|Theory of Knowledge
BonJour's 'Basic Antifoundationalist Argument'
ABSTRACT: BonJour argues that there can be no basic empirical beliefs. But premises three and four jointly entail BonJours Rule ones belief that p is justified only if one justifiably believes the premises of an argument that makes p highly likely which, given human psychology, entails global skepticism. His responses to the charge of skepticism, restricting premise three to basic beliefs and noting that the Rule does not require explicit belief, fail. Moreover, the Rule does not express an epistemic duty. Finally, his argument against this fails since it is false that if an experiential state has representational content, then it is in need of justification. I venture the diagnosis that BonJour mistook the representational content of a cognitive state for the assertive functional role of a belief. Foundationalism may well be false, but not for BonJours reasons.
Laurence BonJour observes that critics of foundationalism tend to argue against it by objecting to "relatively idiosyncratic" versions of it, a strategy which has "proven in the main to be superficial and ultimately ineffective" since answers immune to the objections emerge quickly. (1) BonJour aims to rectify this deficiency. Specifically, he argues that the very soul of foundationalism, "the concept of a basic empirical belief," is incoherent. (2) This is a bold strategy from which we can learn even if, in the end, as I shall argue, it fails.
But, first, what is foundationalism? A person's belief is nonbasic just in case it is justified in virtue of its relation to other justified beliefs; it is basic just in case it is justified but not in virtue of its relation to other justified beliefs. Foundationalism is the view that if one has a nonbasic belief, then in the final analysis it is justified in virtue of its relation to a basic belief. Basic beliefs comprise the foundation of a person's network of justified beliefs. Now to BonJour's argument.
1. The Argument Stated
BonJour summarizes it like this:
What should we make of this argument? (4)
2. Two Objections
2.1 Premise 3 begs the question
Consider premise 3. What is it for a person to be 'in cognitive possession' of a reason why one's belief is likely to be true? BonJour says that "in order for B to be justifed for a particular person A..., it is necessary...that A himself be in cognitive possession of that justification, that is, that he believe the appropriate premises of forms (1) and (2)...," specifically,
So premise 3 is just the claim that a belief is justified for a person only if she believes the premises of a justifying argument. Foundationalism is the denial of this claim. Thus, premise 3 begs the question.
BonJour argues for 3, however. That argument is a more basic antifoundationalist argument than the one BonJour labels as such. I evaluate it in section 3 below. (6)
2.2 Premises 3 and 4 jointly entail global skepticism
According to BonJour, merely believing the premises of a justifying argument is not enough. Premise 4 says one must believe those premises "with justification". Thus, premises 3 and 4 entail what I call 'BonJour's Rule':
Note that BonJour's Rule is perfectly general, so it applies to the beliefs mentioned in conditions (i) and (ii); moreover, it is reiterative, so it entails that a person's belief that p at a time is justified only if she has infinitely many other justified beliefs of a certain sort at that time. Two worries arise: given our present cognitive powers, humans cannot have infinitely many beliefs at once and, even if we could, the type of beliefs generated by the reiterative application of the Rule are too complex for us to grasp them all. Thus, if BonJour's Rule is true, we justifiedly believe nothing since we lack the capacities it requires.
2.3 Four replies
Reply 1. BonJour is aware of this objection. He admits the text suggests that the justifying argument one must have "must always take the form of an explicit metabelief". But this is not what he meant:
I grant the premise: the Rule does not imply that a person's belief is justified only if she explicitly believes the premises of a justifying argument. But it does not follow that the Rule does not require "an infinite hierarchy of metabeliefs"; for unless one at least tacitly or dispositionally or non-occurrently believes the premises of a justifying argument, one breaks the Rule. The specter of skepticism cannot be turned away so easily.
Reply 2. Perhaps we can meet BonJour's Rule by dispositionally believing the premises of a justifying argument. Indeed, since one's dispositionally believing a proposition just is to be such that one would affirm it sincerely and unhesitatingly were one to consider it, and since we have infinitely many such dispositions, then we can have infinitely many beliefs at once.
This reply fails. First, a dispositional belief is not a disposition to believe. (8) Second, even if we can have infinitely many beliefs at once, we cannot have infinitely many of the sort required by BonJour's Rule, namely justified metabeliefs. Allow me to explain.
