1.1. Coverage-supported knowledge and justified belief

For Goldberg, a subject's coverage-supported belief can result in knowledge or a doxastically justified belief. Coverage-supported knowledge differs from coverage-supported justified belief because a subject's coverage-supported belief need only satisfy reliability conditions to count as knowledge but a subject must properly sense whether these reliability conditions obtain to justly infer. So, coverage-supported justified belief involves a first-personal sensitivity that coverage-supported knowledge does not.

By Goldberg's lights, a subject's coverage-supported belief counts as knowledge only if it satisfies (1) five jointly sufficient conditions and (2) a non-Gettier condition. These five jointly sufficient conditions must be satisfied for a coverage-supported belief to count as knowledge because if they are satisfied, then the belief is sufficiently reliable to be properly evaluated as knowledge.

The first condition is the source-existence condition. The source-existence condition obtains if there is a person, group or organization that is disposed to report on matters such as whether ‘Canada has declared war on the US’ (‘W’). So, in the Canada Declares War Case, Victoria's belief that ~W satisfies the source existence condition because sources exist, like the New York Times and Washington Post, that are disposed to report on matters like whether Canada has declared war on the US (‘W’).

The second condition is the reliable-coverage condition. The reliable-coverage condition is satisfied if there is a source who reliably uncovers and publicizes information in a domain of interest to a subject. In the Canada Declares War Case, Victoria's belief that ~W satisfies the reliable-coverage condition because the New York Times reliably uncovers and publicizes W-like information.

The third condition is the sufficient-interval condition. The sufficient-interval condition is satisfied if a source relays information in some domain of interest to a subject according to the “time-related expectations of [the subject], on the one hand, and the abilities of the [source] to make any relevant discoveries, on the other” (Goldberg Reference Goldberg2010: 161). In the Canada Declares War Case Victoria's belief that ~W satisfies the sufficient-interval condition if, say, the New York Times reports W-like information in a time interval such that it meets Victoria's time-related expectations where her time-related expectation is that if W were true, she would expect to see it reported on the New York Times’ website within, say, an hour from when Canada declared war on the US.

The fourth condition is the silence condition. A subject satisfies the silence condition if she has not come across a report regarding her coverage-supported belief. In the Canada Declares War Case Victoria satisfies this condition because she has not come across or encountered any report that W. In other words, Victoria's reliable sources are silent on whether W.

The fifth condition is the receptivity condition. A subject satisfies the receptivity condition if she “would come across whatever relevant reports were offered by the source(s)” (Goldberg Reference Goldberg2010: 164, original emphasis). In the Canada Declares War Case, Victoria satisfies this condition if, say, her internet is working properly and reliably such that if the New York Times did report that W, then she would receive the report that W. In this case, the New York Times has not reported W so she is justified in believing ~W partly because she is properly receptive to any New York Times reports that W.

By Goldberg's lights, a subject is doxastically justified in her coverage-supported belief if she is sensitive to the social conditions that “make it likely” that the five knowledge conditions obtain. These social conditions involve the “various social institutions and practices that form the processes(es) by which news is generated and disseminated in her community” (Goldberg Reference Goldberg2010: 179). For example, Victoria properly expects her news sources such as the New York Times to inform her in a timely and reliable way about whether Canada has declared war on the US. Her expectations are calibrated to the reliability of the processes that compose her community's information gathering and disseminating (Goldberg Reference Goldberg2010). The sufficient reliability of these processes make one of her inference bases sufficiently reliable to make her doxastically justified in her coverage supported inference. This inference base is her belief that ‘if Canada had declared war, then she would have heard about it by now.’ Goldberg calls coverage-supported inference bases of this kind, truth-to-testimony conditionals. They take the general form of ‘if ~p were true, then I would have heard about it by now.’

In Victoria's case, the truth-to-testimony conditional is true. But a subject can be doxastically justified in her coverage-supported inference even if the relevant truth-to-testimony conditional, that serves as an inference base, is false. A false truth-to-testimony-conditional-inference base can confer doxastic justification if the processes that compose a subject's community are reliable enough. Suppose that in Victoria's case that her sources such as the New York Times could not report on whether foreign states have declared war on the US for the hour when she considered whether Canada had declared war on the US. By Goldberg's lights, Victoria is doxastically justified because her belief in the truth-to-testimony conditional would often enough be true despite its falsity in this supposed instant. As a result, Victoria's justified true coverage-supported inference does not count as knowledge.

Suppose that I infer that Canada has not declared war on the US partly on the basis of my belief that ‘if Canada had declared war on the US, then I would have heard about it by now.’ But in this case, Canada has actually declared war on the US. By Goldberg's lights, I am doxastically justified in my false coverage-supported inference if my belief in the truth-to-testimony-conditional-inference base is reliably true. This belief is reliably true if I properly sense the social conditions that make it likely that my sources apprise me of information in the relevant domain.

