Testimony and Non-Evidential Reasons for Belief (A Non-Purist Place for Interpersonalism)

Abstract

Interpersonalist theories of testimony have the theoretical virtue of giving room to the characteristic interpersonal features of testimonial exchange among persons. Nonetheless, it has been argued that they are at a serious disadvantage when it comes to accounting for the way in which testimonial beliefs may be epistemically justified. In this paper, we defend the epistemological credentials of interpersonalism, emphasizing that it is inseparable from the acceptance of non-evidential epistemic reasons to believe, which demands proper conceptual elaborations on the notions of epistemic reasons and of epistemic justification. We offer a proper reading of epistemic reason, and we defend non-purism on justification as the adequate way to conceive the epistemic proposal of interpersonalism on testimony, realizing that only this combination is capable of apprehending certain cases in which there seems to be no way to rule out the idea that the assurance offered by the testifier offers an epistemic reason to believe that it is not evidential.

1. Introduction

The epistemology of testimony as it has developed over the last several years involves, among other debates, the dispute between positions according to which testimony must be explained in terms of the concept of “evidence,” and positions that reject the various forms of evidentialism because they allegedly distort the strictly interpersonal particularities involved in the act of testifying. A prominent interpersonal view of testimony is the one developed by Richard Moran (2006), who defends the idea that testimony must be described as an act whereby a speaker says something to a hearer, presenting him or herself as a guarantor of what is being said. Expressions such as “I assure you” and “Believe me” are always implicit in the act of testifying, and any evidentialist approach to the practice of testimony dismisses their epistemic import by reducing it to that of evidence. As an alternative to interpersonal views, and specifically against Moran's version, authors such as Jennifer Lackey (2008) and Frederick Schmitt (2010) have questioned the epistemological relevance of interpersonal aspects.

We will start from a characterization that appropriately distinguishes interpersonalism and evidentialism on testimony, and, using a thought experiment, we will offer a case in favor of interpersonalism. Next, we will observe that this case prompts us to pay attention to the fact that interpersonalism entails a non-evidentialist conception of what must be admitted as an epistemic reason. From there we will focus on Lackey's arguments for the claim that interpersonal relationships present in testimony are epistemically impotent. We will offer a way of limiting the reach of such arguments, showing that they only succeed if one endorses a purist theory of epistemic justification according to which the latter is only satisfied with evidential reasons. We will also argue that a theory of justification that incorporates a pragmatic condition of justification can help interpersonalism by showing how interpersonal factors can have epistemic import, thus allowing a full explanation of the intuitions that follow from the presented thought experiment. Consequently, in the last section, we outline the minimum content of a non-purist conception of justification that can capture the phenomenon highlighted in our thought experiment, thus giving viability to an interpersonalist conception of testimony.

As will be understood throughout the text, our purpose is to approach the defense of interpersonalism by framing the inquiry about testimony within the debate concerning epistemological purism vs. non-purism.¹ We believe that much of the debate in epistemology of testimony was developed without paying sufficient attention to the various perspectives that become available upon a more careful consideration of the connections between both debates. Our purpose in this paper is to lay the foundations to overcome this theoretical disadvantage.

2. Interpersonalism and evidentialism

It is usually acknowledged that the basic form in which information circulates among people is through an exchange such as the following:

A speaker S asserts that p to hearer, H, and H henceforth believes that p.

Let us begin, then, by briefly characterizing the main traits of both interpersonalism and evidentialism when approaching this kind of exchange.

Interpersonalism affirms that the epistemic value of testimony-originated beliefs stems from some aspect of the interpersonal relationship present in the testimonial exchange. Different theories of testimony stress different aspects. Moran's theory emphasizes the adoption of responsibilities on the part of S regarding the fact that his or her speech act is a reason to believe that p (2006: 190).² According to his version of interpersonalism, this aspect is what grants the hearer a reason to believe, and explains the epistemic value of testimonial exchanges, which in turn explains why, under certain circumstances, testimonial exchanges can give rise to justified beliefs.³ Moran's interpersonalism, as a view concerned with the epistemic value of testimonial beliefs, could be defined as follows:

INTERPERSONALISM: Given a testimonial exchange between a speaker S and a hearer H, where S conveys to H the information that p through the speech act x, whereby H forms the belief that p,

(I-1) The speech act x carries an assurance that S offers to H in relation to p (S explicitly takes on a certain responsibility for the state of S's utterance as a reason for H to believe).⁴

(I-2) Given (I-1), x (or the consideration that x is in place) gives H a Non-Evidential Reason, albeit an epistemic one, to believe that p.

(I-1) specifies the relevant aspects of testimony that are responsible for its epistemic value, and (I-2) explains precisely how those aspects generate epistemic value. INTERPERSONALISM defined in this way does not give sufficient conditions for H to be justified in believing that p, but points to the relevant aspects of testimony that must be present when giving those conditions. It is important to notice that although (I-2) states that testimony provides an epistemic but non-evidential reason to believe that p, such a reason can be insufficient for the belief to be justified. This might require further conditions, such as the speaker being trustworthy, a certain degree of evidential support being reached, or a lack of undefeated defeaters, and so on. All INTERPERSONALISM claims are that the specific epistemic value of testimonial exchanges is grounded in an interpersonal relationship between speaker and hearer, and that this can justify testimonial beliefs (if other conditions are in place), in that it provides an epistemic reason to believe that p.⁵ INTERPERSONALISM, therefore, defends the idea that testimony can offer the hearer an interpersonal reason (IR) to believe the proposition conveyed by the testimony.

