2. Fallibilist readings of Newman

I begin by briefly canvassing recent arguments that Newman’s mature epistemology is committed to weak logical fallibilism and weak epistemic certainty fallibilism. Much of Newman’s work – especially in University Sermons 10–13 and Part II of the Grammar – is dedicated to showing that one can rationally arrive at certain belief or knowledge in a proposition via inference from evidence which does not entail its truth. Dahm quotes a famous passage from the Apologia, in which Newman summarises his argument in the University Sermons and his mature epistemology more generally:

I will return to this quotation later, but for now notice that in this passage, Newman clearly endorses the claim that by considering probabilistic evidence for a proposition which does not ‘reach to a logical certainty’ (i.e., entail its truth), one can arrive at certitude (or knowledge) that the proposition is true. Newman makes similar claims at several points in the latter half of the Grammar, when explaining the process of ‘informal’ inference (i.e., probabilistic reasoning), that is the precursor to certitude. In one passage, he claims that,

Newman elsewhere admits that his contention that informal inference can lead to ‘unconditional acceptance’ (i.e., certitude) may appear to be a ‘paradox’.Footnote ¹³ But Gage and Aquino argue that unless Newman accepts weak logical fallibilism, there is no paradox.Footnote ¹⁴ If one’s evidence for p entails p’s truth, there is no logical ‘gap’ between the strength of one’s evidence and one’s unconditional assent. Yet Newman elsewhere explicitly indicates the presence of just such a ‘gap’:

Clearly, then, Newman accepted weak logical fallibilism. But did he also accept weak epistemic certainty fallibilism? Dahm, Aquino, and Gage believe so, although they do not state their reasoning explicitly. However, I can detect three additional arguments in their work in favour of this view. First, Dahm appears to assume that beliefs formed with what he terms ‘Newman Certitude’ lack epistemic certainty simply because their truth is not entailed by the subject’s available evidence.Footnote ¹⁶ The idea is presumably just that one cannot rationally hold a belief with a perfect degree of credence unless it is entailed by one’s evidence. Since contemporary philosophers, including many subjective and objective Bayesians, typically accept this view, this is a natural assumption for Dahm to make.Footnote ¹⁷ Similarly, Gage points out that on Newman’s account it is crucial to Christian faith’s freedom that the truth of Christian doctrines is not generally entailed by the evidence available to Christians. From the University Sermons onwards, Newman suggests that if available evidence did (clearly) entail the truth of Christian doctrine, then belief or disbelief in Christian doctrine would be compelled by the intellect of those who perceive the evidence and, therefore, not freely embraced.Footnote ¹⁸ If, like Dahm, one assumes that one can only rationally enjoy perfect credence that p if p is entailed by one’s evidence, it follows that Newman was committed to weak epistemic certainty fallibilism. Finally, Aquino and Gage note that Newman sometimes refers to the attitude reached through informal inference as ‘moral’ certitude or certainty. But in Scholastic epistemology, with which Newman had some familiarity, moral certainty means ‘a very high but not complete degree of persuasion’.Footnote ¹⁹

In brief, I take it that all these commentators believe that Newman grants that the epistemic certainty of S’s belief that p is directly proportionate to S’s evidence that p. Since in cases where certitude is attained via probabilistic reasoning the evidence does not entail the truth of the proposition assented to, certitude need not involve belief with perfect credence.

3. Newman on epistemic certainty

Nonetheless, there are powerful reasons to hold that Newman rejected (or would have rejected) weak epistemic certainty fallibilism. Readings of Newman that embrace this view – and concomitantly hold that for Newman, certitude frequently involves merely highly confident belief – obscure the distinctive contours of his epistemology.

