Author Information: Jeremy Fantl, University of Calgary, firstname.lastname@example.org
Fantl, Jeremy. “Interest-Relativity and Testimony.” Social Epistemology Review and Reply Collective 4, no. 11 (2015): 40-46.
Please refer to:
- Freedman, Karyn L. “Testimony and Epistemic Risk: The Dependence Account.” Social Epistemology 29, no. 3 (2015): 251-269.
- Kukla, Rebecca. “Commentary on Karyn Freedman, “Testimony and Risk: The Dependence Account.” Social Epistemology Review and Reply Collective 3, no 11 (2014): 46-52.
- Freedman, Karyn L. “Group Accountability Versus Justified Belief: A Reply to Kukla.” Social Epistemology Review and Reply Collective 4, no. 7 (2015): 6-12.
- Kukla, Rebecca. “A Further Look at Standards of Justification.” Social Epistemology Review and Reply Collective 4, no. 9 (2015): 63-65.
Image credit: smilla4, via flickr
In her “Testimony and Epistemic Risk: The Dependence Account”, Karyn Freedman defends an interest-relative account of justified belief and suggests that the account can contribute to literature on testimony. According to her interest-relative account, your interests in whether p is true can make a difference to whether you justifiedly believe that proposition. Freedman distinguishes her account from earlier versions by allowing a distinctive role for emotional interests and how much we care about whether p is true. 
On the Interest-Relative Account
The account is relevant to the literature on testimony because it helps explain what is intuitively forceful about both credulist and reductivist accounts of the epistemic status conferred by testimony, while still allowing us to decide in favor of one of the two sides. On Freedman’s construal, according to credulism, absent defeaters, a testifier’s say-so that p suffices for a hearer’s belief that p on that say-so to be prima facie justified. According to reductivism, a testifier’s say-so that p is never sufficient for belief that p on that say-so to be prima facie justified: what is required in addition is some non-testimonial reason to believe the testimony.
According to Freedman, the interest-relative account predicts credulist intuitions about cases in which you have minimal interest in the truth of p or have low epistemic risk in believing that p (what I will call “LOW” cases), and reductivist intuitions about cases in which you have strong interest in the truth of p or have high epistemic risk in believing that p (“HIGH” cases). Because credulists tend to support their view using LOW cases while reductivists tend to support their views using HIGH cases, the interest-relative account explains why the cases for each are compelling. However, ultimately the interest-relative account favors reductivism because credulism does not sufficiently respect the requirement that our testimony-based beliefs be supported by evidence. Even in LOW cases, they must be supported by evidence—just not by as much evidence as in HIGH cases.
In an exchange with Freedman in this journal, Rebecca Kukla draws attention to some potential challenges facing Freedman’s view, though does not explicitly address Freedman’s application of the interest-relative view to the debate between reductivists and credulists. I address that point, here, as well as a smaller point that is up for disagreement between Freedman and Kukla regarding how distinctive Freedman’s interest-relative account is, and a larger area of debate between Freedman and Kukla about whether on Freedman’s account we need to know what the interests of our informants are before accepting their testimony.
First, on the smaller point: Kukla worries that Freedman’s employment of emotional concern does not make her interest-relative account distinctive. That’s because if I care about whether p is true, presumably that will have ramifications for how I am rational to act on the assumption that p is true. If I really want the Patriots win the Super Bowl, I will act differently on the assumption that they’ll win than if I don’t care about whether the Patriots will win the Super Bowl. Freedman replies that these differences in how I act do not explain why the standards of justification are lower when I don’t care whether they win.
I think Freedman takes into account emotional responses in a distinctive way. Kukla is right that many advocates of pragmatic encroachment have the resources to allow emotional concern to play a role in determining the threshold of justification. One way is the way Kukla suggests. A second way is to just stipulate that having an emotional response appropriate to the truth of p must be rational if one is to be justified in believing that p. This is what Fantl and McGrath do say in 2009: what you know (and are justified in believing) “is warranted enough to justify you in believing, doing, feeling, wanting, liking, hating, or intending anything at all” (2009, 66). But at times Freedman suggests a very different way that emotional concerns matter. That is, the amount of justification you need to be justified in believing that p depends on how much you care about whether p is true. This takes a slogan suggested by Jason Stanley quite literally (though about justification rather than knowledge): “The more you care, the less you know.” This slogan has always been misleading when it comes to most of the current proponents of pragmatic encroachment. But it is quite apt when it comes to Freedman’s view.
