Lectures on BonJour’s The Structure of Empirical Knowledge Chapter 5


     Copyright © 2013 Bruce W. Hauptli


Part II. Toward A Coherence Theory of Empirical Knowledge:


Chapter 5. The Elements of Coherentism:


5.1 The Very Idea of A Coherence Theory:


88 BonJour distinguishes coherence theories of justification and truth:[1]


-coherence theories of truth: “...hold that truth is to be simply identified with coherence (presumably coherence with some specified sort of system).”  As we saw earlier (p. 4), BonJour is a classical realist and, thus, adheres to a correspondence theory of truth.  BonJour also points out the distinction between theories regarding the nature of truth and theories regarding the criterion of truth.  According to him, the latter theories should, if confusion is to be avoided, be called:


-coherence theories of justification: theories about the criteria or standards or rules “...which should be appealed to in deciding or judging whether or not something is true....” 


5.2 Linear Versus Nonlinear Justification:


90 The standard argument against coherence theories of justification presumes “...that inferential justification is essentially linear in character, that it involves a one-dimensional sequence of beliefs, ordered by the relation of epistemic priority, along which epistemic justification is passed from the earlier to the later beliefs in the sequence via connections of inference.”  BonJour will propose that we accept a nonlinear view which holds that “...despite its linear appearance, [inferential justification] is essentially systematic or holistic in character: beliefs are justified by being inferentially related to other beliefs in the overall context of a coherent system. 


-Robert Fogelin points out that one should not think of systematic or holistic justification in “linear” terms: “nor does the coherentist permit what might be called a circular form of linearity, that is, a structure of reasons that simply loops back on itself.  For the standard coherentist, linear circularity is a bad form of circularity.  In place of such linear conceptions of justification, the coherentist pictures justification using such metaphors as a network, a mesh, a system, or an organic totality of beliefs.  The fundamental idea is that the items in coherent systems of beliefs must stand in relationships of mutual support.”[2] 


91 BonJour also maintains that we must distinguish between “local level” and “global level” justification: at the local level, justification appears linear: “one quickly reached premise-beliefs which are dialectically acceptable in that particular context and which can thus function rather like the foundationalists’ basic beliefs.  (But these contextually basic beliefs...are unlikely to be only or even primarily beliefs which would be classified as basic by any plausible version of foundationalism).” 


-Alternatives to foundationalist and coherentist theories of justification: we have been, discussing foundationalist and coherentist theories of justification as if they are the only alternatives (whether interpreted in an internalist or an externalist manner).  This is not the case.  In addition to these views of epistemic justification, there is a promising orientation which advances a contextualistic view of justification;[3] and in her Evidence and Inquiry: Towards a Reconstruction in Epistemology, Susan Haack develops a view she calls “foundherentism.”[4] 


While justification at the local level may appear linear, BonJour maintains that at the global level there is no linearity.  Instead, we must talk of mutual or reciprocal support:


-92 “...a coherence theory [of justification] will claim, the apparent circle of justification is not in fact vicious because it is not genuinely a circle: the justification of a particular empirical belief finally depends, not on other particular beliefs as the linear conception of justification would have it, but instead on the overall system and its coherence.”  BonJour’s coherence theory of justification involves four distinct steps or stages for the justification of an empirical belief:


(1) The inferability of that particular belief from other particular beliefs and further relations among particular empirical beliefs. 


(2) The coherence of the overall system of empirical beliefs. 


(3) The justification of the overall system of empirical beliefs. 


(4) The justification of the particular belief in question, by virtue of its membership in the system. 


5.3 The Concept of Coherence:


This section identifies five desiderata[5] for coherence:


95 1. Logical consistency is necessary, but not sufficient, for coherence. 


-Criticisms: in his Pyrrhonistic Reflections on Knowledge and Justification, Robert Fogelin maintains that this condition presents problems for BonJour: “BonJour, taking a standard coherentist approach, insists that formal consistency is not a sufficient condition for a system to be coherent, but he does take it to be a necessary condition....He seems not to have noticed that is very unlikely that the belief system of any human being satisfies this necessary condition.  It seems safe to assume that all mature human beings hold at least some beliefs that are inconsistent with each other or at least imply things that are inconsistent with one another.”[6] 


