Is Klein an Infinitist about Doxastic Justification?

Michael Bergmann


Peter Klein deserves the gratitude and respect of the philosophical community for making us work so hard to defend our dismissal of infinitismóa view that many thought was obviously false.  He still hasnít managed to shake my conviction that infinitism is a mistake though he has certainly demonstrated that it isnít an easy matter to refute a canny defender of it.  In this paper, I will argue not only that his latest attempt to defend infinitism fails to defend the conclusion he arrives at but also that the conclusion he arrives at isnít infinitism.

     I confess that Iíve never been much interested in propositional justification.  Insofar as Iíve been interested in epistemic justification, itís doxastic justification that I care about.  Thereís wide agreement between externalists and internalists and between skeptics and nonskeptics that knowledge requires doxastically justified belief.   Thereís less agreement about whether propositional justification is required or at least about whether itís worth mentioning in an analysis of knowledge.  Moreover, as Klein himself points out (in section 6 of his paper), the regress problem is concerned not with propositional justification but with doxastic justification.  Of course, this doesnít show that one shouldnít be interested in propositional justification.  I mention these considerations only to explain why I care about doxastic justification and why Iíll be ignoring propositional justification in these comments.

     In what follows, Iíll focus first on what Klein says about the requirement, for doxastic justification, that a belief be formed in the right way.  Then Iíll make the following three points: (a) Kleinís solution to the regress problem isnít an infinitist solution; (b) Kleinís position on doxastic justification faces a troubling dilemma; and (c) Kleinís objection to foundationalism fails.


I. Doxastic Justification and the Basing Requirement


Klein says in section 2 of his paper that a personís belief is doxastically justified only if it is held in an epistemically responsible manneróthat is, only if the belief is formed in the right way or held for the right reasons.  I have two complaints related to this point.  The first has to do with what Klein avoids saying: he seems to resist saying that the belief must be properly based and he clearly resists saying that it must be caused in the right way.  But I donít see any good reason for resisting either claim.  Klein admits that a beliefís doxastic justification has to do with the manner in which it is held or formed.  And he says that it must be held for the right reasons.  I take it then that he is requiring, for a belief Bís doxastic justification, that B be held because of some adequate reason the person has for it.  But this is just to agree that it must be properly based on some adequate reason the person has for it.  Is there a causal element to this basing?  This isnít the place for a full-fledged discussion of that point.  But, for the record, Iíll just note that I have yet to see an example of a belief that is plausibly thought to be justified in virtue of being based on something, X, even though X isnít a causal contributor to the belief.[1]

     My second complaint in connection with Kleinís section 2 discussion of the basing requirement on doxastic justification is that he later seems to contradict that discussion.  In section 2, he makes it clear that a belief is doxastically justified only if it is formed or held in the right way for the right reasons.  Yet later, in section 5, he speaks of what:

Ö lies at the heart of the infinitistís requirement for doxastic justification, namely that we be able to produce reasons for our beliefs.  The issue is not what causes our beliefs; but rather whether we can cite a reason for our beliefs.[2]


The point here seems to be that doxastic justification requires not that our beliefs be caused in the right way but that we be able to cite a reason for our beliefs.  In what sense must we be able to do this?  Klein concludes that infinitists can reasonably say that having easy (physical) access to an ordinary encyclopedia containing adequate evidence for p, is sufficient for being able to cite a reason for believing p.  But this is a far cry from what he initially said about doxastic justification.  For clearly, the fact that that person has an encyclopedia in the next room containing adequate evidence for p, is consistent with that person forming or holding her belief that p in an epistemically irresponsible manner, in the wrong way and for the wrong reasons.[3]   Contrary to what Klein says, therefore, when it comes to doxastic justification, the issue is not whether there are easily available reasons we could cite; the issue is what our beliefs are in fact based on.


II. Kleinís Solution to the Regress Problem


Letís turn now to an important problem related to doxastic justification: the epistemic regress problem.  Consider the following reductio version of the regress argument for the foundationalist thesis that our beliefs can be noninferentially (doxastically) justified:

1. A belief of ours can be justified only if it is inferentially justified.   [assume for reductio]

2. A belief of ours can be inferentially justified only if the belief from which it is inferred is a justified belief.

3. Therefore, either our beliefs can be justified via circular reasoning or they can be justified via infinite chains of reasoning or it isnít possible for our beliefs to be justified.  [from 1 and 2]

4. Our beliefs canít be justified via circular reasoning.

5. Our beliefs canít be justified via infinite chains of reasoning.

6. It is possible for our beliefs to be justified.

7. Therefore, 3 is false.  [from 4, 5, and 6]

8. 3 is both true and false. [from 3 and 7]

9. Therefore, 1 is false (i.e., our beliefs can noninferentially justified). [by reductio, from 1-8]


The fact that this argument is valid and has only four premises makes it clear that its foundationalist conclusion can be rejected in the following four ways:

(i) The Unjustified Foundations View: our beliefs can be inferentially justified even if the beliefs from which they are inferred are unjustified (so premise 2 is false).