Take some mundane perceptual belief, say, my belief that
According to the Rule, I justifiedly believe B only if I justifiedly believe that my belief that B has feature Ø and beliefs having Ø are highly likely to be true. Suppose, just for the sake of illustration, that feature Ø is the only feature in virtue of which a belief is highly likely to be true. (9) And let's focus on my belief that
Of course, my believing B1 might be tacit or dispositional. But whether I believe B1 tacitly or explicitly, I must believe B1 if I am to believe B in accordance with the Rule, and I must believe B1 justifiedly. Since I must believe B1 justifiedly, the Rule applies to my believing B1. Thus, my belief that B there is a cup before me is justified only if I justifiedly believe the premises of another justifying argument, which includes
By parity of reasoning, I must also justifiedly believe that
And so on, ad infinitum. Again, none of my metabeliefs here must be explicit; they might be dispositional. And on the view we are now evaluating, that comes to the claim that for each of the infinitely many propositions indicated here, if I were to consider it, I would sincerely and unhesitatingly affirm it. That conditional claim must be true if I justifiedly believe B that there is a cup before me given BonJour's Rule glossed by the view under evaluation.
The first worry here is that it is absolutely impossible for me to grasp, much less consider, most of the propositions generated by the Rule. Let n be the number of nannoseconds since the Big Bang, to the n-th power. It seems dubious that there is a possible world at which I even grasp, much less consider, Bn. In that case, it is at best trivially true that I would sincerely and unhesitatingly affirm Bn if I were to entertain it trivially true because anything follows from an impossibility. But even if I'm wrong about this, the conditional claim in question i.e., that for each of the infinitely many propositions indicated above, if I were to consider it, I would sincerely and unhesitatingly affirm it still seems false. For suppose I consider, say, B4, right now. Although it taxes my ability to concentrate and it is difficult to grasp, I can just barely get a grip on it, for a fleeting moment or two. Do I sincerely and unhesitatingly affirm it? Of course not! I can barely comprehend it. Moreover, given my difficulty in even understanding what B4 says, I would not be justified in affiriming it. So the conditional claim in question is false. As a consequence, if BonJour's Rule is true, then my belief that B that there is a cup before me like any other mundane perceptual belief of mine, is not justified. Generalizing, the Rule does not avoid skepticism even if dispositional beliefs are dispositions to believe and we can have infinitely many dispositional beliefs. (10)
Reply 3. A completely different response to my contention that the Rule entails skepticism is to modify the premises of the basic antifoundationalist argument so that they do not entail it. For example, one might restrict premise 3 to basic beliefs "A basic belief is justified for a person only if she is in cognitive possession of such a reason" from which, in conjunction with 4, we get 'BonJour's Rule*':
Unfortunately, this modification of premise 3 is ad hoc; there is no principled way to restrict the 'cognitive possession' of a justifying argument to basic beliefs. (12)
Reply 4. Another response distinguishes the claim that a person must believe with justification the premises of a justifying argument from the claim that a person must be justified were she to believe those premises. (13) That is, we might read premise 4 as
In conjunction with premise 3, we have 'BonJour's Rule**':
It is arguable that difficulties analogous to those for the dispositional construal of the infinite belief thesis arise here. More importantly, however, note that the condition expressed by the new premise 4 and BonJour's Rule** can be met without one believing, even tacitly, the premises of a justifying argument. Thus, if we use the new 4, BonJour's argument becomes invalid. All that follows is that the justification of a person's basic belief depends on the fact that she would justifiedly believe propositions (i) and (ii) were she to consider them. That conclusion is compatible with the idea that basic empirical beliefs are justifed beliefs whose justification does not depend on the justification of any further actual beliefs. So even if this response removes the sting of skepticism, the cost is prohibitive. (14)
I see no way to fix BonJour's 'basic antifoundationalist argument' so that it supports its conclusion without either begging the question or entailing global skepticism. Let's return to premise 3 and examine his argument for it. Perhaps there we can find a more telling antifoundationalist case.
