1. Transmission of justification

I carefully count that there are six chairs in the room, and I know that there being exactly six chairs in the room entails there being fewer than seven chairs in the room. It seems that the belief that there are fewer than seven chairs in the room becomes justified for me in virtue of my counting and acknowledging that necessarily, if there are exactly six chairs in the room, then there are fewer than seven chairs in the room. Thanks to my counting and my knowledge of the entailment from the “exactly-six” belief to the “fewer-than-seven” belief, I thereby acquire justification for the latter belief. When a belief that q gains its justification from the justification that one has for another belief that p in this way, the justification for p transmits to q across the known entailment from p to q. Our epistemic practices, daily and scientific, routinely take it for granted that justification transmits across known entailment, and without that presumption, the sphere of our justified beliefs would be severely limited. Indeed, the very point of reasoning seems to be the transmission of justification, and if it were not for transmitting justification, reasoning would lose much of its attraction: if reasoning does not transmit justification, then why bother reasoning?Footnote ¹

Despite its central significance for our overall epistemic economy, it is widely admitted that transmission of justification across known entailment is not exceptionless. An obvious way in which justification fails to transmit is when the reasoning involves explicit premise circularity, i.e., when the conclusion of the reasoning is one of the premises of that reasoning. Consider the deductively valid argument p therefore p, the most straightforwardly premise-circular argument there is. And suppose that S is justified to believe p, and S knows that p entails p. It seems clear that S cannot get any justification that she does not already have in virtue of the fact that S has justification for p and her recognition of the validity of p therefore p. There is no justification added to S’s justification for p resulting from her recognition of the inferential link from p to p: one cannot increase the justification one has for a proposition in virtue of her recognition that it is self-entailing. And there is no transmission of justification between two propositions if the proposition at the “receiving” end of the process does not gain any additional justification for the subject in virtue of her recognition of that link (even if the subject has justification for the proposition at the “giving” end). So, premise-circular arguments cannot transmit justification.

Justification transmits from one proposition to another only if the latter proposition acquires some justification that it does not already have as a result of that transmission; so, it is entirely predictable and thus uninteresting that premise circularity prevents transmission of justification. Are there interesting cases of transmission failure?Footnote ² Are there, for instance, cases in which transmission fails despite the fact that the argument from p to q is not premise-circular, the subject has justification for p and recognizes that p entails q?

Crispin Wright has famously argued in a series of papers extending over a long period of time that the answer to these questions is “yes.”Footnote ³ There are mainly two different diagnostics that one can identify in Wright’s writings for a deductive argument’s inability to transmit the justification for its premises to its conclusion. One stems from worries regarding epistemic circularity. As we shall see in Section 2, this approach has met considerable resistance, and rightly so in my opinion. Another diagnostic appeals to the idea of a direct justifier. In Section 3, I will present Wright’s “counterfactual” account of direct justification and argue that it is defective. I will also extract in this section a significant idea about transmission from Wright’s directness considerations, which I dub OBJECT OF TRANSMISSION. In Section 4, I will take a closer look at OBJECT OF TRANSMISSION and argue that despite the fact that it has either received little attention or been set aside as inadequate rather prematurely, OBJECT OF TRANSMISSION provides the fundamental tenet for an adequate understanding of the transmission of justification. The main claim I wish to defend is that transmission of justification should be conceived as transmission of a justifier, and the search for cases in which transmission of justification fails should focus its energy on cases in which justifiers fail to be transmitted.

