical realism that there is an independent reality, but disagrees with the notion of “correspondence, agreement, or copying” and, instead, speaks of agreement in terms of “the results of action” (or “workability”) rather than in efficacy of representation.  Of course, there are other alternatives than metaphysical “realism” and “pragmatism” when one considers the possible relationships between us and the “independent” world (if there is such): one might maintain that the proper relationship is one of “worship,” or “aesthetic appreciation.”  Of course, these views might not assign much importance to epistemic justification! 


160-161 BonJour maintains that as a general thesis about reality, metaphysical idealism is seriously deficient (he says “incoherent,” which, of course, has to be the worst thing one can be according to him).  His argument here is worthy of serious study because it is so bad:


-First, on p. 160, he defines “metaphysical realism” as: “...the view that there is an independently existing reality with which our beliefs and statements may correspond or fail to correspond.”  [A bit later, citing Hilary Putnam, BonJour more carefully and correctly characterizes it as the perspective that “the world consists of some fixed totality of mind-independent objects”].[1] 


-Then, on p. 161, he maintains that it is “...the thesis that something or other has an sich reality, that an sich reality generally exists.” 


-Finally, he argues that: “...not everything can have reality only as the object of an act of thought or representation, for the representative act must itself exist an sich if it is to confer representative reality on its object.  Thus although it would make sense at this level of abstraction to hold that an sich reality consists entirely of acts of representation (or, perhaps more plausibly, of the minds which engage in such acts), this is as far as idealism can coherently be pushed, and a view of this sort is still sufficient to vindicate metaphysical realism as understood here.” 


--Clearly, this argument is truly bad.  Following Kant, BonJour minimally wants ‘an sich’ to mean ‘as such’ or ‘in itself’.  He is not as clear as he needs to be however, and this notion of an sich reality, if it is to be tied to a controversy between “metaphysical realists” and “metaphysical idealists” must include the notion of “independent” (better, certainly, “mind-independent”) reality.  Certainly historical idealists (like George Berkeley) can agree that “an sich reality” genuinely exists, if this can mean that minds (and ideas) exist!  In short, his argument goes through without problem if ‘an sich’ means “independent” (but then it is not an argument against metaphysical idealism), and it does not go through if ‘an sich’ means “mind-independent” which is what it must mean if it is to correctly characterize the metaphysical idealist. 


It should be noted, however, that BonJour doesn’t need to argue against metaphysical idealism at this point.  All he needs to do is to make the contrast between these views clear.  He must, however, successfully meet the obvious challenge which confronts someone who wishes to offer a coherence account about the criterion of truth while offering a correspondence account of the nature of truth (that is, while adhering to metaphysical realism)—he must establish that the beliefs which fulfill the coherence conditions set forth by his criterion of truth (that is, the conditions specified in section 7.3) actually are likely to be true (that is, they indeed adequately, appropriately, or sufficiently correspond to the independent reality).  In short, he must successfully offer a “metajustificatory argument.” 


-162-165 BonJour briefly discusses the differences between “semantical realism” and “semantical anti-realism” in an effort to clarify the above discussion.  I believe it actually complicates the discussion, so I will skip over the discussion. 


165 Having briefly discussed the “alternative” to metaphysical realism, BonJour turns to a brief discussion of the problems which acceptance of a correspondence theory of [the nature of] truth engenders.  As he notes, the correspondence theory calls for “...a relation between two terms: first, the thing which is true (the truth-bearer) and, second, some portion or aspect of independent, an sich reality which makes the first term true.” 


-The first problem which correspondence theorists confront, then, is “...what sort of entity is the truth-bearer supposed to be? 


--Alternatives include: “...sentences, statements...occurrent judgments, beliefs, and propositions; and philosophers have worried about which of these is the right choice.  It is hard, however, to see that very much hangs on this question for the issues under discussion here, since a correspondence theory formulated in terms of any (or all) of these...could do the job required....it will be convenient to regard propositions as the primary truth-bearers in the relation of correspondence.”  Of course, serious problems lurk here. 


