Does religion involve belief?
Belief in what? In religious propositions. What are they? Roughly: claims about events or states of affairs that constitute or reveal the meaning or purpose of life or the universe according to some religious narrative.Footnote 1 These range from the highly specific to the highly general. As an instance of a specific claim, the Nicene Creed affirms, for example, that Jesus is the only son of God, was born to the Virgin Mary, and was crucified under Pilate before going on to do other things. As an instance of a more general one, Tim Crane identifies a belief that he regards as common to all religions.
According to William James, ‘Were one asked to characterize the life of religion in the broadest and most general terms, one might say that it consists in the belief that there is an unseen order, and that our supreme good lies in harmoniously adjusting ourselves thereto.’ This belief is what I will call the religious impulse. (Crane (Reference Crane2017), 35)
If James and Crane are right then at least one belief is common to all religions, so all religions involve beliefs. But even if no one belief is common to all religions, each may still involve a belief in this or that religious proposition. It is hard to question the claim that at least one follower of at least one religion believes a religious proposition. But I want to raise doubts, from a Wittgensteinian perspective, over the claim that all or even very many followers of all or even very many religions believe any religious propositions.
Religious belief: for and against
There are two strong reasons to think that most followers of at least the main monotheistic religions, and perhaps also of some others, are believers in religious propositions.
First, religious people themselves avow religious beliefs. Setting aside ritualistic formulations like the Creed, it is natural for most Jews, Christians, and Muslims when asked in everyday settings to describe themselves as believing in propositions that distinguish them if not from one another then at least from the irreligious.Footnote 2 The same goes for some other religions. According to one of its own spiritualists the Dagara people of Burkina Faso
are well known throughout West Africa for beliefs and practices that outsiders find both fascinating and frightening. The Dagara connection with beings from the Spirit World has resulted in the accumulation of first-hand knowledge of subjects regarded in the West as paranormal, magical, or spiritual . . . What in the West might be regarded as fiction, among the Dagara is believed as fact, for we have seen it with our eyes, heard it with our ears, or felt it with our own hands . . . A Westerner will say, for example, that water always makes you wet, yet a native healer who gets into a river and stays for hours doing what healers do might get out just as dry as if he had been working in the Sahara Desert. (Somé (Reference Somé1999), 2, 11; my italics)
The writer could hardly be clearer that he is describing not only the practices of his people but their beliefs. Here and elsewhere there are strong testimonial grounds to think of religious phenomena as typically doxastic.Footnote 3
Second, religious beliefs explain what would otherwise look inexplicable. This follows from a more general principle. At the heart of common-sense or ‘folk-psychological’ explanations of behaviour are the beliefs and desires of those who are doing the behaving. Why did I bet on this rather than that horse? Because I was more confident it would win. Why did you cross the road? Because you wanted to visit the bank and thought it was open. Why did Luther not recant at the Diet of Worms? Because he believed what he wrote. These and many other acts threaten to seem unintelligible if we can't explain them as effects of what people think and want. This goes for religious behaviour as much as any other.Footnote 4
When ordinary people pray it is because they think there is a God up there listening. But whether or not there is a God listening to their prayer isn't itself part of the language-game. The reason people play the language-game of religion is because they think there is something outside the language-game that gives it a point. (John Searle quoted in Magee (Reference Magee1987), 344–345)
Religion very often involves a range of behaviours – pilgrimage, fasting, giving alms, prayer, and so on – that naturally fit into the framework of some set of beliefs that support them.
But there are also arguments that most and perhaps all followers of most and perhaps all religions do not believe any religious propositions. One such argument, due to Rodney Needham, has urged on Wittgensteinian grounds that the concept of belief picks out no single feature of experience but rather a vague and disputable multiplicity of characteristics; and it infers that we ought to discard it altogether (Needham (Reference Needham1972), ch. 7). There is no such thing as belief and a fortiori no such thing as religious belief. I grant the premise but not the conclusion. A concept that has borderline cases or picks out no one common feature that we can describe in other terms needn't be deficient. If anything, it is essential, for example, to our use of chromatic concepts that they admit borderline cases (Wright (Reference Wright1975), 330); it may be essential to our use of ethical concepts that they admit contested cases (Gallie (Reference Gallie1956) ). Why should it not be similar with belief?Footnote 5
Another argument, visible in the work of Wilfred Cantwell Smith, distinguishes faith from religious belief, and observes that whereas belief may be the main expression of faith, it need not be.