3. Bonjour's Defense of Premise 3
Premise 3 says that a person justifiedly believes something only if he believes the premises of a justifying argument. Why believe this? Because, says BonJour, a person "cannot be epistemically responsible in accepting the belief unless he himself has access to the justification; for otherwise, he has no reason for thinking that the belief is at all likely to be true." (15) The idea is this: if one is to be epistemically responsible in believing a proposition, one must have 'access' to a 'reason' for thinking that it is true. BonJour's gloss on 'access' and 'reason' leaves us with the claim that epistemically responsible belief requires believing the premises of a justifying argument with justification. Leaving aside worries about the propriety of applying deontological concepts to beliefs, (16) if this claim is true, there can be no basic beliefs.
This claim, however, is false. First, I can believe responsibly even if there is no justifying argument. Suppose I try as hard as I can to discover whether p is true; upon careful investigation, it seems to me that I have found a knockdown argument. In that case, I am not blameworthy for believing p and contrary to appearances the argument turns on a subtle error and, in fact, there is no good argument for p. For, I have done everything in my power to get at the truth. (17) Second, even if there is a justifying argument, I can believe responsibly without believing it. For example, suppose I believe that I exist on the basis of no grounds at all, propositional or nonpropositional. I believe it, say, because, unbeknownst to me, it is biologically guaranteed by God or Nature. In that case, I cannot rid myself of it. Surely I am not irresponsible if I continue to believe. What else can I do? Third, if epistemic responsibility entails premise 3 that I justifiedly believe p only if I justifiedly believe the premises of a justifying argument then, given the reiterative nature of that premise and the skepticism it induces, I am irresponsible unless I have infinitely many justified metabeliefs or believe nothing. I can do neither. So, since I am responsible only for what I can do, epistemic responsibility does not entail premise 3. (18)
So much for BonJour's argument for premise 3. What might be offered against it?
4. Bonjour's Attack on the Doctrine of the Given
Foundationalists offer the epistemic regress argument, modified to display the defects of coherentism. I need not rehearse it here. Recall that it concludes that we are justified in believing empirical propositions only if there are basic beliefs. BonJour is perplexed by this suggestion: "On what basis is such a belief supposed to be justified, once any appeal to further empirical premises is ruled out?" (19) Foundationalists answer differently. One answer, however, is that "the given" provides a basis. Since the doctrine of the given is understood in various ways, we need to have before us its central thesis, which is simply that experience can play a crucial role in how it is that a basic belief can be justified and thereby end the regress. (20) Thus, the central thesis is compatible with many variations ranging from the artsy "what is given in experience are 'phenomenal entities' ('sense-data') and basic beliefs are infallible or certain or self-justifying 'statements' about them" (21) to the mundane "what is given by way of experience are physical objects, and basic beliefs are about them."
Now, BonJour aims to refute "the idea of the given in general" and thus attacks its central thesis. (22) What's his argument? At the end of each discussion of particular versions of the doctrine of the given, (23) BonJour offers a dilemma, the upshot of which is this:
Premise 2 is pretty clearly false. (24) For if an experiential state were in need of justification because it has representational content, then, since every cognitive state has content, each would be in need of justification. But this consequence false. My imagining what life would be like were I a monk has content but needs no justification. Similarly for wondering whether, entertaining, hypothesizing, being ambivalent about, hoping, having a visual experience, and many other types of cognitive states.