1.2. Features of coverage-reliance ignorance

Coverage-reliance ignorance obtains if (a) a subject infers either a false belief or a doxastically unjustified true belief because (b) she bases her inference in a false belief that sources reliably apprise her of information in a domain in a timely fashion where (c) injustice causes her sources to unreliably apprise her of information in this domain.

I understand injustice to involve asymmetrical power relations between a dominant group and at least one non-dominant group in a given society. Racial injustice, gender injustice, injustice that involves sexual orientation and disability are examples of injustice as I understand it. I take US society as a paradigm case of a society where racial injustice obtains (Omi and Winant Reference Omi and Winant1994; Mills Reference Mills1997; Taylor Reference Taylor2013).

Coverage-reliance ignorance is an epistemically bad case of coverage-supported belief because racial injustice causes, say, Jim's news sources to unreliably report race-and-racial-injustice-related information. If Jose bases his inference that the New York Yankees have not moved to Hartford Connecticut on the belief that the New York Times reliably apprises him of information like this and the New York Times has not reported that the New York Yankees have moved to Hartford, then he is justified in his inference that the Yankees have not moved to Hartford. The absence of a report from the New York Times and his justified belief in the relevant truth-to-testimony conditional justifies him in his inference that it is not the case that the Yankees have moved (Pedersen and Kallestrup Reference Pedersen and Kallestrup2013). This case is epistemically good partly because the New York Times would and does reliably report on matters like these such that the absence of a report justifies him.

Jose's and Jim's cases epistemically differ because Jim bases his inference in an unjustified false belief about his sources’ reliable coverage of racial-injustice-related topics and Jose bases his inference in a justified true belief about his sources’ reliable coverage of sports-related topics. But Jim's case is epistemically bad because of racial injustice. Racial injustice makes Jim's belief about his source's reliability false and unjustified because racial injustice makes his sources unreliable and him insensitive.

1.3. Unjustified true belief case of coverage-reliance ignorance

In the Detroit Police case, Jim infers a false belief at least partly because of racial injustice. But a subject can also infer an unjustified true belief because of racial injustice. Take the following case.

2.1. News organizations qua information channels

News organizations rely on reporters and editors reliably perceiving speakers’ trustworthiness to reliably report in a domain because news organizations function as information channels. An information channel is a medium through which a signal carries information (Dretske Reference Dretske1981). An information channel is reliable if enough of the signals that enter it are transmitted. So, if a news organization functions as a reliable information channel, then this news organization must be composed of processes that reliably transmit information through it.

Following Fred Dretske, Sanford Goldberg roughly defines reliable information transmission in this way:

On this view of reliable information transmission, information is reliably transmitted through a news organization only if a news organization would not have reported that the New York Yankees beat the Boston Red Sox unless the New York Yankees actually beat the Boston Red Sox. Here, the report that the Yankees beat the Red Sox is the signal that is carried through a news organization qua information channel.

News organizations such as the New York Times function as information channels partly because of reporters and editors who instantiate reliable processes and methods. News organization editors and reporters act as gatekeepers who follow methods to reliably transmit information through their newspapers qua information channels. Alvin Goldman characterizes gatekeeping in these ways:

By Goldman's lights, a gatekeeper can be veritistically assessed in terms of whether her practices lead to her audience holding more true beliefs and fewer false beliefs (Goldman Reference Goldman1999: 189). On assessing gatekeepers, Goldman says:

A reporter or editor performs her gatekeeping function veritistically well if she employs practices that elevate the likelihood of true belief and depress the likelihood of false belief among her readers. I assume that practices can include following rules and methods.

John Greco suggests that norms that govern gatekeeping are (1) that information should be let into a social system and (2) that false information should be kept from entering into or continuing to pass through a social system (Greco Reference Greco2016: 492, Reference Greco2017).

In this section, I have argued that a gatekeeper who follows these two gatekeeping norms promotes her information channel's reliability. I have provided the following reasons to support this claim. Reporters and editors are gatekeepers. Information channels are social systems. And these two gatekeeping norms promote the likelihood that information is transmitted through an information channel. The basic idea here is that reporters and editors qua gatekeepers can follow rules or methods that aim to funnel information into the channel and prevent bad information from entering or continuing to pass through the channel (Greco Reference Greco2016, Reference Greco2017).

