Plantinga's ‘No Obligation’ Argument

Plantinga's central argument is both astute and epistemologically important. I call it the ‘No Obligation’ argument, and it goes like this:

1. (1) There is a significant class of beliefs that are ‘properly basic’.

2. (2) Evidentialism requires an adequate account of properly basic belief.

3. (3) Philosophers have yet to provide such an account.

4. (4) Until they do, theists are under no obligation to refrain from believing in (or making claims about) God.

5. (5) Hence, theists are under no obligation to refrain from believing in (or making claims about) God.

There is much that Plantinga gets right here. I mean to show, among other things, that premises (1), (2) and (3) are all correct. Premise (4) is false, though, and dangerously so. My primary aim, however, is to develop a satisfactory account of properly basic belief, and in the process, enhance our collective understanding of what it means to be rationally responsible. This will actually render premise (3) false, and allow us to show that belief in God is not properly basic.

Apologetics and Rational Accountability

Note that Plantinga's argument is easily generalized: if it works to excuse theistic belief, then a comparable argument works to defeat evidentialist objections to other beliefs and claims – dangerously irresponsible ones included. What would we say to a doctor who claimed to be under no obligation to base his diagnoses upon evidence? Or a civil engineer who claimed that he was ‘entirely within his intellectual rights’ to ignore the evidence that a bridge was structurally unsound? Would it excuse either failing to point out that philosophers are still working out the kinks in evidentialism? Of course not. Evidentialism may face theoretical difficulties, but until a better standard comes along, we should deploy it as a heuristic, or rule of thumb. It's the best we have, and far better than nothing. More on this later.

Plantinga, of course, does not intend to excuse irresponsible believing. His argument, however, is easily repurposed to excuse any ideology you like. Just swap in a version of premise (4) that makes reference to your cherished convictions, and the argument will output the needed exemption from evidential standards. Why not use it, for example, to show that believers in astrology are under no obligation to rethink their views? If you don't find this variant of the argument troubling, substitute ‘exploitative capitalism’ for ‘astrology’. Or consider a variant that excuses the ideology of white supremacists. Thanks to Plantinga, ideologues can claim that the core tenets of their ideology are ‘properly basic’ and in no need of supporting argument. In this way, Plantinga's argument appears to open the floodgates to irresponsible cognition.

Things might be different if Plantinga were proposing a workable alternative standard – one that might serve capably in evidentialism's stead. But he isn't.Footnote ² He invalidates the accountability standard that prevails in every domain except religious discourse, and leaves us without a well-functioning surrogate. We are left to wonder how Plantinga and his fellow Christian apologists would combat ideological derangement and other forms of irresponsible cognition.

Plantinga's stance here is symptomatic of a larger failing within Christian apologetics. The overriding purpose of religious apologetics is to validate the hope that believers are ‘within their intellectual rights’ to believe. This concern for their rights though is seldom balanced by a concern for our responsibilities. By contrast, philosophers have a long history of trying to understand the trade-offs between our cognitive rights and responsibilities. The ‘sufficient evidence’ standard may be vague and in need of refinement, but it represents a useful balancing of two legitimate interests. It is, I claim, the standard we should strive to meet until a better one comes along.

What is Properly Basic Belief?

The problematic nature of Plantinga's argument, though, should not blind us to its insights. For it points to a deep problem in our collective understanding of reason's requirements – a problem that evidentialists, above all, need to address. So let's examine the argument carefully, starting with Plantinga's concept of ‘properly basic’ beliefs.

Plantinga defines a basic belief as one that is ‘accepted, but not accepted on the basis of any other beliefs’ (87). Such a belief is proper if it is reasonably or rightly or justifiably believed even so. Plantinga's examples include the belief that ‘2 + 1 = 3’ and the belief that ‘The world has existed for more than 5 minutes’. Under suitable conditions (e.g. standing before a tree in good light), ‘I see a tree’ can be properly basic. Similar things can be said of ‘There is a mild pain in my knee’, and ‘I had breakfast this morning’. These examples illustrate two things of importance. First, we believe many things that we don't know how to argue for – things that we would be hard-pressed to demonstrate to the satisfaction of anyone with serious doubts.Footnote ³ Second – and this is important – we appear to be well within our intellectual rights to believe such things even so. Following Plantinga, I will call such beliefs ‘properly basic’. Evidentialists must concede, I think, that properly basic beliefs (or PBBs) exist; Plantinga's first premise is true.

