2.1 The Triviality Intuition

According to ET, it is necessarily good to conform to epistemic norms. However, an obvious worry with this suggestion is that we can easily imagine cases where there seems to be nothing good in believing the truth, avoiding error, following the evidence, and the like.

First, we can imagine cases where believing in conformity with epistemic norms would be overwhelmingly bad and would not lead to anything good (and where violating epistemic norms would be overwhelmingly good and would not lead to anything bad). Here is one such example:

Second, there are cases where there is seemingly nothing bad and a lot of good in violating epistemic norms as well as nothing good in conforming to epistemic norms, given the triviality of the matter at hand. Here is one example:

Third, there are cases where there is seemingly nothing good in conforming to epistemic norms as well as nothing bad in violating them because of the triviality of the matter at hand. For example:

We could multiply the examples. There does not seem to be necessarily something good in believing the truth, avoiding error, or following the evidence concerning, e.g., the number of blades of grass on your neighbour’s lawn at 11:59 am on 11 May 2016, how long it took the 14^th customer served today at the third most popular coffee shop in New Jersey to drink her coffee, the average number of threads in the carpets of all the blue houses in Ireland, your grandfather’s favourite sexual position when he was 39 years old, and so on. The lesson of such triviality cases is what I will refer to as the ‘triviality intuition’:

Of course, the triviality intuition alone does not suffice to falsify ET. However, it does pose a pressing explanatory challenge or question for teleologists. ^{Footnote 7} If, as ET states, it is necessarily good to conform to epistemic norms, then why are there cases where there is seemingly nothing good in doing so? ^{Footnote 8} The triviality intuition directly challenges ET. So teleologism is only viable if it can give a satisfactory answer to that challenge.

2.2 The Challenge: Explaining the Triviality Intuition

Before examining teleologists’ most promising response to the triviality challenge, it is necessary to first rule out one natural, but inadequate, answer. Call it the ‘epistemic value’ explanation:

Although initially tempting, this answer won’t do.

Merely pointing out the necessary epistemic value of conforming to epistemic norms is not sufficient for vindicating ET and epistemic normativity. To see this, note that when it comes to value and value claims, we need to make a distinction that is analogous to the one between genuine normativity and mere norm-relativity. Call this the distinction between ‘genuine’ or ‘normativity-grounding’ value on the one hand, and mere ‘domain-relative’ value on the other. The former is goodness, which entails normativity and normative reasons. It is the kind of value that is such that if ϕ-ing is good in that sense, then there is a normative reason to ϕ. The latter only implies the trivial claim that something is good relative to or according to the norms of a particular domain. Unlike genuine goodness, mere domain-relative goodness alone does not entail genuine normativity. The sole claim that ϕ-ing is good relative to some domain does not entail that there is a normative reason to ϕ.

For example, there is always something legally good in respecting the law and something legally bad in breaking it. Conformity to the law, in other words, is trivially good from a legal standpoint and violating the law is bad from a legal standpoint. However, it is a further question whether it is necessarily genuinely good to do what is legally good. Similarly, unfashionable outfits are, trivially, bad from a fashion standpoint and fashionable outfits are good from a fashion standpoint. They respectively have fashion disvalue and value. But it is a further question whether it is genuinely good to do what has fashion value and to avoid being unfashionable.

In the same way, it is trivial to claim that there is necessarily something epistemically good in conforming to epistemic norms. Obviously, violating epistemic norms is bad from an epistemic standpoint. But this only vindicates domain-relative value. That is, it only shows that conformity to epistemic norms is good relative to the epistemic domain. Even if we admit that, it is a further question whether it is always genuinely good to do what is epistemically good. It is a further question, in other words, whether epistemic value necessarily constitutes genuine, normativity-grounding value.

