ARISTOTLE ON EPISTEME AND NOUS: THE POSTERIOR ANALYTICS

7/27/2019 ARISTOTLE ON EPISTEME AND NOUS: THE POSTERIOR ANALYTICS

1/49

This is a revised and longer version of a paper which appeared inSouthern Journal of Philosop hy (199 8), Vol. 36, No. 1, pp. 15 46.

ARISTOTLE ON EPISTEME ANDNOUS:

THE POSTERIOR ANALYTICS

MURAT AYDEDE

Phi losophy, Humanities Collegiate Divis ion,Committee on the Conceptual Foundations of Science

THE UNIVERSITY OF CHICAGO

Office address:

The University of ChicagoDepartment of Phi losophy

1010 East 59th S treetCh icago, IL 60637, U.S.A .

E-MAIL:[email protected]

ABSTRACT. According to the standard and largely trad it ional interpreta tion,Aristotles conception of nous, at least as it occurs in th e Posterior Analytics, is

geared against a certain set of skeptical worries about the possibility of scien-tific knowledge, and ultimately of the knowledge of Aristotelian first princi-ples. On th is view, Aristotle in troduces nous as an intuitive faculty that graspsthe first principles once and for all as true in such a way that it does not leaveany room for the skeptic to press his skeptica l point any further. Thus the tradi-tional interpretation views Aristotel ian nous as having an internalist justifica-tory role in Aristotel ian epistemology. In contrast, a minority (empiricist) viewthat has emerged recently holds the same internalist justificatory view ofnousbut rejects its internal ly cert if iable infal libil ity by stressing t he connection be-tween nous and Aristotel ian induction. I argue th at both approaches are flawedin that Aristotles project in the Posterior Analytics is not to answer the skeptic oninternalist justificatory grounds, but rather lay out a largely externalist explica-

tion of scientific knowledge, i.e. what scientific knowledge consists in, withoutworrying as to whether we can ever show the skeptic to h is sat isfaction th at wedo ever possess knowledge so defined.


2/49

ARISTOTL E ON EPISTEME AND NOUS: THE POSTERIORANALYTICS 1

It is di ff icult to know wheth er one knows or not.APo 76a26

Whatever the overall drawbacks and difficulties of the Posterior Analyticsare, one thing is quite clear, it is the most integrated and worked out text everwritten by Aristotle on his ph ilosophy of science and th eory of scientif ic knowl-edge in general. And yet, as it st ands, it is often quite perfunctory. It is not sur-prising th at many critics and interpreters ha ve found it di ff icult to penetrate andobscure in deta i ls especially when compared with h is most important works suchas Physics,Metaphysics, and Nicomachean Ethics. At a lmost every stage in the textone is confronted with serious interpret ive problems. There is, however, oneproblem tha t, because of i ts central importance with in Aristotle s overal l systemof thought, has especia l ly occupied and puzzled the students of Aristotle . Theproblem centers around Aristot les account of scientif ic knowledge (episteme) andits conditions, and thus, it can be seen to arise mainly from the general theoreti-cal structure of the Posterior Analytics. I believe th at the problem is largely an in-terpretive one: i t ar ises only when Aristotle is interpreted in a certa in way undercerta in epistemologica l assumptions whose roots are rather modern. Thus myaim in th is paper wil l be to argue for an interpretat ion of the Posterior Analytics for a possible way of seeing Aristotles general programme in it which is, Ihope, free from the putat ive problem. Hence, I wil l a ttempt to dissolve rath erth an solve t he problem.

It is crucial to be very clear about how it has been supposed to arise andthought to be solved by Aristotle. So my first job in th e f irst section wil l be tostate the problem as presented and interpreted by what I broadly take to be th e

orthodox and tradi tional view. I will be meticulous in my reconstruction, sincemy argumenta ti ve stra tegy wil l essentia l ly depend on the framework I wil l pro-vide. So I want the re ader to bear with me in the fi rst and the second sections.The job of Section II is to survey the broadly empiricist, and thus non-orthodox,at tempts to solve the same problem. As wil l become apparent, a lt hough I amvery sympat het ic to these studies, I believe th ey too have problems. I hope toshow that the inadequacies of both the orthodox and empiricist attempts tosolve what they take to be the basic epistemological problem of the PosteriorAna-lytics have the same source, namely, the epistemological assumptions made indiagnosing the a l leged problem. Once th is is clari fi ed, I will show th at t here isno evidence th at Aristot le makes t hose assumptions (al though t hese a re not for-eign to Aristotle ). In fact, I wil l present evidence to the contrary (V) . Sections

IIII V will reconstruct Aristotles anal ysis ofepisteme. Sections VIVII will f irstanalyze and then locate nous in the emerging picture. I wi ll conclude that Aris-totles epistemology in the Posterior Analytics is free from the putative problem.Let us, then, start wit h wh at the problem is alleged to be.


3/49

MURAT AYDEDE 2

- I -

At the beginning of the second chapter of Book I, Aristotle says that oneknows (epistatai) a thing simpliciter when one knows th e appropriate explanat ionof it and knows th at the t hing cannot be otherwise (71b913). The la tte r re-quirement can also be seen to be a restriction on the scope of what can be knownsimpliciter. And given Aristot les insis tence th at th e objects of science are neces-sary facts, the second clause is properly viewed as a clause in the definition ofscientific knowledge (episteme). 1 Aristotle subsequently expands on the first, ex-planatory, requirement on scientif ic knowledge. Explana tion takes the form ofdemonstrat ion whic h is, in turn, syl logistic in character. To scienti f ica ll y knowthat P is to demonstrate P which amounts to explaining P. Wh at makes a syllo-gism demonstrat ion is determined by t he requirements imposed by Aristot le uponthe premises of the syllogism: Aristotle tells us that the premises must be true,primitive, immediate, as well as better known than and naturally prior to, theconclusion. What makes a demonstration truly expla natory of the fact expressedin by the conclusion is th at t he premises conta in th e proper middle term denotingth e cause of the f act. On Aristot les view, th ere is a unique and natural order ofexplanat ion tha t corresponds to the rea l causal order in nature. The premises ofscientific demonstration should reflect this natural order; hence, they must benecessary, universal, naturally prior to, and properly explanatory of its conclu-sion.

When scienti f ic knowledge is def ined in th is way, one problem immediate lyarises. Scient if ic knowledge is knowledge demonstrated from appropria te prem-ises. But how do we know the premises themselves? Unless they are properly

grounded on further premises, they will not count as pieces of scientific knowl-edge, in which case they will not serve as the appropriate bases for any scien-ti f ic knowledge. If t here is no knowledge of the premises, t here is no knowledgeof the conclusion. In short, i t appears th at nothing counts as scient if ic knowledgeif it is not demonstrated from premises th at are t hemselves scientif ica ll y known.But how far can the demonstrat ion go? As Aristotle expl ici tl y rejects circular andinfinitely regressive demonstration (APo I 3), he says that there are certain

1Hence, we have Mures translation of episteme as scientific knowledge. Tredennickfinds this translation a lso useful in many cases. Although I wil l continue to use it from timeto time, I think this translation is in a certain sense misleading: when understood literallyscientific knowledge involves, as we will see shortly, both the knowledge of demonstratedpropositions and the knowledge of indemonstrable propositions that are first principles,

since the body of a given science, according to Aristotle, typical ly consists of these two typesof propositions (cf. van Fraassen (198 0) and Hintikka (1971)). This might explain, perhaps,why Aristotle sometimes uses the phrase indemonstrable episteme to refer to the kind ofknowledge we have of principles. See, e.g., 71b16, 72a 25b2, 72b1724, 76a1 62 1. So,71b9 1 3 does not define scientific knowledge in general but demonstrative scientific knowl-edge.


