1. A RELEVANT VALUE PROBLEM

Many epistemologists have the intuition that acquiring knowledge about some topic is more valuable than merely coming to a true belief about it. This thought can be traced as far back as Plato where, in the Meno, Socrates and Meno contemplate ways of justifying it. Why, Meno asks, would it be better to know the way to Larissa than to merely have a correct judgment about it (Plato Reference Plato, Hamilton, Cairns and Guthrie1961: 97a–97d)? Either way, someone desiring to get to Larissa would seemingly have the cognitive means of getting there. Plato's dialogue constitutes the first formulation of the value problem of knowledge. There is supposed to be a value problem of knowledge because it is not immediately clear how to explain why knowledge is supposed to be better than mere true belief, especially given that the practical benefits of their attainment seem equal.

Socrates considers a solution involving an analogy with the “tethered” statues of Daedalus (Plato Reference Plato, Hamilton, Cairns and Guthrie1961: 97d–98b). In the last twenty or so years, epistemologists have revived this discussion about the comparative axiological relations between knowledge and mere true belief, offering their own more precise formulations of the value problem along with their own preferred solutions.Footnote ¹ This paper can be seen as an extension of this revival for the purpose of problematizing a deeply engrained assumption about the epistemic value of true belief.

While there are a number of ways that a precise value problem of knowledge might be formulated, the version that I will focus on here goes as follows:

EVP is a particularly epistemic version of the value problem, asking us for an explanation of knowledge's superior status within the epistemic domain. Of course, asking for an explanation of some fact assumes that the purported explanandum is, indeed, a fact that needs explaining. Should we really think that, necessarily, for any proposition, knowing it is epistemically better than merely truly believing it, all other things being equal? The assumption behind EVP might be questionable as it asks for an explanation of a purported fact that is both categorical and necessary. EVP implies that knowledge is epistemically better than mere true belief for all propositions and in all worlds. Some clarification may be needed in order to motivate this assumption.

The implication that knowledge is always epistemically better than mere true belief should not be thought to entail the more controversial claim that, in every instance, knowledge is a genuinely valuable thing to have. The assumption behind EVP is consistent with there being instances of worthless or even genuinely disvaluable bits of knowledge. The clause “epistemically valuable” in EVP should be read attributively and not predicatively, as indicating something about value from the “epistemic perspective” and not something about overall value.Footnote ² Just as a particular gunshot might truly be a good one “from the markman's perspective” while it does not have any more genuine kind of value (perhaps even being genuinely disvaluable), a particular instance of knowledge (or true belief) might be epistemically valuable, or good “from the epistemic perspective,” even if it is worthless or thwarts other, more genuine axiological considerations. In other words, the claim that knowledge is more epistemically valuable than mere true belief is a claim about the relations between the domain-internal value of knowledge and mere true belief and, as Ernest Sosa (Reference Sosa2007) would say, such a claim is “insulated” from considerations about overall value.

The “ceteris paribus” clause in EVP is intended to accommodate worries related to those brought out by Jason Baehr (Reference Baehr, Haddock, Millar and Pritchard2009). Similarly to Baehr, one might worry that knowledge is sometimes less epistemically valuable than mere true belief, especially in cases for which gaining knowledge comes with certain other epistemic costs. For example, if coming to know that p (as compared to, say, simply trusting a lucky hunch about whether p) would take a lot of time and effort that could be better epistemically spent, then a merely true belief that p would seemingly be better, even purely from the epistemic point of view. What EVP assumes is that, insofar as such extra value-laden facts are not built right into the cases imagined, knowledge is going to be epistemically better than mere true belief. That is, insofar as these other epistemic value considerations are equal (time, effort, cognitive load, etc.), knowledge is epistemically preferable.

So, is it indeed the case that knowledge is necessarily and categorically more epistemically valuable than mere true belief? I think that at least most epistemologists would say that it is. Compare two possible relations that agent S might have to a true proposition p: (1) S might (merely) believe that p and (2) S might know that p. Which would be better from the epistemic perspective, all other things being equal? It seems relatively uncontroversial that knowing would be better. EVP is a call for an explanation of that fact. Is it somehow explained by the nature of knowledge? Is it explained by the nature of epistemic evaluation? Something else? EVP is a question that seems deserving of an answer. If it is not a legitimate question, then we are at least owed some explanation as to why it is not. Requesting an answer to EVP by no means entails that an answer must be extraordinarily deep or labyrinthine but any theorist about epistemology should have something to say about it.

