The Theistic Preconditions of Knowledge: A Thumbnail Sketch

James N. Anderson

Version 1.0

One of the distinctive claims of Van Tilian apologists is that human knowledge presupposes the existence of God; therefore, since we know at least some things, it follows that God must exist. In recent months, while surfing the blogosphere, I have encountered several times the insinuation that Van Tilians invariably forward this claim without any argument. For example, one commenter going by the moniker ‘Yo Mama’ (not to be confused with Yo-Yo Ma, the acclaimed Chinese-American cellist) remarked that she had never come across a presuppositionalist who had offered argumentative support for the claim that knowledge presupposes God. In truth, I suspect this tells us more about Yo Mama’s diet of reading than about the efforts of presuppositionalists to defend their arguments. Reasoned support for the claim can be found in the writings of Cornelius Van Til, John Frame, Greg Bahnsen (see also his lectures and debates, particularly his debate with Edward Tabash), and Michael Butler. Similar arguments have been formulated by Alvin Plantinga, Dallas Willard, and Victor Reppert; and while these Christian philosophers would not consider themselves ‘presuppositionalists’ in the conventional sense, their arguments have often been endorsed as supportive of presuppositionalist claims. (See the bibliography below for references.)

However, rather than merely inviting sceptics like Yo Mama to ‘get out more’, I want to take the opportunity to offer one particular line of argument in support of the claim that theism is a precondition of knowledge. I will first argue that there is no place for knowledge within a naturalistic metaphysic, before considering some prominent non-naturalist alternatives. I conclude that only a theistic metaphysic seems to have the features needed to underwrite the preconditions of human knowledge.

Knowledge and Normativity

Let us begin by analysing the notion of knowledge to elucidate some of its essential features. (I will be focusing exclusively here on propositional knowledge: knowledge that such-and-such is the case.) It is almost universally accepted by contemporary epistemologists that ‘truth’ and ‘belief’ are necessary components of knowledge. I cannot know that p if I do not believe that p (for what sense does it make to say, “I know that p even though I don’t believe that p”?). Moreover, even if I believe that p, I cannot know that p if p is false.

Now, one might pause here and observe that even these two basic features of knowledge—belief and truth—are problematic for metaphysical naturalists. Naturalism, as typically defined, is committed to the thesis that only ‘natural’ entities exist, i.e., entities which can be described (at least in principle) in terms of the methods and inventories of the natural sciences (e.g., physics and chemistry). All phenomena are thus explicable (in principle) in objective scientific terms. This applies even to mental phenomena, which must either be reduced to the physical (i.e., explained in terms of more fundamental physical phenomena) or eliminated (i.e., explained away altogether). But it has been argued that beliefs possess intrinsic features, such as subjectivity and intentionality, which cannot be reduced to the physical or eliminated. Similarly, the concept of truth—understood as a property of certain thoughts or propositions or statements—is not susceptible to analysis in terms of purely natural qualities (such as mass, electromagnetic force, or electrical charge). Thus there seems to be no respectable place for ‘beliefs’ and ‘truths’ in a naturalist ontology. Indeed, the manifest difficulties faced by those who try to carve out a place for them have driven some thoroughgoing naturalists (e.g., Paul and Patricia Churchland) to doubt whether there really are such things after all.

But leave these matters aside for now, because I want to focus on a third essential component of knowledge. I’ve noted that truth and belief are necessary for knowledge, but they are clearly not sufficient for it. It is possible to believe that p, and for p to be true, without actually knowing that p. For example, I might believe that it will be sunny on my birthday in a fortnight’s time, simply because I’m an incurable optimist; but even if that belief later turned out to be true, no one would concede that I knew it would be sunny two weeks in advance.

So a third ingredient is needed for knowledge, an ingredient commonly labelled ‘justification’ or ‘warrant’. (For consistency’s sake, I will hereafter use the term ‘warrant’ to refer to this third ingredient.) Contemporary epistemologists have vigorously debated precisely what constitutes ‘warrant’, but fortunately there is no need to take sides in these debates in order to defend the point I want to make here. For there is a common intuition behind all analyses of warrant to the effect that a true belief must be formed or held in the right way, or in an appropriate way, in order to count as knowledge. A warranted belief cannot be formed or held in just any old fashion. There are right or appropriate ways and there are wrong or inappropriate ways. As an example, suppose I come to believe that it is raining outside; suppose further that it is, in fact, raining outside. If this belief is formed on the basis of perception (e.g., I can see and hear the rain through an open window), then the belief is very likely warranted; but if this belief is formed on the basis of a superstitious conviction that it always rains on the days I forget to bring my umbrella, then the belief will not be warranted. The difference is that in the former case the belief is formed in a fitting or appropriate manner, while in the latter case it is not.

Careful reflection on the concept of knowledge in general, and on paradigm cases of knowledge, make it clear that this notion of ‘epistemic rightness’ or ‘epistemic appropriateness’ is an essential feature of knowledge. But observe that this notion is clearly a normative one: it pertains to how beliefs ought to be formed or held (in order to count as knowledge), rather than how beliefs are formed or held. It is not a descriptive notion, but a prescriptive one. It implies that there are epistemic norms which determine (in part) whether or not one’s belief that p is actually knowledge that p.

That the concept of knowledge has an essentially normative aspect, and thus there are such things as epistemic norms (if there is such a thing as knowledge), is a point widely recognised by contemporary epistemologists. For example, Jaegwon Kim writes:

[Epistemic] justification manifestly is normative. If a belief is justified for us, then it is permissible and reasonable, from the epistemic point of view, for us to hold it, and it would be epistemically irresponsible to hold beliefs that contradict it. . . . Epistemology is a normative discipline as much as, and in the same sense as, normative ethics. (Kim, 1988, p. 383, emphasis original)

Knowledge and Naturalism

The fact that there is such a thing as epistemic normativity has interesting implications. In the first place, it poses a serious problem for metaphysical naturalism, for there is no place within a thoroughgoing naturalism for any irreducible normativity. According to the metaphysical naturalist, all phenomena are ultimately explicable in scientific terms (if explicable at all), but science is a purely descriptive discipline. Science describes rather than prescribes. It tells us how things are, as a matter of empirical fact; it has nothing to tell us about how things ought to be. As Alvin Plantinga has remarked:

[Naturalism’s] Achilles’ heel (in addition to its deplorable falsehood) is that it has no room for normativity. There is no room, within naturalism, for right or wrong, or good or bad. (Plantinga, 1998, p. 356, emphasis original)

So naturalism, as a metaphysical position, cannot accommodate the notion of right or wrong ways to form or hold beliefs. Consequently, it cannot accommodate the notion of epistemic warrant. In short: if we know anything at all, then naturalism must be false.

(Remarkably, some naturalists have conceded this point—at least, in a fashion. W. V. Quine and his followers in the school of ‘naturalized epistemology’ have argued that the classical [i.e., commonsense] conception of knowledge should be abandoned altogether, and the normative discipline of epistemology dropped in favour of the descriptive discipline of psychology. In short, we ought to think a whole lot less about how we ought to think and a whole lot more about how we do in fact think. If the reader detects a whiff of self-referential absurdity in this radical recommendation, he is not alone.)

In the remainder of this section, I will press home this general point by considering some different kinds of normativity to see whether epistemic normativity can be analysed in terms of some other (presumably more fundamental) kind of normativity. But before doing that, I want to briefly consider one widespread account of epistemic warrant that may be the first port of call for many common-or-garden naturalists.

It is often claimed that what distinguishes knowledge from mere true belief is that in the former case one possesses adequate reasons for thinking the belief in question to be true. Intuitive though this analysis may seem, there are several difficulties with it. First, it falls foul of Gettier-style counterexamples: it isn’t at all difficult to generate hypothetical scenarios in which one has adequate reasons for holding belief B, but it turns out that B is true only by good fortune (i.e., it is true for reasons other than those held by the believer). Second, this account invites an infinite regress. Presumably one’s reasons for B must themselves be warranted if B is to be warranted. That is to say, B cannot be warranted on the basis of reason R unless one is also warranted in believing R to be true. But if epistemic warrant simply consists in having adequate reasons for beliefs, then one will only know that p if one has reasons for believing that p, and reasons for those reasons, and reasons for those reasons, and so on. Since possessing an infinite non-circular series of reasons is impossible, none of our beliefs could ever be warranted on this account. This is precisely why most contemporary epistemologists accept that at least some of our beliefs (e.g., perceptual beliefs) must be properly basic: epistemically warranted, but not by virtue of our having reasons for taking them to be true.

So this account of epistemic warrant—an account that takes knowledge to consist in having true beliefs based on good reasons—faces considerable objections. Indeed, as a general analysis of knowledge, it is fatally flawed. But even if were not, it wouldn’t offer the metaphysical naturalist a solution to the problem of accommodating epistemic normativity, because on closer inspection the account simply off-loads the normativity in question to a different concept rather than explaining it (or explaining it away) in purely naturalistic terms. For the concept of ‘having a reason’ is no less freighted with epistemic normativity than the concept of ‘having epistemic warrant’. Consider: if belief X gives one adequate reason for belief B, but belief Y does not, isn’t this just because the relationship between X and B meets certain epistemic norms or standards that the relationship between Y and B fails to meet? How could one account for the epistemic difference between these two belief-pairs without reference to some normative criteria?

Commenting on the project of ‘naturalizing’ epistemology, Kim writes:

The implicit requirement has been that the stated conditions must constitute ‘criteria’ of justified belief, and for this it is necessary that the conditions be stated without the use of epistemic terms. Thus, formulating conditions of justified belief in terms such as ‘adequate evidence’, ‘sufficient ground’, ‘good reason’, ‘beyond a reasonable doubt’, and so on, would be merely to issue a promissory note redeemable only when these epistemic terms are themselves explained in a way that accords with the requirement. (Kim, 1998, p. 382, emphasis original)

So the idea that the normativity of knowledge can be explicated simply in terms of having reasons for one’s beliefs in no way removes the problem for the naturalist. (Similar considerations can be applied to the idea that epistemic warrant consists in having ‘evidence’ for one’s beliefs.) Let us therefore turn our attention to various ways in which naturalists might account for epistemic normativity in terms of some other species of normativity.

First of all, there is the familiar notion of deontological normativity: the normativity of ethical duty. It is at least plausible to think of epistemic norms as a subset of ethical norms. On this view, I know that p if and only if p is true and I have flouted no relevant ethical obligations in believing that p (e.g., the obligation not to deliberately avoid considering any evidence against p). Unfortunately, even if this analysis were defensible (and it faces numerous difficulties), it would be of no use to naturalists. For as the ethicist Richard Taylor (among others) has argued, the notion of genuine moral obligation only makes sense in a theistic context. Taylor writes:

A duty is something that is owed, something due, and to be obligated is, literally, to be bound. But something can be owed only to some person or persons. There can be no such thing as a duty in isolation, that is, something that is owed but owed to no person or persons. (Taylor, 1985, p. 75, emphasis original)

[T]he concept of moral obligation [is] unintelligible apart from the idea of God. The words remain, but their meaning is gone. (Ibid., p. 84)

The naturalist may be tempted to reply here that we do have genuine obligations to some persons, but those persons are merely human persons; that is, we humans are bound to certain moral obligations because we impose duties on one another (perhaps by way of social contracts and suchlike). This amounts to saying that ethical norms are essentially conventional and contingent on human cultures and societies. But even if this move satisfies the naturalist ethicist, it won’t save the naturalist epistemologist because it would follow (on the view that epistemic norms reduce to ethical norms) that what counts as knowledge is contingent on human opinion. So it would be possible, at least in principle, for there to be a society in which beliefs formed on the basis of reliable perceptual faculties are not warranted while beliefs formed on the basis of wishful thinking are warranted. (Indeed, perhaps with sufficient propagandist efforts we could bring about such a society.) But clearly this can’t be right. Human opinion is subject to epistemic norms, not constitutive of them.

I mentioned in passing that the claim that epistemic norms are merely ethical norms is open to some serious criticism. Nevertheless, it might be argued that epistemic normativity is analogous to ethical normativity: similar in many ways, but not identical. Perhaps it does make sense to think of epistemic warrant in terms of ‘intellectual blamelessness’, but not quite the sort of ethical blamelessness that, when eschewed, invites a slap on the wrist or some other fitting reproof.

However, this analogized deontologism is hardly a lifeline for the naturalist, for if a naturalist ontology has no place for the notions of ethical duty, guilt, blamelessness, etc., then we have every reason to think it has no place for analogues of duty, guilt, blamelessness, etc. It might be countered that compatibility with naturalism is precisely one of the points of difference between ethical duties and epistemic duties. But this would be a patently ad hoc move with no argumentative merit.

Where else might the naturalist turn? Another possibility is teleological normativity: the normativity of purpose and design. Alvin Plantinga and Michael Bergmann have argued (persuasively, in my judgement) that the kind of normativity involved in epistemic warrant is that of proper function. On this analysis, only beliefs formed by properly functioning cognitive faculties can be warranted. Yet this analysis doesn’t hold much promise for the naturalist either. If some faculty or system has a proper function, then it follows that it was intended or designed to function in certain ways as opposed to other ways; but surely this flies in the face of metaphysical naturalism, one implication of which is that the human brain is not the product of intention or design, but rather the product of unintentional, undirected, unintelligent natural processes (whatever variety of evolutionary naturalist aetiology one favours). So accounting for the normative component of knowledge in teleological terms is not a live option for naturalists.

A naturalist might reply (perhaps with a degree of desperation) that strictly speaking his metaphysic is compatible with the claim that human cognitive faculties are designed, so long as the designer of those faculties is also taken to be a purely natural organism (e.g., an enterprising extra-terrestrial). I readily grant the point. But this response doesn’t resolve the problem of accounting for epistemic normativity within a naturalistic framework; it merely pushes it back one species. The important thing to recognize is that naturalism has no place for an ultimate and irreducible teleology, whether located in our solar system or some far-flung corner of the cosmos. A proper function epistemology can only be accommodated by naturalism if teleological normativity can be reduced to some non-teleological feature of the natural universe. The burden lies with the naturalist to explain how that is possible.

At this point, the naturalist may consider it more productive to approach the problem from behind, as it were, by first asking whether there are any kinds of normativity that can be accommodated by naturalism and then asking how epistemic norms might be accounted for in those terms. For example, it might be thought that natural laws are normative in a certain sense. After all, locutions such as “The ball ought to fall to the ground when I release it” appear to be perfectly intelligible. Here, however, appearances are deceiving. For the ‘ought’ in such cases is merely a manner of speaking; specifically, a manner of speaking about expectations (“The ball is expected to fall to the ground when I release it”). And expectations, understood as probabilistic predictions based on generalizations from past observations, can be explicated in purely descriptive terms without any hint of normativity.

So the ‘normativity’ of natural law is a mere pseudo-normativity and is therefore unsuitable as a framework for explaining the normativity of knowledge. Indeed, this can be seen quite clearly by reflecting on what it would mean to reduce epistemic norms to natural laws. Since naturalism is committed to the claim that all human thinking operates in purely naturalistic terms and thus strictly in accordance with natural laws, it would follow that all human thinking necessarily operates in accordance with epistemic norms. But as anyone who has watched so-called ‘reality television’ will appreciate, this is plainly false. Indeed, on such an analysis it would be impossible to have any unwarranted beliefs.

Another tempting option for the naturalist might be to appeal to statistical norms, that is, norms based on the probabilistic characteristics of naturalistic processes. The obvious route here would be to adopt some form of epistemic reliabilism, according to which a belief is warranted if and only if it has been produced by a cognitive process or faculty that gives rise to predominantly true beliefs. On this analysis, beliefs formed via sense perception would be warranted, while beliefs formed via wishful thinking would not (given that the former is a reliable source of beliefs and the latter is not).

Unfortunately, merely adopting a form of epistemic reliabilism will not alleviate the difficulty of accounting for knowledge within a naturalistic metaphysical framework. In the first place, the naturalist needs to explicate the crucial notion of cognitive reliability in purely naturalistic terms, which in turn requires naturalistic accounts of ‘belief’ and ‘truth’. (How does one explicate the notion of a ‘belief’ being ‘true’ in terms of nothing other than the entities and properties studied by physicists and chemists?)

But problems of naturalistic reductionism aside, reliabilism as such is inadequate as a theory of warrant. For it is easy to generate counterexamples which show that cognitive reliability, while perhaps necessary for warrant, is by no means sufficient. Here is one simple example. Imagine a person—let’s call him Sam—who is unwittingly exposed to a high dose of cosmic rays which dramatically alters the structure of his brain. As a result, on random occasions Sam spontaneously forms the strong impression that certain very large numbers are prime numbers. Furthermore, nine out of ten times these impressions turn out to be true. Sam knows nothing regarding the origin of his beliefs about these numbers, nor does he have any means of verifying them. Nevertheless, the process by which these beliefs are formed is a very reliable one; it gives rise to predominantly true beliefs. It should be obvious, however, that these beliefs are not in the least bit warranted for Sam. He doesn’t genuinely know that the numbers in question are primes.

Alvin Plantinga provides another counterexample to epistemic reliabilism: ‘The Case of the Epistemically Serendipitous Lesion’. Plantinga invites us to imagine a man who develops a rare brain lesion that instigates a number of cognitive processes, most of which produce predominantly false beliefs. However, by sheer dumb luck, one of these cognitive processes results in this man forming the belief that he has a brain lesion. Moreover, we can stipulate this particular cognitive process produces no other beliefs (whether true or false). So the man’s belief that he has a brain lesion arises from a cognitive process that is statistically reliable—indeed, it is 100% reliable. Moreover, there is a relevant causal connection between the belief and the belief’s object: the belief about the brain lesion is actually caused by the lesion. So the causal requirement imposed by some more sophisticated versions of reliabilism is also satisfied. Yet despite these reliabilist conditions being met, it should be clear that the man’s belief is not epistemically warranted; even though the belief is true, the man does not really know that he has a brain lesion.

The clear implication of such counterexamples is that epistemic warrant requires some ingredient in addition to reliability. What is needed here is a further condition or constraint that rules out cases where a belief has been formed by a process or faculty that is only reliable by mere accident. The obvious remedy is to impose a teleological condition: the reliability must be intentional or purposeful. But as we have already noted, a genuine teleological normativity (with respect to human cognition) cannot be countenanced by a thoroughgoing metaphysical naturalism.

A slightly different approach to remedying the shortcomings of epistemic reliabilism simpliciter is offered by virtue epistemology. According to virtue epistemologists, for a belief to be warranted it must not only have been formed by a reliable cognitive process or faculty, but also by the exercise of an intellectual virtue. The notion of intellectual virtues has been defined in various ways by different virtue epistemologists, but one common understanding is that such virtues may be thought of as good intellectual character traits. Examples of such traits would include honesty, reflectiveness, fair-mindedness, thoroughness, and curiosity.

Does virtue epistemology lend itself to a naturalistic account of knowledge? On the face of it, it is difficult to see how metaphysical naturalism can accommodate the existence of normative objective virtues (including intellectual virtues); that is, virtues which do not originate in mere human opinion. Why would one configuration of atoms be more virtuous than another configuration merely on account of its physical properties and relations? But in any case, if one considers how the notion of intellectual virtues needs to be spelled out in order to provide an adequate theory of warrant, we find the naturalist faced with a dilemma. On the one hand, if intellectual virtues reduce merely to cognitive dispositions or processes which typically result in the formation of true beliefs, regardless of the subjective motivations of the believers, then a virtue theory of warrant offers no advance over reliabilism, because it falls foul of the sort of counterexamples raised by Plantinga and others. On the other hand, if the exercise of an intellectual virtue necessarily involves some commendable intention on the part of the believer—say, an intention to attain some worthy epistemic goal, such as the acquisition of truth and the avoidance of falsehood—then we seem compelled again to introduce some teleological dimension to knowledge (e.g., the notion of good epistemic ends). Yet if epistemic relativism is to be avoided then the relevant epistemic telos (whatever it turns out to be) must be objective and not dependent on human feelings or opinions. In other words, it cannot be a telos of our own choosing. But as I noted earlier, the idea of objective teleological normativity should be anathema for any self-respecting naturalist.

If the naturalist wishes to defend some variant of a reliabilist theory of warrant, he will have to address those counterexamples which clearly show that cognitive reliability alone is insufficient for warrant. The most promising approaches to addressing the shortcomings of reliabilism simpliciter are to supplement it with further conditions expressed in terms of cognitive proper function or intellectual virtues. But in order to be successful these supplements must appeal to kinds of normativity that are off-limits for metaphysical naturalism.

To summarise the argument thus far: Any adequate analysis of knowledge must involve some notion of objective epistemic normativity, but metaphysical naturalism has no place for objective norms, that is, norms not dependent on human consciousness. If epistemic normativity can be explained in terms of some other species of normativity, the two most promising candidates are deontological normativity (i.e., norms of duty) and teleological normativity (i.e., norms of purpose or function). Neither of these, however, is welcome in a purely naturalistic universe.

Knowledge and Theism

I have argued that knowledge presupposes the falsity of metaphysical naturalism; therefore, knowledge presupposes the truth of metaphysical supernaturalism (in the broad sense). But this alone does not secure the conclusion that knowledge presupposes theism, for there are other non-naturalist ontologies to consider. In this section, therefore, I will briefly review some major alternatives to theism and assess how each might fare in accounting for epistemic normativity.

As I noted above, only two of the various analyses of epistemic warrant considered earlier would appear to be live options (given the stated shortcomings of the other accounts): deontological epistemology (which would include quasi-deontological notions of epistemic norms) and teleological epistemology. According to these two approaches, epistemic normativity is cashed out either in terms of intellectual duty (or something in that conceptual neighbourhood) or in terms of proper cognitive function (or proper cognitive ends).

It isn’t difficult to see that both of these analyses sit quite comfortably within a theistic framework. God’s character and commands are normative for us by virtue of the fact that we are his creatures and created to reflect the divine image. As Richard Taylor notes, it is natural to think that if there is a God then we are bound to him in an ethical sense: we have obligations toward him (including intellectual obligations). Similarly, certain human epistemic practices are virtuous because they accord with those goals and characteristics valued by God—truth-directedness, consistency, impartiality, prudence, etc.—and his general intentions for human existence. And the notion of cognitive proper function makes perfect sense on the view that humans (including our noetic faculties) have been created by God (whatever view one takes of the process of creation).

But what about the alternatives? Consider first whether a modern-day non-theistic Platonism could underwrite either of these analyses. On this view, epistemic normativity would be grounded in something like non-spatiotemporal abstract ideals (and thus not explicable in purely scientific terms). However, it is difficult to see how mere impersonal abstracta could give rise to epistemic duties or could account for cognitive design, because such notions only make sense in terms of personal relationships. (How could we have obligations of any kind toward non-persons? How could our cognitive faculties have been designed by non-persons?) One of the pressing difficulties here for the Platonist is that of bringing the immanent realm of particulars (which includes us) into meaningful contact with the transcendent realm of universals (which would presumably include the ideals that ground epistemic normativity). Why should transcendent impersonal abstract entities be so much as relevant to the activities of immanent personal concrete entities (such as we are)? After all, even Plato had to appeal to the activity of the Demiurge to account for the relevance of the Forms to the sensible world! But clearly such appeals are bound to move us in a theistic direction.

Another popular alternative to theism (at least in terms of worldwide adherents) is monistic pantheism: the view that all is God and all is One. On this view, all ultimate metaphysical distinctions are illusory. Reality is, at bottom, one undifferentiated entity. It ought to be obvious that such a metaphysic cannot support any cogent analysis of epistemic warrant, let alone the two delineated earlier, since all such accounts presuppose a real distinction between warranted beliefs and unwarranted beliefs. Likewise, paradigm cases of human knowledge take for granted a genuine distinction between the subject of belief and the object of belief. But even if one dispenses with monism, it seems to me that pantheism will still face problems insofar as it fails to uphold a clear distinction between the norm and the normed. For if everything is divine then surely it follows that everything is normative—or, at the very least, everything is normal. What place, then, for ignorance or irrationality? The moment one divides the universe into the normative and that which is subject to the normative, the essence of pantheism has been abandoned.

