Justified true belief

knowledge and the myth of propositions

 

Peter Holmes

July 2017

 

I do not claim originality for any of this. I just want to simplify and clarify some issues in epistemology – for my own benefit, and in a way that may interest you. And I’ll gratefully acknowledge helpful comments. Thanks to Reuben Holmes for suggested improvements.

 

My argument is that talk of propositional knowledge can mislead us into mistaking what we say about things for the way things are; and that the definition of knowledge as justified true belief demonstrates the potency of the myth of propositions.

 

Contents 

 

1  Introduction

2  The truth condition

3  Propositions

4  The truth condition - continued

5  Facts

6  Propositional knowledge

7  Objectivity

8  Gettier's criticism

9  Defining knowledge

10  Foundationalism

 

1  Introduction

 

Knowledge has been defined as justified true belief (JTB). Supposedly, there are three necessary and jointly sufficient conditions for knowledge: truth, belief and justification.

 

In the following version, S is the subject and p is the proposition.

 

S knows that p if and only if:

p is true;

S believes that p; and

S is justified in believing that p.

 

It has been argued that some cases of justified true belief do not amount to knowledge, so that the JTB definition is inadequate. And I discuss this problem in section 8 below: Gettier's criticism.

 

But I suggest there are other problems with the definition, beginning with the truth condition.

 

2  The truth condition

 

The JTB truth condition is: S knows that p if and only if p is true. So if p = the earth orbits the sun, by substitution the truth condition reads:

 

S knows that the earth orbits the sun if and only if the earth orbits the sun is true.

 

But this is nonsense, unless we identify the second use of p as a linguistic expression, perhaps as follows.

 

S knows that the earth orbits the sun if and only if the proposition ‘the earth orbits the sun’ is true.

 

To generalise, this seems to mean: we know a feature of reality if and only if a proposition asserting that feature of reality is true. But if that interpretation is correct, this condition for knowledge seems confused.

 

It is because we know the earth orbits the sun that we know the proposition the earth orbits the sun is true. To say the truth of that proposition is a necessary condition for our knowing that feature of reality is to get things back to front.

 

Of course, we say we know something is the case only if it is, or we think it is, the case. And if it turns out not to be the case, we do not say we have stopped knowing it. We just say we were mistaken. For example, we did not stop knowing the earth is flat.

 

The primary condition for our knowing a feature of reality is that it is the case. If it is the case, a proposition asserting it is true. But the truth of that proposition is a secondary and separate consideration.

 

To put this another way: our knowing that p does not come from the truth of the proposition p. It comes from our knowing the feature of reality that p asserts. We can know that p is true only if we know it correctly asserts that feature of reality, given the way we use the words involved. So in effect, the JTB truth condition asserts that a necessary condition for knowledge is - knowledge. 

 

There are features of reality; there is what we believe or know about them, such as that they are the case; and there is what we say about them, which may be true or false. To muddle these things up is a mistake.

 

The JTB account of knowledge falters at the first fence: the truth condition. But it has been taken seriously, not least by critics such as Edmund Gettier. So it may be interesting to find out where our thinking goes astray.

 

One problem with the definition is its concentration on subjective knowledge – what an individual knows – to the effective exclusion of objective knowledge and its justification. And I consider this in section 7 below: Objectivity.

 

But I also question the role of propositions in a definition of knowledge: S knows that p. Does so-called propositional knowledge (knowing-that something is the case) involve knowing a proposition? What does knowing a proposition mean? And anyway, what are propositions?

 

3  Propositions

 

We can often say the same thing in different ways and languages, such as it is raining, there is aqueous precipitation and il pleut. In philosophy, the one thing expressed in these different ways is called a proposition. A proposition (p) is supposedly what a declarative declares, a statement states, an assertion asserts, and so on.

 

Spoken or written linguistic expressions, such as factual assertions, are real things: sounds, marks on paper, and so on. By contrast, propositions are supposedly abstract (unreal) things that can manifest in different linguistic forms. So we could think of them as undeclared declaratives, unstated statements, unasserted assertions – and other such contradictions in a non-linguistic language.

 

Of course, abstract things such as propositions may exist. But in the absence of evidence for them, there is no reason to believe they do. Nor have we given the words thing and exist clear meanings in this context. More likely, to assert or deny that abstract things exist is to mistake abstract nouns for things that therefore may or may not exist.

 

If we know a linguistic expression such as it is raining, that usually means we can use and understand it, because we speak at least a little English. And French speakers use and understand il pleut in the same or a similar way.

 

But to say we can know an abstract thing, such as a proposition, is to equivocate on the word know. As yet, no one has explained what it means to know a proposition. My guess is that, like concepts and Kant’s noumena, propositions are convenient fictions that surely must somehow exist, even though we cannot quite explain what they are.

