If That Were True I Would Have Heard about it by Now
Abstract and Keywords
This chapter examines beliefs formed on the basis of one's expectation that if things were not so, one would have heard about it by now. It is argued that a proper reliabilist assessment of these beliefs must attend to various cognitive dispositions of the members of the subject's community. On the model offered, such “coverage‐supported” beliefs are a species of inferentially‐acquired beliefs, where the inference is from a premise asserting the subject's expectation of relevant coverage. The epistemic goodness of the subject's expectation, in turn, is determined by reference to prevailing social practices and institutions.
The theme of this book is our epistemic reliance on others—the sort of reliance we exhibit when we depend on others for our knowledge of the world. My claim has been that this reliance has some far‐reaching implications for epistemology in general, and for Process Reliabilism in particular. So far, I have been pursuing these implications by focusing on the paradigmatic sort of epistemic reliance, which is seen in testimony cases. In this chapter, I shift my focus from testimony to one other sort of case in which we rely on others for our knowledge of the world.
In this chapter, I examine the case of beliefs that are formed through epistemic reliance on what I will be calling the coverage‐reliability of one's community. A paradigm example of this sort of reliance is seen in cases in which, reflecting on whether p, one comes to believe that ∼p on the grounds that if p were true one would have heard about it by now.1 I argue that a proper reliabilist assessment of the formation and/or sustainment of the beliefs in question must attend to various cognitive dispositions of one or more of the members of the subject's community. My proposal for doing so will be twofold. First, I will construe “coverage‐supported” beliefs as a species of inferentially acquired beliefs, where the inference is from a premise asserting the subject's (p.155) expectation of relevant coverage. Second, I will argue that the premise‐belief manifesting the subject's expectation of relevant coverage should be assessed in terms of prevailing social practices and institutions. These practices and institutions, I will argue, are part of the “background conditions” on the formation of the “coverage‐supported” beliefs themselves. It is in terms of these conditions that we calibrate the reliability of the beliefs which manifest one's expectations for coverage, and deviations from these conditions constitute a potential knowledge‐undermining element of luck.
I begin with some cases, to fix ideas.
WMD Over lunch, you and a friend are having a discussion about overseas ventures by the US military during the administration of George W. Bush. She raises the question whether weapons of mass destruction (WMDs) were ever found in Iraq. You think for a moment and respond that, no, they were not. (You reason that if such weapons had been found, you would have heard about it by now.)
FOREIGN POLICY As his mind is wandering, it occurs to Smith—out of the blue—to wonder whether the Prime Minister announced a new major change in foreign policy last week. Believing that if the Prime Minister had done so he (Smith) would have heard about it by now, Smith forms the belief that the Prime Minister did not announce a new major foreign policy change last week.
HOLLYWOOD Listening to the locals discuss the goings‐on of various Hollywood celebrities, McSorley overhears a juicy tidbit regarding Toothy Thompson, a particularly famous celebrity who is nearly universally regarded (including by McSorley) as a person of high integrity. According to the speaker, Toothy has actually led the life of a degenerate who has barely escaped legal (p.156) prosecution on various occasions. Given Toothy's reputation and the media's voracious appetite for Hollywood scandal, McSorley rejects the testimony: she thinks to herself that if any of this were true, she would have heard about it by now (from some more familiar source).
I believe that cases of this sort are of great epistemological interest. They make clear that our dependence on others for what we know and justifiably believe outstrips our reliance on their testimony. What is more, they are quite common.2 In this respect the following remarks of John McDowell are apt:
Since I am assuming a reliabilist epistemology, my discussion of these cases will proceed accordingly. My questions will be as follows. Under what conditions are beliefs of this sort reliably formed? Under what conditions do such beliefs amount to (what the reliabilist will recognize as) knowledge, or doxastically justified belief? After addressing these questions in the sections following, I will conclude by returning to the bigger picture, discussing what these sorts of case tell us our reliance on others, and about the organization of our epistemic communities.
Consider someone who keeps himself reasonably well up‐to‐date on events of note; suppose he listens to a reliable radio news broadcast at six o'clock every evening. Can we credit such a person at three o'clock in the afternoon on some date late in the life of, say, Winston Churchill, with knowledge that Churchill is alive? . . . Intuitively, the answer is ‘Yes.’ Something like that is the position we are all in with respect to masses of what we take ourselves to know, concerning reasonably durable but impermanent states of affairs to whose continued obtaining we have only intermittent epistemic access. If challenged, we might say something like ‘If it were no longer so, I would have heard about it’; and we are quite undisturbed, at least until philosophy breaks out, by the time‐lag (p.157) between changes in such states of affairs and our hearing about them. . . . It would be difficult to overstate how much of what ordinarily passes for knowledge would be lost to us, if our epistemology of retained knowledge did not allow that sort of knowably risky policy to issue in acceptable knowledge claims when the risks do not materialize. (McDowell 1994/1998: 422–3)
The sort of belief at issue is one whose formation or sustainment involves an appeal to the conditional If p were true I would have heard about it by now. I will call this the truth‐to‐testimony conditional. Beliefs formed or sustained by appeal to this conditional—a belief in [∼p] formed through the subject's assumption that if p were true she would have heard about it by now—I will call coverage‐supported belief. Although I will have more to say on the matter below, for now I simply note that the relevant notion of coverage is seen in the believer's reliance on a source to be both (i) reliably apprized of the relevant facts in a certain domain and (ii) disposed to offer reliable reports regarding the obtaining of these facts (when they are believed by the source to have obtained). In what follows I will focus on newly formed coverage‐supported belief (FOREIGN POLICY) rather than cases in which coverage reliance supports a previously formed belief (WMD) or rejecting a piece of currently observed testimony to the contrary (HOLLYWOOD). (p.158)
What sort of facts—about one's community, and one's place in that community—would render it likely that a newly formed coverage‐supported belief will be true? Assuming reliabilism about knowledge as well as justification, we can distinguish two questions here. First, under what conditions is such a belief reliably formed, and so doxastically justified? Second, under what conditions is a coverage‐supported belief sufficiently reliable to count as knowledge? (Expanding on my previous terminology, I will call the former sort of reliability ‘G‐reliability,’ and the latter sort of reliability ‘K‐reliability.’3)
I begin with the question regarding knowledge. I want to argue that there are five conditions that are jointly sufficient for K‐reliability in the formation of coverage‐supported belief.4 (In Section 2.2 I will move on to consider how to think of G‐reliability in connection with coverage‐supported belief.)