Do We Need An Account of Properly Basic Belief?

Yes. Clearly, PBBs play a pivotal role in our thinking and reasoning. To understand what it means to be rationally responsible, we need clarity about this role. Definitionally, of course, PBBs must be basic: their legitimacy cannot be inherited, by argument, from yet more basic beliefs (else they sacrifice their claim to be basic). ‘Basic’ here means something like ‘the end of the argumentative line’. Also, PBBs must provide a basis: that is, serve as premises in arguments for other permissible beliefs. This combination of features gives PBBs a very significant role in our epistemic dealings, if only because every argument relies, at some point, on premises which themselves go unargued. (There are versions of evidentialism that seem to imply that reliance on unargued premises is never permissible, but these versions are clearly untenable.) When is it rationally permissible to rely on unargued premises? Why is such reliance OK in some cases, but not others? To answer these questions, we need a working understanding of properly basic belief.

‘We’ here refers to all responsible human beings: anyone who would understand their epistemic obligations needs an understanding of properly basic belief. Arguably, though, evidentialists are particularly obliged to provide such an account. Why? Because evidentialists have led the effort to articulate shared standards of rational accountability. They attempt a general account of rationally permissible belief, and ask us to take it seriously, so more than most, it is they who owe us a workable account of properly basic belief. Plantinga's premise (2), then, is also true.

Are Traditional Accounts of Basic Belief Adequate?

Plantinga needs the claim that philosophers have yet to provide an adequate account of properly basic belief – what I above call premise (3). His argument for this claim is incomplete, but this does not prevent (3) from being true. I'll sketch a better argument (thereby strengthening Plantinga's overall case) because clarity and fair-mindedness demand it. This will illustrate the need for a better understanding of PBBs.

Historically, accounts of properly basic belief have been dubbed ‘foundationalist’ – so called because PBBs are thought to make up the ‘foundation’ of our knowledge. Plantinga summarizes philosophy's 2,500-year effort to understand properly basic belief in a complicated formula:

More simply, philosophers have tended to understand PBBs as self-evident, incorrigible, and/or evident-to-the-senses. This is a fair characterization, but in truth, philosophers have proposed other accounts of proper basicality. For example, Aristotle called PBBs first principles and argued that, on closer examination, they turn out to be inductively validated. The Academic and Pyrrhonian sceptics spotted the problem with this story, and concluded that properly basic beliefs can't exist. (Aristotle conveniently ignored the fact that validation by means of inductive argument would render his ‘first principles’ non-basic – which is to say, not ‘first’ at all. His ‘solution’, in other words, does not so much solve the problem of properly basic belief as evade it.) A radical scepticism set in, and this undermined commitment to rational accountability. The faith-based epistemologies of medieval Christendom were arguably a reaction to this radical scepticism (Norman Reference Norman2010–11). Aquinas concluded that we need articles of faith to function as basic. Descartes felt that a properly basic belief must be indubitable, or impossible to doubt, and nominated ‘I think’ as the cornerstone of all knowledge. For many rationalists, the intrinsically credible axioms of mathematics are exemplars of PBB. Empiricists, by contrast, nominate perceptual beliefs as properly basic. Kant held that synthetic a priori judgements play a foundational role. Logical positivists nominated observation statements justified by the sensory ‘given’. Some pragmatists argued that knowledge rests on a set of starter assumptions that are validated purely by their practical fruitfulness, rather than their epistemic grounding. Wittgenstein, the existentialists, and some postmodern thinkers concluded that only groundless or unjustified beliefs can be basic – a stance that, unfortunately, courts relativism – and so-called ‘reliabilists’ have nominated certain reliably caused beliefs. The sheer number of disparate attempts is a testament to the importance and difficulty of the problem of properly basic belief.

It would take us beyond the scope of this essay to examine these proposals individually, so I'll limit myself to some general observations relevant to present purposes. Plantinga argues – correctly, I think – that foundationalist accounts tend to be self-defeating. (For example, it is hard to see how ‘One is rationally permitted to believe only what can be derived from indubitable premises’ can itself be derived from indubitable premises; Plantinga is surely right that the theist is under no obligation to accept a standard that is self-referentially incoherent.) More generally, philosophers usually opt for highly restrictive understandings of PBB, allowing relatively few beliefs to function as basic. Why is this? One answer is that we have long been enamoured of the power and economy of Euclidean geometry – a system that derives an astonishing variety of theorems from five simple, apparently self-evident axioms. Many have sought, quite explicitly, to axiomatize human knowledge in a similar fashion.

