Introduction
Jeremy Koons has recently offered a radical critique of Alvin Plantinga's ‘reformed epistemology’. Plantinga argues that Christian belief can be a form of undefeated, warranted, basic belief, the epistemic status of which is similar to that of perceptual or memory beliefs. Koons's critique of this position is radical in so far as it claims there is something flawed about the very idea of a basic belief. Koons argues that while there are beliefs that are psychologically non-inferential, there are no beliefs that are epistemically basic in the sense that Plantinga's position requires.Footnote 1
While I, too, am a critic of reformed epistemology, I cannot accept Koons's argument as it stands, since it relies on a misleading account of the theory-ladenness of perceptual beliefs. So the present article will put forward a different line of argument. It will assume that there are epistemically basic beliefs and that Christian beliefs may fall into this category. But it will also insist that their epistemic status is not independent of their probabilistic or explanatory relations with our other, well-established beliefs. It follows that even if Christian belief is warranted, as Plantinga suggests, its epistemic status can be undermined by evidential considerations. In particular, reformed epistemology constitutes no defence against an evidential argument from evil, such as that offered by Paul Draper.
Koons and Plantinga
Let me begin with Koons's suggestion. His strategy is to accept the analogy between Christian belief and perceptual beliefs, upon which Plantinga's argument relies. But he argues that neither Christian belief nor perceptual beliefs can be epistemically basic, since the formation of such beliefs involves bodies of theory. Koons makes his point with the aid of an example, which it will be helpful to reproduce in full.
The phenomenon of St Elmo's fire is a glowing region of atmospheric electricity which appears on pointed objects (church steeples, aircraft wings or propellers, etc.) during thunderstorms. Suppose Smith and Jones observe the same phenomenon during a thunderstorm. Smith is well read in science, is familiar with this type of atmospheric disturbance, and without hesitation judges the observed phenomenon to be St Elmo's fire. A familiar form of theory-ladenness is at work here: because of Smith's background theories, the stimuli he is presented with cause him to form a belief that is consonant with those theories. Jones' world view, on the other hand, is a poorly supported pastiche of superstition and the paranormal, which he has acquired from poorly sourced websites and unreliable supermarket tabloids (tabloids of the sort which specialize in absurd stories about Elvis sightings, people giving birth to alien babies, and bizarre tales of the supernatural). With this background, Jones without hesitation judges the observed phenomenon to be a ghost.Footnote 2
Why do Smith and Jones arrives at such different beliefs? It is, Koons argues, because of the background theories that they bring to their observations. Because such observation reports are dependent on theory, they are not really basic at all: their epistemic status depends upon that of the theories involved. But if this is true of perceptual beliefs, it is also true of allegedly ‘basic’ Christian beliefs. These, too, cannot be ‘properly basic’, but only ‘because no kind of belief can be properly basic’.Footnote 3
Non-inferential, ‘basic’ beliefs
While Koons's example is suggestive, the lesson he draws from it is misleading, since he fails to distinguish between different kinds of perceptual belief. It may be true that all perceptual beliefs are, in some sense, theory-laden, but the theories involved differ. While some are contested, others are uncontroversial: they represent the kind of ‘common-sense’ theory that guides our everyday behaviour. Anthropologist Robin Horton refers to this as ‘primary theory’ and suggests that it remains relatively constant throughout history and across cultures.Footnote 4 Beliefs based on this kind of theory are not incorrigible, but they can play a foundational role. They are beliefs we generally share and that can be referred to in order to settle higher-level theoretical disputes. The clash between Smith and Jones occurs, not at the level of these foundational beliefs, but at a higher level of theory-driven belief formation.
Two levels of belief formation
The point may be illustrated by returning to Koons's example. This presents as perceptual beliefs both Smith's description of the phenomenon as St Elmo's fire and Jones's attribution of the phenomenon to a ghost. In a sense, these are perceptual beliefs, since they arise from perception. But pace Koons, they are not helpfully described as ‘observation reports’. Observation reports are more accurately thought of as ‘low-level’ reports of perceptual experience, which do not involve the kind of theoretical assumptions that Smith and Jones are employing. It is not difficult to formulate a low-level report of this kind, on which these two observers could readily agree. They could agree, for instance, that ‘there is a glowing light around the church steeple’.