-The second problem which correspondence theorists confront is “...about the second term of the relation of correspondence: to what is a true proposition supposed to correspond? 


--166 Alternatives include: facts, things, complexes of objects.  BonJour leaves this largely open, and, of course, serious problems also lurk here. 


-The third problem which correspondence theorists confront is that they must provide an account of the nature of the correspondence relationship—they must give an account of what it means to say that two (very) different sorts of things, metaphysically speaking, “correspond with” each other.  The questions here are: “How can sentences be like facts?” “How can statements be like objects?” “How can beliefs be like things?” 


--BonJour admits that “the account of correspondence offered so far amounts to little if anything more than Aristotle’s dictum: “To say that what is is not, or that what is not is, is false, while to say that what is is, or that what is not is not, is true.”[2] 


--166-167 BonJour points out that it has often been thought that (in addition to giving a more detailed account of the two “relata”), correspondence theorists must also give a more detailed account of the relation (of correspondence).  But, just as he doesn’t feel it is requisite for him to give a detailed conception of the relata, he doesn’t believe he has to give a detailed conception of the relation either.  [167] “Indeed, I am inclined to regard Aristotle’s dictum as adequate by itself....”  That is, the “Correspondence Theory” speaks of “correspondence,” but this is not, as it stands, clear.  ‘Correspondence’ may mean many things, it would seem to be a “relative” (or “contextual”) matter, and it may be sensitive to purposes.  What will suffice for correspondence in one context may be inadequate in another. 


--167 He maintains that “...any possible objective, an sich reality will [also] consist of particulars or things having properties or attributes (including, of course, relational properties).  Despite protestations that any characterization of reality must be merely one among many possibilities and thus “optional,” this one does not seem to be.  (What, after all, would be the alternative?)  And thus we can say that a proposition corresponds to an sich reality if it refers to a particular thing which does exist and attributes to that particular either existence itself or else some specific property that it actually has....” 


---Criticism: this seems far too quick.  First, because the relata are different sorts of things (propositions and facts, for example), the “correspondence” between them will have to be a matter of degree; and clearly not just any degree of correspondence will do.  Clearly, then, more needs to be said here.  Secondly, given the difference in the nature of the relata, a discussion of the relationship seems requisite.  Finally, there is a serious question as to how one is to be able to judge whether or not the correspondence obtains.  Clearly the metajustificatory argument can not be the normal sort of justificatory argument (which, for BonJour, requires appeal to the coherent system of beliefs)—if this were so, then the whole account would be circular.  If it is not such an argument however, he had best not appeal to any of the elements of the system essentially! 


8.3 The Metajustificatory Argument:


(a) Introductory Remarks:


BonJour noted on pp. 9-14 that in addition to providing an account of the standards of justification, he must also provide a “metajustification” of the proposed standards [9] “...by showing the proposed standards to be adequately truth-conducive.”  It is not enough, then, to claim that they are such, he must show that this is the case.  Of course, such a metajustification can not be a justification in the sense in which “ordinary” beliefs are justified according to BonJour—beliefs are justified by appeal to the coherent system of beliefs, and if he offers a metajustification of the truth-conduciveness of the system of beliefs by appealing to the system itself, his account will be viciously circular!  BonJour’s frequent references to the a priori character of his meta-argument in this section show that he is offering a different sort of justification here.  


     This raises two different problems.  First, it is clear that BonJour’s account of justification is a two part conception.  He offers a coherence theory of empirical justification, and he offers a different theory of a priori justification.  His account of justification of beliefs (propositions, etc.) at the a priori level appeals to our intuitive “grasp” or “apprehension” of them:


210-211 ...once an a priori proposition has been understood, nothing further is needed beyond that very understanding—and hence no appeal to any further standard—to “see” or apprehend intuitively (in an optimum case) that the proposition must be true; and hence nothing further is needed for belief in it to be justified. 