The difficulty is that, in being distinctive in this way, Freedman’s view runs into some objections that the more usual way of construing interest-relativity avoids. First, it seems that sometimes caring much more deeply about something can plausibly make that thing more justified for us. Judith Baker (1987), for example, argues that trust in friends is epistemically rational even when that trust “outruns the evidence” (5). But, plausibly, we care much more, say, about whether a person is innocent of some crime if they’re our friend than if they’re not. Therefore, it becomes easier—if rationality can be equated with justification in this case—to believe that our friend is innocent than it is for someone unconcerned with our friend. For Freedman, though, it seems like this would have to be false. Because we care more about whether our friend is innocent, we’d need more evidence in order to believe.
More importantly, Freedman’s view is subject to a puzzle presented by Robert Howell. Howell points out that propositions that we don’t care about at all entail propositions we care very much about. For example, to modify Howell’s example, that my sister is at work entails that my sister isn’t dead (2005, 132). I care very much about the second and don’t care at all about the first. Therefore, the standard of evidence on Freedman’s account is higher for the second than it is for the first. Suppose I don’t have enough evidence that my sister is alive to satisfy the higher standard it must be subject to, but I do have enough evidence to satisfy the standard that applies to the proposition that my sister is at work. Then I will be justified in believing the first, but not the second, even though I know that the first entails the second. This is a violation of at least one epistemic closure principle. While some seem content to deny closure principles, this seems like a particularly egregious violation and, in any case, one would hope that simply adopting an interest-relative account of justification wouldn’t force such violations onto us.
The alternative account of interest-relativity is not subject to this concern. On the alternative account, what’s required for justified belief is that you’re rational in continuing on (emotionally, practically) as if what you justifiedly believe is true. But I am rational in continuing on right now as if my sister isn’t dead, just as I am rational in continuing on right now as if my sister is at work. To act, emote, feel, believe as if either one of them is true is to, here and now, carry on typing out this response to Kukla and Freedman. And that’s exactly what I’m rational in doing.
Credulists, Reductionists and Testimony
The more central part of Freedman’s original paper is the connection between the interest-relative account of justification and the ongoing dispute between credulists and reductionists about testimony. Again, according to Freedman, reductionists tend to support their conclusion with HIGH cases. On the interest-relative account, in these cases, more evidence is needed than simple say-so of our testifier. Therefore, the interest-relative account predicts that the reductivist conclusion is true: we need independent confirmation of the reliability of our informant. But in the LOW cases that the credulist offers us we’ll expect much less evidence to be needed, if the interest-relative account is true and so we’ll have the intuition that the simple say-so of our informant is enough. That’s exactly what we find our intuition to be. As a proponent of an interest-relative account of justification myself, I agree with much of what Freedman says here, and am very sympathetic to the extension of the view to the cases brought forward in support of reductivism and credulism. It’s a terrific insight about why the seemingly conflicting kinds of cases are both so compelling.
Freedman goes on to argue that, while the interest-relativity of justification explains our credulist intuitions regarding LOW cases, we should not conclude that credulism is true. Even though interest-relativity tells us that less evidence is needed than when our interests are pressing, it’s still true on the view that evidence is needed. It’s just that, in those cases, the evidence presented by the mere say-so of our informant is enough. According to Freedman, “credulist accounts are… an inadequate answer to the central epistemic question regarding the normative requirements for beliefs based on testimony, because they overlook the evidence condition that is a basic requirement to justification.” (2015a, 266)
But do credulist accounts overlook the evidence condition that is a basic requirement to justification? Credulists aren’t committed to the position that the testimonial contributions of our informants aren’t evidence. After all, credulists think that testimonial contributions are on par with perceptual and introspective contributions. But perceptual and introspective contributions count as evidence—perceptual evidence and introspective evidence, respectively. And testimonial contributions do too.
