The Thomistic account of rationality -- I'm not talking about its philosophical psychological, but its account of rational thought -- is an interestingly nuanced one. It is famously summarized in Aquinas's preface to his unfinished commentary on Aristotle's Posterior Analytics (although there are other places in which it comes up). In this preface, he is talking about logic, of which the Posterior Analytics considers the summit or peak (namely, demonstrative reasoning, the kind that gives real knowledge). And he notes that logic is structured according to the various acts of reason, and on the basis of this gives a very famous account of the books of the Organon, the foundation of most medieval logic. We start with a general division:
There are three acts of reason, of which the first two belong to reason insofar as it is some kind of understanding (intellectus).
(1) One act of intellect is the understanding (intelligentia) of indivisibles or simples, according to which it conceives what the thing is. And this action is called information of the understanding or intellectual imagination. And to this action of reason is ordered the teaching that Aristotle hands down in the book of Predicaments (= Categories).
(2) The second action of understanding is intellectual composition or division, in which the true or the false is found. And this act of reason is served in the teaching that Aristotle hands down in the book of Perihermeneias (= De Interpretatione).
(3) The third act of reason concerns that which is proper to reason, to wit, discursively going from one thing to another (discurrere ab uno in aliud), such that from what is known one comes to cognition of what is unknown. And this act is considered in the remaining books of logic.
So here we have the basic structure of logic as a rational science: reason starts from basic objects of understanding, organizes these into things that can be true or false, and uses these to discover even more. The first part of logic, then, concerns what we might call concepts or terms; the second, what we might call propositions; and the third what we might call inferences or deductions.
So far, so good. But the third part of logic ends up being more complex than one might have originally assumed, because the act of reason it considers can be necessary, or probable, or fundamentally flawed. If we consider the necessary kind, we have what Aquinas calls logica judicativa, which is also called analytics because it analyzes or resolves things into their principles. This has two parts:
(3a1) The certitude of the judgment, which is had by resolution, is either from the form of the deduction as such, and to this is ordered the book of Prior Analytics, which is about deduction simply speaking;
(3a2) or it is also from the matter, because the propositions posited are are self-evident and necessary (per se et necessariae), and to this is ordered the book of Posterior Analytics, which is about demonstrative deduction.
Prior Analytics is the logical discipline concerned with what it means to say that one claim definitely follows from another; whereas Posterior Analytics concerns what it means to say that we come to know, in the most proper sense, conclusions on the basis of their premises. Thus both of these have to do with knowledge: conditional knowledge in the case of the former and knowledge simply speaking in the latter.
If we consider the kind of reasoning that can be called, in a broad sense, probable, Aquinas calls this logica inventiva -- while 'inventiva' can mean 'inventive' under certain circumstances, it primarily means 'having to do with discovery'. The word is usually translated in this context as 'investigative', but this doesn't quite capture it: this part of logic doesn't merely look into things -- it actually discovers them. Discovery is not all about necessities and certainties; we are at a level less than knowledge here. We are often here not in the realm of knowledge but of belief (fides) or opinion:
(3b1) ...and to this is ordered topics or dialectics. For the dialectical deduction that Aristotle considers in the book of the Topics comes from the probable.
But sometimes calling it belief or opinion is a little strong: what we get is not a definite, albeit imperfect, acceptance of one side of the question, but rather a leaning or inclination to one side. This act of reason Aquinas calls 'suspicion'; as in 'I suspect that the answer is such-and-such'. This gives us a second part of logica inventiva:
(3b2) Sometimes, however, belief or opinion is not completely formed, but a kind of suspicion, because it is not wholly inclined to one part of a contradiction, but more inclined to that one than this one. And to this is ordered Rhetoric.
We can have an even more tenuous result of our inquiry, though:
(3b3) Sometimes a mere estimation according to some representation inclines toward some part of a contradiction, in the way that food is made to be abhorred by a man if it is represented under the likeness of something abhorrent. And to this is ordered Poetics, for the poet is drawing to some virtue by some wholesome representation.