But this appears to be a sort of “foundationalism”—it appears that various particular a priori beliefs are justified directly and not by appeal to other beliefs.  Since such justified beliefs are necessary for the system of a posteriori beliefs to be justified, one might claim, BonJour is, after all, a foundationalist!  As Alvin Goldman points out in his “BonJour’s The Structure of Empirical Knowledge:


...BonJour’s theory of epistemic justification is fundamentally bifurcated.  Coherence is not an essential feature of all justification, only a posteriori justification.  This bifurcation is reinforced by the rejection of a metajustification for a priori justification.  Unlike the case of empirical belief, the cognizer does not need to show that intuitive apprehension [of the a priori beliefs] is truth-conducive. 

  This thoroughgoing bifurcation in BonJour’s theory is troubling.  Epistemic justification, one would have thought, is a unitary concept.  There must be some property (or family of properties) that all justified beliefs have in common; and this common feature or feature is what a theory, or analysis, of epistemic justification should provide....[But for BonJour, it now seems] [c]oherence is not essential for justified belief; nor is possession of a metajustification.[3] 


-In effect, we might say, BonJour is a “rationalist”—he offers a theory of justification which, like those of the continental rationalists (Descartes, Spinoza, and Leibniz, for example) which ultimately “rests” on certain a priori truths which can be established by an exercise of pure reason.  This categorization of his orientation is, certainly, clear given the title of his next book: In Defense of Pure Reason: A Rationalist Account of A Priori Justification.[4]  Clearly, if his justificatory account of empirical knowledge ultimately rests on an a priori account of knowledge, then he is, ultimately, a rationalist. 


     Second, there is the danger of a justificatory regress here.  As Goldman points out, “...BonJour’s move to a priori justification only postpones the difficulty, for won’t there have to be an a priori metajustification of every a priori justified belief.”[5]  While BonJour rejects the need for such a “meta”-metajustification, in the case of the a priori beliefs, it should be noted that his demand for a metajustification is a function of his commitment to internalism and its view of epistemic responsibility.  Clearly BonJour is going to have to convince us that the sort of responsibility which arises at the a posteriori level does not arise at the a priori level. 


For these criticisms to ultimately make sense, however, we must first understand BonJour’s metajustification! 


(b) BonJour’s Meta-Justification:


170 A coherent system of beliefs might either (1) change radically over time, or (2) converge upon a stable set of beliefs and remain relatively stable for a long period of time.  “...it is only in the latter sort of case—the case in which the belief system converges on and eventually presents a relatively stable long-run picture of the world, thus achieving coherence over time as well as at several particular times—that coherence of the system provides any strong reason for thinking that the component beliefs are thereby likely to be true. 


-While BonJour claims there are two options [(a) the system of beliefs will change rapidly, or (b) it will remain largely stable over the long run], Thomas Kuhn champions a third alternative which is important to note here.  He contends that there is an important difference between what he calls “normal science” and “revolutionary science.”  The latter indicates, in his theory, points where the underlying nature of the scientific theorizing undergoes radical, indeed “incommensurable” change.  In his The Structure of Scientific Revolutions, Kuhn offers a detailed picture of the history of science as a series of anomalies and crises which engender revolutions in scientific thinking.  The change from Newtonian to Einsteinian physics, for example, is so radical a revolution that according to him, the new theory can not be understood from the standpoint of the old one, and the old one doesn’t make sense for new theorists.[6]  Indeed it is important to note that instead of taking the question of the “justification of our current theory” as the “starting point” (or goal) of epistemology, one could, instead, take up the question of the justifiability of various proposals for changing one’s theory as the basic problem (or as the starting point). 


“It must also be emphasized that it is the Observation Requirement formulated in Section 7.1 which guarantees that the system of beliefs will receive ongoing observational input, and it is the preservation of coherence in the face of such input which...provides reason for thinking that a system of beliefs is likely to be true. 


