1. REDUCTIONISM, ANTI-REDUCTIONISM AND BEYOND

The epistemic role that others play in our lives can hardly be overstated. A great deal of our beliefs originate in their words. Hence, a key question concerning the epistemology of testimony is: “Under which conditions are we epistemically justified in our testimony-based beliefs (TBBs)?” (henceforth: the testimony question). Traditionally, the answer has either been reductionist or anti-reductionist. According to reductionism, testimony is epistemically neutral. That is, the fact that someone claimed that p is, in and of itself, no reason to believe that p is true. Therefore, the justification for our TBBs has to be ultimately reducible to other epistemic sources – such as perception, memory and inference. Non-testimonial reasons are necessary for the justification of our TBBs.Footnote ¹ Let's assume I ask a stranger for directions. Upon hearing her words, I remember that people have proven trustworthy with regard to this kind of information in the past. This in turn justifies me to believe my interlocutor now, provided I don't have stronger reasons to doubt her sincerity or competence.Footnote ²

In contrast to this, anti-reductionists hold that testimony is not epistemically neutral. Rather, it has its own distinct epistemic force – like perception, memory and inference. Hence, one is prima facie justified to believe what one is told. According to anti-reductionism, non-testimonial reasons are not necessary for the justification of our TBBs. I am, for example, justified to believe a stranger when she tells me where to go, as long as I am not faced with undefeated defeating counter-evidenceFootnote ³ – e.g. the directions she gives me don't match my knowledge of the territory.Footnote ⁴

However, the dichotomy between reductionism and anti-reductionism has been challenged. Philosophers such as Elisabeth Fricker (Reference Fricker1995), Martin Kusch (Reference Kusch2002), John Greco (Reference Greco, Henning and Schweikard2013, Reference Greco, Greco and Henderson2015) and Karyn Freedman (Reference Freedman2015) have argued that a “one size fits all” approach towards testimony is problematic. After all, the testimonies we encounter, as well as the situations in which we do so, are very heterogeneous. Some cases pull us in the direction of anti-reductionism while other cases pull us in the direction of reductionism. Sometimes it seems like we are prima facie justified to believe what we are told. And at other times it seems that we need non-testimonial reasons to be justified to believe what we are told.

If one subscribes to this diagnosis, then this suggests looking for an account that goes beyond reductionism and anti-reductionism in the following sense: One should want an account that can do justice to our reductionist as well as our anti-reductionist intuitions. More precisely, one should want an account that tells us, in a principled way, when we are prima facie justified to believe what we are told and when non-testimonial reasons are necessary.

This is the line of thought I am going to follow in this paper. First, I am going to present some cases that illustrate the tension between our reductionist and our anti-reductionist intuitions. And I am going to argue that, at least prima facie, these cases give us reason to look for an account that can do justice to both kinds of intuitions (§2). Next, I am going to make a short detour to Edward Craig's (Reference Craig1990) genealogy of the concept of knowledge (§3). My reason for doing so is that Craig's insights provide us with a framework that will allow us to make progress towards answering the testimony question. I am going to argue that the interests that give rise to the concept of knowledge, according to Craig, also give rise to an epistemic norm that governs the exchange of information. And I am going to argue further that Craig's insights can also help us determine how strict we should expect this norm to be (§4). Then I am going to use the results gathered so far to account for our conflicting intuitions in the face of different cases of testimony (§5). To anticipate: I am going to argue that the epistemic norm, as well as the hearer's rough understanding of that norm, gives the hearer a presumptive right to believe what she is told in most cases. However, this justification isn't strong enough in high-risk situations. Here, the hearer needs reductive reasons to believe that her interlocutor is especially competent. Finally, I am going to wrap things up with a brief conclusion (§6).

2. THE CASES

Consider the following two cases:

Intuitively, Bob is prima facie justified to believe what he is told. He is justified to believe the pedestrian when she tells him the time, as long as he doesn't notice, for example, that she isn't wearing a watch or that the time told is incompatible with the position of the sun. Hence, Beer with Friends pulls us in the direction of anti-reductionism.