(ii) Linear coherentism: our beliefs can be justified via circular reasoning (so premise 4 is false).[4]

(iii) Infinitism: our beliefs can be justified via an infinite chain of reasoning (so premise 5 is false).

(iv) Radical Skepticism: itís impossible for our beliefs to be justified (so premise 6 is false)


Given that Klein recognizes that the regress problem has to do with doxastic justification and that his stated purpose (in the opening sentence of his paper) is to explain how infinitism solves this problem, one would expect him to endorse option (iii), which denies premise 5.  But what we in fact find is that his solution to the regress problem is to endorse option (i), which rejects premise 2.[5]

     According to Klein, the infinitist thinks that Sís belief that p is doxastically justified if and only if ďS has engaged in tracing the reasons in virtue of which the proposition p is justified far enough forward to satisfy the contextually determined requirementsĒ.[6]  As I understand Kleinís view here, he is saying there is or could be some context C in which a person S holds a belief, B1, and she bases B1 on (holds it because of, infers it from) her further belief, B2.  Her belief B2 is based on her belief B3 and so on down to B10, her belief on which she bases her belief B9.  In context C, by tracing her reasons for B1 this far, S has done enough to satisfy the contextually determined requirements for B1ís doxastic justification.[7] 

     In order to reject the foundationalist solution to the regress problem, Klein must endorse the following view:

K1: For a belief B to be doxastically justified, it must be based on some other belief. 


From this we may conclude that since B10 in the scenario described above isnít based on some other belief, belief B10 isnít justified.  Now suppose Klein combined K1 with:

K2: A belief can be doxastically justified by being based on some other belief only if that other belief is itself doxastically justified.


From this it would follow that B9 in the above scenario isnít justified, since B9 is inferred from B10, which isnít justified.  But then B8 wouldnít be justified either, and so on right down to B1.  Since that clearly conflicts with Kleinís description of the scenario, he must reject K2.  And to reject K2 is to deny premise 2 from the regress argument given earlier.  Itís to endorse the unjustified foundations view.  Hence, Kleinís solution to the regress problem is not an infinitist solution.


III. A Dilemma for Klein


Klein must either accept or reject K2.  Suppose he rejects it, as he seems to in this paper.  Then he endorses the unjustified foundations view.  One problem with embracing this horn of the dilemma is that, as I just noted above, it doesnít provide the infinitist solution to the regress problem that Klein promised to explain.  But he could just take that promise back and acknowledge that his solution to the regress problem appeals to the unjustified foundations view.  However, this isnít the most serious problem with rejecting K2.  The most serious problem is that it commits him to the view that a belief can be justified even if the reason one has for it (i.e., the belief it is based on or inferred from) is an unjustified belief!  One of Kleinís main complaints about foundationalism is that it makes justification inappropriately easy to obtain by saying that beliefs can be justified without reasonsói.e., without being based on other beliefs.  Klein thinks that, in order to be doxastically justified, all beliefs need to be based on reasons in the form of other beliefs.  Given the spirit of his complaints against foundationalism, one would expect him to require these reasons to be good reasons.  But apparently even bad reasons will do, since the reason can be an unjustified belief.[8]   The suggestion that a belief can be doxastically justified via inference from an unjustified belief seems both highly implausible and contrary to the spirit of Kleinís complaint that foundationalism makes justification too easy to come by.

     If, on the other hand, Klein accepts K2 along with K1, then he is committed to requiring for doxastic justification an infinite number of actual beliefs, each of which is based on another in a non-repeating series.  But it seems completely clear that none of us has an infinite number of actual beliefs, each of which is based on another in such a series.  (As a response to this concern, Kleinís claim that we have an infinite number of reasons available to us is beside the point.)  I assume that the reason Klein doesnít accept K2 in this paper (even though doing so would be more in accord with his claim to be an infinitist about doxastic justification) is that he agrees with me that this would make him obviously committed to global skepticism.


IV. In Defense of Foundationalism


Klein makes it clear that he thinks one important consideration in support of infinitism is that its most plausible competitor, foundationalism, is mistaken.  I want to turn next, therefore, to his objection to foundationalism about doxastic justification.  Klein asks us to imagine a conversation between Sally the Skeptic and Fred the Foundationalist.  Fred begins by asserting p and Sally asks his reason for thinking p is true.  Fred gives his reason r and Sally asks what his reason is for thinking r is true.  This continues on until Fred mentions his belief b which is a basic reason, one which is not based on another belief.  A foundationalist will think Fred neednít give or even have another belief on which b is based, for Fredís belief b can be justified noninferentially, in virtue of having some property F on which noninferential justification supervenes.[9]  Here is Kleinís response:

Sally (or Fredís inner voice) can grant that b has F. But she now asks Fred to consider this question: Are propositions with F likely to be true? Fred has only three responses available if he is to remain acting as a responsible epistemic agent seeking to retain only those beliefs worthy of being retained. He could, of course, get bored or see whatís coming and flee Ė reminiscent of some of the characters in a Platonic dialogue. The context is such that if Fred is interested in holding doxastically justified beliefs, these are his possible responses:

1. Propositions with F are not likely to be true.

2. Itís just as likely that propositions with F are true as it is that they are not true.

3. Propositions with F are likely to be true.


Notice that Klein is here making the following claim:

K3: If Fred sees what is coming and refrains from making any of those three responses, he is holding unexamined beliefs and is not an epistemically responsible agent.


