Doxastic Justification and Testimonial Beliefs

Abstract

I argue that a general feature of human psychology provides strong reason to modify or reject anti-reductionism about the epistemology of testimony. Because of the work of what I call “the background” (which is a collection of all of an individual's synthetizations, summarizations, memories of experiences, beliefs, etc.) we cannot help but form testimonial beliefs on the basis of a testifier's say so along with additional evidence, concepts, beliefs, and so on. Given that we arrive at testimonial beliefs through the work of the background, to be justified in holding a testimonial belief, we must not only have a rational speaker's say so, but we must also form such beliefs in a right way. If this is right, then, contrary to typical anti-reductionism, justified testimonial beliefs require more than just a trustworthy testifier's say so – another requirement is that they are formed in a right way.

1. Introduction

You are sitting in your usual corner seat at the local coffee shop where you like to get work done. You like the hustle and bustle as background noise while you read and write. Today, as you are slogging your way through a particularly abstruse chapter of a book, you can't help but eavesdrop on a conversation between what appear to be two college students sitting at an adjacent table. You overhear one say to the other, “You missed a great game last night. The Jaguars came back in the fourth quarter to beat the Titans!”

What you overheard doesn't surprise you – you already knew that the Jaguars were playing the Titans last night, you knew that they were evenly matched, and you knew that playoff games often involve exciting twists, such as fourth-quarter comebacks. More generally, you already knew the meanings of the words that were used, you knew the basics of English grammar, you knew that sports fans tend to converse with one another after exciting games, you knew that it would be odd for a seeming friend to lie or be extremely mistaken about the results of a recent football game, and so on.

Suppose you believe what the college student has said. Are you justified in believing that the Jaguars came back in the fourth quarter to beat the Titans last night? If so, did you need to know all the other things (i.e., that they were playing last night, that they were evenly matched, that the English words mean such and such, etc.) to be justified in believing what you overheard, or was overhearing the assertion enough for you to be justified in so believing? If you are not justified, why not? Is it because the college student was not speaking directly to you? Is it because you have not checked ESPN to verify what you heard? If you are justified, and your background beliefs are unnecessary for you being justified, what role (if any) do your background beliefs play?

Obviously, the foregoing thought experiment raises several important questions in the epistemology of testimony. But consider a variation. Same seat, same coffee shop, same overheard conversation, but this time, upon hearing that the Jaguars came from behind in the fourth quarter to beat the Titans last night, you think to yourself, “If the Jaguars came from behind in the fourth quarter to beat the Titans last night, then that college student just asserted that the Jaguars came from behind in the fourth quarter to beat the Titans last night. But that college student just now asserted that very thing! So, the Jaguars came from behind in the fourth quarter to beat the Titans last night.” In this fashion, you acquire a belief in what you overheard. What about now? Are you justified in believing what you overheard? Does it matter that you affirmed the consequent on your way to acquiring a belief in what you overheard? More generally, does the way in which we arrive at beliefs matter for justification?

Many will give what they take to be the obvious answer: Yes! Of course it matters how you form beliefs! And I wholeheartedly agree. But, despite the widespread acceptance of the claim that the way in which we form beliefs matters for being justified in holding beliefs (and testimonial beliefs in particular), I argue that the way in which we form beliefs has mistakenly been either ignored or dismissed as a mere enabling condition in the epistemology of testimony. Once we appreciate the role of proper belief formation in the justification of testimonial beliefs, we can observe some important ramifications for anti-reductionism in the epistemology of testimony. In the next section, I give a selective overview of anti-reductionist positions, and I explain where my argument will fit in. In Section 3, I present my main argument. In Section 4, I distinguish my argument from recent similar arguments. Finally, in Section 5, I conclude by interacting with a few objections.

2. Situating the argument

In his article “Content Preservation,” Tyler Burge argues for a version of anti-reductionism. Burge's view is that, when a proposition is communicated in an intelligible way to a hearer from a rational source (i.e., the speaker), the hearer has a prima facie justification or entitlement to her newly formed testimonial belief. The reason for this is that rational communication preserves justification or entitlement. In other words, when a speaker asserts that p to a hearer, whatever justification the speaker has in believing that p carries over to the hearer such that the hearer is prima facie justified or entitled in believing that p.