2. BOTTOM-UP APPROACHES TO EVP

In this section, I discuss a particular worry one might have about what I call “bottom-up approaches” to EVP. Bottom-up approaches take tokens of true belief to be fundamentally valuable from the epistemic perspective and then build the value of knowledge up from its parts. Less metaphorically, proponents of the bottom-up approach take the fundamental epistemic value of true belief as relatively basic in their axiological theories and then attempt to explain knowledge's added epistemic value through the value of its other essential properties (e.g., being a certain kind of justified belief, being the product of a reliable process, being an “achievement,” etc.) which purportedly have a relevant kind of derivative epistemic value. According to the bottom-upper, true belief is fundamentally valuable and knowledge is, by its very nature, TB + . According to the story, the extra properties needed in order to turn a mere true belief into knowledge can help us explain why knowledge is better than mere true belief.

Bottom-up approaches have become a very standard way of attacking the value problem and this is partly due to the pervasive view that true belief is a fundamental epistemic value, at least one kind of rightful end to our distinctively epistemic endeavors. Many philosophers take true belief to lay at the bedrock of theorizing about epistemic value. Hilary Kornblith (Reference Kornblith1985: 158) claims that “any account of epistemic evaluation that does not give truth a central role to play is inaccurate.” Speaking about truth from the epistemic point of view, William Alston (Reference Alston2005: 40) states: “[H]ow could any property of a belief be better …?” Michael DePaul (Reference DePaul and Steup2001: 177) takes the fundamental epistemic value of true belief to be “uncontroversial.” Allan Hazlett (Reference Hazlett2013: 262) claims that “it is not coherent to ask whether true belief has epistemic value.”

Consider a paradigmatic case of the bottom-up approach: Alvin Goldman and Erik Olsson's (Reference Goldman, Olsson, Haddock, Millar and Pritchard2009) “conditional probability” solution to the value problem. Goldman and Olsson are veritists, meaning that they take true belief to be the only fundamental value from the epistemic perspective while all other epistemic values are derivative.Footnote ³ A proponent of veritism is forced into attempting a bottom-up solution to the value problem of knowledge because of her self-imposed restrictions on fundamental epistemic value. If true belief is the only fundamental epistemic value and knowledge is supposed to be epistemically better than mere true belief, then there is a need to explain that added value of knowledge in terms of its other properties which are claimed to be derivatively valuable relative to true belief.

As for derivative value relations, I take the following dependence condition to hold: If x and y are valuable with respect to domain D and the domain-internal value of x ontologically depends on the domain-internal value of y (asymmetrically), then x is derivatively valuable with respect to domain D. A domain-internal value is derivatively valuable when its status as a value asymmetrically depends on the domain-internal value of some other thing. Take, for example, one of Sosa's (Reference Sosa2007) “critical domains”: his so-called “world of coffee.” If you are evaluating, say, a reliable coffee machine as a kind of coffee connoisseur, then the value of that particular machine is going to be derivative from that perspective relative to the more fundamental value of delicious and aromatic coffee. Since the coffee-related value of the machine depends on the prior value of delicious, aromatic coffee (by being instrumental toward the existence of such coffee), the domain-internal value of the machine is derivative.Footnote ⁴

Goldman and Olsson are “process reliabilists” about knowledge (Reference Goldman, Olsson, Haddock, Millar and Pritchard2009: 22): “[S]ubject S knows that p if and only if (1) p is true, (2) S believes p to be true, (3) S’s belief that p was produced through a reliable process, and (4) a suitable anti-Gettier clause is satisfied.” Since reliable processes are merely derivatively valuable, Goldman and Olsson are led to the following bottom-up approach (Reference Goldman, Olsson, Haddock, Millar and Pritchard2009: 27–8):

Goldman and Olsson's attempted solution is a bottom-up approach because it begins with the fundamental value of true belief and it explains the added value of knowledge by appealing to the fact that knowledge is essentially a kind of reliably produced true belief. Purportedly, being a reliably produced true belief is better than being a merely true belief because that state of affairs includes a particular kind of derivatively valuable mechanism.

