‘All’ traffic

Next exhibit in my mini-campaign against absurd road signage.


This is the sign that you see if you are travelling in the opposite direction on the very same road as the sign in the last similar post.

Where to begin?

Well, start with the obvious: ‘All’ traffic? Really? What weird idiosyncratic meaning of the word ‘all’ has the authority dreamed up here? The sign says that you can go left. So it isn’t true that all traffic has to go straight ahead – not unless ‘all’ has a special new meaning. Or is there a new and technical sense of ‘traffic’ that they have secretly invented? Is ‘traffic’ now to denote only vehicles travelling on roads that are not at some reasonably proximate distance subject to restrictions affecting motor vehicles?

Or is it – just possibly – that they cannot think, or would not think? (I find it secretly pleasing that this sign seems to confirm my suspicion about the ‘other traffic’ sign that what is really at work is an officious and slightly spooky desire on the part of the council to control free citizens’ behaviour more tightly than do the legal constraints that are in fact applicable to them.)

The added joy here is that, together with the large sign’s unnecessary complexity and inaccuracy/dishonesty, we have one of the appallingly numerous and completely unnecessary ‘traffic enforcement cameras’ warning signs that litter the city. Is anyone unaware that there are cameras all over the place? Where would you have to have been living for the last ten years to be unaware of this? Jupiter? The signs do nothing to affect driving behaviour; they bear little relation to the exact presence of cameras; and motorists don’t need, or deserve, to warned of the presence of cameras. It’s just another instance of the ugly thinking that seems to underlie so much of the hideousness:  ‘Why have just one sign on a post if you can fit another on there?’

(I can forgive the ‘Ring Road’ sign as the sort of thing that is potentially useful for visitors to the city. I am not against all signs: indeed it is a major part of my opposition to excessive, confusing, misleading, and excessively complex signage that it distracts attention from the useful ones.)

‘Eschatological’ verification

A rather different response to the Verificationist assessment of religious language as meaningless from those considered in the last post, and advanced on behalf of a theistic view, is an argument proposed by the English philosopher-theologian John Hick (1922-2012).

Hick’s idea is that we can respond to the Verificationist’s criticism of religious claims by insisting that they are meaningful, because they are after all verifiable, contrary to what Logical Positivists like A J Ayer assumed.

Hick himself adopts a version of verificationism, claiming that a statement is meaningful only if it is ‘factually significant’, and that a statement is factually significant only if its truth or falsity makes a difference to our experience.

Hick therefore holds that the sentence

There is an invisible, odourless, intangible, tasteless, and silent rabbit in this room

is meaningless: its truth or falsity would make no difference to our experience, which would be exactly alike whether it were true or false.

According to Hick, however, religious claims are verifiable; and hence they are meaningful.

They are verifiable, Hick says, by waiting till you’re dead, when (if the claims are true) you will be able to see for yourself that God and Heaven and angels and all the rest of it really do exist, just as the theist claims that they do. It’s true that the religious claims aren’t verifiable by us now; but since they are in principle verifiable in the afterlife, they cannot be dismissed as meaningless, even if the Verification Principle is correct. They are capable of eschatological verification.

(‘Eschatological’: from the Greek eschata = ‘last/final things'; ‘eschatological’ means ‘relating to the end of time, or the last things’ – this might just mean the afterlife, or it might mean something like the Last Judgement, for religious views that contain such a belief.)

Illustration: the Celestial City

The claim is illustrated by Hick in a what the literature has labelled the ‘parable of the Celestial City':

Two people are travelling together along a road. One of them believes that it leads to the Celestial City, the other that it leads nowhere, but since this is the only road there is, both must travel it. Neither has been this way before; therefore, neither is able to say what they will find around each corner. During their journey they meet with moments of refreshment and delight, and with moments of hardship and danger. All the time one of them thinks of the journey as a pilgrimage to the Celestial City. She interprets the pleasant parts as encouragements and the obstacles as trials of her purpose and lessons in endurance, prepared by the sovereign of that city and designed to make of her a worthy citizen of the place when at last she arrives. The other, however, believes none of this, and sees their journey as an unavoidable and aimless ramble. Since he has no choice in the matter, he enjoys the good and endures the bad. For him there is no Celestial City to be reached, no all-encompassing purpose ordaining their journey; there is only the road itself and the luck of the road in good weather and in bad.

