1. Fidelity: a primer

There are two senses in which a belief can “get it right”. The first sense of “getting it right” is truth. A belief “gets it right” when it is true and does not “get it right” when it is false. The second sense is what I call fidelity. A belief “gets it right” when it has fidelity and does not “get it right” when it lacks fidelity.

Fidelity is not some newfangled philosophical posit, though it has at times been presented as such. Philosophers have acknowledged it, in one form or another, for millennia. Plato, in the voice of Socrates, spoke admirably of carving each kind along its natural joints while avoiding any splintering indicative of bad butchery. The medieval scholastics fiercely debated whether the categories (e.g. substance, accident) accurately divide the world or rather reflected conceptual or linguistic biases. More recently, David Lewis stressed the importance of interpreting the words and thoughts of others as tracking the comparatively more natural properties of the world, and Theodore Sider has discussed the importance of employing ideology that matches the world's metaphysical structure.Footnote ¹

These philosophers are all, more or less, articulating the sense of “getting it right” that I call fidelity. (Personally, I favour the nuances behind Sider's articulation of fidelity. But that's not important here.Footnote ²) In each case, the posited correspondence is a form of representing the world. What that representation precisely consists of differs. But, loosely speaking, the representation is organizational. Because the correspondence is to be evaluated by the accuracy of its representation, its success is essentially sensitive to how the world is organized. Furthermore, the way this correspondence “gets it right” is not equivalent to the way truth “gets it right” – not obviously so, anyway. To use an imperfect, but hopefully illuminating, slogan: matters of truth are decided by ‘yes's and ‘no's, whereas matters of fidelity are decided by ‘these's and ‘those's.

As I understand it, fidelity is ultimately a feature of concepts. A concept is a mental representation. So, too, is a belief. But these two kinds of representation are not independent. A belief is constituted by concepts and what a belief represents depends on the concepts that constitute that belief. Consequently, the fidelity of a belief is derived from the fidelity of the concepts that constitute it. Li Si's belief that the bracelet is green is partially constituted by the concept bracelet and the concept green. The fidelity of her belief therefore depends on the fidelity of those two concepts. In this way, fidelity is like other decomposable features of complex wholes.

Of course, not everyone shares my representation-based understanding of beliefs and concepts. Some philosophers may understand belief in terms of something non-representational, like patterns of behaviour or causal relations. There's no need to fight about the nature of beliefs right now. I am interested in the mental activity that occurs in someone's mind when, for example, they just remembered that they left the stove on. I will call that mental activity a belief, though not much rests on whether or not it “really is” a belief. This mental activity is partially constituted by other mental activities. Similarly, though, I don't want to insist at all costs that these smaller activities “really are” concepts. I only assume that there are such representational mental activities.

That's the basic idea behind fidelity. Beyond that, there are more complex questions to address. For instance: does fidelity come in degrees, and if so, how fine-grained is the measurement? I will assume, without argument, that fidelity does come in degrees and that there are two factors that figure into this gradation. First, beliefs typically employ more than one concept, and so a belief can have greater or lesser fidelity insofar as some of its concepts match the world and others do not. Second, each individual concept might have greater or lesser fidelity insofar as they more accurately match the world than rival concepts. I will also assume that there is some threshold above which the fidelity of a belief is “close enough”. Every belief above this threshold is a fidelic, or joint-carving, belief. At the limit, a fidelic belief might just be a belief that has the maximum degree of fidelity. Similarly, I will assume that there is some threshold below which the fidelity of a belief is “too far off”. Every belief below it is a non-fidelic, or gruesome, belief. Finally, I will assume that these two thresholds are one and the same. As a consequence of this “bivalence” about fidelity, every belief is either fidelic or non-fidelic. These assumptions might very well be wrong. But they will help to streamline my discussion of fidelity's epistemic value.

In the example above, the belief that the bracelet is green has greater fidelity than the belief that the bracelet is grue. But fidelity need not be restricted to the natural world. The belief that murder is wrong has greater fidelity than the belief that murder is wrong on Wednesdays because morality does not (ordinarily) depend on the day of the week. Though the day-specific moral belief is true, it misrepresents the universality of the moral facts. Furthermore, some beliefs about social categories such as gender, race, and sexual orientation might have greater fidelity than others, even if these categories are social constructions. Fidelity is applicable to any belief about any topic insofar as that belief reflects a way that the world is organized. Nevertheless, in what follows I will not rely on any specific assumptions about which of our concepts reflect how the world is organized and which do not.

Of course, plenty of philosophers deny that the world is organized.Footnote ³ While such a denial does not automatically constitute a denial of fidelity, it does trivialize it. Consider the analogous situation for truth. Many philosophers hold that truth is a kind of correspondence, a belief being true just in case it “gets it right” with respect to what's going on in the mind-independent world. Those who deny that there is a mind-independent world do not thereby deny this kind of correspondence. But they do trivialize it and are consequently likely to replace the correspondence with some other relation, perhaps even calling that relation “truth”.