Burge's view can be construed as a version of anti-reductionism about the epistemology of testimony in the sense that anti-reductionists typically hold that you are prima facie pro tanto justified in forming your testimonial belief just on the basis of a speaker's say so. More precisely, anti-reductionism about the epistemology of testimony is the position that when a trustworthy speaker, S, tells a hearer, H, that p, H is prima facie pro tanto justified in believing that p just on the basis of S's presentation-as-true that p. That is, the hearer's justification is grounded in the speaker's assertion that p and need not be grounded in anything else. Anti-reductionism does not hold that a speaker's say so (i.e., the presentation-as-true) is the only source of justification; rather, a speaker's say so provides a default prima facie pro tanto justification that can be bolstered by other sources of justification, thereby providing on-balance justification. As long as H does not have any undefeated defeaters that indicate that what S says is either false or unlikely to be true, S's say so alone is sufficient for H to be justified in believing that p (cf. Leonard 2021).

Different versions of anti-reductionism cash out the exact nature of what is required for a believer to be prima facie pro tanto justified in holding a testimonial belief in different ways. Some, such as Burge (1993, 2013), hold that you are justified in holding a testimonial belief when the content of that belief is intelligibly presented to you as true from a testifier who is justified in believing that to which she testifies, and you understand what the speaker says. This is because intelligible propositions presented as true preserve a prima facie pro tanto justification from a rational source – that is, the testifier (cf. Graham 2006). Others hold that testimony is a basic source of justification in the same way as perception, memory, and inference are basic sources of justification – barring undefeated defeaters, nothing is required in addition to the testimony you receive for you to be justified in believing what you hear (cf. Coady 1992). Yet others hold that you are prima facie pro tanto justified in holding a testimonial belief because of the assurance that the content of the belief is true that you receive from a trustworthy testifier who gives you such assurance (e.g., Hinchman 2014; Moran 2005). Then there are middle positions that could be relevant to anti-reductionism, such as Lackey's (2008) view that both the hearer and the speaker have a role to play in producing justification. Lackey is not a full-on anti-reductionist because she affirms that the speaker must provide the hearer with good reasons to believe what the speaker says. However, she is also not a full-on reductionist because she denies that the possession of positive, independent reasons to believe a speaker's say so is sufficient for justification.

What is common to all versions of anti-reductionism, however, is the sufficiency claim: that the speaker's say so is sufficient for the hearer to be prima facie pro tanto justified in holding the testimonial belief that results from the interaction. The speaker is, in one way or another, the sufficient source of the hearer's justification in holding a testimonial belief. That is, anti-reductionists hold not only that the hearer's justification does not need to be reduced to some other source of justification, but also that the speaker's assertion that p is sufficient for prima facie pro tanto justification. For instance, for Burge, as well as for other anti-reductionists, the justification the hearer has in believing that p can be just the speaker's intelligible assertion that p (i.e., the speaker need not present the hearer with reasons to believe that p). What anti-reductionists deny is that a hearer must have other reasons (even other epistemic reasons) to believe that p, in addition to a speaker's say so, to be justified in believing that p.

I argue that anti-reductionists generally have overlooked the fact that one must form a belief in a right way even to be prima facie pro tanto justified in believing that p. What I mean is that the way in which we reach a belief matters for default justification; if things go haywire while forming a belief, this precludes you being justified in holding the belief. Put differently, though there can be several ways of arriving at the same destination while driving on the roads, things are usually not the same way with beliefs – there are correct and incorrect ways of forming beliefs, and the incorrect ways preclude justification. So, in effect, I argue that the prima facie pro tanto justification for our testimonial beliefs always requires something more than just S's say so, contrary to the sufficiency claim of anti-reductionism. I build up to this claim in two steps. First, I argue that all the beliefs we acquire are the results of an automatic belief-forming process that produces new beliefs either through an inference or through some similar basing process. In doing so, I present something similar to recent work from David Henderson, Terrence Horgan, and Matjaž Potrč (e.g., Henderson and Horgan 2020; Henderson et al. 2020). Second, I argue that, given that all our testimonial beliefs are either inferred from or based upon what a speaker says along with background information, forming such beliefs in a right way is necessary for those beliefs to be justified. However, though I think that there are always several things besides what a speaker asserts that go into forming testimonial beliefs, I need not rely on this claim for my main point. Regardless of what goes into forming a testimonial belief, forming such a belief in the wrong way precludes prima facie pro tanto justification.