The first difficulty with reading Newman as espousing weak epistemic certainty fallibilism is that Newman explicitly teaches that certitudes are a species of a distinctive mental act: assent. Precisely, certitude involves assent to the proposition that one knows that some further proposition is true.Footnote ²⁰ For Newman, assent involves ‘unconditional’ acceptance of a proposition as true.Footnote ²¹ Assent is unconditional in at least two ways. Firstly, when one assents to a proposition, one does not hold that proposition on the basis of one’s other beliefs.Footnote ²² Secondly, one holds that proposition ‘absolutely’ – that is to say, one does not merely believe that it is probable to some degree.Footnote ²³ As analytic epistemologists might say, when one assents to a proposition, one’s belief in that proposition is a basic belief that has a perfect level of credence. M. Jamie Ferreira neatly summarises: ‘assent is “indivisible” whereas inference varies in strength’, and ‘assent is “unqualified” adhesion because it is independent of its premises in some sense’.Footnote ²⁴

The unconditional nature of assent makes it implausible that, for Newman, one can have certitude with respect to a proposition if that proposition lacks perfect epistemic certainty. To maintain the latter interpretation of Newman, one would have to hold that in his view, one can properly believe a proposition with greater confidence than its rational certainty. Yet we have already seen Newman’s insistence against Fairbairn and other critics that he firmly believes in the rationality of assent. Indeed, I take it that arguing for the rational legitimacy of assent is one of Newman’s chief purposes in the Grammar. Thus, he explicitly insists that he is ‘not at all insinuating that we are not rational in our certitude’.Footnote ²⁵ Moreover, since Newman identifies certitude with knowledge,Footnote ²⁶ for the same reason it also seems improper to categorise him as a weak epistemic fallibilist, who holds that someone can know a proposition even though that proposition lacks epistemic certainty for them.

The failure to appreciate the distinctive nature of assent for Newman is also unfortunate insofar as it downplays the influence of Scholastic religious epistemology on his thought. The latter is a neglected topic, although it has recently been highlighted by commentators including Michael Shea and Geertan Zuijdwegt.Footnote ²⁷ Newman’s suggestion that assent is necessarily distinct from inference (i.e., belief in a proposition because it follows from one’s other beliefs) has an important parallel in the Late Scholastic analyses of faith that Newman read in some detail after his conversion at Rome, in 1846/7, when he consulted authors, including Suarez, De Lugo, Viva, and Billuart. He explicitly quotes these authors when composing a series of Theses describing his religious epistemology, which he submitted to the leading Roman theologian Giovanni Perrone for approval and subsequently copied several times later in life.Footnote ²⁸

As Newman neatly summarises, according to Late Scholastic consensus, divine faith involves assent to a proposition on the grounds that it is divinely revealed.Footnote ²⁹ Because such assent is supernatural and perfectly certain (‘without [any] doubt or fear of the opposite’Footnote ³⁰), it is not itself grounded in the judgement that naturally perceptible evidence such as the performance of miracles or the fulfilment of prophecy by religious authorities (motives of credibility) indicates that the relevant proposition is divinely revealed.Footnote ³¹ Rather, one believes a proposition with divine faith solely on the basis that it is divinely revealed, but that the belief that it is thus revealed is not itself inferred from one’s wider beliefs or evidence.Footnote ³² Its genesis rather lies in an act of will, enabled by grace.Footnote ³³ Nonetheless, since the assent of faith must be rationally and morally prudent, it should ordinarily be preceded by a ‘judgment of credibility’ (i.e., the rational formation of a natural belief, based on probabilistic evidence, that God has revealed Himself through Christ and the Church).Footnote ³⁴

To those familiar with Newman’s mature epistemology as articulated in the Grammar, it should be clear how Newman’s account of assent and certitude in the Grammar bears close structural resemblance to this Late Scholastic analysis of faith. Just as Scholastics held that divine faith’s certainty and supernatural nature means that it cannot be grounded in a natural and imperfectly certain ‘judgment of credibility’, so Newman holds that given its nature, assent is never grounded in inferential reasoning. But equally, both Newman and the Late Scholastics hold that the ‘basic beliefs’ with which they are each concerned must ordinarily be preceded by reflection on probabilistic evidence in order to be rational. In Newman’s words, ‘though acts of assent require previous acts of inference, they require them, not as adequate causes, but as sine qua non conditions …’.Footnote ³⁵ Since the Scholastic analysis of faith explicitly requires that divine faith involves more than merely ‘moral certainty’ (qua highly confident but imperfect credence, which may plausibly involve a passing element of doubt or fear), it should be unsurprising that Newman likewise envisages certitude as involving more than moral certainty thus understood.

