1. Theories of Truth

Over the years philosophers have proposed many different theories of truth. The following is not an exhaustive list but it does cover the main options, or at least those that have received the most attention historically:

1. 1. The Correspondence Theory of Truth says that a true claim is one that corresponds to the facts or reality. The sentence ‘Snow is white’ is true because, well, it corresponds to the fact that snow really is white. Among others, Bertrand Russell and J. L. Austin have proposed correspondence theories of truth.

2. 2. The Coherence Theory of Truth says that a true claim is one that is consistent or, more strongly, ‘coheres’ with other things we believe. For example, F. H. Bradley and Brand Blanshard have proposed coherence theories of truth.

3. 3. The Pragmatic Theory of Truth says that true claims are those that are useful and dependable. Acting on the basis of a true belief will help you get what you want, while acting on the basis of a false belief will lead to disappointment and regret. William James is best known for defending a pragmatic theory of truth.

4. 4. Minimal Theories of Truth (sometimes called ‘deflationary’ or ‘disquotational’ theories) try to boil truth down to its essence. These theories point out that there is no difference between saying:

   ‘Snow is white.’

   and saying:

   ‘It is true that snow is white.’

   This suggests that the concept of truth is not a profoundly deep metaphysical notion but instead serves some rather simple but essential functions. The most important is that it allows us to make generalizations so we can say things like ‘everything Mary just said is true’. Without the concept of truth we'd have no choice but to repeat everything Mary just said, which would be awkward and time-consuming.

Of course, each of these theories come in different flavours, and the disagreements between defenders of the same general theory can be just as heated as disagreements between defenders of different theories. That point aside, this gives us a reasonably complete picture of the current terrain.

2. The Goldilocks Principle

The Goldilocks Principle is that a theory of truth should neither do too much nor too little. It should be just right. On the one hand (not doing too little) this means that a theory of truth should not leave out important features that are central to the concept of truth. Zebras look a lot like horses, but an account of zebras that only mentions their horse-like appearance, and doesn't mention their distinctive black-and-white pattern, is sadly incomplete. It doesn't do enough.

On the other hand (not doing too much) this means that a theory of truth shouldn't stack the deck. Potentially controversial issues should be settled by additional inquiry, not by definition. In particular, we don't want a definition to be so narrow that it excludes things that should, intuitively, fall under it. This would be like defining ‘language’ as ‘a spoken or written form of communication’ which sounds fine until you realize it excludes the sign languages (such as ASL) used by deaf and hard-of-hearing people. Or defining ‘parent’ as ‘a biological mother or father’ which excludes step- and adoptive parents. Unless there are good independent reasons for these exclusions (maybe one needs a genetic match for medical purposes) definitions like these do too much.

When it comes to truth, the concern is that some theories do too much while others do too little. (It's also possible for theories to do both too much in some ways and too little in others.)

3. Doing Too Little

Minimal theories of truth, for example, are obviously open to the charge of doing too little, of being too minimal. So, while they do capture one of truth's essential roles – namely, allowing us to make useful, time-saving generalizations – they seem to miss other essential functions the concept plays. Among other things, calling a claim ‘true’ makes a speaker responsible for defending that claim. Consider these claims:

and

While in one sense these mean the same thing – they are both claims about the time of my flight – in another sense the second claim means more than the first. The extra effort of adding ‘it's true that …’ gives the second claim a different force. It implies more than the first claim: perhaps that I just checked, or that I have a particularly clear memory, or that I just got a text from the airline. The second claim implies that I have evidence or reasons that I'm prepared to give. Adding ‘it's true that …’ signals that this is a good claim, that it is dependable, and that it can be defended. In contrast, the first claim doesn't have these implications. It's compatible with much weaker claims such as ‘My flight leaves at 12:45, but don't ask me how I know!’

The concept of truth thus plays an important practical role in how we make and judge claims. Adding ‘it's true that …’ raises expectations on speakers: that we can give reasons or cite evidence or otherwise defend our claims. The concept of truth sets a standard for what counts as a solid claim versus idle speculation or musings. Unfortunately, minimal theories of truth often overlook this dimension (and perhaps others too) and, when they do, they do too little. They don't satisfy the Goldilocks Principle.