What has gone wrong here? The best diagnosis I can come up with is this. Recall that while every cognitive state has a representational content, different states play different functional roles. One important distinction between different kinds of roles is that while some types of cognitive states, notably beliefs, are 'assertive' that is, they involve what might be analogically described as a commitment to the truth of a proposition or the obtaining of a state of affairs others are 'nonassertive', e.g. those mentioned in the last paragraph. And here an important fact emerges. An assertive cognitive state needs to be justified because, and only because, it involves a commitment to the truth or obtaining of its content. A nonassertive cognitive state, however, involves no such commitment and so it does not need to be justified. It is not the content of a state like belief that entails its need to be justified; it is its assertive aspect. Thus, since an experiential state has no assertive aspect, it does not need to be justified even if it has a representational content. In short, BonJour seems to have confused the assertive character of a belief with its representational content, thereby wrongly inferring that any state with content is in need of epistemic justification. (25)
For all I have argued, the doctrine of the given deserves the derision it has received lately. More generally, nothing I have said precludes a sensible rejection of foundationalism. I hope to have shown, however, that we can reject neither for BonJour's reasons. (26)
Notes(1) BonJour 1985, p. 17. (2) Ibid., p. 30. (3) Ibid., p. 32, lightly edited. (4) BonJour says foundationalists will attack premise 3 or 4 (pp. 32-33). But consider premise 2. Many epistemologists think a belief can be justified for a person even if there is no reason why it is likely to be true. Just recall those who use demon-world scenarios to argue against reliabilism. Of course, a foundationalist could take this line too. So why cant she just reject premise 2 and thereby avoid BonJours labyrinthine discussions of premises 3 and 4? Indeed, given his aim "to formulate an objection to all foundationalist views" (p. 17), premise 2 looks like a colossal oversight. As it turns out, however, premise 2 is not essential to the argument. For even if there is no reason that makes ones belief likely to be true, one cannot be justified unless there at least seems to be such a reason, or so one might argue. More could be said about the revised 2, but my main concerns lie elsewhere. I will proceed as though 2 is true. (5) Ibid., p. 31, my emphasis. Moreover, BonJour describes those who deny premise 3 of the basic antifoundationalist argument as those who say that "it is not necessary that the person for whom the belief is basic know, or justifiably believe, or even believe at all, the premises of such an argument" (p. 32, my emphasis). And, responding to his critics, he writes: "if I am not justified in believing that I have such a reason, then I do not genuinely have it..." (1989a, p. 277, my emphasis). (6) For corroboration on this point, see McGrew 1995, pp. 63-64; Kelley 1986, p. 198, and Alston 1989, p. 73. McGrew and Kelley say premise 4 is the culprit, but given what BonJour means by 'cognitive possession', it's premise 3. (7) BonJour 1989a, pp. 276-77, my emphasis. (8) Sure enough, we frequently do believe what we are disposed sincerely and unhesitatingly to affirm; and our believing a proposition sometimes best explains that disposition. But, it does not follow from these facts that a dispositional belief is a disposition to believe. In fact, these are not the same state. Suppose I am talking too loudly in a quiet restaurant. Although I would sincerely and without hesitation agree that I am talking too loudly were it brought to my attention, I may not dispositionally believe (or disbelieve) this. For more on the infinite belief thesis, see Foley 1978, pp. 311-312, Fumerton 1995, pp. 58-60, and especially Audi 1982, on whom I have relied in this paragraph. BonJour rejects it in 1985, pp. 23-24. (9) This simplifying assumption is doubtless false, but it makes things less messy and my argument doesn't hang on it. (10) I thank Frances Howard-Snyder for working out with me the argument of this paragraph. (11) In reply to Steup 1989, BonJour takes this approach:
(12) Indeed, nonbasic beliefs are more natural candidates for the Rule since their justification is derived from other beliefs. Moreover, BonJour's explicit reason for endorsing premise 3 (1985, p. 31) applies to nonbasic belief just as well as to basic belief, if it applies to either (see section 3 below). (13) Perhaps this suggestion is lurking in BonJour 1985, as when he stresses that a person need only "have access" to a justifying argument (pp. 23, 31), or that it "be available" to her (p. 19), or that she "be able in principle to rehearse it" (p. 19), or that it only be "tacitly or implicitly grasp[ed]" (p. 152) or "tacitly or implicitly involved in the actual cognitive state of a person" (p. 152). If we take such access, availability, grasp and tacit involvement as something other than belief (even dispositional belief), then we can understand what it is to be in "cognitive possession" of the premises of such an argument in a counterfactual fashion. (14) A third objection results from the second. The skepticism latent in BonJour's basic antifoundationalist argument renders it epistemologically self-defeating. Given premises 3 and 4 we get BonJour's Rule; BonJour's Rule entails that we cannot justifiedly believe anything; if we don't justifiedly believe anything, then we don't justifiedly believe the premises of his argument. So, even if his argument is sound, it gives us no justifying reason to reject foundationalism. We can sensibly think otherwise only if we think one of his premises is false.
