Introduction

It is widely assumed that the pragmatist, coherence, identity and deflationary theories of truth are essentially incompatible and rival views to the correspondence theory (CT), although pluralists disagree (Crispin ^{Wright 1992}; ^{Lynch 2009}; ^{Cory Wright & Pedersen 2010}). In this paper, I will challenge this popular assumption without accepting pluralism. With the exception of some versions of the identity theory (see section 2.2), I find that the alternative theories only contradict CT, either when combined with a rejection of the distinction between thought and reality, or when it is assumed that to offer a ‘theory of truth’ is to offer a theory of the function of the truth-predicate. I propose that CT should not be understood as a theory about the function of the truth-predicate, and that the core ideas of the alternative views, once separated from any anti-realist convictions, can be understood as complementary views about different aspects of a fairly complex phenomenon, notably of how our beliefs relate to their subject matter and how we reason and talk about that relation.

The key is to appreciate that different thinkers approach the issue of truth on the basis of very different ideas about what philosophy in general is all about, often assuming that everyone else does the same. Some people think philosophy is essentially a theory of meaning and pursue the issue of truth as if it were only a question about the meaning of the word ‘truth’. Others think that philosophy only elucidates the content of concepts, and therefore assume that the issue of truth is only a question of how we think about truth, independently of any metaphysical theory about what the world is really like. Yet others are wholly concerned with epistemic issues and approach the issue as only a question of how we warrant claims about truth. Finally, some assume that philosophy is wholly concerned with the question of what the world is really like as opposed to what we might think it is like and how we talk about it. Failure to recognise these conflicting interests and approaches will create confusion and misunderstanding. The surest way to clear up the confusion is to try to understand the issue of truth in a way that makes best sense of all these conflicting interests. That is what I will attempt to do here.

It is important to note that I focus on the core ideas of various truththeories rather than on the details of the views of particular thinkers. This is partly a practical necessity for the big picture approach attempted here, but also a consequence of the fact that there is no canonical version of any particular truth-theory; the various ‘theories’ are at best families of fairly heterogeneous views that share little more than a commitment to some core idea of the kind I present below. The aim is to show that it is possible, even plausible, to think of these core ideas as complementary ideas about different aspects of a very complicated issue, rather than as contrary ideas about one and the same thing. I don’t think my appraisal of the core ideas of each theory is terribly controversial, but since I don’t address the specific details of the views of any particular thinker, one cannot without further ado apply the reasoning and/or conclusions presented here to them. However, I hope that the big picture presented here will provide a novel perspective that can be used in the critical scrutiny of particular versions of each view. In turn, such a scrutiny of particular views may well reveal the weaknesses of the big picture.

Metaphysical Theories of Truth

In addition to CT-which arguably dominates metaphysical thought on truth and is the natural target for alternative theories-I discern two metaphysical views that deserve mention even though they are almost extinct. One is a version of the identity theory and the other a version of the coherence theory. They contradict CT because they explicitly deny that beliefs ever correspond to fact and, therefore, entail that if the truth predicate picks out anything at all, it is not the property of corresponding to fact. For the purposes of identifying genuine rivals to CT, nothing more need be said, but I want to draw out an interesting contrast to more familiar versions that do not deny that beliefs correspond to fact, but instead focus on epistemic issues, or strive to be metaphysically neutral. But let us first turn our attention to CT.

CT is not a single unified body of beliefs about which its proponents agree. It is really an “extended family of theories and, more often, theory sketches” (David 2016: §1). Indeed, the relevant entries of the Stanford Encyclopedia give similar verdicts for the coherence theory (Young, 2008), pragmatist theory (^{Legg and Hookway, 2019}), identity theory (^{Gaskin, 2015}), and deflationary theory (Stoljar and Damnjanovic, 2012). The various sketches share little more than a commitment to some core idea about how to best make sense of the philosophical issue of truth.

The core idea of CT is that sometimes the content of our thoughts accurately represents the features of some subject matter that is distinct from those thoughts themselves. Typically, but not necessarily, the subject matter is a part or feature of objective reality, i.e., a ‘fact’ in the sense of ‘some existing non-linguistic/non-conceptual feature of reality’ (not, ‘true proposition’). For the sake of convenience, I will talk about the content of thoughts as ‘beliefs’ and their subject matter as ‘facts’, even though of course a belief can have another belief as subject matter, and despite the fact that the content of belief can be true or false regardless of what we believe about their truth/falsity.

It is a matter of some controversy what exactly beliefs and facts are, but as far as I can tell that is irrelevant when comparing extant theories of truth. No rival view is built around a significantly different conception of beliefs and/or facts, or about the manner in which beliefs represent facts, except the particular version of the identity theory discussed below (se section 2.2 below).

Other rival views-except those that deny that there is an objective reality to which beliefs can correspond-do not deny that a relation of correspondence can obtain between a belief and fact; they just think that truth is something else than correspondence-say, the coherence or utility of beliefs-or they argue that the function of the word ‘truth’ cannot be linked to any of the above conceptions of truth, or to all of them equally. I find this sub-class of truth-theories to be compatible with CT, contrary to what their proponents believe.

The important thing to note is that truth, according to CT, is not the meaning of a word or content of a concept, but a phenomenon in the world. It is a property that beliefs acquire, in an ontological sense, in virtue of holding a certain kind of relation to whatever they are about. The idea that truth is a relation of this kind between belief and fact, is also the basic idea behind the so-called truthmaker principle; that truth is a relation between truthmaker and truthbearer, such that truthmakers determine the truth-value of truthbearers. Consequently, I include truthmaker theories in the family of CTs and will sometimes use ‘truthbearer’ and ‘truthmaker’ as synonymous with ‘belief’ and ‘fact’.

