1 Introduction

An adequate theory of epistemic justification must be able to account for the differences in the epistemic standings of the subjects in the scenarios below:

Case 1. Last night in her dream, it seemed to Susan as if she were reading an article about a civil war in Iraq. Other than this experience, she is entirely clueless about what is actually going on in Iraq, and Susan does not (come to) believe that there is a civil war in Iraq.

Case 2. John reads an article about a civil war in Iraq in a newspaper that is well-known for its credibility. However, he simply ignores the information he receives from the article and does not come to believe that there is a civil war in Iraq.

Case 3. Michael reads an article about a civil war in Iraq in a newspaper that is well-known for its credibility. Michael comes to believe that there is a civil war in Iraq; however, the reason for which he believes that there is a civil war in Iraq is not his having read that article but his having had a dream similar to Susan’s last night.

Case 4. Mary reads an article about a civil war in Iraq in a newspaper that is well-known for its credibility. Mary comes to believe that there is a civil war in Iraq, and the reason for which she believes that there is a civil war in Iraq is her having read that article.

In cases 1 and 2, neither Susan nor John believes that there is a civil war in Iraq. However, John is, epistemically speaking, better situated with respect to the proposition that there is a civil war in Iraq than Susan. It might plausibly be said that John’s epistemic credentials for the proposition that there is a civil war in Iraq are better than Susan’s. While neither has the belief in question, we might still say that John has something going for the proposition that there is a civil war in Iraq but Susan does not. Examples of this sort can be easily found in the literature, and they are usually deployed to draw the moral that we need a notion of epistemic justification that applies to pairs of subjects and propositions, regardless of whether those subjects believe the corresponding propositions or not; and let us adopt the schema A proposition, p, is (epistemically) justified for a subject, S to articulate the epistemic relation intended to be captured by this notion. So, the proposition that there is a civil war in Iraq is justified for John but not for Susan.¹ It is customary to call this notion of epistemic justification propositional justification.

In cases 3 and 4, both Michael and Mary believe that there is a civil war in Iraq, and both are equally well situated with respect to the proposition that there is a civil war in Iraq. Despite this, Michael’s believing that proposition is epistemically defective in a way in which Mary’s is not. It might be plausibly said that Mary’s believing attitude in question enjoys a more favorable epistemic standing than Michael’s. Examples of this sort are usually deployed to draw the moral that we need a notion of epistemic justification that is different from propositional justification and applies to pairs of subjects and their actual believing attitudes; and let us adopt the schema S’s belief that p is (epistemically) justified to articulate the epistemic relation intended to be captured by this notion. So, the proposition that there is a civil war in Iraq is justified for both Mary and Michael; however, Mary’s believing is justified, but Michael’s is not. It is customary to call this latter notion of epistemic justification doxastic justification.

Once the two notions of epistemic justification are clearly distinguished, it is natural to ask whether one of these notions can be explicated or otherwise accounted for in terms of the other. A standard answer to this question is that propositional justification is explanatorily prior to doxastic justification: a theory of epistemic justification must first explain propositional justification and then explain doxastic justification in terms of propositional justification (plus something else). The fact that doxastic justification requires propositional justification but not vice versa provides some (prima facie) support for this approach. It is uncontroversial that doxastic justification requires at least propositional justification and belief; however, as a quick comparison of cases 3 and 4 above shows, it cannot be plausibly argued that if p is justified for S and S believes p, then S’s believing that p is justified: it is simply false that having propositional justification for p plus believing that p yields doxastic justification for believing p. Rather, there must be a proper connection between S’s believing p and what makes p propositionally justified for S: the former must be based on the latter. According to an orthodox account of the relationship between propositional and doxastic justification, it is necessary and sufficient for a belief that p to be doxastically justified for a subject S, that p is propositionally justified for S and the belief that p is based on what makes p propositionally justified for S.

^{Turri (2010)} argues that the orthodox view of the relationship between propositional and doxastic justification is mistaken.² In particular, Turri maintains that the following principle, which is entailed by the orthodox view, is false:

(Basis) If (i) p is propositionally justified for S in virtue of S’s having reason(s) R, and (ii) S believes p on the basis of R, then S’s belief that p is doxastically justified (p. 314).

