Peter Klein is known in contemporary epistemology as one of the few defenders of infinitism, the thesis that justification requires an infinite series of non-repeating reasons. Recently, however, Michael Bergmann [2007] and Carl Ginet [2005b] have suggested that Klein is not an infinitist, but a defender of what I will call Unjustified Foundationalism, the thesis that a belief can be justified even if it is inferred from an unjustified belief. Ginet finds this “result…so counterintuitive that [he] hesitate[s] to attribute [it] to Klein” [2005b: 155].
So is Klein an infinitist as he claims, or is he a proponent of unjustified foundationalism? In this paper, I argue that the answer is “yes.” Klein is an infinitist about propositional justification and an unjustified foundationalist about doxastic justification; the combination of these theses yields an infinitist solution to the regress problem. Is attributing this view to Klein uncharitable? I argue the answer is “no”. Klein has already rebutted the objections to infinitism and we will see that the objections to unjustified foundationalism are flimsy. In §1, I present Klein’s complete theory of justification. In §2, I consider and rebut Bergmann’s and Ginet’s claim that Klein is not an infinitist. In §3, I consider six objections to unjustified foundationalism and find them all wanting.
The following pair of principles constitute the core of Klein’s epistemology:
pj: A proposition p is justified for S iff there is an infinite series of non-repeating propositions available to S such that, beginning with p, each succeeding member is a reason for the immediately preceeding one. [2007: 11]
dj: S’s belief that p is justified iff p is justified for S, S’s has cited, as a reason for p, the first member of the series of propositions justifying p, and for each proposition in the series, S has cited, as a reason for it, its successor if the epistemic context requires it [2007: 11].
We can gloss dj as follows: S’s belief that p is justified iff S has provided enough reasons along an infinite path of reasons to satisfy currrent epistemic demands. A belief that was once justified can cease to be justified if the epistemic situation becomes more demanding and the agent fails to adduce further reasons.
To understand these two theses, we need a distinction between propositional and doxastic justification. Propositional justification (p-justification) is a property that attaches to propositions relative to persons. It is the property of a proposition’s being justified for a person. Whether a person believes a proposition is irrelevant to whether it is justified for that person. What is relevant is whether a person has sufficient reasons or evidence (in the form of propositions) available for a proposition.The sort of connection between propositions for one to justify the other has traditionally been thought to be a logical or probabilistic relation—that one entails the other, or that the conditional probability of one the other is high [cite??]. But it could also be thought of as the obtaining of some sort of counterfactual: p is justified for S just in case there is some q such that q is available to S and were S to believe p on the basis of q, S would be doxastically justified in believing that p. Doxastic justification (d-justification) is a property that attaches to a person’s beliefs (not the contents of those beliefs). Whether a person has adequate reasons or evidence for the content of her belief is not (directly) relevant to d-justification. What is relevant is whether the belief is held in the right way. The relationship between p- and d-justification is controversial, but what seems to be agreed upon is that it does not follow from p’s being (propositionally) justified for S that S’s belief that p is (doxastically) justified. This can be illustrated by a simple example: Suppose Plum has before him enough evidence to reasonably conclude that Green is the killer. Plum, however, is rather obtuse and doesn’t see this. Suppose Plum has an unreasonable distrust of persons named “Green” and on that basis concludes that Green is the killer. The proposition that Green is the killer is justified for Plum, but his belief in that proposition is not justified.When I say that a proposition is justified, I mean that it is propositionally justified; when I say that a belief is justified, I mean that it is doxastically justified.
With this distinction before us, it is easy to see that pj is a thesis about p-justification and dj is a thesis about d-justification. Let us examine each thesis in turn.
pj raises two important questions: What makes one proposition a reason for another? and What is it for a proposition to be available to someone? Klein remains officially neutral on these questions [2007a: ??]. Though a full infinitist epistemology must address such questions, neutrality is perfectly appropriate for Klein’s project. pj is a thesis about the structure of p-justification, and it is with the structure of justification that Klein has thus far been concerned. All that is essential is the claim that a proposition is justified in virtue of being a member of a set of reasons with the requisite structure—Klein can, with some justification, abstract from questions about the nature of reasons and availability.
