Propositional justification pertains to propositions: it is the sort of justification that a proposition enjoys for an agent when the agent is epistemically justified to believe it. By contrast, doxastic justification is justification of beliefs, i.e., of doxastic states actually instantiated by an agent. The ‘orthodox’ view of the relationship between propositional and doxastic justification is that the latter should be explained in terms of the former, so that an agent’s belief is (doxastically) justified just in case (i) it is a belief in a proposition that is (propositionally) justified for the agent and (ii) it is held on the basis of that which (propositionally) justifies its content. This view has been challenged by John Turri in his paper ‘On the Relationship between Propositional and Doxastic Justification’ (Philosophy and Phenomenological Research, 80, 2010, pp. 312–326). There he offers two putative counterexamples to the orthodoxy and goes on to argue that the order of explanation should be reversed: it is propositional justification that should be explained in terms of doxastic justification, and not vice versa. Though I share Turri’s feeling that there is something amiss with the way a number of contemporary epistemologists talk of propositional and doxastic justification, I do not believe he has managed to put his finger on the real trouble. So the first point I make is that his counterexamples fall short of refuting the orthodox view, which I argue should be maintained. Then I try to diagnose the real source of the trouble. It lies, I suggest, in the way a number of recent epistemologists talk of propositional justification, a way which commits them to the questionable view that the basis upon which facts concerning propositional justification supervene does not include facts concerning the doxastic states of agents, but only facts concerning the evidence to which agents have access. So I observe that there are different degrees of idealisation involved in judgments of ‘propositional justification’, propose to distinguish what an agent is propositionally justified to believe given his overall doxastic state from what an agent is propositionally justified to believe irrespective of his overall doxastic state, and argue that, whenever it is the former relation that is at stake, an agent can be propositionally justified to believe a propositionp at timet only if it is reasonably easy for him to form a doxastically justified belief inp att. It is most likely an awareness of this fact − or of some fact in the vicinity − that encourages the belief that propositional justification should be explained in terms of doxastic justification. But I suggest that this fact, far from being evidence that the notion of doxastic justification is in any sense more fundamental than that of propositional justification, can be accounted for by paying attention to the relationship that links the relevant sort of (epistemic) justification to (epistemic) responsibility. So I conclude that there is an important sense in which facts concerning epistemic justification supervene not merely on facts concerning the evidential states of agents, but on facts concerning their overall doxastic states.

1. 1.

   The terminology is due to Firth (1978, 217 f.). Some writers, including Feldman and Conee (1985, 24), Feldman (2002, 46) and Pryor (2004, 365), talk ofwell-founded rather than ofdoxastically justified beliefs. Here, however, I shall follow the most common usage and stick to the latter phrase.

2. 2.

   Depending on circumstances, an agent may instead be justified todisbelieve, or towithhold belief from, a proposition. For simplicity, in what follows I will not discuss the justificational status of such attitudes, nor will I attend to the complications raised by considering degrees of belief, or different kinds of propositional attitudes, such as trust or acceptance. These omissions will not affect the thrust of the argument.

3. 3.

   Turri keeps switching between formulations in which ‘p’ appears to function as an objectual variable (as in ‘S believesp’) and formulations in which it occurs as a sentential variable, or perhaps a sentence letter (as in ‘S’s belief thatp’). This oscillation doesn’t cause any real harm, but in this paper I consistently adopt the former usage.

4. 4.

   As usual, substitution-instances of ‘<p>’ are used to abbreviate corresponding substitution-instances of ‘the proposition thatp’.

5. 5.

   As I said in Sect.2, it is natural to regard a reason that an agent can have and use as a basis for belief as something accessible from, or internal to, the agent’s perspective on the world – which naturally leads to Turri’s assumption that having reason(s) amounts toknowing (orjustifiedly believing) certain facts or propositions. Here, however, nothing important depends on accepting an internalist construal of justifiers, and the argument of this paragraph could be easily reformulated in terms more acceptable to externalistically-minded philosophers.

6. 6.

   Eugenio Orlandelli has tried to convince me, both in conversation and in correspondence, that Turri’s cases appear to pose a challenge to the orthodox view only because they are set on the background of a misleading account of the nature of that in virtue of which a proposition may be inferentially justified for an agent. Orlandelli’s alternative account may be sketched by applying it to the basketball case. His view is, in a nutshell, that the proposition that the Spurs will win, far from being (propositionally) justified for Mr. Ponens and Mr. F.A. Lacy in virtue of their knowing (or justifiedly believing) P5 and P6, is justified for them in virtue of their being (propositionally) justified to believethat it follows via modus ponensfrom P5 and P6. Orlandelli’s diagnosis is then that Lacy’s belief that the Spurs will win fails to be doxastically justified precisely because it is not based onthat justification. For reasons that will become apparent in due course, I doubt that this account of inferential justification is correct. In any case, it seems able to make sense of our intuitions concerning Turri’s basketball case only on the assumption that Mr. F.A. Lacy neitherknows nor has a doxastically justifiedbelief to the effect that the proposition that the Spurs will win follows viamodus ponens from P5 and P6 – if he did, why would he employmodus profusus to draw the conclusion? This seems somewhat arbitrary, especially if one considers that in the more usual framework adopted by Turri it is possible to make the rather plausible claim that Mr. F.A. Lacy knows (or justifiedly believes) the reasons in virtue of which the proposition that the Spurs will win is justified for him.