How the world determines truth-values is just as unclear as what it is for beliefs to accurately represent their subject matter. For the lack of anything better, it is sometimes said that beliefs are true in virtue of the world, in ways yet to be elucidated (^{Armstrong 2004}: 5). However, the details are largely irrelevant and a full consideration of them all would only serve to obscure the discussion here. I ask readers who find my account frustratingly oversimplified to please bear in mind the difficulties of characterizing a family of theory sketches, and to consider in their criticism whether the absent details really make a difference for the comparison of different theories. When comparing CT to other extant views, all that matters is the core idea that truth is the obtaining of correspondence between the content of belief and whatever that belief is about.

CT is a metaphysical theory, because it concerns what there is and relations between existent entities. It says there is a distinction between belief and fact, and argues for the existence of a relation between them; a relation that is able to obtain independently of whether or not we know it to hold (see, ^{Johansson 2004}; ^{Ingthorsson 2006}). Indeed, it is precisely because the correspondence of belief to fact is ‘unobservable’-we cannot judge whether a belief is true/false merely on the basis of the content of the belief itself-that we have an epistemological problem. Basically, the nature of the correspondence relation is the source and not the solution to our epistemic worries. Indeed, what else could constitute an epistemic problem other than the difficulty of finding out whether a belief corresponds to its subject matter?

Let us now turn our attention to identity theories. The core idea that unifies identity theories of truth (IT) is that “the truth of a judgement consists in the identity of the judgement’s content with a fact” (^{Baldwin, 1991}, 35). Versions of IT count as metaphysical theories in so far as they are motivated by concerns about what there is and relations between what there is, rather than about the way we think or reason. Formulated as such, there is a clear contrast between CT and IT:

The correspondence theory says that there is a ‘gap’ between truth-bearers (thoughts) and something external to them which explains their truth/falsehood. The identity theory says there is no such gap (Sher 2013: 1)

There are at least two versions of how to close the gap. One is the view upheld for a while by Moore and Russell that the very objects that constitute the facts about which we have beliefs are themselves constituents of the beliefs, wherefore there really is no distance between belief and fact (^{Candlish, 1989}). For a belief to be true simply is for a fact to obtain; or, for a belief to be a fact. Moore and Russell later abandoned this view for reasons that are irrelevant here. It is enough that we understand that this version of IT is a genuine rival to CT; beliefs are true when they are identical to their subject matter, not by corresponding to it. However, it is worth noting that Moore and Russell initially intended their theory to be an account of the relation between belief and fact, not a rejection of it. Basically, the idea was that sometimes the distance between belief and reality is nil.

Idealists like F. H. Bradley embraced a different type of identity theory (^{Bradley 1907}; ^{Baldwin 1991}). Whereas Moore and Russell admitted an external material reality whose constituents sometimes become constituents of our beliefs, then Bradley rejected the reality of an external material reality, and therefore denied that the objects that constituted any belief were external to those beliefs. Basically, on Bradley’s view, thoughts are not true in virtue of something else but simply are true. This version of the identity theory is also genuinely contrary to CT.

Note that idealism does not entail IT. ^{Joachim (1906)}, argued that propositions are true when parts of a coherent system of beliefs, while McTaggart endorsed CT (1921: sect. 10). McTaggart accepted the difference between what ideal reality is actually like and what we think it is like, wherefore only some of our ideas correspond to reality. Idealists can admit that there is a gap between thought and reality (as long as both belong to the domain of ideas), and therefore error along with truth.

We can now put these versions of IT aside, having understood how they contradict CT. However, the version associated with John ^{McDowell (1994)}, Jennifer ^{Hornsby (1997)} and Julian ^{Dodd’s (2000)} is not easily identified with either of the views presented above. Partly because they claim it to be compatible with realism and yet postulate an identity of truth-bearer and truth-maker, and partly because they refuse to take an explicit stance on the nature of the entities that supposedly embody the identity. They want to remain metaphysically neutral. Like Candlish & Damnjanovic (2011), Mariá ^{Frápolli (2012: section 2.4)}, and Richard ^{Gaskin (2015)}, I struggle to understand the view, because I struggle to discern the nature of the entities involved. Like the commentators, I am tempted to think that the identity implied in this version of IT is not between belief and fact-as Russell, Moore, and myself understand these notions-but rather between the content of belief and the content of the proposition believed. That is, by ‘fact’ they mean a true proposition and not the constituents of reality represented by the proposition. On this understanding, beliefs are true when identical to true propositions. But then the question remains-at least for a realist like myself-of what makes those propositions true. It is unclear to me whether this version of IT really has any answer, and consequently whether it contradicts CT.

Finally, a few words about the metaphysical coherence theory, according to which truth “in its essential nature is that systematic coherence which is the character of a significant whole” (1906: 76). Importantly, Joachim was an idealist, and so believed that there is no material reality to which either a belief, or significant system of beliefs, could correspond. He appears not to have considered, as McTaggart did, that some of the ideas in the coherent whole could be about other ideas in the whole, and have to them a relation of correspondence in addition to both being parts of a coherent whole. Anyhow, nothing more needs be said to establish that this version of the coherence theory (COH) contradicts CT, since it entails that no belief corresponds to fact. Furthermore, it makes no claim to conform to popular use of the truth-predicate, or make sense of epistemic practices; it is meant to be a consequence of a metaphysical view about the constitution of reality. The contrast to epistemic versions of COH will be useful to have in mind later on (sect. 4.3).

I have now admitted that there exist genuine rivals to CT even though hardly anybody entertains them anymore. Let us now consider a challenge that is based on considerations about the actual uses of the word ‘truth’.