We can better understand dj by considering an example. Suppose Frank believes that p: Rose’s birthday is September 7. Suppose that p is justified for Frank, and that he cites, as a reason for p, his memory of checking his calender.Those who find it implausible that any proposition could be justified for anyone on Klein’s account should consult his [1999] where he argues that we can have available to us infinite chains of reasons. In an ordinary context, Frank need not provide any further reason for his belief that he remembers checking his calender, e.g., a reason for thinking that he’s not misremembering. But if he’s in a context in which the accuracy of his memory is seriously and reasonably in doubt, he may need to provide a reason for thinking that he’s not misremembering in order for his belief. The idea, then, is that whether one must provide a reason for the reason for a justified belief depends on whether the reason is in question in the current epistemic context. Of course, it’s far from clear what an “epistemic context” amounts to or what and how epistemic contexts make requirements. But, here also, Klein is officially neutral. dj, as a thesis thesis about the structure of d-justification, does not provide an account of epistemic contexts.
pj and dj tell us when propositions and beliefs are justified. But lore has it that justification is not an all or nothing affair: both p- and d-justification come in degrees. One proposition (belief) can be more justified than another proposition (belief), and a propostion (belief) can also become more justified than it previously was (or be more justified than it would otherwise have been). Klein’s theory explains one way that a belief can become more justified than it was, but it offers no explanation of how one belief can be more justified than another and no explanation of how propositions can vary in degree of justification. His theory is not to be faulted for these lacunae, for they can only be filled with full accounts of reasons and evidential support. Suppose p is justified for S, and let R be the series of reasons in virtue of which p is justifed. Then Klein claims:
j+: The more propositions from R that S has cited as reasons for their immediate successors, the more justified S is in believing that p.
To explain j+, Klein likes to use the metaphor of distance travelled along an infinite path of reasons.It is for this reason that Klein views the structure of justification as an infinite progress rather than an infinite regress—travelling along the path represents progress. We will use the following slogan: pump up the number of reasons, pump up the justification. It is important to note, though, that j+ cannot be used for deciding which of two beliefs is more justified.It’s not clear whether Klein is sensitive to this point. S’s belief that p might be more justified than her belief that q even though S has travelled farther along her path of reasons for q than she has for p. I have not cited any reasons for my belief that I am appeared to redly, but it is far more justified than my belief that Bloomington will be sunny tomorrow, for which I have provided a reason for which I have provided a reason…. Note also that there are other ways that a belief that p to become more justified: rather than seeking a reason for q (where q is a reason I have cited for p), I might seek an additional reason for p; or I might replace my old reason for p with a stronger one. j+ says nothing about such issues.
It is also rumored that there is such a thing as complete justification (even if no belief ever instantiates it). This could mean one of two things: (i) If a belief is completely justified, then no belief is more justified than it; or (ii) If a belief is completely justified, then it couldn’t become more justified. This is a distinction between unsurpassed and unsurpasable justification. Klein offers an account of unsurpassable justification:
jc: S’s belief that p is completely justified iff p is justified for S, S’s has cited, as a reason for p, the first member of the series of propositions justifying p, and for each proposition in the series, S has cited, as a reason for it, its successor.
Complete justification requires that S actually produce an infinite number of reasons. Klein notes that this is surely impossible; none of our beliefs are completely justified according to jc, for producing a reason takes time and none of us has an infinite amount of time. Unfortunately, this account of complete justification is easily refuted. For suppose some infinite being satisfies the right-hand side of jc. This being could nevertheless become more justified in believing that p: she could move through an additional infinite series of reasons (perhaps she would have to be a transfinite being for this) or she could replace the original series with one containing much stronger reasons. So she is not completely justified, contradicting jc. dj does not, pace Klein, provide the materials for an account of complete justification.
There is no obvious way to derive from pj an account of how p-justification can come in degrees. After all, every justified proposition is a member of a countably infinite series of reasons. Which are more justified than others (and for whom) will depend on the strength of the reasons in the series. As we have seen, pj by itself is silent on such issues.
Let’s summarize Klein’s theory of justification. He offers theories of both p- and d-justification, analyzing the latter in terms of the former. A proposition is justified for someone if it is a member of an infinite and non-repeating series of reasons available to her. Someone’s belief is justified if its propositional content is justified and she has produced enough reasons from the series to satisfy the current context. Klein also acknowledges the received wisdom that justification comes in degrees, offering an account of how a belief can become more justified than it previously was: pump up the number of reasons, pump up the justification. These are the essentials of Klein’s theory of the structure of epistemic justification.For additional details and many defenses against many objections, the reader can consult [Klein 1999; Klein 2003a; Klein 2003b; Klein 2007a; Klein 2007b]. Getting clear on all of this is real progress. But there remains the issue of branding: is Klein in fact an infinitist?
