Why Belief? Varieties of Religious Commitment

1 Theories of belief and the “puzzle”

Crane begins by introducing an account of belief as a mental state with four properties: (a) it has propositional content, (b) it aims to represent the way the world is, (c) it guides the behaviour of the believer, (d) it is inferentially integrated with other beliefs. This fourth component, which plays a key role in generating the “puzzle” about religious belief, provides the focus of discussion: “only mental attitudes which ... are inferentially integrated can count as beliefs” and for an attitude to be a belief “rational or inferential integration is crucial.” I will call an account that requires all these properties (and in particular the fourth) a robust theory of belief. A robust theory contrasts with more modest theories that dispense with (d) in favour of fewer conditions, or ones that are less intellectually demanding on the agent, for example, accounts that focus on the norm or truth,^[2] direction of fit,^[3] or transparency,^[4] i. e., that an agent’s belief that p is fixed by answering the question whether p.

The puzzle is this: on the one hand, many of the propositional commitments distinctive of a religion are, on the face of it, matters of the belief; on the other hand, religious commitments often lack the inferential integrity that are required for belief on the robust account. This puts into question “the categorisation of religious belief as belief.” To make this point, Crane elaborates on what inferential integrity involves and why some religious judgements lack it. It is, first, consistency with other beliefs. For example, if someone comes to believe both that

1. Nobody survives death.

and

1. There is eternal life.

“then they should change their belief” or at least explain the apparent conflict. Second, integrity requires that the believer have some cogent explanation of how the believed propositions can be true. For example, if somebody believes that

1. Our ancestors’ souls are mere spirits.

and

1. Our ancestors’ souls have bodily desires.

“then they are rationally obliged to explain” how this can be so. Since religious commitments seem routinely to fail to meet these demands of consistency and explanatory coherence – Christian doctrine and the views of the Vezo tribe are put forward as evidence – it seems to follow, according to the “puzzle,” that these things are not really believed.

The considerations about the consistency and explanatory coherence of belief are, prima facie, salient to an account of belief rationality. To generate a “puzzle” about whether religious commitments are beliefs, these will need to be set up as a condition of belief possession. How should this be done? Suppose we take consistency and explanatory coherence as normative requirements for belief possession. So, someone finding themselves committed to two propositions (1) and (2), if those propositions are also believed, will need to recognize the tensions between these two claims, see the need for an account of how both p and q can be true, etc., even if they are not themselves in a position to provide it. But on this normative reading of inferential integration, no substantive puzzle about religious beliefs emerges. The evidence that some religious commitments appear to conflict with beliefs – which Crane provides for (1) and (2) and for (3) and (4) – does not establish a failure of norms of consistency or explanatory coherence needed to generate the “puzzle.” To arrive at a puzzle, we will need evidence that people with religious commitments pay no heed to considerations of consistency or explanatory coherence, or even deliberately violate these norms.

Should we, in any case, see inferential integrity as a possession condition for belief? It is defended in some contexts. For example, in philosophical work on delusion inferential integrity has been used to distinguish beliefs from delusions.^[5] But biased, prejudicial, and hypocritical judgments also seem to display similar kinds of irrationality. It is difficult to specify the condition in a way that avoids the counterintuitive results that biased judgements are not beliefs.

A potentially more promising motivation for the “puzzle” comes from the distinctive relationship that religious beliefs appear to have to evidential considerations. This has been the subject of a vast literature, but I will mention one position: George Rey’s meta-atheism.^[6] He argues that religious commitments exhibit various properties to suggest they are not genuine beliefs: their “detail resistance,” or lack of seriousness about filling in the details of how religious claims can be true, and their insensitivity to empirical evidence. This evidence is used to suggest that religious commitments are not full beliefs, but instead a kind of self-deception or fictional belief. While I disagree with Rey’s conclusion,^[7] his version of the “puzzle” has the advantage that it is based on evidence that religious commitments are not aimed at the truth, rather than relying on a robust theory of belief.

2 Cognitivism and non-cognitivism

Crane builds on the account of robust belief to outline a dispute between religious cognitivists and non-cognitivists: cognitivists maintain that religious commitments are beliefs in the robust sense and religious statements are assessed by the standards of rationality and irrationality, consistency and inconsistency. Non-cognitivists maintain that religious commitments are “not subject to rational norms or requirements” and expressions of these commitments are akin to “expressions of hope,” or emotional responses, and when it comes to consistency and rationality “these standards do not apply to what they are calling religious beliefs, since these are expressions of attitudes which are not assessed in this way.” Crane introduces this framework based on treatments of the topic in metaethics, rather than in the philosophy of religion, apparently on the basis that realism/antirealism debates in the two fields are going to be the same. I have some reservations about this approach.