-Note that in commenting on p. 142 above where BonJour was discussing the Observation Requirement, I noted that what he says is that it guarantees at least the appearance of such input.  Will that be sufficient?  Does his requirement guarantee input?  [p. 142 vs. p.170, again] 


171 Thesis M.J.: “A system of beliefs which (a) remains coherent (and stable) over the long run, and (b) continues to satisfy the Observation Requirement is likely, to a degree which is proportional to the degree of coherence (and stability) and the longness of the run, to correspond closely to independent reality.” 


-The [a priori] Meta-Argument for Thesis M.J.:


-“...the kind of situation described [above]...requires an explanation....The coherence-cum-stability of a system of beliefs is complicated and fragile, easily disrupted or destroyed, and thus it is inherently unlikely that a system of beliefs which is constantly receiving the sort of input that is assured by the Observation Requirement would remain coherent from moment to moment without constant revisions which would destroy its stability.  Some explanation is therefore needed for why it continues to do so, and the obvious one is that the beliefs of the system match the independent reality which they purport to describe closely enough to minimize the potential for disruptive input.” 


--Critical points: first, note that it is not enough for the input to be apparent, it must be real input if the presumption contained in this explanatory demand is to be warranted.  Note, secondly, that a “Metaphysical Idealist,” could adopt (with appropriate revisions) everything said thus far.  Berkeley could (indeed, he does) assign a special place to the cognitively spontaneous beliefs, and he could offer a justification for taking the belief system to be truth-conducive (he appeals to the author of these beliefs [a deity]!).  Because he is an idealist, of course, the “explanation” which he would offer, appeals to a “greater mind” (rather than an “independent reality”).  BonJour, of course, relies upon his (bad) argument against metaphysical idealism here.  Thirdly, it must be noted that the argument is premised upon a factual premise—he claims that “this situation” requires an explanation.  But what phenomenon is he referring to?  Clearly he believes that he/we have such a long-run, coherent system of empirical belief.  This requires the “Doxastic Presumption,” and it requires that there be such a system; but he does not argue for this!  Finally, one wonders what “justifies” the “explanatory demand” here.  This is the subject of the following argument (pp. 171-178).


171 A two-premise argument for Thesis M.J:


171 P1 An explanation of coherence-cum-stability-cum-observation is called for. 


-172 According to BonJour, this demand is self-evidently true. 


--Criticism: is it?  While this is a central thesis of the Rationalist (as opposed to Irrationalist) orientation in the Western tradition that there is a [rational] explanation for “everything”—since the ancient Greeks, most philosophers in the Western tradition have held that this is the case (held, that is, that “what is real is rational, and what is rational is real”)—there are theorists who reject this view.[7]  Fideists, for example, hold that there are “things” in the universe which are not rationally explicable.  These fideists not only maintain that there are such things, but they maintain that they are the fundamentally important “things” in the world.  Tertullian, Pascal, and Kierkegaard are all exemplars of this orientation—they hold that some “things” must be taken on faith (and that these things are the truly important things).  For them, then it is definitely not a self-evident, a priori truth that “everything” requires an explanation?  Of course, there could be times where rational explanations are not called for, but the particular “thing” in question might not be among them.  BonJour, however, seems to me to here evince the fundamental Rationalist attitude (in both senses of the term), and while an empiricist might agree that an explanation is called for, an irrationalist could easily reject the claim that it is an a priori truth that an explanation is called for (or, at least, called for here). 


-Moreover, supposing that an a priori explanation is required here, does what BonJour offers fit the characterization offered on pp. 210-211: “...once an a priori proposition has been understood, nothing further is needed beyond that very understanding—and hence no appeal to any further standard—to “see” or apprehend intuitively (in an optimum case) that the proposition must be true; and hence nothing further is needed for belief in it to be justified”? 


171 P2 [clauses transposed] The best explanation for the coherence-cum-stability-cum-observation is that the reliable cognitively spontaneous beliefs are caused by the situations they depict and the entire system of beliefs corresponds [with a reasonable degree of approximation] to independent reality. 


