Aquinas on testimonial justification: faith and opinion.

Siebert, Matthew Kent

A FULL HUMAN LIFE IS IMPOSSIBLE without believing the testimony of others. Without such testimony we would be ignorant of almost all truths about history, geography, and the thoughts and feelings of others. Important practical decisions about what school to attend, what vehicle to buy, whom to marry, and so on, also significantly depend on testimony. Such dependence on testimony was largely glossed over by ancient philosophers, but has been emphasized by C. A. J. Coady, Elizabeth Fricker, Jennifer Lackey, and many other recent philosophers, who often trace the historical roots of their discussion back to an eighteenth-century debate between the Scottish philosophers David Hume and Thomas Reid. Hume took what I will call an "inferentialist" view of belief on testimony. "No kind of reasoning," he said,

   is more common or more useful--even necessary--to human life than
   the kind derived from the testimony of men.... [O]ur confidence in
   any argument of this kind is derived wholly from our observation of
   the truthfulness of human testimony and of how facts usually
   conform to the reports witnesses give of them. (1)

Reid, by contrast, took what I will call a "defaultist" view:

   It is evident that, in the matter of testimony, the balance of
   human judgment is by nature inclined to the side of belief; and
   turns to that side of itself when there is nothing put into the
   opposite scale. If it was not so, no proposition that is uttered in
   discourse would be believed, until it was examined and tried by
   reason; and most men would be unable to find reasons for believing
   the thousandth part of what is told them. Such distrust and
   incredulity would ... place us in a worse condition than that of
   savages. (2)

Sadly, accounts of testimonial justification from before Hume and Reid remain largely unexamined. (3) As C. A. J. Coady notes in his pioneering work Testimony: A Philosophical Study, medieval Christian philosophers were especially disposed to see the value of testimonial belief, for which Aquinas had an "interesting and subtle theory." (4) In this article, I provide the first detailed interpretation and reconstruction of Aquinas's account of testimonial justification. (5) His unique pluralist approach, I argue, does not fit nicely into either the Humean or the Reidian camp. Elizabeth Anscombe and others have recently pointed to another way of believing a speaker--trusting the speaker for the truth--that is not reducible to Humean inference from observations, nor to Reidian default acceptance. (6) I argue that Aquinas has a similar but distinctive interpersonal view of some testimonial belief, as well as a separate account of some testimonial justification that is purely inferentialist.

I begin by highlighting several ways in which everyday testimonial "faith" was important to Aquinas (section I), and by clarifying what he means by "faith" (section II). Then I describe more precisely the defaultist, inferentialist, and interpersonal accounts of testimonial justification (section III). I explain that Aquinas does not take a defaultist approach, but rather, with regard to at least some cases of testimony, he takes an inferentialist approach (section IV). With regard to other cases of testimony he takes an interpersonal approach (section V) that has certain advantages over recent interpersonal approaches (section VI). I conclude with some comments on the advantages of Aquinas's pluralist view (section VII).

I

The Christian tradition with which Aquinas was familiar often dealt with testimony under the heading of "faith" (fides). Apologists in that tradition often drew attention to instances of valuable testimonial belief in everyday life, to support the idea that by analogy we ought to accept religious testimonial belief. For example, in the second century Theophilus of Antioch said:

   Do you not know that faith [pistis] leads the way in all actions?
   ... What sick man can be cured unless he first entrusts himself to
   the physician? What art or science can anyone learn unless he first
   delivers and entrusts himself to the teacher? (7)

In the same vein, Aquinas, in the course of arguing for religious faith, explains three ways in which testimonial belief contributes to human society.

First, following Maimonides, Aquinas says that even when certain talented individuals can demonstrate some important truth for themselves (for example, that God exists), the vast majority of people still need faith with regard to that truth, since most of them will not have the time, talent, or training to perform such a demonstration. (8) And some of those who do have the time, talent, and training necessary will be lazy or make mistakes, and so would be better off having faith anyway. Today we might similarly say that most people must accept important claims of natural science, especially counterintuitive ones about, say, special relativity or quantum indeterminacy, on "faith." (9) Call this a vertical epistemic division of labor, extending from experts down to those who trust them.

Following Aristotle, Aquinas found a vertical division of labor among the experts themselves, since often one science depends on another. Aquinas says that sometimes a "lower" (or "subalternated") science takes on "faith" the conclusions demonstrated by a "higher" science. (10) For instance, some theorems demonstrated in geometry are the starting axioms of optics (a "lower" science), and the optician who does not know the geometrical demonstrations for them must take them on faith from the geometers. We see something similar today in the way biology depends on chemistry, and chemistry on physics.

Second, Aquinas, like Augustine, argues that human society could not function without a horizontal epistemic division of labor.

   For example, singulars and contingents that are removed from our
   senses, such as the acts, sayings, and thoughts of people, are such
   that they can be known to one human and unknown to another. And
   since in human community it ought to be that one person use another
   as himself when he is not self-sufficient, so it ought to be that
   he stand toward those things which another knows, but are unknown
   to him, the way he stands to those he knows himself. Thus faith, by
   which one man believes the sayings of another, is necessary in
   human interaction. It also is the "foundation of justice" as Cicero
   says in De Officiis [1.23], which is why no lie is without sin,
   since every lie detracts from this faith that is so necessary. (11)

Here Aquinas makes the general point that a well-functioning society often requires us to act on what other people know, when not in a position to verify something for ourselves.

Third, later in the same passage, Aquinas argues that faith is necessary for learning the sciences. (12)

   But since by the power of those which we know last are known those
   which we know first, we must from the beginning have some knowledge
   of those things which are better known in themselves, which cannot
   happen except by believing another [credendo]. And we see this in
   the order of the sciences, since the science of the highest causes,
   namely metaphysics, comes last in a man's knowing, and yet in the
   preamble sciences one must suppose certain things which in
   [metaphysics] are known more fully. Hence every science has
   suppositions that the learner must believe. (13)

Here Aquinas says that some of the preexisting knowledge required for learning a science must be testimonial. He appeals to the Aristotelian ordering of the sciences, on which metaphysics is the most fundamental but also the most obscure. Similar considerations apply to contemporary natural science, in which physics is the most fundamental but at the same time arguably the most difficult to understand. Someone setting out to study physics would not get very far without some direction on the basic principles of force, motion, and matter, as physicists today understand them. (14) The faith of a student, unlike that of nonexperts in a vertical epistemic division of labor, is a provisional faith, supporting one's education until one is an expert oneself, and in the ideal case one comes to understand why the principles one accepted at the beginning of one's education are true.

So here we have three areas in which Aquinas recognizes the importance of testimonial belief, under the name "faith": when it enables guidance (1) from experts, (2) from peers who happen to know something you did not know for yourself, and (3) from teachers. Our task will be to see how such faith works, and what, according to Aquinas, makes such belief rational or epistemically justified. (I do not discuss here the difficult question how, on Aquinas's terms, one can have testimonial knowledge.)

II

First we need to make clear what Aquinas means by "faith." Aquinas's terminology can be confusing, since he uses "faith" in both a very broad sense (explained in this section), and a narrow sense (to be explained in section V). Following Augustine, Aquinas says that faith in the broadest sense is just assent to the unseen, by contrast with assent to what one sees to be true. (15) Seeing in this sense has both psychological and epistemic aspects. Psychologically, when one sees that p, one's assent is automatic and involuntary. (16) Epistemically, the idea seems to be that an object seen is as directly accessible to the knower as possible, in as determinate a way as possible, so that none of the features relevant to knowing that p remains unrepresented to the knower. (17) These features of seeing make one's grasp of the object "fixed" and "determined to one," (18) and in this sense "certain." (19) The fixedness or certainty of seeing derives from the natural fit between the intellect and the evidentness of the truth, so it is not merely a psychological feature but also an epistemic feature of the knower.