It is true that even low-level observation reports are theory-laden,Footnote 5 in so far as they require concepts in order to be formulated. This particular report, for instance, requires the concepts of both ‘light’ and ‘steeple’. Any use of such concepts will involve theory, at least of a folk-scientific kind.Footnote 6 But such theory is uncontroversial. More importantly, when observation reports are used to test theories, the common-sense theories upon which they rely are not those being tested. This is why scientists are able to settle disputes by reference to empirical evidence, despite the theory-ladenness of their observation reports.Footnote 7 They can do so because the theory with which their observation reports are laden is not that which is under dispute.
If any beliefs deserve to be described as ‘basic’ beliefs, they are those which correspond to low-level observation reports. The disagreement between Smith and Jones, however, is not at this level, but at the level of the causal attributions they make on the basis of their shared observations. As Koons notes, these causal attributions are clearly influenced by the high-level theories that each observer brings. What is the epistemic status of the ‘basic’ beliefs embodied in low-level observation reports? Do they enjoy a kind of prima facie justification? The answer will depend on one's broader epistemological views. But one could argue that we are justified in holding such beliefs until we are given reason to consider them false.
If there are basic beliefs of this kind, then Koons's argument against reformed epistemology fails. But there are other arguments that seem more promising. The first holds that Christian beliefs more closely resemble the causal attributions made by Smith and Jones. They involve high-level, controversial bodies of theory. One could argue, in other words, that at least these beliefs cannot play the foundational role played by low-level perceptual beliefs.Footnote 8 There is much to be said for this line of argument. One could develop it by pointing out the ways in which religious beliefs differ from perceptual beliefs. It seems, for example, that we could not survive if we were systematically to doubt our low-level perceptual beliefs.Footnote 9 In that direction lies madness. It follows that the question of entitlement does not arise with reference to such beliefs:Footnote 10 one may as well ask if we are entitled to the air we breathe.Footnote 11 This cannot be said of religious beliefs, which we are perfectly capable of living without.
A second way of criticizing reformed epistemology would attack Plantinga's suggestion that Christian beliefs have warrant, arguing that they are not, in fact, produced by a reliable mechanism (or, if they are, that it is not operating in favourable circumstances).Footnote 12 This, too, seems a promising line of argument.Footnote 13 In this context, however, I shall adopt a third approach. I shall concede that Christian beliefs may be ‘basic’ beliefs, in the sense of being non-inferential beliefs that we are entitled to hold ‘until further notice’. I shall also assume that Christian beliefs may have warrant, being produced by a reliable (although not infallible) mechanism.Footnote 14 What I shall argue is that even if this is true, whatever prima facie status Christian beliefs enjoy could be overridden by evidential considerations. The epistemic status of basic beliefs is not independent of their probabilistic or explanatory relations with our other, well-established beliefs.
The epistemic status of basic beliefs
In arguing for this view, I am faced with the problem of competing theories in epistemology. These focus on the question of what turns an instance of true belief into knowledge. The elusive factor has traditionally been referred to as ‘justification’. Justification has generally been understood in an internalist sense: the factors conferring justification, on this view, are factors within the agent's ken. More recently, authors such as Plantinga have come to talk about ‘warrant’, understood in an externalist sense. What confers warrant is the existence of a reliable belief-producing mechanism. To avoid prejudging the issue, I have used neither term, but have referred to the ‘epistemic status’ of beliefs. A true belief has positive epistemic status to the extent that it deserves to be counted as an instance of knowledge. My argument will be that the epistemic status of any belief is not independent of its evidential relations with other, well-established beliefs. In support of this claim, I will survey three views of what it is that confers epistemic status on a belief, showing that this conclusion holds no matter which view one adopts.