Bergmann [2007] complains that Klein’s theory of justification is not a version of infinitism. Carl Ginet has filed a similar complaint [2005b: 155; emphasis added]:
Klein seems to [hold] that the longer the chain of inferential justification for a given belief the greater the justification created, and that, if the chain is long enough (but still finite), the justification can ’increase to the degree required for knowledge.’ This seems to give us the result that knowledge does not require infinitely long chains of inferential justification after all: infinitism gives way to inferentialism.
How are we to make sense of the question of whether Klein is an infinitist? Following Bergmann, I think the question is profitably recast: is Klein’s solution to the epistemic regress problem an infinitist solution? Because Klein offers both theories of p- and d-justification, the question whether Klein is “an infinitist” is ambiguous. But since his theories of both p- and d-justification are integral to his solution to the regress problem, we can sensibly ask whether this solution is an infinitist one. And because Klein’s theoretical goal is a solution to the regress problem, the “ism” that describes his solution is the “ism” that describes Klein.
So what is the argument that Klein’s solution to the regress problem is not an infinitist solution? It is this: of the four non-skeptical solutions to the regress problem, Klein accepts Unjustified Foundationalism and denies Infinitism. For consider a traditional regress argument for skepticism:
Therefore,
Therefore,
This is a valid argument with four premises, so if one wants to avoid the skeptical conclusion, one will have to deny one of the four premises. Corresponding to a denial of each of these premises is a view about the structure of d-justification, and part of a solution to the regress problem:Of course, it would be possible for a theorist to deny multiple premises in the regress argument—but we’re interested in pure solutions that deny only one of the premises. Since each of the premises is initially plausible, the most plausible theories will be those that deny only one of them.
Foundationalism: a belief can be non-inferentially justified (deny premise 1)
Unjustified Foundationalism: a belief can be inferentially justified even if the beliefs from which it is inferred are unjustified (deny premise 2).
Linear Coherentism: a belief can be justified via circular reasoning (deny premise 4).
Infinitism: a beliefs can be justified via infinite chains of reasoning (deny premise 5).
How does Klein reject the regress argument? If we remember Klein’s claim that no belief is completely justified (according to jc), it is clear that he denies “Infinitism.” And if we remember Klein’s dj, it is clear that he accepts Unjustified Foundationalism. A quick conclusion might then be thought to follow: Klein has not offered an infinitist solution to the regress problem.
This quick conclusion does not do justice to Klein’s solution, for it ignores his infinitist account of p-justification. We can grant that Klein is not an infinitst about d-justification—plausibly, in order to be an infinitist about d-justification, one must accept Infinitism. Klein is an Unjustified Foundationalist about d-justification.In his response to Bergmann, Klein is strangely cagey on this point. He claims not to accept the “Unjustified Foundations” view, but then admits that he denies the following claim: “A belief can be doxastically justified by being based on some other belief only if that other belief is itself doxastically justified.” I’m not quite sure what to make of this; my best guess is that he objects to the term “inferentially justified”. This would be because, as noted above, he doesn’t think that justification is a property that can be transferred from one belief to another via inference, but is rather a property that emerges when a reason is provided for a belief. If, as Bergmann does [19], one ignores Klein’s account of p-justification, his promise of offering an infinitist solution will indeed seem to be a ruse. But why should we ignore Klein’s account of p-justification? Bergmann gives two reasons [19]:
Klein accepts that knowledge requires d-justification (2007: 6). But he also holds that d-justification requires p-justification. So while (1) may provide us with a reason for focusing on theories of d-justification (insofar as we take knowledge to be of primary interest in epistemology), in evaluating Klein’s theory of d-justification—his solution to the regress problem—we will have to take into account his theory of p-justification.It’s worth noting that several prominent epistemologists deny (1). Plantinga [1993a] famously denies that d-justification is necessary or sufficient for warrant (that property that turns true belief into knowledge) and, more radically, Cohen [??] and Alston [2005] deny that there is any such thing as justification (propositional or doxastic) and so, by implication as non-skeptics, deny that it is required for knowledge. It’s not clear how mattters stand with Armstrong, Dretske, and Nozick, but at the very least they have not explicitly agreed that d-justification is required for knowledge.