The historical fact is that faith is expressed in a great variety of ways, and that in the Christian case one of the primary and basic expressions has been conceptually, in propositional doctrine; but in other traditions it has been expressed primarily and basically in other forms. What theology is to the Christian Church, a ritual dance may be to an African tribe: a central formulation of the human involvement with final verity. (Smith (Reference Smith1979), 14–15)
Even within Christianity we might expect to find more involvement with doctrine among, for example, Catholics and Presbyterians than among Episcopalians or Quakers.
But if you say that, then you are obliged to spell out what kind of thing counts as ‘faith’, as ‘the human involvement with final verity’. What distinguishes somebody who has faith but not belief from an atheist who deeply enjoys and regularly (‘religiously’) takes part in the ceremonies and rituals of the religion in which she was brought up? And how on this view, or on Needham's view, do we get around the two kinds of evidence on the other side – that most religious people themselves say that they believe, and that supposing that they do explains their behaviour? There is much more to be said for those arguments. But this is not the place to do it: here I mention them only to distinguish them from mine.
It is also worth distinguishing my argument from two kinds of ‘non-cognitivism’ that people sometimes attribute to Wittgenstein or to the ‘Wittgensteinian fideists’ (Cottingham (Reference Cottingham, Glock and Hyman2017), 644–646). I shan't argue that religious ‘belief’ is not belief because religious propositions are not propositions – that is, they say nothing true or false, hence nothing that anyone might believe. On the contrary, I assume that ‘religious propositions’ say exactly what they seem to say and are all either true or false. The question is not whether they are true or false but whether all or many religious people believe they are true. Also, I shan't argue that religious propositions say something so fundamental that they cannot be objects of belief but are rather elements of a ‘world-picture’ (Vasiliou (Reference Vasiliou, Arrington and Addis2001); for discussion see Graham (Reference Graham2014), 60–65). I don't think anything about religious propositions stops them being objects of belief; I just suspect that for the most part they are in fact not.
My argument for that draws on the same Wittgensteinian considerations as does Needham's; but it differs in drawing less sweeping conclusions and by suggesting a reinterpretation of the evidence for religious belief. The next section sets out the underlying conception of belief. The subsequent two sections indicate how it might apply in a cross-cultural religious context. The last section returns to the evidence for belief from (a) first-personal testimony and (b) our explanatory ambitions.
Wittgenstein and belief
One way of making vivid the novelty and interest of Wittgenstein's later views about belief is by contrasting them with his earlier ones.Footnote 6 The contrast is not only biographical. Despite occurring in the context of a highly elaborate account of logic, the early account of belief, like the early account of meaning, presents a very natural and compelling vision of its subject matter. According to this natural idea, whether one believes that things are a certain way is in all cases determinately settled; and what settles it is whether one is in possession of an inner representation of them as being that way.
Wittgenstein's early views about belief are best understood against the background of his overall theory of meaning. He thought that thought and language rested on the possibility of analysing all propositions of ordinary language into logical compounds of atomic propositions. These atomic propositions combined names of atoms or ‘objects’. The atomic proposition describes things as being a certain way. And it is true if the objects that it mentions combine in the same way as their names combine in the atomic sentence (Wittgenstein (Reference Wittgenstein, Pears and McGuinness1961), §2.15).