-173 This second premise can be questioned in either of two ways: (a) Normal Alternative Hypotheses: “...our putative observational beliefs are systematically caused in some way by features of the world, but in which the resulting system of beliefs nevertheless presents a picture of the world in question which is in some important way inaccurate, incomplete, or distorted.”  BonJour maintains that confronted with such alternative hypotheses, we must assess the relative likelihood of the alleged alternatives. 


--173-176 Such “normal alternative hypotheses” might arise either (a1) by having the observation component more or less correct, but the overall system inaccurate; or (a2) by having the whole system (including the observational component) be fundamentally skewed. 


--BonJour maintains that (a1) amounts to a partial agreement with the correspondence theory; moreover, such alternative systems do not seem likely; and, finally, (p. 176) if “...the cognitively spontaneous beliefs which satisfy the observation requirement are indeed objectively reliable, it becomes unlikely, on purely a priori grounds, that a system of beliefs will remain coherent (and stable) in the long run and still fail to depict the unobservable aspects of the world in an at least approximately accurate way.” 


--176-177 BonJour claims that while (a2) might seem to suggest that the view of the world offered by the coherent system was skewed in some way, this does not seem plausible because: [177] there would have to be a general isomorphism between the underlying correct view and the view we supposed to correspond to the world and it is very unlikely [again, on a priori grounds] that there could be such a long-run stable and reliable but massive coincidence. 


--Critical Comment on the Second Premise and “Normal Alternative Cases:” as Alan Goldman points out in his “BonJour’s Coherentism,” “any realist epistemology, BonJour’s included, must eventually defend the claim that the principle of inference to the best explanation itself tends to preserve truth, at least in certain contexts.  (BonJour does not explicitly provide what he might call this meta-metajustificatory argument.)[8] 


--Note, also, that BonJour’s argument here is, and must be if he is to fulfill his justificatory responsibility and make the connection with truth which he seeks, that the best explanation is that the coherent beliefs are true.  But his argument gives us, at best, that the best available explanation is that this is the case!  That is, he needs to establish that there is no other explanation which is better than the one he advances; but he only shows that his explanation is better than those who appeal to “chance,” “normal alternative hypotheses,” and “skeptical [see below!] hypotheses!”  Given that his argument is to be wholly a priori, he must, it would seem, at least argue that these are the only available explanations (though I don’t see how one can determine a priori what the various explanations for any phenomenon might be—at least not given our finite cognitive capacity). 


--In his “BonJour’s Coherence Theory of Justification,” Marshall Swain suggests that the “skewed” beliefs of ordinary, nonscientific individuals, and the belief systems of religiously committed individuals provide a clear problem for BonJour’s claim that it is unlikely on a priori grounds that such “normal” alternatives could arise.[9] 


-173 (b) According to BonJour, the second way in which the second premise of the argument for Thesis M.J could be questioned is via various “skeptical hypotheses.  The next section considers these cases. 


--Criticism: note that there is at least one other way in which the second premise could be questioned: metaphysical idealists, like Berkeley, could, as I have indicated, offer another “explanation” for the long-run coherence of the (or a) dynamically coherent system of beliefs which fulfills the Observation Requirement.  BonJour’s failure to consider this case indicates one problem with his a priori argument: if it is a good argument, it must consider all the a priori possible explanations, and he considers two!  Moreover, he must offer a clear (and a priori) “metric” for evaluating the quality of these “competing alternative explanations.” 