 

Careless talk of abstract things can lead us into confusion. And the JTB definition of knowledge is a chastening example.

 

4  The truth condition - continued

 

Back to the JTB truth condition: S knows that p if and only if p is true. As before, p = the earth orbits the sun.

 

S knows that the earth orbits the sun if and only if the proposition ‘the earth orbits the sun’ is true.

 

Next, we can label the two uses of the earth orbits the sun as p1 and p2, as follows.

 

S knows that  the earth orbits the sun  [p1]

if and only if the proposition  'the earth orbits the sun'  [p2]  is true.

 

Of course, p1 and p2 are the same proposition: the earth orbits the sun. But here, we are using that proposition in radically different ways. Obviously, p2 is meant to be understood as a linguistic expression. But then, what is p1 supposed to be?

 

I suggest that p1 is somehow supposed to be the actual feature of reality (the earth’s orbiting the sun), which just happens to be asserted by the proposition the earth orbits the sun. To put it another way: with p1 we seemingly efface the linguistic nature of p, while with p2 we declare it.

 

And this trick is made possible by the myth of propositions. If we think of them as abstract (unreal) things, it is possible to mistake them for the features of reality they assert. We can project propositions – what we say about things – into the things themselves, blurring the distinction.

 

And a neat little argument seems to justify what we do, as follows.

 

1  A proposition is what an assertion asserts.

2  An assertion asserts a feature of reality.

Conclusion:  A proposition is the feature of reality that an assertion asserts.

 

This is deductively valid. But the first premise is unjustified speculation, and the conclusion is ridiculous. The argument is alluringly simple, but unsound.

 

If instead we recognise that what we call propositions are nothing more than factual assertions – linguistic expressions – we cannot rationally mistake them for the features of reality they assert. The distinction between the way things are, and what we say about them, stands out clearly.

 

To put all of this another way: the assertion, S knows that p if and only if p is true, equivocates on the function of p. It is both a putatively non-linguistic representation of a feature of reality, with no truth value; and also a linguistic factual assertion with truth value. And the irony is that non-linguistic p can function only if linguistic p is true, so its truth is assumed in the assertion, S knows that p.

 

5  Facts

 

This muddle over the word proposition is a niche concern for philosophers. But our everyday use of the word fact is a more striking example of the same mistake.

 

We often talk about the facts of reality, and how truth is that which comports with those facts. But this is to identify facts with features of reality, to think there are facts inherent in reality, or that reality consists of facts.

 

But features of reality just are, neither true nor false. It is what we say about them that can be true or false. And as it happens, we call true assertions about features of reality – facts.

 

Facts are just true factual assertions. They are linguistic expressions. And there are no linguistic expressions inherent in reality. Reality is not linguistic. So it is neither factual nor propositional. The truth is not out there any more than falsehood is.

 

6  Propositional knowledge

 

Supposedly, propositional knowledge has something to do with propositions. And the JTB definition of knowledge is narrowly propositional: S knows that p if and only if p is true, and so on. But this kind of knowledge is also called descriptive or declarative, which is a clue to its nature and function: to express things about features of reality.

 

Propositional knowledge is sometimes called knowing-that, to distinguish it from procedural knowledge (knowing-how) and knowledge by acquaintance (knowing-of). And the integrity of propositional knowledge rests particularly on the distinction between knowing-that and knowing-of. But a little blue-sky thinking calls that distinction into question.

 

If we look at a blue sky, we can know the sky is blue. That is an ordinary use of the word know, to mean something like be aware that. And this is knowledge of a feature of reality: the blueness of the sky.

 

But is our knowing the sky is blue a case of knowledge by acquaintance: knowing-of the blueness of the sky? Or is it a case of propositional knowledge: knowing-that the sky is blue? I suggest this distinction is no more than grammatical. We express knowledge-by-acquaintance by means of noun phrases with no truth value; whereas we express propositional knowledge by means of clauses with truth value.

 

If the sky is blue and we speak English, we know the factual assertion the sky is blue is true, because that is how English speakers use those words. But we would not call this assertional knowledge. Rather, it is knowledge we can express by means of a true factual assertion.

 

In the same way, the phrase propositional knowledge confuses two separate things: what we know and how we express it. Propositional knowledge (knowing-that) is knowledge of reality, expressed by means of true propositions, which are just facts.

 

Of course, we get to know things in different ways. For example, the way we get to know the sky is blue (here, today) may be different from the way we get to know the earth orbits the sun. But still, these are two features of reality about which we can make true or false factual assertions.

 

7  Objectivity

 

True factual assertions constitute the objective knowledge we express in language. And this has a bearing not only on the JTB truth condition, but also on the role of belief and justification. For convenience, here is the JTB definition again.