First, the characterisation of non-cognitivism is difficult to square with the positions one finds in metaethics that go by this description. If ethical non-cognitivists maintained that standards of rationality and consistency did not operate in ethics, the position would be implausible at the outset. Forming an ethical commitment is not a descent into emotional unreason: ethical judgements can be rationally assessed and debated, their implications thought through, arguments and evidence presented for or against them, and ethical sentences can be incompatible with each other, self-contradictory, regimented into valid arguments, and so on. A significant sub-field of metaethics, at least since Simon Blackburn’s work in the 1980 s, has focussed on finding an answer to the “Frege-Geach” problem: why ethical thinking and language, if ethical attitudes have the functional profile of desire, is constrained by considerations of logic. To take one prominent version of non-cognitivism, according to Allan Gibbard moral commitments are planning states that, while not beliefs, are certainly subject to rational consideration, combination, etc.^[8] If non-cognitivism is at all plausible, it needs to explain why ethical thought and language has a degree of logical and rational discipline, not just insist that it doesn’t.

Second, when the proposed cognitive/non-cognitive debate is transferred to religion, it has the unfortunate result of presenting a choice between two theoretical extremes: on the one side, religious attitudes are highly rational beliefs, on the other side, they are non-rational expressions of feeling. This gives a misleading picture of debates about religious commitments and language. It’s like setting up a match where Richard Swinburne and Gordon Kaufmann are the only two competitors, with no other players in contention. There is a large theoretical gap between these two opposing positions. Large enough, in fact, that the best part of a century of debate on the nature of religious commitments can fit into it. I will come to an example of this shortly. But here is one problematic consequence of this way of setting up debates about religion. Take two prominent and well-known positions on religion. First, according to Richard Dawkins and (some) other in the “new atheist” tradition, religious commitments are typically irrational beliefs: “faith...means blind trust, in the absence of evidence, even in the teeth of evidence.”^[9] This is taken to be one of their principal failings. Second, according to some fideists (including, arguably, Kierkegaard)^[10] religious commitments may be held in opposition to the demands of reason and evidence, but this is not a failing: it is a feature of religious commitment, not a bug. Now, on the framework that Crane proposes, these positions will count as non-cognitive theories because religious commitments will fail the standards of inferential integrity. This is a perplexing categorisation as it seems to deny the theoretical space to these positions. If it’s precisely the irrational or non-rational characteristics of religious belief to which these authors are drawing attention. But, on the proposed framework, these characteristics are incompatible with religious commitments being beliefs.

I don’t want to dwell on the issue of (non-)cognitivism, but it is useful to consider why non-cognitivist and expressivist theories, the stock in trade of metaethics, have not had the same prominence in work on religious language and why the debate in metaethics had not usually been seen as offering a promising model for understanding realism-relevant issues in religion. A striking feature of ethical judgements is that they seem to be tied to our motivations: to believe that giving to charity is good, for example, goes along with being motivated (at least to some extent) to give to charity. Moreover, ethical language is distinguished by the use of moral predicates – good, bad, right, wrong, etc. – which typically communicates approbation or disapprobation of the subject matter. Non-cognitivists have a very neat explanation for these facts: ethical commitments are desires (or desire-like attitudes), and desires are intrinsically motivational, and moral utterances conventionally express these desires. This situation is different for religion. For every religious sentence and commitment that looks ethical – “it is a sin to tell a lie” “you should pray five times a day” – there are others that don’t look ethical at all but more closely resemble scientific or historical judgments – “God created the world,” “there will be an afterlife,” etc. These latter judgements do not seem to be tied to a motivational profile, and if the utterances of them conventionally express a non-cognitive state nobody has yet identified a plausible candidate for what it is.^[11]

