Over the past thirty years or so, philosophers in general and philosophers of religion in particular have debated the merits and demerits of foundationalism, its detractors often hastening to announce its demise with the ill-concealed relief of one glad to see the last of a cantankerous neighbour. In the course of the discussions, various figures pop up in the guise of heroes or victims, those confidently taken to be paradigmatic foundationalists or falsely accused of being such. Aquinas is one of these: sometimes named a foundationalist casually and en passant, as if his status as such were obvious; sometimes asserted to be so on the basis of detailed arguments; rather less often it is claimed he is no foundationalist at all, or at most, a foundationalist of a very qualified kind. I propose here to survey these arguments, for and against, with a view to demonstrating that Aquinas is not a foundationalist, for reasons that are rather more radical than the few in the non-foundationalist camp have yet suggested.

I begin with those convinced Aquinas is a foundationalist, some of whom hold him to be deficient or misguided for being so, some of whom attach no culpability to his supposed foundationalism. The two most detailed arguments for Aquinas-as-foundationalist come from two thinkers hostile to foundationalism, Alvin Plantinga and Nicholas Wolterstorff.Footnote ¹Plantinga's claims are most clearly expounded in a 1983 essay, “Reason and Belief in God”, although he assumes a slightly chastened version of the position in later works as well.Footnote ² It should be noted at the outset that the 1983 essay is substantially, though by no means solely, concerned with the issue of Aquinas’ foundationalism;Footnote ³ equally important to Plantinga is to show that foundationalism is both false and self-referentially incoherent and therefore to be rejected (1983: 17 and 90) and it is possible that his reading of Aquinas became skewed by a need to find a representative for views he wished to argue against.

One feature Plantinga's reading shares with most others is its scant attention to the opening question of the Summa theologiae. He opens his examination with a lengthy analysis of Aquinas’ notion of scientia, but one which does not take account of the discussion in ST I, q. 1 (references to the ST hereafter are to the Prima Pars unless stated otherwise). The failure to consider the latter may explain why Plantinga can make the claim that Aquinas seems to follow Aristotle in holding that scientia consists in a body of propositions deduced syllogistically from self-evident first principles (1983: 40). Plantinga does not continue to assert much in the way of the supposedly syllogistically-structured element of Aquinas’ thought (though we might question whether he was too hastily dubbing it specifically syllogistic). Whether or not it is appropriate to saddle Aquinas with the employment of a syllogistic form, though, it is even less clear that the first principles of scientia are for Aquinas supposed to be self-evident. A little later Plantinga provides a more informal version of the claim. Now Aquinas’ picture of knowledge amounts to the claim that we know what we see to be true together with what we infer (1983: 44), an account which resembles that which Plantinga gives of foundationalism: a rational noetic strucutre with a foundation consisting of a set of beliefs not accepted on the basis of others (1983: 52). Thus far, Plantinga's picture of foundationalism is purely structural: it has to do with a particular kind of relationality among beliefs, with no claim of greater certitude being accorded to some of these. Foundationalists however generally require more of their foundational beliefs than that they are not accepted on the basis of others. The latter criterion could apply to any proposition functioning as a postulate, including wildly fanciful ones that would be unacceptable to any self-respecting foundationalist, and Plantinga acknowledges as much when he introduces into his account the notion of proper basicality, ‘properly basic’ being his favoured term for the legitimate starting propositions of a foundationalist body of knowledge. He repeatedly makes claims regarding what Aquinas considers to be properly basic (1983: 55, 57, 58; 1986: 112, 122), the immediate difficulty with which is that it is unclear what they correspond to in Aquinas’ own terms. The issue is not cosmetic: the problem is not simply that Aquinas does not use the phrase ‘properly basic’, but that he does not appeal to any equivalent notion.

Although it is unclear what the concept of proper basicality would apply to in Aquinas’ terms, Plantinga nevertheless introduces a criterion of proper basicality which he attributes to Thomas: Aquinas, he claims, held that a proposition is properly basic for a person only if it is self-evident to him, or evident to the senses (1983: 55). This point seems important to Plantinga, for he articulates it repeatedly (1983: 57, 58; 1986: 112, 122).Footnote ⁴ Nowhere does Plantinga ground this assertion in any text of Aquinas’, however, and it is a puzzling claim, especially in light of the first question of the ST. Certainly, Aquinas believes we acquire knowledge through the senses, but his position is a rejection of innatist or anamnetic theories–we learn by seeing and hearing–and that stance by no means commits him to maintaining that knowledge must be either of sense perceptual data, or self-evident or derived from one of these. The Biblical data from which the chief of all scientiae takes its starting point falls into none of these categories, for example.

Plantinga's case for Aquinas’ foundationalism thus rests, first, on a claim about the structure of foundationalist thought or arguments, and second, on a claim about what can count as foundational (properly basic, in Plantinga's terms) in these structures. The structure he posits is compatible with what Aquinas himself describes, but detached from claims about what can count as a foundational proposition, the structure itself is an at best weak indicator of foundationalism. A second criterion of foundationalism–that which identifies what can count as foundational–is applied to Aquinas with no textual justification, and Plantinga's account of Aquinas’ thought is flatly incompatible with what Aquinas himself claims in a key text. There is a third distinguishing characteristic of foundationalism, one to which Plantinga only alludes in passing in his treatment of Aquinas, but which looms large in other accounts of foundationalism: as Plantinga puts it “other foundationalists” (other than Aquinas, that is) have insisted propositions basic in a rational noetic structure must be certain in some important sense (1983: 58). I will return to this issue of certainty later on; for the moment, we need only note that a criterion Plantinga acknowledges to be widely important is one from which he exempts Aquinas, and his doing so necessarily raises the question whether the criteria for foundationalism do not have to be fiddled if Plantinga is to make Aquinas fit them.