Note that the sort of justification I discuss is doxastic justification, rather than propositional justification. Though the difference between these two is sometimes contested, for the purposes of this paper I understand doxastic justification to refer to justifiedly believing that p (i.e., you are justified in holding a belief that you actually have), while propositional justification refers to the justificatory support the belief that p or the proposition, p, has, which we can refer to as “when believing that p is justified for you,” though you may or may not actually believe that p (cf. Silva and Oliveira forthcoming).

Both reductionism and anti-reductionism could be formulated either as theories of propositional justification or of doxastic justification. Further, it is unclear that one formulation would imply the other and vice versa, such as whether anti-reductionism formulated as a theory of propositional justification would imply anti-reductionism as a theory of doxastic justification and vice versa. My arguments in this paper should not be taken as arguments against anti-reductionism understood only as a theory of propositional justification. Rather, my arguments are aimed at the sufficiency claim of anti-reductionism about doxastic justification of testimonial beliefs.

Assuming a focus on doxastic justification of testimonial beliefs, I interpret the sufficiency claim of anti-reductionism to be a sufficiency claim about doxastic justification. That is, in this paper, anti-reductionism is not the view that the speaker's say so is sufficient for propositional justification (i.e., the proposition that the speaker testifies to, or the belief in what the speaker testifies to, has support), but rather it is the view that the speaker's say so is sufficient for doxastic justification (i.e., when the hearer believes that p on the basis of the speaker's say so, the hearer is justified in holding the belief that p). I contend that this sufficiency claim is false – though the speaker's say so is a necessary condition for doxastically justified testimonial beliefs, it is not sufficient; another necessary condition for doxastically justified testimonial beliefs is that they are formed in a right way.

If my argument is sound, then it gives us strong reason either to modify or to reject anti-reductionism about the epistemology of testimony, since, contrary to the sufficiency claim of anti-reductionism, the prima facie pro tanto justification of hearers holding testimonial beliefs always requires something more than just the speaker's say so. But though we should modify or reject anti-reductionism, my argument should not be taken as claiming that additional reasons (even epistemic reasons) are required to justify a hearer in holding testimonial beliefs. For my claim is not that further reasons are necessary for justification, or that whatever your background information provides is necessary for justification, but rather that testimonial beliefs must be formed in a right way for hearers to be justified in holding them. If the speaker's say so is not sufficient to provide prima facie pro tanto justification for hearers to hold testimonial beliefs, the sufficiency claim of anti-reductionism is false. I leave it open whether the falsity of the sufficiency claim means we should reject anti-reductionism. I acknowledge that some reductionists might take the resulting modified view to be a reductionist view rather than an anti-reductionist view. However, it is likely that there are others who would think that the resulting modified view does not go far enough toward reductionism.

In the next section, I present two claims: first, a weaker claim according to which we inevitably base testimonial beliefs upon what S says along with background information, and second a stronger claim according to which we inevitably infer testimonial beliefs from what S says along with background information. In the rest of the paper, I leave the “prima facie pro tanto” qualification implied unless otherwise indicated. Additionally, throughout the paper I use the term “say so” to capture the different ways anti-reductionists think testifiers/testimony ground our prima facie pro tanto justification.

3. Anti-reductionism and the work of the background

The weaker claim is that when a hearer, H, receives testimony from a speaker, S, that p, and when H comes to believe that p from S's testimony, H cannot help but base the belief that p both on S's testimony as well as on background information. The stronger claim is that H cannot help but infer the belief that p from S's testimony along with background information. The reason why I label these claims as “weaker” and “stronger” is that I take it that the notion of inference at work in the stronger claim is more controversial than the notion of basing at work in the weaker claim. Notice also that the stronger claim builds upon the weaker claim. That is, if the stronger claim is true, then so also is the weaker claim, because the basing relation holds between that from which you infer and that to which you infer (even if it is an irrational inference). However, there perhaps can be a basing relation between your belief that p and a reason (or reasons, or evidence) without you inferring to the belief that p.