As Hick comments:

… [W]hen they turn the last corner, it will be apparent that one of them has been right all the time and the other wrong. Thus, although the issue between them has not been experimental, it has nevertheless been a real issue. They have not merely felt differently about the road, for one was feeling appropriately and the other inappropriately in relation to the actual state of affairs. Their opposed interpretations of the situation have constituted genuinely rival assertions…

In other words, the thing that the theist believed was genuinely meaningful – she was saying and thinking something meaningful that was different from what the atheist said and thought; and as it turned out in the story, the theist’s view was right – she was ‘thinking appropriately’ in relation to the state of affairs that became apparent only at the end of the journey.

It is of course also true, and is admitted by Hick, that if the theist is wrong and there turns out not to be an afterlife, then nothing at all will be verified: you’ll just be dead. For the same reason, religious claims cannot be falsified by experience either: if there’s no afterlife, then you’ll just be dead and having no experience at all; you won’t be having experiences that confirm the falsity of the theist’s view.

Hick on post-mortem personal identity

Hick’s view that religious claims can in principle be verified post mortem in this way depends on its being possible for a person to continue to exist after his death. He has to believe that death does not necessarily mean the end of a person’s existence.

Death does indeed raise interesting questions about personal identity, in particular because of the decomposition and destruction of the body. If we are capable of still being in existence after death, and after the decomposition of our earthly bodies, then these bodies must be inessential to us. Hick seems to assume that post-mortem existence will involve being given a new body. And the task as he sees it is to justify the view that a person could retain his identity through such a change of bodies.

Hick tries to demonstrate that this is logically possible, by considering a series of thought-experimental cases.

Case 1. Imagine that a person vanishes in America, and that an exact duplicate of his simultaneously appears in Australia. The duplicate looks exactly like the original person, has apparent memories as of events in the original person’s life, and in every other way is psychologically just like the original person.

Everyone would say, Hick suggests, that the new person in Australia was the same person as the original person.

Case 2. As in Case 1, except that the original person dies, as opposed to vanishing.

According to Hick, the difference about what happens in the USA cannot mean that we must alter our judgement about the identity of the person in Australia. So here too, Hick thinks, we must say that the new person in Australia is the same person as the original person.

Case 3. As in Case 2, except that the duplicate appears in Heaven rather than Australia. (I thought they were the same place, but apparently this is an error.)

According to Hick, the difference about where the duplicate appears cannot mean that we must alter our judgement about the identity of the duplicate. So here too, Hick thinks, we must say that the new person in Heaven is the same person as the original person.


1. Do we really need this?

The need for the whole argument is highly questionable: I personally find it much more attractive and rational simply to reject the verificationism. If you abandon the verificationism, then obviously there’s no need to argue elaborately for the claim that the truth of the theist’s statements is verifiable after death. You can assert that the statements are meaningful without needing to show that their truth can in principle be verified.

2. If we do need it, then OK… maybe

But it seems right that, if you must accept verificationism, this is a decent way for a theist to maintain that her/his claims are not meaningless. And it seems to me true that if there is an afterlife, then it is not unreasonable to assert that we will indeed be able to verify this for ourselves by experiencing it. A possible objection is that, with no previous experience of Heaven or of God, it is not clear how we would be able to recognise these items as Heaven and as God. It may furthermore be that these things are beyond our comprehension in some way; that will only make the task of recognition harder. But, if these objections can be dealt with, then post-mortem verification of the existence of Heaven and God seems possible in principle.

3. BUT

The arguments concerning personal identity, however, are laughably poor. This is not to say that the conclusion is wrong; only that the arguments’ support for that conclusion is utterly feeble.

In Case 1, I would say, there is no remotely compelling reason to assert that the person in Australia is identical with the original. Anyone who judged that it was the original person would be making a forgivable mistake, at best.

Since the proper verdict in Case 1 is not remotely in line with Hick’s suggestion, the rest of the argument collapses.

What Hick could and should have said is this: if post-mortem existence is possible for us, then it must be that we have a nature such that our bodies are inessential to us: we are essentially immortal souls, which part from the body at death and somehow go to the afterlife, whatever and wherever that is. The fact that the person in the afterlife is identical with the person who died on Earth consists in the fact that it is one and the same soul that is both here and there, before and after death.

