Reformed Epistemology and Proper Function

Nine rungs of arguments, and the whole enterprise has taken one thing for granted: that the way to be entitled to believe in God is to have an argument. Cosmological, teleological, ontological, moral, from religious experience, cumulative; the disagreements were about which arguments work, not about whether arguments were the right currency.

Evidentialism is the assumption underneath that. It says that a belief is rational only if it is supported by evidence, and W. K. Clifford's essay of 1877 put the demand in its most uncompromising form, treating belief on insufficient evidence as a moral failing rather than merely an intellectual one. Natural theology is the attempt to meet the demand. Fideism is one way of refusing it, by holding that faith neither has nor needs rational support, and it is a position with a long history and few contemporary defenders in analytic philosophy.

The position in this rung is neither. It accepts that rationality matters, denies that arguments are required for it, and argues that the demand for evidence, applied consistently, would take almost everything with it.

What the objection was resting on

The evidentialist objection has a hidden second premise, and finding it is the first move.

To say belief in God needs evidential support is to say it is not the kind of belief that can be held without support. Some beliefs plainly can: you do not argue your way to the belief that you had breakfast, or that there is a screen in front of you, or that other people have minds. These are held immediately, and holding them immediately is not a failure. So the evidentialist owes an account of which beliefs may be held without support and which may not, and classical foundationalism was that account. On the version that dominated modern philosophy, a belief may be held without support only if it is self-evident, incorrigible, or evident to the senses.

The reply that Reformed epistemology built itself on is that this criterion fails twice over. It is too narrow, since it excludes memory beliefs, beliefs about other minds, beliefs about the past, and inductive beliefs, none of which are self-evident, incorrigible, or evident to the senses, and none of which anyone proposes to give up. And it is self-referentially incoherent, because the criterion itself is not self-evident, not incorrigible, and not evident to the senses, so by its own lights nobody is entitled to accept it without argument, and no argument for it has ever been produced.

The parallel with the fate of the verification principle in the previous rung is exact, and it is not a coincidence. Both were attempts to draw a boundary around respectable belief using a criterion that fell outside the boundary it drew.

The Reformed proposal

The programme was announced in Faith and Rationality: Reason and Belief in God, edited by Plantinga and Wolterstorff (University of Notre Dame Press, 1983), with Plantinga's long essay "Reason and Belief in God" as its centrepiece, and it had been set out in shorter form in "Is Belief in God Properly Basic?", in Noûs in 1981.

The theological source is Calvin, who holds in the Institutes of the Christian Religion that there is in every human being an awareness of divinity, a seed of religion that is not the conclusion of any reasoning and that accounts for the near-universality of religious practice.

The comparison with perception is doing the argumentative work and should be stated carefully. When you see a tree, you do not infer a tree from a visual experience. You form the belief immediately, and the experience is the ground of the belief without being a premise. Plantinga's claim is that theistic belief can have grounds in exactly this sense: it is not groundless, but its grounds are circumstances rather than propositions, and it is therefore basic without being arbitrary.

Notice the modest form of the conclusion. It is negative, and Plantinga is careful to keep it that way at this stage: the argument removes an objection rather than supplying a reason to believe. Someone who accepts every premise has not been given any ground for theism, only a demonstration that one particular complaint against it does not work.

This is also what separates the position from fideism. Fideism holds that faith stands outside rational assessment. Reformed epistemology holds that theistic belief is assessed by the same standards as any other belief and passes, which is a considerably stronger claim and a considerably more exposed one.

Warrant

Plantinga's later work shifts the question from justification to a different property, developed across Warrant: The Current Debate and Warrant and Proper Function, both published by Oxford University Press in 1993, and applied to religious belief in Warranted Christian Belief (Oxford University Press, 2000).

The move begins from the observation that justification, understood as having fulfilled one's epistemic duties, is not what turns true belief into knowledge. Gettier's cases show that something further is needed, and Plantinga's proposal is that the further thing is proper function: a belief has warrant when it is produced by cognitive faculties functioning properly, in the kind of environment they were designed for, according to a design plan aimed at truth, where the plan is a good one in the sense that beliefs formed that way are objectively likely to be true.

Applied to theism this yields a conditional with a striking shape. If God exists and made us with a sensus divinitatis, then theistic belief produced by it satisfies every clause: proper function, congenial environment, truth-aimed design, high reliability. So if theism is true, theistic belief has warrant without argument. And if theism is false, there is no such faculty and no such warrant.

Plantinga presses the consequence hard in Warranted Christian Belief as the collapse of the de jure question into the de facto one. Objections to religious belief usually come in two forms: that it is false, and that whatever its truth, holding it is irrational or unjustified or unwarranted. The second is supposed to be available to someone who does not want to argue about God's existence. Plantinga's claim is that no such objection survives inspection. Every candidate for a de jure objection turns out either to presuppose that theism is false or to fail. The critic who wants to say belief in God is intellectually improper, without saying it is untrue, has nowhere to stand.

That argument is more consequential than the properly-basic claim it grew out of, because it targets a stance rather than a thesis: the widespread assumption that one can set aside the metaphysics and treat religious belief as an epistemological embarrassment on its own.

The obvious objection

Plantinga raised the objection against himself and answered it in two steps, and both steps are needed.