The first condition which is part of a set of jointly sufficient conditions on K‐reliably formed coverage‐supported belief I call the source‐existence condition. This condition requires that there be some subgroup of members of the hearer's community—we will call this group “the source”—who are disposed to report about the relevant sort of matters. This subgroup might be one that is traditionally recognized by virtually everyone in the community (the traditional print and TV media, for example). Or the subgroup in question might be one to which the subject herself bears some special, personal connection (a group of her friends, say, whose members are particularly interested in, and disposed to publicize to the others what they have learned about, some subject matter). I do not assume that these exhaust the possibilities. (p.159)
In addition to the source‐existence condition, a second condition is what I will call the reliable‐coverage condition.5 This condition requires that the relied‐upon source must be reliable in uncovering and subsequently publicizing truths about the domain in which the subject is exhibiting coverage‐reliance. Let D be a domain of interest to subject H, let p be any proposition in D regarding whose truth H might take an interest, and let α be some source on whom H could rely on matters pertaining to D. Then we can characterize the relevant notion as follows:
CR α is coverage‐reliable in D =def
α (i) will (investigate and) reliably determine whether p, (ii) will be reliable in reporting the outcome of that investigation, and (iii) will satisfy both of the previous two conditions in a timely fashion (more on which below).
GCR There is generic coverage‐reliability in D relative to H =def
There is some source or other in H's community that is coverage‐reliable in D.
As it is formulated, CR captures what we might call a non‐attuned sort of coverage‐reliance, one in which the subject H coverage‐relies on a source, α, who may or may not know that H is so (p.160) relying, and who (even if α knows that H is so relying) may or may not know H's specific informational needs and expectations. But there can be other cases with a source that is attuned to the scope and informational needs and expectations of its audience; and such a source will be one that can explicitly aim to render itself coverage‐reliable relative to that audience. I offer the following as capturing this notion of “Attuned” Coverage‐Reliance:
ACR α exhibits attuned coverage‐reliability in D relative to H =def
(I) α has knowledge of both the scope of the audience φ that relies on it for (some of) their informational needs, and the information‐relevant expectations that members of φ have with respect to α itself; (II) H is in φ; and (III) for any proposition p in D, if it is reasonable for α to suppose, both that (a) members of φ would be interested in the truth‐value of p, and that (b) members of φ are likely to rely on α for the information whether p, then α (i) will (investigate and) reliably determine whether p, (ii) will be reliable in reporting the outcome of that investigation, and (iii) will satisfy both of the previous two conditions in a timely fashion.
Both CR and ACR include a condition, (iii), that requires the relevant discoveries and reports must be made in a ‘timely fashion.’ It is perhaps slightly misleading to speak of “timeliness.” What condition (iii) requires is this: α (the relied‐upon source) must be such that, at the time t at which α is being relied upon by H, (p.161) it is true that, were there some relevant discovery to be made, α would have made the relevant discovery by t, and would have reported on the matter. I speak of ‘timeliness' if only to suggest that there must be some sort of coordination between the time‐related expectations of H, on the one hand, and the abilities of α to make any relevant discoveries, on the other. For this reason we might speak, not of timeliness, but of the requirement that there be an interval of time sufficient for the discovery (were one to be made) of any relevant facts. For this reason I will call condition (iii) the sufficient interval condition.
We can all agree, I suspect, that if a subject forms a coverage‐supported belief at a point in time prior to the relied‐upon source having completed its investigation, then the subject does not count as knowing, even if her belief is true. The explanation is that the sufficient interval condition was not satisfied. More complicated is the issue regarding how to determine what counts as a “sufficient interval” for the completion of a competent inquiry. By what point must a source have reported on any discovery, were there one to be made, so that her reports count as having been made in a “timely” fashion? This is a very complicated matter; here I can only offer some rather plaid generalizations. While timeliness depends to a great extent on hearer expectations, these expectations must be grounded in facts regarding how long it would standardly take the relied‐upon sources to find out about, investigate, and then report on any newsworthy development. With this in mind, I submit that the “timeliness” requirement imposed by the sufficient interval condition should be relativized to the sort of news at issue, to the capacities of the sources on which one is relying for coverage, and to social customs.
Timeliness ought to be relativized to the sort of news at issue. For one thing, the requirements of timeliness will reflect the probability that changes of the sort in question occur in the interval of time since one last heard a relevant report. So, for example, imagine McDowell's subject above only listened to the radio weekly. Even so, if he is within one of the intervals between the weekly radio (p.162) reports he gets on political matters, my own impression is that such a subject would still be in a position to know that the Prime Minister is still alive (given that (s)he still is); and this impression becomes overwhelming if we assume that news of the Prime Minister's death would have quickly made its way to the subject one way or another (even if not at first by radio). In other cases, more regular updating is needed, as when one aims to have knowledge (or even merely true beliefs) about the local weather forecast for the weather three weeks hence: even if one knew ten days ago, on the 1st of the month, what the weather report was for the 20th of the month, once ten days has gone by there is no impression that one continues to know, on the 10th, what the weather report is for the 20th of the month: weather reports change with great frequency. Similarly, some types of news—the outcome of major national elections, the assassination of political leaders, the start of major wars, the outcome of the widely followed championship game, scandals or deaths involving the very famous,6 and so forth—can be expected to be announced immediately, following the relevant events. For these types of news, timeliness requires more or less immediate reporting. For other types of news, by contrast, timeliness merely requires that there be periodic reports (where the time between periods is a function not only of how quickly regular developments arise but also how often hearers expect to be updated on such matters). It is also worth noting that some types of news take longer to investigate, and this will have an effect on the requirements of timeliness. Even for those best situated to investigate such matters, corruption in government can take a very long time to investigate properly; reports on such matters can be timely even if they are issues several years after the alleged corruption took place. (p.163)
In addition, timeliness must be relativized to the capacity of the source(s) on which one is relying for coverage. The Wall Street Journal has the capacity to investigate and report daily on the major goings‐on that affect world business interests (or at least those goings‐on that take place in the major businesses operating in the major business centers of the world); whereas the local paper does not (it probably gets its business news from the Journal itself or the AP newswire). One might expect that the sports section of the local newspaper can get to the bottom of the alleged spat between McBeefy and Coach rather quickly; but your child, who follows the team avidly every Sunday but who does not read the sports pages, cannot be expected to do so.