For example, right now I know, by seeing, that a brown cup is on the table near me. The features of the cup and of the table which are relevant to me knowing that a brown cup is on the table (the cup's and the table's shape, size, color, and so on) are evident to me, making my belief determinate in a way it would not be if I had formed that belief merely on the basis of some natural sign of the cup's presence (for example, its shadow), or on the basis of an inductive argument that a brown cup must be on the table, or on the basis of someone else's assertion that a brown cup is on the table. Aquinas takes the distinction between the seen and the unseen to be sharp: "[A]s soon as something begins to be present or apparent, the object cannot fall under the act of faith," (20) but anything short of seeing makes room for the assent of faith. (21)

Because the object is unseen, the assent of faith is not fully automatic but requires an act of will. (22) In cases of seeing, assent is natural rather than voluntary, so the only voluntary choice an agent can make is whether or not to attend to p sufficiently to see it. (23) But as soon as p is unseen, "the intellect assents not because it is sufficiently moved by its proper object, but by a choice [electionem] inclining toward [one proposition] more than [to its contrary]." (24) So it is essential to faith that one's assent is determined partly by the will. However, this condition does not make Aquinas a crude doxastic voluntarist, since assent is not commanded by the will alone, but by the will under the direction of one's reason. (25) Faith is not just belief on a whim; it is a matter of managing one's assent for the best of the human as a whole, given that one does not see for oneself whether p.

Now let's consider what makes testimonial assent to the unseen justified. We will first consider three kinds of testimonial justification (in section III), and then see how Aquinas's testimonial "opinion" (section IV) is like one of these, and see how his testimonial "faith" in a narrow sense (section V) is like another, allowing him to avoid some problems for recent accounts of testimonial justification (section VII).

III

Three different recent approaches to the justification of testimonial belief each posit a unique kind of testimonial justification. (26)

On the defaultist view, you are entitled to believe the assertions of others by default, that is, provided there is no evidence against such belief. (27) Defaultists often draw an analogy between believing others' testimony and believing your own memory: you are entitled to believe the deliverances of memory, provided there is no sign of memory dysfunction. Testimony, on this view, is yet another knowledge-providing faculty, like memory or perception, and you are entitled to believe testimony provided there is no sign of a speaker's ignorance or deceit. One advantage of this view is that it explains why we usually do not consciously deliberate about whether to accept the testimony of others. One problem for this view is that it seems to entitle hearers to believe any zany statement, so long as one has no evidence against it. (28)

Inferentialists, on the other hand, think you get testimonial justification only by inferring that an assertion is true. (29) For example, you might infer that it is raining in Norway from your evidence that Hildegaard told you by phone that "It is raining," together with your evidence that she is sincere and competent to know whether it is raining or not. One advantage of this view is that it is simpler, in the sense that it explains testimonial knowledge in terms of other kinds of justification that contemporary epistemologists are already familiar with. One problem for this view is that it seems to leave no room for deciding to trust a speaker, something that it seems we are capable of doing.

There is an important debate in the epistemology of testimony between "reductionists" who claim that testimony does not provide a sui generis kind of justification, and "antireductionists" who claim it does. A standard way to be reductionist is to be an inferentialist, while a standard way to be antireductionist is to be a defaultist. (30)

Recently another antireductionist approach has emerged. On an interpersonal view, one has properly testimonial justification only when the interpersonal relationship between speaker and hearer provides the hearer with a noninferential reason to believe the speaker. (31) For example, when a knowledgeable speaker offers me her "assurance" that p, and I "accept" her assurance, then I am justified in believing p, without having to infer that p for myself, and without having a default entitlement to believe what any old speaker tells me. Just as promises are intentional offers to take responsibility for the hearer's well-being, assurances are intentional offers to take responsibility for the hearer's belief being true. Accepting such an assurance from a responsible speaker gives the hearer a right to believe that p without taking precautions that one would otherwise take against being mistaken about p, and a right to defer to the speaker challenges about how belief that p is justified. (32) One advantage of this view is that it explains the reactive attitude of betrayal we have when we find out that someone assured us of something false. One problem for this view is that it seems to say that when no assurance has been given, for example, to an eavesdropper or to a jury, the eavesdropper's or the jury's belief is not testimonially justified, properly speaking, because it is not based on a special interpersonal relationship.

IV

At first glance, one might think that of these three views, Aquinas takes the defaultist approach. Aquinas says we are naturally designed to depend on others' testimony, (33) and this suggests a natural entitlement to believe others by default. (34) Aquinas does not explicitly endorse such a view, but one might think it implicit in his rule that one ought to "presume the good" of others until one knows otherwise. (35) The rule does not require us to presume that others are as good as possible; rather, it requires us not to believe that others are sinning, unless it is obvious that they are. (36) So when others tell you things, your default presumption should be that they are not lying.

But it does not follow that you should presume that they are telling the truth, because they could make honest mistakes. And even if Aquinas were to say that you should presume that others are telling the truth, presumption is not the same as belief or assent. A jury is required to presume, but not to believe, that the defendant is innocent. And such a presumption aims not at the epistemic goal of producing the most accurate belief, but at the practical goal of causing the least harm to others. (37) So "presume the good" is a moral, not an epistemic, rule. One should suspend presuming the good of others to avoid endangering others, but not to avoid cognitive error. (38) Because the rule is not about epistemic justification it does not support a default entitlement view of testimonial justification.

The evidence suggests, instead, that Aquinas thinks that at least some of the time our testimonial beliefs are justified not by default but by inductive inference. He says that "to believe a human without probable reason is to believe too quickly," (39) and that in important matters (like faith and morals), "assent ought not to be given easily." (40) Aquinas recognizes a specifically inferential way of arriving at testimonial beliefs, which he calls "opinion." One way to have faith (that is, to assent to the unseen), he says, is when one is "led [to assent] by human reason, and so strong opinion is called 'faith'." (41) (Here strong opinion contrasts with other cases of opinion in which one does not give full assent to a proposition, and so does not believe it. (42))

What does Aquinas mean by "opinion"? Opinion is the result of inductive or "probable" inference based on "signs," "verisimilitudes," or "probable" (that is, nondemonstrative) syllogisms. (43) Aquinas contrasts opinion with suspicion and doubt. These are cognitive propositional habits (or as we say today, propositional attitudes) that have lower credence than opinion: suspicion is a slight inclination to assent to p, and doubt is an equal inclination to assent to p and to assent to not-p. (44) Further, opinion that p requires an awareness of (or at least a sensitivity to) the fact that not-p is possibly true. (45) So opinion that p is roughly equivalent to taking p to be probable by inductive inference.

A testimonial case of opinion comes up while Aquinas is contrasting opinion with the comprehension of science:

   If someone knows by demonstration that a triangle has three angles
   equal to two right angles, he comprehends it, but if someone else
   accepts his opinion in a probable way [opinionem accipiat
   probabiliter], because it is said by the wise or the many, he does
   not comprehend it, because he does not attain the perfect way of
   knowing it as far as it is knowable. (46)

Here we have testimonial assent based on a dialectical syllogism such as:

   If most people say that p, then probably p.
   Most people say that p
   Therefore p.

Aquinas recommends such reasoning in the case of court testimony. Other things being equal, we should give more weight to p when more witnesses say p than not, because "it is probable that the saying of many contains the truth more than the saying of one." (47)

We can now see that Aquinas has all the resources for an inferentialist account of testimonial belief. He recognizes inferentially justified assent in the form of opinion; he recognizes that some inferentially justified assent (namely, strong opinion) is outright belief; and he recognizes that some opinion is testimonial.

V

However, Aquinas also recognizes cases of testimonial belief justified in an interpersonal way. The most striking case, and the one he elaborates on most fully, is a Christian's faith in God. But Aquinas also describes analogous cases of faith on the testimony of fellow humans, where the justification of one's beliefs is similarly interpersonal. (48) In this section, we will consider the formal object of faith, the special act of will in faith, and a typical reason for having faith. In section VI, we will contrast such faith with demon faith and with the interpersonal forms of justification posited by some recent philosophers.

Faith in the broadest sense (assent to the unseen) includes strong opinion. But from now on, I will use the term "faith" in the narrower interpersonal sense, as Aquinas usually does, to contrast it with opinion. (49) Three features of Christian faith in God, and of the analogous case of faith in one's fellow humans, distinguish faith from opinion. When one has faith that p, one (i) believes the speaker's statement in order to adhere to the speaker, (ii) with a special act of will not present in opinion, and (iii) typically for the reason that the speaker is truthful. Let's look at these three features in turn.