Reason (2) strikes me as controversial, but since Klein is prepared to admit it [2007b: ], we can grant the point. If (2) is true, then a solution to the regress problem would involve commitment to a view about the structure of d-justification. The above Foundationalism, Unjustified Foundationalism, Linear Coherentism, and Infinitism are the only possible accounts of the structure of d-justification (though see footnote 8 above). But though one of them must be part of a solution to the regress problem, none by itself constitutes a solution to the regress problem. A solution must include motivation for its account of the structure of d-justification as well as a fuller account of d-justification. But central to Klein’s motivation of Unjustified Foundationalism about d-justification is his claim that p-justification is necessary for d-justification. So even if the regress problem concers d-justification, understanding Klein’s solution will require understanding his account of p-justification.
When we take Klein’s theory of p-justification into account, we see that his solution to the regress problem, though denying Infinitism about d-justification, deserves nevertheless to be called an infinist solution. Recall that, according to dj, if one cites as a reason for one's belief that p, an unjustified belief that q (i.e. a belief for which one has not provided a reason), then q must be unquestioned in the epistemic context if the belief that p is to be justified. If we recall Klein’s claim that for every belief, there is an epistemic context that would call it into question, we end up with the result that any doxastically justified belief faces an infinite series of potential challenges. Whenever one of these challenges arises, if the agent does not meet it by providing a further reason, then the original belief loses its justificaiton. This is one of the principal reasons that Klein claims that p-justification is necessary for d-justification: p-justification is the guarantor of one’s ability to meet all potential challenges.This isn’t, though, to demand that justified beliefs be indefeasible. By “potential challenge” I simply mean potential contexts in which a reason for a belief is required. It’s perfectly consistent with having p-justification in this sense that a future defeater could undercut that justification. Of course, no agent will ever meet each of an infinite series of challenges: people have finite amounts of time and, anyway, the agent’s context will not generally require going very far. But though the actual practice of providing reasons often stops at beliefs for which no reasons have been provided, any such stopping place is provisional. An agent with a doxastically justified belief is always prepared to continue providing reasons, should the need arise. This picture warrants calling Klein’s solution to the regress problem an infinitist one.
I have argued that although Klein does not offer an infinitist account of d-justification, his solution to the regress problem is nevertheless an infinitist one. One way of offering an infinitist solution to the regress problem is by offering an infinitist account of d-justification (what Bergmann called “Infinitism” above); another, Klein’s, is by offering an unjustified foundations view of d-justification and combining it with (i) an infinitst account of p-justification and (ii) suitable constraints on when an unjustified belief can serve as a reason for a justified belief.
But is unjustfied foundationalism unjustifiable? Bergmann finds the view “highly implausible,” and Ginet finds it “so counterintuitive that [he] hesitates to attribute it to Klein” [2005b: 155]. This attitude dates back at least to BonJour’s [1978]. Apparently listing the possibility of unjustified foundationalism merely for completeness’ sake, BonJour dismisses it in a single sentence [1978: 3]. Although such summary dismissal of unjustified foundationalism is common, arguments against the view are not.citations
In this section I consider six objections to unjustified foundationalism. I rebut all six. While some of these objections may condemn some forms of unjustified foundationalism, none tell against Klein’s form of the view. The first five of the objections either beg the question or require premises that are straightforwardly false. The last objection turns out to be a general externalist objection to internalism. Apparently, the best objection to Unjustified Foundationalism is that it’s a form of internalism. That is surprising.
While I take myself to have established this much, I don’t take myself to have shown that those who reject Unjustified Foundationalism are somehow unwarranted in so doing (and I certainly don’t expect to convince anyone that Unjustified Foundationalism is true). If a theorist finds the view very implausible, as many apparently do, this is sufficient to warrant rejecting the view in favor of a more plausible alternative. But Klein finds the view plausible and can give theoretical motivations for it. A theorist cannot refute Klein by pointing out that she finds Unjustified Foundationalism implausible. She must provide an argument against the view. My aim is to show that the available arguments aren’t up to the task.
The first objection to unjustified foundationalism is one raised by Bergmann [2007: 23]:
Klein thinks that, in order to be doxastically justified, all beliefs need to be based on reasons in the form of other beliefs…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.