2. Epistemic circularity

According to Wright, there are deductively valid arguments that exhibit subtler kinds of objectionable epistemic circularity other than premise circularity that prevent them from transmitting the justification of their premises to their conclusions. The “key thought” on which his discussion of transmission failure is based is that warrant fails to transmit in those cases in which “only someone who took themselves to be rationally persuaded of the conclusion could rationally receive whatever grounds purportedly warranted its premises as doing just that” (Wright Reference Wright2002: 331). Wright’s “information-dependence template” is designed to substantiate this very thought. A piece of evidence, e, provides information-dependent justification to S for p if whether e justifies p for S depends on whether S has independent justification for a different proposition q. And, if e is information-dependent in this way, Wright argues, the justification it provides to p is not transmissible to q, even if (the subject knows that) p entails q. Wright asks us to consider the following argument:

In Soccer, S1 is (defeasible) evidence for S2, and S2 entails S3 (and we assume that the subject knows that S2 entails S3). However, Wright holds that S1 can be evidence for S2 only on the condition that the subject has independent reason for thinking that S3 is true. This point can be supported, Wright holds, by the consideration that if the subject were, for instance, in the vicinity of a film studio and knew that it is just as likely that what she witnesses is staged for the camera as that it is an event in a genuine game, S1 would provide “no warrant any longer for S2” (Reference Wright2002:335). What the subject needs, in order to restore its supporting power, is “some independent corroboration” (Reference Wright2002: 335) of the conclusion. Suppose that she gets that independent support from the testimony of a bystander. In that case, S1 becomes evidence for S2. However, it is exactly because S1 becomes evidence for S2 in virtue of the fact that the subject has independent evidence for S3, the subject does not gain any additional support for S3 in virtue of deriving it from S2. The rationality of the belief that S2 depends on the fact that a belief in S3 is independently rational for the subject, in which case Wright claims the former cannot add to the rationality of the latter via a deductive link.Footnote ⁴

Furthermore, Wright argues that Moore’s infamous “Proof of an External World” suffers from the same circularity problem that afflicts Soccer. Moore’s argument goes as follows:

Wright maintains that “M2 stands to M3 as, e.g., S2 stands to S3” (Reference Wright2002: 337). The idea is that the evidential force for the subject of M1 with respect to M2 is conditional on whether the subject has independent reason to think that M3 is true. This point can be supported, Wright holds, by the consideration that if the subject had evidence, for instance, that there is no material world whose characteristics are mostly, at least in large, disclosed in routine sense experience, then M1 would afford no justification for M2. And, it is exactly because M1 can become evidence for M2 in virtue of the fact that she has independent evidence for M3, the subject does not gain any additional support for M3 in virtue of deriving it from M2.

It is important to note that, according to Wright, whereas justification fails to be transmitted in Soccer and Moore, his diagnosis of these cases as exhibiting a subtle kind of epistemic circularity renders them to be compatible with the idea that justification is closed under known entailment. According to justification closure, a given subject can only be justified to believe that p, which she knows entails q, if she is justified to believe that q: having justification for the conclusion of an argument that she knows to be valid is necessary for having justification for the premise of that argument. In Soccer and Moore, however, according to Wright’s diagnosis, the subject can have justification for the premises only if she already has justification for the conclusion; so, they are automatically not cases in which the subject has justification for the premises without having justification for the conclusion – hence, justification closure is not violated. And that very same idea, which explains why justification closure is not violated in these cases, also explains why transmission fails in these cases: it is because some justification for M3, for instance, must already come from somewhere else in order for her to have justification for M2, so our subject cannot acquire justification for M3 in virtue of having justification for M2 and knowing that M2 entails M3.

There are two important questions regarding Wright’s thesis that it suffices for failure of justification transmission across a valid argument from p, justified by e, to q if the justification of p by e is conditional on the fact that one has independent evidence for q. One is whether Wright has successfully identified a sufficient condition for transmission failure. The other is whether, assuming that the answer to the former question is “yes,” Wright is right in claiming that philosophically interesting arguments such as Moore instantiates the structure articulated by the putatively sufficient condition. A negative answer to each question has its own formidable proponents. Starting with the latter question, Pryor (Reference Pryor2004) agrees with Wright that if independent evidence for the conclusion of an argument is required for the premises of that argument to be justified, then that argument is non-transmissive of justification. However, Pryor holds that the evidential (or justificatory) structure of Moore is not problematically (or “viciously”) circular in the way Wright envisages it to be since the (prima facie) support M1 provides to M2 is not conditional on the fact that M3 (or any other proposition) is independently justified, a thesis that Pryor famously calls dogmatism. On Pryor’s version of dogmatism, M1 provides justification for M2 insofar as the subject lacks justification to believe that M3 is false and not, contra Wright, insofar as she has (independent) justification to believe that M3 is true. There is, on this view, still a type of epistemic circularity Moore instantiates between M2 and M3: evidence against M3 would undermine the support M1 provides to M2. However, this kind of circularity is “epistemologically respectable” (Pryor Reference Pryor2004: 359) and thus does not prevent transmission of justification.