At this stage, I am not (yet) claiming that conforming to epistemic norms is not necessarily genuinely valuable. Instead, I am making two points. First, merely associating epistemic norms with epistemic value leaves open the question whether it is necessarily genuinely good to conform to epistemic norms. Second, the kind of goodness that is relevant to ET and epistemic normativity is normativity-grounding goodness, not mere domain-relative goodness. Hence, the more precise definition of ET is the following:

Epistemic value, in other words, necessarily constitutes normativity-grounding value according to ET. However, what the triviality intuition suggests is precisely that it is not always genuinely good to believe what is epistemically good to believe. This is why the epistemic value explanation won’t do as a response to the triviality challenge. ^{Footnote 9}

A more promising avenue for teleologists is to invoke the difficulty to tell whether something really has no value at all or if it just has very little value. Call this the ‘minimal value’ explanation:

Simply invoking the minimal value explanation is not sufficient for getting teleologists off the hook, however. They must also show that such an explanation is appropriate in the case of epistemic norms. This is because one could invoke the same kind of explanation to defend the necessary authority any norm, including norms that clearly lack necessary normative authority.

Consider examples like the law or etiquette. In response to cases where there is seemingly nothing genuinely good in conforming to the law or etiquette—e.g., cases of immoral, absurd, or obscure laws or etiquette rules—one could give a minimal value explanation and say that these are simply cases where conformity to the law or etiquette is only very minimally good, which is easy to confuse with no value at all. But such an explanation would obviously not be sufficient for vindicating the necessary genuine value of following the law or etiquette. Hence, in addition to invoking the minimal value explanation, teleologists must tell us why, unlike in the case of laws or etiquette, that explanation is warranted in the case of epistemic norms.

In the rest of this paper, I will argue that the minimal value explanation is not warranted in the epistemic case. ^{Footnote 10} It is not plausible to think that necessarily, conforming to epistemic norms is minimally genuinely good. ^{Footnote 11}

3.1 Minimal Prudential Value

Something can be good in the sense of being prudentially good or good for someone. That is, something can be good in the sense that it makes one’s life go well and contributes to one’s well-being or interests. Therefore, a first possible interpretation of the minimal value explanation is a prudential version, which holds the following thesis:

As the examples from Section 2 already indicate, however, it is clearly not always good for us to believe the truth, avoid error, follow the evidence, and the like.

In Dream, the matter at hand is so trivial and disconnected from any of Vincent’s interests that conforming to epistemic norms and suspending his judgement about whether the event in question really happened would have had absolutely no impact on his well-being. Similarly, in Even stars, Myriam’s violating epistemic norms and believing that the number of stars is even only has positive impacts on her well-being and interests. Having that epistemically unjustified belief seems to only make her life go better. Finally, in examples like Illness, not only would violating epistemic norms be overwhelmingly in Rita’s interests, conforming to epistemic norms would be highly detrimental to her well-being. Believing what is epistemically justified for her to believe would only make her life go dramatically worse.

One possible reply is that, although it is not always actually good for us to conform to epistemic norms, it is always potentially so. It might be claimed, for example, that conforming to epistemic norms necessarily provides us with potential premises for successful good actions. It might also be that violating epistemic norms is always potentially bad for us because, for example, any false belief is a potential obstacle to successful good actions. Could teleologists ground epistemic normativity in such potential prudential value? There are three problems with this strategy.

First, this response shows, at best, that conforming to epistemic norms is always potentially good. But this does not mean that there is necessarily a normative reason to conform to epistemic norms. Rather, it means that there is always potentially a normative reason to conform to epistemic norms, which falls short of categorical epistemic normativity. The fact that there might be a reason for you to ϕ does not mean that there is a reason for you to ϕ.

Second, the argument proves too much. Even if we concede that conforming to epistemic norms is always potentially good for us, violations of epistemic norms are also potentially good for us. Even if my belief that my train is leaving at noon is false and epistemically unjustified, there could be a schedule change to noon at the last minute, in which case my unjustified belief would allow me to catch my train. Similarly, any true or epistemically justified belief is also potentially harmful. Therefore, if the potential usefulness of conforming to epistemic norms sufficed to vindicate epistemic normativity, it would also establish, implausibly, that there is necessarily a normative reason to violate epistemic norms.

Third, and most importantly, it is not even the case that conforming to epistemic norms is always potentially good for us. Imagine that in Dream there is an evil demon that will kill Vincent right away if he conforms to epistemic norms and suspends his judgement about the question at hand. In such a situation, it is not even potentially good for Vincent to conform to epistemic norms. And yet, suspending judgement about the question at hand is still the epistemically justified doxastic attitude for Vincent.