8.4 Skeptical Hypotheses:


In his Pyrrhonistic Reflections on Knowledge and Justification, Robert Fogelin helps clarify the nature of BonJour’s argument in this section:


it will be important to get clear what sort of probability argument BonJour is offering.  Obviously, it cannot be an argument based on the probability calculus.  The question “How likely is it that a coherent, stable, observationally rich system corresponds with an sich reality?” is not like the question “If two cards are drawn from a deck, how likely is it that the first is a spade, and the second has a value less than six?”  Furthermore, the probability argument cannot rely on empirical correlations between beliefs within the system and a reality outside it, for it is this very correlation that is at issue.  In place of these standard forms of probabilistic reasoning, which will not serve his purposes, BonJour presents a novel form of a probability argument, namely an a priori argument intended to show that the antecedent probability of a skeptical hypothesis is lower than the antecedent probability of the antiskeptical hypothesis that BonJour is defending. [10]   


179 Evil demons, mad psychologists, and brains in vats are all used by skeptics to call our claims to knowledge into question. 


180 The skeptical problems are not unique (or even special) for the coherence theorists—all epistemologists are (equally) plagued by these problems. 


-Problem: Is this true?  What about externalists?  Doesn’t he note that these theorists will be able to reject the skeptical challenges?  Moreover, there is one epistemological position for which skepticism is not a problem—skepticism itself. 


181 “It is a fact of probability theory, which I assume may be relied upon here,[11] that the relative probability of two hypotheses on the same evidence is a function of the following two factors: first, the probability of that evidence relative to each hypothesis; and second, the antecedent or prior probability of each hypothesis.” [12] 


p(h1|E) = [p(e|h1)*p(h1])]/p(e) 


p(h2|E) = [p(e|h2)*p(h2])]/p(e) 


-the comparison of the probabilities of the rival hypotheses on the same evidence, then reduces to a consideration of the probability of the evidence given the hypotheses and the antecedent probability of the hypotheses.


-“Since it is clear (see below for elaboration) that an appropriately chosen skeptical hypothesis can make the first of these factors (that is, the probability of the evidence given the skeptical hypothesis) at least as great as it is for the correspondence hypothesis, an argument in favor of the greater probability or likelihood of the correspondence hypothesis must be based on the second factor (that is, the antecedent probability of the skeptical hypotheses): that is, it must argue that such skeptical hypotheses are antecedently less likely to be true than is the correspondence hypothesis.  And since any appeal to empirical considerations would obviously beg the question in the present context, the antecedent probability or likelihood in question will have to be entirely a priori in character.” 


-“The basic suggestion, to be elaborated in the balance of this section, is that it is the very versatility of skeptical hypotheses of the variety in question, their ability to explain any sort of experience equally well, which renders them not merely methodologically less satisfactory as explanations but less likely to be true, given the fact of a coherent (and stable) system of beliefs.”[13] 


182-183 Consider the “alternative” that coherent system is the result of chance:


-BonJour’s claim is that [182] “...such a hypothesis is extremely unlikely to be true on a purely a priori basis....” 


-“...the reason why the antecedent, a priori likelihood or probability of the elaborated chance hypothesis [the view that the long-run stability of the whole system is merely a result of chance] is extremely low is fundamentally the same as that given above for the relative unlikelihood of the simple chance hypothesis [the view that the fact that the observation component coheres is just a chance event] vis-a-vis the fact of a coherent (and stable) system of beliefs: the internal tension or probabilistic incompatibility between (a) the claim that the observational beliefs are produced purely by chance, and (b) the claim that they continue even in the long run to satisfy the complicated and demanding pattern required in order to be coherence-conducive.  If these two claims are considered separately, the latter is unlikely relative to the former, and hence also the former relative to evidence constituted by the latter; whereas if they are combined into one hypothesis, this same incompatibility, now internal to the hypothesis itself, makes that hypothesis unlikely to be true on a purely a priori or intrinsic basis.” 