Others have approached the problem of properly basic belief with a decidedly sceptical, anti-theistic axe to grind: they adopt narrow conceptions of PBB in order to invalidate theism (Hume Reference Hume1993; Ayer Reference Ayer1952). Of course, many discover, to their chagrin, that their proposed standards wipe out large swaths of what we normally, in our saner moments, classify as knowledge: moral knowledge, causal knowledge, knowledge of the external world, knowledge of other minds, predictive knowledge, counterfactual knowledge, etc. Plantinga is well within his rights to object to such jury-rigged challenges to theism.

Even when we philosophers don't have an atheist axe to grind, we often err on the side of scepticism. We tend to view humanity as overly credulous (this much is almost certainly true), and we see ourselves as adding value primarily through critical analysis. In our discipline, the incentives for being critical are many, and those for being trusting and assertive are few. Moreover, our sense of purpose is bound up with being skilled debunkers. Consequently, it is in our interests to opt for a highly restrictive concept of properly basic belief. The result is a kind of institutionalized bias in favour of overly demanding concepts of rationality – a fact that explains why philosophy has long been haunted by extreme forms of scepticism. Fair-mindedness demands that we acknowledge this, and seek an account of properly basic belief that is free of such deformations.

‘So far as I know,’ concludes Plantinga, ‘no one has yet developed and articulated … revealing necessary and sufficient conditions for proper basicality’ (2015: 89–93). In this, I take him to be correct; premise (3) is also true. In what follows, I will untangle the conceptual confusion that sustains the problem, develop a practical account of properly basic belief, and in so doing, revitalize the evidentialist challenge to theism.

Justification without Argument

So what would an adequate account of properly basic belief look like? Can evidentialism supply such an account? There are two dimensions of adequacy that we must bear in mind. First, the overall account must be workable: evidentialists exhort us to examine the evidence that underlies propositions, and assent to the propositions when, but only when, the evidence proves ‘sufficient’. To put this proposal into action, we need a working understanding of what sufficient evidence looks like – so we can judge, with some reliability, when the condition is satisfied. Such an understanding appears to entail a working understanding of PBBs. The evidentialist owes us this much, it seems, on practical grounds alone: otherwise, we can't put evidentialism to work.

But evidentialists must also satisfy a theoretical requirement. Why? Because their exhortations are based on the conviction that a belief's evidential grounding is what makes it reasonable (and the lack thereof what makes other beliefs unreasonable). For this conviction to be generally true, it must hold, not only for derivative beliefs, but also for basic ones. It follows that evidentialists must understand properly basic beliefs as evidenced, yet somehow still basic. In other words, evidentialists have an obligation to solve a longstanding conceptual puzzle: How can evidence justify a basic belief without thereby rendering it non-basic?

A common misconception blocks the path to a solution. The mistake is to assume that evidential support must take the form of an argument. Plantinga makes this assumption, and compounds the error by imputing it to evidentialists:

In other words, Plantinga equates having sufficient evidence and having a good reason. Evidentialists, though, are free to dispute this curiously idiosyncratic equation. For evidence frequently takes other forms. Yes, a reason (or ‘proposition’) is often a perfectly good way to call attention to relevant evidence, but it is also possible to present evidence in data tables, charts, graphs, and footnotes. A pointing gesture can serve to evidence something, as can a distinct memory. So can a rhetorical question, a well-chosen example, or an illustrative behaviour. Evidence can take the form of a bag of fossils, a telescope image, an audio recording, a computer simulation, or a signature on a contract. Indeed, anything that reliably indicates something can be evidence. Notice that many forms of evidence cannot enter into inferential relations: a fossil cannot function as a premise, even if a description of it can. The latter is propositionally articulated, the former isn't, but both may be thought of as evidence. A bit of language with sentential structure can be an effective way to draw attention to relevant evidence, but it is not the only way. Simply put, X can evidence Y without X being an argument for Y. Evidential support can be, and often is, non-inferential. Non-inferential justification is one key to understanding properly basic beliefs.