Another common line of defence of minimal prudential value is to construe ET not as a case-based thesis—i.e., not as the claim that conforming to epistemic norms is good for us in all particular cases—but rather as a rule-based thesis about what policies or rules of belief formation are good to adopt in the long run. ^{Footnote 12} Epistemic normativity, on this view, is grounded in the prudential value of a policy of systematically conforming to those norms. So, even if we can be in particular cases where conforming to epistemic norms is not good for us, it remains good for us, over time, to systematically conform to epistemic norms, as a matter of general rule or policy. ^{Footnote 13}

This rule-based strategy is not promising for two reasons. First, for agents like those in triviality cases, a policy of always conforming to epistemic norms except in this one triviality case would clearly be prudentially better than a policy of systematically conforming to such norms. Even in the long run, they would be better off following the former policy since it would give them the benefits of the latter without its costs. Adopting the latter would not add any prudential value. It is not clear, then, why it would be good for these agents to systematically conform to epistemic norms instead of going by a policy that allows them to slip this one time. One might object that a policy of systematically conforming to epistemic norms except in triviality cases is too unrealistic and difficult to follow. But a policy of systematically conforming to epistemic norms is no less difficult to follow. It is just as unrealistic to expect to stick by a policy that requires you to systematically follow your evidence, avoid error, and the like. So this reply is not available to teleologists who want to employ a rule-based strategy.

Second, triviality cases can be modified in ways that cause problems for both case-based and rule-based ET. Suppose that in Dream, Vincent is reminded of the content of his dream after drinking poison while jumping off a plane without a parachute alone in the middle of nowhere. As in the original case, the triviality of the question means that, in the short term, no prudential value would come out of conforming to epistemic norms. But since he is seconds away from a certain death, it would not be good for Vincent in the long run either. Yet, even in this extraordinary situation, Vincent’s belief that his dream really happened would still be epistemically unjustified.

A final possible defence of a prudential version of ET is that having beliefs that conform to epistemic norms is good for humans, viewed as a group. It is plausible, for instance, that having such beliefs and norms allow us to pool and share reliable information. But since having reliable information is necessary for our survival and well-being, epistemic norms and epistemic goods are clearly good for us. ^{Footnote 14}

However, this line of thought is not a defence of categorical epistemic normativity. It is not, in other words, a defence of the claim that there is necessarily a reason for agents to conform to epistemic norms. At best, it shows that it is good, for the human race, to, for example, have the practice of making epistemic claims, use epistemic concepts, study epistemology, conform to epistemic norms most of the time, etc. But none of these conclusions support the necessary normative authority of epistemic norms. Even if we accept that it is good for humans to use epistemic concepts, to make epistemic claims, to generally conform to epistemic norms, and so on, the triviality challenge could still be successful. It could still be the case that conforming to epistemic norms is not necessarily good for agents. After all, the same kind of story could be told about many norms or domains that lack categorical authority. It is probably good for the human race to have developed norms like etiquette, laws, social conventions, and the like. Yet it is not necessarily good for us to conform to these norms.

At this point, one might respond by questioning the notion of epistemic normativity we have been working with. Why think of epistemic normativity as the notion that epistemic norms have necessary or categorical authority for agents? Do we really want or need more than, for example, the conclusion that conforming to epistemic norms is good most of the time? Perhaps we don’t. Perhaps epistemology does have the same kind of authority as etiquette or the law, ^{Footnote 15} and perhaps this weaker kind of authority also deserves the label ‘normativity.’ But whether or not that is the case, it remains true that contemporary epistemologists and normativity theorists typically view epistemology as normative in the categorical or necessary sense. And my concern in this article is only epistemic normativity so understood. ^{Footnote 16}

3.2 Minimal Predicative Value

A second possible sense of ‘X is good’ is a predicative sense, according to which X is good simpliciter or just plain good, even though it might not be good for anyone. Given this second sense of ‘good,’ the minimal value explanation of the triviality intuition can also be given a predicative interpretation that implies the following claim:

This seems to be what many teleologists have in mind, especially those who hold that true belief and knowledge are intrinsically good or good for their own sake. ^{Footnote 17} This second option is also problematic, however. The claim that conforming to epistemic norms is necessarily good simpliciter leads to at least two deeply counterintuitive conclusions.