In relativizing the “sufficient interval” condition to the capacity of the relevant sources, we may have to attend to various highly contingent aspects of the situation. The following ‘absent reporter’ case illustrates. Smith relies for his local news on his town's weekly paper The Community Times. Since Smith is largely bedridden, he is especially dependent on this paper for his news. Wondering whether there have been any important developments in the past two weeks in the local government's attempts to balance the budget, Smith comes to believe that there haven't been any, since if there had been he would have read it in the paper by now (as he has scoured the papers he has received in the interim). Smith knows that the local paper is highly reliable, and that the reporter who covers such news, Wonky, is an excellent reporter. (Smith knows both that Wonky regularly investigates all the relevant news regarding the local scene, and that his reports are quite reliable.) Unbeknownst to Smith, however, Wonky has been out of town for the past month, and the newspaper did not hire anyone to replace him. (The plan was that Wonky would write about all goings‐on in his absence as soon as he returned.) Suppose that it's true that there have been no important developments in the past two weeks in the local government's attempts to balance the budget. Does Smith's belief to this effect amount to knowledge? Insofar as relevant developments were likely in that interim, the clear verdict (p.164) is that Smith's true belief does not amount to knowledge. The explanation for this, I submit, is that there has been something wrong with the coverage Smith has been receiving over this interval. Since conditions (i) and (ii) of the coverage‐reliability condition are satisfied, I submit that the proper explanation is that condition (iii) is not satisfied. Reliable coverage involves timely coverage, and whether coverage is timely in the relevant sense may depend on various contingencies (such as whether there is a substitute for a certain key reporter when that reporter goes on vacation).
Finally, timeliness must be relativized to social customs. Major foreign policy changes are typically announced in standard ways; we might reasonably expect that when these standard ways have not been used then no such policy has been announced. By contrast, there is no standardized way to announce many of the other changes in which we might take an interest.
So far, I have proposed three conditions—the source‐existence condition, the coverage‐reliability condition, and the sufficient‐interval condition (part of the coverage‐reliability condition)—which I am proposing as part of a set of jointly sufficient conditions on K‐reliability in the process by which a coverage‐supported belief is formed. Each of these three conditions are conditions on the subject's community. I turn now to the final two conditions, which are conditions on the coverage‐relying subject, H, herself. First, there is what we might call the silence condition: this condition requires that, in point of fact, H has not encountered any relevant report to date. Second, and relatedly, there is what I will call the receptivity condition, requiring H to be such that she would come across whatever relevant reports were offered by the source(s) on whom she was relying, were one to be made. To be sure, the subject can satisfy the receptivity condition without having to receive the relevant reports directly from the source itself: it may be that there is a more extensive chain of communication linking her to the source. So long as the communication chain is itself both sufficiently reliable (preserving whatever relevant content there (p.165) was in the original report(s)), sufficiently complete (passing on all of the relevant reports), and sufficiently well‐publicized (reaching at least some of the places where the subject is likely to encounter the transmitted message), the subject then would have come across the reports from the source(s) on which she is relying, and so satisfies the receptivity condition.
My claim so far is that these five conditions—the source‐existence, reliable‐coverage, sufficient‐interval, silence, and receptivity conditions—are jointly sufficient for K‐reliability in the acquisition of a newly formed coverage‐supported belief. At the same time, it would seem that the satisfaction of these conditions is not required for the subject to acquire a doxastically justified coverage‐supported belief. To illustrate: in the ‘absent reporter’ case above, it is arguable that Smith's belief is doxastically justified, even though it falls short of knowledge. Other cases involving justified coverage‐supported belief that fail to amount to knowledge would be easy to describe. Perhaps a subject is regularly informed by a source but (unbeknownst to the subject) just happens to be out of the source's reporting range on the day when a report would have been made (and our subject would have had no other way to get the report in question); perhaps sources that are highly reliable in discovering the relevant facts happen to miss a crucial fact (this happens only rarely); and so forth. In these cases it is arguable that the whole process (involving the hearer's coverage‐reliance on the source) was G‐reliable, despite not being K‐reliable. If so, then these would be cases of doxastically justified coverage‐supported belief that fail to amount to knowledge.
How then should we think of the conditions on the acquisition of G‐reliable (doxastically justified) coverage‐supported belief? An initial strategy would be to try to formulate weaker versions of each of the source's, reliable‐coverage, sufficient‐interval, receptivity, and silence conditions, where the weaker versions are satisfied so long as conditions make it likely (though they need not ensure) (p.166) that each of the five conditions on K‐reliable coverage‐supported belief are satisfied. I say that this is an initial strategy for framing the conditions on G‐reliability in the formation of a coverage‐supported belief. Before we can embrace this strategy, however, we must confront what I regard as a challenge—in fact, the central challenge—facing any reliabilist account of the doxastic justification of coverage‐supported belief.
The challenge can be presented in the form of a loose dilemma regarding the nature of the belief‐forming process involved in coverage‐supported belief, as follows. On the one hand, there are reasons to think that coverage‐supported belief should be understood in terms of the extendedness hypothesis introduced in previous chapters. For one thing, the subject who forms a coverage‐supported belief is manifestly exhibiting epistemic reliance on her peers, and, in Chapter 4, I argued that epistemic reliance on one's peers extends the belief‐forming process involved. And, in addition, it seems plain that the cognitive processing that takes place in the mind/brain of the coverage‐relying subject herself is not what does the justifying work for her coverage‐supported belief: the lion's share of that work is done by those on whom the subject is relying for coverage, as it is their work that renders the subject's coverage‐supported belief likely to be true. On the other hand, there are reasons to think that coverage‐supported belief should not be understood in terms of the extendedness hypothesis. Not only would this proposal make the process of coverage‐supported belief one that involves cognitive processing in a widely distributed network of individuals (something that can give us pause even if we accept the idea of extendedness in testimony cases). More importantly, the sort of epistemic reliance exhibited in cases of coverage‐supported belief would appear to be different from that in testimony cases, in ways that bear on the question of the nature of the belief‐forming process in play.