Consider another example of the bottom-up approach: Kurt Sylvan's (Reference SylvanForthcoming) “non-instrumentalist” veritist solution to EVP. Dissatisfied with reliabilist solutions like Goldman and Olsson's, Sylvan takes an alternative approach inspired by the work of Thomas Hurka. According to Sylvan, true belief is the only fundamental value from the epistemic perspective but knowledge is better because it is a way of conforming to the truth by “manifesting a strong respect” for it. The approach is purportedly “non-instrumental” because, unlike reliabilist approaches like Goldman and Olsson's, conforming to the truth in this way is not necessarily instrumental toward true belief. Yet, Sylvan's solution is still an instance of a bottom-up approach because conforming to the truth by manifesting a strong respect for it is derivatively valuable relative to the value of true belief. Because knowledge, by its very nature, is supposed to manifest a way of conforming to the truth via strong respect and is supposed to manifest a way of “properly valuing” some more fundamentally good thing, it has an added domain-internal value, an additional derivative value over and above that of true belief.

In answering EVP, while veritists are necessarily bottom-uppers, bottom-uppers are not necessarily veritists. Take, for example, John Greco's (Reference Greco2010: Ch. 6) virtue reliabilist account. Even though Greco does not take aim at a distinctively epistemic value problem, his proposal can be reformulated as an answer to EVP and, as such, would properly be interpreted as another kind of bottom-up solution. Greco does not commit himself to veritism like Goldman, Olsson, and Sylvan.Footnote ⁵ Greco argues that knowledge manifests a kind of achievement in the sense that, in knowing, an agent deserves credit for her true belief (Reference Greco2010: 71): “(KSA) S knows that p if and only if S believes the truth (with respect to p) because S’s belief that p is produced by intellectual ability.” Knowledge, by its very nature, is supposed to be an intellectually virtuous true belief where the exercise of cognitive ability and virtue figures importantly in a causal explanation of the belief's accuracy. Greco believes that KSA can properly account for the value problem. Supposing that x is a valuable state of affairs, it is better to achieve x than to merely have x obtain luckily and by accident. An accurate belief is an epistemically good thing to have, but achieving accurate belief is better.

A solution in line with Greco's virtue reliabilism is perhaps less obviously a bottom-up approach to EVP. By Greco's account, certain beliefs have an added value because they constitute an achievement of true belief, being accurate because of the exercise of an ability. Constituting an achievement of true belief, however, would be derivative relative to the value of true belief. In the same way that the value of “strongly respecting” a good thing depends on the value of that good thing (à la Sylvan), the value of achieving something also depends on the value of that thing. No one, for example, would be tempted to think that achieving world poverty is good and that is simply because world poverty is not good. The value of achievement exhibits an asymmetric ontological dependence on the value of the object of achievement. Like other bottom-up approaches, then, Greco thinks that knowledge has a derivatively valuable property relative to the value of true belief and it is that property that makes knowledge better than mere true belief.

Again, these kinds of bottom-up approaches have become quite prevalent in the literature on the value problem. Here, I would like to suggest that we should be a bit worried about this prevalence in a way that has not yet been appreciated. In the space provided, while it would be an exercise in futility to attempt to critically and satisfactorily discuss each of these bottom-up solutions to EVP (along with any and all other possible bottom-up solutions to EVP), I at least want to raise a potential worry about the epistemic value of true belief in a way that might also cause us to worry more generally about bottom-up approaches. Given the worry, we should question whether true belief ought to have such an uncontroversial and central standing in our epistemic theorizing.

In this direction, compare the following two cases: (1) Annie occasionally forms her beliefs on the basis of tarot card readings. On the basis of a particular reading, Annie forms the belief that there is a red car parked on the corner of Coral Way and Snapper Creek Road. There is no red car parked on the corner of Coral Way and Snapper Creek Road. (2) Annie occasionally forms her beliefs on the basis of tarot card readings. On the basis of a particular reading, Annie forms the belief that there is a red car parked on the corner of Coral Way and Snapper Creek Road. As things turn out, there is a red car parked on the corner of Coral Way and Snapper Creek Road.

Annie's belief is false in case (1) and true in case (2). The question that I would like to raise is the following: Purely restricting ourselves to epistemic desiderata, is Annie's belief any better in case (2) than it is in case (1)? My own inclination is to say “no, not really.” Epistemically speaking, Annie's beliefs are both of very poor quality and, if they are not equally bad, then they are very nearly so. The fact that Annie's belief happens to correspond to reality in (2) has little to no bearing on its epistemic worth.