Turri provides two counter-examples to (Basis) and concludes that it (and its kin)³ is false. Turri’s counter-examples are designed to show that there are cases in which having good reasons r for believing p plus believing p on the basis of r does not amount to justifiedly believing p. Turri’s cases exploit what he calls a “glaring weakness” in the orthodox view, which does not make any “mention of the way in which the subject performs in forming or sustaining a belief” (p. 315). The first counter-example -the case of Misses Proper and Improper- purports to show that one fails to justifiedly believe p despite having good reasons r for p and believing p on the basis of r, when one bases her belief p on r and some other, epistemically defective reason d. If one comes to believe p “as the result of an episode of explicit, conscious reasoning that features” (p. 316) r and d essentially, then one does not come to justifiedly believe p despite the fact that one has good reasons (namely, r) for believing p and believes p (partially) on the basis of those reasons.⁴ The second counter-example -the case of Messrs. Ponens and Lacy- purports to show that one fails to justifiedly believe p despite having good reasons r for p and believing p solely on the basis of r, when one comes to believe p as a result of deploying a patently invalid inference rule while reasoning from r to p.⁵ Turri concludes that the orthodox view fails to account for the fact that “the way in which the subject performs, the manner in which she makes use of her reasons, fundamentally determines whether her belief is doxastically justified” (p. 318).

Turri maintains that what makes the orthodox view susceptible to such counter-examples is that it tries “to understand doxastic justification in terms of propositional justification” (p. 319). Given the alleged downfall of the orthodoxy, Turri moves on to providing a sketch of an “alternative proposal” (p. 319), which reverses the direction of explanation and “explains propositional justification in terms of doxastic justification” (p. 325). According to Turri’s proposal, “the subject’s intellectual abilities explain why she is in a position to justifiedly believe p” (p. 320, emphasis original), where the relevant intellectual abilities in question involve explicit, conscious reasoning, and where the subject’s being in a position to justifiedly believe p (or her being in a position to have a doxastically justified belief that p) is taken as equivalent to p’s being propositionally justified for that subject.⁶ More specifically, then, Turri endorses both of the following theses:

(PJ) Necessarily, for all S, p and t, if p is propositionally justified for S at t, then p is propositionally justified for S at t because S possesses at t some intellectual abilities such that, were S to believe p by performing those abilities, S’s belief would thereby be doxastically justified.⁷

(R) Propositional justification should be explained in terms of doxastic justification (rather than vice versa).

Turri holds that (PJ) “marks a significant advance in our understanding” (p. 325) of propositional justification and that it is a way of fleshing out the more basic idea captured by (R).

This paper is hereafter divided into four sections. In section 2, I will argue that Turri makes things too easy on himself by defining “the orthodox view” in the way he does: there are good reasons to think that the view that properly deserves the title “orthodox” cannot be what Turri takes it to be. Section 3 is devoted to showing that there are two plausible responses available to the so-called orthodox view to Turri’s counter-examples and, therefore, that even if we take it for granted that Turri’s conception of the orthodox view is along the right lines, his counter- examples are far from being decisive. Section 4 challenges a central assumption in Turri’s argument against the orthodox view, namely, that (PJ) is an alternative proposal to (Basis): I will argue that the two can be simultaneously and consistently held and, accordingly, that the orthodox view can stand tall against Turri’s crusade by accommodating his main proposal. Section 5 sums up the lesson.

2 The Orthodox View Properly So Called

Let us start by granting Turri’s plausible point that an adequate account of doxastic justification should not neglect the manner in which a given subject makes use of her reasons. So, since the manner in which a given subject makes use of her reasons presumably pertains to the manner in which she bases her beliefs on her reasons, an adequate account of the basing relation should not neglect the manner in which she makes use of them. However, let us also note that this point creates a problem for the orthodox view only on the supposition that the orthodox view holds (or entails or presupposes) an obviously implausible conception of the basing relation, according to which the subject bases her belief p on R if she comes to believe p through a reasoning process in which R features somehow (in some manner or other) -but there is no particular manner or way in which R should feature in the reasoning. I doubt that there has been any philosopher that endorses such a crude view as that basing p on R consists simply in taking into consideration R in some way or other in the process of coming to believe p, let alone that such a view deserves to be called as “the orthodox view”. Turri correctly observes that the orthodox view is committed to (Basis), but this is not to say that it is therefore (or, simply, also) committed to that particular implausible conception of the basing relation. To use one of Turri’s own analogies (p. 315), just as we can safely assume that the idea that one can have a well built deck by having the right equipment and using it in some way or other to perform the job has never been the orthodoxy in carpentry, it also seems that we can safely assume that the idea that one can justifiedly believe p in virtue of having good reasons for p and utilizing those reasons in some way or other in the process of coming to believe p has never been the orthodoxy in epistemology.