In turning from Plantinga to his fellow champion of Reformed Epistemology, Nicholas Wolterstorff, we find a similar, if more nuanced, view of Aquinas, though one no less insistent on his foundationalism. Wolterstorff's views modified somewhat over time, his later ones being more moderate, but we begin with the more uncompromising claims of Reason within the Limits of Religion. Here Wolterstorff claims that Aquinas, along with a number of other thinkers, were all foundationalists and that Aquinas offers a classic version of foundationalism (1976/84: 30). Wolterstorff's definition of foundationalism differs slightly from Plantinga's, notably in its recourse to the concept of scientia. Foundationalism, on this account, is the view that a person is warranted in accepting a theory at a certain time if and only if he is then warranted in believing that this theory belongs to genuine scientia (1976/84: 28). This definition merely raises the question of what it means to belong to a genuine scientia, which Wolterstorff clarifies on the next page: a theory belongs to a genuine scientia if and only if it is justified by some foundational proposition and some human being could know with certitude that it is thus justified (1976/84: 29). This account will not, of course, fit what Aquinas maintains about scientia in the larger Summa; with a nod to the latter, Wolterstorff acknowledges that his account of the Thomistic view of properly conducted inquiry is not true of the science of theology because theology is a science of a special sort (1976/84 149 n. 11) and this sort of scientia is lower or subordinate to the first kind (1986: 66–67). Wolterstorff is partly correct: there are two distinct varieties of scientia and sacred doctrine is for Thomas a scientia of the second kind. It is not however an exception to the general principles governing scientiae, nor is it by any means a lone deviation from the first category of scientia, nor do the qualities that make it a scientia of the second sort make it a lesser variety of scientia: quite to the contrary, sacred doctrine is the noblest of all scientiae.Footnote ⁵

Perhaps because of his mistaken assumption about a hierarchy of scientiae, Wolterstorff sets up a distinction between faith and reason which Aquinas does not make: faith complements reason, on Wolterstorff's reading of Aquinas (1976/84: 31), ‘complement’ apparently meaning here that the two operate in distinct realms. From that assumption, it is only a short step to an evaluative position: what follows immediately from the Thomistic view is intellectual elitism (1976/84: 139): those who can, reason, and those who cannot, hold to their unreasoned faith. Faith, Wolterstorff clarifies in a later work, means for Aquinas accepting propositions on the authority of what God reveals (1983: 141). Because Wolterstorff, like Plantinga, is convinced that the Thomistic conception of knowledge limits it to what is self-evident, evident to the senses or demonstrated from these two (1983: 141), it would seem faith and knowledge must occupy separate spheres. Out of the latter conviction comes the germ of an idea Wolterstorff would later develop more fully: Aquinas (along with Anselm) engaged in a “transmutation project” of altering belief or faith into knowledge (1983: 141).

This latter position forms the centrepiece of Wolterstorff's 1986 article “The Migration of Theistic Arguments”. By this time, his views had shifted somewhat, becoming less hostile to Aquinas (no more charges of elitism, for instance), though he remains convinced of the latter's foundationalism. For Thomas, he claims again, scientia has a foundationalist character, which subsists in its being grounded in what is self-evident or evident to the senses.Footnote ⁶ Now, however, he acknowledges that Thomas’ goal differs, at least from that of the evidentialist, since the purpose of natural theology (natural, note) is to enhance happiness (1986: 60). This “telic view of the end of man” is paired off with a foundationalist view of science (1986: 80). Although in Aquinas as in Locke one finds the foundationalist vision of grounding one's views in certitude (1986: 80), Wolterstorff maintains that Aquinas, like Calvin, did not attach foundationalist conditions to the Christian faith (1986: 80). Where does certitude come in? supposedly because for Aquinas faith and revelation are correlatives (1986: 62) and the reason we know we can trust revelation is that it is confirmed by miracles (1986: 64). It is this faith, its rectitude attested by the miraculous, that is the content of sacred doctrine (1986: 66).

Like Plantinga, Wolterstorff provides no textual evidence for his claim that knowledge for Aquinas must be based on what is self-evident or evident to the senses, a position that is all the more puzzling given that he does actively engage with q. 1, as Plantinga did not. Perhaps the notion that sacred doctrine is an exceptional sort of scientia is Wolterstorff's way of reconciling q. 1 with his epistemological claims, even though in there, Aquinas so far from providing support for the notion that sacred doctrine is a lower or subordinate science as Wolterstorff contends, unambiguously states precisely the opposite. Although he is partly correct in holding that the goal of the enterprise diverges from that of evidentialism, he offers no evidence for the claim that its impetus springs from the desire for certitude. For Thomas, certitude pertains not to the foundations of scientia, but its pinnacle, for it is the blessed who will enjoy the fullness of knowledge, and certitude would therefore seem reserved for the next life. If Aquinas holds to the foundationalist vision of beliefs grounded in certitude, then his foundationalism would never be realisable in this life. The notion that miracles make faith rationally acceptable simply has no basis in Aquinas’ thought,Footnote ⁷ nor is there any warrant for equating faith and sacred doctrine, as Wolterstorff does.

In Scott MacDonald, we find a very different sort of case for Aquinas the foundationalist. Like Plantinga, MacDonald barely considers q. 1, though his claims have greater textual justification. In MacDonald's reading, scientia requires foundations (1993: 168) and Aquinas offers foundationalist arguments, at least for the existence of a good creator (1993: 1987). Aquinas is not just an incidental foundationalist, in MacDonald's view; he is a theorist of foundationalism as well, offering two different sorts of argument for the claim that scientia requires foundations. The first of these is negative: it supports foundationalism by default, in holding that inferential justification is only possible if there is non-inferential justification (1993: 166).Footnote ⁸ The foundations of scientia are these non-inferentially known epistemic first principles (1993: 165). MacDonald allows for “non-paradigmatic” scientiae which take non-immediate propositions in their foundations (1993: 178) and claims that Aquinas recognises “a sort of justification” (1993: 179) from dialectical or probable reasoning, as well as the justifiability of holding some beliefs on the basis of reasonable authority (1993: 180). In these acknowledgements, he differs from Plantinga, of whose reading of Aquinas MacDonald is rather critical (1993: 178 and notes 57 and 58 on 193). MacDonald's account nevertheless resembles that of Wolterstorff in contending that a scientia whose foundational principles are not immediate is in some sense bracketable for the purposes of understanding Aquinas’ notion of scientia (these are MacDonald's “non-paradigmatic” scientiae, which correspond to Wolterstorff's “subordinate” ones); like Wolterstorff, MacDonald offers no justification for this claim. Although he more carefully grounds his discussion in Thomas’ work than either Wolterstorff or Plantinga, his evidence comes almost entirely from Thomas’ commentary on Aristotle's Posterior Analytics, with no acknowledgement of the obvious problem with such a procedure: no matter how much an admirer of Aristotle Aquinas may have been, a commentary on someone else's work cannot stand as a reliable sole guide to the commentator's own views. The problem in this instance is compounded by the fact that Aquinas was by no means an uncritical or automatic assimilator of Aristotelian philosophy, and if he parted company with it at crucial points, his generic agreement with it cannot be simply taken for granted.Footnote ⁹