In light of this looming dilemma, let us return to the initial strategy presented above, on which G‐reliability in the formation of coverage‐supported belief is taken to be a matter of the high (p.167) likelihood that the five conditions from Section 2.1 are satisfied. If this strategy is to be vindicated, the proposal that results must accommodate the strength of the reasons on both sides of the issue of extended processing. To this end, let us develop these reasons a bit further, and see whether the initial strategy can accommodate the various insights on both sides.
I begin with the reasons for thinking that coverage‐supported belief should be understood in terms of the extendedness proposal introduced in previous chapters.
One such reason emerges from a comparison between testimonial belief and coverage‐supported belief. Starting with testimonial belief‐formation, consider that part of the process that takes place in the hearer, the consumer of testimony. This process will include all relevant cognitive processing that takes place in her mind/brain as she comprehends the testimony she has observed, monitors it for trustworthiness, and moves to accept it (or not, as the case may be). If Process Individualism is true, then this exhausts the process of testimonial belief‐fixation. But in Chapters 1–5 I argued that the process that takes place ‘in’ the hearer is not what does the lion's share of the cognitive work that renders the belief likely to be true. Take a case where Franklin, reading the morning paper, comes to believe that the President is in Honolulu today, through trusting a report to this effect. As he forms this belief, Franklin is currently thousands of miles from Honolulu (in fact he has never been there), and he has never in his life seen the President in person. How can it be that Franklin's belief that the President is in Honolulu today is likely to be true? (I think that all of us, but especially Orthodox Reliablists, are less impressed with this than we ought to be.) It will be replied that the reliability of Franklin's belief reflects the reliability of the processes through which he comprehends and assesses the testimony from the morning paper. But I think that this answer is ideologically driven: rather than manifesting any insight about the nature of the process that renders Franklin's belief reliable, it reflects nothing so much as a strong prior commitment to Process Individualism. This answer is also (p.168) curious in another respect. Suppose that the reporter's belief that the President is in Honolulu today was acquired at first hand. If we were to ask the Process Reliabilist what renders this (the reporter's) belief reliable, the answer would surely cite the reliability of cognitive processing in the reporter's own mind/brain. Given that Franklin himself is relying on this reliability, it is curious that this part of the story simply drops out of the answer to the question how Franklin's belief achieves its reliability.7 A better answer to our question—by which I mean a more insightful and complete answer—would be that, while processes internal to Franklin are relevant to the reliability of his belief, what does the lion's share of the work here is cognitive processing in the mind/brain of the reporter herself (together with the technologies that enable her to reliably disseminate what she has learned first hand). The extendedness proposal captures this insight, in that it regards the cognitive processing in the reporter's mind/brain to be part of the very process through which Franklin forms his testimonial belief.
Moving on to coverage‐supported belief, consider now that part of the process that takes place in the subject, the one who is relying on her community for reliable coverage. Once again, this process will include all relevant cognitive processing that takes place in her mind/brain, as she forms her belief through endorsing the truth‐to‐testimony conditional. Given Process Individualism, this exhausts the process of coverage‐supported belief‐fixation. But, once again, it is natural to think that the process that takes place ‘in’ the hearer is not what does the lion's share of the cognitive work that renders a subject's coverage‐supported belief likely to be true. To illustrate, take Franklin again. Having just noted that he has not heard from his brother Matt in the three days since Matt was supposed to have moved his whole family to Melbourne, Australia, a worry (p.169) flashes through Franklin's mind: maybe their plane crashed. (Yes, Franklin has these worries occasionally.) Fortunately, reason takes over immediately, and Franklin quickly reassures himself that their plane did not crash, since if it had he would have heard about it by now. Now Franklin is currently hundreds to thousands of miles from any point along the trajectory of the plane that was to have taken Matt's family to Australia; and he is thousands of miles from Australia itself (in fact he's never been there). How can it be that Franklin's belief, that his brother's family arrived safely in Melbourne, is likely to be true? Here the impression is overwhelming that reliability in this belief is not any single individual's cognitive achievement.
Indeed, this impression is even stronger in connection with coverage‐supported belief than it is with testimonial belief. Nor should this come as a surprise; it has to do with the nature of what we might call the ‘individualistic input’ in the two cases—the input into the mind/brain of the subject whose belief is under assessment. In the case of testimony, that input is a perceptual experience as of an assertion that p (or something like this). In such an experience it appears to the subject as if she is receiving some linguistic signal; and the individualistic cognitive processing involved is that of recovering the force and content of the (apparent) speech act, and assessing its credibility, in order to reach a determination whether to accept the content in question. In the case of coverage‐reliance, by contrast, the individualistic ‘input’ is nothing other than an acknowledged absence of any memory impression of having been told that ∼p. By anyone's lights this should seem to be a much more meager basis for believing that p, than the basis one has when (it appears that) one has been told that p straight‐out. It is for this reason that dependence on social factors for the reliability of coverage‐supported belief can seem more substantial than it is in the case of testimonial belief.
This impression can lead us to suppose that, as in the testimony case, so too in the coverage‐reliance case, we ought to capture the subject's epistemic reliance on her peers by postulating that (p.170) the processing involved extends to include processing in the mind/brains of all of those individuals on whom the subject is relying for coverage. We might think to describe coverage‐supported belief as involving reliance on a kind of “extended” or “distributed” memory,8 where the relied‐upon system itself is a massively distributed one that involves cognitive processing in the mind/brains of all of the individuals that contribute to the source's coverage‐reliability, and where the output of this source is silence.9
Unfortunately, there would also appear to be some reasons for resisting the proposal to regard coverage‐supported belief as involving an interpersonally extended belief‐forming process. Although these reasons may not be decisive, they are sufficient to warrant interest in alternative ways of modeling coverage‐supported belief.