The point here can be reinforced by making a distinction between two kinds of basing: psychological and epistemic. A given subject psychologically bases her belief that p on R just in case R is a reason for which she comes to believe p. Psychological basing is not cognitively demanding because it only requires the subject to utilize a reason in some way or other in order for her to base her belief on that reason -not that she utilize it in any specific way. A given subject epistemically bases her belief that p on R, on the other hand, just in case R is a reason for which she comes to believe p and she utilizes R in an epistemically adequate way.⁸ It is clear that epistemic basing is more demanding than mere psychological basing: a given subject might psychologically base her belief that p on R without epistemically basing it on that reason but not vice versa. The question is now whether we should take the notion of basing in (Basis) as psychological or epistemic. It is worth noting that (Basis) does not specify the specific conception of the basing relation the orthodox view might be committed to, which means that it cannot be rejected out of hand that psychological basing captures the orthodox view’s conception of that relation. However, the point is that doxastic justification is an epistemic notion and it would be very odd if the orthodoxy in epistemology about doxastic justification were to hold that what distinguishes doxastic from propositional justification is a sort of non-epistemic basing (such as psychological basing). The overt implausibility of that idea should give us a pause in our assessment of its attribution to a view that properly deserves the title “orthodox,” and (Basis)’s mere silence about the specific sort of the basing relation the orthodox view endorses does not by itself justify attributing it to the orthodox view characterized by its commitment to (Basis). (Basis)’s silence in question is a “glaring weakness”, as Turri calls it, only if it signals a commitment to a notion of non-epistemic basing like psychological basing, but there is no good reason to think that it signals that.

It is instructive here to take a look at ^{Firth (1978)}, where the distinction between propositional and doxastic justification is first clearly introduced and where, given the widely acknowledged significance of the work,⁹ we should search for what the orthodoxy properly so called might look like. There is a certain passage in Firth’s seminal paper that deserves close attention:

S cannot be warranted in believing p unless S arrives at his belief in a way that corresponds, in an appropriate way, to the evidential relationships in virtue of which p is warranted for S (p. 220).

Here Firth claims, as the talk of appropriate correspondence suggests, that there is what one might call a “structural similarity” constraint on doxastic justification (on S’s being justified in believing p), a constraint that originates from the way in which propositional justification is structured (the evidential relationships in virtue of which p is justified for S). The way in which S forms the belief that p must be structurally similar to the way in which p is propositionally justified for S. It is a fair assumption that the orthodox view properly so-called is at least committed to a version of the structural similarity (or “appropriate correspondence”) constraint (in any case, I will make this assumption below).

Now there are at least two ways in which one might conceive the structural similarity constraint -let me call them the weak conception and the strong conception. The weak conception of the structural similarity constraint holds that it is necessary and sufficient for a subject to have doxastic justification that the way she “arrives at” the target belief follows a path that is structurally similar to the one ‘laid out’ by propositional justification. Suppose that S is doxastically justified in believing that p, and suppose that p is propositionally justified for S partly because there is a proposition, R₂, available to S that supports R₁ which in turn supports p. Then the structural similarity constraint conceived along the lines of the weak conception entails that it follows from these facts about S that S arrives at the belief that p on the basis of an inferential path from R to R to p¹⁰. The crucial point is that psychological basing does not require the fulfillment of the structural similarity constraint conceived along the lines of the weak conception. Suppose that the subject arrives at the belief that p through an inferential path from R₁ to R₂ to p, where R₂ supports R₁ but not vice versa. In this case, the structural similarity constraint is not satisfied but the subject psychologically bases her belief that p on R₁ and R₂. So, if we understand the structural similarity constraint along the lines of the weak conception, then psychological basing cannot be the notion of basing that is relevant to the orthodoxy’s understanding of (Basis).