Verificationism and religious language

Some notes primarily written for F1 and F2, but blogged because also relevant for U1.


Verificationism is the best-known candidate theory of meaning according to which religious language would be revealed as meaningless. It is the central tenet of the Logical Positivism developed by the Vienna Circle in the 1920s, and was brought to the attention of English Philosophy by A J Ayer in his 1932 book Language, Truth and Logic.

Verification Principle (“VP”)

A factual statement is meaningful if and only if

either (a) it is a tautology (it is analytic)
or (b) it is empirically verifiable in principle.

Application of the VP to the case of religious propositions like ‘God exists’ yields the result that such statements are meaningless – at least if they are understood in a fairly literal and traditional way.

For example:

•    Insofar as we speak of God as transcendent – existing outside space and time – it seems that we are talking about something whose existence could not possibly be verified by any empirical method. This is because neither our senses nor our scientific probes can reach beyond our world of space and time.
•    Similarly, if the claim is that there is an afterlife in a realm called Heaven or Hell that is similarly supposed to be causally isolated from the world in which we live. Nothing (or nothing that we can experience in this life) could count as empirical verification of the existence of such a world.

•    Similarly, if the claim is that we have immortal immaterial (non-physical) souls, or if the claim is that God, or angels, or any part of the Creation is immaterial. No such entities could be detected by our senses or our science.

Since the religious claims here are propositional in form, they are factual statements. But they are neither analytic nor empirically verifiable; therefore, according to the Verificationist, they are meaningless.


What is the attraction of the Verification Principle?

Well, what motivates it is really two thoughts:

1. If a statement is factual and meaningful, then there must be a real difference in the world between its being true and its not being true. The world is different, in some respect, if it is true, from the way the world would have to be if it were, instead, false.

Now, that thought is an excellent and persuasive one. If the world would be no different whether a statement was true or false, then, surely, that statement isn’t a factual statement about the world. It seems correct that the mark of a meaningful factual statement is that things would have to be one way for the statement to be true, and another – different – way for it to be false. To say that is to say that the truth, as opposed to the falsity, of a genuinely meaningful factual statement would involve a difference in the world – a difference in the ways things were.

2.The second, and more controversial, claim is the claim that scientifically-detectable facts about our world are the only facts that there are; and (?‘hence’) that the only things that we can meaningfully talk about are matters describable by science.

This seems to me much more dubious. First, we surely don’t know that there aren’t facts undetectable by science. Secondly, why can’t the meanings of our words stretch beyond what we can detect to be true or false? After all, we are very adept at manipulating concepts in our thought, by combining and relating and abstracting from simple concepts. We can accordingly construct complex concepts like immaterial realm causally disconnected from our world of space and time. But if we can construct such concepts, then why deny that we can create meaningful statements involving such concepts, even if we cannot tell whether such statements are true?

Objections to verificationism

1. Implausible results in certain areas of Philosophy

Verification leads to implausible results concerning certain subject-areas, including ethics and other minds.

In ethics, verificationists are forced (on pain of classing the whole of ethical discourse as meaningless) to adopt one of two views:

(a)    The ‘naturalist’ view that ethical facts are just facts about what is conducive to human happiness or well-being. This places them within the field detectable by science; but it is far from obvious that this is the best, or even a very plausible, moral theory.
(b)    The ‘non-cognitivist’ view that the meaning of ethical statements is not factual, but ‘expressive’ or ‘prescriptive’. Moral judgements like ‘torturing cats is wrong’ are not really statements of alleged fact, but disguised expressions of the speaker’s attitude, or disguised imperatives or exhortations to do or not to do certain things.

As to other minds, verificationism leads to the implausible ‘behaviourist’ theory that a person’s being in some mental state amounts to no more than his being disposed to behave in certain ways in response to certain stimuli. This is what verificationism has to do, in order to make the existence of other people’s thoughts verifiable by empirical means. Thus:

•    The fact that Jones believes that it is raining amounts to no more than a set of facts about Jones’s actual and possible behaviour: the facts, for example, that, if he goes out, Jones will carry an umbrella (other things being equal); that if he is asked what the weather is like, he will reply (other things being equal) that it is raining; and so on.

This deeply silly theory is now generally regarded as fatally discredited, not least because it seems to leave out what is really central to mentality – namely, what it is like for Jones: how things seem to him within the private, inner, mental, world of his thoughts and feelings.