2.2. Racial prejudice and information transmission

Racial prejudice can make a news organization unreliably transmit information because it can make hearers less reliably perceive speakers’ trustworthiness. A news reporter may improperly take a trustworthy news source as untrustworthy and thus not allow this trustworthy source's information into her news organization qua information channel such that the source reports it. I now defend the claim that racial prejudice can make a news organization qua information channel transmit information unreliably. The following two reasons support this claim. First, racial prejudice can affect whether a reporter allows information into her news organization qua information channel. Second, allowing information into a channel partly constitutes reliable information transmission.

Take the Detroit Police Case. Jim relies on local and national news organizations to reliably report on matters such as whether the Detroit police unjustly treat black folks relative to white folks. Jim's commitment to the truth of this belief involves a commitment to the view that his newspapers’ reporters will likely trust speakers who are actually trustworthy and not trust speakers who are untrustworthy.

Hearers who harbor racial prejudice will tend to perceive, say, black and Latinx speakers as untrustworthy even though they are actually trustworthy (Fricker Reference Fricker2007). Hearers who harbor racial prejudices will tend to misperceive black and Latino speakers this way because racial prejudice is a kind of identity prejudice and identity prejudice can affect hearers’ perceptions of speakers in a subpersonal way (Fricker Reference Fricker2007).

In this case, if a reporter for, say, the Detroit Free Press harbors a racial prejudice, racial prejudice can cause reporters to misperceive trustworthy speakers as untrustworthy and a non-white speaker's attempt to transmit that the Detroit police unjustly treat black folks to a reporter is likely unsuccessful because the reporter likely commits testimonial injustice, then racial prejudice makes the news organizations qua information channel that Jim relies on unreliable. Racial prejudice makes these news organizations unreliable because racial prejudice depresses the likelihood that this speaker will transmit this information to the reporter who could publish this information so that Jim can receive it in a newspaper. So, racial prejudice can make a news organization unreliable because news reporters are constituents of news organizations where racial prejudice influences news reporters’ ability to reliably perceive speaker trustworthiness.

2.3. Racial prejudice and racial injustice information

I now argue that non-dominant-racial-group speakers who are most likely to have information about racial injustice are less likely to have this information reported by news organizations because non-dominant-racial group speakers are less likely to successfully convey information about racial injustice to reporters. I provide three supporting reasons. The first reason is that speakers who are most likely to have information about racial injustice are members of non-dominant racial groups. The second reason is that racial prejudices in societies with entrenched racial injustice tend to be about non-dominant racial groups. And the third reason is that negative racial prejudices reduce the likelihood that non-dominant group speakers successfully convey information about racial injustice.

Non-dominant racial groups such as blacks, Latinx folks and Native Americans are more likely than whites to have information about racial injustice because of their social locations (Harding Reference Harding, Alcoff and Potter1993, Reference Harding2015). Social locations can include any identity dimension such as sexual orientation, economic class or religion. Non-dominant social locations are positions from which subjects are more likely to hypothesize or pose questions that lead to information in certain domains (Harding Reference Harding, Alcoff and Potter1993, Reference Harding2015). Whether an archer can hit her target can depend on whether she is located too far from the target or whether her location is too foggy or windy. Similarly, a social location can differ from others as a location from which to acquire a true belief.

If a group of non-dominant racial subjects experience the effects of racial injustice, then they may be more likely to float hypotheses or ask questions that answer racial injustice related questions or queries than dominant racial group subjects who have likely only experienced the benefits and few of the bad effects of racial injustice. For Nancy Hartsock, women's role in reproduction and caring for family members provides them with a standpoint such that women are more likely than men to know about the nature of oppression in a patriarchal society (Hartsock Reference Hartsock and Harding1987). For Patricia Hill Collins, black women's experience of both racism and sexism elevates the likelihood that black women will understand or know about how racism and sexism function in society (Collins Reference Collins1990). By Linda Martín Alcoff's lights, a subject's identity can bear on whether she accurately believes in certain domains of information and thus proper credibility judgements of speakers should involve identity as a criterion (Alcoff Reference Alcoff1999). According to W.E.B. DuBois, black Americans experience a double consciousness that involves seeing the world from their perspective and seeing the world from a white perspective and as a result black Americans have more knowledge or deeper understanding of how America works and the role that race plays in America (Taylor Reference Taylor2016: 39). A common feature that these views have is that a subject's situation increases the likelihood that she knows or understands about injustice or oppression in her society.