With Ralph McInerny, we shift direction somewhat. McInerny stands between those like Plantinga, Wolterstorff and MacDonald, who are sure Aquinas is a foundationalist, and Eleonore Stump, who is sure he is not. McInerny's view is marked by a distinction between religious truths peculiar to theism and preambles to faith, on the one hand, and articles of faith on the other. For McInerny, Aquinas is a foundationalist with respect to philosophical theism, but not with respect to truths peculiar to Christian faith (1986: 287), foundationalism being characterised simply as the idea that some things are more knowable and basic than others (1986: 282). If that principle is sufficiently definitive of foundationalism, there would be few bodies of thought which would not count as foundationalist, in which case, of course, it matters little whether or not Aquinas is deemed a foundationalist (though McInerny explicitly acknowledges that by this definition, he is [1986: 282]). The blandness of this determination signifies McInerny's real concerns, which are to show Aquinas is neither an evidentialist nor a fideist, fideism being defined as the claim that nothing we know counts for our against faith (1986: 284) and evidentialism as the view that every religious believer has an obligation to seek evidence for his beliefs (1986: 283); since Thomas lays no such obligation on believers (1986: 283), it seems he could not count as an evidentialist.Footnote ¹⁰ For McInerny, Aquinas is a foundationalist who rejects fideism (1986: 284). McInerny's view becomes less clear when we probe the details of his position, however, for the last claim would leave open the possibility that Aquinas is a foundationalist with respect to religious truths as well as philosophical ones, although McInerny seems elsewhere concerned to limit his foundationalism to philosophical theism (1986: 287).

Although McInerny's account responds to some of Wolterstorff's claims (and, implicitly, many of Plantinga's), he does not actually shed much light on the question of Aquinas’ foundationalism, in part because some of his definitions are too broad. This indefinitive quality is obvious in his definition of foundationalism, but also in his glossing of scientia, which he defines as “knowledge in the strongest sense of the term”, although he acknowledges that unlike human knowledge in general, scientia is hard to come by (1986: 282). Concerned though he is to distinguish between the realms of the natural and the supernatural, between the preambles of faith deriving from philosophy and the articles of faith pertaining strictly to religion (1986: 283), he is little concerned to distinguish between kinds of scientiae, and the inexactitude of his account of scientia is arguably where his discussion goes astray, since for Aquinas scientiae are not distinguished by their subject matter, as McInerny would have it. On McInerny's account, Aquinas follows Aristotle in distinguishing between natural science, mathematics, and divine science/metaphysics/natural theology (1986: 287). This tripartite division according to the content of a subject bears little resemblance to the bipartite formal distinction proposed in the ST. If Plantinga and Wolterstorff are mistaken about what counts for Aquinas as the starting point of a scientia, McInerny seems indifferent to its structure.

Like McInerny, John Greco gives a yes-and-no answer to the question of Aquinas’ foundationalism, and like McInerny, his answer is largely the product of his starting definition of foundationalism, although Greco's definition is rather sharper and more telling: foundationalism is an answer to the sceptical problem of infinite regress (1998: 34, 35). On Greco's account, foundationalism requires one must have good reasons for whatever one claims to know, reasons one believes to be true (1998: 35), but these reasons need not be demonstrative, nor need they be certain, irrevisable or indubitable (1998: 39, 36). In Greco's view, many objections to foundationalism are based on false assumptions about the status of foundationalist beliefs, mistakenly requiring more of them than the definition of foundationalism itself requires. Greco can therefore maintain that because Aquinas follows Aristotle in holding that some states get their epistemic status in ways other than inference from other reasons, Aquinas was a foundationalist “for all kinds of positive epistemic states” (1998: 40).

Greco's argument depends on an almost entirely structural account of foundationalism, a description of a two-tier structure of knowledge or belief where some beliefs are properly basic and others are derived from these (1998: 36). Greco not only acknowledges, as Plantinga did, that what can count as properly basic differs from theorist to theorist (1998: 36); he goes further, in his denials that basic beliefs need be demonstrative, certain or guaranteed to be true. Given that Greco declines to give any positive criteria for proper basicality, it would seem that almost any belief could count as basic, as long as the one holding it does not attempt to justify it by recourse to other beliefs. It is not clear why, if halting the infinite regress of justification is crucial to foundationalism, one does not have to posit a stronger barrier to it than Greco seems willing to do, and his account is therefore of questionable internal coherence: regress, it seems, could be halted by brute assertion, and if so, there would be no need to posit a two-tier structure at all.

Aquinas, for Greco, is however not a classical foundationalist in Plantinga's sense because he did not require demonstration for belief to be rational, but only for “evident truths for scientia about God” (1998: 40). Greco cites none of Thomas’ works as the basis for this latter claim, nor does he furnish evidence in support of his subsequent assertion that for Aquinas, beliefs about God are rational because they are based on good testimony, the authority for which is confirmed by miracles “and other aspects of the history of Christianity” (1998: 40). As evidence for the last claim, he cites the work of MacDonald and Wolterstorff which we have already surveyed and which were themselves inadequately documented.Footnote ¹¹

Greco develops his views in distinction to those of Eleonore Stump, whom we will consider shortly and who most clearly denies Aquinas is a foundationalist. Greco maintains that if Stump is correct about Aquinas’ condition for knowledge (“human knowledge is a function of our using the cognitive capacities God created in us as God designed them to be used in the world God created them to be used in”Footnote ¹²) then Aquinas is not a classical foundationalist regarding rational belief or knowledge (1998: 40). Although Greco denies Aquinas is a classical foundationalist, he nevertheless holds that Aquinas required demonstration from evident truths for scientia about God, even though he required no such thing for rational belief itself (1998: 40). It is however because the foundations of knowledge are broader than those required for scientia that Greco can conclude Aquinas is not a classical foundationalist. Nevertheless, the latter determination comes as a surprise, given the assertion earlier on the same page that “it is reasonable to conclude that Aquinas was a foundationalist for all kinds of positive epistemic status” (1998: 40). Although considering whether Aquinas is a foundationalist is one of Greco's stated aims (1998: 35), the essay leaves the reader finally unclear about Greco's position on the matter. Whatever we are to make of his final judgement, however, the fact that he nowhere engages Aquinas’ own writings, but relies solely on other accounts which are problematic in ways we have already noted, calls into question any final determination he might have made. We might additionally query the vagueness of his criteria for foundational and properly basic claims.

