Jonathan Kvanvig's book The Value of Knowledge and the Pursuit of Understanding (2003), is a wonderful example of doing epistemology in a style that Kvanvig himself has termed ‘value‐driven epistemology.’ On this approach, one takes questions about epistemic value to be central to theoretical concerns, including the concern to provide an adequate account of knowledge. This approach yields the demand that theories of knowledge must provide, not just an adequate account of the nature of knowledge, but also an account of the value of knowledge. Given the near‐universal assumption that knowledge has a special kind of value, this demand seems reasonable, though surprisingly hard to satisfy. Another consequence of this approach to doing epistemology is that certain assumptions about epistemic value, like what sorts of things have it and what sorts of things don't, and where such value comes from, become much more salient to the epistemic enterprise. In his book, Kvanvig challenges the assumption that knowledge has some unique store of epistemic value. And he investigates the matter by asking questions about what the bearers of epistemic value are and where they get it. He concludes, of course, that knowledge as we have come to conceive it in twenty‐first century epistemology has no such special value.
Kvanvig frames his task in the book by taking up the question from the Meno, ‘why is knowledge more valuable than true opinion?’ This way of asking the question acknowledges that knowing is valuable, at least in part, because it implies that one has a true belief. It is uncontroversial that we value having true beliefs, so it is uncontroversial that knowledge is valuable at least in virtue of the true belief component of it. But the Meno question is prompted by a conviction that there is something particularly good about knowing, beyond thereby possessing a true belief. It is this value that Kvanvig is after, if it exists.
To pursue this quarry in the terrain of contemporary epistemology, Kvanvig is forced to reformulate the Meno problem slightly. In these post‐Gettier days, we are aware that there might be more than one property that is required in addition to true belief to adequately account for knowledge. Thus, it is not quite sufficiently general to ask why knowledge is more valuable than true belief, because knowledge might require both properties X and Y in addition to true belief. If so, then true belief + X might be more valuable than mere true (p.332) belief, but true belief + (X & Y) might be no more valuable than true belief + X. In that case, knowledge would be more valuable than mere true belief, but only because of the value of a proper subset of its components. In such a case, Kvanvig argues, it is not knowledge per se that is more valuable than true belief, but rather true belief + X, which falls short of knowledge. As a consequence, Kvanvig sets the requirement that must be met for a theory of knowledge to properly answer the Meno question to be that knowledge must be more valuable than any combination of its subparts. Hereafter, I will call this the Meno requirement.
I have explained that the need for generality seems to be what has led Kvanvig to this formulation, but I will argue that it has led him astray. This way of construing the Meno question is too demanding to be a plausible constraint on a theory of knowledge. Ironically, the strongest argument for this arises from some of the radical proposals he makes himself at the end of the book. In these brief remarks, I will make the case that Kvanvig's formulation of the Meno problem makes demands on a conception of knowledge that are too strong, and I will propose an alternative that serves the purpose just as well, while avoiding the problems of his original one.
Let us take a characteristic statement of Kvanvig's formulation of the Meno problem: ‘An adequate account of the value of knowledge must explain why it is more valuable than any subset of its constituents’ (Kvanvig 2003: 112). There are at least two reasons why this formulation is too strong. First, it begs the question against a certain way of thinking about the value of true belief. Second, it does not square with what Kvanvig himself says about the nature and value of understanding at the end of his book. I'll take these considerations in order.
But first, a quick note about what I am not claiming here. I am not claiming that by weakening the Meno requirement, some of Kvanvig's arguments against attempts to account for the value of knowledge thereby fail. It is nonetheless worth correcting the requirement for at least two reasons. One is that it sorts out a tension that I claim is present in the book itself. Kvanvig's formulation of the Meno requirement simply doesn't sit well with his later claims about the nature and value of understanding, and my formulation represents a friendly amendment to bring these two elements of his view into harmony. Another reason is that I hope some version of the credit theory of knowledge will be able to answer Kvanvig's charges and account for the value of knowledge.1 Though I can't make that case today, I at least want to make it easier by watering down—I mean, getting right—the Meno requirement.