Strong evidentialism
A first view of the epistemic status of beliefs is strongly evidentialist. Not only does this view speak of ‘justification’, but it has a strongly internalist view of the factors conferring justification. It holds, in brief, that a subject cannot be epistemically justified except on the basis of considerations of which she herself is aware. A strong version insists that the subject also needs to be aware of the evidential force of these considerations. Perceptual beliefs, for example, may arise from a mechanism that reliably puts us in touch with the world. But on a strongly evidentialist view, this fact alone would not suffice to justify perceptual beliefs. Such beliefs would be justified only if the subject were aware of the mechanism concerned and had reason to think it is reliable.Footnote 15
There are ways in which such a position might be defended. It could be argued, for instance, that we all have a ‘folk-theory’ of perception, which is implicit in our everyday belief-forming practices.Footnote 16 This theory is a causal theory: we understand perceptual experiences to be somehow produced by the object perceived.Footnote 17 On this basis, we assume that perception is a reliable way of forming beliefs, at least in the appropriate circumstances. This is, for the most part, an unreflective assumption, which is ‘hard-wired’ into our way of interacting with the world. It is, nonetheless, no mere ‘act of faith’. Our confidence in the reliability of perception is tacitly confirmed by our everyday experience.Footnote 18 It is confirmed by the way in which the different modalities of sense perception reinforce one another and by the high degree of intersubjective agreement that perceptual beliefs enjoy.Footnote 19
On so strongly evidentialist a view, the mere fact that a belief is ‘basic’ and has warrant (in an externalist sense) will tell us nothing about its epistemic status. Having warrant will be of epistemic significance only if the subject is aware of the mechanism in question and has reasons to believe it is reliable. But if the epistemic status of our beliefs is based on reasons, it can be defeated by other reasons, which speak against it. In the case of perceptual beliefs, such reasons may not undermine our general confidence in the reliability of the process involved, but they may suggest that it is not functioning properly on this occasion. One such reason would be a lack of coherence between the deliverances of our senses and our other, well-established beliefs. It is important to note that a lack of coherence need not involve inconsistency, in the strict, logical sense. ‘Merely’ evidential relations would suffice. If we have reason to believe that what our senses are telling us is highly unlikely to be true, to this extent we have reason to doubt their deliverances.
Foundherentism
That would be one way of making the case. But the reader may be reluctant to accept so strongly evidentialist a view. She may be unconvinced by the idea that we all have some kind of belief in the reliability of sense perception and that this belief is justified. (It might be argued, for instance, that on this view children would have no justified beliefs, which seems wrong.) So here's a second way of approaching the question. It can be argued that so-called ‘basic’ beliefs – such as those formed on the basis of perception or memory – receive their epistemic status from two sources. The first is the fact that they are produced by a reliable mechanism,Footnote 20 while the second has to do with their explanatory or evidential relations with our other, well-established beliefs.
To illustrate this view, we could make use of Susan Haack's crossword puzzle analogy.Footnote 21 The relation of a belief to the reliable mechanism giving rise to it would correspond to that of a word in a crossword puzzle to its clue.Footnote 22 Similarly, the relation of a newly formed belief to our existing, well-established beliefs would correspond to that of the same word to the already discovered words in its vicinity. Even if the mere fact of being produced by a reliable mechanism gives a belief some degree of epistemic status, that status also depends on its evidential relation with our other, well-founded beliefs. Such a view combines elements from both foundationalist and coherentist views of knowledge. Using Haack's ugly but descriptive term, we can describe it as a form of ‘foundherentism’.
On this view, too, the initial status that a belief enjoys on the basis of its origins could be undermined by a lack of coherence with our other beliefs. This would not, of course, be true if the beliefs in question were produced by an infallible mechanism: one that never produced false beliefs. But so strong a form of reliabilism seems untenable, at least when it comes to beliefs about the external world.Footnote 23 There is nothing infallible about the process of forming perceptual beliefs. Even Plantinga, whose posited belief-producing mechanisms are of divine origin, holds that they need to be operating in ‘an appropriate epistemic environment’.Footnote 24 Outside of that environment, even a reliable mechanism may fail to operate as it should. One indication that it is failing would be that the beliefs it produces lack coherence with our other, well-established beliefs.