If there is really nothing to be ashamed of about the justificatory structure of Moore, as Pryor claims, then the dogmatist owes us an account of why it appears to be defective. Why does Moore “sound funny”? How are we to explain “our squeamishness” (Pryor Reference Pryor2004: 351) about it? A good part of the answer to these questions lies, according to Pryor, in the recognition that arguments are “dialectical creatures,” which might be offered to “different audiences” who “come to the table with different commitments and doubts” (369). The dialectical effectiveness of arguments is something that they have relative to audiences, while their justificatory legitimacy is not: a given argument might be dialectically effective for one particular audience but not for another, whereas the epistemic standing of its structure is, unsurprisingly, to be evaluated in terms of the features of its structure, which obtain independently of the commitments of the audience to whom the argument happens to be presented. Pryor’s idea is that Moore is dialectically ineffective specifically against the skeptic, who already has doubts, justified or not, about our knowledge of the external world. However, the fact that Moore is “persuasively crippled” (362) in that regard (that it is not capable of rationally persuading the skeptic) is compatible with the claim that it is “perfectly legitimate” (369) in terms of its justificatory structure. The failure of Moore is dialectical, not epistemic – or so Pryor argues.

Let me now move on to the former question: has Wright successfully identified a sufficient condition for transmission failure? Some philosophers argue that contrary to Wright, the fact that having justification to believe the conclusion of a given argument is among the conditions that make one have the justification one has for its premises need not generate transmission failure. Why should an inference fail to transmit justification if one’s justification for believing a premise does not include the justification for believing the conclusion, but the latter is merely “an enabling condition” (Silins Reference Silins2005: 77) for the former to obtain? It is of course true that if an independent justification for the conclusion is required for the premises to have justification at all, then the argument cannot provide one’s original, first-time justification for the conclusion. However, why think that it cannot provide additional (perhaps second-time or third-time) justification for the conclusion? In other words, why think that an argument transmits justification only if it can provide a first-time justification for its conclusion?Footnote ⁵

Somewhat ironically, in one of his own later discussions of the possibility that transmission might be apt for enhancing justification but not for creating first-time justification, Wright (Reference Wright and Coliva2012) presents a compelling case that substantiates the worry at hand.Footnote ⁶ Consider the following argument:

Suppose that the background against which we consider Flock is as follows:

Under these circumstances, Wright suggests, if F3 were not independently justified for me (by testimony), then I would not be justified in believing F2 on the basis of F1. Despite that, however, Wright acknowledges that transmission might succeed here: I might receive justification, not first-time but second-time, for F3 by inferring it from F2, whose justification on the basis of F1 is in turn conditional on the fact F3 is independently justified. However, if Wright is right that Flock is transmissive of justification, then the information-dependence template he offers is not a correct diagnostic for transmission failure.