This partially dispels the epistemological mystery, but to fully dispel it, we must explain why a properly basic belief's non-inferential justification needn't be spelled out in the form of an argument. What is the operative sense of this ‘needn't’? Plantinga provides a clue when he describes properly basic beliefs as such that the believer is ‘under no obligation to reason to the belief from others’ (2015: 93). For example, we are under no obligation, under normal circumstances, to reason to the belief that ‘2 + 1 = 3’, or to provide an argument for ‘I had breakfast this morning’.

But how are we to understand this lack of obligation? What is its source? What makes it permissible, under certain circumstances, to refuse to provide a justifying reason? More specifically: can one refuse to argue for God's existence, decline to rethink one's belief in him, and claim all the while to be reasonable? Let's find out.

Understanding Proper Basicality

The answer to the riddle of properly basic belief is surprisingly simple: beliefs are properly basic, and not in need of argumentative support, when the burden of proof is not on the believer. In fact, PBBs are properly basic because the burden of proof lies elsewhere. Instead, a burden of disproof falls on any would-be challenger. It is not up to the holder of a PBB to provide reasons for, because it is up to any sceptic to provide reasons against. Burden of proof and presumption, by the way, are two sides of the same coin, the defendant's presumption of innocence being nothing other than the plaintiff's burden of proof. Consequently, we can say, quite simply, that properly basic beliefs are presumptively reasonable, or yet more simply, presumptive.

All of Plantinga's examples of properly basic belief, with the notable exception of some controversial claims about God, are presumptive: ‘I see a tree’, ‘There is a mild pain in my knee’, ‘I had breakfast this morning’, and ‘The world has existed for more than 5 minutes’ are all such that, in just about any context you can imagine, the burden of disproof quite clearly falls on anyone who would challenge them. Plantinga's case for premise (1) – the claim that properly basic beliefs exist – ultimately rests on the fact that there are presumptively reasonable beliefs. In fact, presumptions constitute his evidence for PBBs.

How do we know when a claim is presumptive (or properly basic)? Here's a simple test: Ask yourself if it is appropriate that the claimant bear the burden of proof. If the answer is yes, the claim is not properly basic; if the answer is no, it is properly basic. Need a more refined test? Ask yourself whether the ‘bare’ challenge ‘How do you know?’ ought to be understood as suspending the claimant's right to continue asserting and/or premising it. If the answer is yes, the challenge can be said to ‘arise’, the burden really is on the claimant, and the claim is not properly basic. If instead the challenge can be appropriately turned aside with a simple ‘Why do you doubt it?’, then the claim really is properly basic. To perform this test, one needs judgement and a sense of propriety, but this is no bar to the criterion's being correct: one needs judgement to interpret anything.

Socratic Evidentialism

Let's condense our emerging understanding of properly basic belief. I am proposing that we classify a belief as properly basic just in case:

1. a. It is justified, non-inferentially, by sufficient evidence, and

2. b. It is, for all that, presumptive, because the burden of disproof falls on anyone who would dispute it.

The intuition is that a belief must satisfy condition (a) to be proper – that is, rationally available to deploy as a premise. It must satisfy condition (b) to be basic – that is, not in need of further argument. Proper basicality, I suggest, consists in a belief's having precisely this combination of properties.

Can one be aware of the evidence that justifies a properly basic belief, and still refuse to voice the argument that calls attention to that evidence? It depends. If the question about the basic belief is well intentioned, based on an open and honest desire to know, then it would, of course, be churlish to refuse to share the requested information. But if the belief is clearly presumptive, and the challenge is motivated by overweening scepticism, shallow one-upmanship, or partisan zeal, then the believer is certainly within his rights to deflect the challenge with a simple ‘Why do you doubt it?’ Such a response puts the onus back where it belongs.

I call the epistemological outlook that results Socratic evidentialism. It is evidentialist because it affirms the ‘sufficient evidence’ requirement for reasonable belief. It deserves the appellation ‘Socratic’ for reasons I detail elsewhere (Norman, Reference Norman2010–11), but here's the short version. Properly basic beliefs can be correctly described as justified by evidence, but in the particular conversational context in which they function as basic, the salient aspect is their not being in need of justification. To participants in that conversation, they show up, quite properly, as unproblematic. They're not worth questioning, hence trivially in compliance with the fundamentally Socratic requirement that reasonable beliefs be able to withstand sensible questioning. Thus, non-basic belief can be understood as the special case where the bare challenge ‘How do you know?’ does arise (and justification is had, if at all, by virtue of one or more supporting arguments); and basic belief can be understood as the special case where no challenges arise (and the Socratic requirement is trivially satisfied). Once we understand that presumptions exist – that the need for justifying argument is not universal, but situational – our mental picture of reasonable belief must shift from reasonable-by-virtue-of-a-supporting-reason (arguably Plato's picture of reason) to reasonable-by-virtue-of-being-able-to-withstand-questioning (arguably Socrates’ picture).