4/49


proposit ions whose knowledge does not require any further demonstrat ion. Theyare by their very nature necessary, primitive, immediate, and thus self-

explanatory. Al legedly, t hey command belief t hrough themselves, and thus donot require demonstrat ion to convince us of the ir t ruth. These proposit ions arecal led axioms and first principles. 2 According to Aristotle, as the orthodox viewha s i t, t hey are apprehended by the most accurate and infa ll ible faculty of man,nous. So, demonstration of a scienti fic proposition stops when the demonstrat ionha s t he indemonstrable first principles as i ts premises, which we apprehend bynous.

At t h is point, many crit ics ha ve been truly puzzled. The standard reading isth at the problem Aristotle confronted is a fami li ar one from the history of epis-temology. Knowledge requires justif ica tion. But if anyt hing is to confer justifica -tion on a knowledge cla im, it must itsel f be justi fied. How far can the justif ica -tion go if we are to reject circular and infini tely regressive justi ficat ion? Surelywe must have some sort of strong foundation upon which we can base all ourknowledge. Indeed it is not uncommon to attr ibute the h istorica l roots of founda-tional ism as an epistemological doctrine to Aristotle. However, according to th isreading, the problem in the case of Aristotle is most serious. For, in the f irstplace, Aristotles first principles seem to find their contemporary parallels inth e explanatory laws of current scientif ic theories. But these are neit her ana-lytic nor about anyth ing lik e sense-dat a. They are full -f ledged universal em-pirical proposit ions tha t expla in phenomena. Secondly, on th is reading, Aris-totle requires of appropriate epistemic justification that it be demonstrative, orat l east deductive. Th is would have the effect of demanding th at justif ica tionshould be complete , and apparently, certa in.3 Given the nature of Aristotelian

start ing-points and his insistence th at they should be better known, it is h ard tosee how we can ever ha ve knowledge so strict ly def ined.

2Axioms are also first principles in a certain sense. Aristotle sometimes calls them com-mon axioms or common principles (see APo I 10 76 a37 42). They are meant to be logicalprinciples like the principle of non-contradiction or quasi-logical principles like the onewhich states equals subtracted from equals leave equal remainders. It is not clear whetherAristotle really has intended them to function as premises or simply as rules of inference inderiving the conclusion from premises. See Ross (1965) p.56 and p.531; Ba rnes (199 4), p.1389. In this paper throughout I will concern myself not with the common axioms but only withthe proper first principles that are peculiar to each specific science, especially non-mathematical sciences.

3This is indeed how Irwin (1978, 1988) interprets Aristotles account of scientific

knowledge. On his view, Aris totles main concern is with the complete justification ofknowledge claims: ...because [Aristotle] denies that demonstration requires demonstrablefirs t principles... he implies that in some cases complete justification is non-inferentia l... Non-inferentially justified first pr inciples allow us to cla im knowledge without facing an infiniteregress or a circle... Aristotles rejection of the regress relies on his demand for the completeremoval of possible grounds for challenge (1988, p.135).


5/49

MURAT AYDEDE 4

Indeed, Lesher (1973) points out that, when confronted with the danger of in-finite regress, Aristotles first response (72b1920) falls remarkably short of its

intended target when he says th at not al l knowledge is demonstrat ive . ForLesher says that this only indicates that knowledge of the first principles mustbe indemonstrative, not that we actua lly do h ave the knowledge of them. In asimilar ve in, Grene seems to be utterly puzzled and dissatisf ied wit h Ar istotlesmove:

It is, on the face of it , a most unconvincing argument. How do we know there isknowledge of conclusions? Because there is knowledge of premises. How do weknow there is knowledge of premises? There must be, if there is to be knowledge ofconclusions. And if there are two sources of certainty, nous and episteme, and oneof these is the source of inferential certainty, then the other must be the source of thecertainty of the starting-points of inference. But how do we know there are twosuch sources? Presumably because if there were not, there could not be the wholepremise-inference-conclusion structure we have been describ ing. To the student ofKant there is perhaps nothing ala rming here... The existence ofnous seems to behung simply and solely on the demand that, to have Aristotelian science, we musthave it, and so we do. (1963, p.111)

The main idea that needs to be emphasized here is that scientific knowledge asdefined by Aristotle requires epistemic certainty , so tha t nothing less th an an in-fa ll ible mental f aculty tha t would grasp the principles as true and certain woulddo for Aristotles rigid account of knowledge.

In fact, t here is another aspect of the problem whic h wi ll bring the puzzl ingissue more sharply into focus. Aristot le repeated ly te l ls us tha t we acquireknowledge of fi rst pr inciples through an inductive process (epagoge) star ting from

sense-perception. He even somewha t describes th e process in the notorious lastchapter of the Posterior Analytics.4 However, immediately after saying Clearlythen it must be by induction that we acquire the knowledge of the principles...(100b3), he introduces in the famous passage (100b517), as the orthodox inter-pretation has it, the most accurate and infallible intellectual faculty of man,nous, by wh ich, Aristotle says, we grasp, once and for al l , t he ultimate and neces-sary first principles as such.5

The apparent puzzle this invi tes, as interpreted traditional ly, is th is. If weacquire the knowledge of first principles through induction, one wonders how onecan possibly be so certain, epistemologica ll y, th at one has scienti fic knowledgethat P (when one does). Induction has been well known to lack such a power tosecure epistemic certainty, as Aristotle himself seems to indicate in many

4APo II 19. A parallel account is also given in the beginning ofMet I 1.5Compare also the same apparently puzzling order of Chapters 3 and 6 in Book VI of the

Nicomachean Ethics. In Ch.3 Aristotle says that we acquire the universal principles by in-duction; in Ch.6 he argues that we have only the nous of them.


6/49


places.6 On the other hand, if we have this allegedly infallible faculty, nous,with which to grasp the principles, why should we need induction at all to ac-

quire knowledge of them? And, conversely, if it is induction (however problem-at ic th is might be) t ha t gets us to the fi rst principles, why should we need th isapparently mysterious faculty , nous, t hat, given the Pl atonic tradi tion, seems tooperate independently of sense-perception in its grasping power?7

The tradi tional and to a certa in extent sti l l orthodox way of making sense ofth is apparently puzzling situation is, roughl y, t hi s. There is no doubt th at Aris-totle views induction as indispensable in the path toward the first principles.But given h is strong conditions for scient if ic knowledge and also inductions noto-rious fra ilty in guarantee ing epistemic certainty, induction seems obviously insuf-fic ient to establ ish first principles as knowledge. So it is nous as the infallibleintuitive faculty that gives the strong epistemic warrant to the principles thatth ey require in Aristotles conception. By intuit ive ly and immediatel y graspingth e princip les as true and necessary, nous va lid ates and establishes the results ofinduction as knowledge. In other words, the orthodox view is tha t Aristotle inthe Posterior Analytics defines scienti f ic knowledge so strongly and rigidly th a t,at th e end, seeing tha t normal inductive procedures wil l of necessity fa l l short ofguaranteeing it, he is forced to postulate some sort of infallible and intuitivemental faculty, nous, that would epistemica lly secure the possibi lity that wehave scientific knowledge so defined.8

Although the main idea of the orthodox interpretation is clear enough, asLesher (1973) points out, there are interpretive nuances among scholars abouthow the inductive process is related to the immedia te noetic grasp. Ross (1965),for instance, seems to think that induction has no justificatory role in our knowl-

edge of the principles; al l the justif icat ion th at principles enjoy comes from ourpeculiar noetic grasp of them:

6E.g., APo II 5 91b151 6, b356; 7 92a38b3. Also, cf. APr II 23. See below.7I should emphasize that many of the traditional scholars talk about nous as a faculty or

capacity. Although this is generally true for the De Anima and some other scientific treatisesof Aristotle, in the Posterior Analytics, nous is introduced by Aristotle as a hexis, a mentalstate reached through the realization of a capacity or at the end of a mental process. See APo II 19 10 0b5ff. and below.