As the analogy to truth shows, I need not fully respond to such deniers before exploring the potential value of fidelity, no more than I would need to fully respond to anti-realists before exploring the potential value of truth. Thus, in this paper I will assume without further argument that belief has this second sense of “getting it right” and that in a good number of cases its success or failure is a non-trivial matter.

That being said, it is worth saying one more thing about the fidelity of belief. The details here will be important in the subsequent exploration of its value.

Often, fidelity is given a linguistic gloss. As Sider has put it, “a fact is fundamental when it is stated in joint-carving terms” (Sider Reference Sider2011: i). But this linguistic gloss is inappropriate when it comes to the question at hand. Belief is a mental phenomenon. We sometimes express our beliefs with words, and those words might (or might not) carve the world at its joints. But so far as the fidelity of a belief goes, it is fidelity with respect to our concepts – more specifically, our primitive concepts.

What makes a concept primitive? Here, I want to be as accommodating as I can be and avoid needlessly presupposing any specific theory of concepts. So, in lieu of a full account of conceptual primitiveness, I offer a partial characterization that should suffice for the purposes of this paper.

First, a rough definition: a concept is primitive relative to an agent just in case (i) the agent regularly employs the concept, and (ii) the agent predominantly employs the concept to the exclusion of rival concepts. According to this definition, neither the source of the concept nor the time of its acquisition determines whether or not it is primitive (though, to be sure, these factors often shape which concepts agents have primitively and which they do not). The definition also does not explicitly acknowledge what the agent could do with a concept, what they would ideally do, and so on. The definition is about what the agent in fact does with their concepts. Furthermore, by acknowledging the regularity of the concept's employment, the rough definition allows for change in which concepts are primitive relative to the agent.

Some examples may help to further illuminate what I mean by conceptual primitiveness. Zilin is a mathematician who accepts the successor-based reduction of the natural numbers and the arithmetic functions. This morning, when she calculated the number of eggs to buy for breakfast, she multiplied the number of people eating by the number of eggs each wanted. The cognitive story never involved succession. Assuming that this is what regularly occurs in Zilin's head, multiplication is a primitive concept for her, even though she could and sometimes does “reduce” it to succession. Adam is an American who now lives in New Zealand. At first, whenever he read the weather forecast, he needed to mentally convert from Celsius into Fahrenheit before he could determine whether he should wear a jacket. Now, decades after his relocation, he can determine what to wear without converting. At first, only Fahrenheit was conceptually primitive for Adam. But over time it has been replaced by Celsius. Zilin and Adam meet at an academic conference. While describing her background, Zilin shares a story about the time she and her sister visited their uncle in Chengdu. Adam shares a similar story about the time he and his younger sister visited their father's older brother in San Francisco. Of course, both can speak English and (Mandarin) Chinese. Because they are practicing their non-native languages, Zilin shares her story in English, using the English words ‘sister’ and ‘uncle’, while Adam shares his story in Chinese, using the Chinese words ‘妹妹'’ and ‘伯伯’. Despite what Zilin says in English, she conceives of her sister as her 妹妹 and her uncle as her 伯伯. Despite what Adam says in Chinese, he conceives of his 妹妹 as his sister and his 伯伯 as his uncle. Their primitive conceptions of family are different. Adam's primitive conceptions are based on the gender of his family members, whereas Zilin's primitive conceptions are based on gender as well as birth order.

One belief has a higher degree of fidelity than another to the extent that the primitive concepts employed in that belief better represent the way the world is actually organized. What someone believes can come apart from what they say. So, too, can the fidelity of someone's belief come apart from the fidelity of how they express that belief. Similarly, the fidelity of someone's belief can come apart from the fidelity of the proposition they believe. Some theories of propositions do not distinguish between propositions about younger sisters and propositions about 妹妹. On such theories, the content of Zilin's beliefs may be the same as the content of Adam's beliefs. But clearly something different is going on in their minds. Zilin employs the comparatively more detailed Chinese familial concepts even though she utters English familial words and even if she expresses a comparatively course-grained proposition. This conceptual employment, whatever it is, is what is important in the evaluation of a belief's fidelity.Footnote ⁴

2. The epistemic value of fidelity

Once we acknowledge fidelity as a second sense of “getting it right”, we face a wide range of questions regarding the precise nature of its epistemic value. In this section, I'll articulate some potential answers to these questions. Though some of these answers seem to me to be more plausible than others, I think they all deserve to be considered more fully.