4. Doing Too Much

But theories of truth can also do too much, and I would argue that is where the real danger lies. For one thing adding to a theory is easier than taking away. This means that theories that do too little should be easier to fix than theories that do too much. For another, I'll argue below that the correspondence theory does too much. This is especially troubling because the correspondence theory strikes many people as obviously correct and that can magnify its effects. (In contrast, most people haven't even heard of minimal theories.)

So, how does the correspondence theory of truth do too much? Recall that, according to the correspondence theory, a true claim is one that corresponds to the facts. The problem is figuring out which sorts of claims are capable of being true or false in the first place. (This question is often framed as which claims are ‘truth-apt’.) ‘The cat is on the mat’ is either true or false: either the cat is on the mat or it isn't. And whether it's true or not depends on a clear fact of the matter: the cat, the mat, and whether one is on top of the other. So far so good. But what about claims like ‘this cookie is going to taste so good’? The problem here is that this is a claim about the future. If it's true, according to the correspondence theory, then it must correspond to a future fact. But what is a future fact? Future facts don't seem to exist here and now – they're about the future, after all – so it's hard to see how a claim can correspond to them. And that would suggest that claims about the future can't be true now, though they might become true in the future. In other words, claims about the future can't be true (or false) because there are no future facts to correspond to. Finally, since we can only know things that are true, this would mean that we can't know anything about the future.

Future facts aren't the only facts that get spooky. Consider mathematical claims such as ‘8 is the 1000th digit of π’ or ‘prime numbers get rarer the higher one counts’. If these are both true (they are!) then, the correspondence theory says, they must correspond to a mathematical fact. But what is a mathematical fact? Like future facts, mathematical facts don't seem to exist in any concrete, tangible way. We can't see or experience them in the same way we can experience a cat on a mat. If that's so, then what reason do we have to believe that there are mathematical facts?

Finally, consider ethical claims like ‘lying is sometimes OK’ or ‘it's wrong to eat animals’. As with these other claims, the idea of ethical facts is not straightforward. We know what the fact of a cat on a mat looks like, but what does an ethical fact look like? How do we recognize an ethical fact from an ethical fiction? If ethical truths depend on corresponding ethical facts then we might begin to doubt whether there are any ethical truths at all.

One option is to bite the bullet and concede that claims about the future, maths, and what's right (among others), can't be true or false because there are no facts for them to correspond with. Actually (in fact?), this is a bullet people have been known to bite: for example, many people will deny that there are future facts and will claim, as a result, that we can't really know the future, simply because there aren't any truths to know. This position is not cost-free. For example, science and engineering seem to be in the business of uncovering facts about the future (‘if you mix these two chemicals’ etc.), so if there are no future facts then science and engineering lose some of their point. This means we should proceed cautiously.

Mathematical truths also suggest caution. Few would deny that mathematical truths exist: every maths exam we've ever taken, every time we've balanced our bank statement, we assumed there were mathematical truths. (If I balance my statement twice and get different numbers, I don't just average them and call it good. I know that at least one must be wrong.) Mathematics is so useful, so certain, and so universal that giving up the idea of mathematical truth seems completely reckless. It seems so reckless that I suspect many people would be willing to accept the existence of mathematical facts (spooky as they sound) if doing so is necessary to preserve mathematical truths. But that's problematic too. After all, the original idea was to use facts to help us define truth. But now we seem to be using our belief in mathematical truths to support the existence of mathematical facts. That's the opposite of the original intention and not what the correspondence theory of truth was designed to do.