What is common between the weaker and stronger claims is the idea that, when H forms a testimonial belief that p upon reception of S's testimony that p, there is a relation between H's belief that p not only with S's testimony that p but also with background information. We cannot help but form testimonial beliefs in this way. The reason for this is that we acquire all our beliefs through an automatic belief-forming mechanism that processes both sensory information as well as background information.

Think, for example, of an experience of a guitar string that is out of tune. Upon looking at the object in front of you, you do not merely perceive a silvery, metallic string stretched taut over a wooden structure; rather, you perceive a guitar string. Of course, you do receive certain wavelengths of light, which impinge upon the rods and cones in the retinas of your eyes, which translate the wavelengths into action potentials, which propagate to your occipital lobe, and so on. However, when this series of events occurs, you perceive the wavelengths of light as a guitar string. You are able to do so (in part) because you have relevant concepts such as “string,” “guitar,” “metal,” “color,” “silver,” and even more basic concepts that are “at work” during your sensory experience in the sense that they are cognitively accessible enough (though not necessarily conscious) to render these sensory data intelligible; that is, to translate the sensory data into the perception of a guitar string.

Similarly, when you pluck the guitar string, you receive air vibrations that jostle the cilia in your inner ear, which eventually leads to the translation of such vibrations into action potentials, and so on. When this series of events occurs, you perceive the sensory data as hearing a plucked guitar string, and, more than that, you hear the plucked guitar string as out of tune. You have certain background beliefs, concepts, and so on, that influence how the sensory data are processed and translate them into the perception of a plucked guitar string that is out of tune. In short, you need the background for your experience of the sensory data to be intelligible, and the background is constantly operative to provide such intelligibility.

Just as background information is “at work” in every instance of perception, so also it is continuously “at work” in the reception of testimony in the sense that the background functions as a filter through which testimony passes. But the meaning that the background imposes is not just at the level of rendering words and sentences intelligible; rather, the meaning/intelligibility imposed on testimony that we have in virtue of the work of the background involves situating those intelligible words and sentences within our more general understandings of the world, which include understandings of truth, reliability, and how people tend to interact with each other (cf. Kenyon 2013).

For example, suppose your friend is an avid baseball fan who can spout off pretty much any stat you could care to learn about. Since you are only mildly interested in baseball, you occasionally check in with your friend for your fix of baseball news, and her ability to inform you of the goings on in the baseball world has been uncanny. On one such occasion, you ask your friend how a certain player is performing this season and she tells you that he is batting 0.350. This surprises you, since you remember that this player was, at best, mediocre last season, ending the season with a batting average of 0.130. However, you understand that players can have good and bad seasons, and that 0.350, though remarkable, is within the typical range of batting averages, historically speaking. Your surprise notwithstanding, when your friend tells you that the player is batting 0.350 you believe that this is so.

Let's take a step back and assess this exchange of information between you and your friend. When your friend says, “[Player X] is having a great season! He's batting 0.350 and his team is heading to the playoffs,” you of course understand the meanings of the individual words as well as the meanings of the words collectively in the sentences. Just as the sensory data that you receive when hearing a plucked guitar string that is out of tune are intelligible because of the background, so also the sensory data that you receive when hearing your friend's utterance are intelligible because of the background. That is, you have certain stored concepts, experiences, memories, and so on that render intelligible your experience of your friend's utterance.

Background information does more than just connect meanings with the sounds that you hear. The background also situates your friend's utterance within an already existing cognitive framework, which includes your understanding of baseball, your understanding of your friend, of how people interact, and much more. In arriving at the belief that the player is batting 0.350 this season, you form this belief not just in virtue of hearing your friend's utterance. Rather, you form this belief in virtue of your friend's utterance along with your prior understandings of baseball, your friend's uncanny knowledge of baseball (which, as you remember either consciously or unconsciously, she has demonstrated to you many times before), how people tend to exchange information with each other, and so on.

To drive home the point, consider a variation of the example. Suppose that when you ask your friend about how the player is doing this season she says, “He is having a great season! He has a perfect batting average, and his team is heading to the playoffs.” You are taken aback with what your friend says. It is near-impossible for a player to maintain a perfect batting average through an entire regular season. Perhaps the player has not played the whole season, but only started playing a few days ago? Perhaps the player was traded from a team in the National League to a team in the American League, which reset his average? Or maybe your friend is joking with you? Whatever the explanation, your friend's utterance jolts you into beginning a search for an explanation of what she said (cf. Lipton 2007).