8Indeed, traditionally, intuitive intellect, intuitive reason, intellectual intuition, ormerely intuition were the standard translations of nous. Barnes (1994), however, trans-

lates it as comprehension. Kahn, in his characterization of the traditional interpretation,nicely summarizes the situation thus: ...in that perspective [Aristotles] emphasis on the in-dispensable starting-point in sense perception seems to ally him with the empiricists, whereasthe ultimate appeal to nous then takes on the air of a last-minute betrayal, a sellout to the ra-tionalists particularly if nous is understood as an infallible intuition of self-evidenttruths. (1981, p.386)


7/49

MURAT AYDEDE 6

The induction here is not proof of the principle, but the psychological preparationupon which the knowledge of the principle supervenes. The knowledge of the prin-

ciple is not produced by reasoning but achieved by direct insight... This is in factwhat modern logicians ca ll intuitive induction... The general principle, in such acase, being capable of being known directly on its o wn merits, the particular exam-ples serve merely to direct our attention to the general principle... (p.49)9

On the other hand, it is certainly possible to view nous as involved both in(mechanical ly) get ting the general inductive conclusion from the particulars andin epistemically validating it at the same time by somehow intuitively appre-hending it as true. That is, nous can be seen both as th e (only) justif ica tory sourceand as the psychological mechanism helping the induction from particulars.This seems to be the view held , for instance, by Lee,10 Irwin,11 Grene,12 and in cer-ta in passages, at le ast, by Ross.13

Wha tever t he importance of these deta il s of interpreta tion is, th ey bring outone thing most clear ly: it is nous, and only nous, that epistemical ly justi f ies t heprinciples in the sense required. Now it is important to be clear about th is re-

9But cf. Ross (1945), p.217 . A.E. Taylor a lso seems to be of the same opinion. He clearly

distinguishes between two quite different questions: (1) how we come to reflect on the [prin-ciples], (2) what evidence there is for their truth (1955, p.37). To the first question he re-plies by induction from experience and characterizes the process as simply psychological.But when the induction has done its work in call ing attention to the principle, you have tosee for yourself that the principle is true... by immediate inspection, just as in sense-perceptionyou have to see for yoursel f that the colour before your eyes is red or blue. (p.38) The com-parison of the noetic grasp of principles with our conscious apprehension of sensible quali-ties is indeed striking, and a recurrent theme in the literature in describing the nature of

noetic grasp. We will see its import shortly. The same distinction is also dra wn by Le Blond(1939): ...au sujet de la connaissance des principes, il nous semble que les diffrents essais derponse dAristote peuvent, en gros, se grouper en deux sries: rponses de droit, dun ct,expos de ce que devrait tre la sais ie des principes, pour justifier pleinement la science... Larponse de fait, [dun autre ct,] lessai dexplication de la saisie des principes, de la con-naissance des universels et de la construction des dfinitions est tout autre... (pp.145-6). ButLe Blond nevertheless thinks that in II 19 on reconnait seulement un vigoureux effort de vo-lont pour combler le foss qui spare les expriences psychologiques sur la combinaison desimages et lapprhension infail lible des principes requis pour assurer la science. Mais cestpar un vritable saut qui demeure injustifi, quAristote passe dun point de vue lautre(p.138). See also Grote (187 2): By referring the principia to Intellect, [Aristotle] does notintend to indicate their generating source, but their evidentiary value and dignity when gen-erated and matured (p.293).

10 Lee (1935), see especially p.122.11 Irwin (19 88), see p.135 and n.3. However, cf. his (197 8), pp.214 5.12 Grene (1963): The refinement of perception to make explicit the universal in the indi-

vidual, the species in the specimen: this is the experience that underlies Aristotles confidenceabout nous (p.112).

13 See, for instance, Ross (1945), p.217.


8/49


quired sense. So le t us try to sort out wh at must be involved, according to the out-lined orthodox interpretation, in ones noetic grasp of a certa in principle.

When nous grasps a certain principle, say P, as true, the subject, S, w ho hasthe nous ofP, is supposed to be immedia te ly and intuit ively aware of its trut h.The force of immediate ly here is reasonably clear: Ss awareness of the truth ofP does not depend in any way on her awareness of the truth of any further propo-sition. Her awareness is supposed to be non-inferential, just as, say, ones aware-ness that one is appeared to redly (as the current jargon has it) can be consideredto be non-inferential.

The force of intuiti vely, however, is somewhat mysterious. Wh en used inrelat ion to nous, its force, as I ta ke it, is to bring out th e nature of the infal lib leawareness one has wi th respect to the truth ofP. How can one be infal l ible aboutthe truth ofP when i ts truth-condit ions fa ll outside of ones ken and especia l lywhen P is a universal (empirical ) proposit ion? In the case of infal l ible awarenessof sensible qualities, there is at least a conceivable story to be told, as is wellknown: on one version of it, the truth-conditions of I am appeared to redly aredirectly and immediately accessible to my consciousness, and the nature of thisdirect access is such t hat I am appeared to redly is (becomes) true when and onlywhen such an access is established, which, in turn, guarantees my awareness.Now since Aristotle explic it ly rejects the existence of Platonic innate ideas14 andseems well aware that in any inductive jump the conclusion goes beyond the ini-tial data,15 the infallible noetic grasp of the truth of principles becomes totallymysterious. Nevert heless, the orthodox view has it t ha t however mysterious itmay be Aristot le just postulates some such faculty .16

What is more important, however, is that the nature of this infallible

awareness has been tradit ional ly interpreted in such a way th at wh enS has thenous ofP, S not only knows that P as knowledge is standardly understood atleast as justified true belief17 but also S immediately knows that she knows

14 See, e.g., APo II 19 99b273 6 andMet I 9passim .15 APo II 5 91b1516. Cf. APr II 23, see also Engberg-Pedersens persuasive discussion

of this chapter (1979): he argues that Aristotles concern there is not what Ross has thoughtit to be, namely, perfect induction. Cf. also Hamlyn (197 6), Hintikka (1980), and Kal (198 8).

16 Calling nous an intuitive faculty, to my mind, has no clear and unproblematic sense inthis context apart from saying that nous is just such a faculty that infallibly and immediatelyapprehends once and for all the truth of certain empirical and universal propositions with-out any justificatory recourse to perception. This interpretation is proba bly the result of theinfluence of rationalis t tradition. But even rationalis ts with their nativism are better situ-

ated in giving sense to their intuition. Ross likens nous to the intuition involved in what hecalls the notion of intuitive induction of modern logicians. But this can only be an analogy,and perhaps true at most for Aristotelian mathematical sciences. On the other hand, it seemsto me that many traditional scholars, when characteriz ing Aristotelian nous as having suchan intuitive power, had in mind the Active Intellect of the De Anima III 5.

17 It is easy to see how nous as understood traditionally entails knowledge as justified


9/49

MURAT AYDEDE 8

that P. In other words, one of the immediate concequences of interpret ingnous asdirect and infallible awareness was the committment to a version of the K-K

Thesis.18 So we have :

(i) [Snoei that P (S knows that P & S believes th at S knows that P & Ss be-lief that S knows that P is always true)].