183 Similarly, the “Cartesian demon” alternative is unlikely—on a priori probability considerations:


-184-185 “Just as the relation or tension or probabilistic incompatibility that holds between the simple chance hypothesis and the fact of the long-run coherence (and stability) of my system of beliefs is internalized by the elaborated chance hypothesis, so also the analogous relation of incompatibility which was just argued to exist between simple demon hypotheses and the existence of such a system is internalized by elaborated demon hypotheses, with the result that an elaborated demon hypothesis of the sort just indicated is very unlikely, on purely a priori grounds, to be true.  The likelihood that a demon would have just such desires and purposes (and that these would not change) seems no less great than the unlikelihood that an unsuspected demon would produce just such observations....the elaborated demon hypothesis is, like the elaborated chance hypothesis, extremely unlikely [a priori] to be true.” 


185 BonJour notes, however, that the a priori argument offered so far is not sufficient.  It is not enough to show that the skeptical hypothesis is unlikely, instead one must show that his hypothesis is “the best explanation.” 


-“The principle point at which the correspondence hypothesis seems to be vulnerable to an argument...is in its assertion that the cognitively spontaneous beliefs which are claimed within the system to be reliable are systematically caused by the kinds of external situations which they assert to obtain.  It does not seem especially more or less likely a priori that there should be a world of the sort in question and that it should cause beliefs in some way or other than that there should be a demon which causes beliefs, leaving the two sorts of hypotheses roughly on a par in this respect.  But if this is so, then it can be argued that the correspondence hypothesis is just as unlikely to be true on a purely a priori basis as are demon hypotheses.  It is unlikely, relative to all the possible ways in which beliefs could be caused by the world, that they would be caused in the specific way required by the correspondence hypothesis.” 


--Critical Comment: this, of course, seems to totally undercut the sort of argument he must offer to support the second premise of his two premise a priori argument for the M.J Thesis!  Any response to this “problem,” of course, must be an a priori one if his metajustification argument is to be judged adequate. 


186-187 BonJour’s response to this “problem” is to maintain that “although the elaborated causal hypothesis...is admittedly extremely unlikely relative to the general claim that the beliefs in question are caused by a normal world, there are two correlative reasons why it is less so than the foregoing argument might suggest.  In the first place, it is not the case...that the simple causal hypothesis...makes all or virtually all possible patterns of beliefs and of belief causation equally likely to occur.  While the whole point of a skeptical hypothesis like that of the evil demon is to be completely and equally compatible with any resulting pattern of experience, and thus neither refutable nor disconfirmable by any such pattern, this is not true of the hypothesis that beliefs are caused by a spatio-temporal world of a more or less ordinary sort.” 


-Criticism: this difference between the correspondence and skeptical hypotheses will not do for BonJour’s argument.  His argument must be completely a priori, and the sorts of confirmation he is talking about here (empirical observations) aren’t available a priori. 


187 “Second, and more important, there is available a complicated albeit schematic account in terms of biological evolution and to some extent also cultural and conceptual evolution which explains how cognitive beings whose spontaneous beliefs are connected with the world in the right way could come to exist—an explanation which, speaking very intuitively, arises from within the general picture provided...by the correspondence hypothesis, rather than being arbitrarily imposed from the outside.” 


--Critical Comments: clearly, this consideration is also a posteriori in character?  BonJour needs to confine himself to a priori considerations if his metajustificatory argument is to be successful!  As William Alston maintains in his The Reliability of Sense Perception, “...it would be child’s play to build something into the demon hypothesis as well that would explain why the demon prefers to produce sense experiences in the way we have them, rather than in some other way and rather than not producing any at all.  Any appearance of superiority of the “correspondence hypothesis” on this point stems from surreptitiously supposing ourselves to have independent grounds for accepting the evolutionary account, while lacking any independent grounds for suppositions about a demon.  But, of course, those independent grounds are drawn from...parts of our customary doxastic practices; to appeal to them would land us in epistemic circularity again.”[14]  


188 Overall, then, BonJour contends that the second premise of the metajustification argument is supported (against the “skeptical alternatives”) by the fact that “...the degree of a priori unlikelihood which pertains to the correspondence hypothesis, though admittedly large, is substantially less than that which pertains to elaborated demon hypotheses, making the correspondence hypothesis substantially more likely to be true as an explanation of the long-run coherence (and stability) of my system of beliefs.  And this result, taken together with the argument of the previous section, makes it reasonable to accept Thesis M.J.” 