Negative racial prejudices tend to prevent non-dominant group speakers from successfully conveying racial-injustice information partly because if they are widely held, then not only will testimonial injustice likely widely obtain, but testimonial quieting and testimonial smothering will also likely widely obtain (Dotson Reference Dotson2011). Suppose a black speaker, Michael, tells a reporter, from a newspaper that Jim reads, that the Detroit police unjustly treat black people relative to white people. But this reporter commits testimonial injustice against Michael due to an anti-black racial prejudice she bears. As a result, this reporter does not consider investigating whether black people are unjustly treated in this way. Here the likelihood is that racial injustice information transmission is depressed because of racial prejudice. And here racial prejudice makes it less likely that a non-dominant speaker successfully conveys information about racial injustice to hearers in general and to newspaper reporters in particular. Michael is more likely to have information about racial injustice because of his social location qua black man. Here a non-dominant speaker who is in a better epistemic position to have information in a domain is less likely to successfully convey it because of racial prejudice.

3.1. Racial injustice and a society's unjust structure

The legacy of slavery, Jim Crow and discriminatory hiring practices in the US shape the structure of US society today in ways that negatively affect socio-economic outcomes for blacks, Latinos and Native Americans (Omi and Winant Reference Omi and Winant1994; Mills Reference Mills1997; Taylor Reference Taylor2013). Charles Mills points to the following to justify the claim that US society is structured to favor whites:

This only concerns blacks qua non-dominant-racial group in the US. It does not concern the value diverted from non-dominant groups such as Latinos and Native Americans. If the ways that income and opportunities have been diverted from these groups to whites is considered, then the idea that US society is structured in favor of whites should at least seem plausible. I assume, rather than argue that if racial injustice obtains in a society, then it will likely have an unjust structure, because entire fields of inquiry such as Black Studies, Africana Studies, Latino Studies, Native American studies and parts of entire disciplines such as sociology and history explain the ways these structures have affected people in the past and continue to affect people today.

3.2. Racial injustice information and a society's unjust structure

A society's unjust structure can make news organizations that are more likely to report on racial injustice related topics less likely to reach subjects who lack information about racial injustice. If non-dominant-controlled-news sources are more likely to report on racial injustice related topics, news sources that lack resources such as capital, loans and proper funding are less likely to successfully convey their information to recipients who lack it, a society's unjust structure makes it less likely that non-dominant-group news sources have these resources, then a society's unjust structure makes reports from news sources that are more likely to report racial injustice information less likely to be received by subjects who lack this information.

New York City's first black-owned and edited newspaper, Freedom's Journal, was more likely than its New York City counterparts to publish on issues that took up black experience and racial injustice (Gonzalez and Torres Reference Gonzalez and Torres2011). Freedom's Journal was in circulation from 1827 to 1829 (Gonzalez and Torres Reference Gonzalez and Torres2011). During this time the Freedom's Journal rebuffed falsehoods reported by the mainstream press in New York City. One example of this is when Freedom's Journal provided counterevidence to an article that the New-York Evening Post published that claimed that “the condition of blacks on Southern plantations, despite some ‘occasional exceptions,’ was one of contentment, of gaiety and happiness” (Gonzalez and Torres Reference Gonzalez and Torres2011: 110). The New-York Evening Post went so far as to claim that “the master-slave relationship [was] ‘one of mutual attachment’” (Gonzalez and Torres Reference Gonzalez and Torres2011: 110).

In 1892, Ida B. Wells as editor and part owner of Memphis’ Free Speech periodical published an article that offered counterevidence and counterargument to the false, but widely held, belief that many of the black men who were lynched for raping white women had indeed raped white women (Gonzalez and Torres Reference Gonzalez and Torres2011; Chideya Reference Chideya2018). Wells suggested that these black men were lynched not because they forced themselves on white women, but rather that they were lynched at least partly because white women willingly engaged or wanted to engage in relationships with them (Gonzalez and Torres Reference Gonzalez and Torres2011). Both Free Speech and New York's Freedom's Journal were more likely to uncover and publicize facts about racial-injustice than mainstream newspapers. Both of these periodicals had less funding or capital than their white-owned or mainstream counterparts (Gonzalez and Torres Reference Gonzalez and Torres2011).

If a news source has poorer distribution, advertising, printing material and insufficient staff and these things contribute to a news source's reports more likely reaching subjects, then this news source's reports are less likely read by subjects. If a newspaper has poorer distribution than other newspapers it will reach fewer subjects than these other newspapers because it will be available at fewer venues such as newsstands. Here subjects are not as widely privy to the information that newspapers with poor distribution carry. And even if a subject happens to come across a newspaper that is poorly distributed at a newsstand that carries it, this subject may improperly infer that this newspaper is not epistemically as good as others because it has poor distribution.