The strong conception of the structural similarity constraint holds that it is necessary and sufficient for a subject to have doxastic justification that the way she arrives at the belief that p follow the inferential path that licenses the move from R₂ to p, assuming that R₂ supports p. Suppose that R₂ is a conjunction and the inference rules that license the move from R₂ to p are conjunction elimination and modus ponens, but suppose also the subject arrives at p through deriving p from R₂ by a (blatantly invalid) inference rule according to which any proposition can be derived from any other proposition. In this case, the weak similarity constraint is, but the strong similarity constraint is not, satisfied; and therefore, we can say that S weakly bases but does not strongly base her belief that p on R₂. Given that strong basing is more demanding than weak basing, the conclusion to be drawn is that if we understand the structural similarity constraint along the lines of the strong conception, then psychological basing cannot be the notion of basing that is relevant to the orthodoxy’s understanding of (Basis).

The point I want to make is that the orthodox view in epistemology concerning the relation between propositional and doxastic justification should be conceived as committed to a version of the structural similarity constraint and, given that neither the weak nor the strong conception of that constraint neglects the manner in which the subject reaches the target belief in order for her to be justified, the orthodox view cannot be accused of neglecting it either.

I would like to close this section by clarifying its main point in order to avoid some potential misunderstandings. Turri claims that the orthodox view properly so called about the relation between propositional and doxastic justification is committed to (Basis). For all I have said in this section, Turri’s claim might well be true (and I believe it is true). My point is, however, that it is one thing to hold (Basis) and it is another thing to hold both (Basis) and a non-epistemic (or merely psychological) conception of the basing relation, according to which basing p on R is simply making use of R in some way or other in the process of coming to hold the belief that p. Turri attributes both (Basis) and that conception of the basing relation to the orthodox view properly so called, and I have argued that there are good grounds for thinking that this is a mistake.

3 Two Responses

We can move forward, however, by granting, for the sake of the argument, that what Turri calls “the orthodox view” deserves its title thanks to the number and influence of its adherents, and also that Turri’s objection that that view neglects the manner in which the subject makes use of her reasons does not miss the mark. Now, there are two main sorts of response available to the orthodox view thus understood on account of the fact that the antecedent of (Basis) is a conjunction of two statements, namely (i) and (ii): revise either the conditions for having propositional justification or the conditions for satisfying the basing relation (in such a way that Turri’s counter-examples are neutralized without dropping the thesis that propositional justification is explanatorily prior to doxastic justification). The basic idea is that if we had a version of the orthodox view that entails that the conditions for having propositional justification or satisfying the basing relation are harder to meet than what the original version suggests, then we could reasonably expect that Turri’s counter-examples lose their force: if we had such a version endowed with stronger notions of propositional justification or the basing relation, we would be in a position to say that the subjects that figure in those counter-examples either lack propositional justification or do not satisfy the basing relation.

One way to respond to Turri’s challenge is by working on (ii), i.e. by offering a stronger notion of the basing relation. On the orthodox view, as Turri defines it, the subject bases her belief p on R if she deploys R in some way or other in the process of coming to believe p. Now, consider the following, more demanding notion of basing -proper basing: the subject properly bases her belief p on R if she deploys R in a proper (or adequate/acceptable/legitimate/reliable) way in the process of coming to believe p.¹¹ So, for instance, an explicit process of reasoning from R to p that follows an improper (e.g., invalid) inference rule counts as basing the belief that p on R but not as properly basing that belief on R. Now, the orthodox view can offer a straightforward revision to its initial, crude picture and claim that propositional justification plus the proper basing relation amounts to doxastic justification. The move from basing to proper basing is clearly not ad hoc, given that a main reason why we feel the need to introduce the notion of basing to our philosophical theory of knowledge in the first place is our conviction that in order for a subject to know p, she needs not only to have good reasons R for believing p and believe p but also to form her belief that p by being responsive to R in the right way.¹² If mere basing does not do the job, then it is natural to offer a stronger notion of basing, i.e. proper basing. Furthermore, a revision along these lines has the obvious potential to undermine Turri’s counter-examples, especially the seemingly more challenging case of Messrs. Ponens and Lacy. Let me call this sort of response to Turri’s challenge the proper basing response.¹³