Here the point to emphasize clearly is th at the nature of the non-inferentia l andintuitive justificatory grounds for Ss knowledge that P is such th at they also si-multaneously constitute the conclusive grounds for Ss belief tha t S knows that Pthat are transparent, i.e. directly accessible, to S. Notice tha t on th is interpre-tat ion of Aristoteli an nous any skeptica l cha ll enge is supposed to be completel yblocked when Snoei that P and claims to know that P.19 This is supposed to be oneof those cases where claims to knowledge are not open to any sort of skepticalchallenge, just as, for instance, Descartes clear and distinct ideas and knowledgecla ims about ones own immedia te p henomenal exper ience were thought not opento skeptical cha ll enges.

So much for the truth of pr inciples. But, of course, according to the orthodoxinterpretation, ifnous is to serve i ts purpose in establ ish ing the ultimate and in-fallible foundations of scientific knowledge (and thereby in showing the possi-bil it y of scient if ic knowledge), it is not enough for one to infal l ibly grasp simp lyas true a certain proposition P which happens to be a principle: in addition, oneshould infa ll ibly grasp, or be aware of, P as a principle, or more accurately, as theappropria te principle . And th is involves grasping P as necessary, primitive,immediate , and explanatori ly appropria te. For unless we know the pr inciples as

such, we cannot claim to scienti f ica l ly know anything. But how do we know theprinciples as such? The orthodox view is th at Aris totles answer here is againby means ofnous. Irwin, for example, is very expl ici t about th is: 20

true belief (JTB). Suppose Snoei that P, then S has complete and non-inferential justificationfor her belief that P that objectively guarantees the truth ofP. Then it trivially follows thatthe three standard clauses of JTB analysis of knowledge are satisfied for S. In other words,necessari ly, ifSnoei that P, S knows (JTB) that P.

18 See Hintikka (1962).19 On this point, Irwins interpretation is quite explicit. According to Aristotle, he says,

we must be aware of being better justified in believing [the principles] than in believing any-thing else; that is what makes them best known, and removes any objective ground for chal-lenging them (198 8, p.135).

20 See also his (197 8), where he wri tes: How do the methods for reaching first principlesyield first principles with the right status? Though first principles are not known, becausenot demonstrated, they cannot merely be matters of belief; we must grasp their truth and ne-cessity from themselves apart from their inferentia l relations to other truths... The cognitivecondition in which we grasp all this is intuitive intellect... But how do we reach thisstate? We approach, but do not reach it, by ordinary inductive methods. Observation pro-


10/49


The knower must grasp self-evident principles as such; for if they are grasped non-

inferentially, without any further justification, they must be grasped as true andnecessary when considered in themselves, with no reference to anything else. If firstprinciples are to meet all Aristotles conditions, they must be grasped by intuitionthat certifies that they have the relevant properties. (1988, p.134)

The essential point to bear in mind is that we must know for certain that theprinciples ha ve t he relevant properties i f we are to know for certain tha t wehavescientific knowledge. So, it is thought tha t nous is required as an intuitive andinfallible grasping, or certifying, faculty, because any other sort of justificationfor the belief in the proposition that the principles are necessary, primitive,immediate , and explanatorily appropria te (ca l l th is compound proposit ion R)would ha ve to be discursive, in which case we are again faced wit h the danger of

circular ity or infinite regress. Or, worse sti l l, it is thought tha t since Aristotle isafter showing the possibility of scientific knowledge as he defines it, he needscertainty, hence complete justificat ion; an inferential justification for R on induc-ti ve grounds would not do for him, because it would of necessi ty f al l short of epis-temically guaranteeing the truth ofR. So, as the orthodox reading has it, weneed to know R in a way th at when we do we also h ave to know for sure tha t wedo, and this is supposed to be exactly what the nature of our noetic grasp allowsus to do. So,

(i i) [Snoei that R (S knows th at R & S believes tha t S knows tha t R & Ssbel ief that S knows tha t R is always true)].

Here, again, the noetic nature of the non-inferential and intuitive justificatorygrounds for Ss knowledge that R is such t ha t th ey also simultaneously constitutethe conclusive grounds for Ss belief th at S knows tha t R th at a re transparent, i.e .directly accessible, to S.

Such is, t hen, t he tradi tional and orthodox reading of Aris totles conceptionofepisteme and nous. I find th is read ing unconvincing and contrived, especia l lywith respect to nous. Too much is demanded from nous; it is totally unclear thatAristotle real ly intended nous to do so much epistemological work. Nowhere inh is re levant writ ings, for instance, does Aristotle assign to nous the job of grasp-ing the principles as necessary, primitive, immediate, or explanatorily appro-priate, let alone grasping them intuitively and infallibly as such.21 It is indeed

duces only beliefs... A first principle is not grasped by intuitive intellect just in so far as i tresults from induction... What else is required? (pp.214 15). This something else, Irwinsubmits, is nous.

21 In the Posterior Analytics, for instance, only in seven pla ces up to the last chapter doesAristotle explicitly mention nous or make tacit reference to it: 71b16, 72a35, 72b25, 83b34,85a 2, 88a8, 89b8. And in II 19, it is introduced only in the last paragraph (100b5ff.) where it


11/49

MURAT AYDEDE 10

very interesting to see that such a heavy and demanding interpretation could bebased on such scant, str ictly spea king, non-existent, textual ev idence. Why is

th en such a reading forced upon Aristot le? Part of the reason is clear from what Ihave been saying so far about the traditional view of Aristotles epistemology.What underlies and guides this characterization of Aristotles epistemology,however, h as much to do wit h the interpreters specif ic understanding of wha tth e essentia l task of epistemology is: namely, to successfully defend ones knowl-edge claims against the skeptic, to show that knowledge is possible.22

So, according to this view, one of the main projects Aristotle sets for himselfin the Posterior Analytics is to answer the skeptic who ch allenges the very possi-bil it y of scient if ic knowledge. But this is not a ll. The sort of skeptic that thetradi tional scho lars have in mind is the one th at is known to be very stubborn andthus very hard to satisfy, who requires that successful defence of ones knowledgeclaims be a necessary condition on showing th at knowledge is possible. Butknowledge requires truth. Hence, the skeptic is unlikely to be sat isf ied wit h adefence tha t fa l ls short of establ ish ing the truth of ones cla im to knowledge. Sosuccessful defence in th e eyes of t he skept ic amounts toconclusive defence of onesknowledge cla im. Hence we ha ve, for instance, Irwins insistence tha t Aris totleis after conclusive and complete epistemic justification of claims to scientificknowledge (see the quotes from him above).

So, i t is not dif ficult to understand the ma in rationale beh ind assigning to Ar-istotles nous such a h ea vy ep istemic burden if one star ts wit h some such assump-tions about what the basic task of an epistemological enterprise should be like.For Aristotles strict account of scientific knowledge does not only require truthbut also t he non-demonstra tive knowledge of the appropriate first princip les qua

principles, which involves the knowledge that certain universal propositionshave certain characteristics like necessity, immediacy, primitiveness, explana-toriness. So the orthodox idea is th at since Aristotle saw th at inductive evi -

is expl icit ly mentioned four times. In none of these passages, does Aristotle even come close tosaying that it is nous that intuitively and infallibly grasps the principles as having the re-quired properties. All he says, where he is explicit, is that it is nous that apprehends theprinciples (archai, 1 00b5ff.). In 8 8b3 689 a1, he says, nous is the starting-point ofepisteme,and non-demonstrative episteme is the belief in an immediate premise. Here, however, theredoesnt seem to be any implication that the non-demonstrative episteme is the belief in an im-mediate premise qua immediate. See Lesher (197 3).

22 This understanding is again traditional in a broad sense. Take, for instance, the fol-lowing remarks: There is a certa in disingenuousness in Aris totles rather smug solution.