It is worthwhile asking what leads BonJour to such disastrous epistemic principles. He thinks that epistemic internalism demands those principles. At best, however, he is only half right. That is, at best, he is right that his brand of internalism entails his principles, but he is wrong that internalism per se entails his principles. Indeed, one immensely important thing we can learn from BonJour is that we should not endorse his brand of internalism if we wish to avoid global skepticism. What is it about his particular brand internalism that has such skeptical consequences? First, its object:: one's belief that p is justified only if one has access to an argument from which it follows that p is highly likely to be true. Second, the character of the access required: one has access to that argument only if one justifiedly believes its premises. There are other versions of internalism, of course; but they give us no reason to deny foundationalism.
(15) Ibid., p. 31.
(16) See, e.g., William Alston, "Concepts of Justification" and "The Deontological Conception of Epistemic Justification," in Alston 1989.
(17) BonJour rejects my contention here in the following passage:
This line of thought is equivalent to the following argument: "If I am epistemically responsible in believing p, I have a good reason for thinking it is true; if I have a good reason for thinking p is true, my belief has some feature satisfying the schema; so, . . ." But what does 'good reason for thinking that it is true' mean? Clearly, BonJour means a reason which in fact makes my belief highly likely to be true, for only then could the fact that my reason is 'good' entail that my belief satisfies the schema. In that case, the first premise is false. For I am not responsible for that over which I have no control, and I have no control over whether my reason is 'good' in the sense that it in fact makes my belief highly like to be true. All I can do is try my best to get in a position to find the truth. Having done that, the rest is out of my control. On the other hand, if I can have a 'good reason' for thinking that p is true which does not make my belief highly likely to be true, then the fact that my reason is 'good' does not entail that my belief satisfies the schema in which case the second premise would be false.
(18) It is crucial for my case here that I have not rejected the propriety of the concept of epistemically responsible believing. We may well have all sorts of epistemic duties even if it is false that, for any belief we hold, we have a duty to justifiedly believe the premises of an argument which makes that belief highly likely to be true. I have only argued that we do not have this duty.
(19) BonJour 1985, p. 30.
(20) BonJour writes that, in the context of the epistemic regress argument,
Although BonJour is initially near the mark, he develops the central thesis of the doctrine of the given in three potentially misleading ways.
First, he says that "the most straightforward version of the doctrine" involves the claim that basic beliefs are justified by appeal to certain sorts of experiential states, for example, an "immediate experience," "apprehension" or "intuition" of the state of affairs described by p. This characterization of the doctrine is ambiguous. On the one hand, it might mean that a person's basic belief that p is justified only if he appeals to some experiential state; on the other hand, it might mean that the proponent of the given appeals to some experiential state to explain how a basic belief can be justified. Understood the first way, BonJour characterizes the proponent of the given as either failing to distinguish justifying one's belief that a proposition is true (the activity of justifying) from being justified in believing it (the state of being justified), or insisting that one cannot be justified without showing that one is justified. Neither of these mistakes is central to the doctrine of the given. Understood the second way, however, BonJour is correct.
Second, the proponent of the given may, but need not, claim that the given stops the epistemic regress "by making possible a direct comparision between the basic belief and its object" (my emphasis). She may hold that even if an experiential state does not permit one to compare one's belief with the fact that p, it may still figure in rendering one's belief justified.
Third, the proponent of the given may, but need not, claim that "because I immediately experience the very fact which would make my belief true. . .I am completely justified in holding it" (my emphasis). She may hold that even if my experiential state does not suffice to render my belief completely justified, it can contribute to its being completely justified. Put another way, proponents of the given can say that an experiential state can completely justify a person's belief only if other conditions are met, e.g., she is not aware of any defeaters. And this is what they typically do say.
So, the central doctrine of the given involves neither (a) the claim that one's basic belief is justified only if one appeals to experiential states, nor (b) the claim that such states make it possible to compare one's belief with the state of affairs that makes one's belief true, nor (c) the claim that such states render one's basic belief completely justified.
(21) See, for example, Lewis, 1946, chapter 7; Ayer, 1950; Chisholm, 1964; and Fumerton, 1995.
(22) Ibid., pp. 59-61.
(23) BonJour 1985, pp. 64-65, 69, 78.
(24) Others object otherwise. See, e.g., Steven Rappaport, "BonJour's Objection to Traditional Foundationalism," Dialogue (1989), 433-448, and Evan Fales, A Defense of the Given (Lanham, Maryland: Rowman & Littlefield, 1996), especially pp. 150-72. Their arguments, however, rely on idiosyncratic versions of the doctrine of the given. If I'm right in this section, BonJour's argument fails for much more general reasons.