2. Intellectual arrogance

Verificationism puts all meaningful factual statements within our epistemic reach – in principle, if not  in practice. We can in principle know everything that is a factual matter. But why think this? Why are we entitled to assume  that the world must be such that every aspect of it is accessible to us by empirical means?

Verificationism conflates factuality with our ability to tell whether something is the case. Factuality should (I think) be thought of, instead, simply in terms of truth-conditions: if a statement is factual, then it has truth-conditions: there are ways that the world would have to be if the statement were to be true. But there is no need – intellectual arrogance aside – to assume that we can always in principle tell whether the conditions are satisfied, for literally any and every factual statement. We have (I think) no right to assume that the question whether a factual statement is true or not is one that we must, in every case, be capable in principle of answering.

Verificationism is, as a result, unjustifiably arrogant about our intellectual capacities, and about the capacity of scientific methods to detect all factual matters. Surely, if we are intellectually modest, we should admit that there may be factual matters that are beyond our capacity to verify. Why think that our senses and our science must in principle be capable of detecting everything that is capable of being the case?

3. Intuitively wrong results

Many statements that the VP classes as meaningless don’t intuitively seem to be meaningless.

There is a wholly undetectable, because immaterial, elephant in the room.

There is an immaterial God who transcends time and space and who created the world.

Again, these seem (arguably – it’s a bit controversial) meaningful, at least in that we think that we can understand what would have to be the case for them to be true, even though nothing in our experience could confirm their truth.

4. Difficulties with universal statements

Statements of things like physical laws take the form of universal statements. For example:

•    All gases expand when heated

But it is obviously impossible to demonstrate their truth: it would be an endless task. When would you have finished verifying that every gas expands on every occasion of its being heated?!

5. Hollow victory over theism

Verificationism dismisses religious language as meaningless. But even from an atheist perspective, this is not really the desired result: it amounts to a rather hollow victory, because, since all talk of God is meaningless, if VP is true, the denial that God exists is just as meaningless as the assertion that He does exist.

Surely what we really want, if we are atheists, is to be able to say that we understand the theist’s claims all right; the problem is that there are powerful objections that suggest that the claims are not meaningless, but false.

6. Poorly motivated

Verificationism is poorly motivated, in that it is driven by the rather dubious assumption (see above) that scientifically-detectable facts about our world are the only facts that there are.

7. VP itself meaningless, by its own lights
Finally, there is a standard textbookish criticism that the Verification Principle doesn’t pass its own test of being meaningful.

The Verification Principle is not analytic – you cannot tell that it is true merely by reflecting on the meanings of the terms ‘meaningful’, ‘statement’, etc; so it must be synthetic.

If it is synthetic, then, by its own lights it is meaningful only if its truth can be empirically verified.

But how could the truth of the principle be empirically verified? You would have to identify all the meaningful statements, and check that all of them (and no others) were empirically verifiable. But this is impossible, for at least two reasons:

(a) To identify the meaningful statements, we would need a criterion of meaningfulness in order to do so. If the test were the one in the VP itself, then the exercise would be a pointless one. But if it were another criterion, then that criterion would be a competitor with the criterion given by the VP.

(b) Even if problem (a) could be overcome, the task would be impossible, for there is a potentially infinite number of meaningful statements.  But there is no possibility of empirically verifying an infinite number of statements.

So the VP fails its own test, and is therefore meaningless. It therefore cannot be true.

Magnificent dinosaur

Not Philosophy but there is a wonderful creature currently to be seen eating the multi-storey car park. It is very exciting.


Has the Council finally made a good planning decision? We shall see.

(While taking this photo I met a man who expressed sadness at the demise of what he called this ‘iconic’ building. I laughed in his face.)

Referential opacity


We said that, on a possible interpretation of paragraph 6 of the Second Meditation, Descartes offers there what amounts to an argument for the conclusion that he is not his body.

That argument trades on Leibniz’s Law (also known as the principle of the indiscernibility of identicals) – the principle that a thing and itself have all ‘their’ properties in common. If a is the very same individual as b, then a and b must be exactly alike in all respects. And that is dead obvious: after all, ‘they’ are not two things but one thing; and one thing cannot simultaneously be both this way and that way in the same respect. A thing cannot, for example, be both green all over and red all over, or both cuboid and spherical in shape.