2. HOW MUCH FOR THAT TRUTH?
It has been noted in passing at various points in the literature that there is some intuitive pull to the idea that not all truths are created equal. That is, the value of some true beliefs seems to exceed that of others, based on the contents of those beliefs. Ernest Sosa expresses a view in this neighborhood at least, in ‘The Place of Truth in Epistemology.’
Sosa here points out that the value of a true belief about the number of grains of sand in one's hand is, intuitively (and absent any specific reason to find out such a thing), not very valuable despite its truth. Compared to a true belief about, say, the nature of reality or the moral law, Sosa seems right. Taking for granted for the moment that there is an intuitive value‐discrepancy between true beliefs of these sorts, how shall we account for it?
At the beach on a lazy summer afternoon, we might scoop up a handful of sand and carefully count the grains. This would give us an otherwise unremarked truth, something that on the view before us is at least a positive good, other things equal. This view is hard to take seriously. The number of (p.333) grains would not interest most of us in the slightest. Absent any such antecedent interest, moreover, it is hard to see any sort of value in one's having that truth. (Sosa 2003: 156)
It is possible, of course, that this value‐discrepancy is purely a matter of a difference in non‐epistemic value, in which case it would be irrelevant to a discussion of the Meno problem. After all, it is uncontroversial that a true belief about the correct way to disarm a bomb has considerably more prudential value for me when said bomb is about to go off than does a true belief about tomorrow's weather at that moment. But once we begin to raise questions directly about the sources and bearers of epistemic value, it is at least reasonable to ask whether the intuitive value‐discrepancy between some true beliefs and others can be explained, at least in part, by some difference in epistemic value.
One might think, for example, that certain subjects are intrinsically and epistemically more worthy of contemplation, and even belief, than others. Bob Roberts and Jay Wood argue for this in their book on intellectual virtues. About epistemic goods in general, they say, ‘[e]pistemic goods are not all created equal; in fact, some of them are so far down the value scale as hardly to be goods at all’ (Roberts and Wood 2007: 208–9). They list several criteria according to which a virtuous believer will discriminate the relative value amongst various epistemic goods. One such criterion is the evidential supporting role the belief plays in one's overall system of beliefs.
If beliefs often need support, then beliefs can gain in value by being supports for other beliefs. So not only is having support a sometime desideratum in beliefs; a belief's providing support for other beliefs can also make it more interesting than some others, and a belief can be trivial because of its utter lack of a supporting role. (Roberts and Wood 2007: 156)
But some beliefs can be more valuable than others simply on the basis of their content. Roberts and Wood discuss this under the criterion of the ‘worthiness’ of beliefs. Some beliefs are more worthy than others because of their bearing on important issues beyond the epistemic—beliefs about the guilt or innocence of a defendant at a trial, for instance (Roberts and Wood 2007: 158). But not all differences in value between beliefs is explicable by appeal to non‐epistemic considerations.
[W]e are not saying that knowledge is valuable only if it is ‘practical.’ Some things are worth knowing even if the knowledge has no ‘application.’ Why is it worth knowing how old the universe is, while it is not worth knowing how many grains of sand are in a particular cubic centimetre of the Sahara Desert (assuming that both truths are ‘useless’)? . . . The universe, with all its processes, is worthy of respect. And this worthiness of the objects of knowledge is tied to their particular character—their particular complexity and simplicity, the particulars of their structure and composition and functions. The human genome is interesting because of what it is, whereas the cubic centimetre of the Sahara, simply as so many grains of sand together, is uninteresting because of what it is. (Roberts and Wood 2007: 158)
I am not endorsing any of these views, but I think the fact that they are possible views, and, in my opinion, not obviously false, militates against placing a requirement on theories of knowledge that it be impossible for any true belief to be as valuable or more (p.334) valuable than any instance of knowledge. Such a requirement seems to assume without argument that knowledge exemplifies all possible epistemic value. Otherwise, it would be possible that a belief, true or false, might derive sufficient value from some source other than by being an instance of knowledge, that it becomes as valuable or even more valuable than some instances of knowledge. I see no reason to preclude this possibility until it is shown that either (a) any such value is necessarily going to be less valuable than an instance of knowledge would be, or (b) there is no other such possible source of value.