The proper basing response is, as the discussion in the previous section suggests, what can be straightforwardly gleaned from the standard (or “orthodox”) accounts of the relation between propositional and doxastic justification. Interestingly, Turri does not address the proper basing response to his challenge in any considerable detail. The only relevant comment he makes is this: “I should note that Kvanvig [one of the many philosophers Turri presents as a proponent of the orthodoxy] at times maintains that the belief must be “properly” based in order to count as doxastically justified. But he never explains what “properly” amounts to, and often omits it” (p. 313, fn. 1). However, it is clear that Kvanvig’s alleged failure to explain the notion of proper basing can only function simply as an invitation for further clarification but cannot be treated as dealing a decisive blow to the orthodox view.

In addition to the proper basing response, which works on (ii) (in [Basis]) by offering a stronger notion of the basing relation, there is another response that works on (i) by offering a stronger notion of propositional justification. The orthodox view about propositional justification might be conceived as holding, roughly, that if S has a reason R that supports p, then p is justified for S. So, S’s intellectual abilities such as her reasoning capacity from R to p (or her capacity to make the inferential connection between the two) have nothing to do with whether p is justified for S.¹⁴ This is a perfectly legitimate notion of propositional justification, but it is a weak one and there is room for a stronger one. Note that the weak notion of propositional justification does not accord well with the more fundamental idea that having propositional justification for p puts one in a position to have a doxastically justified belief that p. And, this is because there might well be a subject that has a reason R that supports p but does not have the capacity to reason from R to p. If having propositional justification for p puts one in a position to have a doxastically justified belief that p, then since such a subject lacking the relevant inferential capacity is not in a position to have a doxastically justified belief that p, she does not have propositional justification for p despite the fact that she has R.¹⁵ This suggests the following, strong notion of propositional justification: having propositional justification for p requires not only having good reasons R for p but also the capacity to make the inferential connection between R and p.

Let us call the response to Turri’s challenge that adopts the strong notion of propositional justification the inferential capacity response. The inferential capacity response to Turri’s counter-example illustrating the possibility of deploying a patently invalid inference rule is that it is not described in enough detail to enable us to say whether the subject deploying such an inference is propositionally justified. In that counter-example, the subject has good reasons R to believe p but the inference she makes from R (and only from R) to p is patently invalid. Turri assumes that since she has R to believe p, the subject is propositionally justified to believe p. However, the inferential capacity response rejects that assumption and protests that Turri’s counter-example is under-described because we are not provided with information regarding whether the subject has the capacity to make use of a correct inference rule from R to p.

The proper basing response and the inferential capacity response agree that (Basis) needs to be revised to meet Turri’s counter-examples to the orthodox view. The difference between the two is that the former opts for revising (ii) and the latter opts for revising the requirements for having propositional justification presupposed by (i).¹⁶ So, while the proper basing response delivers the result that in Turri’s counter-examples, the subjects in question do not properly base their beliefs on their reasons, the inferential capacity response delivers the (potential) result that they are not propositionally justified to hold those beliefs. Though these responses stand in need of further refinement and elaboration, it seems to me that they still manage, in their current shape, to tilt the balance in favor of the orthodox view in the face of Turri’s counter-examples: these responses neutralize Turri’s counter-examples (in their current shape, at least) and thereby stave off his attack on the orthodox view.

4 The Orthodox View and Turri’s Proposal

According to Turri, the correct response to his counter-examples is, as noted above, to acknowledge that a given subject’s inferential capacities (or “intellectual abilities”) explain why a given proposition is justified for her. This is supposed to be captured by the ‘because’ clause in (PJ). Turri assumes in his paper that the orthodox view cannot acknowledge this point and is committed to rejecting (PJ). I think this is a mistake. There is nothing in (Basis) that requires a rejection of (PJ), which means that, contra Turri, (Basis) and (PJ) are not alternative theses about the relation between propositional and doxastic justification.