First, faith responds to the person of the speaker as a reason to believe. This feature is the most obscure of the three, and explaining it will take up most of our discussion. It comes up first where Aquinas says that the act of Christian faith is an act of "believing God" (credere Deo). Augustine had a way of making catchy philosophical distinctions in passing, and Peter Lombard then offered many of these up for scholastic interpretation in his Sentences. One of them is a distinction between three different objects of faith's act of belief: credere Deum ("believing that God [exists]"), credere in Deum ("believing in God"), and credere Deo ("believing God"). (50) Aquinas assigns each of these acts to a different part or combination of parts of the soul. Credere Deum is the intellect's act of being determined to the one proposition believed. Credere in Deum is the will's act of believing out of love of God, where the will's object is God himself. Credere Deo ("believing God") is reason's act of inclining the will to assent, where the object is the proposition believed, but only as spoken by that speaker:

   Inasmuch as reason inclines the will to the act of faith, it is
   credere Deo. For the reason the will is inclined to assent to
   things unseen is because God said them; just as a man, in matters
   he does not see, believes the testimony of some good man who sees
   the things he does not. (51)

Aquinas here identifies a reason for belief ("because God said them") that he considers not reducible to an inductive reason for belief of the kind that supports testimonial opinion. Assent is not Christian faith when the proposition is believed

   by some human reasons and natural signs ... but only when one
   believes for this reason [ratione], that it is said by God, which
   is designated by calling it credere Deo. And this specifies faith,
   the way any cognoscitive habit has its species from the reason
   [ratione] by which it assents to anything. (52)

Here "reason" (ratione) need not signal that the believer is drawing an inference. Not all cognitive habits of propositional assent (or propositional attitudes) result from a reasoning process; some, such as prophecy (53) or intellection of first principles, (54) are not mediated by reasoning. Nevertheless, every cognitive habit has a ratio for its assent, in the sense that it has a distinctive character that distinguishes it from other cognitive habits. Here, then, Aquinas is saying that one has Christian faith only when one believes something in light of this, that it is said by God. Aquinas calls the distinctive character of a habit its "formal object," and contrasts the character of faith with the character of inferentially based habits like opinion and science.

In Aquinas's view, every power, habit, and act has both material objects--the items it is directed at--and a formal object--the formal aspect of those items that allows that power (or habit or act) to be directed at them. (55) For example, my will is right now directed at a ripe mango as material object, but according to Aquinas my will can be directed at that mango only insofar as I apprehend the mango as good. (56) Moreover, that same mango can be a material object for various other powers, under other formal aspects, by being visible, movable, understandable, and so on. It can also be a material object for different kinds of willing, more specific than willing in general, as distinguished by different formal aspects under which the mango is good (for example, good as tasty, as opposed to good as nourishing).

The intellect is a single cognitive power aimed at the truth, but different ways of grasping the truth make for different formal objects and so make for different habits. Just as the same mango can be the material object of different kinds of willing, so the same proposition can be the material object of different cognitive habits (or propositional attitudes). And while these different ways of assenting provide different explanations or reasons for one's belief, not all of them are cases of inferential reasoning. For example, Aquinas would distinguish between the way my geometer friend James believes the theorems of geometry as self-evidently following from self-evidently seen first principles, and the way I believe those theorems when I take them on faith from James. James starts from axioms which he simply sees to be true, without any inferential reasoning, (57) and then on that basis reasons his way to the theorems. But when I take those same theorems on faith from James, I neither see that they are true for myself, nor reason my way to believing them. Rather, I believe James, and only thereby come to believe that those theorems are true. So the material objects of my faith are propositions (the same theorems James knows), while the formal object of my faith is believing the speaker (in this case, James), just as the Christian with faith believes God (credere Deo).

Elizabeth Anscombe also talks about believing the speaker, and distinguishes that act of faith from the act of merely believing what the speaker says. (58) When you tell me something I already knew (for example, "George Washington was the first president of the United States"), I believe what you say, but not by believing you. When you present me with a mathematical proof that every integer is the product of its arithmetic mean and harmonic mean, and I follow the proof to its conclusion, I come to believe what you say (the premises and the conclusion), but not by believing you.

Opinion similarly is a matter of believing what the speaker said, but not by believing the speaker. Suppose we have a chemist friend, Sal, and she hypnotizes you to say "Mercuric oxide bonding is endothermic" the next time we meet, and when you say it, I infer that you said it because Sal so hypnotized you, and on that basis infer that the statement is true. My probable inference gives me an opinion that your statement is true. So I believe what you said, but not by believing you. Or suppose you know that I do not trust you, so you try to mislead me by saying something true, in the expectation that I will not believe you; but I see through your ruse and infer that what you say is true. In such a double-bluffing case, I believe what you say, but not by believing you. (59) These cases are far-fetched, but they help us draw the negative point that believing the speaker is not opinion.

Aquinas has his own positive account of what it is to believe the speaker, one that further explains how believing the speaker differs from opinion. He says faith is had by "adhering" to the speaker:

   On the part of the intellect, there are two ways to take the object
   of faith. One is the material object of faith.... Another is the
   formal object, which is like a means [medium] on account of which
   one assents to such a credible [proposition]. In this way the act
   of faith is called credere Deo, because, as said above, the formal
   object is the First Truth [i.e., God], to which a man adheres so
   as, on account of it, to assent to what he believes. (60)

The assent of different propositional cognitive habits (or propositional attitudes) is produced by different means: opinion is produced by inductive inference, science by demonstration, and faith by adherence to the speaker. Aquinas explains adherence further by an analogy:

   Since whoever believes assents to the word of someone, it seems
   that the principal thing in any believing, (61) like an end, is the
   person to whose word one assents, while the items one believes in
   order to assent to that person are secondary. (62)

Assent to a proposition stated by a speaker is like a means to the end of adhering to that speaker.

Nowadays we sometimes use the word "faith" to refer to a kind of trust, as when I say that I have faith that my friend will pay me back; and we sometimes use "faith" to refer to a kind of loyalty, as when I say that my friend kept faith with me by paying me back. Aquinas's talk of adherence, I suggest, combines these notions of faith as trust and faith as loyalty. Aquinas treats faith as trusting belief out of loyalty to the speaker. By contrast, opinion is just a matter of taking something to be probable on one's own evidence, without regard for adhering to the speaker. The loyalty and trust of faith are ways of depending more extensively on a speaker for the truth than one would by opinion alone. For example, faith seems to be incompatible with taking certain precautions against the speaker being wrong, whereas opinion is not. (63) An indication that Aquinas takes this view is that the sin of unbelief is a sin of pride, (64) by which one does not subject one's belief to receiving knowledge from God or from a fellow human. (65) An attempt to establish one's opinion--on some matter one could know about by faith--by considering all the evidence (or at least enough evidence to judge the matter for oneself) is in effect an attempt to replace faith with either opinion or seeing. Such a replacement can be good, epistemically speaking, as when students go from solely relying on the teacher to knowing things for themselves. But it can also be bad, as when one is unjustifiably suspicious of some expert's claim that p merely because p seems improbable on one's own scanty evidence.

We have considered the first feature that distinguishes faith from opinion: faith's formal object as a means to assent. That is, faith requires believing a statement in order to believe the speaker in the sense of adhering to the speaker. Now let's consider the second feature: the choice of faith.

Faith is specially creditable to the believer as something one chooses, whereas opinion and science are not. In several passages Aquinas contrasts faith in this regard with both opinion and science. In one he says that the assent of science is "forced" (cogitur), while opinion "does not have firm assent" and so "does not much seem to have merit" (non multum videtur habere rationem meriti). The assent of faith, by contrast, is subject to free choice, and so can be meritorious. (66) In another he says that in the case of faith the will can

   choose to assent to [a proposition] determinately and distinctly
   because of something sufficient to move the will, but not to move
   the intellect; for example, because it seems good or fitting [bonum
   vel conveniens] to assent to it. And this is the disposition of
   belief, (67) as when someone believes the sayings of some human,
   because it seems to [the audience] appropriate or beneficial
   [decens vel utile]. (68)

Demonstration forces assent, while opinion, I suggest, gives assent no more than is required by the evidence. (69) Even strong opinion, it seems, is stronger only in response to one's evidence. So neither science nor opinion has the special merit of faith.