Moderate foundationalism
Of course, even foundherentism is controversial.Footnote 25 So it is important to note that one does not need to be a foundherentist to accept my argument. The foundherentist claims that coherence with other beliefs is part of what gives beliefs their epistemic status. But even those who deny this may accept that the epistemic status of a belief could be undermined by its evidential relations with our other beliefs. To return for a moment to Haack's crossword analogy, even if the ‘fit’ between a new word and the surrounding words had no positive epistemic value, a lack of fit should lead us to suspect we have not correctly interpreted the clue. In a similar way, the status that a so-called ‘basic’ belief would normally enjoy can be undermined if it is rendered improbable by our other beliefs. All that is required to accept this conclusion is what Robert Audi calls a ‘moderate foundationalism’.Footnote 26
Defending the claim
My criticism of Plantinga's reformed epistemology requires nothing more than this. It requires nothing more than the idea that even a so-called ‘basic’ belief can be defeated by its evidential relations with our other, well-established beliefs. This view seems defensible given quite different conceptions of knowledge. It is also consistent with our pre-analytical intuitions about belief.
Take, for instance, my cat, whose name is Minerva. Normally she sits on my desk while I write, or lounges on the couch, in the sun. Let's say, however, that one morning I glance out my second-floor study window and see what appears to be Minerva, floating in mid-air, looking in at me. If I were to form the belief that the cat is, in fact, out there, floating in mid-air with no tangible means of support, would that belief be justified? This state of affairs is conceivable: ‘logically’ possible, if you like. A belief of this kind would also arise from a reliable mechanism, namely vision, operating in apparently favourable circumstances. If the belief in question would lack justification, it is because of its relation to other, well-founded beliefs that I have. Let's say that the laws of physics, as currently understood, do not make this scenario strictly impossible.Footnote 27 But they do make it extraordinarily unlikely. This fact should lead me to favour an alternative hypothesis, such as that of hallucination or optical illusion.
It might be argued that this is an extreme example, that our so-called ‘basic’ beliefs would rarely, if ever, face an objection of this kind. But the degree of counter-evidence required to undermine a non-inferential, ‘basic’ belief will depend on the degree of justification or warrant that the latter already enjoys. I say ‘justification or warrant’ since both come in degrees. In the case of justification this seems clear, since beliefs can have differing degrees of evidential support. But it is also true for warrant. Any account of the causal mechanisms that confer warrant will include a description of the circumstances in which they need to be operating.Footnote 28 To the extent that those circumstances are less than ideal, to that extent such a belief will lack warrant. If this is the case, it may not require much counter-evidence to call it into question.
This can be illustrated by reference to one of Plantinga's own examples. I am playing poker and see that I have drawn an inside straight. The belief that I have done so, Plantinga argues, ‘doesn't depend, for its warrant, on its being appropriately probable on the rest of what I believe; it has a quite different source of warrant, namely, perception’.Footnote 29 It follows, he continues, that it is not defeated by the fact that drawing such a combination of cards is an improbable event. I agree that such a belief – one that seems antecedently improbable – could arise from a reliable belief-forming mechanism and thus have warrant. But its epistemic status will depend on the strength of the warrant. Let's say, for example, that I am playing poker in a dark, smoky room, and can hardly see the cards. In this situation the epistemic status of my belief that I have drawn an inside straight may be low. In these circumstances, the question of prior probability becomes important: I may wish to move into better light and check my hand before making my next bid.
Against reformed epistemology
These observations allow us to identify the central problem with Plantinga's reformed epistemology. If we are interested in what I am calling the ‘epistemic status’ of a belief, talk of warrant is not sufficient. Even if we were to concede that the existence of warrant confers a positive prima facie status on a belief, our final judgement about its status will depend on its evidential relation to our other beliefs.