For instance, suppose that the fact in question is that a spatio-temporal point x contains matter. This fact relates two objects: the particular point x and the property of being filled with matter, their relation being that the former instantiates the latter. We describe this fact by putting their names in the same relation. To this end we denote objects of a logical type by objects of the same logical type: particulars denote particulars and properties denote properties. We might use the particular a to denote the particular point x and the property of being to the right of F to denote the property of containing matter. In the atomic sentence Fa, which is also a fact, the denoting particular and the denoting property fit together in the same way as the represented particular and property: the former instantiates the latter. This is the picture theory of meaning.Footnote 7
Wittgenstein's early theory of belief applies the picture theory to ‘inner’ states. He writes: ‘It is clear . . . that “A believes that p”, “A has the thought that p”, and “A says p” are of the form ” ‘p’ says p” ’ (Wittgenstein (Reference Wittgenstein, Pears and McGuinness1961), §5.542). By ‘ “A believes that p” has the form ” ‘p’ says p” ’, Wittgenstein means the following. For a thinker to believe that p is for there to occur in his mind, perhaps in his brain, a fact that pictures what p describes. For instance, for A to believe that point x is filled with matter is for the relevant elements of A’s mind or brain to be arranged into a fact depicting that possibility in the sense of the picture theory. ‘[I]f we are told “A believes that p” we are shown what A believes by being told what pictures occur to him’ (Mounce (Reference Mounce1981), 86).Footnote 8
What this means in detail is something of a mystery, but two things are clear enough. (a) An inner representational state achieves its status as a belief independently of whatever outer behaviour it motivates; it is rather a matter of the pictorial form of the inner state itself. (b) Whether any inner state achieves that status, and so whether anyone counts as believing something, is in all cases settled determinately. Either the picture occurs to him, or it does not.Footnote 9
Philosophical Investigations presents in somewhat scattered form a general philosophical psychology that undermines both of (a) and (b). Before stating this, it is worth making three general comments about what explaining a psychological concept involves, according to Wittgenstein's later philosophy.
First, there may be no adequate explanation in terms of necessary and sufficient conditions for the application of the concept. This may be because there are no non-trivial necessary and sufficient conditions at all; but even if there are, they may not be what make us apply the concept to all of its instances. In other words, psychological concepts are in many cases family resemblance concepts (Wittgenstein (Reference Wittgenstein, Anscombe, Hacker and Schulte2009), §65). Failure to recognize that is one pressure that drives us to postulate an inner realm of psychological facts for which no pattern of behaviour could ever be necessary and sufficient.Footnote 10
The second point is that introspection is not a way to understand psychological terms. This is for two reasons. First, an inner act of attending to a certain kind of experience does nothing to constrain your future use of the concept-word that you are trying to understand (ibid., §258). Second: that first-personal understanding of the concept, even if it were possible, would be impotent to generate a third-personal understanding of it. For as Wittgenstein argues in the context of pain, it is not a matter of isolating my own experience and supposing that for another to share my belief is for him to have ‘the same inner experience’ as I am now having (ibid., §301; see also Kripke (Reference Kripke1982), postscript).
The third point is that adequate explanation of a psychological concept demands a statement of its outer criteria, these being distinguished from its symptoms (Wittgenstein (Reference Wittgenstein, Anscombe, Hacker and Schulte2009), §354). On a traditional reading of Wittgenstein, outer criteria for a psychological state are defeasible evidence for that state; but what is not defeasible is their status as evidence of that state. On the other hand, ‘symptoms’ only have contingent evidential bearing on what they are symptomatic of.Footnote 11 Thus crying out, for example, is associated with pain, though it may be faked. But crying out could not quite generally fail to be evidence of pain. If, for instance, we found that crying out was evidentially irrelevant to some supposed neural correlate of pain, say, C-fibre stimulation, we should conclude not that crying out isn't after all a sign of pain but rather that C-fibre stimulation is not. Crying out is thus a criterion of pain.
If we apply these ideas to belief, we get an analysis that is behaviouristic and open-ended. It is behaviouristic because the criteria for belief are what you are disposed to say and do.Footnote 12 This isn't the kind of analytic behaviourism that seeks to identify belief with a behavioural disposition or with a cluster of dispositions. Understanding ‘S believes that p’ is not a matter of identifying its truth-conditions but rather of knowing when it is defeasibly assertible: and this is a matter of the person's behaviour (including verbal behaviour).
Note the contrast with the Tractatus. There, belief was a matter of the presence of inner pictures of facts. Here, nothing of the sort is in question; whether a state of the brain or anything else counts as belief can only be settled on the basis of the person's behavioural dispositions. If we know these, then even God, had He looked into the subject's mind, could not have seen what the person believed any better than we could. Thus Wittgenstein's later account of mental states undermines feature (a) of the earlier one.
But the concept is also open-ended because no one thing is common to all circumstances in which somebody evinces a belief. Of course in the right context, saying that it will rain tomorrow is evidence that you think it will; but there may be no pinning down of all the ‘right’ contexts, and an open-ended variety of other sorts of behaviour might make us doubt the attribution, and might count as grounds that you don't believe it will rain. For instance, if you say that it will rain but show no signs of changing your mind in the face of clear evidence that it won't, or if you don't take your umbrella when you leave the house, or if you go out to water the plants tonight, or . . .