In his “Internalism Exposed,” Alvin Goldman asks:


is there enough decision interval during which justificationally pertinent formal properties can be computed?  Coherentism says that S is justified in believing p only if p coheres with the rest of S’s belief system held at the time.  Assume that coherence implies logical consistency.  Then coherentism requires that the logical consistency or inconsistency of any proposition p with S’s belief system must qualify as a justifier.  But how quickly can consistency or inconsistency be ascertained by mental computation?  As Christopher Cherniak points out, determination of even tautological consistency is a computationally complex task in the general case.  Using the truth-table method to check for consistency of a belief system with 138 independent atomic propositions, even an ideal computer working at “top speed” (checking each row of a truth table in the time it takes a light ray to traverse the diameter of a proton) would take twenty billion years, the estimated time from the “big bang” dawn of the universe to the present.  Presumably, twenty billion years is not an acceptable doxastic decision interval![7] 


In his The End of Faith, Sam Harris maintains that:


…here we encounter a minor computational difficulty: the number of necessary comparisons grows exponentially as each new proposition is added to the list.  How many beliefs could a perfect brain check for logical contradictions?  The answer is surprising.  Even if a computer were as large as the known universe, built of components no larger than protons, with switching speeds as fast as the speed of light, all laboring in parallel from the moment of the big bang up to the present, it would still be fighting to add a 300th belief to the list.  What does this say about the possibility of our ever guaranteeing that our worldview is perfectly free from contradiction: It is not even a dream within a dream.[8] 


2. A system of beliefs is coherent in proportion to its degree of probabilistic consistency. 


-While beliefs might be logically consistent, there could be a probabilistic inconsistency—it could be unlikely that two beliefs are both true although it is logically consistent to suppose that they are such [for example: my belief that p and my belief that it is extremely improbable that p]. 


98 3. The coherence of a system of beliefs is increased by the presence of inferential connections between its component beliefs and increased in proportion to the number and strength of connections. 


4. The coherence of a system of beliefs is diminished to the extent to which it is divided into systems of beliefs which are relatively unconnected to each other by inferential connections. 


-95-97 As BonJour notes, sets of beliefs could be perfectly consistent but have nothing to do with one another.  Thus the requirement that there be “inferential relationships” between the beliefs guarantees truth preservation.  But there are a variety of such relationships:


--mutual entailment (which is the requirement advocated by Blanshard) is not promising since (as Blanshard admits) even Euclidean geometry may not fit this sort of requirement. 


--97 each proposition must be “entailed by the rest” (advocated by Ewing). 


--“congruence” (advocated by C.I. Lewis) holds that the “antecedent probability of any component belief is increased if the remainder can be assumed as premises.”  This sort of orientation is significantly weaker than that of mutual entailment. 


-98-99 BonJour recommends “explanatory” connections between the beliefs: [99] “explanatory connections are not just additional inferential connections among the beliefs of a system, however; they are inferential connections of a particularly pervasive kind.  This is so because the basic goal of scientific explanation is to exhibit events of widely differing kinds as manifestations of a relatively small number of basic explanatory principles.” 


--Criticism: of course much more needs to be said here (and BonJour recognizes this).  Philosophers of science offer extremely different (and complicated) conceptions of explanation, and given the central importance of this notion for the notion of “coherence,” the lack of detail here points to a possible problem for the overall theory. 


-A discussion of scientific explanation helps clarify the sort of connections desired.  Alan Goldman characterizes explanation as follows: “rendering a fact or event intelligible by showing why it should have been expected to obtain or occur.”[9] 


--BonJour offers a traditional conception of scientific explanation which is often called the “covering law model.”  It holds that in scientific explanations, “...particular facts are explained by appeal to other facts and general laws from which a statement of the explanandum[10] fact may be deductively or probabilistically inferred; and lower-level laws and theories are explained in an analogous fashion by showing them to be deducible from more general laws and theories.” 


5. The coherence of a system of beliefs is decreased in proportion to the presence of unexplained anomalies in the believed content of the system. 


100 BonJour notes that we should not be too quick, however, to connect the notions of inferential relatedness and explanation.  His discussion of standing next to a mouse which is three feet from a four foot high pole on which an owl sits, shows that the Pythagorean theorem doesn’t “explain” why the owl is five feet from the mouse.  In short, ...it is a mistake to tie coherence too closely to the idea of explanation.”  BonJour’s point is that nonexplanatory inferential connections can also enhance coherence. 


BonJour notes that the demand for “systematic unification” which arises as we seek scientific explanations provides a powerful engine for conceptual change. 