7. 7.

   A related but not identical distinction between the ‘vertical’ relations that obtain between mental states and the world and the ‘horizontal’ relations that obtain among mental states is made by Zangwill (2005, 4).

8. 8.

   For references to the debate, and a defence of nonreductionism, see Kolodny (2005, 510 f.).

9. 9.

   As I said in Sect.2, for the purposes of this paper it is not necessary to assume that the things that lend justification to propositions have propositional content, can be given or cited in defence of a claim, or are accessible from, or internal to, a specific agent’s perspective on the world. What I think one should grant to Turri is then, in more neutral terms, that (propositional) justification is the epistemic status that a proposition may enjoy for an agent at a time in virtue of there being one or more justification-makers (justifiers) for the agent to believe it.

10. 10.

    This idea was suggested to me by some related considerations made by Lalumera (2013). For the relevant notion of epistemic competence, see Sosa (2007, 22 ff.).

11. 11.

    For an overview of the literature on the transmission of justification and the ways in which it can fail, see Moretti and Piazza (2013). Miss Improper and Mr. F.A. Lacy adopt ways of belief-formation that, employed on a regular basis, would almost inevitably lead them to believe logically inconsistent propositions. So one might also suggest that the former’s belief about Mansour and the latter’s belief about the Spurs are irrational because they are formed by methods of belief-formation which are essentially flawed.

12. 12.

    To many speakers, saying that S is justified to believep att sounds significantly different from saying that S is justifiedin believing p att. Bach (1985, 251) maintains that statements to the effect that someone is justifiedin believing a proposition at a time express claims ofpersonal, as opposed to doxastic, justification, while Kvanvig and Menzel (1990, 240–247) argue that claims of personal justification are in fact logically equivalent to claims of doxastic justification. Be that as it may, it seems clear that the truth of statements to the effect that someone is justifiedto believe, unlike (possibly) that of statements to the effect that someone is justifiedin believing, a proposition at a time does not depend on the agent’s actually believing the proposition at the relevant time.

13. 13.

    As I said, for simplicity I ignore the complications raised by the possibility of considering degrees of belief or different kinds of doxastic and propositional attitudes, as they are immaterial to the problem under consideration.

14. 14.

    Here is a brief description of the two cases.Ron’s case. Ron knows both that invading Iran would be catastrophically stupid, and that if invading Iran would be catastrophically stupid, than the U.S. ought not to invade Iran. But although perfectly familiar withmodus ponens, he will not exploit such pattern of reasoning to form the belief that the U.S. ought not to invade Iran because massive exposition to right-wing propaganda has made him incapable of believing that the U.S. ought not to invadeany country, much less Iran. The intuition elicited by this case is meant to be that the proposition that the U.S. ought not to invade Iran is (propositionally) justified for Ron even if he is currently incapable of forming a (doxastically) justified belief in it.Cedric’s case. Cedric is the most brilliant logician in the world: for any given proposition and any given set of axioms, he is able to discover with relative ease whether the former is a theorem of the latter by applying some clever algorithm he devised. The intuition elicited by this case is meant to be that the true proposition that (say) T is a theorem of A is (propositionally) justified for Cedric even if most competent human reasoners are obviously incapable of forming a (doxastically) justified belief in it. See Turri (2010b, 322–324).

15. 15.

    If you do not believe that it would, just assume it for the sake of the example.

16. 16.

    A similar idea is found in work by Crispin Wright, who takes a proposition to be ‘rationallyavailable’ to an agent when it is ‘consistent with what they believe’ (Wright2004, 181). See also Coliva (2015, 22).

17. 17.

    Pryor’s idea is, essentially, that an agent is prevented from forming arational belief inp on certain grounds whenever he holds a belief inq, such that justification forq would undermine the justification those grounds give him forp (Pryor2004, 364 f.,2012, 285 f.). Pryor applies this idea to Moore’s proof of an external world, whose failure is in his view merely dialectical, a consequence of the fact that whoever rejects its premises or even doubts whether they are true cannot rationally believe its – propositionally justified – conclusion. See Pryor (2004, 368–370,2012, Section 5).

18. 18.

    See the works cited in footnote 16 above.

19. 19.