On the other hand, evidentialism explains the epistemic value of testimonial exchanges in evidential terms, establishing that there is some evidential relation between the testimony that p and p. Hence, it includes two theses parallel to (I-1) and (I-2):

EVIDENTIALISM: Given a testimonial exchange between a speaker S and a hearer H, where S conveys to H the information that p through the speech act x, whereby H forms the belief that p,

(E-1) There is a relationship between x and p, available to H, whereby the conditional probability (objective or subjective) of p given x is higher than the conditional probability that p given non-x.⁶

(E-2) Given (E-1), x (or the belief or the knowledge that x is in place) gives H an Evidential Reason to believe that p.^{7, 8}

Again, EVIDENTIALISM does not give sufficient conditions for a justified testimonial belief, since several conditions could be required.

It is difficult to evaluate virtues and vices of both positions without a more vivid image of how exactly are testimonial beliefs going to be actually justified. How can we know for sure that some IRs can be responsible for conferring epistemic value without knowing under what specific circumstances testimonial beliefs are justified? However, in the next section we will develop a case that gives plausibility to the idea that, at least in some cases, interpersonal aspects of testimony are (at least partially) responsible of the epistemic positive status of the belief, by showing how a belief becomes (intuitively) justified when the person receives an IR.

3. A case in favor of INTERPERSONALISM

Let us consider a widespread situation in our practices of medical consultation. Suppose a specialized physician prescribes a certain imaging diagnostic study, which will enable him or her to define whether we have a serious disease or not. We have the study done and the results are delivered to us along with a report that indicates a very high probability that we do not have the disease. Let us even imagine that we are already familiar with reading the corresponding images and that, in seeing them, we form a relevant belief about the absence of that disease. This situation is very common; we have highly reliable information about our health before going back to the specialist, which is due to our direct reading of the attached results or reports. Still, our usual behavior is to schedule another appointment with the specialist, bringing over the results, and waiting for him or her to give us the diagnosis through direct testimony. The specialist's word makes a substantial difference to the justification of our beliefs that will serve as a rational basis for our decisions about acting or not in a certain way. In these high-risk cases, evidential support (a lot of it) does not seem to be enough for us. So if the word of someone does make the difference, then it might be the case that what it adds is not something evidential in nature.

Drawing on these situations, which could be seen as part of our common sense regarding what is the proper thing to do in these cases, could still be contested by evidentialists, who would want to insist upon the evidential nature of the specialist's word. Even if we had a lot of evidence before the specialist word, it might have not been enough evidence for justification. The specialist word adds that bit of evidence necessary for justification. This is why a case must be presented where it is very difficult to argue that the physician's testimony brings about a new evidential reason in favor of being justified in accepting a certain diagnosis. In particular, we believe that the following case is compelling:

SURGERY. Roberto has been found to have a heart infection compatible only with two diagnoses, A and B. A is fatal if it is not operated on rapidly, while B requires no treatment. The operation in itself involves a considerable risk of dying, so it is important to determine whether the condition is A or B. Roberto, however, is ready to undergo surgery in case it is A. The evidence (scientific papers, medical examinations, etc.) supports proposition s: “Roberto has condition B.” The problem is that such evidence, although strong, is insufficient for Roberto to decide not to undergo surgery. Given the importance of this case, and the risk of supposing that s is false and not undergoing surgery, Roberto is not in a position to make that decision. On the other hand, he still does not have his G.P.'s opinion. Roberto recently took the results to the G.P.'s office, and the physician, in order to calm him down, told him: “Roberto, by next week I will have seen your test results and will have a diagnosis. Do you see these two filing drawers? The one on the left contains the records of my patients that have serious illnesses, and the one on the right has those of the patients that are healthy or have minor conditions. Next week you will be in one of these drawers, and your indecision will be over.” The following week, Roberto is in his doctor's waiting room, waiting to go into her office. The office door is open, and Roberto can clearly see that his doctor gets his file from the drawer on the right and puts it on the desk. This adds new evidence in favor of s (i.e., his doctor believes that s). Still, Roberto does not make a decision. When he goes into the doctor's office, Roberto sits down and listens to what the doctor has to say. “Roberto, you have B, I give you my word.” Roberto feels his muscles relax: he has finally decided not to undergo surgery.

Many things can be said about this case. As we already mentioned, we believe that it is not only imaginable, but highly plausible, at least for many people. The word of others, a personal assurance, can lead someone to form a belief, make a decision, or modify the course of action to be pursued.

What does this case say about justification? We believe that intuitions are very clear in that the evidence that Roberto had before seeing his G.P. is not enough for the (potential or actual) belief in s to be justified. It is also quite clear that the belief is justified after the G.P.'s assurance. Intuitions are less clear regarding when exactly the (potential or actual) belief becomes justified. Would Roberto be justified in believing s after discovering that his file belongs to the “healthy patients” drawer, even if he actually formed the belief after receiving the doctor's assurance? When does the indecision about undergoing surgery become irrational for him? Could the (potential) belief be justified and still be the case that it would not be proper for Roberto to act upon s?