First, if conforming to epistemic norms necessarily led to plainly good states of affairs, there would necessarily be a normative reason to go change the world to make it fit our beliefs. In particular, there would necessarily be such a normative reason precisely and only because it would make our beliefs true or instances of knowledge. As Joseph Raz explains:

But this is a deeply counterintuitive result. Plausibly, there is not necessarily a normative reason to act so as to make our beliefs true just because it would give us true beliefs or knowledge. Most of the time, this is no reason to go change the world. Suppose I falsely and unjustifiably believe that the water in Toronto is poisonous. That belief would become knowledge if I poisoned the water. But that is obviously not a reason for me to poison the water.

Second, minimal predicative value leads to the conclusion that there is necessarily a normative reason for anyone to do anything. This second objection starts from the popular Anscombean thesis that you cannot count as ϕ-ing intentionally unless you know that you are ϕ-ing. As Kieran Setiya explains:

Part of what it is to ϕ intentionally, then, is knowing that you are ϕ-ing. But, if knowledge is necessarily plainly good, it follows that there is necessarily something good in ϕ-ing intentionally, namely the knowledge that you are ϕ-ing. Hence, ϕ-ing intentionally necessarily brings about something good, namely knowledge. Epistemic teleologists are therefore led to an absurd conclusion: since it is necessarily good to intentionally do anything, there is necessarily a normative reason for anyone to intentionally do anything, including the most horrible, absurd, or pointless acts you can imagine.

Of course, one could reject the claim that intentional action essentially involves knowledge. However, even opponents to the Anscombean thesis would have to admit that knowledge is still part of most cases of intentional action. But this weaker claim—that knowledge is involved in most cases of intentional action—suffices to generate an almost equally counterintuitive conclusion, namely that there is almost always a normative reason for anyone to do anything.

Could teleologists bite these two bullets in response? They could, but it would put ET in an unfavourable position in the current dialectic. This would only add to the intuitive cost of ET, which already includes rejecting the triviality intuition. The problem is that the teleological minimal value explanation is up against an alternative explanation, the denial of ET, which does not have these costs. It is hard to see, in that context, why the minimal value explanation would be preferable.

Are there considerations that favour minimal predicative value? Michael Lynch and Jonathan Kvanvig have each offered prominent cases for the claim that true beliefs are intrinsic or final goods. If sound, these arguments could perhaps tip the balance in favour of ET and the minimal predicative value.

According to Lynch, we all care about truth for its own sake and not for the sake of something else. We are committed, in other words, to the final value of believing the truth and avoiding error. To support this claim, Lynch offers the following two thought experiments:

If Lynch is right that we care about truth for its own sake, then we have the basis for a promising argument in favour of ET. In general, the fact that we value something for its own sake is strong evidence that it really is valuable for its own sake. ^{Footnote 22} That argument can be summarized as follows:

1. 1. We all value the truth for its own sake.

2. 2. If we all value the truth for its own sake, then, in all likelihood, the truth really is valuable for its own sake.

3. 3. Therefore, in all likelihood, the truth is valuable for its own sake.

4. 4. Therefore, in all likelihood, it is necessarily good to conform to epistemic norms. ^{Footnote 23}

There are two main problems with this argument, both of which concern premise 1.

First, even if we admit that Lynch’s thought experiments provide evidence that we value the truth for its own sake, our data also includes the triviality intuition. While Lynch’s thought experiments might be informative, we must also remember the triviality cases from Section 2, which suggest precisely that we do not value the truth for its own sake. Hence, even if Lynch’s examples indicate that we value the truth for its own sake, their force is mitigated by the triviality intuition.

Second, and more importantly, these thought experiments do not convincingly show that we care about the truth for its own sake. Lynch is surely right that, in his first example, most of us would decline the neuroscientists’ offer. However, the reason is that it would rob us of many of the things we value most. If I entered the machine, I would no longer, for instance, be in contact with the people I love. I would stop being involved in their lives and they would stop being involved in mine. I would not help anyone ever again, I would stop improving the lives of others, I would not achieve anything anymore, I would not witness the great historical events of my lifetime, I would stop travelling, going to concerts, and so on. Granted, many of these things probably require knowing the truth. However, that does not mean that I value knowing the truth for its own sake. What I value is being in contact with loved ones, helping others, travelling, going to concerts, and so on. Not knowing that I am.