It is fair to say that Wright’s case for the claim that instantiating the information-dependence template is responsible for the failure of transmission of justification in such arguments as Soccer and Moore has proved to be far from being decisive (arguably, by Wright’s own admission). So far as I can see, there are two main sources for resistance against Wright’s proposal under consideration, from which the replies that I have just presented above appear to stem. One is that an attempt like Wright’s to base a claim of transmission failure upon a charge of epistemic circularity is bound to meet resistance, given the extremely problematic and thus rather poorly understood nature of epistemic circularity. It is a well-advised procedure at this point to remind ourselves of Robinson’s contention that the charge of epistemic circularity is “nearly always a muddle, or improper, or both” (Reference Robinson1971: 117). If all epistemic circularity is bad (or “vicious” or “malignant” or “illegitimate”) in that arguments having that feature are structurally incapable of transmitting justification, then having identified a novel type of epistemic circularity instantiated by such arguments as Soccer and Moore, Wright is entitled to conclude that they are non-transmissive of justification. However, it is far from clear that all epistemic circularity is bad. Indeed, someone who wishes to defend that there are some benign (or legitimate) types of epistemic circularity will find herself in excellent company.Footnote ⁷ This is, of course, not to say that Wright’s strategy under consideration fails but only that the move from identifying a type of epistemic circularity to claiming that arguments instantiating that type of circularity cannot transmit justification is not straightforward. White’s (Reference White2006) following remarks succinctly capture the point I am driving at:

The difficulty here is that we currently lack an adequate account of how the distinction between malignant and benign epistemic circularity, assuming that there is such a distinction, is to be understood. In fact, Talbott (Reference Talbott2020) might well be correct in maintaining that “there is no adequate principle in the literature for distinguishing malignant from benign epistemic circularity” (2283). The upshot is that if the goal is to propose a (relatively uncontentious) sufficient condition for (and to come up with some clear cases of) failure of justification transmission, it is not an effective strategy, dialectically speaking, to search for it in the chaotic realm of epistemic circularity.

The second, related source of resistance to Wright’s current proposal stems from the observation that an investigation into the nature of epistemic circularity cannot proceed without an accompanying account of the structure of reasons (that is, an account of what supports what, of the role the background information plays in the obtaining of support relations among propositions, of whether support relations are mutual or unidirectional and bottom out at a level on which the entire structure of those relations rest, and so on). And Wright explicitly bases the information-dependence template upon a specific and controversial conception of this structure, which is “the holism of empirical confirmation famously emphasized by Quine in the last two sections of ‘Two Dogmas’” (Reference Wright2002: 335). This preference of Wright’s excludes, without much argument, alternative conceptions of the structure of reasons, leaving one wondering whether the phenomenon of transmission failure is less of a general epistemological worry and more of a problem that emerges only on some conceptions of that structure. It is undoubtedly an achievement to show that transmission failure occurs on a holistic conception of reasons, but a proponent of the “atomistic” conception (such as Pryor’s dogmatism and other foundationalist accounts like phenomenal conservatism Footnote ⁸ ) might simply remain unmoved by this observation (and perhaps even claim, “This is all the more reason to endorse atomism!”). So, if the goal is to treat transmission failure as a phenomenon that epistemological theories in general should give an account of, as I take (and think Wright takes) it to be, then it is not advisable to tie its proposed condition too closely to one particular theory among many. I conclude that it is preferable to search for a sufficiency condition for transmission failure somewhere else.

3. Directness

The second mechanism that Wright identifies as a source of transmission failure appeals to considerations of directness (or simultaneity) of evidential support. Suppose that in regard to Soccer, we already have the collateral information that scenes of the sort witnessed reliably indicate that real soccer matches (not, for instance, mere film takes) are going on. Against this background, Wright maintains S1 provides evidence for S3 as “directly” (Reference Wright2002: 334) as it does for S2 and thus “simultaneously” (335) with its doing so for S2. And if this is so, then it seems to follow that “it is not because S1 is a warrant for S2 that it is a warrant for S3” (335). And if this is so, then Wright concludes, “there is, arguably, a failure of transmission in this case” (335).