This last question has a standard negative answer, since many philosophers agree in that epistemic states and action are related in a certain way. This is so insofar as we accept some principle that links epistemic justification or knowledge with rational action, such as the following case of justification:

ACTION-JUSTIFICATION (A-J) It is rational to act based on p only if you are justified in believing p.^{9, 10}

So if something like this principle is correct, then we should look further into two things. Recall that Roberto makes up his mind after hearing his doctor's assurance that s. He is ready to act upon p after this. So the first question is: Is acting upon s rational for Roberto after his doctor's assurance? We believe that there is no intuitive idea about rational action, or in the description of the case, that prevents us from giving a positive answer. A second and more interesting question is whether acting upon p was rational for Roberto after seeing that his file was in the “healthy” drawer, but before talking to his G.P. This is, even if his indecision went away only after getting his doctor's assurance, was he being irrational in waiting? Should he have made up his mind right after seeing that he belonged to the right drawer?

Answering this requires several considerations regarding rational action, but it does not seem plausible to hold a characterization of rational action that is so externalist that it prevents the rationality of an action of this type from being partially determined by issues associated with the individual's preferences, his or her aversion to risk, his or her commitments, and the reflexive evaluation of his or her own situation. In this respect, we believe that it is possible and desirable to explain Roberto's decision in terms of his rationality, in the sense that before the doctor's testimony it was not rational for Roberto to not undergo surgery, and that after the testimony it was rational indeed for Roberto not to do it. This does not mean that for another person, in the same “objective” circumstance and with the same evidence, the course of actions could not be different.¹¹

In this sense, we think that there is a strong case (by means of the principle A-J) for the idea that his belief becomes justified only after his doctor's assurance. Now, of course, this is not something that straightforwardly shows that interpersonalism is correct since there could be an evidentialist explanation of the situation. Before the doctor's testimony, there was strong evidence for p, but not enough evidence for p being justified (this is the same dialectical situation we had when considering our general behavior regarding the specialists word). However, we believe that the best description of SURGERY makes a strong case for interpersonalism because it allows us to rule out the only two possible interpretations that evidentialism can offer for understanding the role of doctor's testimony in the move toward justification:

1. (a) The doctor's testimony adds evidence in favor of proposition s;

2. (b) The doctor's testimony:

   1. (bi) is a defeater of defeaters that Roberto could have against s,

   2. (bii) is a defeater of defeaters that Roberto could have against the quality of the evidence in favor of s (in this case, against the consideration that the doctor's behavior is adequate evidence for s),

   3. (biii) eliminates some possibility of error that the observation of such behavior leaves open.

Let's see with a bit more detail why interpretations (a) and (b) are not adequate.¹²

(a) The issue of whether evidential support in favor of s is the same before and after testimony is crucial for evaluating this case as a case that favors interpersonalism. Roberto assumes that his doctor forms medical beliefs in a reliable way, and therefore when it comes to truth, there is no difference as to the degree of support for s before and after testimony. The doctor's word about s does not have a better epistemic relation (in a strict sense) with the truth of s than his beliefs. “The doctor told me s” does not add any evidential content to “The doctor believes s.” If any, as Moran states, the difference would be to the detriment of testimony, not in its favor (278). What is relevant, then, is that the epistemic change is not produced by the incorporation of new evidence, given that the doctor's testimony, considered as mere evidence for s, has the same epistemic weight than the evidence given by his behavior of extracting Roberto's file from the drawer on the right.¹³

Our argument does not reject the idea that the word of the doctor could be taken evidentially. As Moran has pointed out, all tellings can be taken as mere evidence, in the same way as a promise can be taken as mere evidence – because of not trusting the speaker, for example. In those cases, we still infer that the speaker will comply with it, taking as evidential basis for the inference precisely that he made the promise in question.¹⁴ From there, the defender of the idea that the doctor's testimony offers additional evidence to that offered by his previous behavior might want to hold the idea that what Roberto did in that case was take doctor's assertion as mere evidence (without giving weight to the dimension of assurance expressed in clause I-1 of the interpersonalist project). Is this interpretation of SURGERY what we reject, because if that were Roberto's attitude toward the doctor's assertion, he would not have obtained anything that he did not previously have with the observation of his conduct. This alternative evidentialist account is thus powerless to reasonably explain what prompts Roberto to seek the doctor's testimony and why it ultimately succeeds in providing a reason to believe which Roberto previously lacked.