Leibniz’s Law, though completely obvious, is a useful rule that we apply subconsciously every day. For example, if I have forgotten exactly where I parked it, I can pick my bicycle out of the crowd of bicycles not merely by looking for a bike with its features, but also by differentiating it from others: by spotting features of other bikes that I know my own bike does not share. My bike is green; so in scanning the racks I can instantly dismiss all the non-green bikes. My bike hasn’t got mudguards; so I can dismiss all the bikes with mudguards. And so on.

In doing so I can be thought of as subconsciously running arguments of the following kind through my head:

(1) My bike is green

(2) That bike is not green


(3) That bike is not my bike

My bike has a property – being green – that that bike lacks. So that bike is not my bike.

The argument that Descartes presents (if he is to be interpreted as so doing at all) is of similar form:

(1) I can coherently entertain doubt about whether my body exists.

(2) I cannot coherently entertain doubt about whether I exist.


(3) I am not my body.

That is: My body has a property that I do not have: It has the property of being able to be doubted by me. I lack the property of being able to be doubted by me. So I and my body must be two different things.

The fallacy

Descartes’s argument – unlike the one about my bike – is clearly fallacious. It is an instance of what is known as the ‘masked man’ fallacy – so called because it can be presented in a standard way as follows: imagine that you have witnessed a bank robbery by a masked man, and that in considering the possible identity of the robber you reason thus:

(1) I don’t know who the masked man is.

(2) I know who my father is.


(3) The masked man is not my father.

In doing so, you are applying Leibniz’s Law as before: you are reasoning that your father has a property (the property of having his identity known by you) that the masked man lacks; and concluding that your father and the masked man must therefore be different individuals.

But if you reason like that, you are making an obvious mistake: for it could be, for all you know, that it was your father who was robbing the bank. (It would hardly be surprising if he was, given the level of the school fees…)

The explanation of the inapplicability of Leibniz’s Law in these cases is that the principle cannot be applied where the properties on the basis of which a distinction is made are properties that involve reference to a person’s intentional states (doubting and knowing, in the two cases above).

You get a similarly fallacious result if you try to apply Leibniz’s Law on the basis of a difference in modal properties (properties including reference to necessity and possibility), as shown by the following fallacious argument:

(1) Michael Clarke is necessarily Michael Clarke. (True: everyone is necessarily self-identical).

(2) Michael Clarke is not necessarily the captain of the Australian cricket team. (True: the ACB could have appointed someone else)


(3) Michael Clarke is not the captain of the Australian cricket team. (False: he is)

Alleged difference between Michael Clarke and the captain of Australia: one of them but not the other has the property of being necessarily identical with Michael Clarke. But again, the argument is clearly fallacious: the premisses are true, but the conclusion is false.

OK but which properties? How to identify them?

The interesting question is this: What is the mark of these special properties that mean that you cannot apply Leibniz’s Law?

In most sentences, you can replace one name for a person or thing with any other name or definite description that refers to that same person or thing, and the truth value of the sentence will remain unchanged. We say that the different names or descriptions can be substituted, one for another, ‘salva veritate‘ (Latin: literally, ‘with the truth intact’ – i.e. without affecting the truth or falsity of the sentence).

(a) David Mackie is underpaid.

(b) The Philosophy teacher is underpaid.

(c) The father of Nicolas and Andreas is underpaid.

(d) Phoenix’s owner is underpaid.

Given that (a) is true, then (b), (c) and (d) are all equally true. And if (a) is false, then so are (b), (c) and (d). All the underlined phrases refer to the same individual, and using one name or description rather than another makes no difference to the truth-value of the sentence.

Where names and definite descriptions can be interchanged salva veritate, we say that the context in which the name/description appears is ‘referentially transparent’.

In other contexts, you cannot substitute salva veritate in the same way:

For example:

(e) Jones believes that David Mackie is David Mackie.

This is true. Jones would have to be an utter moron not to believe that: even if Jones had never previously heard of me, he would still believe it, because it’s obvious that such statements of identity that repeat the name of an individual must be true.


(f) Jones believes that David Mackie is Phoenix’s owner.