A second, related difficulty with claiming that Newman is a weak epistemic certainty fallibilist is his description of the process of ‘informal inference’ (i.e., probabilistic reasoning), which is the precursor to rational assent in ‘concrete’ matters. In cases where such reasoning leads to assent, Newman explicitly states in the Grammar that assents are ‘as unconditional as if they had that highest evidence’ – that is, the evidence of either ‘intuition or demonstration’.Footnote ³⁶ Similarly, in a private letter, Newman commends a reader of the Apologia who suggests that probabilistic reasoning can give rise to certainty ‘not a jot inferior’ to the certainty produced by mathematical proof.Footnote ³⁷ In the same letter, Newman compares ‘mathematical or strict demonstration’ to an iron rod and ‘moral demonstration, which is an assemblage of probabilities’ to a cable formed of many threads. Although the two are intrinsically different, Newman is clear that both are equally sufficient to produce – and even, ‘rationally demand’ – certitude.

When in the Grammar, Newman makes the claim that assents are ‘unconditional’, he immediately gives several examples of certitudes arrived at in this way, including certitude that, ‘the earth is a globe’, ‘that there are cities in different places on the earth such as Madrid and Paris’, ‘that we had parents, though we can have no memory of our birth’, and ‘that we shall die, though we can have not experience of the future’.Footnote ³⁸ He later adds further examples, including the fact that Britain is an island. Throughout his discussion, Newman takes pains to show that certitudes are (or can be) rationally formed. In other words, he believes that when informal inference gives rise to certitude, it produces a rational belief with (rationally appropriate) confidence equal to logical proof from certain premises: i.e., a perfect degree of credence. Again, since for Newman certitude and knowledge are equivalent, it follows that in his view if one knows or has (rational) certitude that p, one has a belief that p which is perfectly epistemically certain.

The second problem with interpreting Newman as a weak epistemic certainty fallibilist who holds that ‘certitude’ need only involve ‘morally certain’ belief is that it renders his disagreement with Locke insubstantial. As Newman reads him, Locke maintains that, if one loves truth for its own sake, one cannot accept ‘any propositions with greater assurance than the proofs it is built on will warrant’.Footnote ³⁹ Thus, according to this view, ‘absolute assent has no legitimate exercise, except as ratifying acts of intuition or demonstration’ and ‘reasonings in concrete matters […] are never more than probabilities’.Footnote ⁴⁰ Famously, Newman spends Chapter VI of the Grammar explicitly criticising Locke’s account of assent, and this critique continues implicitly for much of the rest of the work.

Yet if Newman merely meant to argue in the Grammar that one can be highly confident in the truth of propositions for which one has considerable evidence, it is difficult to see why he believes that his disagreement with Locke is at all significant. By Locke’s own lights, belief/assent comes in degrees, and it can be highly confident given appropriate evidence. Newman could simply have noted that, pace Locke’s insistence that belief must be proportioned to evidence, one can be highly confident (‘morally certain’, following Gage and Aquino) in beliefs which are supported by strong probabilistic evidence. It is far more plausible that Newman disagrees with Locke robustly, by holding that one can rationally have perfect credence that p (i.e., have ‘certitude’ of their truth) even if one lacks demonstrative evidence of p’s truth.