(b) Here we have three possibilities to consider. The first one (bi) is easily dismissed because, if it were possible to find an example in which the doctor's testimony allowed to rule out a defeater of s (and not merely being evidence in favor of s), it is obvious that this defeating role would already be fulfilled by the doctor's behavior, so that the testimony would not involve the epistemic change that SURGERY presents.¹⁵ The second and third alternatives (bii and biii) are more interesting, but can be tackled together. It is important to note that the testimony of the specialist is not required for Roberto to rule out possibilities that cast doubt over the claim that the observation of the doctor's behavior is evidence in favor of s, and does not rule out some possibility of error (for instance that the doctor might have misplaced the file). Roberto has no doubt that the observation of the doctor getting his file from the drawer on the right is evidence in favor of s (he has no reason for doubting his vision, nor does he have any suspicion about the doctor's account of the arrangement of the files, nor does he fear that the doctor may have misplaced the file this time around, etc.). SURGERY could become more sophisticated by adding all kinds of details so that Roberto would have a huge body of evidence against defeating scenarios and possibilities of error (e.g., evidence in favor of his vision skills, evidence that the doctor is extremely meticulous and never places a folder in the wrong file, etc.). What Roberto needs to determine his course of action is not evidence about the value of the evidence obtained through observation of the doctor's behavior. This point is crucial, not taking it into account could lead to a reading that understands the epistemic gain, obtained by Roberto while receiving his doctor's testimony, in evidentialist terms.

Once (a) and (b) are dismissed, the best interpretation of SURGERY is that the doctor's testimony operates directly as a reason that cannot be understood evidentially and still has the epistemic force to make the belief in s justified. This is a vindication of interpersonalism. According to INTERPERSONALISM, there is a straightforward explanation of why there is epistemic justification only after the doctor's testimony: an epistemic reason, different in kind to all the evidential reasons Roberto had for believing s, has been given by means of the doctor's assurance.

In addition to giving us a motivation to value INTERPERSONALISM positively, what is important to note about SURGERY is that it offers us the opportunity to revise the very idea of “epistemic reason,” given that this case seems to require distinguishing that notion from the one of “evidential reason.” In what follows we will see how the arguments against interpersonalism, especially the one developed by Lackey, by not being able to offer an adequate approach to cases such as SURGERY, show a commitment to a purist conception of justification, that is, one that assimilates the notions of “epistemic reason” and “evidential reason.” SURGERY thus fulfills a double role: on the one hand, it promotes INTERPERSONALISM (at least as the best explanation of some testimonial cases), and, on the other, it forces the incorporation of “non-evidential epistemic reasons,” that is, the adoption of a non-purist approach to justification.

4. Lackey's dilemma and its purist commitments

Lackey offers a characterization of the central thesis common to the different interpersonalist views of testimony:

[1] First (…) the interpersonal relationship between the two parties in a testimonial exchange should be the central focus of the epistemology of testimony. [2] Second, and closely related, certain features of this interpersonal relationship – such as the speaker offering her assurance to the hearer that her testimony is true, or the speaker inviting the hearer to trust her – are (at least sometimes) actually responsible for conferring epistemic value on the testimonial beliefs acquired. [3] Third, the epistemic justification or warrant provided by these features of a testimonial exchange is non-evidential in nature. (221)¹⁶

The most substantive theses from an epistemic perspective are (2) and (3). As we have seen in Section 2, in the INTERPERSONALISM defended by Moran, the interpersonal feature of testimonial exchanges gives the hearer non-evidential reasons to believe (3), which are nevertheless epistemically relevant, in the sense that they are able to affect the status of beliefs in relation to their epistemic justification (2). This means that such non-evidential reasons can, in certain circumstances, turn a previously unjustified belief p into a justified one. In this sense, a specific thesis about the status of IRs derives from INTERPERSONALISM:

NON-EVIDENTIAL EPISTEMIC INTERPERSONAL REASONS (NEEIR): IRs are non-evidential epistemic reasons.

The problem posed by Lackey is how we can explain the relation between the interpersonal aspect of the testimony that p (S offering H his or her personal assurance of the truth of the utterance) and the truth of p. According to Lackey, this problem has the form of a dilemma for any interpersonalist view of testimony:

1st horn: The view is genuinely interpersonal but epistemologically irrelevant.

2nd horn: The view is not epistemologically irrelevant, but it is not genuinely interpersonal either.

In the case of INTERPERSONALISM we can understand both horns of the dilemma as falsifying NEEIR in two different ways. The first horn states that IRs are not epistemic reasons. The second horn states that IRs are not only epistemic reasons, but also evidential ones.

The way Lackey argues for the dilemma is by showing how, first, testimonial exchanges are not enough to grant epistemic value to, or justify, beliefs that result from them. According to Lackey, there is no clear connection between the different aspects of the interpersonal relationship and the truth itself. Neither the assurance provided by S, nor his or her adoption of responsibilities in relation to the truth of p, are sufficient to reasonably suppose that p is true, at least not from an epistemic point of view. Therefore, defending that they are by themselves enough to justify beliefs enables cases of speakers who are systematically unreliable or openly inexperienced about the subject in question to grant epistemic value to beliefs through providing their personal assurance or assuming responsibilities. Nevertheless, we would not say that beliefs originated in their word are epistemically justified. This is why Lackey considers that genuinely interpersonal theories are epistemologically impotent. They cannot really explain how testimonial beliefs acquire epistemic value, since they would justify too much. Thus, Lackey concludes:

…in order for assurance to be of interest to the epistemology of testimony, it must be capable of conferring epistemic value on those beliefs acquired on the basis of testimony. But this can only be accomplished when assurance and its relatives – such as taking responsibility for the truth- are themselves properly grounded epistemically. This can be done in any number of ways: a speaker's assurance that p is true may need to track the truth – either à la Nozick's sensitivity requirement or á la Sosa's safety requirement – or be reliably grounded, or be adequately supported by the available evidence, and so on. The point that is crucial here is not the details of how assurance is connected with the truth, but only that it have a proper connection with the truth. (230)

The conclusion of the first horn of the dilemma is that, in order to explain the epistemic value of testimony, interpersonalist theories must include some conditions that connect the testimonial exchange with the truth in a proper way.¹⁷ But if we take such considerations into account, Lackey argues that the specifically interpersonal contribution becomes theoretically idle, as in no case does it explain the status of testimonially justified beliefs.¹⁸ Hence the second horn of the dilemma, which deserves special attention since several interpersonalist positions present conditions of this sort. In particular, Moran affirms when explaining how guarantees acquire epistemic value:

Of course, as with any public assumption of responsibility, the appropriate abilities and other background conditions must be assumed to be in place for it to amount to anything. For the speaker to be able to do this it must be assumed by both parties that the speaker does indeed satisfy the right conditions for such an act (e.g., that he possesses the relevant knowledge, trustworthiness, and reliability). (2006: 289, emphasis added)

Moran himself acknowledges, then, that for testimony to have some kind of value, we must suppose that certain epistemic conditions are in place, so we can suppose that there is a certain link between the assurance offered by the speaker, and the truth. So in Moran's view, INTERPERSONALISM seems to be reinforced by some reference to epistemic background conditions. Let us call this position INTERPERSONALISM+, defined as a position that, in order to explain the epistemic value of testimony, also brings into scene an evidential relation between the speech act x and p, that must be assumed to take place.¹⁹ Something like (E-1).

Lackey's objection to views such as INTERPERSONALISM+ can be summarized as follows: once we acknowledge that the evidential conditions are in place – for instance, (E-1) – what is the condition (I-1) required for? What is the additional value of the specifically interpersonal aspect, in relation to the epistemic justification of testimony-based belief? We believe Lackey's second horn points to a deep issue in relation to the status of IRs as reasons to believe that are “different in type.” If a consideration such as “S said that p” is impotent without additional evidential considerations, in what sense are IRs not evidential? If a consideration such as “H told me that p” only provides a genuine reason to believe that p when something like (E-1) is brought in, then shouldn't it be obvious that its status as a reason depends on the speech act x being evidence of p? And in that case, is it not obvious that the consideration “S told me that p” is ultimately an evidential reason to believe that p?

Although Moran demands something like (E-1), he does not allow for the consideration “S told me that p” to be interpreted in evidential terms, that is, as one based on a relationship such as the one found in (E-1). In other words, an IR (S assured me that p) is an epistemic reason to believe that p, but this cannot be explained evidentially. When H accepts S′ testimony as a reason to believe p, H is not accepting it because such testimony increases the probability of p. Moran insists, though, on a conceptual distinction between a reason being evidential and its being epistemic, that would, in fact, make room for his position (and other INTERPERSONALISM+ positions) beyond the scope of Lackey's dilemma.

Our example, SURGERY, asks for this conceptual distinction, because, as we have seen, evidentialism does not successfully explain what goes on epistemically in SURGERY. The doctor's testimony is reliable in the sense required by (E-1), but, as we have seen, there seems to be some epistemic value, over and beyond the evidential support Roberto has for p, given by the doctor's assurance that p, that makes the belief justified.

Thus, the example seems to be a strong challenge to Lackey's critique of interpersonalism, to the extent that her dilemma seems to depend on the denial of that distinction (which becomes essential for understanding what SURGERY shows). In what follows, consequently, we will address the question of how it is possible to characterize the aforementioned distinction, giving rise to non-evidential epistemic reasons. To do this, first, we must address a previous question: what makes a reason epistemic?

5. Epistemic reasons

A reason to believe p is a consideration that in some sense counts in favor of believing p. We will not address here the question of how to identify reasons in the first place. Assuming that a certain consideration intuitively counts in favor of believing p and is therefore a reason for believing p, we want to answer the question of whether this reason is epistemic.²⁰ An epistemic reason, in principle, could be explained by saying that a reason is epistemic when it favors believing p in an epistemic sense, or from an epistemically point of view. But what exactly is “epistemic”? According to Lackey,

… “epistemic” applies to a concept to indicate its being truth-conducive or otherwise intimately related to knowledge; thus, if, for instance, a practice has epistemic value, it would be one that somehow furthers the goal of acquiring true beliefs and avoiding false ones. (226)

In a first sense, an epistemic reason for believing p would be something that can work as evidence for p. Something correlated with p evidentially, pointing to p by means of increasing in some way the probability of p, in the sense that if the reason is q, the probability of p given q is higher than the previous probability of p. (This can be characterized in non-probabilistic terms, but the idea is that in some way, that consideration brings us “closer” to the truth of p in a way that can uncontroversially be described as “how facts connect with each other.”) This reading closely connects the notions of “epistemic” and “evidential,” leaving too small a margin to understand how something can be epistemically relevant without being evidential. And this is precisely what gives an initial plausibility to the dilemma presented in the previous section.