Problems: given the emphasis upon “dynamism,” one should note that for such growth one needs inconsistencies, but this undercuts the emphasis upon coherence!  Once anomalies are allowed, there would seem to be no way to measure the degree of consistency! 


-As noted above, in his Pyrrhonistic Reflections on Knowledge and Justification Robert Fogelin maintains that these conditions pose a problem for BonJour: “given this list of standards for coherence, we can ask whether any human system of beliefs has ever satisfied them.”[11] 


5.4 The Doxastic Presumption:


101 BonJour begins the section with a “road map” for Part II by listing what he takes to be the five essential elements of a viable coherence theory of justification:


(1) The idea of nonlinear justification.  [Covered above in section 5.2]


(2) The concept of coherence itself.  [Covered above in 5.3]


(3) The presumption regarding one’s grasp of one’s own system of beliefs...[which] is required...if our coherence theory is to avoid a relapse into externalism.  [Remainder of this Chapter]


(4) The coherentist conception of observation.  [Chapter 6]


(5) The metajustificatory argument....”  [Chapter 8]


[6] At the end of the current chapter, BonJour “raises” three standard objections to the coherence theory of justification [“alternative systems,” “input,” and “truth”] and, thus, a sixth essential item (providing replies to these “standard objections”) is important.  These responses are offered in Chapters 6, 7, and 8. 


-Note, also, that there are also two appendixes: (A) on A Priori Justification, and (B) providing a Survey of earlier Coherence Theories. 


101 Turning to the third task listed above, then, BonJour maintains that for the coherence theorist, “the epistemic justification of an empirical belief derives entirely from its coherence with the believer’s overall system of empirical beliefs and not at all from any sort of factor outside that system.  What we must now ask is whether and how the fact that a belief coheres in this way is cognitively accessible to the believer himself, so that it can give him a reason for accepting the belief.” 


-After all, we saw in his critique of externalist foundationalism, that BonJour adheres to internalist presuppositions and believes, in effect, that the adoption of externalism amounts to the abandonment of epistemology.  One could offer an externalist coherentism, of course, but such an orientation would be subject to all the problems which externalism has already been saddled with:[12]


--101-102 ...“such a view is unacceptable for essentially the same reasons which were offered against foundationalist versions of externalism and, as discussed earlier, seems to run counter to the whole rationale for coherence theories.  (If externalism were acceptable in general, the foundationalist versions would obviously be far simpler and more plausible.)  But if the fact of coherence is to be accessible to the believer, it follows that he must somehow have an adequate grasp of his total system of beliefs, since it is coherence with this system which is at issue.” 


102 Unfortunately for the internalist, however, “...no actual believer possesses an explicit grasp of his overall belief system; if such a grasp exists at all, it must be construed as tacit or implicit, which creates obvious problems for the claim that he is actually, as opposed to potentially, justified....”


-Most coherentists simply take the believer’s grasp of her system for granted. 


103 BonJour relies upon what he calls a “doxastic presumption” which claims that “...the raising of an issue of empirical justification presupposes the existence of some specifiable system of empirical beliefs....the primary justificatory issue is whether or not, under the presumption that I do indeed hold approximately the system of beliefs which I believe myself to hold, those beliefs are justified.  And thus the suggested solution to the problem raised in this section is that the grasp of my system of beliefs which is required if I am to have cognitive access to the fact of coherence is dependent...on this Doxastic Presumption....”


-104 The presumption is only that one’s representation of the overall system of beliefs is approximately correct. 


-The metabelief[13] that one has a system of beliefs does not itself need justification.  The doxastic presumption does not function as a premise in the justificatory argument—instead, it is an unavoidable feature of our cognitive practice: [104-105] “epistemic reflection...begins from a (perhaps tacit) representation of myself as having (approximately) such and such a specific system of beliefs: only relative to such a representation can questions of justification be meaningfully raised and answered.[14]  This representation is presumably a product of something like ordinary introspection...but whereas most introspective beliefs can be justified by appeal to coherence, the metabeliefs which constitute this representation cannot be thus justified in general[15]....the Doxastic Presumption does not...function at all in the normal working of the cognitive system...it simply describes or formulates, from the outside, something that I unavoidably do....” 