The insinuation is clear: beliefs supported bad reasons are of too low a calibre to be justified. This objection can be reconstructed as an argument whose conclusion is the denial of unjustified foundationalism:
This argument does not withstand scrutiny: the second premise is either false or begs the question at issue, depending on how we understand the term “reason.”
There is a perfectly legitimate sense of “reason”—the sense of the term that Klein typically intends—in which a reason is just a proposition, believed or not. For example, regardless of whether anyone believes that there is smoke coming from a house, that there is is a (good) reason for believing that the house is on fire. In this sense, the difference between a good reason g and a bad reason b for p is that g is adequate evidence for p, while b is not. True enough, one must believe g in order to have the reason for p; but not having it doesn’t make it a bad reason. In this sense of “reason”, then, (2) is plainly false: beliefs aren’t the sort of thing that can be reasons.The belief that p, in this sense, is not a reason for anything. But of course the fact that someone believes p can be a reason (e.g., for p, if the person is reliable). Only propositions can be reasons, and in order to be good reasons they need not be justifiedly believed. So, on this reading, the Bad Reasons Objection is unsound.
It might be thought that there is another sense of “reason” according to which reasons are anything that can .I’m dubious that there is such a sense of the term. Good reasons, in this sense, are just those that justify a belief, while bad reasons are just those that don’t. On this reading, (1) is analytic. But (2) is now nothing more than the bald claim that an unjustified belief cannot justify another belief. That’s the very point in question. So, on this reading, the Bad Reasons Objection is question-begging.
I conclude that the Bad Reasons Objection is a bad reason (in either sense!) for rejecting Klein’s unjustified foundationalism.
The next objection is due to BonJour [1978: 3]:
If [unjustified foundationalism] were correct, empirical knowledge would rest ultimately on beliefs which were, from an epistemic standpoint at least, entirely arbitrary and hence incapable of conferring any genuine justification.
This objection can be reconstructed much like the Bad Reasons Objection:
Either the first premise begs the question or the second premise is false, depending on what “arbitrary” is supposed to mean.
To say that a reason is “arbitrary” might mean that it is just as likely as its denial.This can be understood in terms of subjective probabilities, objective probabilities, or epistemic probabilities. If this is what “arbitrary” means, (1) is plausible enough. But (2) is false. The content of an unjustified belief may be justified, and hence more likely than its denial.Klein thinks that p-justification is truth-conducive: if p is justified for S, then p is more likely than not-p for S. Since, according to Klein, the content of any justified belief must be justified, no justified belief is supported by an arbitrary reason—even though many justified beliefs are supported by unjustified beliefs. So if “arbitrary” means “as likely as its denial,” Klein’s version of Unjustified Foundationalism is not subject to the Arbitrariness Objection.This vindicates my claim that Klein’s account of p-justification is important in motivating his Unjustified Foundationalism about d-justification.
Alternatively, to say that a reason is “arbitrary” might mean that it is the content of an unjustified belief. If this is what “arbitrary” means, (2) is analytic. But (1) is plainly question-begging. So if “arbitrary” means “the content of an unjustified belief”, the Arbitrariness Objection has no dialectical force.
I am unable to think of another reading of “arbitrary” on which both premises will be true and neither will be question-begging. I conclude that the Arbitrariness Objection is no reason for rejecting Klein’s Unjustified Foundationalism.
This objection is inspired by an argument Ginet [2005a: 148] employs it against Infinitism, but which can also be run against Unjustified Foundationalism:
This objection is easily dealt with: Klein rejects the first premise, as we saw above. This premise is often taken for granted, but it is rejected by holistic coherentists [BonJour 1985; Lehrer 2000]. Klein is well within his rights to challenge this view of justification. Until compelling arguments for (1) are provided, the No-Transfer Objection is no threat to Klein’s Unjustified Foundationalism.
At this point, we’ve surveyed the existing objections to Unjustified Foundationalism and found them defective. I will now consider three objections that have not, as far as I know, made an explicit appearance in the literature.
We begin with the following thought: Good reasons can be had on the cheap. I can think of numerous good reasons for thinking I’ll win the lottery—too bad I have no reason to think any of them is true! This suggests a new objection to unjustified foundationalism:
The Reasonless Reasons Objection trades on an equivocation: on one reading of “has no reason,” (2) is acceptable to Klein but (1) is false; on another, (1) is true but (2) is false.