Lackey's quotation includes a second sense, that of “being in some way connected to knowledge” or some other epistemically valuable state. Schmitt seems to recover this second sense when characterizing an epistemic reason as follows:

My question is rather whether assurance or its acceptance gives the addressee an epistemic reason to believe, in the sense of a reason that contributes to the addressee's epistemic justification for believing p. (2010: 222)²¹

This reading makes the epistemic nature of a reason depend on whether it is able to affect the epistemic status of beliefs, in the sense of contributing to a belief being epistemically justified (as Lackey suggests: if it is in some way related, in this case, to justification). With this second sense in mind, we can say that while epistemic justification is a truth-conducive state, a reason is epistemic insofar as it contributes to delivering a positive epistemic status like epistemic justification.²²

Both Lackey and Schmitt believe that these senses are interchangeable. But it is actually the second sense that poses the biggest challenge to interpersonalist views, given that what characterizes interpersonal theories of testimony is precisely the thesis that interpersonal aspects are epistemologically relevant in that they have an influence on the epistemic value of at least some testimonial beliefs (i.e., they are relevant for their epistemic status as justified beliefs).²³ Lackey, for instance, shifts naturally from the second sense to the first one in a comment previously quoted in Section 4:

…in order for assurance to be of interest to the epistemology of testimony, it must be capable of conferring epistemic value on those beliefs acquired on the basis of testimony. But this can only be accomplished when assurance and its relatives (…) are themselves properly grounded epistemically. (…) The point that is crucial here is not the details of how assurance is connected with the truth, but only that it have a proper connection with the truth. (230)²⁴

This transition is natural only if one assumes that evidential considerations are the only ones capable of having an effect upon the normative epistemic status of beliefs. This is, if one assumes that evidential relations are the only ones capable of bringing proper connections with truth. If this is the case, then a reason is evidential only when it can affect the normative epistemic status of the belief, and vice versa. As we will see in the following section, this assumption lies behind purist views of justification, but purism is not the only option available.

So, to recapitulate: up to this point we have shown that SURGERY seems to give intuitive plausibility to the idea, explicitly defended by Moran (2006), that IRs are epistemic yet non-evidential reasons (NEEIR). By taking Lackey's dilemma into account, we managed to frame NEEIR in a more general conception of what makes a reason epistemic. In this section we have shown that there are two readings of “epistemic”: an evidential reading, and a broader reading that focuses on the impact the reason might have on the epistemic status of the belief, showing that only under the first reading views such as INTERPERSONALISM+ are incoherent. But this is not enough to show that this distinction is going to help the interpersonalist with its core theoretical thesis. Nothing has been said yet that allows us to understand why is it that IRs can be epistemic reasons. For the remainder of this work, we will try to address this by bringing non-purist theories of justification into the picture.

6. Non-purist theories of justification

In the previous section we saw that, if we suppose evidential reasons are the only relevant ones when it comes to establishing the epistemic status of beliefs, then in order to evaluate if a reason for believing p is epistemically relevant, it is correct to ask if it brings p in some way closer to the truth in a proper evidential way. From this perspective, Lackey's arguments are conclusive.²⁵ This is the case because the existence of epistemic but non-evidential reasons would be contradictory from the get-go. However, SURGERY offers a clear case that not only calls into question EVIDENTIALISM as a conception of testimony, but, more basically, the identification between evidential reasons and epistemic reasons assumed in the formulation of Lackey's dilemma.

In this section, we will show how INTERPERSONALISM+ can begin to make a proper articulation with a more general epistemic position that rejects that only evidential considerations are epistemically relevant. The distinction between “purist” and “non-purist” theories of justification enters center stage.

“Purism” regarding epistemic justification can be characterized as follows:

PURISM J (PJ): Necessarily, if two people have the same evidential support for p, then p is epistemically justified either for both of them, or for neither.²⁶

Although this is the most widely defended position, we can still find its rejection in the literature. “Non-purism” about justification can be characterized as the denial of PJ:

NON-PURISM J (NPJ): It is possible that two people that have the same evidential support for p, differ as to their epistemic justification of p.

There are many ways of making sense of the non-purist claim,²⁷ but one salient version of it, pragmatic encroachment, is known for arguing for a pragmatic condition for justification and/or knowledge, that links epistemic justification and action by means of extracting some specific consequences from the ACTION-JUSTIFICATION principle (which we have already seen in Section 3).

A purist reading of the principle will put all the weight of ACTION-JUSTIFICATION on the action side. Purists see it as a principle that states conditions for rational action, not for justification. Something like “no rational action based on p without justified belief,” leaving the justification of the belief untouched by anything else other than evidence.