This last passage indicates that, even if faith is not inferential, it still requires a reason or explanation for it seeming good to the audience to adhere to the speaker. One such reason that Aquinas points to is the audience's recognition of the truthfulness of the speaker. And this is our third distinguishing feature of faith: a typical motivation for faith is that the speaker is truthful. Aquinas frequently says that one believes God because of God's truthfulness (or because God cannot lie). (70) He also says that a good reason for believing a human speaker is the speaker's conscientious veracity, (71) and that when we consider whether to have Christian faith, even miracles "do not prove [the propositions of] faith directly, but prove the truthfulness of those announcing the [propositions of] faith." (72) These passages probably refer to the Aristotelian virtue of truthfulness (veritas or veracitas). Truthfulness is a moral virtue of accurate self-representation that avoids the extremes of boasting and self-deprecation. Truthfulness is motivated by a love of truth, and falls under the virtue of justice, because it is oriented toward giving others the truth. (73) And just as a minimum degree of friendliness and pleasantness is required for the proper functioning of society, so is a minimum degree of truthfulness. (74) Without it, the epistemic divisions of labor by which one believes experts, peers, and teachers would collapse. (75) Aquinas goes beyond Aristotle by noting that as a virtue of accurate self-representation, truthfulness concerns the particular case of accurately representing one's knowledge. (76) The truthfulness of a speaker thus contributes to the rationality of one's faith by putting one in touch with a speaker's knowledge.

Unlike opinion, which restricts one to one's own evidence, faith in a truthful speaker allows one to "share" and be "joined to" the knowledge of the speaker. (77) Thus faith makes one's belief more sensitive to a truthful speaker's evidence, and less sensitive to one's own evidence. (78) So when presented with a knowledgeable and truthful speaker's statement, the audience that restricts itself to inferring from its own evidence will be less certain and have worse epistemic standing than an audience that has faith. For example, suppose both Fay and Opie are told by their friend Knox that the fastest way to the CN tower in Toronto is to take the second highway exit after the tower. They both assent, but Fay has faith while Opie has opinion. As they separately drive there, they each see what looks like a much more direct route. Opie might either change his mind, or at least be more ready to change his mind than Fay. Supposing Knox does know the way, Fay is in a better epistemic position than Opie. But supposing Knox is wrong and does not know the way, Opie can blame only himself for his mistaken inference, whereas Fay can blame Knox for breaking faith with her by not being truthful. (79) Faith has the drawback of making one dependent on the speaker, but it also has the benefit of linking one's justification to that of the speaker in a way that inductive inference does not. This is because faith is not merely a very strong inferential belief that p, supported by a premise about the truthfulness of the speaker, but rather a way of assenting to what a speaker says by means of adhering to that speaker.

VI

Aquinas does not stick to such humdrum examples as these. His contrast is between demon faith (a kind of strong opinion) and Christian faith. Imagine that a demon sees a prophet make a prediction and perform a miracle (for example, raising a corpse to life) as a sign that God confirms his prediction. Realizing that such a miracle could have been performed only with God's power and permission, the demon unwillingly infers with near certainty that the prophet's prediction will come true. (80) But the demon does not see that it will, since only God sees the future, (81) so the demon's assent is still assent to the unseen, and so is still a case of faith in the broad sense (the sense of "faith" explained in section II). Demons are smarter and more knowledgeable than we are, and have more experience with prophetic predictions, so alternative explanations of the miracle and its relation to the prophet's prediction that a human might consider are not open to the demon: demon faith is super-strong testimonial opinion which compels assent "by the evidentness of signs" (82) and "by the perspicacity of [the demon's] natural intellect." (83) The Christian who does not see the miracle, by contrast, has to trust God's message and his messenger. So there is still room for the human to believe by a special act of will, out of loyalty to the speaker. (84)

Here a puzzle arises. In section II we noted that seeing compels belief, but that the unseen leaves room for the will (under the influence of reason) to command belief. This is true even in the fairly automatic case of opinion. So can belief in something unseen really be "compelled" by the evidence? (85) This puzzle dissolves when we see that opinion can be compelled, but not in the way that seeing is compelled. Ordinary opinion might yield no assent or weak assent, but strong opinion is based on arguments or signs strong enough to minimize one's fear of not-p, and so convince one that p.

   Whenever things accepted are in some way assented to, there must be
   something that inclines one to assent: in assent to per se known
   first principles, it is a naturally endowed light ... and in assent
   to what we opine, some verisimilitudes, which, if they were a
   little stronger, would incline one to believe, as opinion helped by
   reasons is called "faith." (86)

Aquinas's idea seems to be that evidence can provide one with grounds for belief that are sufficient for passing a credence threshold, such that one becomes "convinced." (87) Assent can then be compelled, even though p remains unseen, "by the fact that nothing to the contrary is apparent." (88) So even if the demon does not see that p, it might see that all possible alternatives to p are so improbable that assent to p is clearly the only reasonable option. In a sense, then, it is compelled to assent by the evidence.

Now that we have examined three features of testimonial faith (adherence to the speaker as formal object, the choice of faith, and truthfulness as a reason to have faith), and considered a few examples, we are in a position to compare Aquinas's account of faith with some contemporary interpersonal accounts of testimonial justification. Unlike several recent philosophers inspired by the ordinary language analyses of J. L. Austin and H. P. Grice, (89) Aquinas does not focus on the speaker's role or the specific speech act of "telling" by which a speaker takes responsibility for the justification of an audience's beliefs. At most, Aquinas focuses on the speaker's virtue of truthfulness as a reason for the audience to adhere to the speaker in a way that makes inductive inference with regard to p unnecessary. In this regard, Aquinas's account is more like Amon Keren's or Linda Zagzebski's account of epistemic authority as a "preemptive" reason to believe what an epistemic authority believes. (90) A preemptive reason to believe is a reason to believe what the authority tells you, together with a second-order reason not to base one's belief on certain other evidence that may be available to you. (91) A preemptive reason "replaces" certain other evidence you have, such that you defer to the authority on the question of what you ought to believe, the way a soldier defers to the commanding officer on the question of what to do. (92) The main similarity between such accounts and Aquinas's account of faith is that taking someone's telling you that p as a preemptive reason to believe p makes your belief more sensitive to the speaker's knowledge, just as Aquinas's testimonial faith does, and less sensitive to your own evidence. But Aquinas's account of faith is distinctive in the role it gives to adherence as loyalty to the speaker. And Aquinas's idea that a speaker's truthfulness can be a reason to have faith gives him a distinctive virtue account of testimonial trustworthiness that is worthy of further investigation.