There are two significant ideas Wright’s directness considerations advance. One concerns the object of transmission, the thing that is supposed to be transmitted across known entailments. Wright assumes that it is S1’s justifying power for S3 that is supposed to be transmitted across the entailment from S2 to S3: what would be transmitted across the link between S2 and S3, if this were a successful transmission, is S1’s justifying power. In Soccer, a successful transmission of justification from S2 to S3 would turn S1 for the subject into a justifier of S3; however, this is not so, Wright argues, because S1 is already a justifier for the subject for S3, a justifier for S3 as directly as it is for her for S2. So, the underlying idea is that justification fails to transmit from S2 to S3 because the justifier of S2 – namely, S1 – fails to be transmitted from S2 to S3 (S1 is, as it were, already there with S3, and thus there is no work left to be done for the purposes of transmission). If only S1 were to become a warrant (or justifier) for the subject for S3 because S1 is a warrant for S2 (and the subject knows that S2 entails S3), then we would, by Wright’s lights, have a case of genuine transmission. Since, however, S1 does not become a justifier for S3 in virtue of its being a justifier for S2 and the entailment between S2 and S3 (it is already a justifier for S3), transmission fails. I think this idea of Wright’s (which, as we shall see, however, does not originate with Wright) – which I call OBJECT OF TRANSMISSION – is fundamentally right, and it will play a central role in my argument in the next section.

The other idea is that it is because, under the circumstances described, S1 is a direct justifier for the subject for S3, avoiding the circuitous route via S2, that deduction from S2 to S3 fails to transmit justification. I agree with Wright that directness is a source of transmission failure; so, if S1 is a direct justifier for the subject for S3, then it straightforwardly follows that there is no transmission of justification from S2 to S3 across the known entailment from the former to the latter.Footnote ⁹ If S1 is already a justifier for our subject for S3, as it should be if S1 is a direct justifier for her for S3, then S3 cannot acquire any additional justification in virtue of the fact that she knows that it is entailed by S2 that is also supported by S1. The fact that a given subject is capable of marshalling more deductive steps, thanks to her greater knowledge of entailments, than another subject to arrive at the same conclusion on the basis of the same premises, justified to the same degree for both, does not make that conclusion more justified for the former than the latter.

This gives one main way in which an OBJECT OF TRANSMISSION might fail to be satisfied:

An important question is this: is it really the case that S1 is a direct justifier for a subject for S3? In order to answer this question, we need to take a closer look at Wright’s own account of what it is for a piece of evidence to be a direct justifier for a proposition for a certain subject. The account of direct justifier Wright offers can be properly called “counterfactual,” for it assesses whether a piece of evidence is a direct justifier for a certain proposition on the basis of what would happen – or what we would say – if certain conditions were to obtain. To see this, consider the reason Wright gives for thinking that in Soccer, S1 is a direct justifier for S3:

The idea is that since we would have S1, by itself, as a justifier for S3 under the circumstances in which the referee’s flag were “raised,” we should conclude that even under the (actual) circumstances in which the referee’s flag “remains down,” S1, by itself, is a direct justifier for S3. In other words, since there would be no transmission of justification via the inference from S2 if the referee’s flag were raised (simply because S2 would not be justified under those circumstances), there is no transmission of justification via that inference when, as in the actual case, the referee’s flag remains down.

Let us say that X counterfactually depends on Y just in case if Y did not obtain, X would not obtain. According to COUNTERFACTUAL, the counterfactual dependence of X on Y is necessary for it to be the case that X obtains in virtue of Y. Wright takes it for granted that counterfactual dependence of S3’s justification on S2’s justification is necessary for it to be the case that S3 is justified (at least in part) in virtue of the fact that S2 is justified. Wright’s reasoning, which fits COUNTERFACTUAL, then goes like this: given that S3 would be justified (by S1) if S2 were not justified, then it is not (and not even at least in part) in virtue of the fact that S2 is (actually) justified that S3 is actually justified.