Truth vs. ‘Truth’: Correspondence vs. Use of Words

The philosophy of truth is ripe with discussions about the use of the words ‘truth’ and ‘true’ in natural and formal languages. However, I struggle to understand the implications of any finding about the actual use of the word ‘truth’ for the question of whether beliefs sometimes correspond to fact, and for the epistemic role of such a relation. Arguably, we can find out everything there is to know about the various uses of ‘truth’, and all the intuitions there are to find about its meaning, and still be completely in the dark about the existence and nature of a relation between belief and fact. If truth is a relation between mind and world, then to inquire about its nature by examining actual uses of the word ‘truth/true’ and all our intuitions about truth, is about as effective as to study the nature of the sun by finding out about the uses of the word ‘sun’ and all the intuitions there are about its nature. Basically, given that the relation between belief and fact is not a word, or the content of a concept, we cannot use linguistic or conceptual analysis to reveal its nature. What we can hope to discover is how the word ‘truth’ is actually used, and perhaps uncover that it is not consistently used to denote correspondence of belief to fact.

Galileo Galilei uses a similar argument against those who argued-in response to his observations of sunspots-that the sun cannot have spots or impurities, because the very concept of ‘sun’ was that of a ‘most pure and lucid body’. The idea seems to be that the use of the word, or the meaning attributed to it, determined the nature of the object so named. Galilei points out that the current content of the concept, or the manner in which we use the name, has little bearing on the nature of the phenomenon:

It proves nothing to say… that it is unbelievable for dark spots to exist in the sun because the sun is a most lucid body. So long as men were in fact obliged to call the sun “most pure and most lucid,” no shadows or impurities whatever had been perceived in it; but now it shows itself to us as partly impure and spotty, why should we not call it “spotted and not pure”? For names and attributes must be accommodated to the essence of things, and not the essence to the names, since things come first and names afterwards (cited in ^{Shapin 1996}, 18).

By analogy, it proves nothing about the relation between belief and fact to argue that the current use of the word ‘truth’, as revealed by common usage, is not that of correspondence of belief to fact.

An anonymous referee has objected that the analogy above is flawed. S/he made the case that we can ostensibly identify natural phenomena independently of our theories, but not those phenomena that figure in philosophical notions, like justice, truth, and knowledge. In those cases, we can’t separate the phenomenon from the contents of our concepts as neatly as we can for natural phenomena. I disagree. Most metaphysically interesting phenomena can be picked out ostensibly, independently of our theories-in exactly the same way we pick out the sun-before we start to discuss what they are really like. Consider time, space, matter, property, relation, causation, and, I suggest, the correspondence of belief to fact. Obviously, we cannot think or see the objective nature of the phenomena independently of theories, any more than we can think or see the objective nature of the sun without guidance of theory. Our experience only provides, arguably but not uncontroversially, a common reference for what it is whose nature we disagree about.

Space is not something we recognise just because of some theory we have, but because it appears as an integral part of our experience of the world, in exactly the same way the sun is a part of our experience of the world. To be precise, it is the appearance of sun and space that are part of our experience, and the starting point for our philosophical worries, such as whether there is anything in reality beyond the appearance that corresponds to the sun and space that we experience. The difficulties escalate from then on even among those who agree that space is objectively real; is it absolute or relational, or perhaps substantival, etcetera.

Correspondence of belief to fact is slightly different, I admit, because we don’t intuit putative instances of it in a simple act of perception in the way we can intuit portions of space or instances of causation. But it is still a familiar phenomenon in every person’s life-world. Remember imagining what is inside a present before you unwrap it, and finding out that you imagined accurately what you previously did not perceive? You then infer that your previous belief already corresponded to fact before you perceived it (unreflectively assuming that perception and fact coincide, as naïve realists reflectively do). This is the most natural manner in which we come to believe without any philosophical training that sometimes beliefs correspond to reality. It is then philosophy that makes us mistrust whether there really are any unperceived items and thus whether there really can obtain any relation between belief and the object as it is in itself. This is where our epistemic worries begin and end.

The main point is that we can very well first identify in our life-world the philosophically interesting phenomenon before we start to deliberate how to best account for that feature, or what word to use to refer to it. The only clear cases of where we can’t separate the ‘thing’ from the content of our concept is when our concepts are constitutive of the ‘thing’, such as is arguably the case with various types of social constructions such as democracy. Democracy arguably is to a great extent what we think it is. But correspondence of belief to fact, if it occurs at all, can arguably occur independently of any concept about it. If there is a being able to represent anything, then if one of its representations happens to accurately represent something, then the representation corresponds to the thing regardless of any concept of correspondence that this being may have. The being need not even know that there is a difference between appearance and reality, or between representation and the represented. Indeed, we rarely know when our beliefs actually are true.

CT, then, is the thesis that the best way to make sense of the world is to accept as correct the distinction between our conception of the world and the world itself, and to accept that sometimes our conceptions about some part of the world accurately represent that part. In this form, it is a theory about the constitution of reality, not about the use of words. Obviously, its proponents can argue persuasively that the tradition of using the word ‘truth’ for the correspondence of belief to fact is an old one, perhaps even the original one from which other uses have arguably developed. But the current uses of the truth-predicate are not implied by the content of the theory itself. It is an independent historical accident.

In light of the above, proponents of CT should not worry whether their theory explains the various meanings of ‘truth’ as revealed by actual linguistic practices. The most important thing for a proponent of CT is not, or should not be, the battle for how the word ‘truth’ should properly be used. Nor is it important that a theory about the relation between belief and fact (between mind and world) must conform to popular uses of ‘truth’. What should be the main concern of ontology is the battle for the acknowledgement of the reality of a relation between belief, and fact, and the investigation of the nature of that relation as well as its role in our epistemic practices.