Things look different, however, when we turn our attention to Nadja. Nadja needs reasons speaking in favour of Peter's medical competence, before she can be justified to believe what he says. She has to know that he is an expert in the relevant field of medicine, or that he has successfully treated patients with a similar diagnosis. Lacking such background information, it seems irresponsible for Nadja to simply believe what Peter tells her. Hence, The Operation pulls us in the direction of reductionism.Footnote ⁵

Still, there might be several concerns with the cases presented. In response to Beer with Friends a reductionist might claim that Bob possesses all kinds of relevant background information that bear on him believing the pedestrian. She looks like a local, she answers confidently and without hesitation, Bob knows that people are generally cooperative in these kinds of situations etc. Hence, he possesses ample reductive reasons to believe what he is told.Footnote ⁶

I don't deny that this might be the case. Still, intuitively, it seems that these reasons are not necessary for Bob to be justified in his TBB. The fact that he understands the pedestrian and that he has no concrete reasons to doubt her sincerity or competence seem sufficient for him to get a justified belief out of the exchange. This is why Beer with Friends pulls us in the direction of anti-reductionism, independent from the reductive reasons that might be at Bob's disposal.Footnote ⁷

Conversely, the anti-reductionist might claim that her theory can account for our intuitions in the face of The Operation. As pointed out in §1, anti-reductionists only assume that we are prima facie justified to believe what we are told. That is, we are only justified in our TBBs as long as we are not faced with defeating counterevidence. Once we are faced with defeaters, we need defeater-defeaters.

Consider the following example: I am sitting at the table with several other people. During the conversation, someone asks: “What time is it?”. My neighbour replies: “It's 4.30 pm”. Inadvertently, I look at my own watch. It shows 3.30 pm. After a moment of puzzlement, I remember the clock change and that I haven't adjusted my watch yet. In this example, according to the anti-reductionist, I am at first justified to believe what my neighbour says. However, I lose this justification when I check my own watch. Now I have defeating counterevidence. This defeater, however, is defeated in turn when I remember the clock change.Footnote ⁸ Hence, if The Operation contains a defeater, anti-reductionists might use this to explain why we need reductive reasons to be justified in our TBB.

However, we can simply stipulate that this is not the case. That is, we can assume that Peter answers Nadja straightforwardly, coherently and without hesitation. Moreover, we can assume that Nadja has no concrete reason to distrust Peter and that nothing he says contradicts Nadja's background beliefs. Still, it seems that Nadja would not be justified in simply believing Peter. This intuition can be expanded upon by imagining a malpractice trial after a failed operation. Let's assume Nadja is asked why she chose procedure X. If she answers: “My colleague Peter recommended it to me”, she has to be prepared to say why she believed him and acted on his recommendation. If she can't answer this question or merely adds: “I had no reason not to believe him”, this can only affect the outcome of the trial in a negative way for her. Such an answer simply wouldn't be good enough in this situation.

The anti-reductionist might respond at this point that we don`t always need reasons to distrust our interlocutor specifically for our TBBs to be defeated. Timothy Perrine (Reference Perrine2014: 3230) observes that most average human adults have a wealth of background beliefs about the respective reliability and unreliability of various sources of information. These beliefs can play the role of defeaters as well. Following this line of thought, we can assume that Nadja has the background belief that the average person is in no position to recommend surgeries. And this background belief acts as a defeater when someone recommends a surgery to her. Hence, she needs positive reasons for believing that her interlocutor is not an average person.Footnote ⁹

While I agree that our background beliefs can act as defeaters, I don't believe they do so in The Operation. Nadja is not turning to a random person off the street. Rather, she is (knowingly) asking a colleague of hers. Peter has a medical degree as well and he is working at the same hospital as Nadja. As such, he doesn't fall under the general rubric of the average person who is in no position to recommend surgeries. Therefore, Nadja's background belief that such persons are in no position to recommend surgeries can't play the role of a defeater here.

Still, the anti-reductionist might want to restate the defeater-objection. As a surgeon, Nadja presumably possesses the following background beliefs as well: Different medical practitioners have different fields of expertise. Some are in a position to recommend surgeries to me while others are not. And these beliefs in turn defeat Nadja's TBBs in The Operation.

However, I don't believe that this is the case. After all, Peter might fall into either camp. He might be fit to recommend surgeries to her, or he might not be. Therefore, the background beliefs alluded to above speak neither for nor against Nadja believing what Peter tells her. They are neutral in this regard. Hence, they can't play the role of defeaters either.Footnote ¹⁰

Another line of objection would be to claim that the two cases presented at the beginning of this section are really about distinct phenomena. Beer with Friends is about testimony while The Operation is about advice. Moreover, different rules apply to advice and testimony. In comparing the two, Edward Hinchman states that:

I take Hinchman's point to be that someone who is asking for information can, under certain circumstances, just take the speaker at her word and believe what is said on this basis. However, this is not true for someone seeking advice. In the latter case, one always has to deliberate further and weigh the advice against one's beliefs. One cannot simply take the advisor at her word and act on her recommendation. This alleged difference between testimony and advice could be used to account for our conflicting intuitions in the face of Beer with Friends and The Operation.