There is an Appendix on “A Priori Justification” and an Appendix which provides “A Survey of Coherence Theories.”  The discussion on pp. 210-211 is important for understanding his conception of a priori justification, and the whole of that Appendix should be carefully read. 



Notes: (click on note number to return to text for the note)

[1] BonJour is citing from Hilary Putnam, Reason, Truth, and History (Cambridge: Cambridge U.P., 1981), p. 49. 

[2] BonJour is citing from Aristotle’s Metaphysics 1011 b26.  Cf., The Basic Works of Aristotle, ed. Richard McKeon (N.Y.: Random House, 1941), the translation of the Metaphysics there is by W.D. Ross. 

[3] Alvin Goldman, “BonJour’s The Structure of Empirical Knowledge, in The Current State of the Coherence Theory, ed. J. Bender (Dordrecht: Kluwer, 1989), pp. 105-114, p. 113. 

[4] Laurence BonJour, In Defense of Pure Reason: A Rationalist Account of A Priori Justification (Cambridge: Cambridge U.P., 1998. 

[5] Alvin Goldman, “BonJour’s The Structure of Empirical Knowledge,” op. cit., p. 107. 

[6] Cf., Thomas Kuhn, The Structure of Scientific Revolutions [1962] (second edition) (Chicago: Univ. of Chicago, 1970).  Also relevant here are Stephen Toulmin, Human Understanding: The Collective Use and Evolution of Concepts (Princeton: Princeton U.P., 1972), and Larry Laudan, Progress and Its Problems (Berkeley: Univ. of California, 1977)

[7] I capitalized ‘rationalist’ here (and the correlative term ‘irrationalist’) because I want to draw a distinction here between the use of the term in this context and in the context of the distinction between empiricistically and rationalistically oriented philosophers.  In the latter context ‘rationalism’ is meant to identify those philosophers who offer a theory which assigns epistemic priority to a priori truths (whereas empiricists assign priority to a posteriori ones).  In the current context, however, ‘rationalism’ is intended to capture those philosophers who are committed to the idea that “everything” has a rational explanation (though, of course, we may not happen to know what that explanation is).  The correlative term (‘irrationalism’, or better ‘fideism’) captures thinkers who contend that there are some “things” which can not be rationally explained, and, thus, must be taken on faith. 

[8] Alan Goldman, “BonJour’s Coherentism,” in The Current State of the Coherence Theory, ed., J.W. Bender, op. cit., pp., 125-133, p. 131. 

[9] Cf., Marshall Swain, “BonJour’s Coherence Theory of Justification,” in The Current State of the Coherence Theory, ed. J.W. Bender, op. cit., pp. 115-124, p. 123. 

[10] Robert Fogelin, Pyrrhonistic Reflections on Knowledge and Justification, op. cit., p. 160. 

[11] Well, may he legitimately assume this?  Given his internalism, and his view of epistemic responsibility, how does he avoid having to provide an argument for this “fact?”  Moreover, is it intuitively obvious in the way he says a priori truths should be (cf., pp. 210-211)? 

[12] Note that in addition to the above “fact” of probability theory, additional facts (of mathematics and probability theory) are being employed here.  The same questions as above arise here.  Note, also, that the talk of comparing the (antecedent) probability of rival “hypotheses” is a bit odd given that we are here, of necessity, speaking at the a priori level—we are not speaking of “empirical hypotheses.”

[13] Note that this is not a fact of probability theory.  This claim is essential if this branch of the “metajustification argument” is to successful, and this means it must be justifiable in a non-circular, non-question-begging, a priori manner. 

[14] William Alston, The Reliability of Sense Perception (Ithaca: Cornell U.P., 1993), p. 88. 

Return to PHI 3300 Home page

File revised on 11/07/2013