Eleanore Stump falls most clearly into the camp holding Aquinas is not a foundationalist.Footnote ¹³ Her account is characterised by a clarity about terms and sources whose absence causes some of the problems evident in others we have surveyed, yet like MacDonald, she bases her examination on Aquinas’ commentary on the Posterior Analytics and like MacDonald, does not address the fundamental problem entailed in using a commentary on someone else's work as evidence for Aquinas’ own views. Again like MacDonald, she does not engage the account of scientia in the Summa theologiae.

Nevertheless, she acknowledges that one of the problems in determining whether Aquinas is a foundationalist lies in the precise meaning one attaches to the term, so she offers a limited definition (one intended to be valid for her own purposes). This definition focuses on the structure of foundationalism (some propositions not inferred, others based on these non-inferred ones), and certainty as the objective of the structure. On this point, she follows Keith Lehrer (Stump 1991: 130). Accordingly, for Stump, the certainty of inferred propositions in the foundationalist structure is guaranteed by the certainty of the non-inferred propositions, which thus underwrite the whole.

While she acknowledges Aquinas’ commentary on the Posterior Analytics could give the impression he is a foundationalist,Footnote ¹⁴Stump concludes he is not, in significant part because of the non-equivalence of key Latin terms with their standard English translations. Certitudo, for example, is not the same as ‘certainty’ (1991: 143), scientia is not equivalent to ‘knowledge’ and Aquinas’ theory of scientia therefore cannot be taken as a theory of knowledge (1991: 133, 136).Footnote ¹⁵ Furthermore, some of the propositions guaranteeing knowledge (such as reports of the senses) can be false, according to Aquinas, and error can arise from demonstrations (1991: 137, 143). These considerations alone suffice to show Aquinas is not a foundationalist (1991: 143), although Stump goes on to give a general account of his epistemology which is intended to strengthen her case.

Stump's argument is plausible in many respects and serves as a valuable corrective to some of the flaws of other accounts, flaws which underwrote the conclusion that Aquinas is a foundationalist. A serious weakness of her argument, however, is the absence of recognition of the problems entailed in concluding anything on the basis of a commentary on Aristotle's work. Although, as Stump notes, it might not be unreasonable to suppose Aquinas might share Aristotle's views, it is not valid to assume he does. He follows Aristotle on many points, but also rejects him on others, and it is illegitimate to argue from a commentary to a claim about Aquinas’ own views in the absence of any explicit textual evidence that Aquinas himself held the views on which he comments, evidence which neither Stump nor MacDonald furnishes, any more than either of them acknowledges the issue itself. While her observations regarding Aquinas’ insistence on the fallibility of sense impressions and demonstrations serve as significant caveats, the failure to take account of ST I q. 1 means that what she claims are the foundations of scientia are much more restrictive than Aquinas himself allows.

Thus far, the cases both for and against the notion that Aquinas is a foundationalist are unpersuasive, either founded, like those of Plantinga, Wolterstorff and Greco, on scant engagement with Aquinas’ own work, or, like those of MacDonald and Stump, on a work questionably representative of Aquinas’ own views. The problems raised by Stump–of how exactly one defines foundationalism–creates an added difficulty. What I propose to do is, first, offer a definition of foundationalism against which Aquinas’ thought can be measured, and second, to examine the claims regarding scientia in ST I q. 1, to determine what they indicate of his position, in relation to the baseline definition.

First, however, let me offer an explanation of why the examination of the first question of the ST is crucial to the issue at hand. Perhaps it has received little attention from those on either side of this debate because the disputants regard the issue at hand as an exclusively philosophical one, and therefore take the Summa theologiae to be irrelevant, attesting as it does to theological claims rather than philosophical ones. (I am here attributing a view which is not explicitly stated in the works of any of the authors surveyed; none of those who neglects ST I q. 1 explains why they discuss scientia without engaging one of the most substantial discussions of the term anywhere in Thomas’ works.) If this is the reason for overlooking the ST, then it is not a very cogent one, first because the line between philosophy and theology in medieval thought in general, and Aquinas’ in particular, is rarely drawn so emphatically as to enable one to separate the two out from each other. Second, epistemology is an area particularly ill-suited to polarising ventures: to maintain the structure of knowledge differs depending on whether one is operating in the realm of theological claims or other sorts of propositions would drive in the direction of a double-truth theory, which Aquinas certainly did not do. These grounds for ignoring the ST are unconvincing, therefore.

On the positive side, one further reason to consider the ST is that it was Aquinas’ last work, and therefore embodies his mature thought. Although it does not purport to be a statement of Aquinas’ own views in the sense of these being novel or original to him,Footnote ¹⁶ it can be taken as a reliable guide to views Aquinas himself held, which is not an assumption that can necessarily be made of his commentaries on Aristotle. The most compelling reason to consult the ST, however, is its direct and sustained treatment of scientia in the first question. All the accounts of Aquinas-as-foundationalist surveyed here acknowledged the significance of the category scientia to the issue and in light of that recognition, it is all the odder to ignore a sustained treatment of the subject, especially in a late work. In confining the discussion that follows to ST I q. 1, I nevertheless do not mean to suggest that it is the sole source in the Thomistic corpus relevant to the issue, but only that it is one to which insufficient attention has been paid in examinations of Aquinas’ status as a foundationalist.