Now, you've never doubted her impressive knowledge of baseball before, and she has not even come close to misleading you about such a readily accessible fact about a baseball player. On other occasions when your friend has informed you about the goings on of the baseball season, she has given you information and you have accepted it with few to no questions asked. But this time it is different, and the reason why is that what she said did not mesh with other things that you were taking as given. Put differently, what your friend tells you about the player is in tension with some of your background information.

The background information was not necessarily something you were consciously thinking about as you were listening to your friend. However, it was cognitively accessible enough not only to render intelligible the words and sentences that your friend speaks, but also to provide the understanding that what she says was unusual, was inconsistent with other things you take as settled, was uncharacteristic of the types of information she gives you, and so on.

The variation of the example is an instance in which you may not form a testimonial belief from your interaction with your friend. For the surprise may prevent you from forming a belief about how the player is doing this season until you get some further questions answered. Nevertheless, the general point is that, whenever you receive testimony, the background information is at work to provide meaning/intelligibility to what S says. The meaning-giving function that the background information performs implies that, when we do form testimonial beliefs, we form them not just on S's say so, but also on the basis of our background information.

To be clear, I am not claiming that the background information is only a counterfactual sensitivity to signs of untrustworthiness, inaccuracy, and so on (cf. Goldberg and Henderson 2006). Instead, I am claiming that the background information always has a contribution to our forming testimonial beliefs – specifically, the background information's contribution is that it influences the content of the belief (e.g., whether we take the proposition to be true or false, among other things) and can provide further reasons for that belief, though I do not claim that it must provide further reasons. Even when we receive testimony that does not trigger doubts, and so more or less automatically form a testimonial belief on such testimony, the background information causally contributes to the formation of that testimonial belief.

I am also not claiming that everyone's background information always influences mental representations in the direction of accuracy. The mental representations we end up with as a result of the operation of the background information are dependent upon the types of concepts we have formed, the types of memories we have, and so on. In short, the influence of our background information depends on the constitution of our background information. If I do not have a certain concept, then that concept of course is not going to influence how my background information shapes new beliefs (for as long as I don't have that concept).

Furthermore, some parts of the background information could be more salient than others, and thus could have an outsized influence over new beliefs we form, similar to how anchoring and availability heuristics can influence judgments (cf. Kahneman 2011). So, the operation of the background information does not automatically confer even prima facie pro tanto justification on an agent holding a testimonial belief, though the reasons the background information can provide can confer some sort of justification for an agent holding a testimonial belief.¹

As mentioned, there are two possible claims we can glean from the foregoing picture of how we acquire testimonial beliefs. The weaker claim is that we base testimonial beliefs on S's say so along with relevant background concepts, experiences, memories, and so on. To say that you base a testimonial belief on S's say so and on background information is to say that S's say so and your background information are the (motivating or explanatory, not necessarily normative) reasons for which you hold the testimonial belief. When we base testimonial beliefs in this way, it need not involve any conscious-level processing. Rather, it need only be that the testimonial belief is caused by both what S says and the influence of the background information (cf. McCain 2012; Moser 1989; Turri 2011).

The stronger claim is that we infer to a testimonial belief that p from S's say so that p along with background information.² There are several different theories of inference that are compatible with the way in which I contend we form testimonial beliefs. For example, Hugo Mercier and Dan Sperber define inference as any cognitive process by which we extract information from information we already have (Mercier and Sperber 2017: 53). Other theories of inference are more demanding than Mercier and Sperber's. For example, many theories require that the movement from one or more mental attitudes to a new attitude must occur in accordance with some rule-governed process (e.g., Boghossian 2014; Broome 2014; Harman 1986; Neta 2013; Wright 2014).

It is beyond the scope of this paper to defend a particular theory of inference. Nevertheless, it is sufficient to note that the way in which a hearer moves from background information, such as the beliefs that q, r, and s, concepts t, u, and v, and so on, along with the belief that S has asserted that p, to the belief that p, will count as an inference on several accounts of what inference involves. Theories that would disqualify such a transition as counting as an inference are those that require that the movement from one or more beliefs to a new belief be a conscious-level activity (e.g., Brogaard 2020; Neta 2013). Yet, if a theory of inference that does not require the transition to be conscious is true, then it would follow that the way in which normal human agents form testimonial beliefs is to infer to a belief that p from S's say so that p along with background information.