To affirm that not all knowledge is demonstrable is to reject what he has given every show ofaffirming in the precedent chapter of this very treat ise. Worst of a ll, the necessity which hesays compels us to assert that knowledge of immediate premises is independent of demonstra-tion is itself a consequence of the assumption that knowledge in the strict sense is possible which is the very point at issue G.R. Morrow (1 970 , p.333). For somewhat similarly spir-ited remarks, see also the quotation from Grene (1963) above.


12/49


dence would not sat isfy the skeptic , he ult imate ly made recourse to an intuit iveand infallible faculty, nous, by means of wh ich we not only infall ibly know, but

also infallibly know when and what we know.

- II -

However, even this solution to the epistemological problem Aristotle iscla imed to have faced has been a constant source of embarrassment on the par t ofinterpreters especia ll y in the lig ht of Aristotles known naturali sm and his owninsistence on the empirical nature of a proper scientific inquiry that are well-at tested even by the same tradi tional scholars themselves.23

Indeed, the intuition that there is something wrong with the traditional in-terpretation of Aristotles nous has also been at the source of some of the recentworks especially on the Posterior Analytics. In the last twenty yea rs or so, asworks on Aristotles scientific treatises and generally on his actual methods inempirical research have grown, more and more scholars have come to doubt theaccuracy of the orthodox view of Aristotles epistemology as laid down in thePosterior Analytics.24 Apart from the implications of Aristotles own practice inhis scientific inquiries, there is also his own insistence on the importance of em-pirical data in testing theories, in other words, his own epistemological use ofphainomena.25

23 Couloubaritsis (1980), for instance, after having outlined the traditional interpreta-tion of Aristotles solution, nicely summarizes the embarrassment and its source as fol-lows: Faut-il se contenter dun tel embarras et clore le dbat? ...Dans ces conditions, peut-

on croire qu[Aristote] sest content lui-mme dun pis-aller ou dune sorte de deux ex ma-china pour surmonter son embarras? Ou bien plutt lembarras ne serai t-il pas de notrect, parce que nous ne voyons pas encore trs clairement quelles sont ses conceptions de laconnaissance des principes et de lintellect? En tout cas, il nous parat a ssez tonnant que leStagirite ai t pu assumer une ligne dintervalle pareille contradiction et quil a it ainsi conula connaissance des principes comme lobject de la sensation, par une activit inductive, et enmme temps comme objet dune intuition infaillible par lintellect cest--dire par cette ac-tivit qui constitue en quelque sorte lopposde la sensation. Nous avons l impression que lefil de largumentation dAristote en cet endroit nous chappe (pp.44647).

24 In this connection, I should cite the influential papers by Kosman (1973), Lesher(197 3), Burnyeat (1981), Bolton (1987). Also Barnes commentaries (1994) in general are inthis vein the first edition had appeared in 1975. For important studies on Aristotlesmethods in biological research emphasizing his empiricism and naturalism, see Gotthelf andLennox (1987).

25 There are many passages in Aristotelian corpus that are indeed surprisingly close tomodern scientific methodology. To cite a few, see De Generatione Animalium III 10 7 60b2833; Historia I 6 491a1015; De Partibus I 1 640a1318; APr I 30 46a1727; De Caelo III 7306a618; Ethica Nicomachea I 78 1098b111; De Generatione et Corruptione I 2 316a512,8 325a1 324. To some of these passages we will return in due course. That Aristotle alsodid quite a lot of experimentation has been well known: for a documentation of the extensive


13/49

MURAT AYDEDE 12

In ligh t of th is, many schola rs ha ve been led to re-examine the Posterior Ana-lytics in a much deeper and more detai led manner. Alth ough I am very sympa-

thetic to these studies, I believe, they encounter one common problem that arisesfrom certain well-known passages especially in the last chapter of the PosteriorAnalytics, as we wi ll see short ly . But more important ly, the source of the diff i-culty they face can again be traced back to certain epistemological assumptionsmade with respect to Aristotles main programme in the Posterior Analytics. I twill be important for what I am going to say later on to see how the difficultyarises and wha t its rela tion to the orthodox and tradit ional view is. So let usbriefly examine these recent studies in more deta i l.

The common theme in these studies is to examine in some detail those pas-sages where Aristotle discusses the role of the senses and phenomena in generalin the inductive process in reaching the universal and explanatory principles,and then to find a naturalistic role for nous to play . For instance, Lesher, in hisinfluentia l article , af ter ha ving briefly summarized the tradi tional v iew, char-acterizes h is aim as follows:

Too much is known about... the relations between perception, induction, and uni-versal principles to summarily dismiss Aristotles account for want of cogency; andthe terms intuition and intellectual intuition have too varied a history to bethought of as simple equivalents ofnous or noesis. It is my contention that a more de-tailed examination of nous, noein and related notions supports a rather differentunderstanding of their significance and of Aristotles account of our knowledge offirst principles. ...[B]y pointing out the ways in which nous relates to aisthesis, epa-

goge, and kathalou principles, I hope to show that nous is not properly thought of asintuition or intellectual intuition, at least in any sense of these terms which wouldforce us to distinguish nous from ordinary empirical knowledge. (1973, p.45)

The last sentence here is somewhat surprising. If empirical knowledge is to beconstrued as factual knowledge, i.e. as knowledge of the world interpreted real-istically, then there is no doubt that Aristotles first principles are empirical inth is sense. So Lesher must have had someth ing else in mind. Indeed, as his art i-cle makes clear, the contrast here is between the knowledge of the world ob-ta ined through ordinary empirica l methods, i .e. on the basis of perception, andfactual knowledge obtained a priori, i .e . wi thout any recourse to percept ion.However, even in th is contrast th ere is a certa in ambiguity. For again t he ortho-dox view does not deny that perception plays a necessary role in obtainingknowledge of the principles.26 The question is whether this role has any justifi-catory bearing on our grasp of f irst principles. So wha t Lesher must be rea ll y say -ing is that what turns our belief in first principles into knowledge proper is notsome a priori act of intuit ion tha t operates independently of perception, as the or-range of the experiments and their nature, see Ross (1945), pp.11214.

26 Indeed, Aristotle is very explici t on this point. See APo I 18; De Anima III 8 432a310.


14/49


thodox view has i t, but rath er t he ordinary empirica l and inductive justif icatorygrounds that are also involved in the very obtaining of those beliefs.

Lesher argues for th is conclusion in a rath er sweeping and indirect way. Hismain strategy is to single out the bases for rejecting the traditional idea thatnous is operati ve only in grasping the fi rst principles. For th is reason, he turns hisfocus to those passages where Aris totle seems to describe the process by whic h wereach t he universal in general and, in particular cases, the first principles. Byindicating the obviously empirical character of this process toward the firstprinciples, and, in the light of this, by showing how an alternative interpreta-tion is possible of certain passages traditionally interpreted to the contrary,Lesher concludes that nous is what we have of the universal empirical proposi-tion at every stage whenever induction takes us from wha t is prior relat ive to usto what is prior by nature, whether this proposition be a first principle or not.From this he immediately infers that speaking of acts ofnoesis is just a way ofdescribing how we inductively reach universal conclusions that exceed their ini-ti al data . In other words, he concludes tha t [t]he account ofnous of first princi-ples wh ich concludes the Posterior Analytics is therefore neither ad hoc nor incon-sistent wi th Aristot les empiric ism; on the contrary, i t is a consequence of it .27

Let us see what Aristotle says in the concluding paragraph:

Since of the intellectual states [peri ten dianoian hexeon] by which we grasp truthsome are always true and some admit falsehood (e.g. doxa and logismos whereasepisteme and nous are always true), and no kind other than nous is more a ccurate[akribesteron] than episteme, and the principles of demonstrations are better known[gnorimoterai], and all episteme involves an account there will not be episteme ofthe principles; and since it is not possible for anything to be truer [alethesteron] than

episteme, except nous, there wil l be nous of the principles both if we inquire fromthese facts and because demonstration is not a principle of demonstration so thatepisteme is not of episteme either so if we have no other true[alethes] kind apart from episteme, nous will be the principle ofepisteme. And theprinciple will be of the principle, and as a whole will be similarly re-lated to the whole object. (APo II 19 100b517)28

Now, as far as the epistemological problems in the Posterior Analytics are con-cerned, this passage is the most important passage on which the orthodox ac-count of nous, understood as an infal l ible, certain and most accurate faculty , isbased. (Notice , however, Aristot le is not ta lk ing about a f aculty here. ) For t heeliminative argument involved here to the effect that it is nous by which we

27 Lesher (1973), p.65.28 Throughout in the paper I used various translations most often making some slight

modifications. Barnes (19 75, 1994) and Tredennicks (Loeb Edition) translations, however,are the basis of almost all my quotations from the Posterior Analytics. For the rest, I oftenused the Oxford translation.