If this intuition is correct, then it is going to be one that any bottom-up theorist will have a tough time explaining. If true belief is really supposed to have such an uncontroversial fundamental value within the epistemological critical domain, then we should not be tempted to issue such judgments about the comparative epistemic value obtaining in (1) and (2). If the value of true belief is fundamental in a way that the other relevant epistemic values depend on it, while those other derivative values may surely be important, true belief should still receive at least a significant spotlight in the relevant epistemic value considerations. Given the intuition, this would indicate a kind of cognitive dissonance between certain judgments about epistemic worth and thinking about true belief as having the kind of fundamental epistemic value that the bottom-upper attributes to it.

To be clear, the purported intuition does not imply that Annie's belief in (2) is no better, in any sense, than her belief in (1). Her true belief in (2) could very easily be more valuable than her belief in (1), even if it is not epistemically so. For example, her belief in (2) might be practically better than her belief in (1) if she were to somehow, say, appear on a game show in which the grand prize depended on her correctly guessing whether there is a red car parked at that intersection. After all, the second belief is accurate while the first one is not. What I am suggesting, however, is that it is not obvious that such considerations necessarily bear on the epistemic point of view.

This, of course, is not a bottom-up way of thinking about Annie's case. Consider what someone like Sylvan might say: “Annie's belief has a significant kind of epistemic worth in (2) while it does not in (1). In fact, her belief in (2) is an instance of the only kind of fundamental epistemic good. It is a true belief. Of course, it could be epistemically better since it does not manifest a proper valuing of the truth, but it is good nonetheless.” Or Greco: “Yes, Annie's belief has some epistemic worth in (2), while it does not in (1). Of course, her belief in (2) could be epistemically better since it does not manifest an achievement of true belief, but it is good nonetheless.” My own inclination is to say that, from the epistemic perspective, we just do not value certain kinds of true beliefs. The fact that a belief is accurate, by itself, does not seem to generate a significant amount of epistemic value, much less epistemic value of a fundamental kind.

Now, I do not pretend to have provided any kind of exhaustive refutation of the bottom-up approach. In fact, by merely expressing a handful of intuitions about the comparative value of Annie's beliefs, one might think that I have done little more than a bit of table pounding. I readily grant this. Nonetheless, such considerations (along with some independent worries about the extant bottom-up solutions outlined above) have led me to take a skeptical eye toward views of epistemic evaluation that begin with the fundamental value of true belief. Probably the most appropriate conclusion of this section is a conditional one: If one is moved by the intuitions expressed above or, if one is not fully persuaded by the individual bottom-up solutions provided by Goldman, Olsson, Sylvan, and others, then we should question whether true belief plays the kind of fundamental axiological role that the bottom-upper attributes to it. Alternative approaches to the epistemological critical domain would be warranted.

3. THE GOALS OF INQUIRY AND THE VALUE PROBLEM

My goal in this section is to sketch an alternative answer to EVP, one based on the idea that the values within the epistemological critical domain are properly framed by the constitutive goals of inquiry. This approach will differ from the bottom-up approach in a number of principled ways and I hope to make these ways clear and suggest that the goals of inquiry give us a truly attractive theoretical option. Ultimately, given the worries above, the explanatory power of this inquiry-based, critical domain approach should be seen as some abductive support in favor of the view.

One may think the following: Look, it is just a primitive fact about knowledge that it is epistemically better than mere true belief. It is a fact that is constitutive of the epistemic evaluative domain and there is just no need to say anything more about it. If someone, for example, were to ask us for an explanation of why delicious and aromatic coffee is better than bland and unsavory coffee from within Sosa's “world of coffee,” the proper response might be nothing more than an incredulous stare. Why should we expect anything different in comparing the epistemic value of knowledge and mere true belief? There is, I think, some merit to this approach. However, unlike the evaluation of coffee, there is just no consensus to be found on what the epistemological critical domain is even supposed to be about. Many have the idea that it is supposed to be about some kind of success in cognition but, given that there are both monists and pluralists about fundamental epistemic value, veritists and non-veritists, we are left with some explaining to do.Footnote ⁶ Given the disagreement, it would be best to present some plausible theoretical underpinnings for the value facts at hand.