4. Newman and proper functionalism

However, all this leaves untouched the important question which makes it plausible that Newman embraced weak epistemic certainty fallibilism: how can Newman claim that one can rationally have perfect credence that p, in cases where one’s evidence does not entail p?

Gage and Aquino reject two possible suggestions. One could contest with Martin D’ArcyFootnote ⁴¹ that by Bayesian lights there are cases in which one should have a perfect credence that p given an infinite amount of evidence that p, even though the evidence is logically consistent with not-p. For instance, perhaps the conditional probability that ‘this fair coin will come up heads at least once’, given the evidence that ‘the coin will be tossed an infinite number of times’, is 1, even though it is logically possible that the coin will always show tails.Footnote ⁴² Perhaps it is in such cases that Newman believes one can arrive at certitude via informal inference. Admittedly, Newman sometimes gestures to the idea that in cases of informal inference one perceives an infinite amount of evidence even if one cannot readily articulate that evidence.Footnote ⁴³ Equally, in another passage, he hints that in informal inference the probability of a proposition’s truth given one’s evidence is infinitesimally close to 1.Footnote ⁴⁴ Yet as Gage and Aquino note, it is plausible that in many cases of informal inference, subjects lack an infinite amount of evidence.Footnote ⁴⁵ In particular, this seems to be the case in various instances of certitude that are made on the basis of informal inference, which Newman discusses in the Grammar, such as the belief that historical works of literature can be attributed to particular authors.Footnote ⁴⁶ One might add that Newman himself nowhere explicitly argues that, according to the best formal account of probabilities, infinite evidence guarantees certitude. This would imply that some mathematical account of probability is what determines the rationality of certitude; I will argue below that Newman would likely reject that claim.

Another possibility which Gage and Aquino consider is that ‘the illative sense somehow is able to turn fallible evidence into infallible belief’.Footnote ⁴⁷ In this vein, Joe Milburn holds that when exercised in appropriate circumstances, the illative sense functions infallibly to produce true beliefs, just as when our perceptual faculties function properly they likewise produce true beliefs.Footnote ⁴⁸ Against this, Gage and Aquino object that Newman explicitly holds that the human intellect is not infallible and that this is obviously the case with regard to judgements about such matters as truths of geography, history, or testimony of which Newman believes that we can rationally be certain.

In my judgement, they are hasty to rule out Milburn’s reading of the Grammar. While Newman grants that the human intellect is fallible, he unsurprisingly holds that certitude (i.e., knowledge) is factive: if one has certitude that p, p is true. Thus, Newman suggests that one who is certain need not profess infallibility in general, but merely on this particular occasion (pro hac vice).Footnote ⁴⁹ So perhaps, on Newman’s view, what makes it rational to hold a belief with perfect credence on the basis of probabilistic evidence is not merely the evidence itself together with the proper operation of the illative sense, but additionally the fact that the belief is true and produced in appropriate circumstances for knowledge. Obviously, in such conditions, one’s belief cannot fail to count as knowledge, and so plausibly it possesses perfect epistemic certainty.

If this strikes analytic readers as strange or implausible, it is worth noting that a somewhat similar position has recently been defended by Timothy Williamson. According to Williamson’s ‘knowledge first’ epistemology, one’s total evidence consists of all the propositions that one knows.Footnote ⁵⁰ It follows that every proposition that one knows has unit probability according to one’s evidence: if one knows that p, one’s total evidence entails that p. For this reason, Williamson can be classed as a logical infallibilist even though he does not believe that the processes through which humans form beliefs are infallible. But Williamson further holds that one can arrive at knowledge via ‘induction’: i.e., by reflection on evidence for a hypothesis such that the probability of the hypothesis on one’s evidence is less than one. How so? In Williamson’s view, before one comes to possess the relevant evidence, the conditional property hypothesis given the evidence is imperfect. The probability may remain the same at a further moment, when one has gained the evidence but failed to appreciate its bearing on the truth of the hypothesis. But when one comes to believe the hypothesis on the basis of the probabilistic evidence, one’s belief may count as knowledge. At this point, the belief itself becomes part of one’s knowledge, and it enjoys the highest degree of probability on one’s evidence. In this case, it is not simply one’s ‘internal’ states (e.g., the appreciation of the evidence and the function of one’s cognitive faculties) that make the belief count as knowledge, but ‘external’ states (e.g., the truth of the hypothesis and the fact that one’s belief that p could not easily have been false).Footnote ⁵¹