A first reason for resisting the proposal to regard coverage‐supported belief as involving an interpersonally extended belief‐forming process has to do with the causal relevance of the processing that takes place in others' minds in these cases. Return to the case of Franklin's belief that his brother's family arrived safely in Melbourne. He is epistemically relying on various people for coverage: most saliently, airline and government officials as well as news reporters. (There might be others as well.) It is hard to see how processing going on in the mind/brains of all of these people is causally relevant to the formation of Franklin's coverage‐supported belief. In this respect coverage‐supported belief appears to be unlike testimonial belief. In the case of testimonial belief, the causal relevance of processes in the mind/brains of one's peers is underwritten by the causal transaction that takes place in the (p.171) source's offering and the hearer's reception of the testimony itself. Since coverage‐supported belief does not involve any positive speech contribution from one's peers, its formation involves no such transaction. So even if some of what we might call the ‘connecting counterfactuals' (for which see Chapter 5, Section 3) still hold true—for example: had one encountered testimony that p from one of the relied‐upon sources, one would not currently form the belief that ∼p—it still seems strained to regard the processes of belief‐fixation here as intersubjectively extended. There would appear to be no continuous causal sequence here through which to trace the coverage‐supported belief back to cognitive processing taking place in the mind/brains of members of one's community.
One might respond to this worry about causal relevance by noting that silence itself (as well as other sorts of absences) can be causally relevant in the production of a given effect.10 Here is not the place for an extended discussion of the causal relevance of absences. Instead, I offer the following, which I hope will be uncontroversial: even if silence (absence of testimony) can be causally relevant to belief‐formation, there remain important differences between the causal relevance of testimony and the causal relevance of silence.
To see this, return to our question about the attainment of reliability in Franklin's testimonial belief that the President is in Honolulu today. There I noted that, unless we are assuming Process Individualism, it is natural to think that the proper way to characterize this attainment will appeal to cognitive processing in the mind/brain of his informant, since it is this processing that does the lion's share of the work in rendering Franklin's testimonial belief reliable. There is a general point to be made here about reliability in testimonial belief: the fact that a testimony‐consuming subject's reliance is on a piece of information that (p.172) comes from a particular source motivates the idea that we must cite the relevant cognitive processing in that source's mind/brain, if we are to account for the reliability of testimonial belief itself. The case of coverage‐supported belief stands in stark contrast. In a good many cases there will be no single person on whose reliability the subject is relying for coverage. What is more, even in those cases on which the subject is relying on a single person for coverage‐reliability, his reliance on such a source for coverage does not amount to relying on the reliability with which any particular piece of information was produced by the source, since by hypothesis the relied‐upon source has been silent. In this respect we might say that the coverage‐relying subject's reliance on others does not involve a reliance on any determinate stretch of cognitive operations in another's mind/brain. On the contrary, it would seem more accurate to describe our coverage‐relying subject in this way: he is depending on his whole situation to be such that, were there relevant information, for example, regarding a plane crash (Franklin's case), this information would have made its way to him by now. For this reason, and unlike in the testimony case, coverage reliance might be understood to be reliance on certain general features of one's social circumstances. The appeal to particular features of another's cognitive processing, as part of the cognitive processing involved in the subject's coverage‐supported belief, appears decidedly unmotivated.
On the basis of the foregoing, I submit that the sort of epistemic reliance in play in coverage‐supported belief is importantly different from that involved in testimonial belief, in ways bearing on the relevant processes involved. In Chapter 4, I glossed the sort of epistemic reliance involved in testimony cases as follows (here I call this ‘primary’ epistemic reliance, to set it off from the sort I want to introduce in connection with coverage‐supported belief):
ERPr In forming or sustaining her belief that p, a given subject S exhibits primary epistemic reliance on (the reliability of) some cognitive process π (where π itself may be a component in a more encompassing cognitive process) iff (i) S's belief that p was (p.173) formed and/or sustained through some belief‐forming process π*, (ii) π produced output r, whose content is that q (not necessarily distinct from p), and (iii) the reliability of the belief S formed or sustained through π* depends on the reliability with which π produced r.
ERD In forming or sustaining her belief that p, a given subject S exhibits diffuse epistemic reliance on (the reliability of) some cognitive process π (where π itself may be a component in a more encompassing cognitive process) iff (a) S's belief that p was formed and/or sustained through some belief‐forming process π*, and (b) the reliability of the belief formed or sustained through π* depends on the reliability of the operations of π.
I have just highlighted what I regard as an important difference between the sort of epistemic reliance involved in coverage‐supported belief and that involved in testimonial belief: the former involves a more “diffuse” sort of epistemic reliance than the “primary” epistemic reliance I introduced (in connection with testimonial belief) in Chapter 4. I do not claim that this difference, by itself, warrants a repudiation of the extendedness model for coverage‐supported belief. But I do think that this difference is important enough that it warrants our taking an interest in alternative accounts of coverage‐supported belief. No doubt, any adequate account of the process by which coverage‐supported beliefs are formed must satisfy other desiderata. Two such desiderata are salient. First, an account of the process of coverage‐supported belief‐formation must make clear that (and how) the subject herself is epistemically relying on her source(s) for relevant coverage. (An (p.174) account that fails to satisfy this desideratum fails to treat coverage‐supported belief as a case of diffuse epistemic reliance.) But, second, an account of coverage‐supported belief‐formation must make clear precisely how a subject's sources are to be credited with the lion's share of the cognitive work that goes into rendering her coverage‐supported belief reliable. (An account that fails to satisfy this desideratum will leave it mysterious how coverage‐supported belief depends for its reliability on the coverage‐reliability of one's sources.)
My contention is this: given the recently noted difference in the kind of epistemic reliance in play, we have some reasons to favor an account of the process of coverage‐supported belief that does not endorse the “extendedness” model. In the section following I aim to provide just such an account.
In thinking about the process by which coverage‐supported beliefs are formed, let us return to our initial strategy for thinking about G‐reliability in these cases. The idea, presented at the outset of Section 2.2, was that the G‐reliability of coverage‐supported belief is a matter of the obtaining of conditions (both social and individual) that make it likely (though they need not ensure) that each of the five conditions on K‐reliable coverage‐supported belief are satisfied. In line with this, we might say that a subject's coverage‐supported belief that p is formed through a G‐reliable process if and only if it is formed in such a way as to manifest the subject's sensitivity to features indicating the likelihood that these conditions are satisfied. The challenge presently before us is to develop this idea so as to meet the two desiderata above. I propose to meet this challenge by regarding coverage‐supported belief as a species of inferential belief, where one of the premises involved is none other than (something like) the truth‐to‐testimony conditional itself.