Truth and fidelity are only two among many epistemic features that are sometimes regarded as epistemically valuable. Some others include: accuracy, adequacy, aptness, coherence, creditability, explanatoriness, generality, intellectual virtue (and all of the more specific intellectual virtues such as caution, courage, humility, and perseverance), justification, knowledge, proper functioning, reliability, systematicity, and understanding, as well as social features such as trust, epistemic justice, and epistemic autonomy. (I'm sure there are others that I've missed, and for that I apologize.)

Which of these features are actually valuable and why they are valuable is a matter of dispute.Footnote ⁵ But almost everyone agrees that there are some connections between them. Let me briefly introduce some terminology to help articulate these connections, and then apply that terminology to the case of fidelity.

Some beliefs we value for their own sake. Philosophers disagree as to what feature makes these beliefs valuable for their own sake. But many think that, for at least some of these beliefs, we value them at least in part because they are true. In this sense, truth has final epistemic value.Footnote ⁶ Similarly, some beliefs are epistemically disvaluable because they are false; falsity has final epistemic disvalue. But consider an epistemic feature such as reliability. If some belief of mine comes from a reliable belief-forming process, then that belief is probably true. Because reliability directs me toward that which has final epistemic value – in this case, toward the truth – reliability itself has a sort of indirect epistemic value. Call this indirect value instrumental epistemic value. Many philosophers, though, think that the instrumental value of reliability exhausts its epistemic value; they deny that reliability has final epistemic value. If they're right, we can say that reliability only has instrumental epistemic value, or has mere instrumental epistemic value.

Many of the above-mentioned epistemic features are indirectly connected to each other in the way that reliability is indirectly connected to truth. But some of them have an even more intimate connection. For instance, part of what it is to know something is to have a belief that is true. Plausibly, we value knowledge for its own sake; it has final epistemic value. But some philosophers (e.g. David Reference David, Steup and Sosa2005) think that its final value is entirely comprised of the final value of the truth that it includes. For them, knowledge has only non-basic final epistemic value, whereas truth has basic final epistemic value. Other philosophers (e.g. Greco Reference Greco, DePaul and Zagzebski2003) think that some other component of knowledge contributes final epistemic value. And, in theory, a philosopher could argue that knowledge has basic final value that is greater than the basic final value of its elements.Footnote ⁷

The instrumental/final distinction and the basic/non-basic distinction are importantly different. I won't attempt to give a full account of the difference here, because I do not think that the details matter for what I want to say in this paper. Suffice to say that the connection between the instrumental and the final seems to be something like a causal, correlative, or probabilistic connection and the connection between the basic and the non-basic seems to be something like a constitutive, mereological, or grounding connection. That gloss leaves a lot to be developed, but it should be informative enough to put the distinctions to fruitful use.

To understand the overall epistemic value of fidelity, we need to identify the source of that value. Does fidelity have instrumental value? If so, is that all it has? Or does it also have some sort of final value? If so, is its final value basic or non-basic?

Fidelity might have instrumental value. For instance, because fidelity tracks objective similarities, maybe fidelity promotes more successful inductive inferences. Or maybe fidelic beliefs lead to better theories, which lead to better experimental designs, which lead to more and better discoveries. Maybe not. Maybe, because of our psychological limitations, thinking in fidelic terms actually increases the number of errors we make. (Imagine trying to describe the fall of the Roman empire using the concepts of quantum physics!) Unlike reliability, fidelity lacks analytically specified instrumental connections. Questions about the instrumental value of fidelity are therefore empirical to a degree that renders me unable to do anything more than speculate. Since such speculation has little value, I leave these questions to future empirical studies. I simply note that the claim that fidelity has instrumental epistemic value is not obviously wrong.

Thankfully, I can do a bit more than speculate about fidelity's final value. Many philosophers seem to think that it does have final value. One of the most prominent proponents of this position is Sider. As he forcefully puts it:

According to Sider, the mistake these people make is not a merely instrumental one. When they divide the world into bred regions and rue regions, they make a mistake with respect to what they believe now, regardless of their mistaken beliefs’ consequences. It would have been better if they had used concepts that divided the world into blue regions and red regions. If they had done so, their beliefs would have had greater fidelity. In virtue of that greater fidelity, their beliefs would have had greater final epistemic value.

Yet even if red/blue beliefs have greater basic final value than bred/grue beliefs and even if that difference is because the former have a greater degree of fidelity, it does not follow that fidelity itself has basic final value. I see at least two alternative explanations for the difference in value.