However, I believe there are, at least, two problems with this distinction. Firstly, it seems that we cannot clearly distinguish between seeking someone's testimony and seeking someone's advice. The main difference between the two seems to be that testimony is about belief while advice is about action. But we often ask someone for information precisely so that we might later act on it. Consider the direction-case briefly alluded to before. Asking for directions is a paradigmatic case of seeking someone's testimony. Yet, one typically does so in order to reach a certain destination. Hence, this case can also be described as asking for advice about how to best arrive at the desired place. And if we cannot clearly distinguish between seeking testimony and seeking advice, then it also seems problematic to assume that we are dealing with different phenomena to which different rules apply.

Secondly, it seems problematic to assume that advice can only function as input into doxastic deliberation but never replace it. Consider the following case: Jill and Jack are going to a restaurant together. Jack is rather stuck in his eating habits. He always orders the same. But he wants to change his ways. Therefore, before going out, he decides to let Jill recommend him a dish and to order it no matter what (we can assume, for the sake of argument, that Jack doesn't have any dietary restrictions). Here, Jack is seeking Jill's advice. And he uses her advice as a substitute for doxastic deliberation, rather than as a mere input. Moreover, there seems nothing wrong with Jack doing so. His behaviour doesn't strike one as irrational, immoral or in bad faith. Therefore, the restaurant-case further undermines the alleged distinction between testimony and advice.

Additionally, even if we wanted to uphold the distinction between advice and testimony, this would still not be a problem here. I believe that The Operation is helpful in connection with my investigation, as this scenario is likely to provoke strong intuitions. However, nothing depends on this particular example. Consider the following, slightly modified, version of Steward Cohen's (Reference Cohen1999) well-known airport-case:

Two things have to be noted in connection with Airport. (i) It is clearly a case of testimony. (ii) Intuitively, Mary and John behave correctly. They shouldn't just believe that the flight has a layover in Chicago on someone's say-so. Rather, they need additional reasons for believing their interlocutor. That's why they decide to ask someone whose job it is to be competent with regard to this kind of information. Hence, Airport, like The Operation, pulls us in the direction of reductionism.

Another distinction one might want to draw is the one between justified belief and justified action. That is, in evaluating The Operation and Airport, we have to distinguish between the following questions: Are Nadja, as well as John and Mary, justified to believe what they are told? Are they justified to act based on the resulting belief? Following this distinction, one might argue that our intuitions about the two cases only concern the second question. Nadja, as well as John and Mary, are intuitively not justified to act based on their beliefs. This, however, says nothing about the epistemic status of their beliefs. If we differentiate between the justification of our beliefs and the justification of our actions, then The Operation and Airport miss their mark.Footnote ¹¹

Although this is controversial, I believe there is reason not to distinguish between the justification of our beliefs and the justification of our actions in this way. When we are asked why we did something, it is natural to refer to our relevant beliefs – e.g.: “I went to the fridge, because I believed that there would be beer in the fridge”.Footnote ¹² Hence, one possibility to criticise our actions it to criticise the epistemic status of our action-guiding beliefs. In order to defend ourselves against this criticism, we need to give reasons in favour of the truth of the relevant beliefs. We have already encountered an example for this dynamic above. When faced with a malpractice trial, Nadja needs to give reasons for why she believed she could rely on the words of her colleague. Moreover, the hospital committee will, at least partly, judge the adequacy of her actions based on the reasons she can offer in favour of her relevant beliefs. Somewhat generalizing this result, we can say the following: Whether our actions are justified seems to depend, at least in part, on the epistemic reasons we have in favour of our relevant beliefs.Footnote ¹³

At the beginning of this section, I introduced two cases. While Beer with Friends pulls us in the direction of anti-reductionism, The Operation pulls us in the direction reductionism. Next, I considered some possible replies to these scenarios. While the findings of this section might not provide a knockdown argument against either reductionism or anti-reductionism, I believe that they give us an incentive to look for an account of testimonial justification that can do justice to our reductionist as well as our anti-reductionist intuitions. In order to develop such an account, I am going to use the next section to give a brief summary of Craig's genealogy of the concept of “knowledge”. This in turn will provide us with a framework that will help us to make progress towards answering the testimony question.