Take the Detroit Police Case. Suppose Jim comes across a newspaper that is disposed to report that the Detroit police unjustly treat black people relative to how they treat white people, he may not pick up this newspaper because he infers that if this were a good news source, then he would have seen it at other newsstands. Here, Jim associates wide distribution with epistemic goodness and on this basis he infers that newspapers with good distribution are epistemically better than newspapers with poor distribution. But, of course, newspapers with poor distribution can be epistemically better than newspapers with good distribution

If a newspaper has poorer advertising than other newspapers, then fewer subjects will either attempt to secure information from this newspaper or know that it is even a source from which they could secure information (Wong Reference Wong, Rosengren, Dahlen and Okazaki2013). If a subject does not know that a newspaper exists, then she cannot seek it out for the information it has. And if a subject comes across a newspaper that she has not heard of because it is poorly advertised, then she may infer that it is not epistemically good because either she has not heard that it is an epistemically good newspaper from a reliable source or that newspapers that are poorly advertised likely do not have sufficient funds to provide good news coverage. The basic idea here is that a subject may not pick up an epistemically good newspaper because she is not familiar with it, due to its poor advertising.

If a newspaper's printing materials are poorer than other newspapers, then subjects may improperly take the newspaper as less epistemically good than newspapers that are printed on better materials because subjects may infer that if a newspaper has insufficient resources to print on better materials, then it likely has insufficient resources to provide good news coverage (Marlow and Jansson-Boyd Reference Marlow and Jansson-Boyd2011; Hampel et al. Reference Hampel, Heinrich and Campbell2012; Langer et al. Reference Langer, Fischer, Brune, Rosengren, Dahlen and Okazaki2013). Suppose that Jim sees a newspaper printed in black and white on rough paper and next to it he sees a newspaper printed in color on high quality paper. Jim is likely attracted to the newspaper printed in color on high quality paper even if the newspaper printed in black and white on rough paper has more relevant information or even less false information.

If a newspaper has insufficient staff and it is disposed to report on racial-injustice-related topics, then it may not report on certain racial-injustice-related topics because it has to prioritize its limited staff on some stories rather than others. Suppose that Jim often picks up a newspaper at his newsstand that is disposed to cover racial-injustice-related information, but this newspaper had to prioritize covering city government's decision to underfund education and housing in primarily black neighborhoods over covering how Detroit police are disproportionately pulling over black drivers relative to white drivers. Here one of Jim's news sources does not apprise him of this information because it has insufficient staff.

So far, I have focused on how a society's unjust structure can depress the likelihood that a non-white news organization successfully informs subjects of racial injustice information. But a society's unjust structure can also depress the likelihood that non-white reporters successfully convey racial injustice information to subjects who lack this information.

The structure of US society negatively affects whether non-white reporters in major news organizations such as the New York Times can successfully convey this information because non-white reporters are often underrepresented in major news organizations and the few non-whites who are staff members of such organizations often find that they are not in a position to convey such information. One glaring example of underrepresentation (Frissell et al. Reference Frissell, Ibrahim, Raghavendran and Yang2017) is that despite improvement across the journalism industry since the 1960s (Sylvie Reference Sylvie, Lowrey and Gade2011), in 2016, the New York Times’ political reporting team was composed of 90% whites, 10% blacks and no Latinx or Asians (Chideya Reference Chideya2018). But news organization management often assigned, say, black reporters to cover what management believed were “black” topics (Somani and Hopkinson Reference Somani and Hopkinson2018). And even those few black reporters who are not assigned to such positions, often report on issues of race in ways that are consistent with how major new organizations have tended to report in the domain of racial injustice information (Robinson and Batzen Culver Reference Robinson and Batzen Culver2016).

I have defended the claim that a society's structure can depress the likelihood that subjects will read black-owned news sources. I have provided four supporting reasons for this claim. The first reason is that black-owned news sources have poorer access to capital than white news sources. The second reason is that poorer access to capital makes it less likely that such a newspaper will have good distribution, advertising, printing materials and sufficient staff. The third reason is that society's unjust structure makes it more likely that black newspapers have poorer access to capital than white newspapers. And the fourth is that good distribution, printing material, advertising and sufficient staff make it more likely that a subject will read a news source.

I have also defended the claim that society's unjust structure reduces the likelihood that white subjects’ available news sources will apprise them of information about racial injustice. I have discussed three reasons that support this claim. The first reason is that white subjects are less likely to have information about racial injustice. The second reason is that black subjects are more likely to have information about racial injustice. The third reason is that a society's unjust structure makes it less likely that subjects receive reports from black-owned newspapers.

3.3. Racial injustice's disjunctive effect

So far, I have provided two ways that racial injustice can make news or information sources fail to provide reliable coverage for a subject in the race and racial injustice domain. Racial prejudice is the first way and a society's unjust structure is the second way. In the US, racial injustice affects coverage reliability, or causes coverage-reliance ignorance, simultaneously through both causal routes. But, analytically, racial injustice need only cause this through one route for coverage-reliance ignorance to obtain. Racial injustice can cause a subject to be in a coverage-reliance ignorant state either because of (1) racial prejudice or (2) a society's unjust structure even if, as a matter of empirical fact, both (1) and (2) tend to simultaneously obtain.