    Although there is a sense in which one isrationally committed to believe every implication of the propositions one believes (Volpe2012, 323 f.), it is not the case that one isjustified to believe every proposition that follows from the propositions that one justifiedly believes. This said, among the consequences of justified propositions that an unimpeded but not logically omniscient reasoner will inevitably fail to believe there will likely be some, even many,justified propositions – propositions he might easily have recognized to follow from (or, perhaps, to be made overwhelmingly likely by) propositions he is justified to believe. So an unimpeded reasoner will not be, in general, a reasoner who believes every proposition he is justified to believe (and who disbelieves every contrary of the propositions he is justified to believe).

20. 20.

    This is a technical notion: a doxastic state that provides a reason to believe a proposition merely in virtue of the fact that the agent that is in that state knows or justifiedly believes that he is will count as evidentially idle in this sense. Needless to say, whether a doxastic state is evidentially idle for an agent at a time will depend in part on the agent’s level of logical and probabilistic competence.

21. 21.

    The term ‘evidential situation’ is used to refer to the whole set of reasons possessed by an agent at a time, including the agent’s evidentially non-idle beliefs.

22. 22.

    What I have in mind is not the trivial claim that a proposition that is justified for an agent at a time may fail to be justified for another agent at the same time, or for the same agent at a different time (or that a proposition that is not justified for an agent at a time may be justified for another agent at the same time, or for the same agent at a different time). Nor is it just the claim that the doxastic states of an agent affect what the agent is justified to believe in the sense that a proposition that is justified for an agent at a time might fail to be so justified (or that a proposition that is not justified for an agent at a time might be so justified) if the agent’s overall doxastic profile at that time were relevantly different. It is, rather, the claim that the doxastic states of an agent affect what the agent is justified to believe in the sense that, had his overall doxastic profile been sufficiently different from the overall doxastic profile that he is in at some given time, an agent might have been justified to believe propositions that areinconsistent with the propositions he is actually justified to believe. Nothing of what I say in this paper commits me to subscribe to this form of justification-relativism.

23. 23.

    A third notion might be considered: that of what an epistemic agent is (propositionally) justified to believeirrespective of his overall doxastic profile. But this notion would be of any interest only if Unrestricted Doxastic Independence were true.

24. 24.

    One might venture the claim that, in a natural and important sense, an agent is epistemically justified to believe a proposition at a timejust in case it is epistemically responsible for him to believe that proposition at that time. But for present purposes it is not necessary to defend this claim.

25. 25.

    The considerations sketched in this paragraph are reminiscent of a well-known line of argument employed in support of internalist conceptions of epistemic justification (see, e.g., Ginet1975; BonJour1985; Goldman1999; Foley2005; for criticism, Greco2005, 260–262). Here, however, I am not concerned with presenting an argument in favour of such conceptions, but an explanation of whyEJ holds – when it holds.

26. 26.

    To avoid confusion, Alvin Plantinga has proposed to refer to that which has to be added to true belief to turn it into knowledge as ‘warrant’ (Plantinga1993). But his proposal has not gained wide currency.

27. 27.

    In Wright’s picture of the architecture of perceptual justification, the doxastic justification of many of our beliefs depends, as we have seen, on the existence of a type of propositional justification that is inaccessible to most of those who rely on it. Thus, if Wright’s picture were correct (which I doubt), the notion of what an epistemic agent is propositionally justified to believe irrespective of the evidentially idle features of his overall doxastic profile could well play a role in the analysis of knowledge. However, it would be at most an indirect role, for the general presuppositions of thought about the external world that Wright argues we are entitled to accept would never become, in his view, the content of beliefs that might count as instances of knowledge.

Earlier versions of this paper were presented at a philosophy of language conference held in Gargnano (Brescia) in June 2013 and at Cogito’s Epistemology seminar (Bologna) in January 2014. Thanks to all present for discussion. I am especially grateful to Andrea Borghini, Delia Belleri, Annalisa Coliva, Elisabetta Lalumera, Sebastiano Moruzzi, Ernesto Napoli, Eugenio Orlandelli, Elisa Paganini, Michele Palmira, Marco Santambrogio, Giuseppe Spolaore and Luca Zanetti for many helpful comments and suggestions.

Authors and Affiliations

1. University of Bologna, Bologna, Italy

   Giorgio Volpe

Corresponding author

Correspondence toGiorgio Volpe.

Editors and Affiliations

1. Faculty of Law and Administration, Jagiellonian University, Kraków, Poland

   Bartosz Brożek

2. CIRSFID - Faculty of Law, University of Bologna, Bologna, Italy

   Antonino Rotolo

3. Faculty of Law and Administration, Jagiellonian University, Kraków, Poland

   Jerzy Stelmach

Reprints and permissions

© 2017 Springer International Publishing AG

Policies and ethics