Regarding his second example, I should say that it is not clear to me that we would all prefer the actual world to a world that—unbeknownst to us—just came into existence. After all, a mind-boggling amount of horrible things happened up until yesterday in the actual world. But, if we are in the world imagined by Lynch, then none of these horrible things actually happened. No war, no illness, no genocide, no torture. Clearly, some of us might prefer that.

But, even if we set that aside, it is not clear that this preference would essentially have to do with knowing the truth. I would prefer the actual world because the other option would mean that most of the things that give my life value and meaning never actually happened. It would mean, for instance, that I never improved the life of anyone, that I never fulfilled promises, that I never finished my PhD, that I never built a relationship with my spouse, that the Montreal Canadiens never won the Stanley Cup, and so on.

The point is that, in both examples, our preference for the actual world can plausibly be explained without invoking our valuing the truth for its own sake. Of course, that alone does not entail that we do not, in fact, value the truth for its own sake. However, it does mean that Lynch’s thought experiments fail to show that we do. ^{Footnote 24}

Kvanvig also argues that true belief is valuable for its own sake, but via reflection on what the cognitively ideal agent would be like:

Kvanvig seems to offer the following argument:

1. 1. In cost-free environment where no practical needs are left unmet, a cognitively ideal agent would know all truths.

2. 2. The cognitive ideal in such a world would not consist in knowing all truths if true belief was not valuable for its own sake.

3. 3. Therefore, true belief is valuable for its own sake.

What does ‘cognitive’ in ‘cognitive ideal’ refers to exactly? One natural suggestion is that it just means ‘epistemic.’ On that reading, the cognitively ideal agent just is the agent who is ideal from an epistemic point view. This is a natural reading given that the cognitive is typically cashed out in terms of the ‘mind-to-world’ direction of fit. If ‘cognitive’ is defined in terms of the aim of representing reality accurately, then it is virtually equivalent to ‘epistemic.’

The problem with this reading of ‘cognitive ideal,’ however, is that it makes Kvanvig’s argument invalid. As we have already seen, merely showing that something is epistemically good is not sufficient to show that it is genuinely good and not merely domain-relatively good. It might be true that the epistemic ideal would involve omniscience. However, it is a further question whether it is necessarily genuinely good to do what the epistemic ideal being would do, i.e., to believe what is epistemically best to believe.

Compare this with the idea of the etiquette ideal. Trivially, the ideal agent from the standpoint of etiquette would always conform to every rule of etiquette. But that does not mean that it is always genuinely good to behave in the way that the etiquette ideal would behave. In the same way, no matter how we characterize the epistemic ideal, the crucial question is whether it is always genuinely good to do what the epistemically ideal agent would do. What the triviality intuition suggests is precisely that it is not.

An alternative reading of the first premise is that the ‘cognitive ideal’ does not refer merely to the ‘epistemically’ ideal agent, but rather to the cognition of the genuinely ideal agent. In other words, perhaps the first premise of Kvanvig’s argument should be interpreted as the claim that an agent who always does what is genuinely best always behaves as the epistemically ideal agent would behave. However, if this is how the first premise should be read, then the argument becomes question begging since it takes ET for granted. Instead of showing that epistemic value necessarily constitutes normativity-grounding value, this second reading of the argument uses this claim as part of its first premise. So Kvanvig’s argument is either invalid or question begging.

But perhaps I have misinterpreted Kvanvig’s passage. So far, I have taken his ‘cognitive ideal’ point to be a deductive argument in favour of the value of truth. However, another interpretation is that Kvanvig only aims to provide an intuition pump. Under this interpretation, his point is simply that, when we try to imagine the cognitively ideal agent in a cost-free environment with no practical needs unmet, our intuitions tell us that such an agent is omniscient. Suppose Jon and Joan live in such an environment. Jon knows all truths, except for one completely inconsequential and obscure fact, about which he is wrong. Joan, in contrast, really knows all truths. Isn’t there something better with Joan compared to Jon? Isn’t Jon a little further from the ideal than Joan? But, if that is our intuition, then aren’t we committed to the claim that there is at least something good in any true belief, however pointless?