On one reading, “S has no reason for p” simply means “There is no reason q available to S for p.” If I criticize someone for having no reason for her belief, it is very natural to understand me as complaining that none of the evidence, arguments, or facts available to her are reasons for her belief. On this reading, Klein is committed to (2).I hesitate to say that the second premise is true on this reading because it is something that foundationalists and externalists deny. That Klein is committed to this premise would be sufficient to render the argument dialectically effective, were the first premise true. But (1) is false: all justified beliefs, according to Klein’s dj, have contents that are propositionally justified. And Klein’s infinitist account of propositional justification guarantees that if p is justified for S, then S has available a reason r1 for p, and a reason r2 for r1, etc. If S justifiedly believes p by citing q as a reason for p, then S has available a reason for q. So, on this reading, the argument is unsound.
On another, less natural, reading, “S has no reason for p” means “S did not cite a reason for p.” On this reading, the first premise is indeed true. But the second premise is false because it entails radical skepticism. For, surely, the following is true:
pac: For all p, if S is justified in believing that p, then for all q, if q is in the justificatory history of p for S, then p is not in the justificatory history of q for S.
pac is a denial of linear coherentism about d-justification, which should be uncontroversial. But pac and the current reading of the second premise entail that one must have made an infinite number of unique inferences. But this impossible (even Klein grants as much, as we have seen). So, on this reading, the argument is unsound.
I conclude that there is no consistent reading of (1) and (2) such that both are true. The Reasonless Reasons Objection is unsound.
Many readers might be left with the feeling that the previous objection was on to something. Return to the example used to motivate it: I can conjure up many good reasons for thinking that I will win the lottery tomorrow, but I have no reason to think that any of them are true. Surely, I can’t become jusified in thinking that I’ll win the lottery by citing one of these reasons: I know I’ve no reason to think any of them is true. This yields a new objection to Klein’s unjustified foundationalism:
(2) is plausible on its face.Though foundationalists and externalists should reject it, on certain construals of the term "reason." At any rate, we can argue that Klein is committed to it. Since S knows that she has no reason for q, were she to q as a reason for p, her epistemic context would require S to provide a reason for q. Minimally, one’s epistemic context requires that one be epistemically responsible. Surely, it is irresponsible to draw inferences from propositions you know you have no reason to believe.At least, according to Klein who takes epistemic responsibility to essentially involve providing reasons for your beliefs. But the first premise is thereby false: if S knows she has no reasons for q, she cannot justifiedly believe p by citing q. The Known Reasonless Reasons Objection is unsound.
But perhaps the real resistance to an unjustified foundations view lies not so much in principled arguments as in envisioned counterexamples. In this section, I will consider two such examples.
Green’s wife recently left him for another man. Green is at a friend’s New Year’s party feeling a bit vulnerable. He has been trying to make eye contact with an attractive woman across the room, but she hasn’t returned his gaze. As the result of wishful thinking, Green comes to believe that the woman is wildly attracted to him and that she avoids his eye contact because she finds his presence so—dizzying. Green infers from this that if he were to walk up to the woman and ask her to leave with him, she would accept.
There is a fairly strong intuition that Green is unjustified in thinking that the woman would run away with him. But doesn’t Klein’s Unjustified Foundationalism return the opposite verdict? True enough, this is a case in which one belief is inferred from another, unjustified, belief. And it’s also true that Klein is committed to the claim that a justfied belief can be supported by an unjustified belief. But, of course, Klein is not committed to the claim that any unjustified belief can be cited as a reason for a justified belief. In fact, he denies that. Minimally, the propositional content of an unjustified belief must be a good reason for a belief inferred from it, if the inferred belief is to be justified. But we may suppose this condition is satisfied in the case at hand. Still, according to Klein, two other conditions must be met: (i ) the inferred belief must be propositionally justified; and (ii ) the epistemic context must not require a reason be provided for the unjustified belief. Only if both of these conditions are satisfied will Klein be committed to claiming that Green’s belief is justified. I think it’s plausible that neither condition is satisfied in this case, but I will only argue that Green’s belief fails the first condition. Why it fails the second condition will become apparent when we consider the next putative counterexample to Unjustified Foundationalism.