When Freedman says that there is an evidence condition on justified belief based on testimony, she might mean that justified belief based on testimony requires evidence of some kind or other, or that justified belief based on testimony requires some corroborating non-testimonial evidence. Her words indicate that she means the former:
On the account of justification advanced here, a justified belief is one that depends on evidence for support. This is true with respect to all of our beliefs, not just those based on testimony. Take perception… If I see a cat on the mat, for instance, then in the usual case my belief that there is a cat on the mat will be justified based on the evidence provided by my observation that this is so. Likewise, if I am told that there is a cat on the mat, then in the typical case my belief that there is a cat on the mat will be justified based on the evidence provided by the assertion” (Freedman 2015a, 264).
Here Freedman seems to say that in both perception and testimony, the evidence sufficient to satisfy the evidence condition on justified belief is the perceptual or testimonial evidence itself. With regard to neither source (in the LOW case) do we need independent evidence of perception’s or testimony’s reliability. But, then, how is Freedman’s view in more in keeping with reductivism, which requires that “beliefs based on testimony are justified only on the basis of non-testimonial beliefs” (Freedman 2015a, 252)? True, Freedman distinguishes her view from reductivism, calling it the “Dependence Account”. But if her Dependence Account allows justified belief based on testimony in LOW cases without independent non-testimonial evidence, then we have to ask why her account is supposed to be closer to reductivism than to credulism. Unlike reductivism, her account allows justified belief based on testimony without non-testimonial evidence. Unlike credulism, what? Her answer is that unlike credulism, her account requires that beliefs based on testimony be based on evidence. But, as we’ve seen, there’s nothing stopping credulists from thinking of say-so as a form of evidence. So, if anything, her view seems more like a credulist view than a reductivist one.
On the other hand, Freedman could think of the evidence requirement as a requirement that beliefs based on testimony be—even in LOW cases—supported by non-testimonial evidence. But then her account is not supported by the interest-relative account of justification. For the interest-relative account delivers, if anything, only the verdict that mere testimony allows for justified belief in LOW cases but not in HIGH cases. It tells us nothing about whether, in LOW cases, non-testimonial evidence is required for justified beliefs.
A different way to construe credulism is as the view that the mere say-so of an informant provides defeasible prima facie justification for the proposition testified to. This contrasts with an alternative construal of reductivism: for the say-so of an informant to provide defeasible, prima facie justification for the proposition testified to, we must have non-testimonial evidence that the testifier is reliable. Where each construal differs from Freedman’s is that while Freedman’s construals are about outright justification, these construals are about degrees of justification.
On these construals, I don’t see how the interest-relative account of justification negotiates between these two views or weighs in favor of reductivism. If anything, it seems to weigh in favor of credulism. For, it seems, in LOW cases, all that’s needed for justification is the say-so of the informant. As Jessica Brown (2014, 187-188) observes, the interest-relative account of justification had better not ever allow the standard for justified belief to be so low that any old proposition whose truth we don’t care about gets to be justifiedly believed. Therefore, in LOW cases the say-so of informants has to provide a fair amount of defeasible, prima facie justification—enough so that, in LOW cases, belief is prima face justified.
Credulists, of course, can agree with the interest-relative account that in high stakes cases, more evidence is necessary than is provided by testimony alone. It is not essential to credulism that testimony, prima facie, provide the highest possible degree of justification. It is consistent with their view that the justificatory support provided by testimony can be increased with the addition of further information about the reliability of the source. So the fact that testimony in HIGH cases is not sufficient for justified belief, given the interest-relative account, does not tell against credulism.
But in fact it would be a mistake to infer either credulism or reductivism from all of this, because, as pointed out, the interest-relative account tells us nothing at all about how to treat justified belief in LOW cases. Credulists say that the substantial amount of justification conferred by testimony in LOW cases comes from the mere say-so itself. Reductivists say that the substantial amount of justification comes from independent, non-testimonial evidence. The question is whether testimonial contributions only confer justification because of independent non-testimonial evidence. I don’t see how the interest-relative account of justification helps us resolve this question. Each side can take the account on board.
On Knowing the Testifier’s Interests
Finally, I’d like to take up the discussion between Kukla and Freedman surrounding the issue of whether in order to be justified in believing the say-so of someone else, on the interest-relative account, you must know what the testifier’s interests are. It looks like this is necessary because, as Kukla points out, if a testifier’s interests in whether p are weaker than the hearer’s, then the hearer might well believe that p—and be justified in doing so—on the basis of worse evidence than the hearer would be satisfied by. Worse, because there is an interlocking network of testifiers that stand behind any single instance of testimony, it looks like we’ll have to know what all the relevant interests are of each link in that chain.