3. CRAIG'S GENEALOGY OF “KNOWLEDGE”

According to Craig (Reference Craig1990), we learn something illuminating about our concept of knowledge by asking out of which needs this concept developed and what it has to contain in order to meet these needs.Footnote ¹⁴ In order to answer these questions Craig starts from a fictional state of nature. This will help us to determine the basic needs out of which our concept arose and developed.Footnote ¹⁵

Craig answers as follows: The concept of knowledge arose out of the practical need to have as many relevant true beliefs about our surroundings as possible. In order to meet this need, it is very helpful if we can tap into the beliefs of our fellow human beings. And in order to do so effectively, we need a concept that helps us “to flag good informants”. The concept of knowledge allows us to do just that: “To put it briefly and roughly, the concept of knowledge is used to flag approved sources of information” (Craig Reference Craig1990: 11).Footnote ¹⁶

However, Craig distinguishes between good informants in the state of nature and people we aptly call “knowers”. The reason for this is that the search for information in the state of nature is conducted according to purely subjective criteria. Hence, the answer to the question: “Who is a good informant?” is purely subjective as well: “I am seeking information as to whether or not p, and hence want an informant who is satisfactory for my purposes, here and now, with my present capacities for receiving information” (Craig Reference Craig1990: 84 f.).

This fixation on oneself in the here and now leaves out important factors:

According to Craig, this has the following consequence: “All this is going to edge us towards the idea of someone who is a good informant as to whether p whatever the particular circumstances of the inquirer, whatever rewards and penalties hang over him and whatever his attitude to them” (Craig Reference Craig1990: 91).

Therefore, Craig sees our concept of knowledge as the result of a “process of objectification”. It evolves out of our desire to name good informants for us in the here and now. In its final form, however, it serves the purpose of flagging good informants independent from such circumstantial considerations: “The concept of knowing, our hypothesis must now run, lies at the objectivized end of the process; we can explain why there is such an end, and why it should be found worth making in language” (Craig Reference Craig1990: 90 f.).

For the concept of knowledge to serve this function, it has to be tied to quite high epistemic standards. This is to make sure that those who are called “knowers” can indeed function as good sources of information in a wide variety of practical situations:

However, at the same time, we have reason to believe that the epistemic standards tied to calling someone a knower can't be exceedingly high. In our everyday lives, we, for example, don't expect someone we call a “knower” to be able to exclude sceptical scenarios – a constant dream, evil demons, brains in vats etc. Such scenarios wouldn't even cross our minds. After all, nobody would be able to live up to a maximally strict concept of knowledge. Therefore, such a concept wouldn't help us to identify good informants – there would be nobody left to fit the bill. As David Henderson puts it:

To sum up: The concept of knowledge serves to flag good sources of information. In the state of nature this means flagging good informants relative to one's own practical needs in the here and now. That's how we get a kind of “proto-concept” of knowledge. This proto-concept goes through a process of objectification. At the end of this process, we arrive at a concept that allows us to flag good informants in a wide variety of practical situations. This is due to the fact that calling someone a “knower” is tied to this person fulfilling quite high, though not exceedingly high, epistemic standards. After thus outlining Craig's position, I am now going to apply his insights to testimony.

4. APPLYING CRAIG'S LESSONS TO TESTIMONY

In order to see how Craig's account can help us to answer the testimony question, we have to consider his point of departure. Craig starts with two imminently plausible assumptions:

1. 1. It is in our interest to have as many relevant true beliefs about the world as possible.

2. 2. In order to achieve this goal, it is highly advantageous if we can tap into the epistemic resources of our fellow human beings.

Craig takes these interests to be universal and he refers to them to explain how the concept of knowledge arose and developed across different languages. However, to achieve the original goal – to have as many relevant true beliefs about the world as possible – we need more than just a concept to flag good sources of information. On its own, such a concept only helps us to discriminate between different sources of information. But the concept alone doesn't guarantee that we in fact receive lots of high quality information.

To achieve this goal, we also need norms that regulate, and thereby exert some quality control over, the acquisition and the distribution of information. The latter is what I am going to focus on in this paper. Given our interest in having relevant true beliefs about the world and the role our fellow human beings play in this endeavour, we should expect there to be an epistemic norm governing the exchange of information. Or more precisely, we should expect there to be an epistemic norm governing the speech act of assertion that paradigmatically constitutes giving testimony:Footnote ¹⁷

Moreover, we should take the existence of such a norm to put the speaker under the obligation to meet its requirements, should she want to make an assertion. Conceived this way, EN would exert a certain force over her and prevent her from speaking lightly, that is asserting something when she isn't in the epistemic position to do so. Hence, EN would make it more likely that we receive high quality information.