If racial prejudice makes a subject's news sources unreliable because of how their constituent parts, reporters and editors, reliably misperceive speakers’ trustworthiness, then this subject's sources will not provide reliable coverage in this domain even if the structure of society does not affect information dissemination or delivery. The basic idea here is that even if a source would reliably transmit information in a domain to subjects who lack it, reliable transmission, and thus reliable coverage, requires that a source actually reliably uncover information in this domain.

Similarly, if a society's unjust structure makes a subject's news sources unreliably apprise her of information in the race-and-racial-injustice domain because this structure depresses the likelihood that her sources’ reports reach her, then these sources will likely not provide reliable coverage in this domain even if these sources reliably uncover this information. The idea here is that even if a source reliably uncovers information in this domain, reliable coverage involves that a subject's sources’ reports actually reach her.

That either widespread racial prejudice or a society's unjust structure can cause coverage-reliance ignorance to obtain matters because even if society becomes less racially unjust along either of these dimensions, coverage-reliance ignorance could still obtain. So, even if US society seems much better in terms of, say, society's structure disfavoring, say, black and Latinx folks, racial injustice may still cause these bad epistemic outcomes.

3.4. Other causal explanations

I have so far focused on how racial injustice diminishes news sources’ reliability in the race-and-racial-injustice-information domain through racial prejudice and its unjust structuring of society. These causal explanations differ because one affects subjects’ mental attitudes and the other affects the structure of society that subjects must negotiate.

I have focused on testimonial injustice, quieting and smothering as part of explanations that involve racial injustice's effect on subjects’ mental attitudes. But, unfortunately, there are other such explanations that involve subjects’ mental attitudes. Some of these are that journalists will tend not to inquire into racial injustice related areas because such stories challenge their preferred worldview and that journalists and editors decide to exclude stories about racial injustice because such stories do not rise to the level of what they consider newsworthy (Hall Reference Hall and Hall1997; Dixon and Williams Reference Dixon and Williams2015; Dixon Reference Dixon2017; Frissell et al. Reference Frissell, Ibrahim, Raghavendran and Yang2017; Somani and Hopkinson Reference Somani and Hopkinson2018).

I have focused on how racial injustice unjustly structures society in ways that disfavor non-white news sources. But how society is unjustly structured can affect whether news sources provide subjects with reliable coverage in the domain of racial-injustice information. Some of these are that non-white reporters will not have the social connections that white reporters have such that they can successfully follow up on leads and that non-white journalists will tend to be assigned to columns or beats that only cover ‘race and culture’ which readers take less seriously and whites will tend to avoid.

4.1. News sources and proper sensitivity

I now defend the claim that racial injustice depresses the odds that a subject properly senses these social conditions because a primary way to be sensitive to these conditions is undermined by racial injustice. Three reasons support this claim. The first reason is that a doxastically justified subject is properly sensitive to the social conditions that make it likely that she is reliably apprised of racial-injustice-related information. The second reason is that a primary way to be sensitive to these social conditions is through sensitivity to reports from news sources. The third reason is that racial injustice undermines the reliability of these news sources. If there is more information for a subject to be sensitive to, then she will find it easier to properly calibrate her expectations of her source's coverage reliability to her sources’ likely coverage reliability such that she is justified. Similarly, if there is less information for a subject to be sensitive to, then she will find it more difficult to properly calibrate these expectations to her sources’ likely coverage reliability. So, racial injustice itself undermines a primary way for a subject to be sensitive to whether her news sources reliably report racial injustice information.

Racial injustice undermines what Lauren Woomer calls epistemic tools (Woomer Reference Woomer2019). Subjects use epistemic tools to gain “an accurate view of their environment” (Woomer Reference Woomer2019: 77). Access tools are epistemic tools that subjects use to access evidence. Examples of access tools are subjects’ capacities to attend, store information and receive testimony from speakers. Interpretative tools are epistemic tools that subjects use to interpret evidence. Examples of interpretative tools are collectively shared conceptual resources, evidence standards, ways of reasoning and investigative methods.

Racial injustice undermines access tools in two ways. If news sources are a kind of testimony and testimony is an access tool, then racial injustice undermines subjects’ access to an access tool because racial injustice diminishes the likelihood that subjects actually come to receive news in the domain of race and racial injustice. That is racial injustice depresses the likelihood that news that contains information about race and racial injustice reach subjects. This is the first way that racial injustice undermines access tools.