The best argument in favour of this brusque rejection of purely linguistic approaches to the issue of truth, due to Richard ^{Kirkham (1992: ch. 10)}, is that it is possible to admit everything such linguistic approaches state about the uses of ‘truth’ without admitting that anything has been said about the nature or reality of a relation to reality or its role in our epistemic practices.

It may be true that the word ‘truth’ is used in a multitude of ways in various contexts, some of which cannot be interpreted as involving the ascription to anything of the property of corresponding to reality. For instance, when my younger son says that some particular brand of cereal is good, his older brother may agree by saying ‘that’s true’. He isn’t asserting that his brother’s belief corresponds to some objective goodness possessed by the cereal, because he believes taste is subjective. He is just agreeing, using a common figure of speech. In that particular context, ‘that’s true’ arguably means something closer to ‘I think so too’ rather than ‘your belief corresponds to fact’. Similarly, someone promising in court to tell the truth, the whole truth, and nothing but the truth, can hardly be expected to live up to the promise if we are assuming that truth is correspondence to fact. That would require him/her to be omniscient. The court is only asking the person to honestly describe things as they appeared to them; not how they really were.

We can admit that ‘truth’ is not always, or even ever, used to ascribe the property of corresponding to reality in ordinary language usage, without consequences for the validity of CT. We have only admitted that the word ‘truth’ is used heterogeneously in natural languages, not that belief never corresponds to fact. ‘Truth’ is not unique. Just consider the uses of ‘dry’, for (i) absence of moisture (‘dry’ weather), (ii) absence of sugar (‘dry’ wine), and (iii) absence of emotion (‘dry’ humour). Who would argue that because ‘dry’ is used heterogeneously, there either is no such thing as dry, or that ‘dry’ really means ‘absence of any arbitrary substance’? If we accept the last suggestion, ‘dry’ is never used metaphorically but always literally.

The important thing is that a theory that professes to be a rival and competing theory to CT, must either say that beliefs never relate to facts, or that the relation to facts is somehow significantly different from how it is described by CT. The latter will admittedly be very difficult, because the CT family already includes so many different views about the nature of the relation between belief and fact. In fact, every theory admitting any kind of relation between belief and fact has by default been adopted into CT (David 2016).

In conclusion, CT is not a theory about a word, or concept, or notion. The theory itself can be considered to be a concept about something, and its proponents favour the use of ‘truth’ as a linguistic gimmick to refer to tits subject matter. Truth, according to CT, is something in the world, and we may have many ideas (concepts) about the nature of this something, of which only one can correspond to the real thing.

Correspondence Truth vs. Analytic Truth

An alleged worry for CT is that certain types of propositions are assumed to be obviously true but appear to lack truthmakers. Such propositions are taken to be true merely in virtue of the meaning they have and called ‘analytic truths’. Mathematical truths may serve as a case in point, even though it is not universally agreed that mathematical truths are analytically true as opposed to true by correspondence (for a discussion, see ^{Armstrong 2004}: ch. 8 & 9). The argument goes as follows. It is obviously true that ‘2+2 equals 4’, but it is unclear whether there is, or need be, anything in the world to which the belief corresponds for it to be true. Instead, it is suggested, it is true in virtue of the content of the belief, and its coherence with the rules of arithmetic. If, then, 2+2 equals 4 is not true in virtue of facts then truth in maths must be something else than correspondence, and correspondence cannot be the most basic nature of truth.¹

Fortunately, this worry doesn’t challenge the view that sometimes our beliefs correspond to fact. The argument can at best establish that a subclass of beliefs, notably mathematical beliefs, do not relate to facts. It is only a threat to CT, if it is assumed that a theory of truth must explain all the uses of ‘truth’ that humans have ever seen fit to invent. Indeed, we see here a connection to the underlying assumption of linguistic approaches to the issue of truth discussed in section 3, notably that theories of truth concern the use of the truth-predicate.

To my mind, if it turns out that mathematical ‘truth’ is not correspondence to fact, then according to CT, the term ‘truth’ is used in mathematics to denote a completely different property than the correspondence of belief to fact. That is, mathematics would fall outside the realm of correspondence. It will just have turned out that ‘truth’ has been used inconsistently in philosophical/mathematical/colloquial jargon. We should not be surprised. It is not important which word we use to refer to the correspondence of belief to fact. It would be practical to make a distinction between ‘analytic truth’ and ‘contingent truth’, e.g. to make sure we don’t confuse our epistemic practices in finding out about contingent facts and analytic truth, respectively. I am not yet giving up on the idea that truth in maths may in the end turn out to be correspondence to fact, I am just saying that if it turns out that it is not about correspondence this will merely restrict the scope of correspondence and not wipe it out altogether.

One might have hoped that when technical terms for the relation to reality were adopted -such as, ‘corresponding to’, or ‘being in accordance with’, or ‘in agreement with’- the difficulties of what the word ‘truth’ really denotes would have been separated from discussions about whether or not propositions hold a relation to reality and what the nature of that relation is. But it has not. There are still those who argue that since ‘truth’ is used in so many different ways, or is redundant, or somehow vacuous or useless, then truth is not correspondence and even that there is no such thing as being true. That conclusion isn’t entailed by the argument.

To repeat, only those theories can be considered rivals to CT that (i) provide a substantial challenge to the idea that sometimes beliefs correspond to fact, or (ii) provide an account of a substantially different kind of relation between belief and fact, and (iii) convince us that this is not the relation which our epistemic practices revolve around. I don’t see that appeal to the uses of ‘truth’, or concerns about analytic truth provide either. Perhaps a scrutiny of our epistemic practices may provide challenges of that kind.