Here, I will suggest that the constitutive goals of inquiry act as the fundamental values within the epistemological critical domain. Though inquiry is typically associated with some broad form of asking and answering questions, there are two more precise senses of “inquiry” that are relevant: First, someone might be in a state of inquiring about something. S is in a state of inquiring about question Q just in case S is interested in a legitimate answer to Q. Such a state is initiated by a certain kind of desire for an answer to a question. In that sense, it is a state with what has been called a “world-to-mind” direction of fit, where the satisfaction of it would consist in the realization of its object, namely S actually obtaining a legitimate answer to the question Q. Second, someone might engage in an activity of inquiring, which is something over and above merely being in the state. S might consciously desire a genuine answer to Q without actually attempting to find an answer to Q or even being able to. Those who engage in the activity of inquiring about Q actively probe the world in order to bring about the satisfaction of the state of inquiry that they are in. Engaging in the activity of inquiring is the proper response, ceteris paribus, to being in the state of inquiry in the first sense.

The claim I will make here is that the constitutive goals of inquirers, generally speaking, frame the most fundamental values within the epistemological critical domain. A bit more precisely, for any question Q, if it is possible (in a very broad modal sense) for someone to be in a state of inquiry about it or engage in the activity of inquiring about it, then the constitutive goals of such states and activities are fundamentally valuable from the epistemic perspective. Why should we think that the goals of inquiry play this role? Epistemologists take the goals of inquiry and the interests of inquirers seriously, as it were. Suppose, as a place-holder, that we take the goal of inquiry to be something like information gathering. At least for some of us, it is an interest in our success as information gatherers that causes us to be interested in epistemic evaluation. Someone like the skeptic tries to show us that we are not so successful. Alternatively, non-skeptical epistemologists formulate theories of knowledge, justification, etc. that vindicate our information gathering abilities. Skeptics and non-skeptics alike, though they deliver competing evaluations of our own situations, share a common interest in our capacities as inquirers. It would make sense, then, that such interests would frame the most fundamental values in an associated evaluative practice.

What is most relevant for the purposes of this section is that there is a particular kind of inquiry that could help us in solving EVP. The kind of inquiry that I have in mind is what I will call “propositional inquiry,” where the constitutive goal of such an activity is to come to know some proposition p. The paradigm case of this kind of inquiry is a response to being in a state of inquiring about whether p. When we are in such a state, we have an interest in the possession of an answer to the question of whether p is the case. What would satisfy such a desire? Nothing short of knowledge. In fact, in inquiring about whether p, there are two conditional knowledge-goals present: (1) if p is true, to know that p and (2) if p is false, to know that ¬p. Other instances of propositional inquiry include inquiries about which objects possess some property P (where the goal would be to know some proposition that objects x, y, z, etc. possess P), inquiries about when some event has occurred (where the goal would be to know some proposition including an event and a time as constituents), inquiries about where something is (where the goal would be to know some proposition including an object and a place as constituents), etc. In each of these cases of propositional inquiry, the goal (at the very least) is to know some particularly salient proposition.Footnote ⁷

By saying that knowledge is the goal of propositional inquiry, it is important to note that it is the constitutive goal of that activity, not some contingent property of it. For instance, different people might engage in inquiring about whether p for different reasons. Alice might inquire about whether p merely because she is curious, Brian because it is his job, Chris because she is trying to help a friend, etc. Even though, in some sense, each of these people have different goals and aims when they engage in the activity of inquiring about whether p, what is constitutive of each of their situations is that they have the goal of knowing whether p. Aiming at knowing whether p is what all inquirers about whether p have in common, no matter their contingent circumstances or reasons for inquiring in the first place.

A natural worry about saying that knowledge is the constitutive goal of propositional inquiry would be the following: Why should we think of the goal to be a state of knowledge, in particular, and not, say, a mere state of true belief? If the idea is that, in inquiry, we desire to possess a genuine answer to a question, why is having a relevant true belief not enough to satisfy that desire? Having a true belief about p seems to be one way of thinking about what it would mean to “possess a legitimate answer” to the question of whether p. Some philosophers do think of the goal of propositional inquiry in this way. For example, Duncan Pritchard (Reference Pritchard, Matheson and Vitz2014: 114) claims that “truth is the telos of a properly conducted intellectual inquiry.” Alvin Goldman (Reference Goldman1999: 3–5; Goldman and Olsson Reference Goldman, Olsson, Haddock, Millar and Pritchard2009: 24), Jonathan Kvanvig (Reference Kvanvig2003: 145–50), Michael Lynch (Reference Lynch2004: 13–15; Reference Lynch2009: 225–8), William Alston (Reference Alston2005: 4–7) and others also give a nod to the goal of mere true belief.