Most recent epistemologists of testimony, even when they discuss trust, say little about testimonial trustworthiness. The standard approach is to say that a trustworthy speaker is just one who is both sincere and competent on the topic at hand. (93) Judging a speaker sincere and competent in this way gives good support for testimonial opinion that what the speaker says is true. But a nonvirtuous speaker can be sincere and competent on some occasion, while nevertheless not being worthy of an audience's faith. This is true, for example, of a terrorist who knows where his comrades are hiding and sincerely tells his interrogator where they are, under torture. Sincerity can be compelled by such things as threats, torture, and hypnosis. Even competence on a topic can be compelled by hypnosis, and could conceivably be compelled in science fiction cases by brain lesions or brain-controlling devices. Because speakers compelled in these ways do not speak the truth responsibly out of a virtuous motivation, they are not worthy of the audience's faith. Truthfulness, by contrast, is a virtue and provides a good reason for faith. Truthfulness is also a more fundamental reason to have faith than epistemic authority, since truthfulness governs how a speaker represents his knowledge to others. I can trust truthful Tamara on the topic of motorcycles, even though she is not an epistemic authority on motorcycles, because I can rely on her not to assert things of which she is uncertain, and to admit it when she does not know something. Having faith in a speaker does not require taking the speaker to be an authority, but it does require taking the speaker to be truthful, even when the speaker is an epistemic authority. (94)

VII

We have now seen how Aquinas's take on testimonial justification is not defaultist, but neither is it simply inferentialist or interpersonalist. In this sense, his overall view is pluralist. (95) This is not the place for me to give a detailed assessment of other views and how they compare with Aquinas's, so I will restrict myself to one observation. Other views that explain all testimonial justification in one way are simpler but seem not to account for as much of the phenomena. By including both testimonial opinion and testimonial faith, Aquinas can say (unlike strict inferentialists) that we do sometimes get justified belief by choosing to trust a speaker, and he has some resources for explaining how such faith is justified. He can also say (unlike strict interpersonalists) that inferential belief from eavesdropping or court testimony is nevertheless testimonially based belief, properly speaking, of the strong opinion kind. (96) Aquinas is an antireductionist, in the sense that he would not say that all testimonial justification reduces to other kinds of justification (such as inferential justification). But his account raises distinctive considerations deserving of more discussion, about the role of adherence to the speaker, and the role of truthfulness in supporting such adherence. (97)

University of Toronto

Correspondence to: [email protected].

(1) David Hume, An Enquiry Concerning Human Understanding, 2nd ed. (1777; Indianapolis: Hackett Publishing Company, 2011), sec. 10, pt. 1. Emphasis added. Benjamin McMyler, Testimony, Trust, and Authority (Oxford: Oxford University Press, 2011), chap. 1, sec. 3, argues that Hume's treatise on miracles responds to the account of testimonial justification from eighty years before, in Antoine Arnauld and Pierre Nicole's Port-Royal Logic, pt. 4, chaps. 12-14. Arnauld and Nicole argue that testimony can provide a kind of knowledge (connoissance) that is just as certain as science.

(2) Thomas Reid, An Inquiry into the Human Mind (Indianapolis: Hackett Publishing Company, 1983), chap. 6, sec. 24. Emphasis added.

(3) Some accounts of testimonial justification from before Hume are mentioned in chap. 1 of Linda Zagzebski, Epistemic Authority: A Theory of Trust, Authority, and Autonomy in Belief (New York: Oxford University Press, 2012); chap. 1 of McMyler, Testimony, Trust, and Authority, Robert Pasnau, "Medieval Social Epistemology: Scientia for Mere Mortals," Episteme 7, no. 1 (February 2010): 23-41; and chap. 1 of Coady, Testimony.

(4) Coady, Testimony, 17.

(5) Discussions of Aquinas and testimony have so far focused on his account of faith in God, infused into a human by divine grace. Eleonore Stump, "Faith, Wisdom, and the Transmission of Knowledge through Testimony," in Religious Faith and Intellectual Virtue, ed. Laura Frances Callahan and Timothy O'Connor (New York: Oxford University Press, 2015): 204-30, develops her interpretation of Aquinas on the special case of faith in God, and then offers her own "Thomistic" account of testimonial knowledge. See also John Lamont, Divine Faith (Aldershot: Ashgate, 2004). I have tried to reconstruct Aquinas's general view of testimonial justification from what he says about various testimonial cases, not just the case of faith in God, and I have avoided the controversial question whether, on Aquinas's account of knowledge, testimony can provide knowledge. My interpretation further differs from Stump's Thomistic view by not positing an empathic experience by which one comes to initially trust a speaker, by not counting connatural knowledge as testimonial knowledge, and by not making testimonial trust dependent on connatural knowledge. But I welcome further investigation into the role of connatural knowledge in Aquinas's understanding of testimonial justification.

(6) Elizabeth Anscombe, "What Is It To Believe Someone?" in Rationality and Religious Belief ed. C. F. Delaney (Notre Dame, Ind.: University of Notre Dame Press), 141-51. Other recent writers in this vein include McMyler, Testimony, Trust, and Authority, and Richard Moran, "Getting Told and Being Believed," Philosopher's Imprint 5, no. 5 (August 2005): 1-29.

(7) Theophilus of Antioch, Ad Autolycum, trans. Robert M. Grant (Oxford: Clarendon Press, 1970), 1.8. For further early defenses of faith, see Lamont, Divine Faith; and Avery Dulles, A History of Apologetics (Philadelphia: The Westminster Press, 1971).

(8) Thomas Aquinas, Quaestiones Disputatae de Veritate (henceforth, De veritate), q. 14 a. 10c (Leonine ed. 22); Super Boetium de Trinitate (henceforth, In B. de Trini), q. 3, a. 1 (Leonine ed. 50); Summa contra gentiles (henceforth, SCG), bk. 1, c. 4 (Leonine ed. 13); Summa theologiae (henceforth, ST) I, q. 1 a. 1 (Leonine ed. 4); ST II-II, q. 2, a. 4 (Leonine ed. 8).

(9) W. V. O. Quine and J. S. Ullian, The Web of Belief, 2nd ed. (New York: Random House, 1978), 61-62.

(10) In B. de Trin., q. 2, a. 2, ad 7: "The ultimate first principle of any science is always understanding (intellectus), but this is not always the proximate principle; rather, sometimes faith is the proximate principle of a science, as in the subalternated sciences, in which conclusions come proximately from faith in the things supposed from a superior science, but come ultimately from the understanding of the superior knower, who has certainty, through understanding, about the things [merely] believed [in the subalternated science]. And similarly the proximate principle of [theology] is faith, but the first principle is the divine intellect, which we believe. And yet the goal of our faith is that we come to understand what we believe, just as if a subordinate scientist [inferior sciens] were to learn the science of a superior scientist, such that things previously only believed would become understood and known." All translations of Aquinas's works are my own. The classic work on this topic is Marie-Dominique Chenu, La theologie comme science au XIIIe siecle (Paris: J. Vrin, 1969), but see also John Jenkins, Knowledge and Faith in Thomas Aquinas (Cambridge: Cambridge University Press, 1997); Robert Pasnau, "Divisions of Epistemic Labour: Some Remarks on the History of Fideism and Esotericism," Continuity and Innovation in Medieval and Modern Philosophy: Knowledge, Mind, and Language, Proceedings of the British Academy 189 (Oxford: Oxford University Press, 2013): 83-117; and John Hawthorne's criticism of Pasnau, "Aquinas on Faith and Knowledge: Response to Robert Pasnau," in the same volume, 119-33. For further discussion in Aquinas, see especially Expositio libri Posteriorum, bk. 1 (henceforth In I Post.), lect. 15 and lect. 25 (Leonine ed. 1.2). Other relevant passages include In 1 Sent., prol., q. 3, a. 2, ad 2 (Mandonnet ed.), In 3 Sent., d. 24, q. 1, a. 2, qc. 2, ad 3 (Mandonnet ed.); De veritate, q. 14, a. 9, ad 3; In B. de Trin., q. 2, a. 2 and q. 3, a. 1c; In I Post., lect. 17, n. 3 and lect. 41, n. 2; ST I, q. 1, a. 2c; ST I, q. 79, a. 9c; ST D-D, q. 9, a. 2, ad 3.

(11) In B. de Trin., q. 3, a. 1c. See also STII-II, q. 109, a. 3.

(12) See also De veritate, q. 14, a. 10c; In B. de Trin., q. 3, a. lc; Lectura super Ioannem (henceforth In Ioh.) 8.4 (Vives 19); SCG, bk. 3, c. 152, n. 4 (Leonine ed. 11); ST II-II, q. 2, a. 3c (Leonine ed. 8); Expositio et lectura super Epistolas Pauli (on Hebrews, lect. 11, n. 1, Vives 21).

(13) In B. de Trin., q. 3, a. 1c.

(14) Here is an illustration: A physicist friend of mine told me of a physics undergraduate who had spent ten years on his undergraduate degree, because he spent most of his time in the library double-checking everything his teachers told him.