However, norms don't necessarily come with an obligation to meet its requirements. It might be a mafia-norm that its members commit acts of violence to further the cause of their organization. Still, we wouldn't want to say that the members are thereby obliged to commit acts of violence.Footnote ¹⁹

Let me respond to this worry by stressing an important difference between EN and the norms of the mafia. Following Craig, I take the interests that give rise to EN to be universal. We have them because we are human beings with limited cognitive resources and it is not up to us to change our predicament. This universality explains why we should expect EN to come with an obligation to follow it. In contrast to this, mafia-norms only exist relative to the interests of the mafia. And, intuitively, there is strong reason not to be a member of this organization in the first place.

While it is often just assumed in the literature that the fulfilment of an epistemic norm is constitutive of asserting properly, we can use Craig's insights to explain why we should in fact expect there to be such a norm. But the parallels don't end here. We can also draw on Craig's account of how the proto-concept of knowledge turns into a concept of knowledge proper. This will help us to specify the epistemic requirements one has to fulfil to assert properly.

Like Craig, we can start by imagining a small community of people in the state of nature. Here, everybody knows everybody and knows about the other's practical needs. Hence, the speaker can tailor her assertion to the hearer's needs. Therefore, we get a “subjective norm of assertion”. However, such a state of nature is significantly different from a complex modern society. Here, we have to be prepared to function as informants for a variety of people with widely different practical needs. Moreover, we cannot expect that we will often be in a position to know about these needs.

In order to meet these needs, we can again appeal to a process of objectification. We start with an epistemic norm that is tied to a specific hearer's needs in the here and now and end with an epistemic norm that caters to the needs of different hearers in a wide variety of practical situations. And, analogously to the concept of knowledge, we can assume that this process of objectification raises the epistemic standards. A speaker has to fulfil relatively high epistemic demands in order to assert properly. This is to make sure that her testimony will serve a wide variety of hearers irrespective of their practical needs. But, as with the concept of knowledge, the epistemic standards for asserting properly mustn't be exceedingly high. Such high demands would too often prevent us from exchanging information. This in turn would be detrimental to our original goal of having as many relevant true beliefs about the world as possible.

To sum up: The same need that gives rise to our concept of knowledge according to Craig, can also be used to explain the existence of an epistemic norm governing assertoric speech. Moreover, we can stipulate that this norm of assertion undergoes a process of objectification. At its end lies a norm that is tied to quite high, though not exceedingly high, epistemic standards. This shall make sure that our testimony can serve as a source of information in a wide variety of practical situations. In the next section, I will use the framework developed so far to account for our conflicting intuitions in the face of different cases of testimony.

5. ACCOUNTING FOR OUR CONFLICTING INTUITIONS

In the last section, I have argued for two things: (1) The speech act of assertion, by which we paradigmatically convey information, is governed by an epistemic norm and (2) this norm is tied to relatively high epistemic standards. This allows us to explain why we are often prima facie justified to believe what we are told. Any speaker conveying information is subject to the epistemic norm of assertion. Hence, we can assume that she will not speak lightly. She will only make an assertion, if she is in a strong enough epistemic position to do so. Moreover, it is part of an average human being's language competence to have some rough understanding of this norm.Footnote ²⁰ That this is the case can be seen from the fact that we are prone to reproach a speaker if we find out that she did speak lightly, without being able to properly back up her assertion. In the words of Sanford Goldberg:

These considerations allow us to straightforwardly account for the anti-reductionist intuitions evoked by Beer with Friends. Again:

Intuitively, Bob is prima facie justified to believe the pedestrian when she tells him the time. According to the account developed so far, this can be explained as follows: In answering Bob, the pedestrian commits herself to upholding the epistemic norm that governs the speech act of assertion. She will not speak lightly, but only answer Bob if she is in the epistemic position to do so. Otherwise, she will say something like: “As far as I know” or: “You should really ask someone else”. Moreover, Bob – being a competent speaker – has a rough and ready understanding of the epistemic norm in play. Therefore, Bob has a prima facie reason to believe what he is told.