I think that we should be skeptical of the idea that mere true belief is the goal of propositional inquiry. Suppose that Kate finds herself interested in whether p (for simplicity, let p be some empirically knowable proposition) and that, in the course of inquiring about whether p, she comes to believe, luckily but truly, that p. Suppose that this belief of hers manifests itself in the following way: She is disposed to judge that p is true, she uses p as a premise in both her practical and theoretical reasoning, she behaves as if p is true, but when she is asked why she thinks that p is true, she never makes any transitions in thought from anything that would really support the truth of p outside of the fact that she just has a gut feeling that p is true. Suppose that she even thinks that she knows that p is true.

Now, clearly, Kate does not know that p and, if she were to close her inquiry about whether p, she would do so illegitimately in some sense of the term.Footnote ⁸ For the sake of argument, I want to grant that this consideration, by itself, is not enough to establish that knowledge is the goal of such inquiry instead of mere true belief. As Pritchard (Reference Pritchard, Matheson and Vitz2014: 123–6) argues, even if an activity can only be legitimately closed by being in a certain state, that does not entail that being in that state is the goal of such an activity.Footnote ⁹ Nonetheless, we can still ask the question of whether Kate has indeed satisfied the goal involved in her inquiring about whether p.

Of course, given the state that Kate is in after she comes to her belief, she at least thinks that she has satisfied the goal. But, has she in fact satisfied it? I doubt it. At least part of what we want when we inquire about whether p is to adequately track whether p, coming to possess the correct answer in a relevantly unlucky way. Kate, in some sense, does possess the right answer about whether p, but she has not properly tracked whether p. Her accurate belief is lucky in a way that detracts from her success as an inquirer. If someone were able to later convince Kate that her belief was only luckily true, she would presumably be disappointed about her previous doxastic attitude vis-à-vis her goal of inquiring about whether p, even though that belief was in fact true. Thinking of knowledge, and not true belief, as the proper goal of propositional inquiry accommodates this intuition.Footnote ¹⁰

When we think about what we want in propositional inquiry, it is knowledge and not mere true belief that is the ideal end of propositional inquiry. The propositional inquirer categorically prefers knowledge to mere true belief in satiating her relevant inquiry-related desires. Hence, I think that it is natural to think of it as the proper goal, even if falling short of that goal will sometimes turn out to be acceptable given an inquirer's other interests.Footnote ¹¹ The goal of the propositional inquirer is to find out the truth of some proposition in whatever way she can (whether it be via perception, memory, testimony, deductive inference, some a priori method, etc.) and, as Alan Millar (Reference Millar, Haddock, Millar and Pritchard2010: 164) states, “[f]inding out is nothing less than coming to know.”Footnote ¹²

An interesting result now follows. Let us call it the “value of knowledge thesis:”

VKT is a natural result of the claim that the constitutive goals of inquiry explain the fundamental values within the epistemological critical domain. If it is the case that, for every possible inquiry-activity, the goal of such an activity is of fundamental epistemic value, at the very least, knowing whether p will be valuable for all semantically coherent propositions.Footnote ¹³ There may be many other inquiry-related interests that would be relevant to the epistemological critical domain, perhaps some that are not reducible to a knowledge-goal; but, at the very least, the relevant interests will include knowing whether p, for all p.Footnote ¹⁴

Given VKT, we can now put it to work in answering EVP. We can now give a relevant explanation for why tokens of knowledge are necessarily more epistemically valuable than tokens of mere true belief. The nature of propositional inquiry dictates that it is knowledge that is the ideal end and it is the nature of inquiry that explains knowledge's fundamental epistemic value. Mere true belief has a lesser standing (if it has any standing at all). Recall Kate who propositionally inquires about whether p and luckily comes to have a true belief about whether p. Kate's belief, though true, is less than optimal vis-à-vis the goals of propositional inquiry. Her belief is lucky in a way that detracts from her success as an inquirer. She has not tracked the truth in the way that she would have if she came to know whether p. In that way, knowledge is better than mere true belief from the epistemic perspective. Further, knowledge is epistemically better than any other cognitive state that might fall short of it.

This kind of answer to EVP is a unique one in that it does not appeal to any kind of metaphysical analysis of knowledge. In that way, it is quite unlike the bottom-up solutions encountered earlier. Instead, the solution appeals to facts about the goals of inquiry and knowledge's extrinsic (but necessary) relation to those goals. We need not have a particular metaphysical account of knowledge on the table and we need not reductively specify the essence of knowledge in order to see those relations. This is an avenue that I think many theorists about epistemic evaluation have missed. If epistemic evaluation comprises the kind of critical domain that is framed by a certain set of fundamental domain-internal values, then it need not be some feature of an analysis of knowledge that explains why it is necessarily better than mere true belief. More generally, when we are trying to explain the relative domain-internal value of two different kinds of things, we do not always need to appeal to some real definition of those things.