Plantinga on justification
It is true that Plantinga does not speak exclusively of warrant; he also discusses the question of justification. But when speaking of the justification of Christian belief he uses the term in a weak sense. The justification he speaks of is ‘deontic’: it has to do with whether an individual is within her epistemic rights in believing as she does. This allows him to argue that it is a condition that is easily fulfilled, even in the case of the Christian.
If your belief is a result of the inward instigation of the Holy Spirit, it may seem obviously true, even after reflection on the various sorts of objections that have been offered. Clearly, one is then violating no intellectual obligations in accepting it. No doubt there are intellectual obligations and duties in the neighbourhood; when you note that others disagree with you, for example, perhaps there is a duty to pay attention to them and to their objections, a duty to think again, reflect more deeply, consult others, look for and consider other possible defeaters. If you have done these things and still find the belief utterly compelling, however, you are not violating duty or obligation.Footnote 30
But all that this requires is that the Christian be acting in good faith. Someone could be justified in this ‘deontic’ sense even if there were no truth-indicative connection between the perceptual experience and the object to which it refers. What we might call ‘objective’ justification requires something more. It requires a person to have, not just reasons, but good reasons for her beliefs, that she form those beliefs in ways that are objectively appropriate.Footnote 31
Plantinga is not unaware of the question of objective justification. Indeed he discusses one version of it, namely that defended by William Alston. On Alston's view, a person is justified in believing that p if she does so on grounds that make it ‘objectively likely’ that p is true.Footnote 32 Those grounds can be other beliefs, from which what one believes is inferred. But they may also involve non-inferential relations with what is believed or simple facts about the propositions in question. The likelihood in question, Alston insists, is an objective probability, based on ‘the lawful structure of the world’,Footnote 33 not merely my existing beliefs about it.
Plantinga rejects this view of objective justification on two grounds. He argues, first of all, that ‘several important sorts of beliefs – a priori belief and memory belief in particular – do not seem to have a ground in Alston's sense at all’.Footnote 34 Such beliefs are neither inferred from other beliefs nor based on experience, at least in the sense of ‘sensuous experience or phenomenal imagery’.Footnote 35 His second objection is that on Alston's account, ‘any grounded belief in any necessary proposition p is justified’, since ‘the objective conditional probability of [a necessary proposition] p on any proposition will be 1’.Footnote 36 This means that if the proposition ‘God exists’ is necessarily true, as many theists have affirmed, then my belief in God would be objectively justified even if I actually believe for reasons that do not seem reasonable, perhaps ‘just to please my friends, or because I am brainwashed or hypnotized’.Footnote 37
These arguments fail to undermine Alston's conception of objective justification. First, even if a priori and memory beliefs are not based on ‘sensuous experience or phenomenal imagery’, they may be grounded in some other kind of experience. Indeed elsewhere in the same work Plantinga effectively concedes this point, writing about another kind of ‘phenomenal experience’ that has evidential force and which forms the basis of a priori and memory beliefs.Footnote 38
Plantinga's second argument is no more successful, although it is more difficult to see why. How should we understand Alston's idea that the grounds for belief should make it ‘objectively likely’ that p is true? Let's call the grounds for belief the ‘evidence’, e. It is not enough that the probability of p, given e, be greater than a certain threshold, say 0.5. It should also be the case that e specifies the reason why the probability of p is as high as it is. In other words the probability of p, given e, should be higher than the probability of p, given not-e. The problem in the case of a necessarily true proposition is that e will make no difference. The probability of p will always be 1. But all this shows is that if we interpret Alston's account in terms of the standard probability calculus, it will not work for necessarily true propositions, a fact that he freely admits and discusses elsewhere.Footnote 39 It does not show that belief in such a proposition would be justified on any grounds whatsoever, as Plantinga suggests.Footnote 40
An independence thesis
Let me try to set out more formally Plantinga's view of the epistemic status of properly basic beliefs. His vigorous rejection of what he calls ‘classical foundationalism’ can be misleading.Footnote 41 It can lead us to believe that he is rejecting foundationalism tout court. In fact, however, his reformed epistemology represents a particularly strong form of foundationalism. It holds (in the words of one of his commentators) that a Christian's ‘basic’ belief ‘has its epistemic value independently of its relation to other beliefs’.Footnote 42 It is this that enables Plantinga to maintain that Christian belief remains undefeated by ‘merely’ evidential considerations.