4. Distinguishing from other views

In this section, I present the main details of the arguments given in Henderson et al. (2020) and Henderson and Horgan (2020), which are similar but importantly distinct from my argument. (Note that these authors build upon the “iceberg epistemology” presented in Henderson and Horgan (2000).)

According to Henderson et al., when it comes to belief fixation and belief maintenance, the cognitive processing that is accessible to conscious awareness is not the whole story – there is other cognitive processing that occurs “below” conscious awareness, but of which (at least some of which) we can nevertheless be implicitly, phenomenologically aware (the authors call this “coloring” of conscious experience by such cognitive processing “chromatic illumination” (Henderson et al. 2020: 241)).

Much, if not all, of the cognitive processing that occurs “below” conscious awareness utilizes what the authors call morphological content (Ibid.: 237–38), which I refer to as “the background information.” Morphological content, or background information, refers to concepts, memories, skills, sensitivities, and other information that we most often are not concurrently, consciously aware of, but which interact with new thoughts and experiences in some way and provides meaning/intelligibility to all new thoughts, experiences, and so on. In other words, background information functions as a filter through which new thoughts and experiences pass, whereby they are made intelligible. When it comes to testimonial beliefs (as well as other sorts of beliefs, such as perceptual beliefs), these are abductively inferential (Ibid.: 234) meaning that they are caused by both the conscious-level experience in tandem with the morphological content.

Morphological content contributes to the justification of a testimonial belief by providing evidential support for the belief (i.e., evidential support that is in addition to the basic experience, if that also counts as evidence). The combination of the experience (i.e., the reception of testimony, the perception, etc.) and the support from the morphological content provides the believer with a default entitlement to the belief. This provision comes from “good bootstrapping,” which involves multiple different experiential states functioning as checks on accuracy for other experiential states and doxastic states (Henderson and Horgan 2020: 4917). Henderson and Horgan (2020) call this default entitlement “evidentially embedded epistemic entitlement” (Henderson and Horgan 2020: 4908).

I agree with much of what Henderson et al. argue. For instance, I agree that morphological content (the background information) can provide further support to hearers holding testimonial beliefs, and I agree that it can also influence how we experience things at a conscious level (“chromatic illumination”), even if such influence is not directly consciously accessible.

However, there is one main aspect of the way I am employing the notion of the background information that differs from Henderson et al., as well as from the function of “epistemic competences” in Henderson (2008). One of the things I think we should notice about the way in which background information influences new testimonial beliefs is that there are always more things that go into the formation of a testimonial belief than just what the speaker says. Given this, my focus is on how we use the testimony we receive, along with whatever the background information provides, to arrive at a testimonial belief. What is required even for prima facie pro tanto justification is that we move from inputs to belief in a right sort of way. As mentioned, it is this emphasis on the importance of forming beliefs in a right way that seems to be missing from anti-reductionism. In the previous section, I made a case for focusing on forming beliefs in a right way. In the next section, I defend against a few objections.

5. Objections and conclusion

I begin with a fundamental worry about whether forming a belief in a right way is as central to doxastic justification as I say it is. Following Burge (1993), you might argue that the requirement that testimonial beliefs be formed in a right way is just an enabling condition for justification. That is, you might agree that how we form testimonial beliefs is important, but you might say that it just enables the production (or, as some would say, the transmission) of justification. For instance, when my friend whispers to me that I have some spinach between my teeth, if I am to be justified in holding the resulting testimonial belief, I must remember what she has just whispered, I must understand the meanings of the words she uses, I must understand that she is referring to me, and so on. But all these things do not positively contribute to doxastic justification – rather, they must be in place so that other things can produce or ground or transmit justification.

However, note that forming a belief in a right way – understood broadly as a correct “movement” from premises to conclusion – is dissimilar to the other things just identified as enabling conditions, such as a properly functioning memory, knowledge of the meanings of words, and so on: the former is always a necessary condition for you to be justified in believing that p and the latter are not. In other words, forming a belief in a right way positively contributes to justification, whereas enabling conditions do not. Put yet another way, forming a belief in a right way is a necessary aspect of justification, whereas mere enabling conditions are aspects of belief formation. Both the speaker's say so and the way in which we form beliefs rationalize the resultant testimonial beliefs (i.e., they both are rational answers to the question “Why does he believe that?”).