It is not customary with us to consider rhetoric and poetics as part of the rational sciences, or as concerned with deduction, but it was commonplace in the Middle Ages. In rhetoric and poetics, as in analytics or dialectics, we move from one thing to another in such a way as to discover the unknown on the basis of the known. This is a process subject to some uncertainty, and thus in itself only deals with probabilities, in a large sense of the term. Dialectics deals with the probable in the strict sense: the things that happen for the most part, the conclusions where we have almost eliminated all the possible alternatives, etc. But we can also have reasoning that deals with the probable in a looser sense. In rhetorical deduction we are in essence determining whether things fit with general opinion, whether they seem like they might be likely, and so forth. And in poetic deduction we are concerned with imaginative representation -- what is imaginatively plausible, what is attractive and repulsive, and so forth. Both rhetoric and poetics have an especially practical aspect, and the very fact that the reasoning they consider are the weakest kinds of reasoning is the source of their practical strength. They don't strictly have to consider truth or falsehood in the usual sense. Rhetoric is realm of reasoning where we are concerned with 'good enough for our purposes'; poetic is the realm of reasoning where we are concerned with 'good enough for representation'.
Aquinas doesn't develop the idea of poetic deduction or inference at length, but he is almost certainly thinking of Avicenna here. In Avicenna's account, we are all very familiar with poetic deductions or syllogisms: that is what metaphors are. Metaphors are the enthymemes of poetics. And Avicenna argues at some length that you can expand every metaphor to a syllogistic deduction. Now, the premises of these syllogisms may well be false in the strict sense, but Avicenna also holds that this is not an impediment here: what poetic syllogism requires is imaginative assent, what we might call acceptability of the representation; and this is true of the conclusion, as well. For whatever reason, Avicenna always uses a particular Sufi example, which certainly livens up his discussion of poetic syllogisms: "A rose is a mule's anus with dung in the middle." The point of this is not that this is some scientific account of a rose. The point rather is that this is a representation of the rose as worthless, and it is the kind of representation that will be acceptable if, for instance, you are a Sufi discussing the value of transient and earthly things in comparison to the value of eternal and divine things.
Aquinas never commits to anything as elaborate as Avicenna, and one suspects that if he had ever commented on the Poetics he would have made some modifications. But it does seem that the general idea is operative here, and it's almost inevitable that it would be: both Avicenna and Averroes treat Aristotle's Poetics as a logical book. Borges has a famous little short story in which he depicts Averroes trying to figure out the Poetics without any knowledge of what drama is. It's not a very accurate depiction of the Averroist approach to the Poetics, but it is certainly true that the approach of the commentators is very different from what we would think the obvious one. We approach Aristotle's Poetics as about creating and performing dramatic situations. But the Islamic commentators thought that this was a secondary feature of the work; for them, what the Poetics primarily describes is the way in which a mind manifests ideas, to itself or another mind, through imaginative representation. Avicenna might conceivably take liberties, but Averroes is a very literal commentator. If he reads it as a text on reasoning, it's because the work really can be read as a text on reasoning. It certainly is the case that this reading require some generalization of Aristotle; but this easily done, since all it requires is that we read Aristotle as discussing general truths through the specific examples provided by Greek culture. Arabic culture is in many ways radically different, but you can see that some of what Aristotle says about tragedy would obviously carry over to Arabic poetry. Further, we have scattered comments from Aristotle himself that encourage this kind of reading, such as his discussion of metaphor, or the famous claim that poetry is more philosophical than history because it is more universal.
In any case, we need rhetoric and poetics both for the full Thomistic account of rationality. We are still missing one part, though. That is, full accounting of rationality requires us to diagnose fundamentally flawed reasoning.
(3c) The third process of reasoning is served by that part of logic that is called sophistics, which Aristotle considers in the book of Elenchuses (= Sophistical Refutations).
And this gives us our complete set, which is, to recap:
- Logic as concerned with understanding: Predicaments or Categories
- Logic as concerned with judgment or propositions: Interpretation
- Logic as concerned with discursive reasoning
- Necessary / Judging
- Formal knowledge: Prior Analytics
- Formal and material knowledge: Posterior Analytics
- Probable / Discovering
- Belief or opinion: Topics or Dialectic
- Suspicion: Rhetoric
- Imaginative representation: Poetics
- Defective: Sophistical Refutations
- Necessary / Judging
The Thomistic account of rationality is thus:
(a) pluralistic: It does not reduce rational thought to one kind of thinking, but recognizes that one may be rational in very diverse ways.
(b) systematic: The pluralism does not lead to a hodge-podge, but is highly structured, both as to its principles, which are the kinds of rational activity themselves, and as to its intrinsic end, which is demonstrative knowledge.
(c) practical as well as theoretical: The rational sciences cover both practical reason and speculative reason; there is no sharp division here, and practice can be as logical as pure reason. This is most obvious in cases like rhetoric or poetics, where the practical is always important, but there are practical and speculative forms of all of them.
(d) an account of inquiry itself: All parts of inquiry have some place in the logical scheme; none are left out. This is a result of the inclusion of constructive and corrective, as well as probative, parts.