One might object at this point that young children don't have such a “rough and ready” understanding of the epistemic norm in play. Therefore, this account implies that young children can't acquire justified beliefs through testimony. This is especially problematic since many anti-reductionists are (partly) motivated by the intuition that young children can indeed acquire justified beliefs through testimony. Does the account developed here fail to retain this motivation?Footnote ²¹

In response to this, it has to be noted that young children already show some competence in dealing with assertoric speech. Infants as young as 16 months attempt to correct human adult speakers when they observe the latter to falsely label objects both have visual access to (Koenig and Echols Reference Koenig and Echols2003). This can be taken as suggesting that infants have a basic understanding of assertions.Footnote ²² Moreover, 3- and 4-year-olds prove competent in identifying accurate vs. inaccurate informants (Koenig et al. Reference Koenig, Clément and Harris2004). And a further study shows that 3-year-olds explicitly understand the difference between assertoric and imperative speech. In the case of assertions, they criticise the speaker for saying something wrong. In the case of imperatives, they criticise the addressee for not doing what she is told (Rakoczy and Tomasello Reference Rakoczy and Tomasello2009). These studies indicate that young children are more sophisticated in their understanding of language than we give them credit for. Still, one has to be careful not to overinterpret these studies. They might just suggest that small children have a grasp of assertoric speech aiming at truth.Footnote ²³ This in turn doesn't show that they actually have an understanding of EN.

However, even if one remains agnostic on the last point, the following has to be emphasized. The justification under consideration here is internalist in nature. The hearer is prima facie justified to believe the speaker because she understands that, in testifying, the speaker commits herself to upholding an epistemic norm. Even if young children are lacking such internalist justification, this doesn't mean that their TBBs are devoid of any positive epistemic status. The account presented here is compatible with them being entitled to believe what others tell them in an externalist sense. More precisely, the existence of an epistemic norm gives them an epistemic right to their TBBs, although this right might not be reflectively accessible to them.Footnote ²⁴ Hence, the account developed in this paper is more “child-friendly” than it might first seem.Footnote ²⁵

Another line of objection would be that it seems problematic to simply assume that any given individual will comply with the epistemic norm of assertion. Philip Nickel worries that: “From an epistemic point of view it might seem stupid to rely on presumed compliance with an epistemic norm or rule as a free-standing basis for forming a belief” (Nickel Reference Nickel2013: 211). And Casey Johnson claims that: “[T]he constitutive norm of assertion neither guarantees nor makes it more likely that asserters stand in the appropriate epistemic relation to the content they assert” (Johnson Reference Johnson2015: 360). As there is always the possibility that we will encounter a speaker who violates the norm of assertion, such a norm can't explain why we should be justified to believe any given speaker. What we need is additional reasons to believe that our interlocutor does in fact act according to this norm.

However, I believe that this objection misses the function that norms play in our everyday lives. Consider the following analogy: There are norms and rules that regulate road traffic. Therefore, a driver doesn't need additional reasons to believe that her fellow drivers will also act in accordance with the road traffic licensing regulations – e.g. they will drive on the right side of the road, keep within the speed limit, respect right before left. Rather, this is a background assumption while going from A to B. Similarly, one should be able to assume that one's interlocutors act in accordance with the epistemic norm of assertion.

Still, one might wonder whether the testimony-case and the traffic-case are similar enough to draw far reaching conclusions. After all, there seems to be an important disanalogy. In the traffic-case, it is in everybody's self-interest to uphold the norms – e.g. to drive on the right side of the road. In the testimony-case, however, it is often in a person's self-interest to violate the norm – to deceive one's interlocutor. Therefore, in the testimony-case, unlike the traffic-case, one always needs further reasons to believe that one's interlocutor will behave in a cooperative way.Footnote ²⁶

However, I don't believe that this conclusion follows. It seems frankly paranoid to always have to consider whether one's interlocutor might have a hidden agenda. Why, for example, should Bob have to wonder whether it is in his interlocutor's interest to deceive him when he is asking for the time? Rather, the following seems more appropriate. One only needs further reasons to believe that one's interlocutor is following the norm of assertion if one has concrete reasons to believe that she has a motive to deceive. As Alvin Plantinga puts it: “I believe you when you tell me about your summer vacation, but not when you tout on television the marvellous virtues of the deodorant you have been hired to sell” (Plantinga Reference Plantinga1993: 79).

This result is compatible with anti-reductionism. After all, here one is only prima facie justified to believe what one is told. As pointed out in §2, this justification can be defeated. This is the case when one has concrete reasons to believe that it might be in one's interlocutor's interest to speak deceptively. Such defeaters call for additional reasons that act as defeater-defeaters in order to restore justification. In Plantinga's advertising-case, for example, one would need additional reasons to believe that the actor would never endorse a product she doesn't believe in.