Objections and Replies

Before applying this account to the question of God-belief, let's see if it withstands scrutiny. Test it, first of all, against Plantinga's examples of properly basic belief. Indeed, test it against any cases you like. You will find, I think, that it fits the evidence beautifully. More precisely, the account includes uncontroversial cases of properly basic belief, and excludes uncontroversial cases of non-basic belief. (The account should classify as basic claims that may rightly function as an unargued premise, and it should classify as non-basic claims that should not so function – and provably, it does precisely this.) Controversial cases – like belief in God – of course, should not be used in this testing. Indeed, only someone with a partisan axe to grind would use them in this capacity. The objective here is to develop a shared understanding of responsible belief, and for this purpose, we must form a community of inquiry that includes both believers and non-believers, and build from the ground up (as Plantinga himself notes: 93). More precisely, we must build outward from common ground.

Let's consider some common concerns about this proposal, and see if any constitute a genuine problem. First, some will worry that it opens the door to dogmatism. After all, it does allow some people, in some circumstances, to make assertions, then refuse to provide justifying reasons. Is this not a recipe for dogmatism?

No. Politely to decline to shoulder a burden of proof that isn't yours is by no means tantamount to unreasonable tenacity. More precisely, one can recognize the situational impropriety of a ‘bare’ demand for supporting evidence (such challenges presume, sometimes incorrectly, that the onus is on the claimant) without refusing to countenance any ‘onus-bearing’ challenge that might come along. One may decline to provide a reason for without closing one's mind to the reasons against that might be out there. The latter really is ethically problematic, but that in no way makes the former so. Acknowledging the reality of presumption is by no means tantamount to closed-mindedness.

Second, some will worry that this account makes rationality ‘relative’. And it is true that it makes reason sensitive to relevant features of the evidential, doxastic, and conversational context. I will not deny that, when two Christians discuss their experiences of God, and neither is remotely interested in the question of God's existence, they are within their rights to take ‘God exists’ as properly basic. For them, in that context. The story will prove quite different, though, for the case where a Christian and a secularist debate the wisdom of a public policy proposal based on scriptural dictates. Here, it would not be appropriate for the Christian to insist on the proper basicality of ‘God exists’, and argue for his policy position on that basis. (Plantinga, I suspect, would concede this.)

I contend that an account of reason's requirements should be sensitive to relevant features of context. This kind of sensitivity may be caricatured as ‘relativist’, but it is in no way pernicious. It's possible, of course, to relativize reason in a way that prevents it from functioning properly in the resolution of differences. When Plantinga claims that ‘the Christian community is responsible to its set of examples [of PBB], not to theirs [that is, those of atheists]’, he makes rationality relative to the articles of faith that prevail within a somewhat insular community (2015: 93). Clearly, this relativizes reason in a way that would prevent it from functioning properly in interfaith dialogue. This way of relativizing reason is pernicious, but it is distinctly different from the scrupulously engaged, fair-minded dialogical sensitivity I'm advocating.

Others will complain that the ‘sufficient’ in ‘sufficient evidence’ is vague. This is true, but it is by no means a failing. To see this, notice first that the notion of ‘sufficiency’ is linguistically indexed to an appropriate threshold: just as ‘too demanding’ is definitionally overmuch, ‘sufficient’ is definitionally just enough. By linguistic convention, sufficiency automatically occupies the ‘Goldilocks zone’.

Of course, we'd like a simple and useful answer to the question ‘How much evidence should we require?’ and evidentialists tend to be cagily, even frustratingly, vague on the subject. But they have excellent reasons for doing so. For one, evidential thresholds vary, depending on what's at stake. Thus, we expect a cancer diagnosis to meet a higher evidential standard than a claim about the greatest halfback ever to play the game. The contextual nature of evidence thresholds is recognized by the American legal system, which employs a ‘beyond reasonable doubt’ standard in criminal trials, and a ‘preponderance of evidence’ standard in civil proceedings. The main point, however, is that educated judgement intuitively adjusts evidential expectations to fit the circumstances. There can be real wisdom in such judgement, even when it is unthinking and automatic. And this wisdom can be hard to formulate explicitly.