It may be, though, that we can modify Kvanvig's formulation of the Meno requirement in a way that preserves the generality that prompted the problematic formulation, yet allows for the possibilities mentioned above. Perhaps we could say that an adequate account of the value of knowledge must explain why
(∀s)(∀p)[Value(sKp)>Value(sRp)] (where R is some relation comprising elements of K, and R ≠ K)
After doing some necessary spade work, Kvanvig narrows down our commonsense usages of the term ‘understanding’ to two that are particularly relevant to theoretical epistemology. ‘My suggestion . . . is that we focus on understanding in two central uses: when understanding is claimed for some object, such as some subject matter, and when it involves understanding that something is the case’ (Kvanvig 2003: 189). This leaves out common uses of the term to indicate that one has grasped the meaning of someone's utterance, or that one understands how to do something. I do not take issue with Kvanvig's choices here, so I will continue on the assumption that these two uses are the ones that are of primary epistemological interest.
Kvanvig has a lot of interesting things to say about understanding, but there are three points of particular relevance for my purposes. First, he claims that understanding, like knowledge, is factive. Thus it makes no sense to say that someone understands X, though X is false. But it is not entirely clear what this means when applied to the notion of understanding a subject matter. Kvanvig acknowledges this awkwardness, but dismisses it quickly,
(p.335) This appears to explain the sense in which one's understanding, which comprises, perhaps among other things, some beliefs, is factive in a derivative sense. Whatever beliefs are implicated in one's understanding of X must be true.
Objectual understanding is, of course, not straightforwardly factive, for only propositions can be true or false. Still, the uses I wish to focus on are ones in which factivity is in the background. For example, to understand politics is to have beliefs about it, and for this objectual understanding to be the kind of interest here requires that these beliefs be true. (Kvanvig 2003: 191)
This strikes me as an implausibly strong requirement for understanding, especially when conjoined with the second of Kvanvig's claims about understanding that are relevant to my present point—that understanding comes in degrees. This second point seems right, since both the degree of explanatory coherence (which Kvanvig takes to be a likely way to account for understanding) as well as the amount of information present in someone's understanding can vary. Why not, then, allow that one can have objectual understanding that includes false beliefs, while acknowledging that such false beliefs lower the degree of understanding one has?
Kvanvig himself actually appears to soften the factivity requirement a bit later in his discussion of understanding. For instance, he says,
My objection here may just be a quibble, but I am not sure that we can make sense of what Kvanvig here proposes. First of all, there seem to be cases in which one can have understanding even when one believes something false that is not ‘peripheral,’ but rather central. Secondly, I'm not sure how to understand the sense in which we can talk about the understanding one has of a subject matter once one has dropped out of consideration some perhaps fairly key beliefs that bear evidential and explanatory connections with other beliefs, your grasp of which constitutes your understanding. Consider an example.