Epistemic Theories of Truth

Admittedly, the idea that truth is correspondence to fact does not sit easily with the ways we reason and talk about truth, in the sense that we don’t justify ascription of truth by appeal to the correspondence of belief to fact, but to some epistemic warrant. Consequently, attempts have been made to tie the notion of truth to the various ways we warrant claims about something being true, or ways we come to accept something as true, rather than to the correspondence of belief to fact. I’ll begin by discussing this idea generally, before addressing specific ideas about valid warrants, such as coherence to a system of beliefs, and/or successful practices.

Remember that I have excluded from the discussion those views that are wedded to anti-realism. I only focus on the view that truth is based in some valid form of justification for believing in some or other proposition, without denying that beliefs correspond to fact, i.e. views of the following kind:

Epistemic conceptions of truth are those tying the truth of a proposition to some function of its warrant for belief. They are grounded in the intuition that when one ascribes truth to a proposition, one is primarily signaling that the proposition ought to be believed. […] According to epistemic theories of truth, therefore, it is the epistemic/normative role of truth ascription, rather than, say, correspondence to an ontologically given fact, that supplies the substance of our concept of truth (^{Cox 2001}: 473).

Note that the intuition on which epistemic theories (ET) is grounded, says that the truth of a proposition is the result of something we do. Accordingly, a belief ought to be considered true only when it is ‘warranted’ in some way or other, and its correspondence to fact is neither here nor there. Consequently, ETs do not necessarily deny that beliefs correspond to fact, they just mean to say that it is the warranty and not correspondence that commits us to the truth of a belief. I fail to see that anything in this view contradicts CT; it merely states that we should use the term ‘truth’ to denote warranted beliefs, rather than beliefs that actually correspond to their subject matter. If this is true, the quarrel that ET have with CT boils down to proper use of words.

The quarrel about the use of ‘truth’ can be settled by simply agreeing to use the word for warranted belief, and not correspondence. But there are consequences. If we decide that ‘true’ actually means ‘justified’ but without implying correspondence, we need to change the definition of knowledge accordingly; we would have to call it the ‘true corresponding belief’ conception of knowledge. The definition of knowledge as justified true belief assumed that ‘true’ means corresponds to fact. Would it be enough to say that knowledge is ‘justified belief’? This would be an acceptable option for those who deny the reality of an objective reality to which beliefs can correspond. However the suggestion should hold no appeal for those who believe in an objective reality, think knowledge is ‘of the world’, and accept that justification is fallible. To claim that truth is warrantability but doesn’t involve correspondence, only raises the question of what the relationship can be between knowledge and reality, and of course about the purpose of epistemic warrant. What is it we warrant, when we warrant belief?

The idea that knowledge requires both justification and truth (in the sense of corresponds to fact) is motivated by the idea that justification seems separate from truth, but also the idea that it is truth we are after in the end. Justification is not a goal in itself, but a means to achieve truth. Equate truth with justification, and we have turned the means into a goal; are the means after all a goal unto itself? The suggestion seems incoherent to me. Suppose we accept that the truth of a belief is constituted by the warrant for it, then it appears that what we believe when we believe that ‘P’ is true, is that ‘P’ is warranted. We have now turned ET into a platitude, and one that doesn’t address at all the question of what the relation is between our beliefs and their subject matter or how the knowledge that ‘P’ somehow constitutes knowledge of the world.

No one is denying that we are supposed to believe or not to believe on the basis of various warrants, but I simply cannot make much sense of the suggestion that what we believe when we believe that ‘P’ is true (and even less if we merely believe that ‘P’) is that ‘P’ is justified/warranted. Nor can I understand why the fact that we believe in ‘P’ on the basis of various warrants should imply that it is the warrant and not the correspondence of ‘P’ to fact that makes ‘P’ true. Surely it is possible to think both that (i) some thoughts correspond to reality, and (ii) our reasons for believing that they do so correspond to reality are based on something else than correspondence. So we can justify our belief that ‘P’ corresponds to fact by appeal to warrant, and yet think that it isn’t the warrant that makes ‘P’ true; the warrant is what justifies our belief that ‘P’ corresponds to fact.

We can reformulate the same worry in connection to one concern that may motivate the rejection of correspondence as a basis for truth. This is the complaint that it makes little sense to attribute a property that we can never know for certain is actually possessed by any of the things we call ‘true’. Instead we should only use the word for what can be ascertained of any belief, like utility and/or coherence with other beliefs. There are two objections that come to mind. First, why should we think that the truth of a belief is a matter of attribution of a property to the belief? For those who believe there is an external reality to which our beliefs sometime correspond, correspondence is not the attribution of a property to beliefs, but the having of a property by a belief, and that there is nothing we can do about that except to try to have ideas. We don’t make beliefs true; truthmakers do. The epistemic problem then has to do with how we justify our conviction that the belief has that property in virtue of some or other truthmaker.

Second, as Carnap observes (1949), the skeptical argument applies equally to every property we can think of, and would, if sound, render most words, if not all, useless by the same measure. We cannot with absolute certainty know whether any property ascribed to any object is really possessed by the object (or if the object really exists), and hence all words used to denote properties should be judged useless/meaningless; not to forget the property of being ‘warranted’. These objections are of course intended to reveal the absurdity of the position by revealing its backfiring consequences. It says that either we demand absolute certainty for the application of any term or make due with a degree of uncertainty about them all. Since we do accept some degree of uncertainty about ascriptions of knowledge-in so far as we accept that knowledge is fallible-and about the ascription of every property known to mankind, then why should we not accept such uncertainty about truth? The main point is that properties are simply had by their bearers whatever we chose to attribute to them. It is not their having of a property that is uncertain, but our knowledge about them having it.