(15) "'Faith' properly speaking is when someone assents to those things which he does not see." In 3 Sent., d. 23, q. 3, a. 4, qc. 3 expos. Aquinas uses the term "assent" the way philosophers today use the term "belief," while he typically reserves the term "believe" (credere) for the act of faith, either in the broad sense explained here, or in the narrow sense explained in section V.

(16) In 3 Sent., d. 17, q. 1, a. 2, qc. le; ST I-II, q. 17, a. 6c; STII-II, q. 1, a. 4c.

(17) STI, a. 85, q. 3c. Eleonore Stump, "Faith, Wisdom, Testimony," suggests that the way faith provides knowledge without direct acquaintance is analogous to blindsight, as described in Lawrence Weiskrantz, Blindsight (Oxford: Oxford University Press, 2009). Following a later scholastic distinction, we could also say that faith can provide notitia abstractiva but not notitia intuitiva.

(18) De veritate, q. 12, a. 1c; q. 14, a. le; ST I, q. 12, a. 13, ad 3; ST II-II, q. 2, a. 1, ad 3; q. 4, a. 1c.

(19) "Certitude is nothing other than determination of the intellect to one: the stronger the determination, the greater the certitude." In 3 Sent., d. 23, q. 2, a. 2, qc. 3c. Aquinas implies here and elsewhere (In I Post., lect. 1, n. 1 and lect. 41, n. 2; ST II-II, q. 5, a. 4c and ad 2; ST II-II, q. 70, a. 2c) that certainty comes in degrees. And Aquinas identifies three different kinds of certainty relevant to determination of the intellect: certitude of evidentness, when the object is seen; certitude of adhesion, when the will is fully determined to assent; and certitude of cause, when the cause of one's assent is perfectly reliable or efficacious (as in the case of divinely infused Christian faith). Science has the certitude of evidentness, which divinely infused Christian faith lacks, but they both have the certitude of adhesion, and sometimes Aquinas says that Christian faith's adhesion is stronger than that of science, due to its cause (In 3 Sent., d. 23, q. 2, a. 2, qc. 3, ad 1, ad 2; De veritate, q. 14, a. 1c and ad 7; ST II-II, q. 4, a. 8c and ad 1). Aquinas has a few puzzling passages where he says that the articles of faith are self-evident (per se nota) to the faithful (In 1 Sent., prol., q. 1, a. 3, ad 2; ST I-II, q. 100, a. 3, ad 1 and a. 4, ad 1). But they cannot be self-evident in the sense that they are seen to be true (ST II-II, q. 1, a. 4, a. 5), so the most charitable way to interpret these passages is to say that the articles of faith (a) play the role of first principles, just as in the subalternated sciences (ST I, q. 1, a. 2), and (b) have as much certainty of cause and adhesion as the first principles of a scientia, thanks to the divine infusion of grace. On this point, I am inclined to follow Chenu, La theologie comme science au XIIIe siecle, and the earliest Thomists, who interpreted Aquinas's claim that theology is a scientia to mean that it has a scientific character, even though the articles of faith do not have the self-evidence required for a scientia, strictly speaking. For alternative views, see M. V. Dougherty, "Aquinas on the Self-Evidence of the Articles of Faith," The Heythrop Journal 46 (2005): 167-80; and Jenkins, Knowledge and Faith in Thomas Aquinas, chap. 2.

(20) De veritate, q. 14, a. 9c.

(21) In 3 Sent., d. 24, q. 1, a. 3, qc. 3, ad 3; In B. de Trin., q. 2, a. 1, ad 5; ST II-II, q. 2, a. 9, ad 3.

(22) See In 2 Sent., d. 7, q. 2, a. 2, ad 5; In B. de Trin., q. 3, a.1c; SCG, bk. 3, c. 154, n. 1; ST I, q. 12, a. 13, ad 3; De veritate, q. 12, a. 1c.

(23) In 1 Sent., d. 17, q. 1, a. 4c; ST II-II, q. 2, a. 9, ad 2. By saying that assent is "natural," Aquinas does not mean it is "natural and easy," as Eleonore Stump says in Aquinas (New York: Routledge, 2003), 362, but just that it is a naturally automatic process, not subject to voluntary choice. Stump's example of a mother who "finds herself assenting, whether she wants to do so or not, to the proposition that the judge dislikes her son's performance" is not an example of seeing, but of strong opinion, akin to demon faith. I discuss strong opinion and demon faith in section V below.

(24) ST II-II, q. 1, a. 4c.

(25) ST I-II, q. 17, a. 6c: "[R]eason reflects on itself, so it can also order its own act just as it orders the acts of other powers. Hence its act can also be commanded. But some things are apprehended which do not convince the intellect so much that one cannot for some reason assent or dissent, or at least suspend assent or dissent, and in such cases the assent or dissent is in our power, and falls under one's command." For a contemporary defense of this idea, see Zagzebski, Epistemic Authority, 5, 63; and McMyler, Testimony, Trust, and Authority, 150.

(26) I here give a simplified survey of the growing literature on the epistemology of testimony. For fuller surveys, see Jennifer Lackey, "Knowing from Testimony," Philosophy Compass 1, no. 5 (2006): 432-48; Axel Gelfert, A Critical Introduction to Testimony (New York: Bloomsbury, 2014); and John Greco, "Recent Work on Testimonial Knowledge," American Philosophical Quarterly 49, no. 1 (January 2012): 15-28.

(27) See, for example, Coady, Testimony, and Tyler Burge, "Content Preservation," The Philosophical Review 102, no. 4 (October 1993): 457-88.

(28) It is difficult to come up with a test case in which the hearer has no evidence for or against a speaker's trustworthiness or a speaker's claim. Suppose you ask a stranger for directions, or ask the time. You are likely to know from the stranger's behavior, clothing, and so on, that the stranger is from a society where the members have internalized a norm against lying about such things. On the other hand, you have probably been lied to (at least as a joke) by fellow members of such a society. So even when speaking with strangers who seem to have no motive to lie, you already have evidence for and against what they tell you. Jennifer Lackey, Learning from Words (Oxford: Oxford University Press, 2008), 168-70, tries to avoid such complications by giving an example of someone reading an alien's diary. For an argument that ultimately the defaultist and the inferentialist must agree that some basic monitoring of a speaker for trustworthiness is required, see David Henderson, "Testimonial Beliefs and Epistemic Competence," Nous 42, no. 2 (June 2008): 190-221.

(29) See, for example, Elizabeth Flicker, "Against Gullibility," in Knowing From Words, ed. Bimal Krishna Matilal and Arindam Chakrabarti (Dordrecht: Kluwer Academic Publishers, 1994): 125-61; and Elizabeth Flicker, "Second-Hand Knowledge," Philosophy and Phenomenological Research 73, no. 3 (November 2006): 592-618.

(30) On the various ways of being a reductionist or antireductionist, see Greco, "Recent Work on Testimonial Knowledge."

(31) For example, see McMyler, Testimony, Trust, and Authority and Lackey's criticism of such views in Learning from Words, chap. 8.

(32) McMyler, Testimony, Trust, and Authority, chap. 2, argues that from an assurance a hearer gains a right to defer challenges to the speaker. He argues in the rest of the book that this right results from the speaker giving the hearer a "second-personal reason" to believe what is said. Amon Keren, "Trust and Belief: A Preemptive Reasons Account," Synthese 191 (August 2014): 2593-615, argues that from an assurance a hearer gains a right not to take precautions against being wrong, as part of a "preemptive reason" to believe what is said. However, Keren also argues that "the justification of trust-based beliefs depends on the justification of non-trust-based beliefs," such as that the speaker is trustworthy (2613). So in the end Keren's view is reductionist.

(33) See In B. de Trin., q. 3, a. 1c, quoted in section I above, and ST II-II, q. 109, a. 3, ad 1.

(34) am grateful to Martin Pickave for pointing out this line of thought.

(35) ST II-II, q. 60, a. 4c: "[F]rom the fact that someone has a bad opinion of someone [else] without sufficient cause, he does him an injustice and despises him. But no one ought to despise or bring about whatever harm without being forced to [absque causa cogente]. So where no manifest signs of malice in someone appear, we should take him to be good, interpreting for the best [in meliorem partem] what is doubtful."