15/49

MURAT AYDEDE 14

grasp first principles or that we have the nous, and not the episteme, of the f irstprinciples to work, at least the following should be true:

(i i i) (Snoei that PP is true),

which seems to be the basis of attributing infallibility to nous. Aristotle alsosays that nous is truer and more accurate than episteme. Now it is also true th at

(iv) (Sepistatai that PP is true),

and since episteme is demonstrated knowledge whereas first principles are inde-monstrable and onlynous and episteme are always true, it is nous by which we ap-prehend first principles.

As Barnes points out, th e e liminati ve argument can actually be f il led out inmore th an one way.29 But the basic idea is clear: what we have of the first prin-ciples is nous, which is always true. Th is passage has been the main supportingground for the orthodox view,30 and many have thought t hat it constitutes a ma-jor block for the Lesher- type non-orthodox broadly empir icist interpreta tions.But, thi s being so, wha t is most curious is th at Lesher is the only writer I know ofwho elaborately addresses this difficulty while arguing for a non-orthodox in-terpretation ofnous:31

29 Barnes (1994), p.270. For a similar eliminative argument to the same effect, see EthicaNicomachea VI 6. Also cf. APo I 33 89a13.

30 However, notice also that there is nothing in this passage that supports (ii). Also, al -

though (i) entails (iii), the converse is not true.31 Engberg-Pedersen (1979) says with respect to II 19 that: ...nous, in addition to denot-

ing thefaculty that is responsible for the universal principles being seen as a result of epa-gogic attention being given to particular cases, is now stated to denote the state that obtainswhen the principle is seen. ...Epagoge is attending to particular cases with the consequencethat a universal point is seen, for which the faculty ofnous taken as a generalizing ability isresponsible... In Aristotle epagoge raises no question as to the certainty of the universalproposition... nor does he wish to introduce nous as an abil ity that guarantees the truth of theuniversal proposition (this much for one traditiona l interpretation ofPosterior Analytics II19) (pp.3171 8). But Engberg-Pedersen, somewhat surpris ingly, says nothing about the lastparagraph of II 19. He does not even discuss how it can be made to fit into his general ac-count. This is very curious especially given that that paragraph has traditionally been takento be the main support against the kind of interpretation he is pushing for.

Kosman, in his quite insightful paper (1973), argues that nous is to be properly viewedas the insight we gain when we see and understand certain universal propositions in thelight of the explanatory capacity they have with respect to phenomena which makes us graspthe propositions as the principles. Nous is just the kind of ordinary insight when we under-stand the scientific explanatory power of certain universal propositions, and we come to thisunderstanding by ordinary inductive and empirical means, i.e. while we are actually doingempirical science. However, Kosman explicit ly declines to discuss II 19: ...there are many


16/49


But some difficulties remain. In the final pa ragraph of II, 19, it i s argued that it i snous (and nothing else) which we possess of first principles, but there are some fea-

tures of Aristotles argument which make it hard to reconcile with the empiricism ofthe preceding genetic account. Nous is characterized as being akribesteron than anyother kind of knowledge... and even alethesteron than episteme, and these quali tiesof unsurpassed accuracy and infallibility have been seen as the marks of a fac-ulty of intuition... (1973, pp.623)

Subsequently, Lesher gives a series of arguments intended to show that Aristotledid not mean in that passage that nous is the most accurate and infallible stateof mind as these epith ets a re understood in the ir ordinary senses. In brief , Lesherseems to reject (ii i) . His al ternati ve interpreta tion is not actual ly wit hout force,and I tend to believe th at it is in fact true. But it fal ls short of establ ish ing hismain point.32 If t here is any argument in Aristotles passage, then it requires (ii i )wit hout which i t is not clear what to make out of th at very passage.33

things I dont understand about that chapter and its relation to the rest of the Analytics. Sowith many other questions which Ive left unanswered, only a part of the reason for whichhas been the intentionally broad scope of my discussion (p.391).

Burnyeats (1981) line of argument against the orthodox view is very similar to Kos-mans. He thinks that episteme is to be understood not as scientific knowledge but as scien-tific understanding that comes with the nomological explanation of the phenomena. With re-spect to II 19, he writes: A faculty of intuitive discovery is not needed because discovering...first principles is a matter for induction... Aristotle sees no Humean problem about a leapfrom inductive evidence to knowledge. ...Hence, as he sees the problem of our grasp of firstprinciples, the difficulty is not a lack of evidence to transform inductive belief into certainknowledge. That inductive belief is already knowledge. What it is not yet is understanding

and that kind ofgnosis which goes with understanding. To acquire this a t the level of firs tprinciples what we need is greater familiarity, perhaps some more dialectical practice; inshort, intellectual habituation (p.131). This habituation, he says, finds its home in Aris-totles conception ofnous. But this sounds like we should make ourselves believe that we gotwhat we want. In any case, this wil l not do. Aristotle seems clear about the nature of induc-tive evidence: [In division] it never becomes necessary that the thing is necessary when suchand such holds, but [division] proves in the way in which a person who does an inductionproves (APo II 5 91b151 7). See also APo II 5 91b3536; APo II 7 92a38b3. Cf. APo I 473a323 4; I 24 86a11 14; I 31 87b3035; Top V 3 131b193 9 and VII 5. In this connection,Engberg-Pedersens (1979 ) discussion of Aristotelian induction is very helpful. Aristotle isalso very cautious about hasty scientific generalizations. In De Generatione Animalium III 5756a26, for instance, he says: The cause of the ignorance of those who take that view isthat, while the differences among animals with regard to copulation and procreation aremanifold and obvious, those people observe a few cases and think that things must be the

same in all cases. For similar remarks, see also Physica VIII 1 252a22b4.32 Furthermore, his suggestions about how to properly understand the alleged character-

istics ofnous are not incompatible with the truth of (iii).33 Perhaps this is the reason why Lesher seems also to reject, tacitly, a reading of the pas-

sage as contain ing an eliminative argument, see p.64, n.54. But this is unconvincing, we havea parallel of the argument in Ethica Nicomachea VI 6 that obviously requires (iii ). Cf. also


17/49

MURAT AYDEDE 16

Barnes interpretat ion of II 19 is somewha t simi lar t o Leshers but less clearand rath er brusque. According to Barnes, Aristot le makes a verbal point aboutnous at 100b5ff. in response to one of two questions he ra ises in the beginning of I I19: how do the principles become known and what is the state that knows them?(99b179) Aris totles answer to th e first question, according to Barnes, is bymeans of induction, as given in the genetic account of II 19 (99b30100b4). Andth e whole purpose of the l ast paragraph of I I 19 is just to answer the second ques-tion and noth ing more. Just as episteme can be said to be the state we arrive atthrough demonstrative syllogism, nous is just the sta te we are in a t t he end of aninductive process. Barnes says, Nous has no phi losoph ica l importance in APst.(1994, p.270). But th is is, to say the leas t, puzzling. Ifnous is the end state of aninductive process, then the propositional content of the end state one is in anytime one happens to engage in an inductive process will be true, since this endstate is claimed to be nous, and thus, is, by (i i i) , a lways true. But not every endstate of an inductive process is a state whose content is true. Induction is a fall i -ble process.34 Barnes does not say much on the issue even though he gives t he im-pression tha t he cla ims to have solved the mystery ofnous in APo II 19.