Williamson’s identification of evidence with knowledge and his views on evidential probability are, of course, highly controversial.Footnote ⁵² But for present purposes it is useful to note that like Newman, Williamson holds that through reflection on merely probabilistic evidence that p, one can arrive at belief that p which has perfect epistemic certainty (or at least, belief that is entailed by one’s evidence). As Newman believes that inference and assent are distinct acts, so too Williamson seems to hold that there is a fundamental epistemic distinction between merely believing that p on the basis of some evidence and knowing that p is true. This should give pause to interpretations that hold that Newman must be a weak epistemic certainty fallibilist merely because he believes that one can gain knowledge through informal inference or because he grants that the illative sense is fallible. However, I am not suggesting that this is the best way to interpret Newman historically. Williamson’s view of evidence as knowledge is idiosyncratic, and Newman nowhere outlines such a view. Williamson’s discussion of evidence would, I believe, strike Newman as far too theoretical. Additionally, Williamson has various broader commitments which clearly conflict with Newman’s views: for instance, he suggests that one can know that p even if one rationally believes with a high degree of certainty that one does not know that p.Footnote ⁵³

Rather than countenancing that Newman would defend the rationality of certitude reached by informal inference in either of the two ways examined above, I instead suggest that analytic theologians can better understand his thought by turning to a recent but somewhat neglected account of epistemic probability, outlined by Alvin Plantinga. Notably, while Plantinga is himself a fallibilist, he rejects the account of epistemic probability provided by his Bayesian contemporaries.Footnote ⁵⁴ In its place, Plantinga offers a new account of epistemic probability that is centred on the proper function of epistemic agents: how they are supposed or designed to form beliefs.Footnote ⁵⁵ Famously, for Plantinga, a belief has warrant if it is produced by an agent’s cognitive faculties functioning properly, according to a design plan successfully aimed at truth, in an appropriate environment.Footnote ⁵⁶ Drawing on this theory of warrant, Plantinga constructs a theory of conditional epistemic probability. Roughly, this theory holds that the epistemic probability of a proposition A on another proposition B is the probability that someone whose cognitive faculties were functioning properly would accord to A if she were certain of B and if she considered A’s probability solely in light of B. More precisely,

Crucially, nothing in Plantinga’s account of epistemic probability requires that for A/B to have a conditional probability of 1, B must entail A’s truth. Nor, more generally, is the probability of A/B to be analysed in logical or mathematical terms. Rather, epistemic probability is determined by the agent’s proper function; accordingly, it may be species- or even agent-relative. If we follow Plantinga, then it is perfectly possible that God (or perhaps, unguided evolution) has designed humans so that under appropriate conditions, we would form perfectly certain beliefs on the basis of non-demonstrative evidence.

Unfortunately, Newman does not himself lay out a theory of conditional epistemic probability. As such, ascribing Plantinga’s views to him is evidently anachronistic. However, in my view, Newman is likely to have welcomed Plantinga’s account, and it is plausible that he anticipated it to some degree. Newman embraces what Roderick Chisholm termed ‘particularism’: the view that we should begin reflection on which beliefs count as rational by considering paradigmatic instances of rational belief rather than by applying some preconceived methodology which specifies which beliefs are rational.Footnote ⁵⁸ More precisely, when Newman considers what makes belief rational, he does not merely appeal to paradigmatic examples of rational belief. Rather, he appeals to what he takes to be human nature, as designed and created by God. Thus, as Newman explains at the start of the Grammar, ‘[w]e do but fulfil our nature in doubting, inferring, and assenting; and our duty is, not to abstain from the exercise of any function of our nature, but to do what is in itself right rightly.’ Similarly, later in the work he maintains that,