It is true that some proponents of ET have rejected the relation to reality. But it is difficult to see that the basic ideas of ETs exclude that there could also be a relation to reality. That is, it might well be the case that we reason and talk about truth in terms of various ‘warrants’, and yet that the whole point of these ‘warrants’ is to justify our belief that some idea corresponds to fact. On that understanding, epistemic theories simply are theories about valid reasons for believing that something is true, but not theories about what makes something true. So understood, CT and ET are strictly speaking about different things and perfectly compatible.

Indeed, the strength of ET would be greatly reduced if they were cashed out as genuine rivals to correspondence-as explicit rejections of a relation between belief and fact-because then they turn out to be versions of extreme skepticism or idealism. Only CT explicitly states anything about a relation to reality, at least provided we think of reality as the spatio-temporal world, and not the realm of propositions. Indeed, all theories that have posited a relation to reality, but have used other names for that theory, have been adopted into CT. After this general reflection, let us have a closer look at a couple of popular suggestions about what kind of ‘warrant’ could constitute truth.

After these general remarks, let us focus on the pragmatist theory of truth (PT), which is often expressed as the view that to be true is to be useful, i.e. that ‘a proposition is true if it is useful to believe that it is true’ (^{Schmitt 1999}: 107; ^{Lewis 2001}). The utility in question is measured in terms of the success of actions performed on the basis of those beliefs. As so many thinkers have pointed out over the years, this version of PT is too implausible to be taken seriously because it is all too clear that utility and truth do not coincide (^{Russell 1910}: 121; ^{Horwich 1994}: xiii). A belief may be a useful guide to action, and nevertheless be false. To believe that the stores close at 5 p.m. may prevent a person from ever failing to buy the groceries in time, even though in fact the stores stay open until 9 p.m. Every semi-successful scientific theory that has been superseded by another theory is an example of the (limited) utility of some false ideas. A more plausible understanding of the relation between utility and truth is that utility is a consequence of truth, and therefore an indication for the truth of a proposition, but not as being truth itself.

Susan Haack has argued that the idea that pragmatists equate truth with utility is mistaken (1976). According to her, neither Peirce, James, nor Dewey deny that truth is correspondence, but instead admit that it is correspondence. Haack suggest that they are merely considering the practical difference between having true and false beliefs, giving the answer that true propositions will not only be useful in the short run, but also resist what later thinkers have called falsification. This is not to say that utility constitutes truth, but to assume truth is correspondence and to draw out the consequences of having beliefs that correspond to reality, notably (i) they will be useful, and (ii) we can (fallibly) warrant our belief in the correspondence of an idea to world by appeal to how well it guides us in our practical endeavours in the world. I suggest this should be the appropriate stance.

Admittedly, it may perhaps be necessary to read the collected works of the pragmatists ‘charitable’ in order to render them consistent on this point. But one cannot expect any thinker to remain consistent in their writing from birth to death. The pragmatists may have considered, at some point, the possibility of discarding altogether the correspondence of belief to fact. But when considering the substitution of correspondence for utility they have been at a loss for explaining why some propositions are useful and not others. Consequently, some kind of relation between belief and fact has crept into their thinking again.

On Haack’s understanding, PT is not a rival theory to CT. It is a complementary theory about the practical consequences of having beliefs that correspond to fact, which entails the epistemic thesis that utility is an indicator of truth. Consequently, pragmatism can very easily be seen as an ET about reasons for believing or assuming that something may correspond to fact. Indeed, if all true beliefs are useful, but only some false beliefs are useful (and only in the short run), it appears as if utility is an indication for the truth of propositions, albeit not an infallible one.

Let us now turn to the coherence theory of truth (COH). I am very much inclined to accept the standard objection to COH, considered as an epistemic and not a metaphysical theory, notably, that however important coherence may be as an indication of the truth of a proposition, it is not itself constitutive of truth (^{Pap 1949}: 356; ^{Rescher 1985}; Dorothy ^{Emmet 1992}: 20; ^{Horwich 1994}). A belief may be coherent with all our other beliefs about the world, but nevertheless be false. After all, our worldview may be to a large extent false. But, arguably it is more unlikely that all our beliefs about the world are false, than that one particular belief is false. Therefore, if a belief is consistent with the rest of our beliefs, and if its assumed falsity would imply that our world-view by and large is false, we find it more likely that the individual belief is true than that all our other beliefs are false.

Very plausibly, COH is a theory about how we reason about truth, rather than what truth is. The idea that coherence provides a good reason for believing in the truth of a propositions has been criticised, e.g. by Erik ^{Olsson (2005)}, but we need not consider that point here. My point is that even if coherence is assumed to provide good reasons to believe in something, it would not establish that beliefs never correspond to fact, nor would it provide an alternative understanding of the relation between belief and fact. Indeed, if, for the sake of argument, we remove the correspondence of belief to fact, then the idea that coherence constitutes truth turns the claim ‘the coherence of P with some system of beliefs S, warrants our belief that P is true’ into the platitude that ‘the coherence of P with some system of beliefs S, warrants our belief that P is a coherent part of S’. A similar platitude appears for straw man position that utility constitutes truth, notably ‘the utility of P warrants our belief that P is useful’. The result is that not just truth, but knowledge as well, is reduced to a tautology about the connection between belief and warrant, one that says nothing about the relation between knowledge and world.

To treat PT and COH as rivals to CT would require us to understand them as contrary theories. Either as saying (i) there is no such thing as a relation to reality, (ii) the relation to reality is irrelevant, or (iii) the relation is not correspondence. In the first two cases, beliefs are true because they are useful/coherent regardless of any relation to reality at all. In the third case, there must be some ‘utility relation’, or ‘coherence relation’ to reality, quite different from correspondence. As far as I know, there are no suggestions about alternative relations, at least not to that part of reality that isn’t propositional, and as we have seen the correspondence relation has utility built into it (but not vice versa).