Similar considerations apply to the traffic-case as well. A driver is also only prima facie justified to believe that the traffic-norms are upheld. She has to exert special caution if she has concrete reasons to believe that someone else is violating the norms – e.g. there was an announcement on the radio warning against a wrong-way driver. In this situation, she should drive extra carefully until the all-clear is given.

The testimony-case, as well as the traffic-case, point to a general feature of the way that norms affect our everyday lives. Such norms afford us with a background of common expectations in front of which our social interactions can flourish. They are there precisely so that we don't always have to ask ourselves how the others will behave. We only have to enter into a more reflexive mode of assessment once we have concrete reasons to doubt that a certain norm is upheld in a given situation.Footnote ²⁷ Consequently, we have no reason to assume that the reference to an epistemic norm of assertion should be insufficient to explain why we are prima facie justified to believe what we are told.

Still, it is important to notice that the prima facie justification outlined so far has its limits. As argued in the last section, we should expect the epistemic norm of assertion to be tied to relatively high epistemic standards. This is to make sure that we function as good informants for each other in a wide variety of practical situations. However, at the same time, we shouldn't expect these epistemic standards to be exceedingly high. If the latter was the case, we would often have to abstain from giving each other information. This in turn would undermine the very function of the epistemic norm of assertion. It would now hinder the flow of information instead of just regulating it.

This has the following consequence: Someone seeking information in a high-risk situation can't assume that the speaker meets her epistemic needs. Although she may assume that the speaker conforms to the epistemic norm of assertion, she can't assume that this norm is tailored to her special needs. The existing norm of assertion is likely tied to lower epistemic standards than the ones that the hearer is subject to. Hence, the hearer isn't prima facie justified to believe the speaker. She needs additional reasons. More precisely, the hearer needs reasons to believe that the speaker is in a better epistemic position than the speaker would usually need to be in to assert properly. The hearer needs reasons to believe that the speaker is especially well equipped to give the hearer the desired information.

These considerations allow us to account for the reductionist intuitions provoked by The Operation and Airport. Again:

Nadja is in a high-risk situation. Her patient's life is in danger. Therefore, the epistemic standards she is subject to are higher than the epistemic standards that usually govern assertions. Hence, she can't just take her colleague at his word. In order to be justified in believing him, she needs additional reasons speaking in favour of his special competence. The same holds true for Mary and John. A lot depends for them on being at Chicago airport on time. Therefore, they can't just take Smith at his word. Rather, they need to ask someone whom they have reason to believe to be highly reliable on the subject at issue.Footnote ²⁹

Let me briefly elaborate on two kinds of reasons a hearer can have for believing the speaker to be especially competent with regard to the desired information:

1. (i) The hearer has reason to believe that the speaker knows about the hearer's high practical risks.

As pointed out before, asserting properly is tied to relatively high epistemic standards. This is to make sure that we can function as good informants for each other in a wide variety of situations. This general concern for each other's needs suggests the following extension of the account developed so far: A speaker should consider her interlocutor's special practical risks if she knows about them.

Again, we can take our cue from Craig's remarks concerning the standards that are tied to calling someone a “knower”: “Where, as in the murder trial, we positively know that the most serious consequences turn on it, our inclination is to wind the standard up yet another notch” (Reference Craig1990: 94). Analogously, we can say that if we know that for our interlocutor a lot depends on receiving accurate information, then we have to apply even stricter standards to our assertions than we would do normally.

This has important consequences for the hearer. If the hearer has reason to believe that the speaker knows about the hearer's heightened risks, then the hearer may assume that the speaker will consider these risks and adapt her assertion accordingly. That is, the hearer may assume the speaker will only make a straight-out assertion if the speaker is absolutely certain. The Operation and Airport can help us to flash out these considerations. Let's assume that Nadja doesn't just give Peter a broad description of the case. Moreover, she emphasizes how serious her patient's condition is and that the operation is a matter of life and death. Due to her vivid description, Nadja may assume that her colleague is cognizant of the seriousness of the situation. Therefore, she has reason to believe Peter if he answers her directly and doesn't qualify his answer with phrases like: “I am not an expert in the relevant field of medicine” or: “I haven't treated this kind of case before” (moreover, we can assume that the no-undefeated-defeater-condition is fulfilled). Similar considerations apply to Mary and John. Let's assume they tell Smith how important it is for them to be in Chicago on time. Now, they can assume that Smith knows what's at stake for them. If he still sticks to his guns and doesn't retract or qualify his previous assertion, then this gives them reason to believe him.