A first thing to note is that if Kvanvig is only offering an intuitive case, then it is doubtful that it can suffice to vindicate minimal predicative value in the present context. Even if we admit Kvanvig’s intuition, the predicative version of ET still has the counterintuitive implications outlined above and we still have the triviality cases. So, even if it provides some support for minimal predicative value, Kvanvig’s intuition likely won’t be enough to tip the scale in teleologists’ favour.

What should we make of Kvanvig’s intuition? Once again, we must distinguish between what it takes for an agent to be ideal from an epistemic standpoint and what the cognition of a genuinely ideal agent would be like. As we have already seen, there is plausibility to the idea that the agent who’s ideal from an epistemic standpoint would be omniscient. So, in a sense, it is plausible that there is something worse or less than ideal with Jon compared to Joan: he is epistemically worse than her. He is less than epistemically ideal. But, once again, this claim only entails the domain-relative claim that knowing the truth is epistemically valuable.

What teleologists need is the intuition that Joan is genuinely better than Jon. But, once we set aside domain-relative value, it becomes less obvious that there must be something just plain worse with Jon. If the only difference between the two is really just a single, completely pointless piece of knowledge, then it is hard to see why Joan’s situation would be genuinely better. One possible response is that, if we had to choose between being in Jon’s situation and Joan’s situation, most of us would choose Joan’s. This might be true, but it is not clear that our answer to this question can be trusted as a guide to what is genuinely valuable.

To see this, consider an alternative version of the Jon and Joan case. Imagine that Jon conforms to all the rules of Victorian table etiquette, except for one completely inconsequential, unnoticeable, and practically unenforceable rule. Joan, in contrast, conforms to all the rules of Victorian table etiquette, including the obscure one that Jon violates. Once again, both agents are in a cost-free environment where no practical needs are left unmet. Hence, there are no costs or benefits associated with following or violating the obscure rule in question. But what would we do if we had to choose between being in Jon’s situation and Joan’s situation? I suspect that most of us would choose Joan’s. After all, there is no practical cost associated with conforming to that additional rule and no benefit to violating it. So why not? Yet conforming to Victorian table etiquette is obviously not necessarily good simpliciter.

So, for all these reasons, Kvanvig’s ‘cognitive ideal’ point does not show that conforming to epistemic norms is necessarily good simpliciter. ^{Footnote 26} More generally, given the failure of the two most prominent cases in its favour as well as the problems outlined above, I conclude that the predicative version of the minimal value explanation is implausible. Conformity to epistemic norms does not appear to be necessarily just plain good.

3.3 Minimal Attributive Value

This leaves the third, attributive sense of ‘good.’ This is the sense of ‘good’ that we use to refer to something being good qua member of a kind K. To be a good member of K in this sense is to do or be able to do what members of K are supposed to do qua members of K. It is, in other words, to achieve or be able to achieve the defining aim, function, or telos of members of that kind. We use this attributive sense of ‘good’ when we judge, for instance, that X is a good toaster, a good football player, a good teacher, and so on. Corresponding to this third kind of value is a final possible version of the minimal value explanation, which holds the following:

There are several possible ways to flesh out this idea, but the most obvious one is in terms of good and bad beliefs. Perhaps beliefs that conform to epistemic norms are necessarily good, well-functioning, or successful qua beliefs. Such beliefs, the suggestion goes, are as they are supposed to be qua beliefs. Just like toasters that toast well, beliefs that conform to epistemic norms fulfil the constitutive aim or function of beliefs. ^{Footnote 27}

The problem, however, is that, even if it is true, this attributive value claim alone cannot vindicate ET and epistemic normativity. This is because goodness qua member of a kind is not necessarily genuine or normativity-grounding goodness. The mere fact that X is well-functioning or successful qua member of kind K does not entail that it is genuinely good to be X, to have X, or to promote X. It leaves open the question of whether it is genuinely good to be a good member of K or to do what good members of that kind do. Hence, it only entails mere domain-relative value.