Consider first the proposal to regard coverage‐supported belief as a type of inferential belief. The relevant inference would be from (p.175) the subject's currently formed belief that she has no memory of having been informed that ∼p, together with her belief in the relevant instance of truth‐to‐testimony conditional, to the conclusion that p.11 (This inference need not be explicit or conscious; her reliance on it might be tacit.) Such a proposal has several virtues. First, it avoids all of the objections that can be made against the proposal to treat coverage‐supported belief as employing an intersubjectively extended belief‐forming process. (The process of drawing an inference itself is not an intersubjectively extended one.) In addition, the inferential proposal makes clear the role of the truth‐to‐testimony conditional in coverage‐supported belief: this conditional serves as a premise in the inference through which the coverage‐supported belief is arrived at. Third, the inferential proposal represents the epistemic goodness of coverage‐supported belief—e.g., its G‐reliability and K‐reliability—as reflecting the epistemic goodness of the inference itself, together with the epistemic goodness of the premises of the inference. Since inferentially acquired belief is belief acquired through a belief‐dependent process, we get a further happy result: the reliability of a subject's coverage‐supported belief will be seen to depend on the reliability of the subject's belief in the relevant instance of the truth‐to‐testimony conditional.
I call this last result happy, since it is here that we can see the role that social factors play in the doxastic justification of coverage‐supported belief. In particular, the doxastic justification of coverage‐supported belief will depend on the doxastic justification of the subject's belief in the relevant truth‐to‐testimony conditional; and this, in turn, will depend on whether the subject is sensitive to the conditions under which she has relevant reliable coverage. Suppose she expects coverage on matters where coverage is unlikely. In that case her belief in the truth‐to‐testimony conditional will be false; and if she formed it under conditions in (p.176) which she was oblivious to the reigning social practices and institutions, her belief in this conditional will have been G‐unreliably formed as well.12 Since such a belief is a premise in the inference through which the coverage‐supported belief is formed, the resulting coverage‐supported belief will be G‐unreliable as well. This result is happy, since that is precisely what one would want in a case in which a coverage‐supported belief is formed under conditions in which the subject herself is not sensitive to the prevailing relevant social institutions and practices.
The inferential account of coverage‐supported belief has a fourth virtue: it accounts for another important difference between the epistemology of testimonial belief and that of coverage‐supported belief. Silence and testimony function differently, epistemically speaking.13 This difference between the epistemic roles of testimony and silence is related to the asymmetry between having positive evidence for p, and lacking any evidence against p. In the case in which one has positive evidence for a claim, one's evidence supports one's claim, which (understood in Process Reliabilist terms) might be cashed out as follows: the process involved in forming the belief that p through evidence e (where e is evidence for p) is G‐reliable.14 But in the case in which one has no negative evidence against a claim, one's belief in the claim is not thereby epistemically supported. That is, it is not in general the case that the process involved in forming the belief that p, through a lack of evidence that ∼p, is G‐reliable. After all, one might lack evidence for all sorts of reasons: one might not have looked hard enough, or in the right places, or at all, etc. It would thus appear that the reliability of belief in p, formed on the basis of a lack of evidence (p.177) to the contrary, requires more than the mere lack of evidence. It is plausible to suppose that there is an implicit sort of inference that goes on here: one infers that p from the claims (first) that there is no evidence to the contrary, and (second) that had it been the case that ∼p one would have likely encountered such evidence (that is, evidence for [∼p]). Since coverage‐supported belief that p is belief grounded in a lack of testimony that ∼p, we might then expect that this sort of belief depends on the corresponding sort of inference. This is precisely what the inferential account predicts.
Let us turn, then, to the two desiderata mentioned at the end of Section 2.2. Does the present proposal make clear that (and how) the coverage‐relying subject is exhibiting epistemic reliance on her peers? And does it make clear that (and how) the vast majority of the cognitive work that ensures the reliability of coverage‐supported belief is in the cognitive (including social) dispositions of those on whom the subject is relying for coverage? More generally, does it honor the insight that the reliability of coverage‐supported belief is not the cognitive achievement of any single individual?
To address these questions, we would do well to examine that process by which the expectations of the coverage‐relying individual are calibrated so as to be brought in line with the prevailing social practices and institutions in her community. After all, it is in such a process that an individual learns when, where, and for what sorts of thing she can rely on relevant coverage in her community. By focusing on this process we might think to find out how an individual ‘locks onto’ those social features that underwrite the reliability of her coverage‐supported beliefs. And in so doing we might hope to shed light on both the individualistic aspect of the achievement involved in reliable coverage‐supported belief—the process by which the subject herself discerns when she is likely to have the relevant coverage—as well as the social aspect of that achievement—the relevant social practices and institutions that ensure relevant coverage. (p.178)
Let us begin with those practices and institutions themselves. I submit that these should be seen as constituting the all‐important background conditions on the formation of coverage‐supported belief. In this connection we do well to highlight the following comment from Hilary Kornblith. Speaking of the epistemic relevance of the social environment within which our belief‐forming processes operate, he writes:
Following Kornblith's suggestion, I propose that the various social practices and institutions that are relied upon in coverage‐supported belief form part of “the social environment” in which coverage‐supported beliefs are formed. The idea is this. We live in a world with a dizzying variety of information sources available to us. What is more, many of us structure our environments in such a way that we have easy access to the reports of a select subset of these sources. The more traditional among us read the daily paper (hard copy), watch the evening news, or—even more old school!—listen to the news on the radio. The more technologically savvy among us follow certain Twitter streams, receive News Alerts via email or other web‐based technologies such as Google Reader, read the daily paper online, follow certain blogs, and so forth. And all of us rely, sometimes explicitly, sometimes implicitly, on less “official” sources for news: friends, neighbors, and so forth. The sorts of news‐generating and ‐publicizing practices and institutions that are prevalent in one's community constitute a core part of one's “social” environment. To a large extent, epistemic subjects are at the mercy of such an environment: a subject who lives in (p.179) a community with few or no channels for the communication of news, or where such channels as exist are highly unreliable in their reports, will be a subject who is seriously epistemically impoverished. In the former case, she will miss out on lots of truths, some of which may be of great interest to her; in the latter case, she will run an increased risk of acquiring false beliefs. My current point, however, is that in such a news‐diminished community, the epistemic significance of silence will be affected as well. After all, the silence of one's sources in a community whose sources rarely report anything interesting may not tell us very much; whereas the silence of one's otherwise coverage‐reliable sources can tell one a great deal.