Theorists of pragmatic encroachment (Fantl and McGrath 2002, 2009, 2012; Hawthorne and Stanley 2008; Stanley 2005, among others²⁸) broaden the philosophical meaningfulness of the relation between belief and action, as they understand that not only is it the case that to evaluate the status of an action we must pay attention to the epistemic “quality” of the mental state in which it is based: it is equally true that to assess the epistemic “quality” of a mental state, we must evaluate the possible actions related to it.²⁹ Their reading of ACTION-JUSTIFICATION has as a consequence that not only evidential considerations will be epistemically relevant, since in order to evaluate the epistemic status of a belief we have to take into account everything that might have an impact on the normative status of actions (such as those related to pragmatic context, people's preferences, their aversion to risk, etc.). The movement toward this reading is supported by arguments that show that if we assume purism about justification, we cannot make sense of some basic fallibilist intuitions. So, at least for fallibilism, non-purism has to be true. This reading of ACTION-JUSTIFICATION can be translated into a PRAGMATIC CONDITION for justification, which can be defined in several ways, such as:

(PC): A belief p is justified for S only if improving the epistemic support for p is pointless from a rational action perspective.³⁰

Pragmatic encroachment has generally been understood broadly as a theory that not only argues that a pragmatic condition for justification is in place, but also that there is only one specific way in which pragmatic factors have an impact on justification: by lowering or increasing the amount of evidence needed for the belief to be justified.³¹ We can call this position evidentialist pragmatic encroachment view (EPEV). We believe that EPEV is not mandatory, since nothing in the argument against purism, or in PC, implies it.

In fact, SURGERY could be seen as showing how EPEV is too restrictive. It is important to note that the viability of SURGERY depends in principle on the adoption of some form of pragmatic encroachment, at least understood merely as involving the idea that a higher risk situation demands more reasons for a subject to achieve sufficient epistemic justification as the basis for her rational action. Indeed, it is the high risk implied by the possibility of having the disease in question, which makes Roberto require more reasons than the evidential one given by the observation of the doctor's behavior in front of the filing drawers. This type of admission of pragmatic encroachment is not controversial in principle if what is concluded from it is that the greater the risk the greater the evidential reasons required (i.e., EPEV). But what SURGERY motivates is a commitment to the claim that, at least in certain circumstances and for certain rational agents, the greater risk imposes a need for more reasons that is not met with more evidence but with IRs (e.g., doctor's assurance) that do not provide further evidence. Consequently, SURGERY calls for a transition toward a position that not merely involves the type of pragmatic encroachment mentioned above, but that also incorporates the thesis that there are situations where the epistemic reasons that make a difference in justification are not evidential ones. We can call this position, the one that admits both evidential and non-evidential epistemic reasons, the broad pragmatic encroachment view (BPEV). Thus, since there are many “pragmatic encroachers” we have to be very explicit as to what is it that we consider helpful here from pragmatic encroachment philosophers: their arguments for the existence of a pragmatic condition like PC for justification. Nothing else. Once PC is admitted, SURGERY drives us toward BPEV, surpassing the more restrictive EPEV.

SURGERY allows us to assume the claim that the IR has had an impact on the belief's epistemic status. According to what we have been defending, Roberto was not justified in believing s before receiving the doctor's testimony, and he is after doing so. He was not justified because, before the doctor's testimony, condition PC was not in place (adding more epistemic support for s would change the rational course of action, say from “look for more epistemic support” to “do not undergo surgery”), and after the testimony it is (more epistemic support for s would not change the decision not to undergo surgery). SURGERY thus shows that, at least in some cases, IRs are able to modify the epistemic status of beliefs in relation to their justification. There is a clear sense in which IRs could be epistemically relevant reasons even though they are not evidential ones: that is, if they had an impact on the epistemic justification of someone by virtue of being relevant to the course of rational actions to be pursued. This shows that, according to a non-purist theory of justification, IRs have an epistemic value indeed. Thus, SURGERY leads to BPEV by requiring the assumption of NEEIR for an adequate explanation of how PC is achieved only after the physician's testimony.

Now, to emphasize the point that we have been defending, it is important for us to highlight that the type of non-purism that SURGERY seems to motivate, the one that assumes the acceptance of the existence of non-evidential epistemic reasons, should not be confused with some ways of alluding to pragmatic reasons as epistemically relevant. In particular, it should not be confused with the pragmatism defended by Maguire and Woods (2020), which argues for the existence of pragmatic reasons (which have to be differentiated from epistemic reasons) that play a normative authoritative role in the “belief game.”³² Indeed, according to Maguire and Woods, it must be distinguished between reasons for having or not having a belief that is epistemically correct, and reasons for the epistemic correctness or incorrectness of a belief. Only the first reasons have an authoritative role and, according to the authors, they are always of a practical nature. Without those kinds of reasons there is no point in playing the epistemic game; but once inside the game, it is only the evidential reasons that play the relevant role. In this way, the pragmatism of Maguire and Woods continues to be evidentialist after identifying which reasons are, strictly, epistemic reasons.

For our part, regardless of our adhesion or not to the defense made by the authors of the existence of authoritative reasons that are of a purely practical nature, what we show, through SURGERY, is something stronger than what their pragmatism defends: that the IR that Roberto obtains is clearly epistemic in the sense of Maguire and Woods; it is not the case that it fulfills the authoritative role that the authors emphasize for practical reasons. What SURGERY shows is that there are epistemic reasons that are not identifiable with evidential reasons. In this way, the non-purism to which SURGERY leads us, the one that assumes that there are non-evidential reasons to believe and, consequently, assumes NEEIR, is not incompatible with the pragmatism outlined by Maguire and Woods, since, in principle, it is neutral with regarding whether or not to accept the existence of the authoritative reasons to which pragmatism alludes (given that they are in turn distinguishable from the recognition of the impact of risk for the requirement of epistemic reasons, which is assumed by our non-purism in question given our adherence to BPEV). This non-incompatibility disappears if pragmatism implies that only evidential reasons can play the epistemic role within the justification practice. That sort of pragmatism would imply a rejection of NEEIR, leaving the phenomenon described in SURGERY unexplained, and thus becoming a position as affected by the argument of this paper as evidentialist purism and EPEV.