(25) One might object that so long as an experiential state plays a role in the justification of a belief, the question of whether there is any reason to think that its content is accurate or correct can be legitimately raised. And without a positive answer to this question the capacity of such a state to contribute to a beliefs justification is decisively undermined. (See BonJour 1985, p. 78.)
We must be careful here. Suppose there is a positive answer to the question of whether the content of some (token) experiential state accurately represents "nonconceptual reality," as BonJour puts it. Specifically, suppose the answer is yes. Is it then able to contribute to a beliefs justification? Presumably, BonJour will say no. Rather, he will say that not only must there be such a reason, but one must have it. Perhaps he would say, as he does with respect to belief, that it "is necessary not merely that a reason along the above lines exist in the abstract but also that you yourself be in cognitive possession of that reason, that is, that you believe that reason and this belief be justifed for you." This suggestion is tantamount to the following thesis:
What should we make of this thesis? Consider, firstly, the perspective it requires: on the one hand, there is my experiential state with its content; on the other hand, there is nonconceptual reality; now, by comparing the two, I am supposed to be able to tell whether the first depicts or corresponds to the second. Of course, this is a pipe-dream. There is no such first-person perspective from which such a comparison can be made; and even if there were, it would involve yet another experiential state and the demand would be reiterated, ad infinitum. The fundamental question here, then, is this: why accept this condition on the justicatory role of experiential states? Why suppose that in order for a nonassertive experiential state to contribute to the justification of one's belief one has to be able to tell whether its content matches up with nonconceptual reality? BonJour gives us no answer.
(26) For comments on earlier drafts, I thank William Alston, Terence Cuneo, Kevin Guilfoy, Frances Howard-Snyder, Hud Hudson, Steve Layman, John OLeary-Hawthorne, and Paul Tidman. I benefited from reading to the Northwest Conference for Philosophy, University of Portland, 1996, where William Rottschaefer, my commentator, helped me clarify some things. A Faculty Research Grant from Seattle Pacific University, 1996, supported work on this paper.
Alston, W.P., 1989, Epistemic Justification: Essays in the Theory of Knowledge, Ithaca, New York: Cornell University Press.
Audi, R., 1982, "Believing and Affirming," Mind 91, 115-120.
Ayer, A.J., 1950, "Basic Propositions," in M. Black, ed., Philosophical Analysis, Englewood Cliffs, New Jersey: Prentice-Hall, 60-74.
BonJour, L., 1985, The Structure of Empirical Knowledge, Cambridge, Massachusetts: Harvard University Press.
BonJour, L., 1989a, "Replies and Clarifications," in J.W. Bender, ed, The Current State of the Coherence Theory, Dordrecht: Kluwer, 276-292.
BonJour, L., 1989b, "Reply to Steup," Philosophical Studies 55, 57-63.
Chisholm, R., 1964, "The Myth of the Given," in Philosophy, Englewood Cliffs, New Jersey: Prentice-Hall, 261-286, reprinted in Paul Moser, ed., Empirical Knowledge: Readings in Contemporary Epistemology, Lanham, Maryland: Rowman & Littlefield, 1996, 2nd edition, 73-96.
Feldman, R., 1985, "Reliability and Justification," Monist 68.
Foley, R., 1978, "Inferential Justification and the Infinite Regress," American Philosophical Quarterly 15.
Fumerton, R., 1995, Metaepistemology and Skepticism, Lanham, Maryland: Rowman & Littlefield.
Ginet, C., 1985, "Contra Reliabilism," Monist 68.
Kelley, D., 1986, The Evidence of the Senses, Baton Rouge, Louisiana: Louisiana State University Press.
Lewis, C.I., 1946, An Analysis of Knowledge and Valuation, La Salle: Open Court.
McGrew, T., 1995,The Foundations of Knowledge, Lanham, Maryland: Rowman and Littlefield.
Sosa, E., 1991, "Reliablism and Intellectual Virtue," in Ernest Sosa, Knowledge in Perspective, New York: Cambridge University Press.
Steup, M., "The Regress of Metajustification," Philosophical Studies 55 (1989), 41-56.