It bears to mention that PT and COH need not be rival theories either. A belief can be supported both by appeal to its practical utility and its coherence with an already accepted body of beliefs, and then we have two good (but fallible) reasons to believe that it is true, even though it may in fact turn out to be false anyway.

Deflationary Theories of Truth

It is difficult to give a general characterization of deflationary theories (DT) because it is not obvious that there is a single core idea that unifies that family of views, except their opposition to any attempt to provide a single unified answer to the question ‘what is truth’.

DT is sometimes presented as the view that the only interesting thing to know, or say, about truth is that propositions specify their own conditions for being true. However, as Richard Kirkham notes, this is an oversimplification of DT. Kirkham distinguishes what he calls the ‘deflationary thesis’, the claim that truth is not a genuine property, from various ‘speech act projects’, which profess to establish the ‘redundancy thesis’, i.e. that the word ‘truth’ is redundant (^{Kirkham 1992}: ch. 10). Kirkham points out that the speech act projects do not really support the deflationary thesis. It is possible to accept the redundant function of the word ‘true’, and yet accept that truth is a genuine property; ergo, the deflationary and redundancy theses are independent of each other. Propositions can arguably correspond to reality whether or not we ascribe truth to them. Indeed, if knowledge is fallible, we can only talk hypothetically about the truth of propositions, which turns ascriptions of ‘truth’ into expressions of the speaker’s endorsement of a proposition, his/hers insurance of being sincere, or his/hers belief that the proposition counts as ‘true’ in light of some warrant. Instead, Kirkham suggests, the proponents of the speech act project really commit to the deflationary thesis on the basis of prior commitment to some form of anti-realism and not as a result of the perceived redundancy of the truth-predicate.

The deflationary thesis is contrary to CT, since it says that there is no such thing as property of truth. However, aside from various speech act projects, it is not clear that any version of DT endorses it. Paul Horwich’s minimalism, often labelled as a DT, is neither a rejection of a property of truth nor is it a part of the speech act project. On the one hand Horwich readily acknowledges that “truths do correspond-in some sense-to the facts; it [minimalism] acknowledges that statements owe their truth to the nature of reality” (1998: 104). But, on the other hand, Horwich admits that minimalism “does not explain what truth is in any such way”, and nevertheless he claims that it is a theory of truth itself, not just of the function of the word ‘truth’. He is not, I hope, guilty of contradiction. He is merely trying to explain truth in a non-standard way, notably implicitly. Minimalism, arguably, brings a new meaning to the term ‘deflationism’. It aims to deflate the idea that a theory of truth should be in some sense ‘substantive’.

Unfortunately, due to the implicitness of the theory, the finer points of Horwich’s minimalism escape me, and I remain unconvinced that minimalism is as good a theory of truth as we can ever hope to get. Indeed, the emergence of pluralism is a manifestation of the conviction that a more substantial theory is possible. Pluralism arguably arose from Crispin Wright’s attempt to understand how Horwich’s claim that truth is a genuine property could be reconciled with his claim that the minimalist implicit definition of truth is as close as we can get to explaining its nature (1992: Ch. 1). Wright argues that minimalism is right not to deny that truth is a genuine property, because it is incoherent to deny this, and at the same time make normative claims about the correct use of ‘truth’. He considers the possibility of construing minimalism as the claim that there is no single metaphysically significant explanation of the property that ‘true’ ascribes, because ‘true’ may ascribe many different properties. In other words, any normative theory about the correct use of ‘true’ requires there to be a genuine property, but, Wright suggests, it may not always be the same property.

On Wright’s reading, minimalism is implicitly a pluralism about what kind of property the word ‘true’ ascribes to propositions. In some contexts, it may be correct to say that something is true because it is coherent (e.g., in mathematics), sometimes because it is scientifically verified to correspond to reality (natural science), sometimes even because it fits to some moral or aesthetic norm (humanities). That is, the relevant property may vary with the context, and may not always belong to the proposition in question in virtue of corresponding to some fact. This is why we should not expect, or look for, a single metaphysically significant property that all true propositions have in common (like correspondence). This could be the motivation behind the minimalist claim that all we can ever hope to achieve is a general schema-the T schema-that fits all the uses of the word ‘true’, and which is consistent with the ontologies of all the different theories of truth. Pluralism, then, is neither deflationary with respect to a property of truth (in fact it is inflationary) nor a theory of truth and should therefore be considered a separate family of views.

Pluralism

The pluralists that have followed in Wrights wake are not as focused on making sense of minimalism, and many enough prefer to think of all true claims as having the same property-being true-while they have it in virtue of different grounds (^{Lynch 2009}; Cory Wright & Pedersen 2010; ^{Edwards 2013}). Contemporary pluralists take themselves to be making sense of two observations about the state of the dispute about truth. First, that the plausibility of various truth-theories varies across philosophical discourses: CT is popular in metaphysics, DT in philosophy of language, ET in epistemology, and COH appeals to antirealists and idealists. Second, that the problems and limitation of each theory emerge as their proponents attempt to overgeneralise their preferred view across the board. Pluralists take this to naturally suggest that claims in each domain are true in different ways. Using Pedersen’s terminology, claims in different domains of discourse can all be truth-apt but they acquire the property of being true in different ways (2014). Empirical claims about contingent matters of fact are true in virtue of corresponding to fact, epistemic claims about beliefs being warranted are true in virtue of those warrants, claims like ‘2+2=4’ and ‘bachelors are unmarried men’ are analytically true in virtue of the concepts they express, and moral and aesthetic claims could be true in virtue of moral/aesthetic norms.