We do believe, however, that our position is quite close to Schroeder's (2021),³³ even though he explicitly denies that strong principles like ACTION-JUSTIFICATION are true (so he wouldn't actually accept PC straightforwardly and technically his position would not be a case of BPEV) and, consequently, our argument is independent of his position. Indeed, his Pragmatic Intellectualism might be understood as an example of non-purism, and his approach toward epistemic reasons and their role in justification and rational belief might illuminate the way IRs can be epistemic. If (just as Schroeder argues) the way epistemic reasons counterbalance each other is what determines how and when a belief is rational or justified, his position entails the admission of non-evidential epistemic reasons. In particular, Schroeder defends that, for example, risk has epistemic relevance by being an epistemic reason to withhold belief. Still, he does not accept that there are non-evidential epistemic reasons for believing (just for withholding belief). In this way, his non-purism differs substantially from ours, which, as we have seen, relies on SURGERY as an example that can only be understood by accepting that there are non-evidential reasons for the adoption of a belief. Thus, we believe, Schroeder's rejection of non-evidential epistemic reasons for believing makes his version of non-purism face a series of difficulties when explaining cases like SURGERY. Complementing Schroeder's perspective with our own, we believe that we can finally arrive at the best way to describe SURGERY, this is as a case in which the doctor's assurance is a reason that has the power of surmounting (together with all the evidential reasons Roberto already has) the epistemic reason to withhold belief associated with the risk of the situation.

7. Conclusion

The debate between evidentialist and interpersonalist views can be described emphasizing either one of the two tasks that any epistemological theory of testimony must perform: first, specifying what the nature of that phenomenon is, including the aspects (psychological, interpersonal, social, etc.) that enable successful testimonial exchanges; and, second, providing an account of the conditions under which these aspects allow for testimonial beliefs to have a positive epistemic status, such as justification or knowledge.

Throughout the work we have focused on the second task, and we have defended the epistemological credentials of interpersonalism, realizing that it is inseparable from the acceptance of non-evidential epistemic reasons to believe, which demands proper conceptual elaborations on the notions of epistemic reasons and of epistemic justification. We offered a proper reading of epistemic reason, and we have also defended non-purism on justification as the adequate way to conceive the epistemic proposal of interpersonalism on testimony, realizing that only this combination is capable of apprehending certain cases in which there seems to be no way to rule out the idea that the assurance offered by the testifier offers an epistemic reason to believe that it is not evidential.

We want to finish this text with two final observations that leave open questions to work on in the future.

1. (1) Is it correct to point out that for every act of testifying a possible non-linguistic behavior of the testifier can be thought of (in the current or in another possible world) with the same evidential value as her act of telling (something analogous to the behavior of the doctor in front of the drawers in SURGERY)? If so, would this imply that the value of interpersonalism is not restricted to capturing some specific cases of testimony (such as that of SURGERY) but to all of them?

2. (2) As for the theoretical task of specifying the nature of testimony, which is the other goal of a proper theory of testimony, would it be useful to pay attention to cases like SURGERY as a way of preferring interpersonal theories over evidential ones? Giving an adequate account of the nature of testimony and its role in the epistemic economy of social life seems crucial as a touchstone when it comes to choosing between theories.

With this in mind, we can consider the following case:

BASKETBALL. Peter is in the same circumstances as Roberto in SURGERY, with only one difference: while Peter is in the doctor's office waiting for his diagnosis, a basketball game is taking place which Peter has some kind of interest in watching on TV because it looks like it will be an interesting game, although he is not particularly keen on that sport. When he sees the physician taking his file from the drawer on the right, he is still doubting whether or not to undergo surgery, and he decides to wait for the physician's words. At that moment he receives a message from a friend, who tells him that the game is indeed very interesting. With that information, Peter considers that the physician's behavior he saw is enough for him. He cancels his appointment, and leaves the office to watch the game at home, having decided not to undergo surgery.

We believe that BASKETBALL (although there can be doubts about the rationality of Peter's final decision) leaves space for us to evaluate patient's behavior as, at least, very strange. That Peter's shift from indecision to decision, from needing the physician's word to being indifferent to it, is based on undertaking an action with a very low degree of subjective utility (even in circumstances where s were the case), seems enough to suspect of him at least as a participant in social life regarding testimonial exchanges of information. His conduct would seem to go against the role that testimony typically plays in our lives. A theory of testimony unable to let us consider Peter's behavior as strange would probably have difficulties in explaining the nature of testimony.

An adequate development of the questions formulated in (1) and of the intuition outlined in (2) demands, of course, a deep investigation, which exceeds the scope of this text.³⁴