To recap: why is an empiricist interpretat ion ofnous as t he ordinary grasp ofth e deliverances as t he end state of induction supposed to be somehow prob-lemat ic? The reason, I submit, is th is: Aristot le is explici t about (i i i) , but howcan anyth ing remotely li ke ( i ii ) be true given the notorious fra il ty of induction?Ifnous ofP is wha t we have w hen we infer or justify (or, somehow arrive at ) P onthe basis of inductive and empirical evidence, then there is no epistemic guaran-tee that P is true, no matter how good our evidence is; hence, there is apparent lyno necessity th at w hen someone noei that P, P is true ifnous is related to induc-

tion in the way Lesher and many others ha ve thought . So, the traditional em-barrassment seems to be sti l l with us.The traditional and orthodox view has put the emphasis on the nous as an

infallible intuitive faculty and tried to solve the epistemological problemsraised in the Posterior Analytics, as it understood th em, by means of this nous .This was overkil l. Indeed, what is known about Aristotles empiric ism and hi sscientific pract ice makes it very hard to swal low such an interpretat ion ofnous,

APo I 33 89a13.34 This a rgument works better against Lesher since he is, unlike Barnes, very clear about

the relation between nous and Aristotelian induction: The relation between nous and epa-goge turns out to be a typically Aristotelian one: there is one activity, grasping the universal

principle, but it admits of various descriptions; to speak of it as an act ofnoesis is to give anepistemological characterization, while to characterize it as epagoge is to speak of method-ology. This account of nous and epagoge coincides with Aristotles view that experienceprovides us with principles which we then endeavor to structure within syllogistic form, andit makes perfectly good sense ofnous as the source of scientific knowledge since it is nouswhich supplies us in general w ith such principles (1973, p.58).


18/49


especia ll y wi th respect to the kind of knowledge of the first principles requiredfor episteme. The non-orthodox view, on the oth er h and, has emphasized t he em-

piricism of Aristotle, and viewed him as almost a thorough-going empiricist.35But, as we ha ve seen, it a lso faces dif ficult ies. I agree with Couloubarits is (1980)th at Aristotles th read of thought seems to have somehow escaped us so far. Sowha t to do?

In fact, I tend to believe that the non-orthodox interpretation of Aristotlesepistemology is very close to truth , if any such there be. Not th at I view Aris-totle as an empiricist in, say, a Lockean or Humean sense; but I think, Aristotlehas adopted a scientific methodology whic h is empiri cist in a broad sense. Butwha t about the general epistemological problems th at seem to be raised by wha tAristotle says about nous?

Why exactly can we not grant (iii) to Aristotle while holding at the sametime that the only evidence we can have for our knowledge of the f irst principlesis empirical and inductive wi thout postulat ing an intuitive anda priori justifyingfaculty? Now it is true that inductive ev idence by its very nature fa ll s short ofestabl ish ing the truth of our beliefs. But our inductive bel iefs are sometimesknowledge, only that we cannot always correctly tell when our beliefs are knowl-edge. In other words, we cannot necessari ly know for sure when (and general lywhat ) we know when our al l and only evidence is inductive . To put it sl igh t lydif ferently, we may not necessari ly be able to conclusively defend our inductivelygrounded knowledge claims when cha l lenged. But why should a conclusive defenceof our knowledge cla ims be a necessary condit ion on the possibili ty of our ha vingknowledge? Such a demand has been made by the trad it ional and notoriousskeptic we are all famil iar wi th through the history of epistemology.36 What

35 Barnes, for instance, writes: the answer Aristotle gives to the first question is whole-heartedly empiricist; and only a failure to distinguish between the two questions of B 19 willpermit a rationalistic interpretation. Nous is an answer to the second question, not a rival,rationalistic, answer to the first: intuition as a mode of discovery is absent from APst.(1994, p.270).

36 Although the conception of knowledge as justified true belief (JTB) has been tacitthroughout the history of traditional epistemology, the almost universal interpretation of thejustification clause, until recently, was such that it required complete/conclusive justifica-tion that was subjective in the sense that all the grounds for complete justification are meantto be accessible to the one who has knowledge this is what made the traditional account ofepistemic justification internalist. But this complete justification was also meant to be objec-tive in that it would guarantee the truth of what is claimed to be known: this is what made itcomplete and conclusive. With this alleged certainty, the traditional account ofknowledgewas also internalist (see fn.18 above). To my mind, one of the great discoveries of epistemol-ogy in this century was the realization that proper epistemic justification need not be com-plete and conclusive. Gettier (19 63) partly owes his success to the account of knowledge thatwas the result of dropping the requirement of complete justification: one does have knowledgeas long as the three conditions given for knowledge are in fact true even if ones (incomplete)justification depends on actual ly fa lse bel iefs. The decision to allow for incomplete justifica-


19/49

MURAT AYDEDE 18

reason is there to think that Aristotle is trying to respond to such a skeptic,rat her th an giving an account of wha t t he nature of scienti f ic knowledge is? Par t

of my argument in what fol lows will consist in an attempt to show th at there isli tt le, if any, reason.37

We may remember that Aristotle begins the Posterior Analytics by definingepisteme as demonstra ted knowledge. Now demonstra ted knowledge is not just akind of justif ied true belief whose justif icat ion happens to be deductive. It mustbe demonstrative, and Aris totle , to a grea t extent, can be seen to be cash ing out inthe first book of the Posterior Analytics just wh at is involved in demonstration. Insuch a context, (iv), given above, is not found problematic: necessarily, we haveknowledge only i f what we bel ieve is true. Wh at e lse could be less problematic?Truth is a def init ional ch aracterist ic of any kind of knowledge. But also, neces-sarily, we h ave scientif ic knowledge tha t P only if we ha ve the demonstrat ion ofP from principles th at we also know. Wh at is t he nature of our knowledge ofprinciples? We are told that i t is not demonstrated knowledge, hence it cannot beepisteme. Aristotle tells us th at one has the nous of principles. One does not,strictly speaking, epistastai a principle, but noei it . So one can agree with Barnesthat nous is just the name of the hexis we are in when we apprehend the princi-ples. But this hexis is a knowledge sta te. Wh y th en should (i i i) be thought to beproblematic in th is context? It is thought to be problematic , I submit, only if onefails to distinguish between two different issues: establishing ones claims toknowledge which has primarily to do with internal epistemic justification onthe one hand, and on the other, giving an explicative account of the nature ofknowledge which has to do with capturing those conditions whose sat isfaction isboth necessary and suff icient for one to have knowledge. Now these two issues

coll apse into one, of course, if the sat isfaction of those conditions is to be infal li -bly accessible to one who knows. Apart from the general question of the very pos-sibil i ty of such an access ever (especia ll y in the cases where the objects of knowl-edge are universal and empirical facts), is there any evidence that Aristotle as-sumes, or requires, or even makes an at tempt to show, that the sa tis fact ion of t heconditions he specifies for scientific knowledge are accessible in this way to one

tion resulted in externalist accounts of knowledge. However, the enormous difficulties in-volved in the attempts to solve Gettier problems soon forced many researchers in the direc-tion of developing externalist theories of epistemic justification too. One way of seeing myproject in this paper is as an attempt to combine Aristotles general and well-known natural-ism regarding his scientific methodology with a naturalist interpretation of his epistemology,more specifically, with an externali st interpretation of his account of scientific knowledge. It

would be an interesting exercise to see just to what extent his psychological account ofnousin De Anima (especially the Active Intellect of III 5) can be seen as, inter alia, a version of agenerali zed externalis t account of epistemic justification. See VII below.