Take another example: Suppose that I wanted to compare the domain-internal value of two different kinds of tennis racket tensions from the perspective of someone interested in competitive tennis. I want to know, say, why a racket strung at 55 pounds is better than a racket strung at 20 pounds. We would be traveling down the wrong road if we were to begin explaining such facts by formulating and comparing exhaustive accounts of what it is for a racket to be strung at each tension level. In giving such an explanation, real definitions of particular tensile strengths would only play a role insofar as they help us specify each strength's relation to the relevant goals of the domain. What would explain the superior value of a racket strung at 55 pounds is an explanation of how certain physical facts about tensile strengths relate to the goal of winning tennis matches. A racket strung at 55 pounds allows for the right balance between power and control, better equipping a player to make winning shots. The relevant relations to the domain-internal goal play the most crucial role in such explanations and we can give them without completely specifying what it is to be a tensile strength of a certain magnitude.

Similarly, we can explain knowledge's epistemic superiority to mere true belief through the associated goals of the domain. The relative relations between each cognitive state and the goals of propositional inquiry seem to provide exactly the right touch. Knowledge is necessarily better than mere true belief with respect to those goals. We should not conflate value theoretic considerations and metaphysical ones. This is especially easy to do when we are comparing the relative epistemic merits of knowledge and mere true belief because the latter is typically taken to be a necessary condition on the former. This causes an automatic association with metaphysical analysis. Yet, insofar as epistemic evaluation comprises a critical domain, the value theoretic facts can be divorced, at least to a certain extent, from exhaustive metaphysical accounts of the items of value. In this case, knowledge's extrinsic relation to the goals of inquiry gives us all we need in responding to EVP.

Supposing that this story about the fundamental epistemic value of knowledge is correct, what then about mere true belief (i.e., true belief that does not instantiate knowledge)? Does it always, like knowledge, have epistemic value? If the intuitions about Annie's beliefs above are right, then it seems as though true belief does not. Can an inquiry-based critical domain meet the proper demands here? Bottom-up approaches begin with the fundamental epistemic value of true belief and then build up the value of knowledge through necessary derivative value relations. We are currently asking about what happens when we flip the script. We are asking about the status of mere true belief relative to the fundamental epistemic value of knowledge (perhaps best thought of as a “top-down approach” to the epistemic value of true belief). What kinds of necessary value relations might obtain between the two when knowledge plays the more central role?

A natural starting point in answering this question would be to survey some plausible necessary relations between mere true belief and knowledge and to see whether any of those relations generate legitimate derivative value. The relations already encountered do not obviously obtain in any necessary way when we take knowledge to be the more fundamental value. Mere true beliefs at least initially do not seem to have any necessary instrumental relation toward obtaining knowledge. After all, if a belief that p is a merely true belief, then that rules out the possibility that the belief that p instantiates knowledge. More, beliefs that are merely based on other true beliefs also will not instantiate knowledge, at least insofar as those other true beliefs lack some further epistemic merit. And even though mere true beliefs might sometimes be a causal means to the obtainment of knowledge in a way that is deviant vis-à-vis basing, such causal relationships only hold contingently, allowing for cases of mere true belief that are not instrumental in this respect. Mere true beliefs also do not seem to manifest a respect for knowledge or manifest an achievement of knowledge, especially in the way that Sylvan and Greco think of those relations. If there are necessary derivative value relations obtaining between knowledge and mere true belief, we will have to look elsewhere.

One might think that true belief, by itself, is epistemically good because it is an essential or constitutive part of knowledge and essential parts of good things are also good. Consider an example from Wayne Riggs (Reference Riggs, Hendricks and Pritchard2008: 316): “If an MRI machine is valuable because it helps save lives, then the individual components of it that make it work are also valuable.” Riggs calls this kind of mereological, derivative value relation “contributory” value. The issue, though, is that true belief is not literally a part of knowledge. Knowledge is not a mereological sum of true belief and other things and true belief does not “contribute” to knowledge in the same way that, say, an MRI machine's magnetic coils contribute to it. Insofar as being an MRI machine is good, so is being a certain kind of magnetic coil. We could literally replace a magnetic coil in an MRI machine with one of the same type and it would contribute toward the machine's operation in the same way. But, to speak in this way about mere true belief in its relation to knowledge would only be to engage in some lofty metaphor.