 

S knows that p if and only if:

p is true;

S believes that p; and

S is justified in believing that p.

 

In what follows, I assume that S, the subject, is an individual person. But S represents any subject, and so all subjects. So we can use the words we, us and our when talking about what S knows.

 

In this section, I follow the convention of talking about true and false propositions - language embedded in the JTB definition. The implications of my objection to the whole idea of propositional knowledge will become evident in the next section: Gettier's criticism

 

p is true

 

The JTB theory defines knowledge as what an individual knows: S knows that p... . So this seems to be an exclusively subjective view. It may even suggest knowledge is something each of us has in epistemic isolation.

 

But we also use the word knowledge objectively, to refer to propositions that are true, regardless of what anyone knows. And the first JTB condition acknowledges this: S knows that p if and only if p is true. The truth of p is independent of our knowing it.

 

Each of us is an individual subject, with our own beliefs and understanding. But objective knowledge, expressed in the form of true propositions, can free us from subjective, epistemic isolation.

 

S believes that p

 

The second JTB condition is: S knows that p if and only if S believes that p. And here believes means something like accepts or agrees that p.

 

One way of expressing this is to say: belief is a necessary condition for knowledge. And another expression is: knowledge is a subset of belief. But these apply only to subjective knowledge, to what an individual knows.

 

Objective knowledge is independent of belief. That is what the word objective means. So, if the proposition p is true, what anyone believes about it is irrelevant. But each of us may have access to objective knowledge.

 

The JTB belief condition is misleading for two reasons.

 

First: to make belief a necessary condition for knowledge is to ignore objectivity. And far from being a subset of belief, objective knowledge renders belief irrelevant. The earth orbits the sun, regardless of what anyone believes, knows or claims to know.

 

And second: in the expression true belief, the modifier true is misattached. A belief is an attitude, and attitudes have no truth value. A proposition may be true or false, but beliefs about that proposition are neither. Obviously, in the condition – S believes that p – it is p that is true, not S’s belief that p.

 

In the justifed true belief definition, the word true really refers to an unspecified proposition, such as the earth orbits the sun. To explain what S knows, the JTB definition should be the JBTP definition: justified belief in true propositions.

 

S is justified in believing that p

 

The third JTB condition is: S knows that p if and only if S is justified in believing that p. And here, being justified means having at least one sound reason for believing that p.

 

But if the proposition p is true (the first condition), justification for believing that p comes ready-supplied. We need no other reason for believing that p, than that the proposition p is true. If it is, the requirement that S needs any other justification for believing that p is redundant.

 

In other words, just as factual knowledge is objective, its justification is also objective, which means independent of what we believe. To feel justified in believing that p may not mean that we actually are justified – how ever hard and sincerely we try to be.

 

Objective knowledge of features of reality, expressed by means of true propositions such as the earth orbits the sun, renders individual belief and its justification redundant. But, fortunately, an individual may have access to objective knowledge.

 

Of course, it is individuals who establish objective knowledge and express it by means of true propositions. But this is a collective, accumulative and (by and large) self-correcting enterprise, made possible by our use of language.

 

8  Gettier's criticism

 

In 1963 Edmund Gettier pointed out that sometimes justified true belief does not amount to knowledge, so that the JTB definition is inadequate. And discussion of the so-called Gettier problem has continued since then.

 

A Gettier-case is a little story with dramatic irony. Given that the story is fictional, we Gettier-spectators know the complete situation. We have, as it were, objective knowledge of the features of reality in the story. But the protagonist does not have this knowledge. Here is an example.

 

A woman sees a group of people and mistakes one of them, a stranger, for her friend. So she believes her friend is there. And as it happens, her friend really is there, but hidden. So what she believes is the case. But does she know her friend is there?

 

As with all Gettier-cases, this story is supposed to show justified true belief failing to amount to knowledge. But I suggest it really demonstrates the confusion caused by the myth of propositions.

 

The point is, what happens in the story has nothing to do with propositions. The woman’s mistake does not come from a false premise. She just believes the stranger is her friend, which is not the case. And her belief that her friend is there is not propositional. Propositional belief is as muddled an idea as propositional knowledge. There are just beliefs and knowledge-claims expressed by means of propositions.

 

We want to say that what she believes is true, because her friend really is there. But that is the myth of propositions at work. What she mistakenly believes to be the case is a feature of reality. And a feature of reality is not a proposition. When we believe or know a feature of reality is the case, we do not believe or know a proposition. So we do not believe or know something that is true or false.

  

The woman does not know her friend is there because she lacks objective knowledge of that feature of reality. And afterwards, apprised of the situation and her mistake, she would not say she knew her friend was there. That is not how we use the word know. She would say she believed the stranger was her friend, but was mistaken.

 

Gettier-cases recycle the JTB’s misleading concentration on subjective knowledge. But they also contain the solution to the Gettier problem. Protagonists believe things for reasons that do not objectively justify their beliefs, which is why their beliefs do not amount to knowledge. So objective justification is what really matters.

 

The distinction between what are called internalist and externalist accounts of justification may mirror the distinction between subjective and objective knowledge. But, pending evidence to the contrary, we have no reason to believe there is more than one reality. So there are features of reality about which we can have objectively justified knowledge, such as the knowledge we have as Gettier-case spectators.

 

The scalar approach to justification ties in with Hume’s line that a wise man proportions his belief to the evidence. But the features of reality about which we may form more or less justified beliefs just are or are not.

 

9  Defining knowledge

 

To define words such as knowledge and know is to explain how we use them, or could use them. By contrast, to define a thing such as knowledge is to describe it, which is a radically different operation. A definition of a thing is just a description.

 

We describe a real thing by making factual assertions about its properties. And perhaps the three JTB conditions for knowledge are supposed to be analogous to the properties by which we identify a real thing. But this is to heap up equivocations.

 

The word knowledge is a noun, so it looks like the name of something whose origin and nature we can try to describe or analyse. But until we have evidence that knowledge is that kind of thing, such a description or analysis is tendentious. So how do we salvage epistemology as a respectable discipline?

 

One approach is to say knowledge is an abstract thing. But claiming to describe or analyse an abstract thing involves more equivocation. Another approach is to say knowledge is a concept. But a concept is another abstract thing. Talk of concepts and other abstract things lures us down the rabbit hole.

 

If instead we concentrate on the various ways we use words, we can free ourselves from the myth of abstract things. For example, the difference between knowing-that, knowing-how and knowing-of, need not puzzle us. We just use the word knowledge and its cognates in explicably different ways.

 

It seems we can be dazzled by important words such as knowledgejustification and truth. We ask: what are knowledge, justification and truth, and where do they come from? We analyse, theorise and endlessly debate different answers, or conclude there are no clear answers.

 

We are asking the wrong questions.

 

10  Foundationalism

  

It is rational to want a foundation and coherence for what we believe and claim to know. And the attempt to define knowledge arises from our desire to establish its foundation. For example, the JTB definition asserts that knowledge is founded on true propositions: S knows that p if and only if p is true.

 

The curious thing is that the JTB definition of knowledge both acknowledges and proceeds to ignore what really matters: the objectivity of knowledge expressed by means of true factual assertions. We can know that factual assertions such as the earth orbits the sun are true.

 

But those assertions are linguistic expressions. So their foundation is in – and their coherence comes from the consistency of – our linguistic practices. Our knowledge is not linguistic. But its linguistic expression, including what we say about knowledge, can be nothing other than linguistic.

 

In epistemology, classical foundationalism has involved claims that knowledge is founded on, for example, experience (empiricism) or reason or thought (rationalism). And the traditional distinction between a posteriori and a priori knowledge reflects such ideas.

 

But talk about knowledge is just that: talk. Language is so potent and pervasive in our lives that we can forget what it is and mistake what we say about things for the way things are.

 

In philosophy, we cannot get beyond language through language. (In real life, language is not normally in the way.) But we have tended to ignore the role of language in all our explanations, including our explanation of knowledge – hence the myth of propositions.

 

So here is an outline of what could be called linguistic foundationalism.  

 

Linguistic foundationalism – an outline

 

1  When we say, write or otherwise express anything, we use words or other signs. And they are real things, such as sounds, marks on paper or gestures.

 

2  Among the many things we do with language, we make factual assertions about features of reality. But such assertions are not the only kind that matter to us.

 

3  We name, describe and quantify features of reality as we talk about them. They do not name, describe and quantify themselves. (People are an exception.)

 

4 There is nothing propositional about the existence and nature of things. Reality is not linguistic. The truth is not out there any more than falsehood is.

 

5  Logic deals with language, not reality. Other discourses deal with reality, such as the natural and social sciences. But the rules of logic apply to all uses of language.

 

6  A factual assertion is true if it correctly asserts a feature of reality, given the way we use the words or other signs involved. And we call true factual assertions facts.

 

7  Facts constitute the objective knowledge we express in language. But they are nothing more than linguistic expressions, grounded in our linguistic practices.

 

8  We build and repair this knowledge on foundations and with materials of our own making. But that does not mean the edifice must be shaky. That we can always say more does not mean we can never say enough.

 

 

Peter Holmes

 

July 2017

Amended, January 2018

Amended, May 2018

Amended, June 2018