Sentence (f) may be false, even if (e) is true. If Jones doesn’t know that I own a thing called Phoenix, then he won’t believe that I am Phoenix’s owner; and so (f) will be false. So the substitution of ‘Phoenix’s owner’ for ‘David Mackie’ makes (or may make) a difference to the truth value of the sentence. You cannot substitute salva veritate.


(g) It is impossible that David Mackie should be someone other than David Mackie.

Obviously true: no one can be anyone other than the person that (s)he is.

(h) It is impossible that Phoenix’s owner should be someone other than David Mackie.

Obviously false: I could sell Phoenix.

Sentence (g) is true, but sentence (h) is false. But the only difference is that, once again, we substituted the definite description ‘Phoenix’s owner’ in place of ‘David Mackie’. So in this context too, you cannot substitute salva veritate.

Contexts in which you cannot substitute salva veritate are labelled ‘referentially opaque’. As you can see in the examples above, intentional contexts ‘Jones believes that…’ and modal contexts ‘It is impossible that…’ are the culprits.

Why does this happen?

Well, the point about the referentially opaque contexts is that they are contexts in which whether what is said of some individual is true or false depends on the way in which the individual is referred to – the truth or falsity of the sentence depends on how the individual is named or described.

The properties with which you cannot invoke Leibniz’s Law to demonstrate a non-identity are precisely the properties that create referentially opaque contexts.


All of the above is just a technical and fussy way of spelling out what is very obvious anyway, at least insofar as the Cartesian argument (if that is what it is) is concerned.

This is because, of course, Descartes is in no position to deny the possibility that his thought may in fact depend on his possession of a body: for all he knows, thinking requires the brain (as we all believe). The fact that he is unable to doubt his own existence, but at the same time apparently able to doubt the existence of his body, reveals only that he is ignorant of the essential link between thinking and having a body. If he knew all the facts, including the fact (if it is a fact) that thinking requires a brain, then he would no longer be able to doubt the existence of his body, given that he cannot doubt his own existence.

Jim and Jemima

Provoked by events today, a few more notes in the hope of clarifying this theory of knowledge as requiring true belief plus ‘epistemic virtue’.


It needs to be clear in your mind that this is just one among a whole series of rival theories, each aiming to offer a watertight analysis of what it is to possess propositional knowledge (to know that p, where p is some proposition.)

This part of the syllabus – the analysis of propositional knowledge – aims to give you an insight into the main rival theories that have been advanced, reflecting the fact that the recent history of this part of epistemology is one involving the sequential introduction of a series of theories, each of which has been introduced because it was supposed to be an improvement on its faulty predecessors.

The aim of such theories is to specify the necessary and sufficient conditions for knowing that p.

That is to say: each proposed theory claims that unless you satisfy all of the conditions that it specifies, you don’t know (each condition is necessary for knowledge); and that if you satisfy all the conditions, then you do know (the conditions are jointly sufficient for knowledge).

Each theory is tested by intuition: a good theory is one that gives the intuitively right answer in all imaginable cases. Theories are rejected when its is shown by this method that they give the wrong answer in some imaginable case or other.

Thus, for example:

The old tripartite (TB+J) theory says that a person knows that p iff (s)he has a true, justified, belief that p.

Gettier is generally taken to have shown that that theory is not good enough: there are imaginable cases in which a person has a true, justified, belief that p, but does not know that p. So the conditions laid down by the TB+J theory are not sufficient for knowledge.

So the old TB+J theory got refined: some people suggested the addition of a ‘no false lemmas clause’.

But there seemed to be imaginable cases where that revised theory didn’t give the right answers either. For example, the fake barns case: the subject there has a true and justified belief, not obviously inferred from any false belief, that she is looking at a barn; but the subject still doesn’t know that she is looking at a barn.

Back to the drawing board: another new theory gets proposed; some clever person comes up with what seems to be a counterexample. And so it goes on.

What this means for you is that if you are ever asked to talk about the TB + EV account of knowledge, you need to state matters in a way that at very least makes it clear that and why this theory is an improvement on the tripartite (TB+J) account.

What is ‘epistemic virtue’?

The fundamental idea behind this theory is that when we say that someone ‘knows’ something, as opposed to saying merely that they believe it and that it is true, we are saying that they deserve particular credit for believing the truth.

Why would such credit be deserved?

This is where you need to be careful to be absolutely precise.