When we learn third-personal attributions like ‘S believes that p’, what we learn are these criteria. We grasp the family resemblance concept of a ‘criterion for attributing belief’, just as, when we learn the English word ‘game’, we grasp a family resemblance concept. Since these criteria have the diversity typical of family resemblance concepts, it can happen that all the relevant information is available, and yet leaves open the question whether the concept applies. The concept can therefore create puzzles that seem unanswerable not because they are profound but because they call for semantic stipulation.
Wittgenstein himself discusses some examples of this, of which two are instructive, the first because it is simple, and the second because it covers a mental state. First (Wittgenstein (Reference Wittgenstein, Anscombe, Hacker and Schulte2009), §80): I see a chair over there; but when I go to fetch it, it disappears from sight. ‘So it wasn't a chair.’ But a few seconds later I see it again, am able to touch it, etc. ‘So it was there after all, and the disappearance is some kind of illusion.’ But now it disappears again. What are we to say?
It is not that the answer is so deeply buried that we cannot see it. You can say what you please, as long as it does not prevent you from seeing the facts. The facts are: something chair-shaped appears at these times and disappears at those other times. Determining whether there really was a chair there is not filling any gap in our knowledge but just laying down the semantic law. The case being so unusual, our normal criteria of chair-existence do not cover it, so we need new rules at this point. ‘But do we miss them when we use the word ‘chair’? And are we to say that we do not really attach any meaning to this word, because we are not equipped with rules for every possible application of it?’ (Wittgenstein (Reference Wittgenstein, Anscombe, Hacker and Schulte2009), §80; I made the same point with this example in Ahmed (Reference Ahmed2014), 167–168).
A more complex case concerns the mental concept of ‘reading’, by which Wittgenstein doesn't mean understanding of what is read but only ‘the activity of rendering out loud what is written or printed; but also of writing from dictation, copying something printed, playing from sheet music, and so on’ (Wittgenstein (Reference Wittgenstein, Anscombe, Hacker and Schulte2009), §156). Humans learn some kind of reading very early on in life. But when do we start to do it?
Take the case of a pupil who has so far not participated in the training: if he is shown a written word, he will sometimes produce random sounds, and now and again the sounds will ‘accidentally’ come out roughly right. A third person hears this pupil on such an occasion and says, ‘He is reading’. But the teacher says, ‘No, he isn't reading; that was just an accident.’ – But let's suppose that the pupil continues to react to further words that are put before him. After a while, the teacher says, ‘Now he can read!’ – But what of that first word? Is the teacher to say, ‘I was wrong, he did read it after all’ – or, ‘He only began really to read later on’? – When did he begin to read? Which was the first word that he read? This question makes no sense here . . . Nor can the teacher here say of the trainee, ‘Perhaps he was already reading when he said that word’. For there was no doubt about what he did. – The change when the pupil began to read was a change in his behaviour; and it makes no sense here to speak of ‘a first word in his new state’. (ibid., §157)
can say if we like that ‘either the pupil was reading that word or he was not’; but the criteria are behavioural, and nothing in the pupil's behaviour settles whether he was reading.
The indeterminacy, over whether it was really a chair or whether the pupil was reading that first time, also arises for belief. That is, there are cases where a person's behaviour, including her linguistic behaviour, leave it unsettled whether she believes something – and not because we haven't observed enough behaviour but because it falls into a pattern on which the criteria are silent.
Imagine an observer who, as it were automatically, says what he is observing. Of course he hears himself talk, but, so to speak, he takes no notice of that. He sees that the enemy is approaching and reports it, describes it, but like a machine. What would that be like? Well, he does not act according to his observation. Of him, one might say that he speaks what he sees, but that he does not believe it. It does not, so to speak, get inside him. (Wittgenstein (Reference Wittgenstein1980), §813)
‘[O]ne might say’ that he does not believe that the enemy is approaching. Does he, or doesn't he? Well, we know what he sees, and what he says and does: does this amount to belief that the enemy is approaching? ‘You can say what you like, so long as this does not prevent you from seeing the facts’ and I have just said what the facts are – what he says and sees and does – and if these don't settle it, neither could ‘looking within’ at the pictures in his head.