We should also proceed cautiously when it comes to ethical claims. It is tempting to take the spookiness of ethical facts as evidence that there are no ethical truths. However, there are costs here too. The fact that we debate, discuss, reason about and often arrive at consensus on ethical claims (here's an obvious one: slavery is wrong) would suggest that we generally do treat ethical claims as being true or false. We use them to make sense of our own and others’ behaviour, to justify our actions and choices, to organize society, and to punish wrongdoers and reward the virtuous. It's not too much to say that ethical truths, and our attitudes towards them, are a large part of what makes us human. If it turns out that they don't exist, if we can't appeal to them to organize our society, or justify our actions, or to assign praise and blame, then we lose a bit of our humanity.

By itself none of this means that there are ethical truths (or truths about the future, or mathematics, etc.). It just means that we should be careful before we reject their existence. The costs can be considerable.

The general concern is this: correspondence theories of truth define truth in terms of corresponding facts. These facts are what make a claim true (or false). This seems to work well when we're talking about physical objects in the here and now, such as cats and mats. But, in many other cases, it's not clear that the corresponding facts exist. If the facts don't exist, or if we have trouble seeing how they could, then the corresponding claims cannot be true or false. And if these claims cannot be true or false then it's impossible to have knowledge of what these claims are about. This casts a chill over our attempts to investigate, understand, debate, and act on these claims. To bring this back to the Goldilocks Principle, this means the correspondence theory does too much. It stacks the deck against certain types of claims and makes it impossible for us to have the knowledge we want.

5. Diagnosis

What went wrong? The problem is not that there are limits on what's true and whether we can know it. The problem is with how we got to this outcome. Usually, if there's a topic where we can't know the truth (if there is one), then that's because the topic has some feature that puts it beyond reach. The topic creates the limitation, not our using a particular theory of truth. For example, if I don't know what you're getting me for my birthday – or even whether you're getting me anything for my birthday – that's because you're really good at keeping secrets. It's not because I have a particular theory of truth.

It's the same in other cases, too. If we don't know whether there's intelligent life in the universe, that's because of the enormous distance any message would have to travel. If we can't predict this evening's traffic, that's because traffic is a non-linear dynamic system, like the weather. If there's no way of knowing the next spin of a (fair) roulette wheel, that's because every outcome is equally likely. If you don't know whether to tell a rather innocent lie or a rather painful truth, then that's because there are costs and benefits either way. If your boss can't know what's in your e-mail, that's because you used PGP encryption, and so on. The common thread is that if there are things we cannot know, truths we cannot access, we generally treat this as a feature of the thing under discussion and not as the consequence of having a particular theory of truth. Restrictions on truth and knowledge are not set by definition. Instead, they are usually encountered through hard-won experience, when our best efforts at understanding or prediction are thwarted. The fact that we can inquire into and explain the reasons for these restrictions is further evidence that their basis is not a matter of definition but rather depends on the world and what we're trying to understand. In this case theory – a theory of truth – shouldn't take precedence over our practical experience.

6. Where Next?

At the beginning I said that the Goldilocks Principle – neither doing too much nor too little – sets a standard that any theory of truth should meet. We've now seen how theories of truth run afoul of this principle: minimal theories for doing too little, correspondence theories for doing too much. I've had more to say about the correspondence theory for two reasons. The first is its popularity: because the correspondence theory can seem obviously correct it's important to stress that it has less obvious, but still considerable, costs. The second is its scope: because the correspondence theory does too much, it's important to describe what ‘too much’ means here. This does not mean that other theories (coherence, pragmatic, etc.) get off scot-free. But since they have fewer defenders they have not been my focus here.

Truth is a fundamental concept that plays a key role whenever we make decisions, judge others, and act on our beliefs. The Goldilocks Principle encourages us to consider what we want from a theory of truth: both what we want it to do and what we want it not to do. It helps us triangulate on the most important features of truth itself, the features that a theory of truth should incorporate and preserve.

Today the concept of truth seems under attack. Politicians and others brazenly mislead and lie, as if making a claim often enough or loudly enough or on Twitter enough can actually make it true. But we also do harm to truth when we restrict the concept and place limits on the kinds of claims that can be true. Some theories of truth, I've argued, risk doing exactly that. When we think about truth – when we think about what it is, how to recognize it, how to discover it – it pays to keep the Goldilocks Principle in mind.