There is still something of a problem here, though, for it is hard to resist the view that understanding may be correctly ascribed even in the presence of some false beliefs concerning the subject matter . . . When the falsehoods are peripheral, we can ascribe understanding based on the rest of the information grasped that is true and contains no falsehoods. In such a case, the false beliefs are not a part of the understanding the person has, even though they concern the very material regarding which the person has understanding. So in this way, the factive character of understanding can be preserved without having to say that a person with false beliefs about a subject matter can have no understanding of it. (Kvanvig 2003: 201–2)
Suppose I understand my wife. I can reliably predict her behavior in a wide range of circumstances, I possess a lot of facts about her past and about her present dispositions, and so on. Included in this set of my beliefs about her past is that she suffered a very traumatic boating accident when she was 3 years old that resulted in her refusal ever to travel on water. Her lifelong attempts to conquer her fear have failed over and over, and all this has caused a certain amount of guilt and pain and humiliation, all of which I have witnessed, and I have a deep appreciation of how all this has affected her character, etc. But, as it happens, she never had a boating accident. Instead, her parents neglectfully let her wander off while they were on vacation in Mexico when she was 2 (not 3), and she fell off a pier into the ocean. As she got older, her parents explained her fear of the water by making up a ‘boating accident’ that did not render them culpable for the resulting psychological damage.
Obviously, my belief about her boating accident (as well as her belief) is false, and if understanding is strongly factive, it cannot be included in my understanding of my wife. Yet this belief is central to my understanding of my wife's history, experience, and personality. In other words, it hardly seems peripheral. Moreover, even if I was willing to grant that it should be ‘deleted’ from what constitutes my understanding of my wife, (p.336) it is hard for me to imagine what it would be like to ‘subtract’ this one belief from my total understanding of her. What are we to do with all the evidential and explanatory connections that would be ‘left dangling’? Are they to be abandoned as well? If so, it looks as though my understanding of my wife suffers a terrible blow once we subtract my false belief about her alleged boating accident. But it does not seem plausible to me that such a judgment is reasonable. The lesson I take from this example is that my understanding of my wife survives the falsehood of even this central belief, and so even the more lenient constraint of factivity that Kvanvig admits must be loosened some more.
The final, and most significant, of Kvanvig's claims about understanding that I want to highlight is that it is not a species of knowledge. Kvanvig at first acknowledges the intuitive pull of the idea that it is, but then offers several arguments against it. The main argument here is that the primary determinants of knowledge and of understanding are quite different.
Kvanvig buttresses this general argument by appeal to some specific examples, but since I want to grant this point, I need not belabor it here.
Note that the crucial features just discussed concerning understanding draw attention to things other than what is central to knowledge . . . [O]nce we move past its factivity, the grasping of relations between items of information is central to the nature of understanding. By contrast, when we move past the factivity of knowledge, the central features involve nonaccidental connections between mind and world. (Kvanvig 2003: 197)
The two conclusions I want to leave this section of the essay with are that, by Kvanvig's own lights, he ought to acknowledge that some, and perhaps even a significant, departure from factivity is consistent with even a very high degree of understanding, and Kvanvig's avowal that one can have an understanding of X even in the absence of knowledge about X.
4. THE VALUE OF UNDERSTANDING
There is one more point about understanding that needs making before I can return to the issue of how to properly formulate the Meno requirement. I take it to be uncontroversial that understanding is epistemically valuable. But what Kvanvig calls objectual understanding, at least, is a complex and, more to the point, composite thing. Even remaining neutral with respect to whether understanding M is exhausted by a list of beliefs or known information, it is typical that understanding M requires that S have some beliefs about the subject matter that alone do not constitute the whole of S's understanding. Therefore, understanding has parts. And it is valuable. Though the literature on extrinsic value is sparse on the subject, it is customary to suppose that an object that is a contributing part of a valuable whole derives value from that relationship. The fuel injection system in my car is valuable to me because it contributes to the ability of my car to provide transportation, which is valuable to me. Similarly, the beliefs that partly constitute my understanding of M are valuable to me because they contribute to that understanding. Therefore, being a part of one's understanding of some subject matter, M, can be a source of epistemic value for one's beliefs.