The aim of this section was to highlight the difference between the account of belief that Wittgenstein endorses in the Tractatus and a second account that arises from the discussions of family resemblance concepts and criteria for mental state concepts that feature prominently in Philosophical Investigations. The central theses of this later account are as follows:
(i) Belief is not an inner state that exists independently of the subject's dispositions.
(ii) What justify third-personal attributions of belief as right or wrong are criteria covering behavioural (including verbal) dispositions.
(iii) These criteria may admit of borderline cases.Footnote 13
(i)–(iii) constitute the Wittgensteinian perspective from which I aim to cast doubt on the claim that being religious involves belief in religious propositions, either invariably or for the most part.
In the next section I'll describe two central criteria of belief: I'll call them evidentialism and penetration. Unlike the foregoing, the material from this point on builds on Wittgenstein's thought in ways that find relatively little inspiration in his actual writings. Certainly there is no attempt to claim that the following ideas are Wittgensteinian in anything more than the loosest sense.
The criteria of belief
Evidentialism says that what a person believes is sensitive to their evidence. ‘Evidence’ here means the kinds of inputs and processes that inform our own beliefs about other people's minds. This includes perception: if you see that I see a horse then you have a basic ground for attributing to me the belief that there is a horse before me. It includes memory and testimony. If you know that I saw a horse a minute ago then you have a basic ground for attributing to me the belief that there was a horse before me. If you know that somebody has told me that Desert Orchid won the Cheltenham Gold Cup this year then you have a basic ground for attributing to me that belief. It includes simple reasoning. If you know that I know from testimony that Desert Orchid won the Cheltenham Gold Cup this year, and by memory that Desert Orchid won the Grand National last year, then you have basic ground for attributing to me the belief that at least one horse has won both races.
In those examples I have what you would regard as evidence that p. Evidentialism also allows that what is evidence for me may not be evidence for you. This might happen, for example, if you think that I have superior discernment: my perception, memory and reasoning about the current condition of the track and the experience of the horses and riders gives me (but not you) evidence concerning the outcome of the next race; and you might treat this as a basic sort of ground for attributing to me at least a high degree of confidence about who will win the next race. Alternatively you might think me wrongheaded in treating this or that as evidence that p: what we commonly perceive may be evidence for me that Desert Orchid will win the race, but for you evidence that he won't. This is still grounds for you to attribute to me the former belief.
There is a limit to how wrongheaded you can regard me as being. A crazy pattern of dependence of what I say on what I see won't count as a relationship between belief and evidence at all. Highly relevant here is the systematicity of the relation. If I say that there was a horse in this field when I see the paw-prints of a dog then it might be reasonable to attribute to me both the belief that a horse was here and a systematic error about what counts as evidence for that. But if I say this on alternate Mondays depending on whether I see paw-prints, on alternate Mondays and Wednesday afternoons depending on whether there is heavy cloud cover, on Sunday between 1130 and 1145 a.m. depending on whether buttercups are growing in the field, and so on, then you have grounds for thinking not that I am expressing beliefs that respond to evidence in some highly deviant way but rather that I am playing some non-doxastic game. Still, there may be no sharp line between systematic and non-systematic dependence; nor therefore between belief and mental states that shade into it.Footnote 14
The second criterion, penetration, concerns not the coherence of what I see with what I say but of what I say with what else I do, including what else I say. My saying p is basic grounds for attributing to me that belief when accompanied by behaviour that it rationalizes, or at least in the absence of behaviour that it makes mysterious. Thus it is a criterion for my thinking that Desert Orchid will win that I am prepared to bet on it, or at least not to bet much on another horse. Of course the condition isn't necessary or sufficient: not necessary because I might be averse to gambling; not sufficient because I might want to lose money. Still, part of learning belief-attribution is learning what sort of behaviour a belief typically rationalizes.
More generally, if behaviour evinces a belief then the belief must penetrate the whole range of one's behaviour. Suppose a senior Politburo official affirms the viability of communism and the inevitability of revolution and works towards it at the office – but at home he expresses extreme scepticism about the whole idea. Nobody would say that he thinks communism is viable even when he is at the office – if so, the belief would penetrate his behaviour in ways that it actually does not. Of course there may be contexts where we should expect an agent who believes that p not to show it; for instance, even if the Politburo official really is convinced that communism is a failure, we'd still expect him not to do anything that reveals this when he is at work. But in the absence of special reasons to conceal the belief that p, overt behaviour that it rationalizes is criterial for the attribution of it.
