by John M. Frame

A Review Article

[An article reviewing John L. Pollock’s Contemporary Theories of Knowledge (Totowa, NJ: Rowman and Littlefield, 1986). Originally published in Westminster Theological Journal  52:1 (Spring, 1990), 131-141. Used by permission.]


Theologians have traditionally taken an interest in philosophical epistemology because of their concern with the knowledge of God. Sometimes they have sought to use secular epistemological theories to their advantage; sometimes they have sought to refute such theories. But the interaction has often been vigorous. This history suggests the value to theologians of keeping current in the field. We are still writing quite a bit about the classical epistemologies of Plato and Aristotle (against the background of Parmenides and the sophists), about traditional rationalism and empiricism, Kant and Hegel. Some theologians have also developed interest in certain twentieth century developments, particularly those associated with logical positivism, the later Wittgenstein and the existentialists, and especially the movement away from “objective” knowledge represented in different ways by Thomas Kuhn, Michael Polanyi, Norwood Hanson, Paul Feyerabend, Alasdair MacIntyre, D. Z. Phillips, and others.

Like most theological works, my own Doctrine of the Knowledge of God1 only goes this far.2 Of course, its purpose was not to survey secular theories but to set forth some biblical teachings about knowledge. But some comparison between biblical and secular notions was inevitable, and I regret now that I did not in that book refer at greater length to more current developments in the secular field.

John Pollock’s Contemporary Theories of Knowledge is an excellent recent survey of the present-day epistemologies of secular philosophy. Roderick Chisholm, perhaps the best known contemporary epistemologist, calls the book “A thorough and accurate survey of the present state of the subject, [Pollock's book] is also an original contribution of first importance. I know of no better introduction to contemporary theories of knowledge” (back cover).  I agree with Chisholm’s estimate, and I think this book is a very useful tool for bringing theological readers up to date in this area and a good focal point for some Christian evaluations of the contemporary theories.

It is, for the most part, a highly technical book, difficult to read, a book which takes the reader more deeply into the details of its arguments than many of us would prefer to go. At times, however, Pollock wakes us up with vividillustrations and convenient summaries of his argument. In the former category is the opening of the book, a three-page suspense tale ending with the discovery of Harry. Harry’s brain has been surgically removed from his body and placed in a vat of nutrients, where it continues to live. A computer sends impulses over wires attached to the brain, which give Harry the impression that he is living his normal, pre-vat life. The narrator concludes, “racked by the suspicion that I am really a brain in a vat and all this I see around me is just a figment of the computer” (p. 3).

From this tale, one might anticipate that the book would consist largely of reflection upon skepticism. Actually, however, Pollock deals with the skeptic in short fashion. The skeptical conclusion, that we know precisely nothing, is, to Pollock, so implausible, so unlikely, that it actually functions as a reductio. If an argument logically entails skepticism, he maintains, there must be something wrong with the premises. Pure skepticism, of course, is irrefutable, since the skeptic allows his critic no knowledge on the basis of which to debate. But we know that the skeptic is wrong; for if we don’t know that, we don’t know anything else. And if we do know that, it is evident that we know some things (e.g. skepticism is false) which we cannot prove.

Skepticism as such, then, is not of much interest to Pollock. But skeptical arguments, he says, are useful; for they alert us to false premises. If a premise leads to skepticism, it cannot be accepted. So skeptical arguments are of considerable negative value (p. 7). From them we can learn various things about what knowledge involves and does not involve.

From here, Pollock takes it for granted, not only that we have knowledge, but also that we have various kinds of knowledge: perceptual knowledge, memory knowledge, knowledge by induction and deduction (pp. 10ff). These are the four kinds of knowledge on which the book focuses. Pollock also seems to believe that we may have a priori knowledge and moral knowledge, but he notes candidly that these are highly problematic in modern epistemology, and he says nothing more about them through the book. He also ignores, after mentioning it briefly (p. 10), knowledge of other minds. He does not mention the possibility of knowledge coming through the testimony of other persons, which Iconsider important and sufficiently distinct from the other forms to deserve separate treatment.3

Nor does he say anything about knowledge through divine revelation. God plays no role in Pollock’s epistemology whatsoever, and one gathers that when Pollock describes his position as “naturalistic” (pp. 168ff, elsewhere) he means to reject not only the Cartesian ego, the “ghost in (the) machine” (p. 161), but to reject any dependence on religious or supernatural concepts. He claims an advantage to his view in the fact that his concept of knowledge can be applied to a “cognitive machine” (p. 149), and he spends some time speculating on how such a robot might be made to function (pp. 149ff). Pollock’s discussion of the cognitive robot is not satisfying to me. He proposes that “Oscar” be given “sense organs” (149), “‘reasoning’ faculties, both deductive and inductive” (149), “pain sensors” (150), “a ‘language of thought’” (150), “pain-sensor sensors” (151), perceptual organ activation sensors (155), cognitive process sensors (155), mental representations for objects and self (156-161). But he doesn’t give us any suggestion as to how these remarkable faculties might be built into a robot. Until he does, the way is open for a critic to argue that such abilities can only be performed by a spirit, even a “Cartesian ego.” If Pollock is simply trying to illustrate his epistemological proposal, perhaps Oscar serves a purpose; but if he is presenting this as an argument for naturalism, it certainly does notsucceed.

What is knowledge? Before 1963, most all analytic philosophers defined knowledge as “justified, true belief.” In 1963, however, there appeared Edmund Gettier’s article “Is Justified True Belief Knowledge?”4 Gettier suggested bycounterexamples that not every case of justified true belief was knowledge. Here is one of Gettier’s counterexamples, paraphrased by Pollock (p. 180):

…consider Smith who believes falsely but with good reason that Jones owns a Ford. Smith has no idea where Brown is, but he arbitrarily picks Barcelona and infers from the putative fact that Jones owns a Ford that either Jones owns a Ford or Brown is in Barcelona . It happens by chance that Brown is in Barcelona , so this disjunction is true. Furthermore, as Smith has good reason to believe that Jones owns a Ford, he is justified in believing this disjunction. But as his evidence does not pertain to the true disjunct of the disjunction, we would not regard Smith as knowing that either Jones owns a Ford or Brown is in Barcelona.

Many others published articles trying to solve the “Gettier problem,” mostly by adding a fourth condition to knowledge besides justification, truth, and belief (p. 9). But still others found counterexamples to those fourth conditions, and so the debate continues to this day.

Pollock’s own solution to the Gettier problem involves some reconstruction of the concept of “justification,” to which we should now turn. “A justified belief,” says Pollock, “is one that is ‘epistemically permissible’ to hold” (p. 7). Hedistinguishes epistemic permissibility from both prudential and moral permissibility. I am not persuaded by these distinctions. Pollock argues only by giving examples of beliefs that he thinks are prudentially or morally right but epistemically wrong and vice versa. E.g., someone promises not to think ill of another; in this case, thinking ill might be morally wrong, though epistemically right. My own analysis of this case, however, is that such a promise is invalid, since it pledges something that cannot be pledged. It is never right to promise someone that I will regard the truth as false or vice versa. Therefore thinking ill (when epistemically justified) is both epistemically and morally right, and the illustration does nothing to show that the former is not a subset of the latter. Even Pollock’s evaluation, however, entails only that epistemic permissibility is not the only kind of prudential or moral permissibility, not that epistemic permissibility is outside these two realms. Surely epistemic permissibility is founded upon our ethical obligation to believe only the truth. If it is not, then I don’t know what basis it might have. (I shall discuss Pollock’s basis at a later point.)5

Pollock also distinguishes epistemic permission from any concept of epistemic obligation:

Epistemic norms never tell us that it is epistemically obligatory to believe something– only that it is epistemically permissible to do so. It is not true, for example, that if I believe both P and “if P then Q” then, in the absence ofconflicting reasons, I ought to believe Q. This is because I might not care about Q. (P. 84, emphasis his; cf. pp. 124, 185.)

Of course, we might be morally obligated to care about Q, which would prevent us from bringing our uncaringness as an excuse. Apart from that, however, Pollock’s argument does present a good reason why we would not in this case be obligated to believe Q consciously. Most of our beliefs, however, are not being entertained consciously at a particular moment. “Caring” is one reason (among others) why we might consciously attend to a particular belief at a particular time; but it doesn’t seem to have much to do with what we believe or don’t believe.

My own account of this is that if someone believes P and “if P then Q,” in one sense he already believes Q, since Q does not include any information not included in the premises; in another sense he will believe it if and when he has (at least once) become conscious of the entailment. The element of obligation becomes evident when someone tries to deny what he knows– denying it either to himself or to someone else. At that point, it becomes legitimate to say “you ought to believe Q; and when asked in an appropriate forum, you ought to admit that you believe Q.”

One reason Pollock seems to resist any subjection of reasoning to moral evaluation may be his view that we “do not literally ‘decide’ what to believe” (p. 22). On p.80, he adds,

We do not have voluntary control over our beliefs. We cannot just decide to believe that 2 + 2 = 5 and thereby do it. We have at most indirect control over what we believe. We can try to get ourselves to believe something byrepeatedly rehearsing the evidence for it, or putting countervailing evidence out of our minds, or by deliberately seeking new evidence for it, but we cannot voluntarily make ourselves believe something in the same sense that we can voluntarily clench our fists.

There is much truth in this. It may be that we simply believe what we believe, and apparent “struggles to decide what to believe” are either struggles to form a new belief (by confronting evidence as Pollock outlines above), struggles to determine which of two or more inconsistent beliefs will prevail in our thinking, or introspective struggles to determine what we already believe in our heart of hearts.6

Nevertheless, if there is no voluntary decision concerning what to believe as such, there certainly are voluntary decisions to be made as to whether and how a belief is to be confessed, applied, implemented, etc. Especially when peoplerefuse to acknowledge what they know to be true (“exchanging the truth for a lie”), the will is obviously active. And there are voluntary decisions concerning the use of evidence as Pollock mentions in the above quote. Thus there is plenty of room for moral evaluation in the epistemic sphere. And since epistemology deals not only with the beliefs we actually have but also with the processes by which we confess, defend, implement, apply, resist, deny our beliefs, it is not wrong to say that there are certain beliefs which we “ought” to hold,7 beliefs which are justified by a kind of moral rightness.

The final thing to note about Pollock’s view of epistemic justification is that it is subjective, rather than objective (p. 10; cf. p. 183). He also characterizes this concept as the “belief-guiding” or “reason-guiding” sense of justification (p. 10). It helps us “in determining what to believe.”8 A justification gives us reasons for adopting a particular belief.

Justifications in this sense are person-variable. A good reason for one person to believe P will not necessarily be a good reason for someone else to believe it. A child may believe in the existence of Santa Claus because his mother has testified to that proposition. He has found his mother to be trustworthy, and so he rightly believes he has good reasons for his conclusion. But the child’s father, having had much broader experience of Christmas celebrations in their cultural context, would not be right to accept the child’s justification for belief.

Now on this view of justification, one may be justified in believing something, say P, even though P is false. That fact is illustrated in the previous paragraph. The child has good reasons to believe in Santa and no good reasons to deny that he exists. Therefore he is justified in believing a proposition which, most of us would say, is objectively false.

The rest of the book focuses on the exploration of this concept of justification. Pollock insists that “The central topic of epistemology is epistemic justification rather than knowledge” (p. 9). He is right as to the importance of this concept in the epistemology literature.9 I would agree that subjective justification is an important category and therefore deserves study. I don’t understand, however, why this concept dominates the literature (including the present volume) to the extent that it does. It is certainly not the only kind of epistemic justification, and it may not even be the most important kind.

Consider again the child who believes in Santa Claus on his mother’s testimony. Is he justified in believing in Santa Claus? I have explained how we could answer “yes” to this question, by taking “justified” in the subjective sense. But is it not obvious that in another sense the child’s belief is not justified? Is it not common for freshmen entering college to be told that beliefs uncritically acquired at mother’s knee are not adequately justified for the purposes of highereducation?

Is it not also common for a reviewer of, say, a biography, to criticize the author for making “unjustified allegations?” In that context, the reviewer is not referring to whatever private, subjective reasons the author may have had forhis disputed beliefs. Rather, the reviewer is expressing disappointment that the author has not given reasons in the book sufficient to convince others to believe as he does.

I think that usually when we10 speak of justified beliefs, we are speaking of beliefs well enough grounded to stand the scrutiny, not only of those holding the beliefs, but of those who know the subject best.11

Remarkably, Pollock, in the Appendix where he considers the Gettier problem, notes that subjective justification as he has earlier defined it is not the kind of justification necessary for knowledge. In response to the Gettier challenge, Pollock develops a concept of “objective” justification which he thinks will do justice to our intuition that justification is a necessary condition of knowledge. After exploring several possibilities, he settles on this one:

S knows P if and only if S instantiates some argument A supporting P which is (1) ultimately undefeated relative to the set of all truths, and (2) ultimately undefeated relative to the set of all truths socially sensitive for S. (p. 193).

“Instantiates” here roughly means “accepts,” but see p. 188 for more precision. “Ultimately undefeated” means that all of the potential refutations to the argument can themselves be decisively refuted. “Socially sensitive” truths are truths which S is “expected to know” by others in his social group. If S is expected to know a proposition Q which if true would defeat P, and if he does not know an adequate defeater for Q, then S does not know P even if Q is false. But if S’s argument defeats such Q’s and all other potential defeaters, then S knows P. This concept of “objective justification” makes more precise the concept which I sketched intuitively in the previous paragraphs.

Well, if Pollock is right that objective, not subjective justification is the kind of justification necessary to knowledge, and if I am right that objective justification is at least as important to epistemology as subjective, then I cannotunderstand why Pollock devotes 95% of the book to subjective justification! Is it perhaps that, having eliminated any role for God in this epistemology, he is thus unable to give any cogent account of “objective truth?”

Let me try to show how a theistic commitment would modify his perspective and make it more cogent. It is perhaps significant that in describing objective justification, Pollock gives a role to the knower’s social group, to what the knower is “expected to know.” If Pollock were sufficiently broadminded to accept the membership of God in such a social group, then “expected to know” would take on moral significance (contrary to Pollock’s earlier insistence) and we would have some concrete guidance on how to evaluate claims to objective knowledge: since God is omniscient, anyone who meets condition (2) will automatically meet condition (1). We can then judge A’s claim to knowledge on the basis of our knowledge of what God expects A to know  = knowledge of God’s revelation. (Otherwise, we would need to be omniscient ourselves to judge whether someone has met condition (1).)12

The main body of the book is devoted to a survey of contemporary epistemologies in which Pollock defends one type of epistemology and attempts to refute the others. By “epistemologies” we are here to understand views of subjective justification, and the term “justification” will henceforth refer to subjective justification unless I indicate otherwise. He first distinguishes between “doxastic” and “nondoxastic” theories. (His taxonomy is found on pp. 19-25.) The former holds that justification of a belief for S is entirely a function of the other beliefs held by S. On a doxastic view, one justifies his beliefs by relating them (by comparison, deduction, induction, etc.) to other beliefs one holds. On a nondoxastic view, one is not limited to this sort of justification. E.g., on one kind of nondoxastic view, a belief derived “directly” from perception is justified because perception is a legitimate cognitive process, whether or not we have beliefs about the origin of the belief and the legitimacy of perception.

Pollock’s taxonomy of epistemological views is as follows:

I. Doxastic

A. Foundationalism

B. Coherentism

1. Linear positive

2. Holistic positive

3. Negative

II. Nondoxastic

A. Externalism

1. Probabilism

2. Reliabilism

B. Internalism

Direct Realism- Pollock’s view

Under the doxastic category are two distinct views, foundationalism and coherentism. On a foundationalist view, beliefs are ultimately justified by reference to “foundational” or “basic” beliefs. Among all the beliefs we hold, some are more fundamental to justification than others. Through the history of philosophy, various sorts of beliefs have been considered

“foundational:” we will recall Descartes’ “clear and distinct ideas,” Spinoza’s “axioms,” Leibniz’s “laws of thought, Hume’s “impressions,”13 Thomas Reid’s “common sense,” the “logical atoms” of Russell and the early Wittgenstein. We may recall also the recent proposal of Alvin Plantinga and Nicholas Wolterstorff that belief in God be considered “epistemologically basic.”14 The most common view today, however, is a variety of empiricism in which the foundation consists of reports of sense-experience, or at least of reports of the “appearances” with which we are acquainted.15 These beliefs are considered self-justifying, and all other beliefs must be justified in relation to them. Thenonfoundational beliefs are derived from the foundational by some kind of “reasoning.”

Pollock tries very hard to present views he ultimately rejects in their very best form. His practice in this respect is commendable. J. Gresham Machen was also very good at this, and present day theologians would do well to follow such examples. I will not, obviously, be able to reproduce Pollock’s expositions and arguments in detail, but I will try to summarize accurately.

The following is a highly condensed summary of Pollock’s argument against foundationalism: (1) We do not always know how we are being appeared to; indeed, we can be wrong about that. After an accident, let us say, witnesses often correct their first impressions of how they were appeared to (pp. 59-61). Therefore appearance beliefs are not self justifying. (2) We rarely have beliefs about how we are appeared to. Evidence of our senses does not take the form of beliefs (p. 61). Therefore, if sense-perception plays some basic role in justification, it is not by way of beliefs about our perceptions, as on a foundationalist theory. (3) Should we say that epistemologically basis beliefs are, not incorrigible or self-justifying, but only prima facie justified? To say that is to say that such beliefs are justified until we have reason to disbelieve them, “innocent until proved guilty.” But there is no better reason to make this claim for sensory appearance-beliefs than for any other kind of belief; so there is no reason, on this basis, to make any particular kind of belief epistemologically basic. Without “basic” beliefs, what we have is a coherence theory, not a foundations theory (pp. 60-66). (This is, I think, what Pollock would say in reply to the type of foundation theory proposed by Plantinga and Wolterstorff.) (4) Even granting the incorrigibility of sense beliefs, we can reason from them to other beliefs only by way of memory, and that requires memory to be an additional source of “basic” beliefs (in which case the above problems recur) or to function nondoxastically (pp. 46-57).

It is of course normal for us, when questioned about our reason for holding a belief, to derive that belief inductively or deductively from a belief of which we are more sure. This is the intuitive basis for foundationalism. But we should keep reminding ourselves of the specific question Pollock is addressing. That is the question of subjective justification. Thus it is not particularly relevant how we seek to justify our beliefs to other people; that is more in the category of “objective” justification. Rather, he is asking how we gain “epistemicpermission” to believe what we believe. And he is arguing, therefore, that whatever role “assured beliefs” may play in our epistemic self-defenses, they are not the reason why, in general our beliefs are subjectively justified. Not every justified belief can be derived from a sensory “basic” belief, and sensory beliefs are not in any meaningful sense self-justifying.

A coherence theory differs from a foundations theory in that for the coherentist there are no “basic” beliefs, no epistemically privileged propositions. A person justifies a belief by relating it somehow to all his other beliefs. (“Beliefs,” because like foundationalism, coherentism is doxastic.) If that belief “coheres” with the rest, then it is justified; otherwise, not. Pollock distinguishes positive coherence theories, in which positive support is required for all beliefs, andnegative coherence theories, in which all beliefs are “innocent until proved guilty” and are to be abandoned only by sufficient negative reasons (pp. 71ff). Another distinction is between linear and holistic coherence theories. In the former, our basic reason for believing a proposition is a small set of beliefs which, to be sure, when we ask reasons for reasons, expands to include our entire stock of beliefs. In the latter, one cannot reduce the justification for a belief to any such linear chain.

Against linear positive coherence theories, Pollock argues that these cannot produce any “plausible candidates for reasons for beliefs that result directly from perception” (p. 77). On a coherence view, all reasoning is by inference; but “…perception is not inference. When I believe on the basis of perception that the book is red, I do not infer that belief from something else that I believe. Perception is a causal process that inputs beliefs into our doxastic system without their being inferred from or justified on the basis of other beliefs we already have” (p. 75).

Remember again that Pollock is concerned with subjective justification, with how we acquire epistemic permission to believe, not with how we defend our beliefs to others. In fact we may always choose to defend our beliefs by inference; but that cannot be the way our original subjective justification comes about.

Against holistic coherence theories, Pollock objects that in fact we do not derive all our beliefs from beliefs about the coherence of our belief system; in fact we very rarely have such beliefs about coherence. And if we did, how would we get those beliefs? In order to believe P, we would first have to believe Q, namely that P coheres with our other beliefs. But to believe Q, we would first have to believe R, namely that Q coheres. The result is infinite regress.

Against negative coherence theories, Pollock replies that if we consider all beliefs justified until defeated or refuted, then reasons play no positive role in justify beliefs, only the negative role of rebutting. But this means that no sense can be made out of the notion of believing something for a reason, a notion crucially important to the concept of subjective justification (pp. 83-87).

Under nondoxastic theories, Pollock explores externalism and internalism. A nondoxastic theory says that our beliefs are subjectively justified not only by means of otherbeliefs, but also by some states of affairs about which we may not have beliefs. In an internalist view, those states of affairs are only internal to us. In an externalist view, they may be external to us as well.

He examines two types of externalism, probabilism and reliabilism. Probabilism, the view that beliefs are justified when they have a sufficiently high probability, gets Pollock into some complicated mathematics, from which he concludes that there is “no appropriate kind of probability for use in probabilist theories of knowledge” (p. 113). The more intuitive, ordinary-English concept of epistemological probability, he says, “is defined in terms of epistemic justification, so this provides no analysis of epistemic justification and no support for probabilism” (P. 113).

Reliabilism teaches that “a belief is justified if and only if it is produced by a reliable cognitive process” (p. 114). Pollock rejects this principle also, on the ground that reliability of processes has nothing to do with epistemic(=subjective) justification. Poor Harry, the brain-in-the-vat, has unreliable perceptive faculties. But he has no reason to think his faculties are unreliable, so he has no alternative but to trust them. In other words, Harry’s beliefs about his “normal life” are, in general, subjectively justified, though mostly false. More fundamentally, Pollock argues, most “reliable cognitive processes” (color vision is his example) are reliable only in certain circumstances. But if we narrow the circumstances too far (such as by presupposing the truth-value of the belief under consideration) we reach the conclusion that the belief is justifiable only if it is true, which is not the way subjective justification is supposed to work.

Against all externalism, Pollock insists that in our moment to moment belief formation we do not always have access to data concerning the reliability of cognitive processes or concerning the probability of propositions. We do, of course, have access to the cognitive processes themselves; but those are internal rather than external.

By process of elimination, then, we are left with a form of internalism. Pollock calls his version of internalism “direct realism,” because it is nondoxastic: in it we gain justification for beliefs without the mediating presence of other beliefs. Our reasons for believing, fundamentally, are our mental processes themselves. We believe because our mental processes lead us to believe as we do. We do not need to have beliefs about our mental processes (e.g. about their reliability) for this to happen. If we sometimes appeal to reliability or probability, or to “basic” ideas or to systematic coherence, that is just the way our mental processes sometimes work. So the fundamental justification is not an appeal to reliability or whatever; it is simply that our mental processes work in this particular way.

But aren’t our mental processes sometimes fallible? Yes, but we discover that by means of mental processes themselves, one checking another. And more importantly: don’t forget that we are here talking about subjectivejustification. There is often a lack of correlation between subjective justification and truth. So the fallibility of our mental processes is irrelevant to their primacy in subjective justification.

Consider poor Harry, the brain-in-a-vat. His faculties are supremely unreliable; yet because he himself has no reason to doubt them, his beliefs about his living a normal life are subjectively justified. He is subjectively justified because that is simply the way his faculties work.

This does not mean that all of our beliefs are justified. Some of our beliefs are chosen arbitrarily, the result of wishful thinking, etc. They are not the result of the working of our cognitive faculties. (But I wonder how on Pollock’s basis one can distinguish properly cognitive faculties from other psychologically actual means of forming beliefs.)

Pollock includes much more exposition and argument in favor of his internalism, but I will stop my exposition here. Internalism appears to be almost the inevitable conclusion of the book, once the reader gains clarity as to the concept of “subjective justification.” Certainly it is true that we use foundationalist, coherentist, reliability and probabilistic arguments to justify our beliefs to others. Sometimes we use these methods of justifying our beliefs to also ourselves in situations where we gain some detachment from our own belief commitments. But these are justifications which aim at showing the objective truth of these ideas. They are not the original means by which we acquire such beliefs.

In the original acquiring of beliefs, much is mysterious. Pollock is right; we rarely argue explicitly with ourselves. We rarely appeal to “foundations” or coherence or probability or reliability. Rather, we just find ourselves believing. And when we are assured of them, we cannot always say why or how we are assured. Rather, our minds are simply “programmed” to give us assurance in certain situations.

Foundationalism, coherentism, probabilism and reliabilism therefore really confuse subjective with objective justification to some extent. Internalism is the only fully subjective mechanism available for subjective justification.

Having agreed with Pollock’s main point, however, I would like to add something about its vacuity. For when I ask “how am I justified in believing P?” Pollock’s answer seems to boil down to this: “You are justified by the justificatory faculties of your mind.” That’s a bit like the scholastics’ explanation of falling bodies by reference to “falling tendencies” within those bodies. Now Pollock isn’t quite as bad as all that. He does present some illuminating psychological description of how we come to make up our minds about beliefs. But most of that description is negative, telling us what we don’t do. And, we recall, Pollock does no more than to reflect on the general ignorance on such important matters as a priori and moral knowledge. (If my arguments in DKG are correct, without moral knowledge there is no knowledge of anything.) He says nothing positive about them. What he does say positively is mostly theaffirmation of such things as “sensors” (as in the Oscar chapter); but we didn’t need him to tell us we had such faculties. So in general, he leaves our mental processes (and hence subjective justification) under a great veil of mystery. Inthe end, we think as we do because we think as we do. And as Pollock’s main conclusion that seems too vacuous to be of help.

This reminds me of what some philosophers have called the “paradox of analysis:” Often when someone tries very hard to analyze something and insists on a rigorous equation between the analysis and the analysandum, he comes up with something uninformative. What are black holes? The only way to come up with a perfect equation between analysis and analysandum is to say that a black hole is a black hole. Has Pollock tried to seek too much analytical perfection?

It also reminds me of Cornelius Van Til’s point that philosophical rationalism, insofar as it seeks an exhaustive explanation of reality, leads us to see the world in terms of “blank identity,” like Parmenides’ “being,” Plato’s “good,”Aristotle’s “thought thinking thought,” Plotinus’ “one,” etc.

What might be an alternative, motivated by Christian theism? Well, we can certainly concede that the mind works the way it works! But we should insist that subjective justification is only a small part of the story in epistemology. Objective justification is part of our responsibility to “test all things” and to seek and apply God’s truth rightly. That deserves far more analysis than Pollock gives it. And we have epistemic obligations from the living God;16 not just epistemic permissions, as Pollock insists.

As for subjective justification, well, a Christian view would stress that God made man’s mind to know him and to know his truth. So in man as originally created, subjective justification and objective justification coincided. Unlike Harry’s brain, our mental faculties were reliable. Sin, however, led man to flee from the truth, especially from God. This fact introduced distortions, self-deceptions.

A Christian analysis would have to discuss this process of self-deception17 and also the restoration of sound thinking as part of God’s redemptive grace.

We can learn much, certainly, from the thoroughness and rigor of Pollock’s argument. But clearly we need to learn much more about human knowledge than Pollock (and, I gather, modern philosophical epistemology in general) has to teach us.



Phillipsburg, N. J.: Presbyterian and Reformed, 1987. Henceforth DKG.

There is, however, some allusion to more recent developments in the Appendix dealing with Plantinga and Wolterstorff.

Cf. Thomas Reid’s “credulity principle.”

Analysis 23:121-23.

As for prudence, I believe as a Christian that ultimately prudence and morality concur. Pragmatism reduces epistemic norms to prudential ones which is not, in my view, wrong in itself. But to make this work one must have a Christian concept of prudence, which was lacking in most forms of pragmatism.

I say this “may” be the case. My actual view of the matter is more complicated. I think that there are different levels of belief. At the most fundamental level, we know God and we know the world as God made it. We thus have a limited, though adequate stock of true beliefs. But sin causes us to develop additional beliefs inconsistent with those true beliefs– i.e., to “exchange the truth for a lie.” Regeneration enables us to resist this process. Belief at the most fundamental level is involuntary; but in the process of belief-substitution in the states both of sin and of regeneration, the will is quite active indeed (though even in those states the process is not always conscious).

Here I take “hold” as a broad term including the possessing, acknowledging, confessing, applying, implementing, of the belief in question.

Evidently, as we’ve seen, we should ignore any moral connotations which we might hear in this phrase.>

See my DKG, pp. 389-391, 395-398 on Plantinga and Wolterstorff.

10 As indicated earlier, the “we” here will have to exclude professional epistemologists.

11 Another possibility, suggested in DKG, pp. 397f, is that there are multiple “levels” of justification corresponding to the kinds of cogency demanded by persons in various contexts (scholarly societies vs. children’s Sunday Schools). The highest social level on this scale would be our relationship with God, before whom we always stand, and who continually challenges us to be satisfied with nothing less than an absolute justification, a justification based on his Word.

12 “Knowing what God expects us to know” is of course itself problematic. My DKG is intended in part to show how we come to know and apply God’s revelation.

13 Hume’s view might, however, be better understood as nondoxastic, in Pollock’s vocabulary.

14 Plantinga and Wolterstorff, ed., Faith and Rationality (Notre Dame, Ind.: Univ. of Notre Dame Press, 1983). Discussed in DKG, 382-402.

15 Chisholm’s famous “I am appeared to redly,” which Pollock also employs, is a way to avoid prejudicing the question of what reality, if any, the appearance refers to.

16 See DKG.

17 See Greg Bahnsen, Self-Deception, a doctoral dissertation for the Philosophy Department of the University of Southern California.