(36) For example, a servant cannot give alms on his master's behalf on the principle that it would be a good thing to do so, because "we ought not to presume any good of anyone, but only that good without which the person would sin, since we should not believe bad of anyone unless the person's malice is clear to us" (In 4 Sent., d. 15, q. 2, a. 5, qc. 3, ad 1; compare In 4 Sent., d. 27, q. 1, a. 2, qc. 4, ad 2).

(37) See ST II-II, q. 60, a. 4c and In Ioh. 2.3. Presuming the best takes two forms, neither of which is a case of outright belief. Either one "supposes" the best of another to avoid the injustice of contempt (just as one "supposes the worst" in a medical case in order to make sure of a cure), or one "determines" a case where one's evidence is incomplete (like a judge deciding a case in accordance with the evidence presented, regardless of what the judge believes). ST II-II, q. 60, a. 4, ad 3.

(38) ST II-II, q. 70, a. 3, ad 2: "One should presume good of everyone unless the contrary is apparent, so long as it does not tend toward anyone's danger; for then one should apply caution that one not believe anyone easily." ST II-II, q. 60, a. 4, ad 1: "It can happen that he who interprets for the best is deceived rather often [frequentius fallitur]. But it is better that someone be deceived frequently, having a good opinion of some bad man, than that he be deceived less often, having a bad opinion of some good man, because in the latter case he does an injustice to someone, but not in the first."

(39) In 3 Sent., d. 24, q. 1, a. 3, qc. 2, ad 1.

(40) Quodlibeta 3, q. 4, a. 2c (Leonine ed. 25).

(41) In 3 Sent., d. 23, q. 3, a. 4, qc. 3 expos.

(42) Aquinas says in De veritate, q. 14, a. 1, that opinion does not yield "assent" because it does not produce firm acceptance of p rather than not-p. But elsewhere he says that opinion yields assent, and that if the grounds for opinion were "a little bit stronger, they would incline one to believe, in the sense that 'faith' is said to be opinion helped by reasons" (In B. de Trin., q. 3, a. 1, ad 4). Elsewhere Aquinas glosses "strong opinion" as opinion "strengthened by reasons" (In 1 Sent., prol., q. 1, a. 3, qc. 3, ad 1). So Aquinas seems to accept the thesis that credence above a threshold becomes full assent, and so, in one sense, "faith." I discuss strong opinion further in section V below.

(43) In 1 Post., lect. 1, n. 6 is Aquinas's fullest discussion on the way opinio is justified, but see also the way he contrasts it with science in In 1 Post., lect. 44; STI-n, q. 51, a. 3c; In 3 Sent., d. 17, q. 1, a. 2, qc. 1c; and In B. de Trin., q. 3, a. 1, ad 4.

(44) "STII-II, q. 2, a. 1c; In 1 Post., lect. 1.

(45) Aquinas calls this "fear of the contrary," and says that "it is of the nature of opinion that what one thinks, one thinks possible to be otherwise" (ST I-II, q. 67, a. 3c; ST II-II, q. 1, a. 5, ad 4). The "faith or opinion" of In 1 Post., lect. 1, I suspect, is identical with what Aquinas elsewhere calls "strong opinion" (see sections IV and V below). On this topic, see Martin Pickave, "Human Knowledge," in The Oxford Handbook of Aquinas, ed. Eleanore Stump and Brian Davies (Oxford: Oxford University Press, 2012), 311-26, 325 n. 30; and Francis Martin Tyrrell, The Role of Assent in Judgment: A Thomistic Study (Washington, D.C.: The Catholic University of America Press, 1948), 87.

(46) ST I, q. 12, a. 7c.

(47) ST II-II, q. 70, a. 2c.

(48) I should note that Aquinas emphasizes the distinctive interpersonal way of believing other humans when using them as cases that help us understand faith in God by analogy. Elsewhere, Aquinas sometimes ignores the interpersonal aspects of believing a teacher in his explanations of learning (ST I, q. 117, a. 1; In 2 Sent., d. 9, q. 2, ad 4), or exalts faith in God over faith in human testifiers in a way that suggests that outright trusting of human testifiers is irrational "because a human can deceive and be deceived" (De rationibus fidei, chap. 7, Leonine ed. 40).

(49) For the view that faith is opinion, see Thomas Hobbes, Leviathan, ed. C. B. Macpherson (London: Penguin Books, 1968), 1.7. For the view that believing a speaker is merely a matter of probability, see John Locke, An Essay Concerning Human Understanding, ed. Roger Woolhouse (London: Penguin Books, 1997), bk. 4, chap. 15.

(50) "Credere deum" refers to all the propositions believed on faith, not just the proposition that God exists. See In 3 Sent., d. 23, q. 2, a. 2, qc. 2c; ST II-II, q. 2, a. 2c.

(51) In 3 Sent., d. 23, q. 2, a. 2, qc. 2c; In B. de Trin., q. 3, a. 1, ad 5.

(52) Expositio et lectura super Epistolas Pauli Apostoli (part on Romans, lect. 4, n. 1, Vives 21).

(53) De veritate, q. 12, a. 1c; ST II-II, q. 171, a. 1c.

(54) ST I-II, q. 57, a. 2c; I-II, q. 51, a. 3c; II-II, q. 4, a. 8c; In 1 Post., lect. 4, n. 8. See also In 2 Post., lect. 20. On the induction involved in intellection of first principles see Scott MacDonald, "Theory of Knowledge," in The Cambridge Companion to Aquinas, ed. Norman Kretzmann and Eleonore Stump (New York: Cambridge University Press, 1993), 183, 194-95.

(55) In 3 Sent., d. 27, q. 2, a. 4, qc. 1, ad 3: "Material diversity of objects suffices for distinguishing acts numerically, but they are not distinguished according to species except by diversity of formal objects. But the diversity of formal objects is according to the notion which a habit or power principally attends to." ST I, q. 59, a. 2, ad 2: "Powers are not distinguished by a distinction of material objects, but by a formal distinction, following the notion of the object. And thus a notional diversity of good and true suffices for distinguishing intellect from will."

(56) ST I, q. 59, a. 4c; I-II, q. 8, a. Ic; I-II, q. 9, aa. 1-2.

(57) See section II above.

(58) Anscombe, "What Is It to Believe Someone?" 144-45. We might find a similar distinction at work in ST II-II, q. 129, a. 6c, where Aquinas says "it pertains to faith to believe something and someone," referring respectively to the proposition believed and the person believed.

(59) Anscombe, "What Is It to Believe Someone?" 145, describes a double-bluffing case.

(60) ST II-II, q. 2, a. 2c

(61) Note that in this context "believing" refers specifically to the act of faith in a narrow sense (ST II-II, q. 2, a. 1 and following), not to the assent that can also be found in other cognitive habits like knowledge from demonstration or opinion from inductive inference.

(62) ST II-II, q. 11, a. 1c.

(63) See Keren, "Trust and Belief' for an account of testimonial trust on which trusting someone for the truth requires seeing oneself as having a reason not to take precautions against the speaker's testimony being false.

(64) ST II-II, q. 10, a. 1, ad 3.

(65) ST II-II, q. 162, a. 3, ad 1.

(66) ST II-II, q. 2, a. 9, ad 2.

(67) Here again "belief' refers specifically to the act of faith in a narrow sense.

(68) De veritate, q. 14, a. 1c.

(69) The passage just quoted is in tension with ST II-II, q. 1, a. 4c, where Aquinas says that even opinion involves a "choice." I suggest that this "choice" involved in opinion is merely a choice to assent to the degree justified inductively on one's own evidence, and is fairly automatic. On the automatic role of the will in Aquinas, see also Stump, "Faith, Wisdom, Testimony."

(70) Expositio et lectura super Epistolas Pauli Apostoli (part on 2 Timothy, lect. 1, n. 4, and part on Galatians, lect. 6, n. 2, Vives 21); ST II-II, q. 2, a. 4c; ST II-II, a. 5, q. 2c; ST II-II, q. 89, a. 1c.