When we left our hero, the Meno requirement, it looked like this:
(∀s)(∀p)[Value(sKp)>Value(sRp)] (where R is some relation comprising elements of K, and R ≠ K)
All we need to show that the above formulation is problematic is that it is possible that S's believing but not knowing that p could have equal or greater epistemic value for S than S's knowing that p. And the things we learned about understanding will allow us to show just such possibilities. Imagine that S understands some subject matter, M, to a very high degree. Included as a part of this understanding is S's belief that p. Let us further suppose that p is fairly central to S's understanding of M. As we have seen, S's belief that p will derive some epistemic value from its inclusion in S's understanding of M. It is at least possible that this value could be equal to or greater than the value of S's knowing that p, in a case where that item of knowledge is not a part of S's understanding of M or of anything else. If so, then we have a case in which S's belief that p has equal or greater value than S's knowing that p.
Of course, a lot has been left unexplained here. In particular, I have given you no indication of how we are supposed to measure and compare the ‘total epistemic value’ present in each of these two possible cases. Indeed, there is no guarantee that these two different kinds of epistemic value are even commensurable, allowing a meaningful comparison to be made. But all I am trying to show here is that even the reformulated Meno requirement begs the question against such possibilities as described above. As long as there is a possible source of epistemic value for a belief that is independent of whether that belief constitutes knowledge, then there will be the possibility that the belief could derive enough value from that alternative source that it is equally or more valuable than it would be if it were known. Being a part of one's understanding of a subject matter has been shown, by Kvanvig's own lights, to be just such a possible alternative source of epistemic value. So it is unreasonable to require that S's knowing that p always be more valuable than S's bearing some relationship short of knowledge to p, because S's bearing that other relationship to p is compatible with p deriving value from an alternative source. As I indicated before, the only way such a requirement would make sense is if it were already determined either that there was no other source of epistemic value for beliefs besides being true or being an instance of knowledge, or that such alternative value can never rival the value derived from the more traditional sources. I see no reason to accept either of those views at this point.
As a brief aside, notice that it is also possible, at least, that a false belief could derive value from being a part of S's understanding of M, assuming I am right that the factivity requirement on understanding should be liberalized somewhat. Though it is less plausible, it is still possible that even such a false belief could derive enough value from being part of S's understanding that M, that it is more valuable than at least some instances of S's knowing that p.
(p.338) We need yet another reformulation of the Meno requirement, preferably one that is still sufficiently general to satisfy the circumstances of contemporary epistemology, but that does not impose unreasonable demands on a theory of knowledge. The problems raised for the previous formulations arose from the fact that truth and knowledge may not be the only sources of epistemic value out there. If not, then formulating the Meno requirement in terms of the relative amount of value exhibited by a mere belief on the one hand and an item of knowledge on the other seems a poor way to proceed. It will always be possible that the mere belief could derive enough value from elsewhere to compare favorably with the item of knowledge in terms of the amount of value involved. I think the better way to construe the Meno requirement is that an item of knowledge must have a kind of epistemic value, or perhaps a source of epistemic value, that no belief that fails to count as knowledge has. This does not guarantee that knowledge is always more valuable than whatever falls short of knowledge, but it does insure that knowledge always has a special kind of value that nothing that falls short of it can have.
Formulating the requirement in this way avoids all the problems raised for previous versions. All those problems stemmed from the fact that the requirement was put in terms of relative quantity of epistemic value between what is known and what is believed but not known, which is not guaranteed always to come out in favor of the former. Yet I believe that it captures the spirit of the original Meno requirement. For instance, one still has an answer to the question, ‘why should I prefer knowing to not‐knowing?’ The answer is that only by achieving knowledge can I attain epistemic value of a certain sort. Thus, all else being equal, knowing will be more valuable than not‐knowing. But we have to acknowledge that all else is not always equal. We could, perhaps, have gotten to this point more quickly by simply adding the ‘all else being equal’ caveat to the second formulation of the Meno requirement, but that would have left obscure the kinds of things that must be equal, and so we would have been left with a poorer understanding of the value of knowledge. Also, putting the Meno requirement in terms of a unique kind or source of epistemic value that only comes with knowing highlights the fact that knowing still has an exalted position for a reason. It gives us something we can't get anywhere else.