Religious belief
Do their religious assertions express the beliefs of all or many of the religious people who make them? Settling that would require a cross-cultural empirical investigation that would far exceed the space available here. In any case we needn't expect a single answer to be valid across all traditions. Nothing would be less surprising than that different religious traditions meet different criteria at different times: we might find religious propositions that Christians once believed but no longer do, or propositions that in the present day some Jews believe and others do not, though all profess them. But I can indicate what grounds we might expect to find for saying that many modern religious communities and practitioners don't, universally or for the most part, believe religious propositions. These grounds concern the satisfaction or violation of the two criteria.
On evidentialism, distinguish two questions. The first is whether there is in fact evidence for religious propositions concerning, for example, miracles or the existence of God. The second is whether people's willingness to assert religious claims, and more generally to take part in religious life, is responsive to the evidence for those claims. The first has been disputed for centuries but is not relevant here. What matters is the second.
On this point, probably much religious engagement is not sensitive to evidence in the way you would expect if it were expression of belief. Crane, writing about the modern debate between the religious and the ‘new atheists’, says:
It will be obvious to all readers that apart from its bad temper, the most striking feature of this debate is its stagnation. The New Atheists pile argument upon argument against religion, and the religious are as unmoved by them as the New Atheists are unmoved by any part of the defence of religion. We don't really have a debate at all, in fact; just people talking past each other or shouting at each other. (Crane (Reference Crane2017), 33)
You might – as Crane does – take the fact that the religious are unmoved by these arguments to be evidence that their beliefs are insensitive to the evidence; but I think it suggests rather that what the religious are unmoved from isn't really belief at all.
This isn't to deny the possibility of wishful thinking – genuinely believing something either in the face of clear evidence against it or without caring what the evidence is. But it is to say two more moderate things. First: this attitude is prima facie grounds for denying that the subject's attitude is doxastic. These grounds could be overturned, for instance in the presence of penetration; but they need not be. Second: more extreme and flagrant violations of evidentialism are stronger grounds for thinking that the attitudes that violate it are not beliefs. If somebody asserts ‘It is raining’ on a plainly bright and cloudless day we should be inclined to think either that he means something other than that it is raining or that for him, assertion in this context is not the expression of belief; in either case, we do have strong grounds to think that he does not believe that it is raining.
Turning to penetration: again, we might easily find religious subjects whose behaviour does not seem to be penetrated by some religious or quasi-religious belief in a proposition that they affirm.
I read, among many similar examples, of a Rain-King in Africa to whom the people pray for rain when the rainy period comes. But surely that means that they do not really believe that he can make it rain, otherwise they would do it in the dry periods of the year in which the land is ‘a parched and arid desert’. For if one assumes that the people formerly instituted this office of Rain-King out of stupidity, it is nevertheless certainly clear that they had previously experienced that the rains begin in March, and then they would have had the Rain-King function for the other part of the year. (Wittgenstein (Reference Wittgenstein, Rhees, Klagge and Nordmann1993), 157)
It isn't too fanciful to suspect that many modern Jews, Christians, and Muslims exhibit a similar sort of disconnection between what they do and feel and what we should expect them to do and feel if they really believed what they profess. Perhaps they fear death more than we should expect, for instance, or perhaps they seek natural explanations for what previous generations would have regarded as miraculous. Such disconnections are perhaps what attract some people to the idea that religion is a separate ‘language game’ from, for example, scientific investigation or everyday life: observing that these and many other activities proceed quite independently, perhaps even at odds with, belief in the propositions that religious assertions are supposed to express, they have inferred that religious assertions do not in fact mean what they seem to mean. The alternative defended here is to deny not the default semantics for these sentences but rather that their assertion in these contexts expresses any belief at all.
I don't have the space here to prove, even if I could, that religious belief really is much less widespread than religion itself. The aim was only to sketch how things might look if it were, and to suggest that we should doubt the contrary claim that all or most followers of all or most religions really do believe the religious claims that they affirm.
Avowal and explanation
I turn finally to our initial motivations for attributing beliefs to the religious: that they say they have them and that it explains what they do.