For example, part of being a good torturer is to make one’s victims suffer intensely for a long time. This is what torturers are supposed to do qua torturers. Similarly, part of being a good hired killer is to murder one’s victims quickly and without leaving a trace. This is what hired killers are supposed to do qua hired killer. But those attributive value claims do not make it genuinely good to make people suffer intensely for a long time or to murder swiftly and without leaving a trace. Even if we accept that this is what good torturers and hired killers do, such attributive value claims leave open the question of whether it is genuinely good to be a good torturer or a good hired killer. This means, in turn, that it leaves open the normative question of whether there is any normative reason to do what good torturers or good hired killers do.

Therefore, showing that conforming to epistemic norms means having well-functioning beliefs won’t suffice for vindicating ET and epistemic normativity. Even if the underlying attributive value claim is true, it still leaves open the normative question of whether it is necessarily genuinely good to have beliefs that are good qua beliefs and to avoid beliefs that are bad qua beliefs. But this is what teleologists need to show. What the triviality intuition suggests is precisely that there is sometimes no normativity-grounding value in having well-functioning beliefs and in avoiding defective beliefs.

The same problem arises if we claim instead that agents who conform to epistemic norms are necessarily good qua believers, qua inquirers, qua epistemic agents, and the like. Even if they are true, these attributive value claims only entail domain-relative goodness since they leave open the normative question of whether it is necessarily genuinely good to be a good believer, a good inquirer, a good epistemic agent, and the like. Once again, this is precisely what the triviality intuition calls into question.

Could teleologists claim, instead, that conforming to epistemic norms necessarily makes us well-functioning qua human beings? This may be more promising since it is admittedly harder to deny that being good qua human agent is necessarily genuinely good. The problem, however, is that it is implausible that violating epistemic norms necessarily makes people defective qua human beings and that conforming to epistemic norms necessarily contributes to making one a well-functioning human agent. As we have already seen, to say that X is a good member of a kind K is to say that it does what members of that kind are supposed to do. It is to say, in other words, that it can achieve the constitutive, essential, or defining aim, function, or telos of members of that kind. But what is the essential telos of human agents? When are humans defective or well-functioning?

One natural answer is that we are well-functioning insofar as we flourish. As we have already seen, however, conforming to epistemic norms is not necessarily good for us and violating epistemic norms is not necessarily bad for us. But it is hard to see how ϕ-ing could be necessary for our flourishing without being also good for us. The reasons that it is not necessarily good for us to conform to epistemic norms are also plausibly reasons that it is not necessarily conducive to our flourishing.

Another natural answer would be that the essential aim of human agency is autonomy and rational action. Agents are well-functioning, in other words, insofar as they can intentionally do what they think they ought to do. But you can very well violate epistemic norms and still be autonomous in that way. Perhaps you cannot be autonomous if you are generally incompetent at conforming to epistemic norms. However, there can certainly be cases where failing to follow the evidence about particularly trivial and insignificant questions won’t interfere with your autonomy or ability to act as you think you ought. Hence, it won’t suffice to vindicate the necessary normative authority of epistemic norms.

Could conformity to epistemic norms itself be a constitutive telos of human beings? Could knowing the truth and avoiding error about any matter (including complete trivialities) be a defining aim of human agency? It is not clear why this would be the case. It cannot be because we all have the capacity to systematically conform to epistemic norms. We do not. Most people have deeply entrenched biases and limitations that routinely prevent them from, for example, following their evidence. And at least some of those people are unable to get rid of those biases because they do not care or because it is too difficult.

Neither can it be because we all actually want to know the truth and avoid error about any question with which we might be confronted. At least some human beings do not care to know the truth about, say, completely trivial and obscure questions. Once we add to this the fact that conforming to epistemic norms can often be detrimental to people’s interests, well-being, and flourishing, it is simply implausible to suppose that human agents qua human agents are supposed to systematically conform to epistemic norms. If human agency must really have an epistemically relevant telos, other candidates seem more plausible. Why isn’t it to, e.g., conform to epistemic norms often enough to flourish, be autonomous, or maintain optimal physical and psychological health? After all, this is the kind of thing that most of us would want and could at least come close to achieving.

So the claim that it is necessarily attributively good to conform to epistemic norms is either false or insufficient for grounding epistemic normativity. Hence, minimal attributive value cannot be invoked to meet the triviality challenge either.