A third alternative to consider would be panentheism, sometimes known as ‘process theism’. According to panentheists, God contains the universe but is not identical with it (i.e., the universe is a proper part of God). The panentheist conception of God is often analogized to the dualist conception of human nature: the universe is thought of as God’s ‘body’, while his transcendent immaterial aspect is thought of as God’s ‘mind’ or ‘soul’. And one of the crucial tenets of panentheism is that neither aspect of the duality is absolute or independent of the other: they are mutually dependent and the interaction between them accounts for God’s temporal progression from potentiality to actuality.

How then would one account for objective epistemic normativity within a panentheistic scheme? In some strains of panentheism, God is not thought to be a personal being, in which case (to recycle my earlier point) it is very hard to see how objective epistemic duties would apply to us or how the notions of purpose and design would pertain to our cognitive faculties. But even among those process theists who hold that God is a person, there is a reluctance to ground moral obligations in the character or will of God. Equally unwelcome is the notion that our cognitive faculties are the product of divine design. (I know of no process theist who does not affirm some version of the neo-Darwinian account of human origins; indeed, one of the attractions of panentheism is that it purports to reconcile an unthreatening supernaturalism with the methodological naturalism of modern scientific orthodoxy.) So even if the panentheist’s God is a personal being after all, this personal being is not the source or ground of objective epistemic norms for human thought (whether construed deontologically or teleologically). But then what would account for the existence of these norms? What is perhaps the most tempting route—an appeal to transcendent abstract ideals—offers no advance on the non-theistic Platonism considered above.

In any case, I believe a more general problem for panentheism can be raised. For reasons similar to those rehearsed earlier in discussing metaphysical naturalism, epistemic norms within a panentheistic scheme would need to be grounded in the transcendent aspect of God rather than the immanent aspect—in the ‘mind’ of God, as it were, as opposed to the ‘body’. Cognitive agents within the universe (such as human beings) are subject to epistemic norms. And norms necessarily transcend that which is subject to those norms. Thus, in the epistemic arena, the transcendent aspect of the panentheist's God would (in some respect) serve as the norm while the immanent aspect would serve as the normed. But wouldn’t this introduce a fundamental asymmetry of privilege into the panentheist’s metaphysic? Surely any norm enjoys a certain priority or superiority over that which is subject to the norm. One serves as the ideal; the other is judged with respect that ideal (and may well fall short of it). But if God’s transcendent aspect is privileged in this way over God’s immanent aspect, would it not be better to reserve the title ‘God’ for the transcendent aspect alone? All this is just to say that once this fundamental asymmetry with respect to normativity has been conceded, the resultant scheme would be better characterised as a form of theism than a form of panentheism.

Some general conclusions can be drawn from all this. It appears that any worldview whose ontology can underwrite a viable account of epistemic warrant must exhibit the following features: (1) it must be non-naturalist (i.e., it must make room for real objective normativity); (2) it must posit a fundamental ontological distinction between that which grounds or originates epistemic norms and that which is subject to epistemic norms; and (3) it must posit a ground of epistemic normativity that is personal (i.e., exhibiting features such as intellect and volition). In short, we are looking for a worldview with an ontology that includes a supernatural personal being whose character or intentions give rise to norms for human thought; that is, a broadly theistic worldview.

In conclusion, then, we have solid reasons for believing that if human knowledge is possible then there must be a God. Knowledge presupposes the existence of objective epistemic normativity, which in turn presupposes an ontology that can account for the existence of such normativity. Naturalism, as many of its contemporary advocates now acknowledge, has no place for objective epistemic normativity. And non-theistic non-naturalisms fall short on other grounds: by trying to ground epistemic normativity in the non-personal, or by failing to distinguish the normative from the normed, or by leaving unexplained the connection between the normative and the normed. Only theistic worldviews have the metaphysical resources to underwrite the most defensible analyses of epistemic warrant. In four words: if knowledge, then God.

Selected Bibliography