Let’s consider whether the content of Green’s belief is justified for him, according to Klein’s infinitist account of p-justification. This is tantamount to asking whether Green has available an infinite series of propositions such that, beginning with the proposition that the woman would accept his offer, each member of the series is a reason for the immediately preceeding member. He has a reason available for his belief that the woman would accept his proposition: that she is wildly attracted to him. But, as the case is described, he has available to him no reason for thinking that the woman is wildly attracted to him. Were he to seek a reason, he’d come up empty-handed. I take it this is typically how it goes with wishful thinking. So the content of Green’s belief is not propositionally justified for him. Hence, Klein’s dj, which embodies his version of unjustified foundationalism, entails that Green’s belief is not justified, in perfect accord with our intuitions.
Suppose Plum has before him enough evidence to justifiedly conclude that Green is the killer. Plum, however, is rather obtuse and doesn’t see this. Suppose Plum would derive great satisfaction from knowing that Green is the killer, and comes to believe that Green is the killer purely as the result of wishful thinking. From this, Plum infers that Green has seen the murderee’s bedroom.
In this case, the intuition is that Plum is not justified in believing that Green is a murderer. But doesn’t Klein’s unjustified foundationalism return the opposite verdict? The answer here is “no.” We can grant that the content of Plum’s belief is justified, since he had enough evidence to justifiedly conclude that Green is the killer. But Plum’s epistemic context surely requires that an adequate reason be provided for any belief about who murderer's identity. Plum, however, has provided no reason at all for his belief that Green is the killer. But then, by condition (ii) above, no belief inferred from this unjustified belief can be justified. So, according to Klein’s dj, Plum is not justified in believing that Green has seen the victim’s bedroom.
The case of unjustifiable wishful thinking was no counterexample to Klein’s Unjustified Foundationalism because the agent lacked the requisite propositional justification. The case of justifiable wishful thinking was no counterexample because the agent failed to live up to the demands of his epistemic context. These two sorts of cases are, I think, standard examples of wishful thinking. So far, cases of wishful thinking do not threaten Klein’s Unjustified Foundationalism.
But perhaps one can construct a case in which (i) S believes that q as a result of wishful thinking, (ii) q is propositionally justified for S, (iii) S’s epistemic context does not require that she provide a reason for q, and (iii) S infers p from q. In such cases, Klein will be committed to claiming that S is justified in believing that p. Let’s look at such a case:
Suppose Green is panning for gold in the Klamath river. Peering into the riverbed, Green has a visual experience representing a large gold nugget resting in the sand. Green forms the belief that he sees a gold nugget in the riverbed. But he doesn’t cite this visual experience as a reason for his belief--he believes it purely as a result of wishful thinking and cites no reason for it. Now, from his belief that he sees a gold nugget in the riverbed, Green infers that he’s going to make a pretty penny when he returns to town.
In this case, we can suppose that Green’s visual experience is an adequate reason for thinking that he sees a gold nugget in the riverbed. And we can suppose that he has a reason for this reason, and a reason for that reason, and so on. It also seems reasonable to assume that Green’s epistemic context does not require that he provide a reason for thinking that he sees a gold nuggett—that’s not the sort of thing we ordinarily need to provide a reason for (nor the sort of thing that we ordinarily do provide a reason for). And, of course, we can assume that believing you see a gold nugget within your reach is an adequate reason for thinking that you’ll make a pretty penny. So, according to Klein’s version of Unjustified Foundationalism, Green is justified in believing that he will make pretty penny when he returns to town. Is the incorrect result?
Intuitions aside, Klein can offer a clear theoretical motivation for his assessment of the case. Klein is interested in an epistemic property that involves providing justification (reasons) for one’s beliefs. This is an essentially dialectical epistemic property (though sometime the dialectic involves just one person). Instantiating the property with respect to a belief requires providing as many reasons for the belief as the current context demands, and being prepared to continue providing reasons, come what may. We can, for the moment, call this property “Klein-justification.” Klein has attempted to capture Klein-justification with his dj (whether dj is adequate to this task is an interesting question, but one that we are not examining here). It is perfectly clear that Green instantiates Klein-justification: he has provided enough reasons for his belief for current purposes, but is prepared to go on providing reasons should that become necessary. Some version of unjustified foundationalism seems to be the correct account of Klein-justification, and the example above only strengthens this point.