However, COUNTERFACTUAL is false: from the fact that X would obtain even if Y did not obtain, we cannot derive the conclusion that it is not in virtue of Y that X actually obtains. That is, the counterfactual dependence of X on Y is not necessary for it to be the case that X obtains in virtue of Y. My wife bought chocolate candies for me, so I have chocolate candies in virtue of the fact that she bought me some. However, even if she did not buy any chocolate candies for me, I would still have some chocolate candies – I would buy them for myself! I already had formed a resolute intention to buy chocolate candies, and if I were not stopped on my way to the market by my wife holding candies, I would buy them. So, contra COUNTERFACTUAL, the following conjunction appears to be true: I actually have chocolate candies in virtue of the fact that my wife bought me some, despite the fact that I would still have them if she did not buy me some.

The problem with COUNTERFACTUAL is a version of the preemption problem that afflicts counterfactual analyses of causation. On a (reductionist) counterfactual analysis of causation, causation can be explained, without remainder, in terms of counterfactual dependence: counterfactual dependence is both necessary and sufficient for causation. The preemption problem concerns the necessity claim. Suppose that Shooter 2 had fired a little later than Shooter 1 and that the dictator was already dead by the time Shooter 2’s bullet arrived. However, if Shooter 1 had not fired, Shooter 2’s bullet would have killed the dictator. In this case, it is clear that it is Shooter 1’s firing that caused the death of the dictator, despite the fact that the death of the dictator does not counterfactually depend on Shooter 1’s firing (it is not true that if Shooter 1 had not fired, the dictator would not have died since, in that case, Shooter 2’s bullet would have killed him). Shooter 1’s firing preempts Shooter 2’s firing. It is widely acknowledged that preemption cases such as this one “do not play nicely with” (Broadbent Reference Broadbent and Dowden2026) and proved to be “the biggest bugbear for” (Menzies and Beebee, Reference Menzies, Beebee, Zalta and Nodelman2024) counterfactual analyses of causation. And, as the chocolate candies case above is intended to show, preemption is as deadly a problem with COUNTERFACTUAL (which is intended to serve a necessary condition for the in-virtue-of relation rather than for causation) as it is with counterfactual analyses of causation.

Given that there are good reasons to think that COUNTERFACTUAL is false, then, from the fact that S3 would be justified even if S2 were not justified, we cannot conclude, contra Wright, that it is not in virtue of the fact that S2 is justified that S3 is actually justified. That is to say, the evidential structure that obtains among S1-S2-S3 when the background information is as it actually is (that is, when the referee’s flag remains down) need not be the same as the evidential structure that obtains among them if the background information were different (that is, when the referee’s flag were raised): it might be because S1 is evidence for S2 that S3 is justified when the referee’s flag remains down, despite the fact that S3 would still be justified when the referee’s flag were raised. I conclude that while Wright is right that directness generates transmission failure, his account of the conditions for directness rests on a mistaken principle like COUNTERFACTUAL and thus is defective. Since, without an adequate account of directness, we cannot correctly identify arguments that are non-transmissive due to directness, Wright’s claim that arguments such as Soccer (when the background information is fixed in a certain way) suffer from transmission failure lacks persuasiveness.

I would like to conclude this section by emphasizing three points that have emerged. Firstly, a significant idea in Wright’s discussion of directness is OBJECT OF TRANSMISSION, and a central way in which justification might fail to be transmitted across known deduction is due to a piece of evidence being a direct justifier for the subject for the conclusion of the argument. Secondly, Wright’s counterfactual account of direct justifier (for a given subject) is defective. Thirdly, we need (but currently lack) an adequate account of direct justifier.

4. Transmission of justifiers

In the preceding section, I have extracted a significant idea from Wright’s discussion of directness that I have dubbed OBJECT OF TRANSMISSION, which goes, let’s remember, as follows:

There are some passages in Wright’s writings that clearly indicate that he endorses OBJECT OF TRANSMISSION. Consider, for instance, what Wright says with respect to Dretske’s famous Zebra case:

In this passage, Wright says that the justification that one has for “Those animals are zebras” does not transmit to “They are not mules disguised as zebras” (despite the fact that the former is justified and the subject knows that the former entails the latter) because the justifier of the former (that is, “the look of the beasts”) is not a justifier for the latter (or “have no bearing on this possibility”). In the Zebra case, the OBJECT OF TRANSMISSION is not fulfilled because the look of the beasts does not become a justifier for the subject for the proposition “they are not mules disguised as zebras” (which is in turn simply because that look is not a justifier simpliciter for that proposition). And it is because the OBJECT OF TRANSMISSION is not fulfilled, Wright effectively argues, that there is failure of justification transmission in this case.