The first is that no criterion of proper basicality is available to anyone, and demanding one from him while having none oneself is not a fair request. The way to arrive at such a criterion is to assemble clear examples of beliefs that are properly basic and beliefs that are not, and to frame a hypothesis that fits them. That method is inductive, and it starts from the examples one has. The Christian will include theistic beliefs among the examples; the classical foundationalist will not. Neither is proceeding illegitimately, and there is no neutral ground from which to run the induction.

The second is that the account does not license every belief, because it requires grounds. Belief in the Great Pumpkin has no ground: no circumstance triggers it, no faculty produces it, nobody forms it on any occasion. Belief in God, on the model, has grounds in a wide range of circumstances and a faculty that produces it. The parity argument therefore does not go through, because the parody case does not have the features the model requires.

Defeaters

The reply just given puts the weight on a notion that has to be made precise, and it is the notion that does most of the work in what follows.

This is where the real argument about religious belief is conducted once the formal objection is set aside, and Plantinga accepts the shift. If belief in God is properly basic but defeasible, then everything turns on whether there are defeaters, and the candidates are not obscure.

The problem of evil is the leading rebutting defeater, and the remainder of this tier is largely occupied with it: whether the distribution of suffering is evidence against theism, whether a theist may plead ignorance of God's reasons, and whether particular evils resist any accounting. Religious diversity supplies a different candidate, since a believer who learns that equally thoughtful people form incompatible beliefs by the same route has some reason to doubt the route. Naturalistic explanations of religious belief supply a third, since an account of why the belief arises that makes no reference to its truth threatens to undercut rather than rebut it.

Two of these are the subject of later rungs, and it is worth noticing what the shift means for how they should be read. On the Reformed view, the arguments in the rest of this tier are not attempts to establish theism from a neutral start. They are the exchange over defeaters, and the theist's task is not to prove the belief but to defend it against undermining.

The other route: perception

Plantinga's is not the only version of the programme, and William Alston's is different enough to be worth separating, because it makes a claim the Plantingan account deliberately avoids.

In Perceiving God: The Epistemology of Religious Experience (Cornell University Press, 1991), Alston argues that religious experience functions as a perceptual practice: people take themselves to be directly aware of God, form beliefs on the basis of that awareness, and check those beliefs against a body of accumulated tradition in something like the way perceptual beliefs are checked against each other. His central move is to compare that practice with ordinary sense perception and to ask what entitles us to trust the latter.

The answer, he argues, is nothing that could not equally be said of the former. Any attempt to justify reliance on sense perception uses sense perception, so the justification is circular. We rely on it anyway, because it is an established practice that is internally consistent, socially shared, and self-supporting, and because there is no non-circular alternative for any basic source of belief. Alston's claim is that a religious perceptual practice meets the same conditions, and that rejecting it while accepting sense perception requires a difference between them that critics have not supplied.

The account is more exposed than Plantinga's in one respect and less in another. It is more exposed because perceptual practices can be checked, and the checks available for religious experience are visibly weaker: there is nothing corresponding to the ability of any competent observer to look again. Alston does not deny this and argues that the disanalogy is a matter of degree rather than a difference in kind, which critics dispute. It is less exposed because it makes contact with something people actually report, rather than positing a faculty whose operation nobody notices.

Alston also faces the diversity problem in a sharper form than Plantinga does, and he says so. Different traditions produce incompatible beliefs through practices with equally good internal credentials, and if his argument works for the Christian practice it works for each of them. His response is that a practitioner is entitled to continue in the practice they are in until given reason to stop, which is a defensible position and also a thin one, and the objection that it makes the truth of one's religion a function of where one was born is the standing complaint against it. That is the material of rung 20.

What it settles

Philip Quinn pressed the most careful early objection in "In Search of the Foundations of Theism", in Faith and Philosophy in 1985: an adult intellectually alive to the modern world is exposed to a substantial body of potential defeaters, so even granting the account, most educated theists are not in the position it describes, and they will need the arguments after all. Plantinga replied in the same journal, and the exchange set the shape of the subsequent literature.

Three limits are worth stating plainly, and the first two are ones Plantinga accepts.

The account is conditional. It establishes that theistic belief has warrant if theism is true. It does not establish that theism is true, and Plantinga does not claim it does. Someone hoping for a reason to become a theist will not find one here.

The account is defeasible, so it does not remove the need to answer objections. It relocates the burden rather than lifting it.

And the account is available to others. Any belief system that can tell a story about a faculty producing its central beliefs can run the same argument, with the same conditional structure. Plantinga treats this as a fair result rather than a problem, since each such system is then in the same position, with its warrant contingent on its truth. Critics regard the symmetry as showing that the argument establishes less than it appears to: an argument that vindicates every view on the condition that the view is correct has not distinguished among them.

Next

The Reformed picture treats theistic belief as produced by a faculty rather than inferred from evidence, which raises a question it does not itself answer: what happens when theism is treated as an explanatory hypothesis after all, competing with naturalism over the same data? The next rung takes the two as rival frameworks and examines the disputes that follow, over simplicity, over scope, and over whether appealing to God to explain what science has not yet explained is a legitimate inference or a standing mistake.