Case 1 – randomly true belief

If you form your beliefs at random – just arbitrarily deciding to believe this or that, with no regard to any evidence that there may be – then clearly you deserve no credit at all for being right on the occasions when you are right. You are no better than a gambler at roulette who happens to have picked the winning number on some spin of the wheel. The gambler didn’t know that that number would come up. He was just luckily right.

Case 2 – the scam

More deserving of at least some epistemic credit (though not, perhaps, of moral credit) is a gambler who has a particular reason for thinking that his chosen number will come up. Consider, for example, a gambler (David) who has taken advantage of a special friendship that he has with the man (Jim) who makes the roulette wheels for the local casino. These two old friends, each of whom would do anything for the other, have agreed that Jim will ‘fix’ a particular wheel in such a way that, when David bets on the number 27, the ball will be caused to fall into the desired slot. The required technology is tested and found to be perfect in its operation. David bets on 27, believing (because of the agreement and because of the evidence that the scheme and the technology have worked in testing) that 27 will win; and it does. David had a true justified belief that 27 would win. Arguably, he also knew that it would win.

Case 3 – Gettier variant

But true justified belief alone is not enough for knowledge, as a Gettier-style variant on the same case shows. Suppose that things are as above, except that, unknown to David, Jim has fallen hopelessly in love with the girl (Jemima) who spins the wheel at the local casino. Jemima has promised to run away to France with Jim if he will fix the wheel in such a way that the ball will fall into whichever slot she chooses on some particular occasion. Jim’s hopeless infatuation with Jemima leads him to fall in with this new proposal and, abandoning any sense of loyalty to his old friend David, he therefore sets up a new secret mechanism that makes the roulette wheel responsive to Jemima’s wishes only. He completely neglects to tell David of these new arrangements. Such is the power of love.

David enters the casino and bets all his money on 27. The wheel is spun, and the ball lands on 27. (Hurrah!) David had a true and justified belief that it would do so, for the same reasons as before. But in fact, David was just lucky. The ball didn’t fall into the right slot because of the arrangement that David had with Jim, but only because the wheel had been fixed to respond to Jemima’s choices. David merely had the very good luck that Jemima happened also to bet on the number 27. Even if he knew what would happen in Case 2, David clearly didn’t know in Case 3 that the number 27 would win.

(That’s how easy it is to invent Gettier cases.)

So what and where is ‘epistemic virtue’?

It is against this kind of background that we need to view the TB + EV theory. There must be a difference between Case 2 and Case 3 that explains the fact that in Case 2, David knew that 27 would win, but in Case 3 he clearly didn’t.

David can’t reasonably be criticised for believing what he believed in Case 3. He formed his belief on certain strong evidence that he had – evidence consisting (as in Case 2) of knowledge of the scam plotted with Jim, evidence of the reliable technology invented for that scam, and evidence of Jim’s longstanding friendship and loyalty, which led him to believe – reasonably – that the scam would be successfully implemented. His belief that 27 would come up was undoubtedly justified, and this is an awful lot better than believing for no reason at all that 27 rather than any other number will come up. So David deserves some epistemic credit. His belief is not based on stupid reasons, or no reasons, like that of the random gambler in Case 1.

But this is not to say that David’s belief in Case 3 has ‘epistemic virtue’ in the technical sense required by the new TB + EV theory. That new theory is meant to be an improvement on the tripartite (TB+J) theory that seems to be refuted by Case 3 above, in which David has a true, justified, belief, but one that is too luckily right to count as knowledge. So ‘epistemic virtue’ must mean something special – something that goes beyond the mere justification that David had in the case described.

What more is required for ‘epistemic virtue’, beyond mere justification?

Well, the idea is that your true belief also has epistemic virtue if the reason why your belief is true is that you have used the justifying evidence available to you.

This cannot be said of David, in Case 3. His belief is true; but it is true only because of the amazing good luck that Jemima also happened to choose to bet on the number 27, out of the 37 options available to her. In particular it could not be said that David’s belief is true because of the arrangements with Jim, knowledge of which constitutes David’s justification for his belief.

So epistemic virtue is a very specific technically-defined phenomenon: it is a matter of owing the truth of your belief to the quality of the reasons that you have for believing it. You need to be right because of the care you take in forming the belief. If we count David as knowing in Case 2, it is because we believe that it is thanks to his knowledge of the precautions that he has taken to ensure that 27 comes up that he is right in thinking that it will.