But racial injustice not only depresses the likelihood that subjects receive news sources with racial injustice and race information, it also depresses the likelihood that the news sources they do receive actually report such information. This is the second way that racial injustice undermines access tools.

By Woomer's lights, a subject is agentially insensitive if she does not use her epistemic tools such that her beliefs accord with her evidence. She explains white ignorance in terms of agent insensitivity (Woomer Reference Woomer2019). The explanation of coverage-reliance ignorance that I proffer complements her explanation of white ignorance because if testimony regarding racial injustice from news organizations is undermined by racial injustice, then the likelihood that a subject is a white ignorant state will be higher.

In the Native American News Case, one of the primary ways that Thomas can be sensitive to whether the social conditions obtain such that his news sources reliably report information about racial-injustice-related topics is to be sensitive to reports from these same news sources. So, Thomas in part relies on information from sources that are unreliable for information about these sources’ coverage reliability. But, of course, if these sources are unreliable, then they may not provide information for him to be sensitive to such that his expectations about coverage reliability in this domain are properly calibrated. The basic idea here is that racial injustice is a social phenomenon that obtains at a society-wide scale such that it is difficult for subjects to be sensitive to it without relying on news sources that can inform them about the world beyond the reach of their individual perceptual capacities.

4.2. Testimonial injustice, testimonial quieting and proper sensitivity

A subject is less likely properly sensitive to whether these social conditions obtain because non-dominant speakers who are more likely to have this information are likely either taken as not trustworthy or not recognized as knowers because of racial prejudice. One reason that supports this claim is that a doxastically justified subject's sensitivity involves sensitivity to information from speakers about whether the social conditions obtain that make it likely that she is reliably apprised of information on racial injustice. A second reason is that this subject likely commits either testimonial injustice or quieting against non-white speakers who would convey information that these social conditions do not obtain. And a third reason is that these non-dominant speakers are more likely to have information about whether these social conditions obtain.

4.3. Racial injustice and dominant-group psychology

White subjects are unlikely properly sensitive to information that concerns the social conditions that make it likely that her news sources reliably apprise her of racial-injustice-related topics. One reason that supports this claim is that a doxastically justified subject's sensitivity involves sensitivity to information from speakers about whether the social conditions obtain that make it likely that she is reliably apprised of information regarding racial injustice. A second reason is that dominant group subjects are unlikely to revise their initial beliefs regarding a domain. A third reason is that dominant-group subjects are unlikely to believe information that conflicts with their dominant-group status. A fourth reason is that dominant-group subjects form beliefs that support their dominant status at an early age. Here, I assume that the set of beliefs that a subject learns early in childhood overlaps with the set of beliefs that she has first come to hold in a domain.

Suppose Thomas encountered well-supported information that newspapers such as the Pennsylvania Gazette are not likely to report instances of Europeans attacking Native American villages because of racial injustice's effect on what Colonial-American newspapers publish. And suppose Thomas was taught at an early age that proper Anglo-Christian Americans would publicly address wrongs done to any of God's children where this includes Native Americans. Thomas would find it difficult to believe this new well-supported information he encountered because it conflicts with a belief that he acquired at an early age. The belief perseverance effect predicts that Thomas would find it difficult to believe this new information (Nisbett and Ross Reference Nisbett and Ross1980). The belief perseverance effect is a tendency that subjects have where they stubbornly retain beliefs that are their first beliefs, or some of their earliest beliefs, on a topic (Nisbett and Ross Reference Nisbett and Ross1980). That is, subjects’ initial beliefs regarding a domain tend to be stubbornly resistant to counterevidence in comparison to beliefs in the same domain formed much later.

Another psychological feature of subjects that predicts that Thomas would find it difficult to believe this well-supported information is ‘identity protective cognition’ (Stanley Reference Stanley2015: 230). “[Identity] protective cognition is motivated reasoning with the goal of ‘affirming one's membership in an important reference group’” (Stanley Reference Stanley2015: 230). According to this psychological feature of subjects, white subjects are unlikely receptive to information that conflicts with their dominant-group status because they want to protect this status. Similarly, social psychological research on dissonance suggests subjects’ reception of “disconfirming evidence … actually causes discomfort” (Elliot and Devine Reference Elliot and Devine1994; Mandelbaum Reference Mandelbaum2018: 11). And, by Eric Mandelbaum's lights, the social psychology literature suggests that if a subject encounters evidence that either is inconsistent with a belief that they identify with or threatens their positive sense of self, then not only will they reject this evidence but they may strengthen their confidence in the belief that they identify with (Mandelbaum Reference Mandelbaum2016, Reference Mandelbaum2018). The basic idea here is that subjects have a psychological mechanism that either aims to preserve a subject's positive self-conception or aims to avoid the discomfort that is associated with believing negative things about oneself. And if this psychological mechanism obtains, then Thomas is unlikely sensitive to this information about Europeans attacking Native Americans that either undermines his positive self-conception or is likely to cause him to believe something negative about himself.