We may sum up this view of the status of properly basic beliefs as an independence thesis (IT).
IT: A properly basic belief has epistemic status independently of its evidential relations with other, well-established beliefs.
If epistemic status is thought of simply in terms of warrant, then there is some truth in the independence thesis. Whether a belief has warrant does not depend on its evidential relations with other beliefs; it depends on the existence of a reliable cognitive mechanism that gives rise to the belief. But what I have argued is that the mere existence of a reliable cognitive mechanism is not enough. It does not settle the question of epistemic status, since even a warranted belief may be defeated by evidential considerations.
I have formulated IT in my own words because it difficult to find an unambiguous statement of this position within Plantinga's work. But his claim that the Christian faith remains undefeated by evidential arguments seems to require it. Take, for instance, his response to the argument from religious diversity. Does it not seem improbable that among the variety of the world's religions only Christianity will be true? Plantinga responds that it may seem improbable, even to the Christian, against the background of her other beliefs. But this objection
(like so many others) seems to make sense only if the believer, to be rational, must hold her Christian beliefs on the basis of their relations to other beliefs she has – or, at any rate, only if those Christian beliefs are probable with respect to those beliefs. One of the main burdens of this book, however, is that the believer can be perfectly rational in acquiring some of her beliefs in the basic way – not on the basis (probabilistic or otherwise) of other beliefs.Footnote 43
There is both a truth and a falsity here. The truth is that a properly basic belief is not arrived at on the basis of inference from other beliefs. The falsity is the suggestion that such a belief cannot be defeated by its evidential relations with our other beliefs.
Once again, the latter is only a suggestion, but it is a suggestion Plantinga makes repeatedly. It is, perhaps, clearest in his response to Paul Draper's evidential argument from evil. This holds that the distribution of pleasure and pain among sentient beings is better explained by what Draper calls the ‘hypothesis of indifference’ than by theism.Footnote 44 Plantinga argues that even if the argument were sound, the most it would show is that Christian belief is ‘evidentially challenged’.Footnote 45 But, he continues, this does not constitute a defeater for a belief that has independent warrant. He illustrates his point with a number of examples, of which the following two are representative.
My friend has a cat named Maynard; I believe that Maynard is a cat and also (as my friend reports) that Maynard likes cooked green beans; the latter, however, is much more likely on the … alternative hypothesis that Maynard is a Frisian, or possibly a Frenchman; so the belief that Maynard is a cat is evidentially challenged for me. I believe (naturally enough) that you are a human being; you and I are on a walk in the woods, however, so I also believe that you are in a forest; of course that proposition is vastly more likely on the … alternative hypothesis that you are a tree; so the belief that you are a human being is evidentially challenged for me.Footnote 46
It follows, Plantinga argues, that a properly basic belief is not defeated by the fact that there exists an evidentially well-supported hypothesis that implies it is false.
Let me note, in passing, that these are very odd examples. Take the case in which you and I are walking in the woods. I can see you; you are talking with me; I recognize you as a person with whom I have interacted before; a moment ago I may have assisted you in stepping over a log. In these circumstances I have naturally (and non-inferentially) formed the belief that you are a human being. Certainly, the fact that you are in a forest is more likely on the hypothesis that you are a tree, but that hypothesis has nothing else to recommend it. It would explain no other aspect of my experience. If I needed an explanation of my experience, by far the most plausible would be that you are a human being. So there is no significant sense in which my belief is ‘evidentially challenged’. Similar considerations may be applied to the case of Maynard the cat. That he is a Frisian, or a Frenchman, might explain his liking for green beans, but it certainly does not explain his cat-like shape, the sound of his miaow, or the warmth of his fur. This, too, hardly represents a situation of evidential challenge.