Before turning to the ST's treatment of scientia, however, it is important to clarify the sense of foundationalism against which Aquinas’ account will be measured. Accounts of foundationalism commonly emphasise its structure: some propositions are inferred from others, while some propositions are independent of such inference. The non-inferred propositions provide epistemic justification, evidence or certitude for those inferred from them. Those offering definitions do not agree over the the strength of the claim being made on behalf of the basic or foundational propositions, and there is obviously a large difference between providing some evidence for a proposition and guaranteeing its certitude. Commentators often note that there is little agreement over what can count as a foundational proposition, or what the criteria for such propositions are, or how justification is transferred from basic to non-basic propositions, and these deficits of agreement are sometimes taken as indications that these criteria are not constitutive of foundationalism itself. Nevertheless, the non-inferred propositions are always taken as having a more than sheerly postulatory status: they provide some level of guarantee of truth, if not necessarily epistemic immunity (Alston 1992: 146).Footnote ¹⁷ Some accounts stress foundationalism's relation to the problem of infinite regress: the function of the foundational propositions is to halt the regress of justificatory grounds (Alston 1992: 144; Greco 1998: 34, 35 BonJour in BonJour and Sosa 2003: 9–12; Audi 1988: 408), a position which, as has already been noted, implies epistemic strength in the propositions doing the halting.

The difficulty that has already emerged from the small number of writers considered here reflects a broader problem: as one moves farther away from the classical roots of the idea in Aristotle and Descartes, the criteria for the foundations are modified and chastened to the point where it becomes difficult to determine what could not count as a basic belief. The common characteristic of accounts surveyed here is also the criterion of foundationalism enjoying the broadest acceptance: a structure in which some propositions derive their justification from others, while some are not so derived. The more rigorist defintions hold the underived or basic propositions to some standard of epistemic solidity, ranging from the narrow definitions of basicality limiting it to, for example, the self-evident or evident to the senses, to accounts which merely insist the basic propositions need not be certain or indubitable. The difficulty with these moderate versions of foundationalism is the coyness of the merely negative claims made on behalf of the foundations:Footnote ¹⁸ the category of not-certain/not-indubitable might range from the kind of reports of the senses whose reliability we take for granted in everyday life (“I am about to walk into a lamppost”) to precisely the kind of claim that Plantinga has been accused of not excluding definitively enough from the category of the properly basic (“The Great Pumpkin returns every Hallowe’en”). Until moderate foundationalists are prepared to offer criteria by which acceptably basic propositions can be distinguished from those which could not function as foundational, claims that this or that thinker is foundationalist will be merely otiose.Footnote ¹⁹ I propose therefore to add to the structural criterion of foundationalism a telic one: the purpose of the non-inferred or basic propositions is to impart to the structure as a whole a measure of certainty. The purpose of a two-tier structure whose starting propositions are not meant to secure certitude of some degree is not legitimately described as foundationalist. With this baseline in mind, we turn to the first question of the larger Summa.

The immediate relevance of this article is evident from the fact that although its subject matter is sacred doctrine, sacred doctrine is treated as a scientia, and this category of scientia is crucial to the designation of Aquinas as a foundationalist. The point at which Thomas’ account of scientiae most clearly resembles foundationalist epistemologies is its structure, for a scientia, in his view, proceeds from accepted starting principles.Footnote ²⁰ He does not explicitly say that other claims are inferred from these principles, though we may reasonably take this point to be implied. What he does not do is claim that the starting principles provide a guarantee of the rectitude of the structure or the inferred propositions. That the starting propositions in some sense underwrite inferences further along in the chain of reasoning is implied, but Aquinas makes clear that the starting principles can function simply as postulates, for he acknowledges that a scientia cannot make an argument that will satisfy one who disputes its starting principles and these starting principles are therefore not taken as self-evident, even by those content to reason from them. Metaphysicians cannot dispute with someone who denies their principles, nor can the theologian make any headway with someone who will not grant at least some of the truths of divine revelation (q. 1, a. 8).

Although the common feature of scientiae is that they proceed from starting principles and work towards others, they are grouped into two categories according to the source of these principles. One kind of scientia proceeds from “a principle known by the natural light of intelligence” (such a principle is self-evident, q. 1, a. 2 resp. and ad 1 and cannot be proved, q. 1, a. 6 ad 2); the other kind takes its starting point from some other science (q. 1, a. 2). It is important to grasp that the first kind of scientia, which proceeds from self-evident principles, is not superior to the second, in Aquinas’ view.Footnote ²¹ The subordination Aquinas claims is not of all the second kind of scientiae to the first kind, but of each one of the second kind to its own starting principles. On this reasoning, sacred doctrine is subordinate to the knowledge of God and the blessed, from which it takes its starting point, but not to arithmetic or geometry, which are the starting points of music and perspective. In fact, sacred doctrine, which numbers among the second kind of scientiae, transcends all others (q. 1, a. 5), so there is no inherent advantage in starting from self-evident principles as opposed to principles derived from another science. Aquinas deems sacred doctrine to possess greater certitude than other speculative sciences because it derives from the light of divine knowledge: the higher science from which it derives its starting principles is the science of God and the blessed (q. 1, a. 5). This last science is the surest of all, yet its claims are still not demonstrable and the certitude Aquinas claims of it is of a highly qualified kind, as we shall see.

At this point, it is important to notice that the word scientia is being used in two quite distinct senses in this question. In the first sense, as when Aquinas distinguishes between two kinds of scientiae on the basis of their differing starting points, the nearest English equivalent would be ‘discipline’ or ‘subject’. In the second sense, as when he speaks of the science of God and the blessed, the word signifies ‘knowledge’.Footnote ²² It is only scientia in the first sense which resembles foundationalism, and the similarity is sheerly structural. Nowhere in q. 1 does Aquinas make any claim about the structure of knowledge generally, or about the justification of epistemic claims. He is concerned with certitude only in a highly qualified sense, and not with warrants for justified belief and in these respects, differs markedly from contemporary philosophical epistemologies of almost every stripe.

The first question of the larger Summa is often taken as identifying the starting principles of sacred doctrine with scripture, and if we are to grasp Aquinas’ notion of scientia, it is vital to understand the precise role of scripture in sacred doctrine. The very first article of I.1 establishes the need for sacred doctrine to employ scripture: because we are directed to an end which surpasses the grasp of our own reason, we need the help of revelation; philosophy cannot suffice for this purpose (q. 1, a. 1). Far from justifying claims at the bar of reason, then, the structure of the highest scientia serves to enable us to transcend the limitations of human reason. One could of course dismiss these features of sacred doctrine as irrelevant to understanding Aquinas’ notion of scientia, since only one of the scientiae bases itself on revelation, but for the fact that Aquinas sees sacred doctrine as the most paradigmatic of all scientiae. Moreover sacred doctrine is superior to other scientiae in that it stands over them in an evaluative capacity. Its task is not to prove the principles of other scientiae, but to judge them, and where other scientiae differ from sacred doctrine, they are deemed false (q. 1, a. 6 ad 2). One cannot, therefore, bracket the discussion of sacred doctrine in attempting to understand scientia and once one acknowledges sacred doctrine's significance for the conception of scientia, one is forced to see that Aquinas’ notion of scientia is rather remote from the concerns of foundationalism and and the justification of religious claims at the bar of reason or the justification of epistemic claims at all.

Nevertheless, what Aquinas does not do is what he is often taken as doing, namely ultimately grounding sacred doctrine in scripture itself. The higher science on which this scientia is based is, once again, the scientia, or knowledge, of God and the blessed. It is this scientia which is surer than all others, not scripture per se. The role of scripture is not to provide a certain warrant for sacred doctrine, but to mediate that which is truly certain, the divine knowledge to which we have access via scripture.Footnote ²³ If philosophical accounts of the Thomistic version of scientia have often erred in bypassing the treatment of sacred doctrine, theological accounts have often fallen short of the strictly accurate by making Aquinas seem more Biblicist than he actually is, by making it seem, in other words, that scripture itself somehow underwrites the truth of sacred doctrine, in his view. Scripture actually lies at two removes from sacred doctrine: first, doctrine's sure foundation is divine knowledge, but second, its principles are the articles of faith (q. 1, a. 7). These ultimately derive from scripture (II-II. q. 1, a. 9 obj. 1 and ad 1), but they cannot be equated straightforwardly with scripture itself. Although the knowledge of God and the blessed is true beyond all doubt, we do not in this life have direct access to these, and the process of reasoning from what we do have access to–scripture–is by no means straightforward.Footnote ²⁴

No more can sacred doctrine be equated with, or taken as straightforwardly ‘enclosing’, the divine knowledge which provides its starting point and serves to guarantee its rectitude. Sacred doctrine bears the stamp [quaedam impressio] of divine knowledge, but the latter is necessarily simple, being the divine nature ipse (q. 1, a. 3 ad 2). This apparently unremarkable observation bears significant import for our question, for Aquinas essentially claims that the structure of this scientia (which is, remember, the queen of all scientiae) is built upon knowledge that is fundamentally unlike that of its corresponding scientia itself. Because the human mind cannot re-present the simplicity of divine knowledge, it necessarily carves this knowledge up into discrete bits, among which it perceives and states connections. This mosaic representation is the only kind of which we are capable, given our embodied and composite nature, yet it misrepresents the simple divine nature, and this, God, is its chief subject (q. 1, a. 7 s.c.). The starting point of the scientia that is sacred doctrine is therefore neither scripture nor divine knowledge, but principles that echo or reflect divine knowledge, which we know via scripture and derive from scripture and which we incorporate into the scientia by means of propositions whose very multiplicity necessarily distinguishes them from their source.

The starting principles of this scientia, the articles of faith which summarise scripture and divine self-knowledge, function as postulates in relation to the propositions that follow from them, for sacred doctrine does not argue in proof of its principles, but goes on from them to demonstrate something else (q. 1, a. 8 resp.). In this respect, sacred doctrine seems no different from any other scientia of the second kind, as is clear from the other examples Aquinas gives of the latter: perspective begins from the principles of geometry, music from those of arithmetic (q. 1, a. 2). It is not the job of the draftsman to prove Euclid any more than it is the harpist's job to prove arithmetical principles. The fact that another science may be able to provide such proofs is irrelevant: the salient point is that a scientia can be based on propositions functioning for its own purposes sheerly as postulates. These postulates cannot count as foundational in the weak sense that they are immediately justified, since they are derived from another body of knowledge, nor in the strong sense of providing justification, inasmuch as Aquinas acknowledges that they could be rejected outright.

Scientiae of the second kind could be interpreted as assuming a form of externalism: although the practioners of these scientiae cannot demonstrate the principles from which they depart, someone else can, the access of such persons to the starting beliefs being what justifies them epistemically. While one could not rule out such an interpretation, it does not sit easily in the framework of Aquinas’ other assumptions. First, while it might work for other scientiae of the second type, it cannot work for the paradigmatic scientia of sacred doctrine, because no living human being can properly speaking demonstrate its starting principles. Its truth is guaranteed solely by God's own knowledge in this life; we trust that scripture and the articles of faith accurately mediate this knowledge. Second, although Aquinas could easily have claimed a degree of certitude for the second kind of scientiae on the basis of the certitude of their starting principles, he does not do so. Even though Aquinas maintains sacred doctrine's basis in God's own self-knowledge is surer than any other principle could be, Aquinas never claims that this certitude is transmitted to the rest of the scientia. He simply does not make claims regarding the transmission of epistemic justification. What his account emphasises is simply the dependence of the structure of the whole on the starting principles and he shows no interest in justifying the latter for the benefit of those disinclined to accept them. When, therefore, Aquinas claims we must accept what is revealed by God on faith (q. 1, a. 1 ad 1), it is not clear that he thereby makes sacred doctrine sui generis, a special case of scientia, one in which one abandons normal reasoning in favour of tow-headed deference to authority. There is an at least strong analogy to this acceptance on faith in the draftsman's and musician's departures from mathematics of principles which they need not, and quite possibly cannot, demonstrate. The purpose of the structure appears to be to get somewhere one could not get if one had to argue for the starting principles, not to export justification from these to other parts of the structure.

At this point, the proponent of foundationalism might object–as do some its defenders–that there is no need for every individual to grasp why some foundational principles may be taken as not requiring justification; for example, someone might not be able to understand a self-evident proposition adequately (Audi 2003: 213). If so, the foundationalist essentially proposes, not just a two-tier structure of knowledge, but a two-tier structure of knowers: those who understand their beliefs are justified and those who accept their beliefs are validly grounded because someone else can justify them. Although such a picture may bear some formal resemblance to Aquinas’ notion of sacra doctrina's resting on the knowledge of God and the blessed, it is again the purpose of positing the structure which distinguishes the two accounts. For the foundationalist (at least the modified kind just mentioned) the point is to assert the certitude of propositions on the basis of some knowers’ grasp of their ground. Aquinas does not invest the propositions which make up sacred doctrine with certitude; indeed, he has relatively little to say of the latter in q. 1 and to get a sense of its significance for him, we must turn elsewhere in the Summa, to the treatment of faith in the Secunda Secundae. For Aquinas, certitude is a quality linked not to propositions, but to persons, to the devotion which is found in those believers who have greater faith than others (II-II. q. 6, a. 4). We are freed from doubt and uncertainty in divine matters by way of faith (II-II. q. 2, a. 4), but even those matters brought forth in support of the authority of faith fall short of vision, and do not cease to be unseen (II-II. q. 2, a. 10 ad 2). On Aquinas’ account, in this life there is not so much a hierarchy of knowers as a community of trusters.

These trusters, moreover, live doubly by faith, since their ability to reason rightly from the starting principles is frankly acknowledged to be shaky; on this point he insists in both the treatment of scientia in I. q. 1 and of faith in II-II. qq. 1–7. Sacred doctrine's need for divine revelation, in Aquinas’ view, derives not only from its supernatural end, but from the perennial yet very human propensity to error: had knowledge of the truth been left to human reason, the truth would be known only to the few who had the native ability and inclination to pursue these truths and then only after a long time, and even then, only with the possibility of an admixture of error (q. 1, a. 1; cf. II-II. q. 1, a. 9 ad 1). Aquinas does not ascribe error to the human formulations which are the articles of faith, or to the divine revelation which is the authoritative source of these articles, but the very process of theological reasoning is theoretically open to error; at the very least, Aquinas is not willing to deem either the church's tradition or philosophy as possessing any kind of certitude (q. 1, a. 8 ad 2). If what in Aquinas’ thought most nearly corresponds to foundationalism is the two-tier structure of scientiae, Aquinas builds into the latter an acknowledgement of potential fragility that is markedly at odds with the very purpose of asserting foundationalism.

This awareness of fragility manifests itself elsewhere in the ST's first question, also, for this is the note on which Aquinas concludes his discussion of sacred doctrine and scientiae. If the philosophical commentators tend to ignore the larger Summa's discussion of sacred doctrine altogether, the theologians who are at pains to stress its significance pay scant attention to the discussion of metaphor in the last two articles of q. 1.Footnote ²⁵ Perhaps one reason for the lack of attention these have received is the apparent abruptness of the change in foci. From talk about reason, arguments, and disciplines in the first eight articles, we suddenly in q. 1, aa. 9 and 10 find a discussion of figurative language in scripture. At one level, this discussion could be taken as no more than the usual medieval exploration of what is ‘fitting’ (conveniens), a harmless, if to the modern mind irritating, game theologians in the Middle Ages insisted on playing. The principles that lead to the conclusion of fittingness in this instance nevertheless carry over from earlier parts of the question.

According to Aquinas, scriptural use of metaphor suits our embodied nature, which more readily grasps intellectual propositions when presented in the garb of sensible objects, since all our knowledge originates from sense (q. 1, a. 9). Certainly, this presentation helps the simple, but it is not merely a concession to them, so much as a reflection of the structure of all human knowing, based as it is on the inescapable fact of human embodiment. Though Aquinas does not point to the need for, or use of, metaphor in other disciplines, the implications of what he says about the link between human embodiment and our knowledge and ability to grasp propositions potentially applies to any field of inquiry. The resting of metaphor's appeal to the embodied means that scripture's use of metaphor does not make it a special case, something remote from the usual ways of pursuing or rendering knowledge.Footnote ²⁶ This principle implies another, which Aquinas quickly makes explicit: metaphor is not mere rhetorical decoration.Footnote ²⁷ Metaphor is more integral to sacred doctrine than it is to poetry, being both theologically useful and necessary (q. 1, a. 9 ad 1).Footnote ²⁸ The shift from the focus in the body of the article to this last claim about sacred doctrine in the reply to the first objection signals another continuity between q. 1, a. 9 and the preceding articles: because of sacred doctrine's dependence on scripture as the medium of divine self-disclosure, as scripture expresses itself, so must its dependent discipline. The scriptural texts on which sacred doctrine draws are riddled with metaphor and so is sacred doctrine itself. Aquinas seems unconcerned about acknowledging that the obliqueness and ambiguity inherent in figurative language are native to the paradigmatic scientia, an insouciance that suggests constructing a structure solid in its foundations and transparent in its justificatory grounds was not his concern.

Aquinas regards the metaphorical conveyance of this divine revelation as both veiling and unveiling truth. Although he insists that what is taught in one part of scripture metaphorically must also be taught elsewhere in more straightforward fashion if it is necessary to faith, the very purpose of the veiled quality of metaphor is to beckon the reader onward: the minds of those to whom revelation has been made do not rest in these metaphors, but are raised towards knowledge of truth (q. 1, a. 9 ad 2). Obscurity, to put it bluntly, encourages reflection, prods the reader to figure out the mystery: “the very hiding of the truth in figures is useful for the exercise of thoughtful minds” (q. 1, a. 9 ad 2).Footnote ²⁹

The reliance of scripture, and therefore sacred doctrine, on metaphor nevertheless poses a significant problem, one which Aquinas addresses directly in the last article of ST q. 1, but which he had already signalled in q. 1, a. 9: its allusive and therefore elusive quality. Specifically, in the tenth article Aquinas is concerned with the forms or levels of Biblical interpretation which had by the Middle Ages become more or less standard (the allegorical, moral, anagogical and literal). One can interpret the same passage in these differing ways because the things signified by the words of scripture can themselves function as signifiers (q. 1, a. 10). Aquinas is quick to attach some important qualifications to this claim. First, he maintains that the sense of any passage is not multiplied because particular words in it can signify several things, and so the multiplicity of senses does not produce equivocation because all senses are founded on the literal. Anything necessary to faith will always be stated literally somewhere in scripture, even if it is elsewhere stated figuratively or spiritually (q. 1, a. 10 ad 1).