What I have been arguing throughout this paper is really that pluralism is not the only way to explain the variation in appeal of different theories in different domains. A better explanation, I think, is that thinkers in different domains approach the issue of truth on the basis of very different ideas about what they are trying to explain. As a result the truth-predicate is used in different ways in contexts; ways that have as little in common as the different ways we use the world ‘dry’. However, as with ‘dry’, I think we can identify an original, or alternatively, a core sense of ‘true’, and show that it is in light of that core sense we can understand in what way the other uses are metaphorical, i.e. figures of speech in which the truth-predicate is used to suggest a likeness or analogy to the core sense. The understanding I have in mind is roughly this. The idea that beliefs owe their truth to what the world is really like-which is the core idea of the truthmaker principle-naturally invites the idea that truth is about the way things really are. Any claim that can give the impression of being ‘about the way things really are’, is therefore likely to be understood as truth-apt, even when on reflection we may suspect that there is no fact of the matter, or when it is clear that the claim can only represent the speakers subjective opinion. Claims like ‘meat is murder’ or ‘babies are cute’, do grammatically look like assertions of fact and are not regarded by everyone as mere opinion. Obviously, claims that we regard as established knowledge, i.e. more than mere assertions about fact, come closest to being not merely truth-apt but true. Indeed, when we consider that it is on the basis of various epistemic warrants that we decide to believe or not to believe in the truth of a claim, then the suggestion comes naturally that it is the warrant and not the correspondence to fact that is of greater importance for our deliberations. However, just like it is possible to accept that the truth-predicate is redundant, then we can also accept that the actual correspondence of a belief to fact is never among the reasons we can have to believe that the belief so corresponds. This suggests that there is a clear difference between (i) what truth is, (ii) how we use the truth-predicate, and (iii) how we decide to believe that the content of belief is true. If we nevertheless assume that it is the warrant for deciding to believe that makes the belief true, we are left with the question of how beliefs, and the practice of warranting beliefs, helps us relate to the world.

I can detect a similar reasoning among pluralists, but where they identify the core understanding of ‘truth’ as ‘valid claim’ rather than ‘how things really are’ (^{Pedersen 2014}; ^{Strollo 2018}). According to this suggestion, the common feature of true claims across discourses are that they are the ones that are perceived as valid in that particular discourse, while it varies between discourses in what way they are judged to be valid. I find this to be a plausible suggestion about what many philosophers actually think, but I also think that this cannot be the correct account of truth. First of all, validity and truth come apart in logic, and in scientific discourse generally. If they did not, then all valid arguments would also be sound. We need correspondence truth to distinguish between sound and unsound arguments. Second, because the correspondence relation can hold (or not) independently of any reasoning we can muster in support of the truth of a belief; CT does not identify true claims with valid claims, and so could never be among the views united under the banner of pluralism. Third, because in epistemology valid reasoning coincides with the kind of reasoning that warrants belief, and then my earlier worries about collapsing the distinction between truth and knowledge come to the fore. The only explanation I can think of, that adequately explains the heterogeneity of the use of ‘truth/true’ without raising more questions than are answered, is that the truth-predicate has become associated with the core understanding ‘how things really are’ (as opposed to what corresponds to how things really are) and then used metaphorically to denote any claim that appears to somehow assert, rightly or wrongly, how things really are (in the domains of empirical, epistemic, aesthetic, moral, and logical discourse).

Now, I want to repeat that CT doesn’t really stand or fall with the plausibility of my explanation of the heterogeneity in the use of the truth predicate. No matter whether the core connotation of ‘truth’, as it is used across domains of discourse, is ‘how things really is’ or ‘valid claim’, then it remains true that the actual heterogeneity in the uses of ‘truth’, has nothing much at all to do with the question of whether and how our beliefs sometime relate to reality, or what our epistemic practices revolve around. And in both cases one can still complain that it would be useful to make certain analytic distinctions with regards to the different uses of ‘truth’. Perhaps even suggest that other expressions than ‘it’s true’ preferably be used in certain contexts to express sincerity, conviction, judgement of taste and coherence, just for the sake of clarity. For instance, that, when appropriate, we say ‘I am being sincere/honest’, ‘I am certain/convinced about this’, ‘I really like this cereal too’, and ‘this is coherent with the rules of arithmetic’, or ‘this is genuinely surreal’. That would help to improve clarity in all domains of discourse.

Conclusion

Many things can be said, and have been said, about the problems of construing a correspondence theory of truth (^{Mulligan, Simons & Smith 1984}; ^{Kirkham 1992}; ^{Sher 2004}). What is the nature of the facts that make propositions true? What are the proper bearers of truth? What exactly is correspondence? What makes negations true? Are there negative facts? My intention is not to trivialise these problems, or to answer them. I admit that the correspondence theory is not much more than a projected program, or an idea whose strength is based almost solely on the difficulties we encounter when trying to do without the idea that truth is correspondence. What I hope to have achieved is a more nuanced view of what it takes for two theories of truth to genuinely contradict each other, and also a more nuanced view of the parameters to consider in the scrutiny of what a particular theory of truth is really saying. Furthermore, I hope to have shown that it is possible to construe the various theories as complementary ideas about the issue ‘what is truth and how do we reason and talk about it?’ Indeed, if there is such a thing as a substantial theory of truth, it would better include both metaphysical, epistemological, and semantic aspects. Here is my version: truth is a property of those beliefs whose content correspond to the facts they represent, and which we come to believe are true on the basis of a variety of factors that warrant that belief, such as their coherence with the system of already accepted beliefs and whether they reliably lead to successful practices.