There are, however, circumstances in which non-inferential, ‘basic’ beliefs may be evidentially challenged. If I glanced out of my window on what had earlier been a hot summer's day and saw what appears to be snow falling, I would certainly be astonished. Such an event seems extraordinarily unlikely. Should I nonetheless believe it is snowing? If I went outside and felt the cold, watched the falling substance melt as it lands on my hands, and found others remarking on this extraordinary event, then the evidence in support of the snow-belief would simply outweigh its antecedent improbability. If, on the other hand, I went outside and felt it was still warm and saw that no-one else had noticed what seemed to me to be snow, then the belief that it is snowing would have little evidential support. In these circumstances I would be obliged to consider an alternative hypothesis. Perhaps I am dreaming, hallucinating, or experiencing an optical illusion.
Weighing the evidence
What is required, in other words, is a weighing of the evidence. Plantinga, however, appears to reject the idea that there is any weighing required. At one point he does ask what would follow if ‘the rest of what I believe offered evidence against theism and none for it’. His answer is, ‘Not much’.Footnote 47 The implication is that such a state of affairs would not undermine the epistemic status of a Christian's belief, given that the latter is properly basic. But whether this is the case will depend on the initial status of one's Christian belief and how strongly the other evidence weighs against it.
I have argued that Plantinga's appeal to warrant addresses only one of the factors that contribute to the epistemic status of a belief. It neglects the question of objective justification and overlooks the fact that a warranted belief can be defeated by evidential considerations. But his appeal to warrant is problematic for another reason, since it rests on a claim that is practically trivial. Plantinga famously suggests that if what Christians believe is true, then it has warrant.Footnote 48 But what does this mean? What we call ‘the Christian faith’ is, in fact, a set of beliefs, some of which (regarding the sensus divinitatis and the work of the Holy SpiritFootnote 49) are beliefs about warrant. It follows that to say ‘if Christian belief is true, it has warrant’ is to make a claim of the form ‘if (p and q), then q’ or ‘if it is Tuesday and it is raining, then it is raining’. So not only does reformed epistemology fail to avert evidential challenges, but the insight upon which it is based is of little philosophical interest.
The central argument of this article has been that in so far as Plantinga's reformed epistemology relies on an independence thesis, it represents an implausibly strong form of foundationalism. A more plausible form of foundationalism would accept that even a properly basic belief can be defeated, not merely by showing it to be inconsistent with other, well-established beliefs (as in the case of a logical argument from evil), but also by its evidential relations with those beliefs. Whether a properly basic belief is defeated in this way will depend on the degree of justification that it already enjoys. Plantinga, it might be noted, has given us no reason to believe that non-inferential, ‘basic’ Christian beliefs enjoy any degree of (objective) justification at all.
It may be, of course, that such reasons can be given. The Christian's beliefs may enjoy some degree of justification. That justification might even be a prima facie justification, comparable to that enjoyed by perceptual beliefs. (I am not convinced it is, but I have conceded this possibility for the sake of the argument.) Even if this were true, however, the Christian would not be entitled to ignore the evidence that counts against her beliefs. The problem with theism may well be that ‘belief in God just does not hang together very well with so many of the other things that we have come to believe about the world and about ourselves’.Footnote 50 This remains a problem even if, as Plantinga argues, Christian faith is a form of warranted, basic belief.
Conclusion
Let me conclude by returning to Jeremy Koons's argument. In a sense, I am in agreement with the idea that no belief is epistemically basic, if ‘being basic’ means being justified independently of its probabilistic or explanatory relations to our other beliefs. But I have argued that there may well be beliefs that are ‘basic’ in the sense of playing a foundational role in the structure of our knowledge. My argument against reformed epistemology rests on the idea that the question of the epistemic status of a belief is not settled by pointing out that it is produced by a reliable mechanism. That status also depends on the evidential relations between the belief in question and our other, well-founded beliefs. Evidential considerations may defeat whatever degree of prima facie justification, or warrant, a basic belief enjoys. While not entirely defeating reformed epistemology, this argument does undercut much of its apparent utility, from the theist's point of view. In particular, it restores the force of ‘merely’ evidential arguments from evil.