Further, it would be inappropriate to label forming a belief in a right way an enabling condition because this standard would cast too many things as enabling conditions. That is, if we were to label forming a belief in a right way as an enabling condition, we could use the same standard to hold that the speaker's say so, too, is just an enabling condition for justification, which is unacceptable. For if something that positively contributes (or constitutes) to justification could be an enabling condition, then most aspects of a testimonial exchange, including the speaker's say so, would seem to qualify as an enabling condition. For these reasons, I think we should set aside the worry that forming a belief in a right way is an enabling condition.

For another objection, you might claim that H does not need the background information to be justified in believing that p, since, in principle, S's say so alone could justify H in believing that p (cf. Graham 2006). However, the purpose of the foregoing discussion about the function of the background information is to highlight the fact that there are always other things that the background provides when we form testimonial beliefs. Even if the extra things (i.e., things in addition to the speaker's assertion that p) are just enabling conditions for the speaker's say so to “do its work” in grounding doxastic justification, it still matters how we “move” from the speaker's say so and the things that the background information provides to the testimonial belief. The improper formation of a testimonial belief precludes justification regardless of what feeds into that belief.

Given the foregoing picture of how testimonial beliefs are formed, neither just the speaker's say so nor the reasons a hearer has from both the speaker and the background information are enough to justify a hearer in holding a testimonial belief. The reason for this is that the inappropriate formation of a belief is sufficient to preclude doxastic justification. Put differently, even if a hearer has the right sorts of reasons that propositionally justify the content of a belief that p, if the hearer does not form the belief that p in a right way, that hearer will not be justified in believing that p. A classic example of this is where two people both believe that p and they believe that if p, then q. One person appropriately infers q following modus ponens, whereas the other person inappropriately infers q following “modus profuses” (i.e., from any p, q, and r, infer r from p and q). All else being equal, though the belief that q is propositionally justified for both, only the one who follows modus ponens is doxastically justified in believing that q.

For a more concrete example of what it looks like to move from inputs to belief inappropriately, suppose that Naomi believes that, if her professor sends the class an email saying that he is sick, then class will be canceled. Today, she is talking with one of her classmates prior to class. Though Naomi has not checked her emails today, her classmate has. Naomi's classmate informs her that their professor has not sent out an email saying that he is sick today. Naomi then concludes that their class will not be canceled today. Assume that Naomi does not have any other relevant knowledge concerning whether their class is canceled. In this sort of case, Naomi erroneously concludes that their class will not be canceled today. She is not justified in believing that their class will not be canceled today, in part because she moves from inputs to belief in an inappropriate way: she erroneously takes the evidence that her professor has not sent an email saying that he is sick today to imply that their class will not be canceled today. Even if it turns out that their class is not canceled today, given how Naomi forms her belief, she is still not justified in believing that her class is not canceled today. Though there are other factors that could preclude her from being justified, at least part of the explanation must be that she has not formed her belief in a right sort of way.

But a “right” way of forming a belief doesn't have to involve consciously following a logical rule. Rather, a right way of forming a belief could involve, as Ernest Sosa (2015) and others have argued, the competent manifestation of epistemic competence, which is a disposition to form beliefs in a competent way that is appropriately aligned with your evidence. Note that Sosa's understanding of the competent manifestation of epistemic competence differs from Henderson's (2008) understanding of epistemic competences. However, the function of Henderson's epistemic competences could also lead to forming a belief in a right way. It is important to understand, though, that “epistemic competences” is not synonymous with “forming a belief in a right way.”

I am not claiming that the competent manifestation of epistemic competence is the only “right” way to form a belief. For, apart from the above general characterization, I do not take a stand in this paper on what it means to form a belief in a “right” way. Nevertheless, it is relatively uncontroversial that there are right and wrong ways of arriving at beliefs, and I mention the competent manifestation of epistemic competence as a possible account of a right way to arrive at a belief.