On the first point: it is true that many religious people will affirm belief in some distinctively religious proposition. But what does this show? It is important not to think of belief on a model that Wittgenstein criticizes throughout his later writings on philosophy of mind. This is the ‘Cartesian’ model on which avowals of belief are reports of an inner state to which you have a kind of privileged access. But first-personal avowal of belief that p need not be based on observation of an inner state, but rather on whatever evidence supports p in the first place. It is, as Evans put it, transparent:
Wittgenstein is reported to have said . . .: ‘If a man says to me, looking at the sky, “I think it is going to rain, therefore I exist”, I do not understand him.’ . . . Wittgenstein was trying to undermine the temptation to adopt a Cartesian position, by forcing us to look more closely at the nature of our knowledge of our own mental properties, and, in particular, by forcing us to abandon the idea that it always involves an inward glance at the states and doings of something to which only the person himself has access. The crucial point is the one I have italicized: in making a self-ascription of belief, one's eyes are, so to speak, or occasionally literally, directed outward – upon the world. If someone asks me ‘Do you think there is going to be a third world war?’, I must attend, in answering him, to precisely the same outward phenomena as I would attend to if I were answering the question ‘Will there be a third world war?’ I get myself in a position to answer the question whether I believe that p by putting into operation whatever procedure I have for answering the question whether p. (Evans (Reference Evans1980), 225)
If that is right, then what underlies your avowal of the belief that p is not privileged access to that belief but evidence for p itself. But then your avowal that you believe p is no stronger grounds for our saying that you do than is your assertion of p itself. To be sure, this is grounds for our saying this, but it needn't be decisive: nothing can stop it from being outweighed by other sorts of behaviour, for instance your responsiveness to evidence or whether behaviour is what we should expect the belief that p to rationalize.
The second point concerns whether beliefs are explanatory of the things that people do. On the present conception of belief, it is misleading to think that their beliefs explain what people do. The reason is that we identify a belief by its behavioural manifestations. The appeal to a belief as the explanation of that behaviour is therefore at best only superficially explanatory. Saying ‘He did X because he believed Y’ is like saying ‘He did X because he is in a state of the sort that typically produces behaviour like X.’
An analogy might help get this across. Suppose we call something poisonous to humans if its ingestion typically causes certain physiological responses. We now observe some such responses in a subject. What explains this? Well, we might mention the ingestion of something poisonous. But that points to where the explanation lies. A fuller explanation would describe the chemical properties of what the person ingested and the mechanism by which something with these properties might cause such a reaction. But in this fuller explanation any appeal to its being ‘poisonous’ would have vanished: ‘ingestion of something poisonous’ does not give but only locates the explanation.
Similarly, to explain behaviour as motivated by belief isn't to explain it but to point to the explanation. ‘Why did she take her umbrella?’ ‘Because she thought it was going to rain.’ This ‘explanation’ adjusts our expectations about where to look for the explanation: namely, in a state whose causes and other effects lie in a given range. For instance, the causes might include her having seen a weather report forecasting rain, or her being in the Lake District. The other effects might include her not bothering to water the plants or putting on waterproof boots.
Rejecting the hypothesis that somebody acts out of a belief therefore has no tendency to frustrate our explanatory ambitions, because on the present conception of belief, the hypothesis itself was never an explanation but only an indication of where one could be found. In rejecting the hypothesis of belief, we are not giving up on explaining behaviour but only adjusting our expectations about where such explanations might best be sought. Why do people pray? In denying that they believe that God exists, we are suggesting that the explanation is not their evidence of God's existence, or any state that also causes them not to fear death, or to see miracles everywhere. But other explanations remain available: maybe they pray because it is a habit, or because they were taught to pray as children, or because it gives them a feeling of comfort (or any of many other possibilities).Footnote 15
Conclusion
I have not offered conclusive grounds for thinking that all or most followers of all or most religions lack religious beliefs. But I have described a Wittgensteinian theory of belief on which things could look that way, and I have pointed out the kinds of facts that would, if we found them, suggest this radical conclusion. The conclusion here is therefore not that we should ‘refuse to concede [to belief] any . . . empirical value as an index to the inner life of men’ (Needham (Reference Needham1972), 234) but rather that in the case of religious attitudes it is a matter for doubt whether such a concession is ever warranted. What could settle this open empirical question are the facts to which Wittgenstein's analysis directs us.Footnote 16