By contrast, he sharply criticises Locke and others for not ‘going by the testimony of psychological facts, and thereby determining our constitutive faculties and our proper condition, and being content with the mind as God has made it’.Footnote ⁶⁰

As the reference to our ‘proper’ condition in this last quotation intimates, Newman does not naively assume that the way in which humans typically reason is always rational. This is fortunate, for otherwise he would have to allow that common mistakes in reasoning (e.g., base-rate fallacies) are actually legitimate forms of inference. Like Plantinga, Newman holds that human cognitive faculties, including the ‘illative sense’ (i.e., the cognitive faculty that allows the mind to reach certainty through informal inference), can fail to operate as they should. Like other mechanisms, the illative sense can fail to operate properly, although this fallibility is no proof that it cannot achieve its proper function of producing perfectly certain beliefs:

Newman does not specify precisely what ‘proper function’ involves: he does not explicitly mention a ‘design plan’ as Plantinga does. However, it is likely that Newman assumes that the proper function of our cognitive faculties follows the way in which they are created and intended to operate by God. As we have already seen in the Apologia, Newman maintains that probabilistic reasoning can yield certitude, ‘according to the constitution of the human mind and the will of its maker’,Footnote ⁶² and in the Grammar, he criticises Locke for not resting ‘content with the mind as God has made it’.Footnote ⁶³ Equally, in his earlier works Newman countenances the possibility that humans should rationally believe in God or revelation, without any evidence, if that were how God intended us to come to faith.Footnote ⁶⁴

Newman’s underlying epistemology, then, does not assess the rationality of beliefs in terms of their conformity to some set of logical rules or mathematical calculus. As commentators such as Ferreira and Aquino himself have noted, this position is not entirely novel: it is rather a continuation of the ‘British Naturalist’ tradition of thought that Ferreira traces through John Wilkins, David Hume, and Thomas Reid.Footnote ⁶⁵ As Ferreira writes, these authors share a commitment to the ‘sanction of the natural’ (i.e., ‘what is evident in what we in fact as humans say and do’) as a response to scepticism.Footnote ⁶⁶ Arguably, Plantinga himself also belongs to this tradition: he frequently quotes Reid, including in his discussion of epistemic probability in Warrant and Proper Function. There, he approvingly mentions Reid’s remark that, ‘in most cases, we measure the degrees of evidence by the effect they have upon a sound understanding, when comprehended clearly and without prejudice’,Footnote ⁶⁷ while noting that Reid unfortunately failed to develop this insight into an adequate account of epistemic certainty. We can perhaps therefore read Plantinga as offering a development of the account of epistemic certainty that was already latent in the epistemological tradition to which Newman belonged.

In light of all this, I suggest that just as Newman suggests that the rationality of beliefs is to be determined by human nature (i.e., the proper function of our cognitive apparatus), if pressed he would affirm with Plantinga that epistemic probability is to be determined by what our nature disposes us to believe when it is functioning as God intends. Of course, neither Newman nor Plantinga deny that the laws of logic or the probability calculus can be important for human rationality. Thus, Plantinga affirms that, ‘[T]here are areas in which conformity to the probability calculus is a sort of ideal for us. This is once more a reflection of the enormously elaborate and highly articulate nature of our noetic design plan’. Nonetheless, it is consonant with Plantinga’s account of epistemic probability that one can believe a proposition with certainty. Since, as I have argued, Newman believes that certitude involves belief with perfect credence, we can draw on Plantinga’s account of epistemic probability to show how Newman can reject weak epistemic certainty fallibilism whilst also affirming the possibility of arriving at certitude via probabilistic reasoning.