However, note also that I am not understanding “forming a belief in a right way” to be equivalent to “being appropriately responsive to reasons.” There is a difference between these two, for you could be sensitive and/or responsive to all the right reasons in your situation and yet still use those reasons in a wrong sort of way to reach a conclusion. The focus in the phrase “forming a belief in a right way” is on how you move from reasons/premises to a conclusion, rather than on the inputs you use when moving to a conclusion (though, as argued above, it seems there are always several inputs involved when reaching a conclusion).³

If a belief is not formed in a right way, whether that be a conscious-level or an unconscious-level process, then the believer is not even prima facie pro tanto justified in holding the newly acquired belief. For instance, the absence of the competent manifestation of epistemic competence is enough to preclude doxastic justification. For if a hearer moves from the reasons acquired from a speaker and the reasons the background information provides to the new testimonial belief in an inappropriate way, the hearer is not justified in holding that testimonial belief.

For a further example, suppose that your friend, who is selling you his old kitchen table, tells you that, apart from a few scuff marks on the edges, there is nothing wrong with the table. Let's suppose that you acquire the belief that there is nothing wrong with the table. If the foregoing argumentation is sound, you acquired your new testimonial belief by either basing it or inferring it from both the reason or assurance you have from your friend and from reasons or other data that come from your background information. If you are to be doxastically justified in holding your new testimonial belief, it is not enough for you just to have reasons or assurance from your trustworthy friend, nor reasons from your background information. For it could be that the way in which you arrived at your belief that there is nothing wrong with the table was inappropriate. For instance, you could have applied the “modus profuses” rule to the reasons you possessed, thereby precluding you from being doxastically justified in holding the belief; and “modus profuses” is not the only way you could go wrong in acquiring a belief – there are several sorts of irrational movements from reasons to belief. For instance, another way in which you could have gone wrong in arriving at your testimonial belief is if you take the warm, affectionate feeling you have toward your friend to imply that there is nothing wrong with the table he is trying to sell to you.

In general, it seems that forming testimonial beliefs inappropriately, regardless of which reasons you have, precludes doxastic justification. But if you could have undefeated reasons to believe that p from a trustworthy speaker (grounded in the speaker's say so) and yet not be prima facie pro tanto justified in believing that p, then it seems that, contrary to the sufficiency claim of anti-reductionism, the speaker's say so is not sufficient for doxastic justification.

Of course, an anti-reductionist could easily absorb my criticism by agreeing that forming testimonial beliefs in a right way is necessary for us to be doxastically justified in holding those beliefs. However, doing so would come at the cost of abandoning the sufficiency claim that characterizes the anti-reductionist position. For if forming testimonial beliefs in a right way is necessary for even prima facie pro tanto justification, then a speaker's say so is not sufficient for justification. A modification of the anti-reductionist position to accommodate my criticism could preserve the view's emphasis on the support testimonial beliefs receive from a speaker, but such a modification would also shift more focus onto how believers form their beliefs (e.g., as in virtue epistemology). However, as mentioned above, such a modification might seem to be too much of a departure from anti-reductionism and instead may be taken to be more aligned with reductionism.

One reason for thinking that my arguments should be taken as warranting only a modification rather than a rejection of anti-reductionism is that, once modified, the resulting view maintains the spirit of anti-reductionism. As mentioned above, most anti-reductionists hold that justification for testimonial beliefs works the same way as justification for other sorts of beliefs, such as perceptual beliefs and memorial beliefs. It seems that my arguments highlighting the necessity of forming a belief in a right way for doxastic justification could extend beyond doxastic justification for testimonial beliefs to perceptual and memorial beliefs as well. It is beyond the scope of this paper to contend that my arguments can be so extended. Nevertheless, if they can be so extended, this result would fit well with the spirit of anti-reductionism.

In conclusion, my claim is that a general feature of human psychology gives us strong reason to modify or reject anti-reductionism about the epistemology of testimony. Because of the work of the background information in how it influences the beliefs we acquire, we cannot help but acquire testimonial beliefs on the basis of a testifier's say so along with additional evidence, concepts, beliefs, and so on. As such, it matters for doxastic justification not only that a rational speaker asserts that p, but it is also necessary that you form your belief in a right way. If this is right, then the sufficiency claim of anti-reductionism is wrong, since you are not doxastically justified in believing that p just on the basis of the speaker's say so.⁴