First, the red/blue beliefs might constitute a complex epistemic feature that itself has basic final value. Whether or not complex epistemic features have basic final value is a controversy I don't intend to settle here. But suppose that they do. Arguably, fidelity is a constitutive element of some complex epistemic features. Consider the state of understanding. When an individual comes to understand something, they come to acquire something that goes beyond mere truth. Perhaps in some cases this acquisition involves fidelity.Footnote ⁸ To understand something is, in part, to possess a genuine explanation of that thing. These explanations may be causal – as when a chemist explains the reactive dispositions of the elements – or they may be non-causal – as when a philosopher explains that goodness is what the gods love. These explanations only reflect a genuine understanding when they are stated in joint-carving terminology. In other words, fidelity is a constitutive element of understanding. So, if understanding has basic final value, then the final value of fidelity might be merely non-basic.

Second, the red/blue beliefs might form an organic unity that is more epistemically valuable than the organic unity formed by the bred/rue beliefs. As Moore (Reference Moore1903) puts it: “The value of a whole must not be assumed to be the same as the sum of the values of its parts” (p. 79). Moore suggests that the (non-epistemic) value of someone admiring something beautiful is greater than the admiration and the beauty taken on their own. But the value of an organic unity need not be an “amplification” of the value of its parts. The value of a whole may be positive even if one or all of its parts lacks value. Perhaps, for example, musical rests lack (non-epistemic) value even though music is generally more (non-epistemically) valuable with rests than without. We might say something similar about the value of complex epistemic features and their constituents.Footnote ⁹ But the point stands even for beliefs that do not constitute complex epistemic features. Even if a belief that is both true and fidelic is more valuable than a competing belief that is true and non-fidelic, it does not follow that fidelity itself has value, let alone basic final value.

This fact raises a troubling possibility. “Getting it right” is epistemically valuable and true beliefs are more valuable when they are also fidelic. But what if that's simply because such beliefs are a specific way of getting the truth right? In other words, what if the value of “getting it right” derives entirely from the basic value of truth? To that possibility I now turn.

3. Veritism unmotivated

As I mentioned earlier, many philosophers think that truth is a basic epistemic value. Many of those philosophers go even further and think that truth is the only basic epistemic value. Such philosophers – veritists – think that the apparent epistemic value of all other epistemic features can be either explained away or explained in terms of truth. Kristoffer Ahlstrom-Vij, for example, argues that the state of understanding is epistemically valuable only as an instrumental means to further truths; any apparent surplus value derives either from the practical value of having systematized beliefs or from the prudential value of feeling a sense of closure and completeness.Footnote ¹⁰

At first glance, fidelity seems amenable to a similar treatment. Even if fidelity is a distinct sense of “getting it right”, its apparent value might be explained in terms of truth. Look at how contemporary metaphysicians talk about the importance of fidelity: they talk about the importance of acquiring fundamental truths, the importance of securing inductive inferences, the importance of establishing substantive disputes where one disputant affirms and the other denies the same claim, and so on. In short, their talk looks suspiciously truth-centric.

But looks can be deceiving. Below, I sketch five initially plausible strategies a veritist may use to explain the value of fidelity in terms of the value of truth. Each strategy faces problems without straightforward solutions. A firmly committed veritist might find a way to solve these problems. Yet I think the problems are significant enough to show that the value of fidelity is not straightforwardly reducible to the value of truth. As it stands, then, the veritist's commitment is unmotivated.

The first strategy takes the value of fidelity to be merely instrumental to the basic value of truth. If someone uses joint-carving concepts, they are more likely to acquire true beliefs and avoid false beliefs. Similarly, if someone uses gruesome concepts, they are less likely to acquire true beliefs and avoid false beliefs. As I mentioned before, there is nothing inherent to the notion of fidelity that analytically guarantees such instrumentality. So in the absence of empirical evidence there is no general reason to think that fidelity is instrumental to the truth. However, there is a connection between fidelity, objective similarities, and inductive inferences. Inductive inferences succeed because the future resembles the past; they succeed because in general objects that are objectively similar remain objectively similar. Thus, the green objects we observe now will be like the green objects in the distant future, whereas the grue objects we observe now might not be like the grue objects in the distant future. Consequently, inductive inferences made on the basis of joint-carving concepts are more likely to lead to true beliefs; similarly, inductive inferences made on the basis of gruesome concepts are liable to lead to false beliefs.

Unfortunately, the connection between fidelity and inductive inferences is not quite that simple. Many of the stock examples of joint-carving concepts and their gruesome counterparts involve temporal “shuffling”. But fidelity can sometimes be a matter of specificity. The concept green has greater fidelity than the concept green or blue. Yet inductive inferences that move from “This F is green” to “All Fs are green” are less reliable than inductive inferences that move from “This F is green or blue” to “All Fs are green or blue”. The opposite is true when the form of the inference is reversed. Importantly, though, more specific concepts are not always more joint-carving. As I suggested earlier, the concept wrong is epistemically better than the concept wrong-on-Wednesday because the former better represents the universal, non-specific, nature of morality. All of this points to the conclusion that the value of fidelity cannot be fully explained in terms of its instrumental value in inductive inferences. Perhaps there is some other reason to think that fidelity has instrumental value. But the burden of proof rests with the veritist.