So the vagueness in evidentialism's ‘sufficient evidence’ standard is not motivated by evasion. It is motivated by the recognition that situated judgement must be accorded the freedom to do what it does best. Theory must know its place, and not usurp the place of practical judgement. Evidentialism makes a very high-level claim, and one should not expect more precision at that level of generality than the subject matter allows. Evidential thresholds are, and ought to be, contextually variable; the ‘sufficient’ in ‘sufficient evidence’ is appropriately vague.

Note that this does not entail that evidential thresholds are arbitrary. They may be context-sensitive, but that doesn't mean that it's OK for an individual or a group to adopt any evidential threshold he, she or it likes. Engineers who make claims about the structural soundness of bridges are held to shared, evidence-based standards suited to the discipline of bridge-building. Scientists, doctors, litigants and ship-owners each face accountability norms tailored (however imperfectly) to the needs of their domain. In each case, truly accountable cognition answers to standards that are appropriately sensitive to salient features of the situation. Of course evidential thresholds vary; that doesn't make them arbitrary.

Finally, some will find fault with this account for deploying normative language – for using terms like ‘should’ and ‘ought’ to explain properly basic belief – thereby compelling users of the account to exercise normative judgement. This concern, too, rests on an error. For it is not theory's job to reduce normative concepts to non-normative ones. Nor is it theory's job to relieve epistemic agents of responsibility to exercise judgement. Theoreticians can direct attention to relevant features of context, but it is practitioners that must exercise their judgement – including their normative judgement – in applying it.

Is Belief in God Properly Basic?

If two adherents of the same religion choose to take God's existence for granted so they can focus their attention on a higher-level question (such as whether it's OK for them to use electricity or play football on the Sabbath), then we might signal their right to do so by classifying ‘God exists’ as properly basic for them, in that situation. Whether God-belief is properly basic as such (or generally to be regarded as available for premising, or rightly regarded as responsibly held by the global community of inquiry), though, is another matter entirely. To answer that question, we must apply our new account. And that means asking, first of all, whether we have sufficient evidence for God's existence.

Reflect for a moment on what ‘sufficient’ ought to mean in this context. What's an appropriate evidential threshold for a matter such as this? Notice that the stakes here are, on the believer's own account, incredibly high: we're talking about the most important being in (or out of) the universe, a being whose need to be believed is so great that we mortals must organize our lives around his worship. Nothing is more important than understanding that he is, for the fate of our eternal souls hangs in the balance. Non-believers, by the way, should agree that the stakes are high, for God-belief colours the perception of billions of people, and rearranges human priorities in ways that tangibly affect the well-being of us all. Does it make sense to set a low evidence threshold for such an important matter? Should second-rate evidence satisfy us? Clearly not. We wouldn't accept second-rate evidence for the claim that the Great Pumpkin will save us from ecosystem collapse, and we shouldn't accept second-rate evidence for God's existence. Extraordinary claims require extraordinary evidence, especially when they purport to describe an aspect of reality that transcends well-established physical laws. Miracles, virgin births, divine creation, the afterlife, bodily resurrection, eternal hellfire: these are not claims that merit exemption from evidentiary standards, they're claims that require exceptionally conclusive evidence.

Notice also that, if they wish to take rational accountability seriously, religious apologists need an evidentiary threshold high enough to invalidate the many thousands of faith claims made by rival religious traditions, but low enough to permit their own cherished convictions. Have religious apologists ever offered such a proposal? I'm not aware of any such attempt.