What could be more appropriately said about true belief in its relation to knowledge is that it is a necessary condition on knowledge. However, is being a necessary condition on some good thing, by itself, always a relation that generates derivative value? Consider a moral example: Benevolent actions are morally valuable and being an action is a necessary condition on being a benevolent action. Does that mean that there is some derivative moral value accruing to any action at all in virtue of it being an action? Of course not. It seems as though, again, if there is some necessary derivative value relation obtaining between knowledge and mere true belief, we will have to look elsewhere.

Tamar Szabó Gendler (Reference Gendler2002) presents a more promising strategy. When we evaluate certain kinds of things, we occasionally treat “impure instances” of good things as relevantly similar to “pure” ones (Reference Gendler2002: 47):

Supposing that squares are fundamentally good as a certain kind of ideal, certain approximations of squares might also be derivatively good in virtue of their square-likeness. Perhaps, then, supposing that knowledge is fundamentally good as an ideal end of propositional inquiry, mere true belief might also be derivatively good in virtue of its knowledge-likeness.

However, we need to be at least a little careful with this. Even if being a member of kind K is good and even given that K-likeness generates derivative value, there will still be gradations of goodness that are generated by degrees of K-likeness. Things can be more or less similar to other things in ways that are relevant to derivative value generation. For example, if squares are good in the way of geometrical beauty, then more derivative value will accrue to a nearly squared rectangle than to a four-sided figure that is not nearly square at all. Of course, it is difficult to differentiate degrees of similarity in a principled way (e.g., see Deutsch Reference Deutsch1998) but, in many cases where a certain gradational standard is fixed, it is relatively obvious that certain things are greatly similar to members of kind K, while others are intermediately similar to members of kind K, and while others yet are minimally similar to members of kind K, etc. Presumably, when derivative value is generated via K-likeness, the amount of value generated corresponds to the relevant degree of similarity. Only a minimal amount of value would accrue to things that are only minimally similar to members of kind K, more would accrue to things that are intermediately similar, and so on.

So, how similar to knowledge is mere true belief? Take, for example, Annie's true belief that there is a red car parked at the corner of Coral Way and Snapper Creek Road. There are certain respects in which Annie's belief is similar to a relevant token of knowledge; they are both cognitive states, they both represent the world as being a certain way, they are both veridical, etc. Yet, there are many respects in which they are dissimilar. Knowing that there is a red car parked at that intersection would have a particular kind of causal production, it would have a certain kind of justificatory status and relation to an agent's cognitive abilities, it would only endure in a circumscribed set of environments, etc.; all of these are features that Annie's mere true belief lacks. I would be tempted to say that mere true beliefs, those like Annie's that do not exhibit any further epistemic merit, are more like irregular trapezoids than they are like nearly square rectangles with respect to square-likeness. They are only minimally similar to the ideal of knowledge. This would mean that mere true beliefs have only a minimal derivative value in virtue of their knowledge-likeness.Footnote ¹⁵ As I have suggested above, however, this may be seen as exactly the right verdict, one that bottom-up approaches seem unable to replicate.

4. CONCLUDING REMARKS

Given the claims above, I think that it is well worth questioning whether true beliefs have a fundamental epistemic value. This is not to say that there is nothing to the slogan that we “aim at the truth” in our distinctively epistemic endeavors; it is merely to say that we might not be able to interpret this slogan as simply meaning that we aim at true belief. We look to grasp the truth in a stronger sense than merely believing it. The lesson is that, if the comparative evaluations of epistemic worth suggested above are legitimate, then we need to own up to the fact that, by focusing on true belief, many epistemologists have misidentified what is really at the heart of such evaluations. Our attention needs to be redirected. At least conditionally, these considerations should be seen as generating at least some abductive support in favor of the view that the constitutive goals of inquiry are fundamentally valuable from the epistemic perspective. Not only does this inquiry-centered way of seeing the epistemic evaluative practice provide an abstract, general solution to EVP, it can render fine-grained verdicts about particular cases in ways that some may think are more satisfactory than the most popular, truth-centric alternatives. At the very least, the proposal here sets a relatively high bar that any alternative theory must meet.Footnote ¹⁶