The Minimal Core: Truth, Assertion and Belief

Truth, as I said, is a very simple concept and a very simple property. There does not seem to be much more in it than the trivial equivalence: That P is true if and only if P, and the equivalence between asserting that P is true and asserting P. But isn’t there something more metaphysically and epistemologically loaded in the concept of truth? If truths are facts, doesn’t that mean that there is something in the world which has to be such that it corresponds to the truth when we say that something is a fact? And don’t we need to think something true if our thoughts are supposed to be facts? Since on this view ‘A true thought is a fact’ states an identity, it can be read from left to right, or from right to left: either by nudging our thoughts towards the world, or by nudging the world towards our thoughts. The first line suggests that we should be realists about truth, the second that we should be idealists. Another familiar idea is that true thoughts are successful: it is much more useful to have true beliefs than false ones. These observations have led to the idea that truth is a metaphysically loaded notion. But do we need to accept such metaphysical views to understand the concept of truth?

It seems that in order to understand the very concept of truth, we do not need to have a sophisticated substantive theory of how our thoughts relate to the world. We need only the simple equivalence It is true that P if and only if P. If you say something true, you have stated a fact, if you say that ‘“London is in England” is true’, you have said nothing other than that London is in England (Horwich, Reference Horwich1991). But we may agree that the concept of truth is associated with various marks, which are not definitions, but which can be taken as criteria of its use:

1. (i) Equivalence: The belief (assertion) that P is true if and only if P.

2. (ii) Objectivity: The belief that P is true if and only if things are as P says they are.

3. (iii) Warrant and coherence: Some beliefs can be true but not warranted nor coherent and some can be warranted or coherent without being true.

4. (iv) Norm: It is correct to believe that P if and only if P is true.

Taken individually, none of these marks define truth. Neither do they define truth, in any substantive sense, taken together. This is why many philosophers have held that this gives only a ‘nominal’ definition of truth.

The most important thing in these truisms is the relationship between belief, assertion and truth: assertions express beliefs, which are true. Let us call this the basic triangle (Williams Reference Williams2002). The best way to see this is to attend to a paradox known as Moore’s paradox (Moore Reference Moore and Baldwin1993). To say: ‘I believe that it rains, but it does not rain’ is at least odd, or contradictory.

These truisms help to see where normativity comes in. It does not come with the notion of truth nor with the notion of propositional content: propositions are true or false. It comes with the notion of belief, and the expression of belief in assertion, and with the notion of correction:

Can we derive from this correctness condition any norm of truth for belief? It seems so. ‘Correct’ expresses a normative property. But which one? The most plausible candidate is that it is an ‘ought’, i.e., a deontic norm:

rather than ‘good’ or ‘valuable’, an axiological norm:

since ‘it is correct’ does not mean ‘it is good’ (something can be correct without being good, and vice versa). (NT) registers a requirement and implies an obligation.

The problem, when we try to say in what sense truth is a norm, is familiar from the case of moral norms: norms are useless if they do not move us to do or think in certain ways. A norm must motivate. It must require and guide, and in this case it must tell us what to believe and how. This is what may be called the problem of normative force. But not all normative notions are able to move or guide in the same way. Deontic norms (such as ought, forbidden) are strongly prescriptive, and they have a potentially universal scope, they apply everywhere. Values (good, bad, admirable, loathsome) are less prescriptive and can be relative (good for X, good in context Y). Rules and conventions (such as rules of games, codes of etiquette) are prescriptive in domain-specific ways (social, cultural, epistemic) and are optional (they can be removed and replaced by others).

(NT) says that we ought to believe all truths and avoid all falsehoods. But this seems to fly in the face of common sense: should we believe all truths? Including those that are trivial or uninteresting? And should we really avoid believing all falsehoods? There are cases where it can be good, or obligatory, to believe false things (for instance and in some cases it may be good to believe that one can recover from an illness, or obligatory to believe what the boss says, because she or he is the boss). There are also some propositions that it is impossible to believe, such as that I do not exist. So it seems that the best we can say is that (TN) holds typically or normally.

It also seems too demanding to ask that we believe what is true, for we do not always hit the truth. We have to believe what we have evidence for. So the norm for belief might be weaker than (TN), such as:

But evidence is not enough for truth. And it may be that one ought to believe what is true, while having no evidence.

Moreover the norm of evidence (NE) can clash with the norm of truth. If you have misleading evidence for a proposition (say, that this drug will cure you) you may be in the position of following (NE) but this is not what you ought to believe: if the drug is dangerous, you ought not to believe that it will cure you. So (NE) can conflict with (NT).

Moreover if (NT) were the right norm, it would be impossible to suspend judgement. One could adopt instead a weaker norm of truth:

But now (NT*) is too weak: a mere permission to believe does not have the normative pull that a norm must have. These objections miss the target. (NT) is not a prescriptive norm, in the sense of an imperative giving instructions about how and when to believe a truth. There is no such norm.

The alternative is to consider (NT) not as a prescription, but as a criterion of rightness, which tells us what it is to believe. This is actually what the notion of correctness involved in (C) says: it says that a belief is correct if and only if true, which spells out what belief is, and what commitment one incurs when one has a belief. But does it solve the problem of motivation? No, for it cannot be understood as giving us any guidance. If it did, it would satisfy two conditions on norms: first, that one be aware of the norm, and conscious of what it prescribes; second, that one feels compelled and appropriately moved by the prescription. But (C) does not satisfy these conditions. To see this, suppose one reasons thus:

1. (1) If P is true, then I ought to believe that P; if P is false, then I ought not to believe that P [= C].

2. (2) P is true.

3. (3) So, I will believe that P.

But the conclusion is absurd. If one accepts (2) one already believes that P. What would be the use of knowing that it is correct to believe that P only if it is true?

The Relativist and the Pragmatist Threats

Perhaps the source of our difficulties lies in the fact that philosophers tend to concentrate on the objective of finding one aim for belief, and one norm to which all beliefs should obey. This seems to suggest that there is only one route towards truth. But there are many kinds of beliefs, and many kinds of truth. And these can be assessed with respect to many criteria, just like restaurants, which can be assessed for the quality of food, their variety or exotic character, the atmosphere, the price, etc.

Beliefs similarly can be assessed on the basis of their utility for various purposes: whether they are commonsensical and in some sense needing no justification, whether they are well supported by evidence, whether they are useful for life or for health, etc. The criteria of assessment differ also depending on the kind of beliefs and the kind of domains on which they bear: empirical, conceptual, moral, political, aesthetic, etc. Why should we expect that they will be assessed through a single norm? Maybe there are as many norms as there are domains of inquiry. This suggests that if beliefs are not assessed for truth but for utility, then truth is not the only norm or value along which beliefs are to be evaluated. Or perhaps there is no norm at all. After all there are many kinds of beliefs, including strange ones, irrational ones, or delusive ones, of which it can be asked whether they are actually beliefs and whether they follow any norm or rule.

Let us just consider the hypothesis that there are no norms for belief: belief is too messy an affair to be subject to a norm. Maybe beliefs are like moods: they come and go, and we do not know why. This may be supported by a number of irrational and incoherent beliefs, which are common. After all, don’t people believe absurd things: conspiracy theories and such. Maybe these are not beliefs, but sort of beliefs, which have some properties of beliefs (e.g., they represent certain things to us) but not the central one: they are not regulated by the norm of truth, or are so regulated sometimes, but not always. Let’s call these odd ‘beliefs’ schmeliefs.

Let us imagine a tribe of schmelievers who follow the norm of truth on Mondays, then the norm of believing what pleases them on Tuesdays, then the norm of believing what is cute on Wednesday, etc. These people would be incoherent. But, one could say, what’s the problem with incoherent and conflicting norms? After all there are ethical disagreements all the time. But can there be epistemic disagreement on norms of thought? The problem would be: Would these people believe anything? Would they have beliefs at all? The answer is no. They would have states which resemble beliefs, but which are not beliefs. So this nihilistic stance does not make real sense.

The pluralist and pragmatist line – there are many norms for belief, depending on our practical aims and contexts – is more attractive. To take up the same example: in some cases I have to believe that I have no illness, in others to believe what my doctor tells me. Sometimes I have to be serious, at other moments I am allowed to joke. But does it make sense? For there to be a pluralism of dimensions of assessment of beliefs, it should be the case that one kind of norm not only coexists with the norm of truth, but also can compete and override this norm. This would imply that the main criterion of assessment for beliefs is not epistemic, but sometimes practical. But this is absurd: truth does not just happen to be one criterion among many, it is what beliefs are for, the central aim of belief, just as lungs are for breathing, government is for governing.

Take the case of error: if one discovers that one has made a mistake, one has to change one’s belief. And the fact that a belief is false is a fatal flaw for a belief. The primary dimension of assessment of beliefs is truth, and derivatively, evidence. One can agree that there are other dimensions of assessment, but can they be in competition with the epistemic dimension? It is doubtful. Let us take an example other than the one before: it is often said that sometimes, in public contexts, one has to balance one’s beliefs with others’ opinions, even if they are weird, just for the sake of tolerance and (as it is said) democratic debate. The marketplace of ideas should make us honest and tolerant of other and opposing opinions, and we are never sure that we are right.

This is a case where it is more useful, or prudent, to avoid asserting (one’s beliefs about) what is true, and better to accept (or do as if one accepted) the beliefs of others. In such cases, it seems that I can compare my evidence that I have of a belief with the utility of believing the contrary or of not believing it. Can we conclude here that the right kind of reasons for believing are pragmatic, and not epistemic? No, for even if my belief happens to be useful in the circumstance, it needs to be true. I can in some cases usefully acquire false beliefs, but that cannot be the rule. Such cases show that practical reasons for belief can coexist with epistemic reasons, but not that the former can override the latter. The epistemic norm is always the default stance. In other words, there are practical reasons for belief, but they are never the right kind of reason for believing.

The Norm of Truth is a Norm of Knowledge

We have seen that if we take belief to be governed and regulated by a norm such as (TN), we encounter a dilemma: on the one hand the norm of truth, if it is strongly prescriptive, is too strong, and if it is reduced to a correctness condition on belief it does not motivate or guide, since for that we would have to have access to truth in most cases; and on the other hand, if we side for the weaker norm of evidence we risk being subjective. Moreover, one must in some sense be responsive to the norm, but for that it seems that we should have access to it. This suggests that the norm of belief must be both objective and sensitive to evidence.

The obvious proposal is that the norm of belief is not truth, and not evidence, but something that is both true and well supported by evidence: knowledge. When we believe something, we aim at knowing it. Certainly we can say: ‘You believe P, but you don’t know it’. But this seems to imply that if we believe it, we ought to know it, or at least try to know it. Our beliefs are not just beliefs about what we take to be true, they are also attempts at knowing. Many of these attempts will be unsuccessful, for many of our opinions are false, but we can accept that knowledge is the aim of belief, and actually its norm (Williamson Reference Williamson2000; Engel Reference Engel2002). In other words, the norm is this:

But certainly this seems odd, since if we know something, why should we believe it, or only believe it, exchange the gold of knowledge for the bronze of belief? So the norm should be different:

This still seems to be a very strong norm. How can one obey it? In many cases, we aren’t even in a position to know, either in practice or in principle. But how to guarantee such a strong condition?

Obviously something has to be added. The knowledge in question has to bear certain guarantees or warrants: to be secure, safe and reliable. I shall not try here to develop the nature of these guarantees, but the most plausible condition for such guaranteed knowledge is that the knower has certain kinds of dispositions, and must be able to receive some credit for acquiring it. In traditional terms this means that knowledge must be based on certain good dispositions, or virtues exemplified by the knowers, and not simply true and warranted, or rather: it has to be produced by knowers who are reliable (Sosa Reference Sosa2007).

Norm and Value

Taking this line entails returning to the suggestion that was set aside previously, that the relation between belief and truth has to be a norm, but also a value. Certainly the very same difficulties that we met when trying to understand how a norm of truth can motivate or guide our believing can be formulated if we say that truth is a value for belief, perhaps the supreme value: if the value is objective, how can we access to it, and how can it guide us? Values are, in a sense, as difficult to reach as norms. But the solution just proposed, that the norm of belief is not truth but knowledge, provides a way out. For knowledge, if it is produced by dispositions and virtues, is less transcendent than a norm. It can be secured by certain kinds of qualities, which are themselves knowable.

We can know what it means to be sensitive to the value of knowledge, for instance to be curious, courageous, honest and humble. These are the traditional intellectual virtues of the Aristotelian tradition, and one can learn to cultivate them. Even institutions such as universities and scientific research can exemplify it (sometimes). One may object that knowledge remains, just like truth, transcendent and unknowable: how do we know that we know? Science has renounced the aim of certainty. But that does not prevent knowledge from having the status of an ideal. Just as truth is the ideal of belief, knowledge is the ideal of belief.