But there is a further question: is Klein-justification the same as justification? I imagine some have a strong intuition that Green, in the example above, is not justified in his belief that he will make a pretty penny, even though he is Klein-justified. They will be inclined to take this as a reason for thinking that Klein-justification is not sufficient for justification. I must admit some degree of sympathy with these intuitions, but I think it is important to point out that they are a brand of externalist intuitions. According to one sort of internalism, a person is always in a position to tell whether they are justified and, what is more, to avoid being unjustified (in the best case by justifiedly believing, in the worst by suspending judgment) [Chisholm ????]. What’s important here, though, is not the ability to control whether one is justified, but the ability to tell whether one is justified. Klein seems to accept just this sort of internalism [2007: 6]:
[E]pistemically responsible agents examine their beliefs in order to determine which, if any, are worthy of being kept&helip;[but] it could be that we just can’t adjust our beliefs so that all and only those that we take to be worthy of believing are, in fact, believed. A responsible epistemic agent strives to believe all and only those propositions worthy of belief.
In the case of pan-handling Green, the intuition that he is unjustified stems from the unsavory provenance of his belief: it is inferred from a belief resulting from wishful thinking. But Green is in no position to discover this. If he were, then his epistemic context would surely require that he provide a reason for his belief that he sees a gold nuggett (which he hasn’t) and so Klein wouldn't be committed to the claim that he’s justified in believing that he will get a pretty penny. Of course, it’s not enough that Green cannot tell that his belief is the result of wishful thinking. Some beliefs resulting from wishful thinking are irresponsibly held; sometimes a belief that requires a reason is instead based solely on wishful thinking.The problem with this sort of belief, then, isn’t so much that it is the result of wishful thinking, but rather that the agent has not provided a reason for the belief. Klein seems to think that the etiology of a belief is largely irrrelevant to its epistemic assessment. But this belief isn’t like that: ordinary epistemic contexts do not require that responsible agents provide reasons for their perceptual beliefs. Since (i) Green was in no position to tell that his belief that he saw Gold was the result of wishful thinking and (ii) his epistemic context did not require that he provide a reason for this belief, Green was perfectly responsible to infer that he would make a pretty penny when he returned to town. From the internalist perspective, then, that Green’s belief stems from a belief based on wishful thinking is no strike against it.
No doubt some epistemologists will count this as a strike against this sort of internalism. But then the problem with Klein’s unjustified foundationalism isn’t that it’s a form of unjustified foundationalism but that it’s a form of internalism. And as such, it finds itself in some distinguished company. Isn’t that progress?
I said that the problem with Klein’s unjustified foundationalism is that it’s a form of internalism (granting for the sake of argument that it is a problem). It’s easy to see why. We could weaken the internalist strictures by adding an externalist constraint to Klein’s dj
dj*: S’s belief that p is justified iff p is justified for S, S’s has cited, as a reason for p, the first member of the series of propositions justifying p, and for each proposition in the series, S has cited, as a reason for it, its successor if the epistemic context requires it, and every belief formed to satisfy the epistemic context is formed using a reliable belief-forming process (or a properly functioning cognitive mechanism, or whatever).
dj* would still be a version of unjustified foundationalism, since it would allow for beliefs to be justified by the citing of unjustified beliefs. Critically, though, dj would avoid the sort of counterexample envisioned above. After all, wishful thinking is not a reliable belief-forming process.
We considered five arguments against unjustified foundationalism and three alleged counterexamples. The arguments all turned upon equivocation and question-begging; Klein’s brand of unjustified foundationalism generated the intuitive results in the first two counterexamples; and the last counterexample turned out to count equally against a wide swath of internalist epistemologies. We saw that unjustified foundationalism itself could be modified to avoid this sort of counterexample (at the cost of adding externalist constraints). If Klein’s unjustified foundationalism fares as well as any other form of internalism, then, pace Ginet, I don’t think we should “hesitate to attribute it to Klein.” Especially when it’s a mere epicycle away from avoiding potentially problematic counterexamples. Unjustified foundationalism at least merits further exploration.
Though there is more to say on this subject—perhaps infinitely more—it’s time to conclude. We have seen that Klein is an infinitist about propositional justificaiton and an Unjustified Foundationalist about d-justification. We have seen that he offers an infinitist solution to the regress problem. And, most surprisingly, we have seen that Klein’s unjustified foundationalism has the resources to deal with the few existing objections to the view. I conclude that Klein has travelled far enough along an (hopefully) infinite chain of reasons in defense of pj and dj. What remains to be seen is whether he can develop a compelling epistemology from these structural theses.