To see why this is so, it is instructive (and plausible anyway) to think of inferential capacities as dispositions to form new beliefs given some other beliefs and to think of exercising those inferential capacities in the process of forming new beliefs as manifestations.¹⁷ Now, just as a glass has a disposition to shatter when struck and what explains the actual shattering of a glass in certain circumstances is its disposition in question (i.e. fragility),¹⁸ a subject might have a disposition to form new beliefs given some other beliefs and what explains her actually coming to form those new beliefs in certain circumstances is her disposition in question (e.g., her mastery of modus ponens). A plausible analogy is, then, that manifestation of intellectual capacities is to actual shattering as having those capacities is to fragility. If so, (PJ) might be plausibly read as (PJ*):

(PJ) Necessarily, for all S, p and t, if p is propositionally justified for S at t, then p is propositionally justified for S at t because S possesses at t some intellectual abilities such that, were S to believe p by performing those abilities, S’s belief would thereby be doxastically justified.

(PJ*) Necessarily, for all S, p and t, if p is propositionally justified for S at t, then p is propositionally justified for S at t because S has at t a particular disposition such that, were S to believe p, S’s belief would thereby be doxastically justified.

Now, let us suppose, for the sake of analogy, that the following thesis captures “the orthodox view” about the relation, say, between fragility (a specific disposition) and shattering (its manifestation):

(Fragility) If (i) x is fragile, and (ii) x strikes (a suitably solid surface, say, the ground), then x shatters.

(Fragility) is crude and needs massive refinement. But that is not our concern in this paper. The point is that (Fragility) is an analogue of (Basis) in the debate over the nature of the relation between fragility and shattering, and the question is whether an adherent of (Fragility) as such is committed to rejecting a thesis along the following lines, which is purported to be an analogue of (PJ*):

(DM) Necessarily, for all x, and t, if x is fragile at t, then x is fragile at t because x has at t a particular disposition such that, were x struck, x would thereby shatter.

I think it is obvious that (Fragility) and (DM) are not alternative theses: there is simply no reason why an adherent of (Fragility) cannot consistently endorse (DM). (DM) simply specifies the reason why something is fragile, i.e. why (i) in (Fragility) is true when x is replaced by an individual constant. But this by itself presents no threat to the adequacy or truth of (Fragility).

Now, just as (Fragility) and (DM) are not alternative theses, (Basis) and (PJ*) are not alternative theses either:

(Basis) If (i) p is propositionally justified for S in virtue of S’s having reason(s) R, and (ii) S believes p on the basis of R, then S’s belief that p is doxastically justified.

(PJ*) Necessarily, for all S, p and t, if p is propositionally justified for S at t, then p is propositionally justified for S at t because S has at t a particular disposition such that, were S to believe p, S’s belief would thereby be doxastically justified.

(PJ*) specifies the reason why a given proposition is propositionally justified, i.e. why (i) in (Basis) is true when p is replaced by a particular proposition. But this by itself presents no threat to the adequacy or truth of (Basis).

Moving on to Turri’s (R) and its bearing on the orthodox view, I want to make two points (recall that [R] is the thesis that propositional justification should be explained in terms of doxastic justification (rather than vice versa)). First, (R) is purported to capture the main idea expressed by (PJ) (or (PJ*)); and if, as I have argued above, (PJ) (or (PJ*)) is consistent with (Basis), then (R) in the sense intended by Turri must also be consistent with (Basis). Second, it is not clear what exactly (R) amounts to. However, given that (Basis) is consistent with (PJ) (or (PJ*)) and therefore with (R), then, assuming that Turri’s claim regarding the orthodox direction of explanation is true, the sense in which that claim is true cannot be the sense in which it is inconsistent with (R).

5 Conclusion

To sum up the entire discussion above, let me present the main lesson in the form of a dilemma: either the orthodox view about the relationship between propositional and doxastic justification is not what Turri takes it to be, or it is. If the former is true, then Turri’s objection misfires. If the latter is true, then Turri’s preferred proposal is not an alternative to the orthodox view and can be accommodated by it. In either case, the orthodox view remains unscathed.