According to a second strategy, fidelity is an indicator for something else that has epistemic value but fidelity itself does not have epistemic value. Consider the psychology of Li Si, who looks at Zhang San's bracelet and forms the belief that the bracelet is grue. Most likely, she also believes that the bracelet is green. If she doesn't believe that the bracelet is green, then something has gone wrong. The only reliable way to form the belief that the bracelet is grue is to first observe that the bracelet is green. But if Li Si does not believe that the bracelet is green then she is failing to believe something that her perceptual evidence shows is obviously true. Thus, the second strategy provides a forked explanation. Either someone who has a gruesome belief also has its joint-carving counterpart, or they do not. If they do, then the gruesome belief only has a sort of non-basic value based on the basic value of the truth of a distinct belief. If they do not believe the joint-carving counterpart, then they are failing to base their beliefs on the evidence that they are given. That failure is instrumentally disvaluable insofar as it will prevent someone from acquiring true beliefs and avoiding false beliefs. Thus, fidelity appears to have value because joint-carving beliefs indicate that the believer is not employing an instrumentally disvaluable process of belief formation. But the indication is merely a correlation. In this way, the second strategy tries to explain away the apparent value of fidelity.

This explanation makes sense in the specific case of Li Si looking at a jade bracelet, but it fails to generalize. First, Li Si could form the belief that the bracelet is grue on the basis of Zhang San's testimony that the bracelet is grue. The veritist may respond by saying that even in that circumstance Li Si should be able to deduce that the bracelet is green, and so if she does not believe that the bracelet is green then there is something wrong with how she is arriving at the truth. Fair enough. But the issue depends on which concepts are primitive for the agent. For normal humans such as Li Si, the concept green is primitive and the concept grue is conceptually defined in terms of green. Thus, it would be strange for Li Si to believe that the bracelet was grue but not believe that the bracelet was green. Yet there could be aliens for whom grue is primitive. They could even lack the concept green. For such aliens, the second strategy's explanation for the merely apparent value of fidelity fails.

And yet we don't need aliens to show that the explanation fails to generalize. Li Si believes that the bracelet is made of jade. But Li Si is not a chemist and does not know the chemical make-up of the bracelet. Thus, she does not believe that the bracelet is made of nephrite, that it contains silicon, and so on. As a matter of fact, everything that is made of jade also contains silicon. Nevertheless, there is nothing unusual with how Li Si forms her belief. Consequently, a veritist who adopts the second strategy would have to deny that Li Si's belief that the bracelet is made of jade is less valuable than the belief that the bracelet is made of nephrite.

The third strategy employs the notion of a fidelity belief. A fidelity belief is a belief that explicitly maintains that some concept is joint-carving. The strategy says that every individual believes the fidelity beliefs that support their primitive concepts. Perhaps, even, they must believe them. Plausibly, every individual must be immodest in that they believe their beliefs to be more accurate than anyone else's – if they were not immodest, then they would just adopt the beliefs of those they judged to be more accurate, in which case they would still believe their beliefs to be the most accurate, albeit indirectly.Footnote ¹¹ The third strategy extends this need for immodesty to fidelity: every individual must believe that their concepts are more joint-carving than the alternatives – otherwise, they would come to employ those alternative concepts. So, this third strategy concludes, an individual's fidelic belief has a sort of non-basic epistemic value insofar as it entails that they have a true fidelity belief.

But this strategy seems psychologically unrealistic. It seems to me that plenty of people have beliefs such that they have no opinion about the fidelity of the concepts employed in those beliefs. Children, for instance, have beliefs but lack an explicit concept of fidelity. Even those who do have an explicit concept of fidelity may be ambivalent about the fidelity of some of their beliefs. Recall that an individual can lack full voluntary control over which concepts are primitive to them. When Adam first moved to New Zealand, he could not help but think of temperature in terms of Fahrenheit. The third strategy would say that Adam had a fidelity belief to the effect that the Imperial measurement system is joint-carving. And yet Adam may have viewed the Imperial measurement system that his brain internalized as an unfortunate burden to be shed, rather than as a joint-carving success. I admit that there is some kind of tension between using a concept and regarding that concept as defective. But that tension is not a matter of inconsistency in belief.