I’m not sure to what extent this is true. Certainly, to know whether each person is justified in believing what they say, on the interest-relative account, you might need some evidence about what their interests are. To an extent we do know such things. Most people’s interests in mundane propositions can be justly assumed to be approximately the same. Some people have bizarre interests, of course, but I don’t see why we have to be totally at a loss when it comes to what the standards are for justified belief in any individual case. But more importantly, to trust someone’s testimony, we don’t have to know whether they’re justified in believing. We just have to assume that they were justified in asserting. And the norm for justified assertion need not be the same as the norm for justified belief.
This is particularly so if an interest-relative account of justified belief is true. If an interest-relative account of justified belief is true, then this would have to be accounted for in socially approved norms of assertion. We would expect socially approved norms of assertion to require that we only assert when we have enough evidence that the assertion can be spread out into the world and attended to by people with a range of interests. All that’s necessary for a hearer to gain knowledge through testimony, then, is what this default asserter-expectation is. If you know that your standards are higher than what’s required by the general asserter-norm, then you’ll take that under advisement when listening to a testifier.
Of course, there are certain kinds of facts, and certain fora on which facts are spread where people routinely violate this socially approved standardizing norm. Social media and the internet generally is one such forum. Politically charged claims are one such kind of fact. But it’s not like we don’t become aware of these difficulties and so learn what kind of evidential weight to give the testimony that comes out of such fora and to testimony in support of such facts. The system isn’t perfect, of course, and it might be that justified belief becomes very difficult with respect to certain kinds of facts coming out of certain kinds of fora. But that seems to be exactly the right result.
Baker, Judith. “Trust and Rationality.” Pacific Philosophical Quarterly 68 (1987): 1-13.
Brown, Jessica. “Impurism, Practical Reasoning, and the Threshold Problem.” Noûs 48, no 2 (2014): 179-192.
Fantl, Jeremy and Matthew McGrath. “Evidence, Pragmatics, and Justification.” The Philosophical Review 111, no. 1 (2002): 67-94.
Fantl, Jeremy and Matthew McGrath. Knowledge in an Uncertain World. Oxford, Oxford University Press, 2009.
Freedman, Karyn. “Testimony and Epistemic Risk: The Dependence Account.” Social Epistemology 29, no. 3 (2015a): 251-269.
Freedman, Karyn L. “Group Accountability Versus Justified Belief: A Reply to Kukla.” Social Epistemology Review and Reply Collective 4, no. 7 (2015b): 6-12.
Hawthorne, John . Knowledge and Lotteries. Oxford, Oxford University Press, 2004.
Howell, Robert. “A Puzzle for Pragmatism.” American Philosophical Quarterly 42, no.2 (2005): 131-136.
Kukla, Rebecca. “Commentary on Karyn Freedman, ‘Testimony and Risk: The Dependence Account’.” Social Epistemology Review and Reply Collective 3, no. 11 (2014): 46-52.
Kukla, Rebecca. “A Further Look at Standards of Justification.” Social Epistemology Review and Reply Collective 4, no. 9 (2015): 63-65.
Stanley, Jason (2005). Knowledge and Practical Interests. Oxford, Oxford University Press, 2005.
 See, for example, (Hawthorne 2004), (Stanley 2005), (Fantl and McGrath 2009), and (Fantl and McGrath 2002).
 It’s not clear whether this new construal of credulism is an answer to the second of the two questions Freedman distinguishes between. What she calls the “central epistemological question” (Freedman 2015a, 252) is the first:
1) Does testimony, absent defeaters, prima facie justify beliefs formed on that basis?
The second, she says, is different:
2) Is testimony a fundamental source of knowledge on par with perception and introspection?
On this new construal of credulism, it is neither an answer to question 1 (because it only says that some degree of justification is conferred), nor is it an answer to question 2 (because it only says that testimony is a fundamental source of justification, not knowledge). To make it more of an answer to question 1, the new construal could be supplemented with the claim that the degree of justification conferred by testimony, absent defeaters, is sometimes sufficient to prima facie justify beliefs formed on that basis.
Categories: Critical Replies