4.4. A generality problem variant

Someone could object that whether a subject is doxastically justified in his coverage-supported inference depends on how broadly he takes a news source's reliability-information domain. Take Jim's case. Jim infers that the Detroit police do not unjustly pull over black folks while driving (~D) on the basis of (1) an absence of a report that D and (2) that if D were true, then he would have heard about it by now from the Detroit Free Press. Jim is justified in believing (2) if he is sensitive to the reliability of the Detroit Free Press’ reports in the domain of racial injustice information. And suppose that Jim falsely infers here. So, Jim infers a false and unjustified belief because he is not justified in believing (2) which serves as his inference base.

According to this objection, if the information domain that Jim is sensitive to is the information domain of news in general, then Jim will infer a false but justified belief. This inference is justified because the Detroit Free Press is generally reliable regarding information in the general-news-information domain even though it isn't reliable in the racial injustice domain. If the general news information domain does not largely consist in racial injustice information, then the Detroit Free Press can reliably report in the general news information domain even if they are unreliable regarding racial injustice information.

And that Jim properly senses the Detroit Free Press’ reliability in this domain makes him doxastically justified in his inference. Here racial injustice does not undermine Jim's doxastic justification. Seemingly this is in tension with the central claim I defend, namely that racial injustice undermines subjects’ capacity to be doxastically justified in beliefs in the information domain of racial injustice.

Someone could rebut that racial injustice information may make up the majority of newsworthy information in the domain of general news information in locales with pervasive and persistent racial injustice such as Detroit. So, if racial injustice properly composes most of the information in the general news domain, then Jim would not be doxastically justified in his false inference. In such a case, the objection fails because racial injustice does undermine his capacity to be doxastically justified in racial injustice information domain inferences.

Someone could also object that Goldberg's account of coverage-supported inference justification faces a variant of the generality problem. The generality problem is a problem that at least reliabilist theories of justification face (Goldman Reference Goldman and Pappas1979; Bishop Reference Bishop2010; Kampa Reference Kampa2018; Lyons Reference Lyons2019). This problem is that reliabilist theories of justification do not offer any principled way to determine which process type is causally relevant when assessing whether a particular belief is reliably formed. If one cannot determine a belief's causally relevant process type, then one cannot evaluate whether a subject is justified because according to reliabilism a subject is justified if and only if a process type that reliably produces true beliefs causes her belief.

Coverage-supported inference justification faces a variant of the generality problem rather than the traditional generality problem because if one evaluates whether a subject is doxastically justified in her coverage-supported inference, one must have a principled way to determine a coverage-supported inference's relevant information domain. This matters because, by Goldberg's lights, whether a subject is doxastically justified in her coverage-supported inference depends on whether she is properly sensitive to her information source's coverage reliability in a particular information domain. And this matters for my central claim because if an evaluator cannot determine a coverage-supported inference's relevant domain, then it is unclear how racial injustice can undermine a subject's justification because it will be unclear whether racial injustice undermines a subject's sensitivity to this relevant domain.

Someone could rebut that this generality problem variant does not present an issue for my claim because this variant is an instance of meta-blindness (Medina Reference Medina2013a, Reference Medina2013b). By Jose Medina's lights, meta-blindness obtains if a subject is blind to her own blindness. Jurors who falsely judge a black witness as not credible exhibit meta-blindness if they are unaware that they could render such a judgement (Medina Reference Medina2013a, Reference Medina2013b). On the other hand, a juror that senses that they might issue such a credibility judgement is not meta-blind.

Consider the case where Jim seems doxastically justified because he is sensitive to the Detroit Free Press’ reliability in the domain of general news information. Here what seems like justification-conferring sensitivity is actually an instance of meta-blindness because Jim cannot sense that his news source could be unreliable on racial injustice information. Put differently, Jim exhibits meta-blindness because he is unaware that there could even be an epistemic problem regarding coverage-supported inferences. If Jim is meta-blind, racial injustice causes this meta-blindness and this meta-blindness diminishes Jim's capacity to be doxastically justified in coverage-supported inferences, then racial injustice contributes to the undermining of Jim's capacity to be doxastically justified. Ipso facto, that there is a generality problem variant does not defeat my central claim. Regarding coverage-supported inferences in the domain of racial injustice information, I point to this generality problem variant to highlight a way that racial injustice diminishes subjects’ capacity to be doxastically justified.