Just as we wish to know whether our natural inferential tendencies are likely to give us an accurate or a distorted picture of the world, we also need to know whether our social institutions and practices are helping to inform us or to misinform us. And just as we need to examine our perceptual and inferential equipment against the background of the natural environment in which they operate, we also need to investigate these mechanisms against the background of the social environment in which they operate. Such investigations are straightforward extensions of the naturalistic project in epistemology. (Kornblith 1994: 97)
It is in this spirit that I claim that the news‐generating and ‐disseminating practices and institutions in a community constitute the background against which we assess her coverage‐supported beliefs for reliability. The take‐home point of the proposal is that a reliabilist assessment of coverage‐supported belief can't just examine the belief‐forming process that takes place within the mind/brain of the coverage‐relying subject herself; such an assessment must also assess the various social institutions and practices that form the process(es) by which news is generated and disseminated in her community. Two different coverage‐relying subjects, as alike skin‐in as any two distinct individuals can be, might nevertheless differ in the G‐reliability of their respective coverage‐supported beliefs, as one subject lives in a community in which these institutions and practices provide her with highly reliable coverage on the issue at hand, whereas the other lives in a community where the coverage is less highly reliable (and where there are more issues of interest to her that are not covered). Whatever difference there is in the G‐reliability of their respective beliefs supervenes on more than what is going on in their respective heads: it also supervenes on the social practices and institutions that surround them.
Let us turn now from the social to the individualistic aspect of the achievement in successful coverage‐supported belief. While the social institutions and practices I have been discussing constitute the (p.180) background conditions on a subject's coverage‐supported belief, it is the subject's sensitivity to the existence and nature of these institutions and practices, and her sense of what they portend in terms of the coverage that she is receiving, that determine the G‐reliability of her coverage‐supported beliefs. More specifically, a subject who is sensitive in this way, and whose expectations of coverage are well calibrated to her surrounding community, will have reliable beliefs regarding whether she is likely to have coverage on a given topic. Any particular beliefs she has in the truth‐to‐testimony conditional in a given case will likely be true.
In this respect it is noteworthy that ordinary subjects typically do form coverage‐supported beliefs only under conditions in which background conditions are propitious—suggesting that ordinary subjects are indeed calibrated to the prevailing social institutions and practices. Thus it is no surprise to learn that, while a good many subjects form coverage‐supported beliefs regarding major news issues, or regarding topics where one is aware of a group interested in disseminating relevant information, subjects don't typically form coverage‐supported beliefs on topics regarding which no one could reasonably expect coverage—the size of average fingernail clippings in the eastern United States, for example. You probably believe (or are disposed to believe) that it is not the case that a new major world war has been initiated in the last forty‐eight hours, on the grounds that something that newsworthy would have been reported if it had; but you probably don't believe (are not disposed to believe) that in the last forty‐eight hours there have been no minor skirmishes between the county governments of Pulaski and Massac Counties, Illinois (two counties that border one another in the southernmost part of my home state). Jones believes that her conceited and garrulous colleague X has not won a major research prize (if he had everyone would have heard about it by now); but she has no belief, and no disposition to believe, that her quiet and unassuming colleague Y has not won a major research prize (as she realizes that it would be just like the humble Y to hide such (p.181) a fact from others). These examples—and many others like them could be offered—appear to support the idea that our dispositions to form coverage‐supported belief do tend more or less to vary with the presence of the conditions that make it likely that these beliefs will be true. No doubt, this is because we are socialized to appreciate when we are likely to be relevantly supported, and when not. This is so much grist for my mill.
What, then, does this tell us about the G‐reliability of coverage‐supported belief? Take a subject who is sensitive to the prevailing social conditions (and their effects on relevant coverage). Such a subject, I said, will have a reliable belief in the relevant truth‐to‐testimony conditional: she will believe, reliably, that if ∼p were true she would have heard about it by now. Suppose too that she hasn't heard about it by now, and that her belief to this effect was reliably formed. In that case, her coverage‐supported belief in p, inferred from her belief in the relevant conditional and her belief that she hasn't heard anything to the contrary, would then be G‐reliably formed. For, in that case, her coverage‐supported belief was acquired through a conditionally reliable inference involving reliably formed beliefs as premises. Such is the way to understand G‐reliability in the formation of coverage‐supported belief.
No doubt, there is great variation in social practices and institutions worldwide, and even between local communities there can be relevant differences. What is more, social conditions are not as stable and long‐lasting as, for example, the environmental invariances that constitute the background conditions for our perceptual systems. But these facts are no objection to the proposal to regard the relevant social conditions as background conditions on coverage‐supported belief. For one thing, social variation and social change themselves are normal features of our environment and our world. For another, it is a normal part of socialization within a community that one becomes sensitive to prevailing social arrangements, and apprized of important social changes. Included in this socialization process is the process of learning what counts (p.182) around here as newsworthy, what sorts of news one can expect (and where and when one can expect this news), and so forth.15 Acknowledging all of this merely forces us to acknowledge that the background conditions that obtain in any given case depend on the time in question and the community in play.
This acknowledgment is not without implications. Once we acknowledge the fact of social change and recognize variation in social practices and institutions themselves, we would predict that complicated issues could arise when we assess the coverage‐relying beliefs of subjects who move between communities. But in this respect our predictions appear to be borne out. Imagine a subject who grew up in a heavily news‐intensive community where access to news of all sorts was pervasive, who then goes on to move into an informationally arid community where access to the news is much more limited (and must be sought out). If early in her time in the new community our subject forms a coverage‐supported belief appropriate to her original community, is her belief doxastically justified? If we think not, this might be because we think the issue of doxastic justification turns on how well she has calibrated her coverage‐related expectations to her present community: since she has not yet calibrated herself to her new community, and so is relying on coverage that in fact she is not likely to get there, we might regard her coverage‐supported belief as not justified. On the other hand, we might be inclined to say that in this case her belief is justified, but fails to amount to knowledge: we would say this if we think that there remain reasons to assess the G‐reliability of her current coverage‐relying belief relative to what is normal conditions for her original community. I don't want to decide this matter here; my present point is merely that the present account predicts that there will be complications in cases involving subjects who travel between communities, and this prediction is borne (p.183) out (our intuitions about the proper verdict in these cases run in various different directions).