K3 is a crucial component in Kleinís response.  If K3 is mistaken, then this objection to foundationalism doesnít go through.  And K3 is mistaken.  The foundationalist may be aware of Kleinís claim that ďunexamined beliefs arenít worth havingĒ.  However, as Klein notes, the foundationalist can see whatís coming.  For either the allegedly required examination could be satisfied by basing b on an unjustified belief or the allegedly required examination takes K2 for granted.  But examination of oneís beliefs that involves inference from an unjustified belief is clearly useless. And requiring examination that takes K2 for granted leads immediately to global skepticism, which makes such a requirement seem implausible.  So Fred the foundationalist will reject Kleinís examination requirement and refrain from asserting or believing any of 1-3.  In doing so, he is not being an epistemically irresponsible agent.  Heís simply rejecting a highly implausible examination requirement.  Thus K3 is false and Kleinís objection to foundationalism doesnít go through.[10]



Annis, David. 1978.  ďA Contextualist Theory of Epistemic Justification.Ē  American Philosophical Quarterly 15: 213-19.


Bergmann, Michael.  2004.  ďWhatís NOT Wrong with Foundationalism.Ē Philosophy and Phenomenological Research 68: 161-65.


DeRose, Keith. 1992.  ďContextualism and Knowledge Attributions.Ē Philosophy and Phenomenological Research 52: 913-29.


Klein, Peter.  200?.  ďHuman Knowledge and the Infinite Progress of Reasoning.Ē  ??


Lewis, David.  1996.  ďElusive Knowledge.Ē Australasian Journal of Philosophy 74: 549-67.


Plantinga, Alvin. 1993.  Warrant: The Current Debate.  New York: Oxford University Press.



[1] A priori beliefs in necessary mathematical truths donít provide obvious examples of the sort I say I canít find.  It is sometimes pointed out that mathematical beliefs are about the abstract realm of numbers and that things in that realm are causally inert, so that a priori mathematical beliefs canít be causally based on them.  But according to what I take to be the usual understanding of belief grounds (the things on which beliefs are based), they are other mental states of the subject.  So an a priori mathematical belief is based on, for example, a mental state such as the seeming obviousness of its content, not on any abstract item in the realm of numbers.  And this seeming obviousness is itself a mental state that can enter into causal relations.

[2] See section 5, question iii of Kleinís paper.

[3] Klein also requires that the subject is such that she would check the encyclopedia for this evidence were doing so required in the context.  But the satisfaction of that further condition (i.e., being such that she would do such checking) isóeven when combined with the satisfaction of the easy access requirementóalso consistent with the person forming or holding her belief that p in an epistemically irresponsible manner, in the wrong way and for the wrong reasons. 

[4] This differs from the holistic coherentist position which accepts the foundationalist thesis.  According to holistic coherentists, a belief can be (doxastically) justified in virtue of being a member of a coherent set of beliefs, even if it isnít inferred from or based on any of those other beliefs.  See Plantingaís 1993, ch. 4 for further discussion of how coherentists who reject circular reasoning are a species of foundationalists.

[5] In fact, his position is a contextualist version of the unjustified foundations view.  This contextualism isnít the recently discussed variety defended by Lewis (1996) and DeRose (1992).  Rather, itís more like the old-fashioned sort of contextualism defended by David Annis (1978).

[6] See the beginning of section 5 of Kleinís paper.

[7] Obviously Iím assuming that ďtracing out oneís reasonsĒ involves basing oneís beliefs on the further beliefs offered as reasons.  But this seems eminently plausible given that this tracing out of reasons is supposed to contribute to doxastic justification, something that Klein thinks of as depending on oneís holding the belief in question for the right reasons.

[8] In correspondence, Klein has said that not only does Bís doxastic inferential justification not require the justification of the belief on which B is based, it also doesnít require that the person holding B have propositional justification for Bís content (which can be had, according to Klein, only if that person has available an infinite number of non-repeating reasons for Bís content).  Thus, he thinks Sís belief B can be doxastically justified despite the fact that (a) Sís belief B is based on an unjustified belief and (b) S lacks propositional justification for Bís content.

[9] For my views on how Fred should continue this discussion, see my 2004.

[10] Thanks to Jeffrey Brower and Michael Rea for comments on an earlier draft and to Peter Klein for many helpful conversations concerning his views.