Here we have another way (that is, distinct from FAILURE 1, observed in the previous section) in which OBJECT OF TRANSMISSION might fail to be satisfied:

If OBJECT OF TRANSMISSION is true, then justification fails to be transmitted from p, justified by e, to q, across known entailment, unless e is a justifier simpliciter for q. This is what happens in the Zebra case according to Wright: the look of the beasts is not a justifier simpliciter for the “not-disguised-mules” proposition, despite the fact that it is a justifier simpliciter for the “zebras” proposition, which the subject knows entails the not-disguised-mules proposition; and this is why transmission fails in this case.

The gist of OBJECT OF TRANSMISSION is that transmission of justification is transmission of a justifier: the justifier of a given proposition is transmitted to what it knowingly entails as a justifier.Footnote ¹⁰ When transmission succeeds, what is a justifier for a proposition becomes a justifier for the subject for what the subject knows that proposition entails. Before proceeding further, I would like to make two points concerning this notion of “becoming” and how it is related to the notion of “being a justifier for a subject.” Firstly, only those things that are not already F can become F by going through a certain process. So, only those things that are not already justifiers for a subject for a certain proposition can become justifiers for her for that proposition across known entailment. This is the point, as we have seen in the previous section, on which directness considerations and FAILURE 1 rest.

Secondly, the fact that a piece of evidence is a justifier simpliciter for a proposition obtains irrespective of the contingencies that might befall us as mere mortals, irrespective in particular of whether a given subject comes to know whether a given proposition entails another. So, there is no such thing as a given piece of evidence becoming a justifier simpliciter for a proposition in virtue of the fact that the subject comes to know that a proposition entails another. When transmission succeeds, what happens is that what is a justifier simpliciter for a proposition becomes a justifier for the subject for that proposition: the justificatory relations that obtain among pieces of evidence remain the same as they have always been; what changes through transmission of justification is that what has not been a justifier for the subject for a proposition becomes a justifier for her for that proposition (more on this below). And, accordingly, if a given piece of evidence is not a justifier simpliciter for a proposition, then it cannot become a justifier for the subject in virtue of the subject’s knowledge that that proposition is entailed by a proposition for which that evidence is a justifier. And this is the point captured by FAILURE 2.

OBJECT OF TRANSMISSION is an old idea, and various defenses of some of its versions can be traced back at least to Dretske’s (Reference Dretske1969, Reference Dretske1970) attack on closure of justification (and also, notably, to Thalberg’s (Reference Thalberg1969, Reference Thalberg1974) defense of the justified true belief analysis of knowledge against Gettier’s counterexamples). Reflecting “nearly half a century later” (Reference Dretske, Steup and Sosa2014: 29) on the relevant issues his earlier works addressed, Dretske notes that “a more revealing way” of making his main point has emerged as a result of work by Wright and Davies. Dretske writes:

It is clear that Dretske here subscribes to OBJECT OF TRANSMISSION. It is, Dretske holds, because reasons (or evidence) for believing that what is in the bottle is wine are not reasons (or evidence) that what is in the bottle is not merely colored water, so justification fails to transmit from the former proposition to the latter. Dretske simply takes it for granted that transmission of justification is transmission of the justifier.

It is generally acknowledged that if OBJECT OF TRANSMISSION is true, then there are more or less intuitively clear cases such as Dretske’s Zebra and Wine cases in which transmission of justification seems to fail. Here are two representative passages, the former about the Zebra case and the latter about the Wine case:

The important question is, then, whether the OBJECT OF TRANSMISSION is true. In the next subsection, I argue that there are good reasons to think that the answer is yes.