So claims of God's existence require extraordinary evidence. Do we have such evidence? Well, many people have powerful religious experiences. Can't moments of transcendence, transformative revelations, profound experiences and mystical insights count as evidence of a God? Again, no. Such ‘evidence’ fails on at least three counts. First, to function properly as evidence, the information conveyed in these experiences would have to rule out competing hypotheses. (This is built in to evidence's job description: if a bit of information doesn't rule out, or at least weigh against, competing hypotheses, then it's not evidence at all.) And it never does. In fact, there are many (and many more likely) hypotheses compatible with such information. After all, our minds are known to play tricks on us. We know that emotional needs and wishful thinking powerfully alter human perception. People ‘see what they want to see’. We know that children have imaginary friends; mightn't religion be a grown-up game of ‘let's pretend’? We know that powerful biases – confirmation bias, motivated reasoning, groupthink – warp human judgement. Science provides an increasingly detailed picture of how the brain generates religious experiences, and the probabilities quite clearly favour the hypothesis that God-belief is an artefact of a decidedly quirky nervous system. For a religious experience to count as evidence for God's existence, it would have to rule out such a hypothesis, or at least weigh against it. Why? Because information needs to count against relevant alternatives in order to even count as evidence.

Second, evidence should be open to public scrutiny. Why? For the same reason that scientific experiments should be replicable – to provide an independent check. Religious experiences, though, are notoriously private. Why does God only appear in the fevered imaginings of self-proclaimed prophets and the private musings of the devout? If he were real, you'd think he'd make the occasional public appearance.

Third, real evidence should be such as to satisfy a dispassionate and fair-minded observer. The testimony of those who are emotionally invested in their God must be discounted for the same reason that a witness with a major stake in the outcome of a trial must be discounted: self-interest tends to compromise objectivity. Fair-mindedness comes, not from years of devotional practice, but from the disciplined practice of impartial habits of mind. The methods of philosophy and the sciences generate real evidence; the methods of religion do not.

God-belief's failure to meet the sufficient evidence condition is reason enough to deny it properly basic status. But it also fails to meet the burden of proof condition. For the burden of proof lies squarely on the theist. How do we know? Well, apply the test: should a believer in a certain god – Poseidon, say – be permitted to assert that Poseidon exists, claim that such belief is properly basic, and thereby compel the rest of us to accept whatever he can rationally derive from that premise? Or are we Poseidon-doubters within our rights to ask ‘How do you know?’ and expect a direct and sufficiently compelling answer before we acquiesce? Should a devotee of Poseidon be able to dismiss such a challenge with a casual ‘Why do you doubt it?’? Of course not. Our question should be understood as suspending his right to rely on ‘Poseidon exists’ in his thinking and reasoning, at least until a fair review of the evidence indicates otherwise. And if this is the correct answer for Poseidon, why is it not also the correct answer for Jehovah, Allah, and Christ?

Our concept of properly basic belief should function to designate the class of beliefs and claims we have an unproblematic right to rely on without further argument as we together build common understanding. PBBs are in essence the common ground that believers and non-believers, Hindus and Muslims, Protestants and Catholics, liberals and conservatives, Poseidon-devotees and Poseidon-doubters should turn to for premises. Divisive and highly improbable claims are the last things we should treat as properly basic.

Thousands of years of concerted effort haven't produced any onus-shifting arguments for God's existence. (The argument from design might once have placed the onus on non-believers, but Darwin turned the tables on that argument long ago.) Religious apologetics is a marvellous testament to motivated reasoning, but it is no testament at all to the existence of God. Meanwhile, doubters have long since shifted any burden of disproof they might once have shouldered. Indeed, they've raised challenges that would scuttle any ordinary hypothesis. The problem of evil, for example, points out that the standard Christian God-hypothesis makes a testable prediction: that we will not observe any undeserved suffering. The fact that we observe lots of it would lead anyone with a modicum of fair-mindedness to discard the God-hypothesis, but that is not what religious apologists do. Instead, they engage in the intellectual contortions known as theodicy. In fact, our modern concept of God bespeaks a tortured effort to evade evidential accountability: why else would He have evolved, over thousands of years, into an invisible, inaudible, intangible, inscrutable, supernatural, mysterious, and faith-based god-of-the-gaps?

No, Virginia, belief in God is not properly basic. As long as our beliefs and opinions have implications for the well-being of others, we will have intellectual responsibilities as well as rights. At a minimum, being responsible means observing the same standards we would have others observe, and if we want those standards to be truth-conducive, they will involve an evidence requirement. (This follows directly if we define evidence as a reliable indicator of the way things are.) The ‘sufficient evidence’ standard may lack specificity, but it remains essential, not just to scientific progress, but also to civilized co-existence and moral growth. God-belief is a clear case of irresponsible cognition – one that flouts civilizing standards of rational accountability. It requires an epitaph, and scripture, I think, provides it: