CHAPTER 7

 

Strong Higher-Level Theistic Evidentialism:

Elucidations and Defense

 

 

 

 

 

 

 

 

 

            The higher-level evidentialist requirement stands in need of further analysis. First, there is a need to spell out the epistemic relevance of reflective rationality, especially its relation to lower-level justification. In Part I of this chapter I explain the epistemic merit of reflective rationality by showing the ways it may affect S’s being justified at the lower level. In Part II I direct my attention to the plausibility of the strong higher-level evidentialist requirement. The plausibility of such a requirement rests on the impossibility of being immediately justified in beliefs about the epistemic status of our beliefs, or at least the impossibility of this with respect to theistic belief in particular. I will consider several arguments against such a requirement - arguments to the effect that one can be immediately justified in believing that one’s belief or some proposition is immediately (or mediately) justified. The last part of the chapter returns to Plantinga’s religious epistemology and considers the extent to which strong higher-level evidentialism is compatible with it. The essential insight here will be that externalism in one’s concept of justification or warrant is compatible with strong internalist requirements for the conditions of being justified or warranted in particular beliefs. Therefore there is no intrinsic incompatibility between Plantinga’s externalist theory of warrant and the strong higher-level evidentialist requirement.

 

 

I. The Epistemic Relevance of Reflective Rationality

 

A. Epistemic Level Relations

 

            In the last chapter I articulated the Janus-faced character of reflective rationality, contrasting it with unreflective rationality. The conditions necessary (and sufficient) for being justified in higher-level beliefs (or propositions) are far more stringent and demanding “evidentially speaking” (on both internalist and externalist construals) than what can plausibly be imposed on justification at the lower level. This is especially true given the sorts of difficulties generated by second-order internalist requirements for justification as such canvassed in chapter 1. By failing to distinguish between beliefs that p and beliefs that <p is Q>, where Q = rational, justified, or whatever epistemic merit one selects, the classical evidentialist tradition imposed strong evidentialist requirements at the wrong level. Now one may read this history in two ways. The sorts of considerations I have introduced may lead one to think that the Enlightenment tradition was fundamentally misguided and so jettison the whole package. I think that Reformed epistemologists have had a tendency to take this path. Alternatively, my suggestion is that the internalist intuitions guiding the classical evidentialist are roughly correct. But these intuitions are more directly relevant to a cognitive desideratum distinct from being justified (or warranted) in some belief that p.  The appropriate target of those intuitions is the cognitive desideratum of reflective rationality (and the related notion of procedural epistemic rationality). And this desideratum is not necessary (at least not in any unqualified sense) for being justified in some belief that p. But what sort of justificatory or epistemic relation might obtain between these two desiderata? And in what ways, if at all, might reflective rationality be epistemically meritorious? It seems that reflective rationality will be epistemically relevant vis-à-vis beliefs that p just if it is epistemically meritorious vis-à-vis lower-level justification.

            One possibility here is that

   (1)     S is justified in believing that p

is entailed by

   (2)     S is justified in believing that <S is justified in believing that p>.

            Let’s call this the trickle-down justification theory (hereafter TDJ theory).

            Let’s first consider the TDJ theory on an externalist view according to which a belief’s being justified involves actual truth-conducivity. If S is justified in believing that p, then S’s belief that p is likely to be true. The sense of “likely to be true” here is (as explained in chapter 1, section II.D) spelled out in terms of objective (statistical) probability. By this the externalist understands a contingent truth about the number of true beliefs or the ratio of true beliefs to false beliefs produced by a particular cognitive process (or based upon certain grounds). So the TDJ theory is true if and only if

 

(1*)      S’s belief that p is produced (or sustained) by a cognitive process X which usually produces (or sustains) true beliefs of the form “p”

 

is entailed by

 

 (2*)     S’s belief that <S’s belief that p is justified> is produced (or sustained) by a cognitive process Y which usually produces (or sustains) true beliefs of the form “p is justified”.

            But it doesn’t seem that (1*) is entailed by (2*). The cognitive process that produces the belief that p will either be the same cognitive process which produces the belief that <S’s belief that p is justified> or it will be a different cognitive process. The fact that a given cognitive process may yield totally different truth ratios for two different beliefs suggests that even if process Y reliably produces a higher-order belief B* it may not reliably produce the lower-level correlate B.  On the other hand, if we suppose that process X (which yields B) is not the same process as process Y (which yields B*), we have a similar difficulty. It is surely possible that at time t1 cognitive process Y reliably produces the belief B*, but that at time t1 cognitive process X does not reliably produce beliefs of the form B.  It is not clear how the reliability of one cognitive process entails the reliability of another cognitive process. Since process Y entails that S has a justified belief about S’s being justified in believing that p, one might think that if this doxastic output, B*, of process Y itself constituted an adequate ground for the belief B the previous objections might be circumvented. But even if the “adequacy” of B* as a ground of B could be spelled out in externalist terms, we would have a remaining difficulty. I have argued in earlier chapters that S’s having a reason R for believing a proposition p, though it creates a disposition to believe p on the basis of R, does not entail that R is causally operative in sustaining S’s belief that p.  In cases where R is something like “evidence e makes p likely to be true”, R will frequently factor into at least the partial basis of the belief that p, but this is a weaker concession than the entailment requirement. Since for the externalist causal origin crucial, the fact that S’s holding B* is psychologically compatible with S’s holding B on at least the partial basis of inadequate grounds closes even this way of thinking about an externalist construal of the TDJ theory. So, given externalism, the TDJ theory doesn’t appear to be true.

            For the most part we get a similar result on the internalist account. Suppose S is justified in holding some theistic belief ht only if the probability of ht on evidence e (plus background knowledge k) is greater than 1/2, and where the probability relation is a function of one’s own inductive criteria (subjective probability) or correct inductive standards (epistemic probability). Surely one can investigate and critically examine one’s evidence base and the extent to which the relevant set of inductive criteria make ht probable given e and k, and yet in fact ht is not rendered probable on e and k.  Our assessment of the evidential probability relations is one thing; the actuality of such support relations in another.[1] In other terms, let ht* = <the probability of ht given e and k is greater than 1/2>. We may have good evidence e* that makes it evidentially probable that ht*, and yet despite the strong probabilifying evidence in support of the higher-level probabilistc judgement ht*, that judgement may not be true. So we may not in fact be justified in the lower-level belief that ht which requires that ht* be true (i.e., it be true that ht is probable given e and k).

            The logical theory of probability, though, presents a different result. Here every statement about some h being probable on some evidence e is either necessarily true or necessarily false. So if the higher-level probability judgement is true it will be necessarily true. This is one way in which [it is probable that <it is probable that h>] entails <it is probable that h>. Where A = [P(h/e & k) > 1/2], if A is true, then Nec.A. And if Nec.A, then the logical unconditional probability of A is 1, so P(A) = 1. But then, being a logical truth, A’s logical conditional probability on any evidence will also be 1, since every logical truth is entailed by any bit of evidence. So let B = [P((A)/e* & k) > 1/2], and where evidence e* indicates the evidence which makes it evidentially probable that <h is probable on e and k>. It follows that e* and k will entail (A), and so the evidence e* which makes it epistemically probable that <h is (epistemically) probable> also entails that <h is (epistemically probable> by the conditions of logical probability. So, for a logically omniscient being, a true statement that [it is probable that <it is probable that h>] will always entail that <it is probable that h>. If it is true that the reasons that make probable the former are necessary for being justified in believing that one is justified in believing that h, then one will be justified in believing that h.  So the TDJ theory is true on the theory of logical probability, but as explained in chapter 1 that is not the account of probability which is relevant to epistemic justification.

B. The Epistemic Merit of Reflective Rationality

 

            One way to resolve the issue of the TDJ theory is to stipulate various senses of first-level justification such that synchronic and diachronic forms of reflective rationality have first-level relevance simply because they constitute states of being justified in believing that p which fall on a justification continuum. This would typically be an internalist move since the epistemic point of view would be something like “S’s having good reasons for believing that p is likely to be true” rather than some actual high number of true beliefs or ratio of true beliefs to false beliefs produced by the appropriate cognitive mechanism(s).  Justification would be charted in terms of variations on subjective and epistemic evidential probability, as well as distinguishing between synchronic and diachronic modes of reflective rationality (and distinguishing various degrees or levels of objectivity in the diachronic sphere). For instance, this would include S’s investigating to see whether a given probability relation between evidence e and proposition h does hold, S’s investigation being adequate given his own inductive standards vs. adequate by correct standards, and similarly with S’s investigation into his inductive criteria (to determine whether his criteria are correct or not).

            Let me spell this out a bit more. At time t1 S may be justified by virtue of having evidence e which makes S’s theistic belief probable. S’s justification here may be by way of subjective or epistemic evidential probability (so there is something like an indexical which can be added, justificationSP or justificationEP). But at time t2 S may embark upon an investigation and acquire additional evidence e* about the adequacy of e, where e includes logical evidence about whether e makes theistic belief probable and whether the inductive criteria being employed are correct ones. As a result of the investigation, S comes to believe that he is justified since he now has evidence regarding the conditions that must obtain if he is justified. Call this latter state justification*. Futhermore, we can evaluate the diachronic dimension (and any synchronic state resulting from it) in terms of S’s having checked to see that his investigation was adequate given either his own inductive standards or correct criteria. And so we have something like justification**, justification***, and so on, where “*” indicates an index of degrees of reflective rationality. The justification continuum can begin with justificationSP and justificationEP and continues through a series of “*” indexed justification states (some of which will include either justificationSP or justificationEP).

            I noted in I.A that there may be a discontinuity between beliefs about (subjective or epistemic) evidential probability relations and the actual obtaining of such probability relations.The same holds true for any degree of investigation. But as long as we do not restrict justification to the actual obtaining of probability relations (especially “epistemic” probability relations where it is assumed that the inductive criteria are all true), we can include the discovery of new logical evidence e* about whether the probability relation in question does hold as sufficient for being justified in an epistemically relevant sense. For “*” indexed justification states will be states in which S has a belief about the justified status of his belief based on some degree of investigation and discovery of new logical evidence. Even if a “*” indexed justification state is not sufficient for justificationEP (perhaps some of S’s inductive standards are not true) or for a justificationSP state (perhaps S”s own inductive standards do not make the proposition in question probable given evidence e), S nonetheless has engaged in an activity of critical reflection on the belief in question in the way of perspectival truth-conducivity. This does seem to count toward the rationality of a lower-level belief.[2]

            How does this bear on the externalist package? It is interesting to note that some externalists (Alston 1989c, pp. 27, 113, 244) recognize that having justified higher-level beliefs constitutes good epistemic reasons or evidence for believing the correlated lower-level proposition p.  Although, for the externalist, the higher-level belief will be about the actual truth-conducive adequacy of the source of the lower-level belief, this still suggests some kind of internalist concession. By implication, there seems to be a recognition of the value (even if restricted) of “perspectival” truth-conducivity over against “actual” truth-conducivity. A pure externalist will not find this a felicitous situation. He demurs even negative internalist constraints whereby what falls within the believer’s perspective (other beliefs or knowledge) may serve to defeat justification. Inasmuch as I have been looking specifically at Plantinga’s religious epistemology, we are not dealing with a pure externalism (and I have given reasons in chapter 1 for rejecting pure externalism, certainly as far as the central claim of this thesis is concerned). I have argued in previous chapters two important consequences of Plantinga’s recent epistemology. First, reasons may defeat the (proper function) rationality of a belief. Secondly, reasons may be necessary for a person to remain epistemically warranted in holding some theistic belief by virtue of contributing to the causal sustenance of (firm) belief in the presence of (otherwise) defeating conditions. Here causal relevance translates into epistemic significance.

            I think the typical scenario will be one in which defeaters call for additional support. A ground G may be sufficient for S to be warranted in holding some theistic belief Pt at t1 but not when the belief that Pt is conjoined with a defeater at time t2. If S acquires a reason to believe that God does not exist (say, a version of the atheological argument from evil), considerations about whether some evidence e really makes it probable that God does exist may be crucial. At time t1 S believes that God exists (at least in part) on the basis of evidence e (temporal regularities in the Universe). At time t2, S comes upon an evidential argument from evil. One move open to S, and perhaps required (depending on how strong the atheological objection strikes him), is to investigate the extent to which he is justified in holding to his evidence, and - in the case of beliefs in the existence of simple, intelligible scientific laws - whether such evidence really is to be expected given that God exists, or some probabilistic matter like that.  In other words, faced with objections to the truth of one’s beliefs, one’s continuing to hold a warranted belief (perhaps with the confidence one does) may very well depend on one’s investigating the evidence for one’s belief and its evidential force (perhaps discovering new logical truths about one’s evidence base, e.g., new deductive entailments from one’s evidence). Alternatively, and perhaps more directly relevant, suppose that the defeater is a reason for, say, believing that temporal regularities do not increase the probability of God’s existence, or that this evidence conjoined with other evidence does not make the existence of God more probable than not. The undercutting force of a defeater that asserts that one’s evidence of design is not adequate will find one important defeater-defeater in rebutter form. Evidence e* will “backs up”, so to speak, the evidential force of the teleological considerations by constituting reasons for supposing that e does have the probabilifying strength required for the corresponding justificatory status of theistic belief. The same is true where some experiential ground is claimed to be an inadequate source for theistic belief. S’s coming to see (on the basis of reasons) that his ground is adequate may provide at least partial causal support for the belief.

            Now it seems to me that what is typically happening in situations like those briefly outlined here is not that one ground is replacing another ground, but that another ground is added to one’s total evidence-base.[3] Perhaps this added ground was initially an overdetermining factor within S’s noetic structure, or maybe S discovers this (potential) ground when investigating the adequacy of his putative grounds or evidence for belief in God having been moved to do so by defeating conditions. At time t1 S believes that Pt on some ground G with a fairly high degree of firmness.  At t2 S encounters an undercutting (or rebutting) defeater. As explained in chapter 4, in such a scenario - and barring theistic belief being an intrinsic defeater-defeater - if S is to continue rationally holding his theistic belief, S will need a defeater-defeater. The set of relevant defeater-defeaters will include reasons for supposing that his ground is adequate or that there is other evidence for God’s existence and that evidence is adequate. At t3 higher-level reasons provide partial support for the belief that Pt by providing support for the adequacy of the original grounds or evidence. This suggests that even if reflective rationality is not unqualifiedly necessary for lower-level justification, there is a principle of its restricted necessity under certain defeating conditions. My becoming reflectively rational becomes an issue when I have (subjectively) good grounds to believe that what I have taken as evidence for God’s existence is not adequate evidence. In such cases, reflective rationality is epistemically meritorious just because S’s remaining (or in some cases acquiring) epistemically warranted belief in God depends in part on S’s having reasons for believing that evidence e is adequate and where these reasons outweigh the reasons for supposing the evidence to be inadequate.[4]

            I think we can conclude that the preceding considerations suggest why, even on an externalist view, it is good thing (cognitively speaking) for a person to be reflectively rational. Moreover, it may be àla Plantinga that one’s becoming reflectively rational is actually an aspect of cognitive proper function. Given a certain range of defeating conditions C*, the design plan dictates that we engage in those epistemically relevant activities which constitute diachronic reflective rationality. This could be an important dimension to the proper functioning of one’s defeater system. In this way procedural epistemic rationality may be combined with the notions of warrant and proper function to establish the importance and even necessity of regulating our cognitive life through voluntarily engaging in the activities which constitute diachronic reflective rationality, especially where the acquisition of defeater-defeaters depends on such activities.

 

II. The Plausibility of Higher-Level Evidentialism

 

            The argument for higher-level theistic evidentialism was from [E*] to [ET*] by way of universal instantiation, from the principle that if any belief that <p is justified> is justified it is mediately justified to this requirement holding for theistic beliefs in particular. I noted in the last chapter that the higher-level evidentialist requirement is by no means uncontroversial.[5] At this stage it is necessary to investigate to what extent higher-level evidentialism is plausible as an epistemological thesis, for it might be thought that surely higher-level beliefs, like lower-level beliefs, could be immediately justified for some people. If this is so, then [E*] is false. And unless we have some reason for holding that the evidentialist requirement in [E*] is applicable to theistic belief in particular, the general thesis of chapter 6 expressed by [ET*] fails. Rather than beginning with a set of arguments for [E*], I want to work from the other direction by considering the main reasons for supposing that higher-level beliefs are susceptible to an immediate justification.

 

A. High-Accessibility Internalism

 

            One reason for supposing that higher-level epistemic beliefs are immediately justified is on the assumptions that (a) some beliefs are immediately justified and (b) what justifies a belief at level L1 justifies correlated epistemic beliefs on level Ln+1 . According to one prominent epistemological tradition, that of high accessibility internalism, whenever a person is justified in some belief that p, the person can tell just upon reflection that he is justified. So rather naturally if S is immediately justified in believing that p, S is immediately justified in believing that <S is immediately justified in believing that p>, subject perhaps to the condition that S consider the higher-level proposition.[6] Hence, suppose that I am immediately justified in the belief that <I am tired>. Here I will be justified in the higher-level belief that [my belief that <I am tired> is justified] just by virtue of my being justified in the original introspective belief, since the same thing confers justification on both levels. The difficulties with this position are manifold.

            First, one must hold that justificatory status is either irreducibly evaluative (on some form of intuitionism in value theory) or determined by concepts involved in the belief in question. This will be unacceptable from the standpoint of naturalized epistemologies according to which epistemic principles are not (solely) a priori but empirical. Secondly, from the mere fact that a person considers the proposition that he is justified in believing that p, it does not follow that if he is justified in believing that p the same thing justifies the lower- and higher-level beliefs. Perhaps something in the process of reflection (i.e., other beliefs) was involved in conferring justification on the higher-level belief. Thirdly, and following partly on the first point, if one saddles one’s account of justification with this requirement, one will have to exclude certain beliefs and/or models of immediate justification from the category of proper basicality. Typically those who argue for high-accessibility internalism restrict properly basic beliefs to those which can be expected to survive the transition of justification to the higher-level - so called “self-presenting states” (first person intentional and perceptual states). If what generates justification for p must also generate justification for <p is justified> certain candidates for immediate justification will be given the axe. As I shall argue shortly, beliefs formed through a process of reliable belief formation will be excluded, since the status of justification by origin seems to rest upon questions of empirical fact which are not cognitively accessible just upon reflection. Positively put, Alston explains that the preoccupation many foundationalists have had with immediate awareness, the directly evident, and self-evidence seems to be a consequence of the “uncritical assumption that correlated propositions on two levels enjoy the same justification” (1989c, p. 159). It would equally seem to follow then that if one restricts oneself to immediate justifiers that can in fact generate justification on both levels, the kinds of beliefs one takes to be susceptible to an immediate justification will also be restricted. They will be restricted to beliefs about one’s current conscious states and/or the immediate data of sense experience.[7]

            Now if we consider the epistemology of theistic belief in the light of this argument we are left with either of two possibilities. Theistic belief will either have to satisfy the conditions required to survive trans-level justification or it will be excluded from the class of properly basic beliefs. Putative models of immediate justification for theistic belief do not seem to sit well with surviving trans-level justification. Consider Alston’s theory of the nonsensory perceptual experience of God. Here, though the ground is cognitively accessible, its adequacy is not. So it will be impossible, with this externalist twist, to tell just on reflection that one's belief is justified (since that involves a claim about the adequacy of grounds). Moreover, to say that one may be experientially presented with God doing such and such in one’s life is not to say that one can be presented with epistemic facts or statuses. There is a significant difference between being immediately aware of the presence of God by virtue of God being presented to one’s experience and being immediately aware of the proposition that this is an adequate basis for justified M-beliefs.[8] This is all the more apparent once we take a view on the adequacy of grounds which involves claims about empirical processes which are not (for humans at least) cognitively accessible just on reflection.

            Of course this follows from an (at least modest) externalist account of justification, and one might feel that this unfairly stacks the cards in favour of the desired thesis. But as Alston has explained (1989c, pp. 158-162), the types of beliefs which have figured prominently in discussions where it has been assumed that what justifies p also justifies <p is justified> have been beliefs about one’s current states of consciousness or self-presenting propositions. And although one might argue that such beliefs are justified by being produced by maximally reliable cognitive mechanisms, the road of high accessibility internalism has turned on notions such as self-warrant and truth-warrant.[9] Now it is certainly feasible to generate self-warranted beliefs that make reference to God. For instance, some person S may believe that she is thinking about God. I will also grant that here S could also be immediately justified in believing that she is thinking about God. But the kind of fact that is relevant to the self-warrant of beliefs on both levels is an introspective fact. And this hardly suggests that S is justified in believing that God exists (or believing propositions which self-evidently entail that God exists), still less that S is (immediately) justified in believing that she is justified in believing in God.

            The case for immediately justified epistemic beliefs from high accessibility internalism then is compatible with [ET*]. We can mount an argument to [ET*] without [E*] by arguing from

 

[D1]     A justifier J for a belief b is a T-justifier = Df. J sufficiently confers justification on both the belief b and the correlated higher-level belief B*.

 

 

and the conjunction of

 

 

[A]       Given any person S, S is immediately justified in the higher-level epistemic belief that <p is justified> only if[10] (a) S is immediately justified in the belief that p and (b) the belief that p is justified by a T-justifier.

 

[B]       Given any person S, if S is immediately justified in the belief that p, then if S’s correlated higher-level epistemic belief that <p is immediately justified> is justified, the belief that <p is immediately justified> is mediately justified, unless the justifier of p is a T-justifier.

 

[C]       Given any person S, if S is immediately justified in the belief that Pt, the justifier of Pt is not a T-justifier.

 

 to

 

 

[ET*]   Given any person S, S’s belief that <S’s belief that Pt is immediately justified> is justified only if S’s belief that <S’s belief that Pt is immediately justified> is based upon adequate reasons.

 

 

            Strong higher-level theistic evidentialism is still true, even if strong higher-level evidentialism as such is not true.

 

B. Swinburne and the Principle of Credulity

 

            A second objection to strong higher-level evidentialism may be mounted from an application of the principle of credulity. Since Richard Swinburne presented this case to me, I will follow his epistemological argument. Swinburne takes it as a principle of rationality that “(in the absence of special considerations) if it seems (epistemically) to a subject that x is present, then probably x is present; and what one seems to perceive is probably so” (Swinburne 1991, p. 254).[11]  From this principle of credulity (hereafter, POC) it follows that if it seems to some person S that <S's belief that Pt is justified>, then probably S's belief that Pt is justified. Swinburne adopts a view of basic propositions according to which these are propositions which (a) seem to S to be true and (b) S is inclined to believe them but not solely on the grounds of other propositions (which make them probable) (1981, pp. 20, 33, and 36).[12]  Moreover, he is explicit about there being no restrictions on what may count as a basic proposition (1981, p. 22). And such basic propositions become basic beliefs so long as none of S’s other basic beliefs render them improbable. On this account, epistemic beliefs may be immediately justified just by virtue of being basic beliefs (via POC) which are not rendered improbable by S’s other (basic) beliefs.

            The plausibility of this argument rests in large part on the sense given to “seems” in the statement of POC. Alston takes the sense of “seems” in POC to be presentational. If I seem to be experientially presented with X’s being F, then I am prima facie justified in believing that X is F.  But there is some difficulty in seeing how one could be experientially presented with beliefs being justified. An epistemic interpretation yields two possibilities. First, [It seems epistemically to S that <S’s belief that Pt is justified>] (call this proposition [O]) may mean that S is justified in supposing that S’s belief that Pt is justified. But then the plausibility of [O] will depend on what justification amounts to. As already noted, on a truth-conducive account, the plausibility of [O] would be greatly reduced.  On the other hand, we might take [O] to mean one is strongly inclined to believe that one’s belief in God is justified.[13]  This is how Swinburne takes it. But now the question switches as to how plausible the overall epistemological theory is.

            If we take “seems” in POC to mean simply that one is inclined (perhaps strongly) to believe, because of experience or reason, that one's belief that Pt is justified, it would seem that we have licensed a far too liberal position in our epistemology. The principle would give carte blanche (at least prima facie) to all propositions we are strongly inclined to believe - whatever the source of that motivation. If we adopt a source-relevant conception of justification, this will not do. People are inclined to believe all kinds of propositions for less than epistemically adequate reasons, such as fear or wishful thinking. But where justification requires “the justifier” to be causally operative in generating or sustaining the belief, a person may well be inclined to believe that a belief of his is justified when in fact it is not (since its causal ground is not efficacious vis-à-vis the epistemic point of view). What does seem to be true is that people often take it without reason that their beliefs are justified or rational, though this may not be as widespread as some might think.[14]  Some people, children for instance, lack the appropriate concepts to form beliefs about their beliefs, let alone the epistemic status of their beliefs.[15]  We have no reason to believe that the lower animals do. Nevertheless, among those who do hold views about the their beliefs, some of them take such beliefs as basic. But this fact supports only descriptive or psychological foundationalism, not epistemic foundationalism. An immediate or basic belief is a belief formed or held in some manner other than solely on the basis of other beliefs. This belief will also be immediately justified if justification is conferred on it by something other than some other belief(s).

            Swinburne’s position on basic beliefs seems to reduce to a version of negative coherence theory - unless S’s belief that p is made improbable by some other (basic) belief of S’s, S’s belief that p is justified. But an epistemologist who wants something positive in the way of evidence or grounds, even for basic beliefs, will not find this adequate. Of course, suppose that having an immediately justified epistemic belief amounts to (i) being inclined to believe that one’s belief is justified, but not on the (sole) basis of other beliefs and (ii) none of S’s basic beliefs render the belief that <S’s belief that p is justified> improbable. Then we can safely accept the point. However, we need only point out that either this is not an adequate account of justification or it fails to account for a broad range of other epistemic statuses and merits that would call for positive support by reasons.  Interestingly enough, one might read Swinburne’s main account of rational belief in terms of the mere possession of evidence, where that evidence is constituted by a person’s set of basic beliefs (1981, p. 33). And one wonders how this fits with the negative coherence position taken on basic beliefs. It is not clear how it is that basic beliefs are rational other than by being undefeated at any given time given S’s set of basic beliefs.[16]   However, Swinburne recognizes various senses of rational belief based on the possession of different kinds of evidence. Since at times he appears to require only that a person have certain (propositional) evidence to achieve a certain kind of rational belief (as opposed to basing one’s belief on the evidence), positive epistemic coherentism would be compatible with Swinburne’s psychological foundationalism.  So we might say that a basic belief is minimally rational just if one has no reasons that render it improbable. And a basic belief will enjoy greater degrees of rationality to the extent that a person has certain evidence that provides positive evidential support for it, even if the belief is not based on that evidence. This suggests one of two things depending on how we slice the epistemological cake. We may conclude that for some people epistemic beliefs may be immediately justified in the sense that they take it that some putative belief of theirs is justified (but not because of any other beliefs they hold) and that none of their other beliefs render this epistemic belief improbable. Alternatively, we may recognize the psychological basicality of epistemic beliefs but assess the degree of their rationality solely in terms of the (propositional) evidence a person has for it. Either way, I do not see that the thesis of this chapter is adversely affected by the argument from POC.

            Lastly, I suggest a final response to the present argument. This stems from a certain perspective on how we are to apply POC. Swinburne takes it that this principle of rationality applies to individual beliefs. All propositions (which strike a person as true) are innocent until proven guilty. Alston takes a different approach (1991c, p. 195). For Alston, POC does not properly apply to individual beliefs taken in isolation but rather to socially established practices of belief formation and sustenance that give rise to our individual beliefs. These practices may be generational (i.e., the practice of forming beliefs from nondoxastic input) or transformational practices (i.e., forming beliefs from doxastic input). Point being here that on Alston’s construal, it is not a single belief that is innocent until proven guilty but a well-established doxastic practice. With this approach to POC, it will be necessary to show, not just that it strikes some people that some of their beliefs are rational, but that we can locate a social practice of forming epistemic beliefs in a basic way. Without moving into the topic of how we are to individuate practices, I appeal to the intuitive plausibility of perceptual, memory, introspective, and inferential practices (which we distinguish by the sufficiently varied psychological states associated with them). I modestly suggest that we have no adequately similar well-established psychological and social practice of forming epistemic beliefs in a basic way. If anything, it seems that justified epistemic beliefs arise as a result of transformational practice. On a doxastic approach to epistemology, then, POC may be used to show that if we are ever justified in our beliefs about attributions of justification it is because these higher-level beliefs arise from a practice of associating such beliefs with propositional evidence or believing them on the basis of such evidence. In which case, the immediate justification of epistemic beliefs is unwarranted, even if its psychological counterpart is true.

 

C. Reliabilism and Higher-Level Justification

 

            One might suppose that on a reliabilist view of justification, higher-level beliefs could be immediately justified. So long as a higher-level belief is produced by a reliable process of belief formation then it is justified. And if this reliable process does not involve mediation through other beliefs or knowledge we would have a case of the immediate justification of higher-level beliefs.

            In the first place, it is rather evident that if we intend to carry over the assumption that the same thing justifies a lower-level belief and its correlated higher-level belief reliabilism will offer us no real case. Even if my being appeared to redly is an adequate ground for my sensory perceptual belief that <this is a red object in front of me>, it is difficult to see how this experiential input equally serves as a justifier for a belief which attributes justification to the original belief. In the one case we are involved in the identification of a physical object and the attribution of physical properties to this object through various phenomenal qualia. In the other case we are evaluating the belief so formed in terms of its epistemic status. The beliefs simply have content of a very different kind, and it seems that the sort of input that justifies the first is not sufficient to justify the second. As Alston points out, from the perspective of reliabilism “to believe justifiably that the belief that p is based on an adequate ground is to base that higher-level belief on adequate evidence that the ground of the belief that p is sufficiently indicative of its truth” (Alston 1991a, p. 17). In this case, though, to have a justification for a higher-level belief is to have in one’s possession the right kind of empirical evidence that supports the belief that the relevant belief-forming process involved in the production of the belief is a generally reliable one. And it is hard to see how the evidential basis for the lower-level belief sufficiently supplies this.

            Secondly, the sort of evidence which is required for higher-level justification does not generally accompany most beliefs, and certainly they are not - at least on the externalist view - matters one can come to grasp just upon reflection. It is just not the case that people are typically already in possession of the relevant evidence. The average epistemic subject is simply in the dark about what mechanism produced such and such a belief, let alone whether it was reliably produced. And if, as seems to be the case, being justified in higher-level claims along these lines also requires general beliefs about reliability claims, the prospects of having a justified belief for PJ* for most beliefs that p becomes even more implausible. The mere fact that a belief has been reliably produced seems wholly inadequate as a justifying ground for the belief that it has so been produced. It seems that we would have to adopts a particular form of internalism, high accessibility internalism,[17] according to which people have some sort of privileged cognitive access to epistemic facts about themselves, say just upon sufficient reflection. Only if we did this would there be some degree of plausibility in the immediate justification of (at least some) epistemic beliefs. But as already noted, such a case does not seem to carry force for theistic belief and is problematic on a truth-conducive view of justification.

            It might be urged that we just have mechanisms that yield epistemic beliefs in a reliable fashion, and so the prospects remain (or at least it is logically possible).  However, in the absence of evidence that shows that this is so, we are best advised to avoid such a commitment. After all, the evidence we do have suggests something quite to the contrary. The empirical evidence seems to support the view that typically the reliable (or at least most reliable) mode of forming higher-level beliefs is mediate - a claim which I take to be true no less in ordinary life as in science, philosophy, and theology.[18]  So whatever is logically possible, in fact humans seem to be constructed such that they do best in the cognitive enterprize vis-à-vis higher-level beliefs when they have considered the relevant evidence for reliability and/or which are the valid principles of epistemic justification and to what extent the belief in question satisfies the conditions laid down by such principles.[19]  Whatever might be the case for Alpha Centaurians, the human situation is such that the reliable process for the formation and justification of higher-level (theistic) beliefs is mediacy or ratiocination.[20]

 

III. Higher-Level Evidentialism and Reformed Epistemology

 

A. Plantinga and Higher-Level Theistic Evidentialism

           

            The question that now arises is whether the strong higher-level evidentialist requirement is compatible with Plantinga’s religious epistemology.  To the extent that Plantinga’s anti-evidentialism is construed only as a lower-level anti-evidentialism, the thesis of higher-level theistic evidentialism remains untouched. In chapter 3 we considered Plantinga’s position with respect to the de jure question of theistic (and Christian) belief.  Theistic belief may be held in a basic way without a person’s violating any intellectual duties and - more importantly - without being a consequence of cognitive malfunction of some sort. But there is an important distinction between imposing evidentialist requirements on belief in God and imposing it on belief in the rationality of one’s belief in God. Anti-evidentialism with respect to the former does not entail anti-evidentialism with respect to the latter.

            A more potentially difficult problem is rooted in the externalist dimension to Plantinga’s epistemology. Although Plantinga’s “warrant” and Alston’s “justification” coincide with respect to a reliabilist constraint, they diverge on the issue of whether grounds are necessary for positive epistemic status (or the epistemic desideratum that each is targeting). This internalist feature of “grounds” is a crucial point of divergence between Alstonian justification and Plantingian warrant. There is not space to defend Alston’s view that justification always involves grounds (which are cognitively accessible), and fortunately there is no need to do so. First, for Plantinga warrant is the crucial condition which transforms true belief into knowledge, but Alston (distinguishing between justification and knowledge) holds that one may have knowledge as a result solely of a reliable process of belief formation. Here there is substantial agreement on the absence of internalist requirements for knowledge (though Plantinga’s externalism includes proper function, not just reliability). Secondly, many putative cases of Plantinga’s properly basic theistic beliefs apparently correspond to Alston’s nonsensory perceptual awareness of God, and so involve grounds. (Though - as explained in chapter 3 - Plantinga has also suggested that belief in God may sometimes be more akin to memory or a priori beliefs which as he sees it do not have anything in the way of Alstonian grounds). Fair enough. But any form of mediate justification, or what Plantinga calls propositional warrant, involves the relation of one belief to at least one other belief (as a ground) and so introduces a mild internalist condition. Plantinga recognizes that we sometimes form beliefs on the basis of other beliefs (1993b, p.137-139), and even that sometimes the reasons in question include beliefs in the adequacy or supports relation (1993b, p. 44), and that beliefs so formed are warranted. So thirdly, even if there is no internalist feature like grounds in Plantinga’s concept of warrant, there will still be conditions under which a belief is warranted only if it has grounds. That is precisely the case with what Plantinga calls propositional warrant. As I have explained earlier in the thesis, one may be a thoroughgoing externalist with respect to one’s concept of justification (or warrant), but recognize a wide range of internalist constraints on the conditions for the justification (or warrant) of beliefs of this or that sort.

            The account of higher-level evidentialism has been developed mainly along Alstonian-internalist lines, but it is rich enough to be adapted to Plantinga’s more externalist approach. How would that go?  Plantinga’s view of epistemic conditional probability involves an objective and normative component. The objective component is a kind of logical probability, but one not merely involving conditionalization on the evidence, but conditionalization on other propositions as well.[21] The normative component is a matter of the degree of confidence a rational person would have in some proposition A given B. And the other conditions of warrant must apply as well. A person’s relevant cognitive faculties must be functioning properly in the appropriate environment, etc. So what is required for warrant by epistemic conditional probability is not merely an evidential relation between propositions but the fact that we are functioning properly when we so form beliefs (1993b, pp. 192-193). Moreover, because of the function that the design plan plays, epistemic conditional probability will not involve noncontingent statements that are either necessarily true or necessarily false. Plantinga holds that epistemic probabilities are contingent (1993b, pp. 172-173). It is a feature of our design plan as rational beings and as such could have been otherwise. Once again, warrant is externalistic. For take the case where warrant is conferred on S's belief that p by some other belief that q (by virtue of q making p epistemically probable), and where S holds the belief that p on the basis of the belief that q.  In this case another belief forms the ground of the belief that p, and so there is an internalist dimension in this case. But what is also required is proper function. It is only by virtue of externalist factors that epistemic conditional probability is efficacious in producing warranted beliefs.       

It is should be clear that even if Plantinga’s position does not clearly support higher-level theistic evidentialism it is not inconsistent with it. The main reason is this. Plantinga concedes that there is warrant by epistemic conditional probability. So there are circumstances in which the design plan dictates that a belief B is warranted when formed (or sustained) on the basis of another proposition which renders B probable. It is possible then to state further that we are designed in such a way that higher-order beliefs only have warrant for us when they are based on the appropriate (deductive or inductive) propositional evidence, or maybe this is the case with theistic belief in particular. In other words, suppose that the design plan stipulates that for rational human agents, that segment of their noetic structure which is responsible for the production and sustenance of higher-level beliefs is functioning properly only if higher-level beliefs are based on other beliefs that constitute good evidence for them. We typically regard beliefs about the warranted status of some belief as itself warranted primarily because we have considered the relevant evidence and have found that evidence to support it. Plantinga of course distinguishes between normal functioning and proper functioning, and it might be thought that all that this establishes is an empirical case of how we functioning normally.  The question is, then, how do we go about determining just what the design plan is?

            Plantinga has emphasized that the de jure question of theistic belief cannot be answered indepedently of the de facto question. Our metaphysics will govern what we think a human person is and hence govern our perspective on proper function in our cognitive life (e.g., which beliefs are properly basic and properly nonbasic, etc). Although there is some sense to this with respect to the kinds of beliefs we ought to hold, it is precarious indeed to argue this with respect to the mode in which we hold certain beliefs. This was after all one of the consequences of the analysis and argumentation of chapters 3-5. The suggestion that nonbasic theistic belief is epistemically defective or epistemically inferior just doesn’t seem plausible, even given a wide range of theistic or religious assumptions. All the more am I at a loss, then, to see how a theistic or Christian metaphysics has anything to say about whether the strong higher-level evidentialist requirement is true.  I think that the rather close tie between the reliable generation of higher-level beliefs and propositional grounds provides at least a presumption for regarding this mode of belief formation as an instance of the proper functioning of one aspect of our cognitive establishment. Moreover, perhaps we could work inductively (as Plantinga has suggested with reference to determining which beliefs are properly basic) to arrive at some degree of confirmation for the hypothesis that higher-level evidentialism is true, or at least is true with reference to theistic belief.

 

B. The Analogical Argument from Basic Theistic Sources

 

            The only remotely plausible argument against higher-level theistic evidentialism from the perspective of Reformed theology is what I will call the analogical argument from basic theistic sources, which leads to higher-level Reformed epistemology (the proper basicality of higher-order theistic beliefs).[22] I have already spent some time (in chapter 5) developing the sensus divinitatis as an immediate mode of theistic belief formation. What I have not discussed is another apparent mode of religious belief formation mentioned by Calvin, and that is the inner testimony of the Holy Spirit. Calvin seems to look at this phenomenon as a belief-forming mechanism (apparently supernatural, as opposed to the natural sensus divinitatis) which is responsible for the formation and sustenance of immediate belief in the divine authority of the sacred Scriptures. Actually, to be more precise, the witness of the Spirit is the ground of the certainty the believer is said to have regarding the authority and divine origin of Scripture (cf. Institutes I.vii.4,5). With this position in mind, one might try to extend the epistemic functions of either the sensus divinitatis or the inner testimony of the Holy Spirit to cover the range of higher-level beliefs. If we are designed so that one of the ways in which we form (lower-level) theistic beliefs is on the basis of the inner testimony of the Holy Spirit (or the sensus divinitatis), surely it is reasonable to think that we could be designed in such a way that the Holy Spirit also testifies to us (or the sensus divinitatis grounds) the fact that the putative lower-level theistic belief is reliably formed, justified, or warranted.

            The obvious question is why we should think that the internal testimony of the Spirit (or the sensus divinitatis) is a source for higher-level beliefs in the first place. I think the only remotely plausible case here must draw on reasons similar to those adduced by Calvin with reference to the assurance that Scripture is the Word of God. Suppose that second-order doubts lead to first-order doubts. The argument can be worked out two ways. Suppose that (i) higher-level beliefs are hard to come by or (ii) we altogether lack a source for certainty about our epistemic beliefs. In both cases, problems at the higher-level might be thought to affect matters on the lower level - a kind of trickle-down skepticism. Higher-level evidentialism at the very least implies that higher-level justification is hard work (at least requires some moderate degree of reflection and consideration of evidence). This leaves many theists in the cold with respect to higher-level justification. And even if we work through the arguments of natural theology, we don’t end up with certainty. And will not the lack of certainty about a belief’s being justified lead to doubt about the belief itself? These points suggest that certainty regarding higher-level beliefs is religiously significant. Perhaps the certainty of faith will be jeopardized if we didn’t have an internal testimony to higher-level beliefs. In short, there is a concern here about matters at the higher-level adversely affecting beliefs at the lower level. And if requirements at the higher-level are made too stiff, this will cause epistemic deflation at the lower-level, thereby affecting the degree to which a person holds a belief at the lower level.

            But the argument seems to be vitiated by several epistemological confusions.

            For one, there seems to be an ambiguity present in the argument. It is necessary to distinguish between (a) lacking the belief that <p is justified> and (b) possessing the belief that <p is not justified>. Although one might be able to argue that (b) could raise doubts in a subject about the belief that p, the same is not an implication of (a). It does seem a bit odd to say that S is strongly inclined to believe that <p is not justified> but that S is at the same strongly inclined to believe that p (though I suspect that one’s concept of belief and justification is crucial for deciding the plausibility of this).  Surely, though, it is eminently plausible to say that S may be strongly inclined to believe that p without S’s having been so reflective as to have thought about or formed a belief about the epistemic status of the belief that p. In other words, S may simply lack the higher-level belief that <p is justified> and yet “be disposed to feel it true that p” upon considering p.  The lack of having a higher-level belief (justified or not) simply does not bear on the possession of a belief that p or the degree to which one believes that p.  Moreover, I am inclined to think that many contemporary theists are not ordinarily reflective about their beliefs. And among those who either do believe that their religious beliefs are rational or would believe it upon reflection, it is because they have “thought it over” and believe for a reason, however implicit the reasons are or however unsophisticated their arguments may be. And this would suggest that the higher-level belief is based on reasons.

            But the real question is whether such beliefs are immediately justified.  We must consider what sorts of sources could be reasonable as immediate justifiers. The last chapter considered the prospects for this for several putative immediate justifiers and came up with a negative verdict. One possibility not discussed there was testimony, and especially the testimony of the Holy Spirit. But the testimony of the Spirit seems radically implausible as a source for immediate justification. Not only are there important disanalogies between human testimony and what is called the testimony of the Spirit, but most fundamentally testimony itself is not (typically) an immediate mode of belief formation. The point may be contested but it does strike me that when we come to hold a belief on the say-so of another person typically what actually happens is that we come to hold the proposition in question on the basis of other beliefs. For instance, I will usually hold the belief that p on the basis of my belief that S asserted that p, that S believes that p, that S is justified in believing that p, or that S can be relied on as a credible witness. At least this seems to be true for the testimony principle as it operates in adults. In this case, a person will be justified in holding some proposition on testimony only if he is also justified in these other beliefs and believe the relevant proposition on at least the partial basis of one of these other beliefs.[23]  I think what has led to thinking of testimony as an immediate source is equating mediate justification with a process of inference or argument, in which case testimony is obviously not mediate. But this is simply too narrow of a conception of mediate justification. So even if the Holy Spirit could function as a testimonial basis for justified epistemic beliefs, this would not suffice for regarding such beliefs as immediately justified.[24]

            The analogical argument from basic theistic sources seems to me implausible and obviously unnecessary. And in the absence of any worked out argument from theistic metaphysics to the proper basicality of epistemic beliefs (and in the light of the failure of other arguments against higher-level theistic evidentialism canvassed earlier), the reasonable position is to regard the strong version of higher-level evidentialism as established. Moreover, I would suggest that a more plausible function for the Holy Spirit (besides grounding firm first-order religious beliefs) would be to assist a person in seeing the force of certain evidence (including a grasp of certain inductive standards) that may be relevant for one’s becoming justified or warranted at the higher level. So, as I suggested with reference to the sensus divinitatis, the epistemic function of the Holy Spirit need not be restricted to immediate modes of belief formation and sustenance.

 

IV. Conclusion

 

            In this chapter I have concluded my examination of the strong evidentialist requirement and the related cognitive desideratum of reflective rationality. The requirement is found to be epistemically relevant and significant by virtue of the ways reflective rationality affects justificational status at the lower level. Once again, defeating conditions dictate the way we think of epistemic level relations and evidentialist requirements for belief. Moreover, the philosophical and theological cases against strong higher-level theistic evidentialism have been sufficiently answered. There is no substantial case against the justification of beliefs involving the attribution of justification requiring mediate justification. In the final place, the internalist conditions for justification that are entailed by higher-level evidentialism are compatible with most externalist theories of justification and knowledge, and especially with Plantinga’s theory of warrant and proper function.