My Appendix to 5.4: Problems with the Doxastic Presumption:


(A) Note that it would appear that BonJour has a deep problem with regard to the status of his “doxastic presumption:” it seems to be either (i) a “meta-belief” which is “available” to help justify his beliefs generally—by providing “access” to both (a) the individual belief to be justified and (b) the general coherent system of beliefs which it must cohere with.  But, then, the presumption itself is in need of justification (and the regress problem has not been resolved).  Alternatively, (ii) the doxastic presumption is a “presumption,” rather than a belief; and, thus, not in need of justification.  But, then, it does seem to be legitimate to appeal to it to provide a justification for his beliefs generally.  This problem seems very much parallel to the problem which BonJour adduces for the “givenist” foundationalists—in each case the theorists (givenist foundationalists and BonJourian coherentists) seem to appeal to “basic considerations” which must both provide justification without themselves raising justificational queries.  But to do the former, such considerations seem to need “content” which raises justificational queries; while to do the latter, such considerations seem to lack content, and thus don’t seem “available” as a justificatory basis. 


(B) Michael Williams maintains that the doxastic presumption is problematic because it reintroduces the very notion of epistemic priority which BonJour seeks to reject (see p.90).[16]  According to Williams, such


...metabeliefs must themselves amount to knowledge, or at least be justified.  If they are not justified, appealing to them will do nothing towards providing a global justification of my beliefs about the world....[Thus] the coherence theory extends the concept of appearances.  Appearances are not confined to sensory appearances but cover the entire way we take the world to be: our “accepted world,” as Blanshard would say.  But the metabeliefs that capture our knowledge of appearances, in this extended sense, must still be [epistemologically] privileged.  They must be in some general and fully objective way epistemologically prior to knowledge of the world.  If they are not, grounding knowledge of the world on them will do nothing to improve its epistemological status.[17] 


(C) Indeed, Williams offers a criticism related to (A) above maintaining that the doxastic presumption suffers a fate similar to that which the “givenists” suffer according to BonJour:


if the metabeliefs...can be justified in neither a foundational nor a coherentist fashion, then from BonJour’s internalist standpoint they cannot be justified at all.  Seen in this light, the term, “Doxastic Presumption,” seems to indicate a desire to have things both ways.  While, officially, the metabeliefs the coherence theorist must take for granted do not possess any kind of intrinsic warrant, talk of a “presumption” in favor of their truth is surely suggestive of some measure of intrinsic credibility, albeit defeasible.  It would be clearer to say that coherentist justification is justification under the Doxastic Assumption.  But since no internalist can allow justification to arise out of a brute assumption, putting things this way would simply underline the impossibility of being an internalist without being a foundationalist.[18] 


(D) Note that BonJour’s discussion on pp. 104-105 opens the door for a skeptical challenge.  The possible distinction between (i) our system of beliefs and (ii) our meta-belief(s) about (or our “representation of,” or our “grasp” of) this system legitimates the skeptic’s question as to whether or not our “representation,” “grasp,” or “meta-beliefs about” the system properly mirror the system.  If they do not, then the whole coherentist project dissolves.  In his Pyrrhonian Reflections On Knowledge and Justification, Robert Fogelin maintains that BonJour’s statement [on p. 105] that the coherence theory can not answer the form of skepticism which questions whether our representation of our system of beliefs is accurate (which is, after all, what the Doxastic Presumption asserts):


with disarming candor, BonJour acknowledges that the coherentist cannot meet one form of skepticism—that form of skepticism that challenges the doxastic presumption.  It seems, then, that BonJour’s position has foundered on Agrippa’s third mode: that of hypothesis or arbitrary assumption.  If that’s so, then his project has failed!  Period!  There is nothing to be said next if BonJour’s task is, as he indicated earlier, to refute skepticism.  BonJour has given away the store, yet continues advertising goods for sale.[19] 


(E) note that the Doxastic Presumption Is not only a presumption that we have access to an approximately correct representation of the overall system of beliefs, and to the particular belief in question, and whether or not the latter “coheres with” the former; the presumption must also be regarding the “coherence” of the overall system.  Given critiques about how many statements may be assessed for such coherence (in the minimalistic sense of logical consistency), this is a serious problem for the presumption! 


5.5 The Standard Objections [to the Coherence Theory are Briefly Restated and Clarified:


Returning to the text, in this section BonJour again briefly characterizes the three “standard” objections which are traditionally raised against coherence theories.  Throughout the remainder of the book, he will try to answer these objections and thus further clarify and defend this orientation. 