The reader could be forgiven for viewing this last claim with scepticism, given Aquinas’ use of scripture in the many pages of the Summa that follow: even by the standards of medieval Biblical interpretation, many of his appeals to scriptural warrant seem rather less than literal. Furthermore, he seems by this assertion to have contradicted what he said earlier regarding the necessity of metaphor: if everything important is stated literally somewhere, in what sense could metaphor be necessary? The possibility of not reading him as merely self-contradictory rests, I think, on the limited nature of the claim in q. 1, a. 10 ad 1: he does not say every truth in scripture is stated somewhere literally, but only those necessary to faith. The category ‘necessary to faith’ by no means covers all theology, and even if we are to view sacra doctrina as a very limited form of theology, it still seems unlikely Aquinas thought everything he expounded in the Summa, for example, was necessary to faith. If so, we must assume that at very many points, this scientia must find its way through the tangle of metaphors in its narrative source, and express the truths it finds there in the medium of yet more metaphors, often without the guidance of plainly-expressed propositions.

Where does this leave us? where Aquinas signalled we would be in the very first article of q. 1: feeling our way along with the rather imperfect crutch of human reason. If Aquinas acknowledges that after all the intellect's labours we are still apt to make mistakes, and if even the communal wisdom of the church's past is less than indubitable, and our surest guide to the truths most necessary to our happiness is a text rich in the productive ambiguity of poetic speech, it would appear he has seriously reckoned with the possibility of our going astray, or at the very least, of our articulating truths in a rather unsatisfactory fashion. Had he claimed sacred doctrine is grounded directly in divine self-knowledge, we would have to wonder at such a conclusion, but he did not. God's own knowledge is the surest thing there is, but its certitude does not function as the guarantor of the truth of the scientia that is built upon it, because this scientia represents a process of reasoning which is itself prone to miscarriage, via a text suited to the limitations of the human mind, with all its tendency to picture metaphysical truths in material images which in some measure betray the notions they attempt to convey.Footnote ³⁰ Far from serving as a means whereby the certitude of starting propositions can transmit justification to propositions deduced from them, Aquinas’ theoretical structure introduces the possibility of wobbling at almost every link. The resultant picture is not of an impenetrable fortress, but a delicate structure whose architect shows a vivid awareness of its fragility.

Why then propose the structure at all, if it is unlikely to lead to indubitable truth? First, the awareness of fragility should not be taken as an admission of the hopelessness of the enterprise, or of the general unreliability of the propositions within it. The fact that there are many opportunities for error does not mean the whole structure is doomed–most buildings have their weaknesses, after all, but most manage to stand and function largely as they should. The point is that the purpose of the structure is not to supply certitude or guarantees. Why then engage in sacred doctrine–or any scientia for that matter–in the first place? Because scientia, like the interpretation of metaphorical speech, provides exercise for thoughtful minds. The knowledge it yields befits us in this life (q. 1, a. 9 ad 3), and as such, is of a different character from that of the blessed, whose knowledge ranks alongside God's.Footnote ³¹ The fragility of knowledge in this life is not a permanent state, but an adequate one, one meet for the moment. We might conjecture that it is the very possibility of questioning which sacred doctrine allows that keeps human minds in this life attentive: faced with the beatific vision, no one's attention would waver, but given that our glimpse of the divine in this life is less compelling, the aporiae and debatability of sacred doctrine–indeed of scripture itself–keep the wandering mind wondering and in that sense, more fixed on the things of God than it might otherwise be. The model of scientia outlined in the Summa theologiae thus has both fragility and provisionality built into it. Aquinas does not simply claim that human reasoning can err in making inferences from the relevant data, but that the data themselves present ambiguities, making them difficult to interpret. These difficulties have a positive purpose, however, inasmuch as they provoke reflection. It seems that in this life, it is given to us to mull and to wonder, rather than to relish certainties.

The immediate objection to this summary of the I. q. 1 would be that this description fits sacred doctrine, but not any kind of scientia whatsoever, and that sacred doctrine is a special form of scientia which cannot be taken as paradigmatic of other scientiae. The problem with this line of argument is that Aquinas nowhere singles out sacred doctrine as an exception to the general characteristics of scientiae. Granted, other disciplines need not argue from scripture and can therefore avoid the particular difficulties encountered in Biblical interpretation. Against this acknowledgement, however, we must weigh two considerations: first, Aquinas holds that the ultimate basis for sacred doctrine yields greater certainty than others and that sacred doctrine is superior to and indeed judges all other scientiae, and second, that because metaphor reflects the roots of our knowing in human embodiment, the difficulties posed by the figurative nature of human language would seem to affect far more than this one discipline. Aquinas’ willingness to deem sacred doctrine the queen of all scientiae even as he painstakingly delineates its manifold fragilities indicates that whatever the similarity between the structure of scientiae that he proposes and the structure attributed to foundationalism, the purpose of each is quite different. Aquinas’ scientiae would exhibit a logical structure, but their logic does not serve to transmit justification from premises to inferences. The very goal of foundationalist structures is what Aquinas neither espouses nor what his scientiae attain. Indeed, he sees not only the inevitability of human error, but a postive value in incertitude, when it beckons beyond the conclusions we can reach through reason. The case for Aquinas’ foundationalism rests solely on the similiarity between the structure of scientia which he posits and the structure of justified belief or knowledge in the foundationalist's account, and structure by itself does not identify an epistemology as foundationalist. The purpose of the two structures is entirely different: for the foundationalist, the basic propositions underwrite the epistemic legitimacy of the whole and some degree of certitude must therefore be accorded to them. For Aquinas, the very allusiveness of the knowledge we have in this life beckons us onward to the only certain knowledge there is, knowledge which we will have only in the next life. If for Quine cognitive discourse is an open space in the tropical jungle of ordinary language, created by clearing the tropes away (1981: 189), then for Aquinas the jungle is cognitive discourse's native habitat, and territory he himself is more than content to inhabit.