I see parallels here with the situation that arises when content anti‐individualists assess the meanings and mental contents of a subject who travels between communities with different linguistic practices—so‐called ‘world‐switching’ cases.16 If this parallel between these two types of world‐switching cases is a good one, we would anticipate the epistemic significance of some of the emerging technologies (BlackBerries and iPhones, Twitter and Facebook, News Alerts, and Google Reader): they enable subjects to remain apprized of relevant information at any time and at any place, making it harder for any technologically savvy individual to be the victim of a relevant sort of ‘world‐switching’ regimen.17 (What are the effects of these technologies, if any, on the semantics of world‐switching?) But even for subjects without these technologies the point remains: the facts (one) that over time there is a great deal of change in the social world, and (two) that even at a time there is wide variation in social practices and institutions, do not undermine the proposal to regard social conditions as part of the ‘background conditions' for the formation of coverage‐supported belief.
Does the present proposal acknowledge that the reliability of coverage‐supported belief is not any single individual's cognitive achievement? It does: to regard the relevant social practices and institutions as background conditions is precisely to highlight that role.
In sum: the challenge facing any reliabilist account of the G‐reliability of coverage‐supported belief is to acknowledge the (p.184) various considerations that appear to be relevant—both those that emphasize the subject's epistemic reliance on her social peers (for relevant coverage), and those that suggest that this epistemic reliance does not favor treating the process of coverage‐supported belief‐formation as an interpersonally extended one. I have tried to honor all of these considerations by treating coverage‐supported belief as a species of inference‐based belief, and by regarding the various social conditions relevant to the truth of the truth‐to‐testimony conditional as part of the ‘background conditions' for coverage‐supported belief. While this proposal involves treating coverage‐supported belief in a way that does not parallel my earlier treatment of testimonial belief, this otherwise‐unhappy lack of symmetry in my treatment is warranted (I have argued) by the different sort of epistemic reliance that is in play in the two cases.
I turn, finally, to the big question: how does the phenomenon of coverage‐supported belief bear on the book's central theme of our epistemic reliance on others? It is a second instance of my contention that reliabilist epistemic assessment must take stock of ineliminably social factors. On the proposal I have offered here, these factors in question enter, not as part of the belief‐forming process itself, but as part of the ‘background conditions' on the process. What this suggests is that there are cases in which the reliability of what everyone can agree is an ‘individualistic’ belief‐forming process nevertheless depends on aspects of the prevailing social environment.
(1) I thank Frank Döring for suggesting to me (in conversation) the importance of this sort of belief.
(2) A recent typical example comes from the New York Times sports section of August 6, 2008. Yankees' manager Joe Girardi, commenting on Joba Chamberlain, a great young pitcher who had to leave the previous game due to experiencing a sharp pain in his shoulder, “guessed” that Chamberlain still felt strong, saying “I would have heard if he didn't.”
(3) G‐reliability is global reliability; and it would be natural for a reliabilist to think that K‐reliability involves both local and global reliability.
(4) I think that they are individually necessary as well, but I will not be arguing for this in each case.
(5) It is something very near this condition that the New York Times claims to satisfy, in its claim to publish “All of the News that's Fit to Print.”
(6) The death of Michael Jackson on June 25, 2009, is a case in point. He was pronounced dead by doctors at UCLA Medical Center at 2.26 p.m. (LA time), and less than twenty minutes later TMZ.com, an LA‐based celebrity news website, posted a report of Jackson's death. Interestingly, attempts to post the news on Wikipedia soon thereafter overwhelmed that site. See “Michael Jackson's Death Roils Wikipedia,” posted June 25, 2009, on cnet news: >http://news.cnet.com/8301‐10233‐10273277‐93.html< (accessed November 4, 2009).
(7) Lackey (2007b) makes a point very much in the spirit of this one as part of her criticism of the view that knowledge is true belief for which the subject deserves credit. Her point there is that we don't deserve the lion's share of the credit for (at least some) testimonial knowledge, since the lion's share of the credit belongs to the subject's informant.
(8) When I speak of ‘extended memory’ here, a point made in Chapter 5 should be borne in mind: this has no direct bearing on the cognitive‐psychological taxonomy of memory processes. The point is rather that, insofar as we are interested in a reliabilist assessment of the coverage‐supported belief, we need to assess, e.g., the coverage‐reliability of the subject's source(s).
(9) I thank Alan Millar, John Burgess, and an anonymous referee from Oxford University Press for this suggestion.
(10) For this point, I thank John Burgess, as well as an anonymous referee from Oxford University Press.
(12) What is the process through which one forms a belief in a truth‐to‐testimony conditional? This is a special case of a more general question: what is the process through which one forms beliefs in subjunctive conditionals? For one characterization of this process, see Williamson (2007: Chapter 5). For my part I will assume that there is some reliabilist story to tell here, but, as I don't know what that story is, and as it would take me far afield anyway, I will not pursue such matters further.
(14) I don't pretend for a moment that this is an adequate description of the process. Let it be a stand‐in for whatever process is used in a case of forming beliefs on evidence.
(15) Indeed, so far as I can tell, the epistemological issues surrounding the notion of newsworthiness have received no attention whatsoever from epistemologists (as opposed to scholars in journalism schools). I regard this as a lacuna in the epistemology literature. I hope to address this at some future time.
(17) A nice example from The Chicago Tribune of March 25, 2009: a rider of Metra, the local high‐speed suburban train to Chicago, has arranged to have information regarding any current train delays delivered to him in real time via Twitter (whereas previously he had to go online to find this information, to Metra's website). The result is that as long as he has his hand‐held unit (and as long as Metra reliably and quickly updates the postings of delays), he is in a position to form G‐reliable coverage‐supported beliefs regarding delays on the Metra line wherever he is: the lack of relevant tweets can be the basis for a G‐reliable belief that there are no relevant delays.