A variation of the third strategy fails for a similar reason. Suppose that a fidelic belief entails a true belief about the objective organization of the world and a gruesome belief entails a false belief about the objective organization of the world. Then the apparent value of fidelic beliefs can be explained in terms of the final value of these entailed beliefs. Yet these entailments do not hold in general. I believe that the jade bracelet is green, I believe that the jade bracelet was observed before AD 3000, and I believe that something is grue iff either it is green and observed before AD 3000 or blue and not observed before AD 3000. So I naturally also believe that the jade bracelet is grue. Yet I also explicitly reject that the world is objectively organized in a grue way. So either I hold contradictory beliefs or, more realistically, in general a belief does not entail a belief about objective organization.

The fourth strategy identifies a value discrepancy in true beliefs. Many philosophers (though certainly not all) maintain that some beliefs are more valuable than others. Often, this difference in value is explained by the comparative significance of the beliefs.Footnote ¹² The first strategy, then, claims that fidelity tracks significance. For example, the belief that all emeralds are green is more valuable than the belief that all emeralds are grue because the former is more significant than the latter. Likewise, beliefs about chemistry (e.g. that an oxygen atom has eight protons) are more valuable than beliefs about mixology (e.g. that a sidecar has cognac, orange liqueur, and lemon juice) because the concepts inherent to chemistry have greater fidelity than the concepts inherent to mixology and that greater fidelity indicates greater significance. And by thinking in terms of bred and rue, Sider's imaginary group of people missed an opportunity to attain a “deeper”, more significant, truth about their world.

This strategy supplies incorrect value judgements. Significance amplifies value: a true significant belief has greater epistemic value than its insignificant counterpart, while a false significant belief has greater epistemic disvalue than its counterpart. But fidelity does not amplify value. It adds value. A true belief that is also fidelic has greater epistemic value than a true belief that is non-fidelic. But a false belief that is fidelic has less disvalue than a belief that is both false and non-fidelic.

Think about specific comparisons. Probably most people have the intuition that the belief that the bracelet is green has more value than the belief that the bracelet is grue. Speaking personally, I also have the intuition that the belief that the bracelet is not green has less disvalue than the belief that the bracelet is not grue. But I might be unusual in that respect. So let me address two motivations a veritist may give for the opposite intuition and then offer a motivation for my intuition.Footnote ¹³

First, the veritist may say that it's worse to believe that the bracelet is not green because it is more blameworthy – epistemically blameworthy – to form that type of belief. In most circumstances, it's obvious that the bracelet is green, and so someone who believed that it was not green must have made a careless mistake. In contrast, being grue is a complicated property. It is more excusable for someone to make a mistake about which things are grue.Footnote ¹⁴

This motivation assumes an implausible connection between epistemic value and epistemic deontology. To be sure, the extent to which someone is criticizable for their belief has something to do with the disvalue of that belief. But the fact that one belief is less criticizable than another is insufficient to show that the former is less disvaluable than the latter. Recall the aliens with the grue-ified conceptual scheme. For them, green is comparatively more complicated, and thus it is more excusable for them to make a mistake about which things are green. But the value (or disvalue) of the belief is unchanged.Footnote ¹⁵

A second potential motivation is based on a more sophisticated understanding of truth. Many philosophers talk as if the amount of truth contained within a belief is readily discernible from the surface grammar of the sentence that expresses that belief. But that has to be a simplifying assumption. Arguably, for instance, the belief that Socrates is dead is a logically complex belief that contains the belief that Socrates was alive and the belief that Socrates is not alive. In this particular case, ‘green’ and ‘grue’ are single words. But perhaps the concept green is more informationally rich than the concept grue. If so, then the belief that the bracelet is green would contain more truth than the belief that the bracelet is grue. Similarly, the belief that the bracelet is not green would contain more falsehood than the belief that the bracelet is not grue. Consequently, the true joint-carving belief would be more valuable than the true gruesome belief, and the false joint-carving belief would be more disvaluable than the false gruesome beliefFootnote ¹⁶

Decisively discrediting this motivation would require a fully developed account for measuring truth; no one, myself included, has one of those. Still, I find this motivation suspicious. Why should we think that green is more informationally rich than grue? We often describe grue as a disjunctive concept. But being disjunctive doesn't necessarily entail being less informative. In fact, it seems to me that saying something is grue communicates the same amount of information communicated by saying something is green. The only difference is that in the case of grue the information is mixed together in a way that we find somewhat confusing. Think of it this way. There are four possible states for the bracelet to be in: (1) green and observed before AD 3000, (2) green and not observed before AD 3000, (3) blue and observed before AD 3000, and (4) blue and not observed before AD 3000.Footnote ¹⁷ All of the beliefs under consideration rule out exactly two of these possibilities. The belief that the bracelet is green rules out (3) and (4); the belief that the bracelet is not green rules out (1) and (2). The belief that the bracelet is grue rules out (2) and (3); the belief that the bracelet is not true rules out (1) and (4). Thus, on further examination it seems that the beliefs contain the same amount of information, and therefore would contain the same amount of truth or falsehood.