(71) In 3 Sent, d. 23, q. 3, a. 4, qc. 3 expos.

(72) In 3 Sent., d. 24, q. 1, a. 2, qc. 2, ad 4.

(73) For Aquinas on truthfulness, see Sententia libri Ethicorum, lib. 4, lect. 15 (Leonine ed. 47); ST I, q. 16, a. 4, ad 3; and ST II-II, q. 109. See also Kevin White, "The Virtues of Man the 'animal sociale,' 'affabilitas' and 'veritas' in Aquinas," The Thomist 57, no. 4 (October 1993): 641-53; and Kevin Flannery, "Being Truthful (or Lying to) Others about Oneself," in Aquinas and the "Nicomachean Ethics", ed. Tobias Hoffmann, Jbm Muller, and Matthias Perkams (New York: Cambridge University Press, 2013): 129-45.

(74) ST II-II, q. 114, a. 2, ad 1.

(75) In ST II-II, q. 109, a. 3, ad 1, Aquinas says this of social dependence for the truth generally, but does not cite the three epistemic divisions of labor mentioned in In B. de Trin., q. 3, a. 1 which I discussed in section I above.

(76) ST II-II, q. 109, a. 3, ad 3: "But since knowable truths, inasmuch as they are known by us, are about us and pertain to us, in this way the truthfulness [veritas] of teaching can pertain to this virtue, and whatever other truthfulness [veritas] by which one manifests by word or deed what one knows."

(77) Aquinas says that Christian faith joins us to God's knowledge in De veritate, q. 14, a. 8c; SCG, bk. 1, c. 4; ST I, q. 1, a. 1c.

(78) ST II-II, q. 4, a. 8, ad 2: "Other things being equal, vision is more certain than hearing. But if [the authority] of the speaker greatly exceeds the vision of the seer, then hearing is more certain than vision. Someone of little knowledge is made more certain [magis certificatur] about something he hears from someone extremely knowledgeable [scientissimo] than about something seen by his own reason. And man is much more certain about what he hears from God, who cannot be deceived, than about what he sees with his own reason, which can be deceived." See Keren, "Trust and Belief," 2611-12, on the way trust makes one "less sensitive to evidence available to one" but "more sensitive to evidence available to the speaker."

(79) To be more precise, Opie can blame both himself for his mistaken inference and Knox for undermining the minimal faith (in yet another sense of "faith," namely, trustworthiness, In 3 Sent., d. 23, q. 3, a. 4, qc. 3 expos.) required for the good functioning of society (in B. de Trin., q. 3, a. 1c; ST II-II, q. 109, a. 3, ad 1). Fay, on the other hand, can fully blame Knox, unless for some reason she should not have taken Knox to be a truthful person (at least on the subject at hand). For a similar distinction, see Miranda Flicker, "Group Testimony? The Making of a Collective Good Informant," Philosophy and Phenomenological Research 84, no. 2 (2012): 257, on the difference between "ethical let-down" (appropriate for Opie to feel when Knox is wrong about something he said) and "betrayal" (appropriate for Fay to feel when Knox is wrong about something he assured her of).

(80) See ST II-II, q. 5, a. 2c; ST II-II, q. 178, a. 2; In Ioh., 9.3; De veritate, q. 14, a. 10, ad 11.

(81) ST I, q. 86, a. 4c. See also In 2 Sent, d. 7, q. 2, a. 2c; De veritate, q. 8, a. 12c and ad 3; De veritate, q. 12, a. 10c; SCG, bk. 1, c. 63, n. 5; ST I, q. 57, a. 3c; Quaestiones disputatae de malo, q. 16, a. 7c (Leonine ed. 23).

(82) De veritate, q. 14, a. 9, ad 4: "The demons do not assent voluntarily to the things they are said to believe, but are coerced by the evidentness of signs [sed coacti evidentia signorum], from which they are convinced that what the faithful believe is true, although those signs do not make visible what is believed so that through them they could be said to have vision of the things they believe. Thus 'credere' is said almost equivocally [quasi aequivoce) of faithful humans and of demons."

(83) ST II-II, q. 5, a. 2, ad 2.

(84) A similar case for Aquinas is the case of the disciple Thomas, who both saw Jesus resurrected and had faith in him. Aquinas says that "Thomas saw one thing and believed another. He saw [Jesus] the human and, believing, confessed him to be God, when he said 'My lord and my God'." ST II-II, q. 1, a. 4, ad 1.

(85) Terence Penelhum, "The Analysis of Faith in St. Thomas Aquinas," Religious Studies 13, no. 2 (1977): 133-54, draws attention to this puzzle.

(86) In B. de Trin., q. 3, a. 1, ad 4.

(87) Aquinas seems to use "be convinced" to mean "give full assent to," as when he says that "there can be something active that totally overcomes the corresponding passive power, as when one per se nota proposition convinces the intellect to assent firmly to a conclusion." STI-II, q. 51, a. 3c. See also ST I-II, q. 17, a. 6c. He says that by faith one is "convinced" in ST II-II, q. 4, a. 1c; ST II-II, q. 5, a. 2; De veritate, q. 14, a. 2, ad 14; In Heb., 11.1.

(88) In 3 Sent., d. 23, q. 3, a. 3, qc. lc.

(89) See, for example, Moran, "Getting Told and Being Believed"; Edward Hinchman, "Telling as Inviting to Trust," Philosophy and Phenomenological Research 70, no. 3 (2005): 562-87; McMyler, Testimony, Trust, and Authority; Paul Faulkner, Knowledge on Trust (Oxford: Oxford University Press, 2011), chap. 6.

(90) See Zagzebski, Epistemic Authority, especially chaps. 5-7 and 9, and Keren, "Trust and Belief."

(91) take part of this formulation of a preemptive reason from Keren, "Trust and Belief," 2600.

(92) McMyler, Testimony, Trust, and Authority, 146-49, gives this analogy in support of the view that a telling can give you a "second-personal reason" to believe what the speaker says.

(93) See, for example, Elizabeth Flicker, "Telling and Trusting: Reductionism and Anti-Reductionism in the Epistemology of Testimony," Mind 104, no. 414 (April 1995): 393-411; McMyler, Testimony, Trust, and Authority, 93-94. Katherine Hawley, "Trust, Distrust and Commitment" Nous 48, no. 1 (2014): 120, has a more nuanced account of testimonial trustworthiness, but one that still fails to provide a sufficiently virtue-based account of trustworthiness to licence the kind of trust required for testimonial faith. Zagzebski, Epistemic Authority, chap. 6, comes closest to Aquinas in this regard, by requiring not just "sincerity" and "accuracy" of belief for testimonial trustworthiness, but "conscientious" sincerity and accuracy.

(94) For a more detailed argument that speaker trustworthiness should be understood as a Thomistic virtue of truthfulness, see Matthew Kent Siebert, "Truthfulness and Trust," American Catholic Philosophical Quarterly (forthcoming).

(95) For different pluralist approaches to testimonial justification, see John Greco, "Testimonial Knowledge and the Flow of Information," in Epistemic Evaluation: Purposeful Epistemology, ed. David K. Henderson and John Greco (Oxford: Oxford University Press, 2015): 274-90; Mikkel Gerken, "Internalism and Externalism in the Epistemology of Testimony," Philosophy and Phenomenological Research 87, no. 3 (November 2013): 532-57; Zagzebski, Epistemic Authority, chap. 6; and Faulkner, Knowledge on Trust.

(96) To be fair, those who take the interpersonalist view, like McMyler, do recognize that it is appropriate in some less than proper sense to say that such belief is "testimonial" and that it still can be a source of knowledge.

(97) I am grateful to an audience at the Pacific meeting of the American Philosophical Association, April 2013, and to an audience at Saint Louis University, December 2014, for helpful comments. I am also very grateful to Martin Pickave, Peter King, and Jennifer Nagel for comments on earlier drafts of this paper. This research was made possible through the support of a grant from the John Templeton Foundation, through a scholarship from the University of Toronto, and through the support of Saint Louis University.