107 (1) The Alternative Coherent Systems Objection:


Clearly one can’t simply negate the beliefs of a coherent system and assume that another coherent system will arise—once one recognizes that logical consistency is not all there is to coherence, this will not do.  Nonetheless, alternative coherent systems seem a real possibility.  BonJour responds to this objection in Chapter 7. 


108 (2) The Input Objection:


“Coherence is purely a matter of the internal relations between the components of the belief system....Hence if, as a coherence theory claims, coherence is the sole basis for empirical justification, it follows that a system of empirical beliefs might be adequately justified...in spite of being utterly out of contact with the world that it purports to describe.  BonJour responds to this objection in Chapters 6 and 7. 


(3) The Problem of Truth:


109 It must be shown how coherence is truth-conducive.  Absolute idealists, of course, avoid this problem by equating truth with long-run coherence and success, but this begs the question.  BonJour responds to this objection in Chapter 8. 


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

[1] Other theorists (Blanshard and Nicholas Rescher, for example) draw the distinction as one between theories regarding the nature and theories regarding the criteria for truth.  Cf., Brand Blanshard, The Nature of Thought v. 2 (London: Allen and Unwin, 1939), pp. 267-268; and Nicholas Rescher, “Fundamental Aspects of the Coherence Theory of Truth” [1973], in Contemporary Readings in Epistemology, eds. Michael Goodman and Robert Snyder (Englewood Cliffs: Prentice Hall, 1993), pp. 174-185. 

[2] Robert Fogelin, Pyrrhonistic Reflections on Knowledge and Justification, op. cit., pp. 146-147. 

[3] Cf., David Annis, “A Contextualistic Theory of Epistemic Justification,” American Philosophical Quarterly v. 15 (1978), pp. 213-219.  Reprinted in The Theory of Knowledge: Classical and Contemporary Readings (third edition), ed. Louis Pojman (Belmont: Wadsworth, 2003, pp. 248-254. 

[4] Cf., Susan Haack, “A Foundherentist Theory of Empirical Knowledge,” [1997] in The Theory of Knowledge: Classical and Contemporary Readings (third edition), ed. Louis Pojman, op. cit., pp. 237-247, and her Towards a Reconstruction in Epistemology, op. cit. 

[5] That is, a desirable or centrally important (though not, absolutely necessary, characteristic). 

[6] Robert Fogelin, Pyrrhonistic Reflections on Knowledge and Justification, op. cit., p. 149.  Cf., also, Richard Fumerton, Metaepistemology and Skepticism (Lanham: Rowman, 1995), pp. 144-147. 

[7] Alan Goldman, “Internalism Exposed,” Journal of Philosophy v. 96 (1999), pp. 271-293, p. 284. 

[8] Sam Harris, The End of Faith (N.Y.: W.W. Norton, 2004), p. 57. 

[9] Alan Goldman, Empirical Knowledge, op. cit., p. 23. 

[10] Explanandum = what is to be explained. 

[11] Robert Fogelin, Pyrrhonistic Reflections on Knowledge and Justification, op. cit., p. 149. 

[12] In his Pyrrhonistic Reflections on Knowledge and Justification, op. cit., Robert Fogelin discusses Donald Davidson’s orientation as a version of an externalist coherentism in Chapter 9. 

[13] A “meta-belief” is a belief about one’s beliefs (or about one’s system of beliefs). 

[14] The reader must ask herself whether or not this suggests that this “presumption” plays a “foundational” role in BonJour’s theory. 

[15] The reader must ask whether this means that this presumption is an “arbitrary” one.  Clearly, of course, BonJour does not intend this, but what ensures that it is not such? 

[16] As Williams contends, the acceptance of the “doctrine” of “epistemic priority” is a key element of the foundationalists’ theory—they accept the claim that some beliefs are “more basic” than others (that they are [“epistemologically”] prior to the other beliefs).  Coherentists, on the other hand, contend that beliefs are all dependent (at least in terms of their “epistemic status”) on other beliefs (and thus, all on the same “epistemic” level). 

[17] Michael Williams, Unnatural Doubts, op. cit., p. 293. 

[18] Ibid., p. 297 

[19] Robert Fogelin, Pyrrhonistic Reflections On Knowledge and Justification, op. cit., p. 153. 

Go To Hauptli's Supplement for Chapter 6

Return to PHI 3300 Home-page  

File revised on 10/28/2013