Finally, here's how I would motivate my intuition that the belief that the bracelet is not green has less disvalue than the belief that the bracelet is not grue. Generally speaking, we want to improve the value of our beliefs. But if improving the fidelity of our beliefs sometimes worsens their value, then we will often find ourselves in highly counter-intuitive situations. Suppose Zilin believes that her current scientific theory is false. (Perhaps she is motivated by the pessimistic meta-induction.) She knows that, for whatever reason, the true scientific theory is beyond her reach. My intuition is that in this situation Zilin should stick with her current theory. But someone who accepts the opposite intuition would have to say that Zilin would be better off grue-ifying her theory by replacing its concepts with less fidelic counterparts. For the same reason, teaching philosophy would become a risky business. By teaching my students that the wrongness of murder is not relative to the day of the week, I risk the possibility that some students will use the right moral concepts for the wrong moral beliefs. From the perspective of epistemic value, then, my lectures could cause serious harm! The strangeness of these situations suggests my intuition is the right one. It is better to have a false belief that is joint-carving, even if that belief is still overall disvaluable.

The fifth and final strategy I'll consider attempts to explain away the apparent epistemic value of fidelity. While fidelity is valuable, it would be a mistake to take that value to be epistemic. Perhaps thinking about the world with joint-carving concepts is practically valuable. Consider, for example, the progress made in theories of physics, where old concepts are replaced by new, seemingly more joint-carving, concepts. This advancement in our concepts has enabled us to develop technologies such as lasers, atomic bombs, and smartphones. Or perhaps thinking about the world with joint-carving concepts creates a sense of “getting it right” that is pleasurable or otherwise prudentially valuable. Finally, perhaps it is valuable (either practically or prudentially) for someone to be “on the same page” as other people when it comes to the concepts that they employ. Fidelity may then serve as a convenient means of coordinating our use of concepts.

I do think that fidelity has non-epistemic value in some of these ways. But I doubt that they can adequately explain away all of its apparent value. These non-epistemic explanations leave large swaths of philosophy inexplicable. Metaphysicians fiercely disagree about the fundamental nature of reality. Yet many of them go to great lengths to show that their disagreements lack practical consequences. Certainly, such metaphysicians often take pleasure at “getting it right”. But their pleasure is not merely a matter of fidelity. They also take pleasure in believing the truth! Thus, in response to the fourth strategy, I offer the following tu quoque: if the non-epistemic value of fidelity should lead us to doubt that fidelity has epistemic value, then the non-epistemic value of truth should also lead us to doubt that truth has epistemic value.

There might be ways for veritists to respond to my arguments against these five strategies. There also might be other strategies for explaining the apparent epistemic value of fidelity that I haven't considered. Still, my arguments should have shifted the burden of proof onto the veritists. In the meantime, we can consider the consequences of accepting that fidelity has basic final value.

4. The balance of epistemic value

By now, I hope to have motivated the claim that the value of “getting it right” is determined along two distinct dimensions: the value of truth and the value of fidelity. If so, then it's worth asking how fidelity is best pursued as well as how its pursuit should be balanced with the pursuit of truth. We want to “get it right”. But, if “getting it right” means pursuing both truth and fidelity, then what do we do when the two come into conflict? Should we prioritize truth over fidelity? If so, how much truth is worth an off-setting lack of fidelity?

These questions may look familiar. They are the same sorts of questions we ask when considering the pursuit of the truth. As William James famously declared, no one simply “pursues the truth”. Such a description obscures an important complexity. In reality, there are two separable pursuits: seeking true beliefs and avoiding false beliefs. How someone balances these pursuits will shape their overall strategy for belief formation. If they privilege seeking true beliefs, then they should just believe everything. Much of what they would come to believe would be false, but they would be guaranteed to believe everything that is true. Similarly, if they privilege avoiding false beliefs, then they should maintain total agnosticism. They would entirely lack true beliefs, but they would also succeed in their goal to never form a false belief.Footnote ¹⁸

To give this dual-natured pursuit a less misleading label, call it the pursuit of alethic value. Now, some philosophers think that we lack direct control over our beliefs. If they're right, then the “optimal” strategies suggested above are not so easily implemented. Nevertheless, even those philosophers would agree that there is a wide range of available strategies for pursuing alethic value. These strategies influence how sceptical or dogmatic someone is, how often they seek further evidence, and so on. Philosophers disagree as to which of these strategies are rational, and some even argue that there is only one rational strategy for the pursuit of alethic value.