However, this reason doesn't apply if the hearer doesn't have the opportunity to tell the speaker about her situation and if, moreover, the hearer doesn't possess independent reasons to assume that the speaker knows about the hearer's situation. If – for all the hearer knows – the speaker isn't cognizant of the hearer's special situation, then the hearer can't assume that the speaker will take it into account. In this case, the hearer needs a different kind of reason to believe the speaker:

1. (ii) The hearer has independent reasons to believe that the speaker is an expert with regard to the desired information.

If (ii) is the case, then the hearer may assume that the speaker satisfies very high epistemic standards. After all, one is an expert in virtue of having lots of knowledge about a certain subject. And this knowledge will be reflected in her assertions concerning this subject.

Once again, The Operation and Airport can be used for the purpose of illustration. Let's assume that Nadja didn't have the opportunity to tell Peter about her patient in detail. However, due to her research, she knows that Peter is a leading figure in the relevant field of medicine, who has never lost a patient. This background knowledge can also justify Nadja in believing what Peter says. And let's assume that Mary and John don't tell Smith about their predicament. While they wouldn't be justified to believe Smith in this case, they would still be justified in believing the airport agent. This is due to the fact that it is part of an airport agent's job to be an expert with regard to the kind of information that Mary and John are seeking.

Although I discussed (i) and (ii) separately, it is of course possible that both reasons are available to a hearer simultaneously. It is possible that we find ourselves in a high-risk situation in which we are seeking information from someone whom we have reason to believe knows about our predicament and whom we know to be an expert. Moreover, there might be cases where the risks are so high that only a combination of (i) and (ii) will do.

6. CONCLUDING REMARKS

The aim of this paper was to give an answer to the question: “Under which conditions are we epistemically justified to believe that what other people tell us is true?” More precisely, the aim was to develop an answer to this question that looks beyond reductionism and anti-reductionism in so far as it does justice to our reductionist as well as our anti-reductionist intuitions. In this connection, I have argued for the following position: In most cases, we are prima facie justified to believe what other people tell us. This is due to the fact that the speech act of assertion is governed by an epistemic norm of which we have a rough and ready understanding. This norm evolved in such a way as to make sure that we can function as good informants for each other in a wide variety of practical situations. However, this norm did not evolve with an eye to high-risk situations. Hence, a hearer in a high-risk situation can't simply assume that the speaker's assertion is tailored to the hearer's special needs. Therefore, the hearer isn't prima facie justified to believe what she is told in these situations. Here, the hearer needs reductive reasons to believe that the speaker is especially suitable to give the hearer the desired information.

Let me end by addressing a general worry. One might wonder whether the conclusions reached in this paper are specific to testimony. After all, various philosophers have argued that practical risks affect the strength of the reasons we need to acquire justified beliefs or knowledge. This is a perfectly general thesis (henceforth: GT). If it turns out to be true, then it applies to testimonial reasons as well. So why devote a paper to specifically arguing that the justification of our TBBs depends on practical risks?Footnote ³⁰

In response to this, I want to point out two ways in which my account moves beyond GT and provides insights that are unique to testimony. Firstly, philosophers typically argue for GT by making an inference to the best explanation – the intuitions certain cases provoke in us are best explained by assuming that high risks raise the demands on our epistemic reasons.Footnote ³¹ However, such an inference to the best explanation, in and of itself, doesn't explain why high risks should have this effect. In contrast to this, the account developed in this paper gives an explanation as to why we should expect cases like The Operation and Airport to require Nadja, as well as Mary and John, to possess reductive reasons. Moreover, this explanation is specific to testimony as it refers to the epistemic norm that governs the exchange of information.

Secondly, GT only says that high risks require us to possess stronger epistemic reasons than usual. It doesn't say which kinds of reasons are required for different epistemic sources such as testimony, perception, memory and inference. Again, the account developed in this paper is more specific. It points out that the hearer needs reasons speaking in favour of her interlocutor's special competence with regard to the topic at issue. And it further points out what these reasons consist of. So even if the claim that the justification of our TBBs is influenced by practical risks is a special instance of GT, the account developed in this paper can still deepen our understanding of how we acquire justified beliefs from the words of others.Footnote ³²