If research on religious judgements has not been primarily concerned with cognitivism/non-cognitivism, where has it been focussed? Here is one example (for others see Scott (2022)). A recent debate has focussed on the nature of propositional religious faith. Faith is in one respect a theoretically more interesting attitude to investigate than religious belief because it is seen as a distinctive and prominent element of religious commitment and, in some cases, the possession of it is an entry requirement for membership of a religion. Propositional faith seems to resemble belief, not least because it is a commitment to the propositional content being true. But, unlike belief, people with faith also sometimes go through extended periods of doubt. One way of accommodating this evidence is to say that faith that p is available to people with a non-doxastic attitude that p. Non-doxasticists take different views on how to specify the non-doxastic attitude required for faith: there are a spectrum of cases that more or less closely resemble the functional profile of belief. For example, according to William Alston, to have faith that p, one needs at least to accept p. Acceptance is understood, in line with Cohen,^[12] as the use of the proposition in one’s theoretical and practical reasoning, along with affirming it, calling it “true,” etc. As Alston puts it, “a sizable proportion of contemporary, sincere, devout Christians are accepters rather than believers.”^[13] Other, related, accounts of the non-doxastic attitude include: assent,^[14] assumption,^[15] acquiescence.^[16]

To be clear, I am discussing non-doxasticism not because I support it. I think acceptance and the other proposed non-doxastic attitudes are too pragmatic and transactional to satisfy other characteristics requisite for faith, in particular, the tendency of people with faith to resist evidence and psychological pressures to give up on their faith.^[17] I also worry that a non-doxastic theory puts into question the sincerity with which speakers assert their faith.^[18] However, the conception of belief that I (and I expect most other doxasticists) see as needed for faith is less demanding than the one Crane proposes. Neither robust belief nor non-cognitive attitudes are at issue in this debate about the nature of religious commitment.

3 Religious language and commitments

What are Crane’s recommendations for an approach to religious language and belief? Primarily, to pursue a more descriptive account of religious attitudes. We should “not try and legislate on these phenomena, and say how some theory says they should be, but describe them as they are.” Many religious attitudes, it seems, are only partially integrated, but may non-religious beliefs are like this too: “this is what many of our complex beliefs are like.” Religious beliefs “are like many other items in the swamp of belief.” Rather than wheel in some heavy-duty conception of what belief is and judge whether religious attitudes are beliefs according to this standard, we should “describe a person’s psychological reality as it is”.

These recommendations are ones that I expect that many will find agreeable, but also see as pushing against an already open door. Most contemporary accounts in the philosophy of religion of religious language or commitments are, at least to some extent, driven by the evidence of what people with these commitments say and do. Indeed, descriptive approaches to religious attitudes trace back at least as far as Wittgenstein.^[19] Wittgenstein points out the differences and similarities between religious beliefs and historical or scientific beliefs by focussing on examples. On the one hand, religious believers say that they believe these things, and affirm and uphold religious propositions as truths, and employ these propositions (at least to some extent) in their reasoning. But these beliefs also exhibit a degree of imperviousness to contrary evidence and are differently related to affective attitudes. Wittgenstein does not draw out any explicit conclusions from this – a fact that should not surprise anyone familiar with his writings – but the descriptive strategy, rather than insisting on some philosophical test for belief, is one that has been adopted by many working on religious language and the attitudes of speakers in the intervening years.

While a more descriptive approach to nature of religious commitment has much to recommend it, it also raises a preliminary problem about how to determine the scope of the enterprise. I’ll conclude with some comments on this issue.

How should we determine whether a given judgement or utterance is religious, and thereby falls into the range of examples that our theory of religious commitment should explain? Opinion on this matter tends to fall into two camps: on the one hand, many philosophers (me included) focus on sentences and judgements with a religious subject matter. We do not need to specify precisely what those subjects are to get an enquiry underway, but they include God, deities, angels, spirits, miracles, redemption, grace, holiness, sinfulness, etc. On the other hand, some philosophers delineate the field of interest by the context of utterance. For example, Alston proposes that religious discourse includes “language (any language) in connection with the practice of religion—in prayer, worship, praise, thanksgiving, confession, ritual, preaching, instruction, exhortation, theological reflection, and so on”.^[20] The scope of the topic will vary significantly depending on which approach we take. If, in the middle of a sermon, a vicar recommends that the congregants vote for a particular political party in an upcoming election, is that a religious utterance? Not on the former theory (unless the vicar says something like “God commands you ...”) because it does not have a religious subject matter. But it is said in a religious context and in the course of the practice of religion, so presumably counts as religious on the latter theory.

There are benefits and disadvantages to considering either context or content to delineate the field of enquiry. But Crane takes both approaches. The “puzzle” is set up for “the things people say in religious contexts”. But later the discussion moves to “credal statements” that, while they are not defined, presumably refer to declarative sentences with a religious subject matter. But credal statements are often uttered outside religious context, and non-credal statements are commonly used in religious contexts. We will need some way of identifying the relevant linguistic and psychological phenomena.