In fact, there is an analogous complexity in the pursuit of fidelity. A joint-carving belief has positive epistemic value and a gruesome belief has negative epistemic value. The pursuit of fidelic value includes seeking joint-carving beliefs and avoiding gruesome beliefs. Their “optimal” strategies might also be beyond our ability to implement. Some philosophers think that, in one way or another, we lack full control over what concepts we employ.Footnote ¹⁹ If they're right, then in some situations we are unable to acquire joint-carving beliefs or avoid having gruesome beliefs. And yet there is still a wide range of available strategies that influence how readily someone adopts new concepts, how hesitant they are to abandon their old ones, and so on.

Philosophers may disagree as to which fidelic balancing strategies are rational. But because they have not explicitly acknowledged the separability of the two pursuits, they in fact have had little to say about the balance between them. As a consequence, many of the recent disputes in metaphysics over ideology – that is, disputes over metaphysical terminology – can be traced to unarticulated differences about the proper balance between seeking the joints and avoiding the gruesome. Here's one example. In a discussion of the comparative fidelity of logical ideology, Sider advocates for an “egalitarianism” according to which “both ∃ and ∀ carve at the joints” (Sider Reference Sider2011: 258). He does not seem overly bothered by the possibility that he is carving too aggressively. Others, however, seem more cautious. To avoid arbitrariness in his choice of ideology, Cowling (Reference Cowling2013) offers a kind-based principle of ideological parsimony; to avoid both arbitrariness and redundancy, McSweeney (Reference McSweeney2019) motivates a sort of agnosticism about the fidelity of our logical concepts. Their reasoning can be seen as motivated by an underlying aversion to bad butchery. It's a move they judge to be worth making, even if they thereby risk missing one of reality's joints. If there is no uniquely rational balance between seeking fidelic beliefs and avoiding non-fidelic beliefs, then there might be no fact of the matter about which of these strategies is best.

So “getting it right” is a multi-faceted affair. The pursuit of alethic value requires a balance between seeking true beliefs and avoiding false beliefs. The pursuit of fidelic value requires a balance between seeking fidelic beliefs and avoiding non-fidelic beliefs. And the pursuit of the value of “getting it right” requires a balance between the pursuit of alethic value and the pursuit of fidelic value. I don't mean to suggest that the analogies are as perfect as I've drawn them here. I am sure that there are subtle differences worth exploring. But the point stands: the Jamesian framework neatly characterizes the complexity inherent to “getting it right”.

And we are once again confronted with a new version of the same question: which strategies for balancing the pursuit of alethic value and the pursuit of fidelic value are rational? I don't have a detailed answer to that question. But I do have an answer to which strategies are not rational. Adapting some useful terminology from Arrhenius and Rabinowicz (Reference Arrhenius, Rabinowicz, Hirose and Olson2015), we can distinguish between five separate principles regarding the relationship between alethic value and fidelic value:

Informally, the first two principles take alethic value to be a “higher epistemic good” than fidelic value. According to them, even though more fidelity is always better than less, there's no amount of fidelity that outvalues a (certain level of) truth. So, for example, the addition of a single truth to someone's body of beliefs would be more valuable than the replacement of those beliefs with perfectly joint-carving ones, even if the replacement did not detract from the alethic value of the beliefs. The same is true for the last two principles, but in reverse: avoiding a million false beliefs wouldn't be worth the smallest decrease in fidelity. In contrast, according to the Archimedian principle the values have the same upper bound (which might be infinite) and consequently there will always be a trade-off worth making.

I suspect that most philosophers will find both Weak Fidelic Superiority and Strong Fidelic Superiority implausible. But I also suspect that many philosophers may be inclined to privilege the pursuit of alethic value over the pursuit of fidelic value by endorsing either Strong Alethic Superiority or Weak Alethic Superiority. I think that would be a mistake.

Consider Putnam's model-theoretic argument.Footnote ²⁰ Any theory, before we can say whether the theory is true or false, needs to be interpreted. But, on one way of construing Putnam's argument, almost any extensional theory can be interpreted in a way that makes the theory flawlessly true. So far as the truth of our theories is concerned, it doesn't matter what the world is really like. The same can be said for interpretations of what we believe. This result threatens the supposed value of “getting it right”.

One standard response to the argument is to claim that not all interpretations are created equal. Some interpretations are better than others in virtue of the fact that the former pick up on objective samenesses and differences. Here's how David Lewis puts it:

Lewis's use of the phrase “a worthy goal” is suggestive. Putnam's model-theoretic argument shows how we can secure maximal alethic value by deprioritizing our pursuit of fidelic value. If either Strong Alethic Superiority or Weak Alethic Superiority were true, then we should hail the argument as our epistemic salvation – how fortuitous it is that the highest good is acquired at so little cost! But we shouldn't. The worthy goal of “getting it right” depends on the non-infinitesimal value of fidelity. Strong Alethic Superiority and Weak Alethic Superiority are false and the treatment of alethic value as a higher epistemic good is irrational.