37 See C.C.W. Taylor (1990) who also claims that Aristotle in his epistemology was notconcerned with responding to a Cartesian-style skeptic. However, he nevertheless expressessurprise about Aristotles alleged understanding of the relation between nous and induction.


20/49


who has scienti f ic knowledge? I do not think th at t here is. It seems to me th a tAristotle is innocent of the problem th at h is interpreters ha ve found in his writ -

ings, especia l ly in the PosteriorAnalytics.

- III -

I think, it is time to see in more detail what the conditions of having scien-ti f ic knowledge are as specif ied by Aristotle. In this and the next section (I IIIV ), I wi ll reconstruct Aristot les understanding of demonstrat ion as discussed es-pecial ly in APo I 29. For, as we wil l see, th is is th e key notion in having scien-ti f ic knowledge. My aim is to be very clear and explicit about exactly wha t i t isto have scientific knowledge according to Aristotle, i.e. about the necessary andsuffic ient conditions of an explica ti ve analysis of scientif ic knowledge. I wil l re-cast Aristotles discussion in a reconstructive fashion and, at the end, suggest adefini tion of scientif ic knowledge. Along the way, we wil l recover many themesessenti al for a proper understanding of Aristotles epis temology. Once we dothat, how to fit nous in the emerging picture, its role and place, will become ap-parent. Th is is th e job of sections VIVII , prior to which, in V, I wi ll brief lyreturn to the orthodox interpreta tion ofnous for more cri ticism.

In the very beginning of th e second chapter of Book I of th e Posterior Analytics,Aristotle gives his first formulation of the conditions on having episteme:38

We think we s-know [epistasthai] a thing in the unqualified sense [haplos] (and notin the sophistic manner, i.e. incidentally) when we think [1] that we know[gignoskein] the cause from which the fact results, and [2] that the fact cannot beotherwise. (71b913)

In the rest of t he ch apter and in Chap ter 3, Aristotle seems to elaborate on clause[1]. And then, in Chap ters 46, he ta kes up the oth er clause [2]. But as we wil lsee, wha t he sa ys here about the f irst clause is preli minary and not very i l lumi-nating. Th is is not very surprising, since, as I wi ll argue, Aristotle needs wha t heha s to say about second clause in a proper char acterizat ion of clause [1] which isthe core of having scientific knowledge.

One interesting feature of this passage is the we think we know construc-tion. Aristotle is very clear and explic it about th is construction, and his use of i tin many other passages39 where he wants to reformulate the one or the other

38 In what follows I will use the somewhat barbaric s-know and s-knowledge for

epistasthai and episteme respectively just to mark them off from other cognate expressionssuch as gignoskein and eidenai for which I will most of the time use just know. For therelation between these three verbs as they occur in Aris totles writings, see Burnyeat (1981).

39 For instance, cf. APo I 6 75a1518, 9 76a2829, 33 89a59, II 11 94a2021; Met I 3983a2425; Ethica Nicomachea VI 3 1139b2021, 1139b3334; Physica I 1 184a1213, II 3194b1920.


21/49

MURAT AYDEDE 20

clause in some fashion makes it clear that such a construction is not accidental orstylist ic here. Wh at might its import be? Barnes fi rst considers the idea th at

Aristotle gives a consensus argument for h is def ini tion: we th ink [episteme] to beso-and-so; therefore [episteme] is so-and-so (1994, p.91). Al though he seems tothink that there is some force in this interpretation, he, at the end, rejects it onthe basis that so interpreted, Aristotles argument... is factually mistaken; for[the passage] does not represent a correct analysis of the actual use of the Greekverb epistasthai (p.91). He says th at th is construction should be taken to indica teth at the passage gives a stipulat ive definit ion ofepisteme. I think this is basi-cal ly righ t. But there is more to it especia l ly when viewed in the l ight of someother passages as we will see later on. Suffice it to say, for the moment, th at ifthe definition is intended to be more or less stipulative, it must be relativelyclear that Aristotle sets his project in terms suitable for giving an explicative ac-count ofepisteme, and not necessari ly wi th an eye to show against the skeptic t hepossibil it y of h aving scienti f ic knowledge as the tradi tional commentators havethought it.

Let us recast th e essence of Aristot les def inition in more explic it terms:40

S s-knows that P if , and only if ,[1] S knows that is the explana tion ofP, and[2] S knows that P cannot be otherwise.

How should we analyze [1] and [2]? In particular, what should we include in anaccount of the ordinary knowledge (gignoskein in Aristotles passage) in [1] and[2]? The best thing to do in the way of answering these questions, I think, is to

proceed as Aristotle does in the subsequent chapters. So, then, le t us begin wi thAristotles prel iminary discussion of clause [1] in Cha pter 2.It is instructive to see that Aristotles first allusion to nous in the Posterior

Analytics is found almost right af ter h is defini tion ofepisteme. He says at 71b169:

Now whether there is also another type of s-knowing we shall say later; but we saynow that we do know [eidenai] through demonstration. By demonstration I mean ascientific syllogism [sullogismon ep istemonikon]; and by scientific I mean one in vir-tue of which, by having it, we s-know something.

It is clear that the reference here in the first sentence is to nous. It is interestingto see Aristot le using epistasthai in alluding to nous especial ly as he immediate ly

40 Since a free variable occurs in the definiens, the follo wing definition, str ictly speaking,is technically flawed. But I will leave it this way, since I do not want to bother the readerwith unnecessary complications that the quantificational issues might produce. I think thedefinition as given is intuitively clear enough to do its job for what fol lows.


22/49


proceeds to specify that episteme is just wha t we have in virtue of hav ing a dem-onstrat ion. I th ink Aristotle is writ ing here somewha t loosely (but see fn.1

above). In clause [1] of h is defini tion ofepisteme above, Aristot le does not mentiondemonstrat ion, but th is passage and the one immediate ly fol lowing th is make itfairly clear that to have the explanation ofP is just to have the appropriatedemonstration ofP:

If, then, s-knowledge is as we posited, it is necessary for demonstrative s-knowledgein particular to depend on things which are true, pr imitive, immediate, better knownthan, prior to, and explanator y of the conclusion (for in this way the principles wil lalso be appropriate to what is being proved). For there will be syllogism evenwithout these, but there will not be demonstration; for it will not produce s-knowledge. Now must be true because one cannot s-know what isnot the case... And be primitive and non-demonstrable because you will not s-know if you do not have a demonstration of them; for to s-

know non-incidentally that of which there is a demonstration is to have a demons-tration. They must be both explanatory and better known and prior explanatorybecause we only s-know when we know the explanation; and prior, if they are ex-planatory; and already known not only in the sense of understanding their meaningbut also of knowing that they are. (71b203 4)

So to know the explanation ofP is just to have the demonstrat ion of it. We learnth at demonstration is a specia l sort of syl logism, but i ts specialness is not char ac-terized by means of some sort of formal properties. A syl logism is a demonstra-tion because of non-formal properties of i ts premises. We have th e episteme ofP inso far as we syl logistica ll y infer i t from premises th at are true, primitive, imme-diate, better known than, prior to, and explanatory ofP.

However, th is is not a rigorous character iza tion. For it is not necessary th a teach premise in the syllogism be immediate and primitive for its conclusion to beepisteme. Aristot le elsewhere al lows th at a lready demonstrated premises,whenever required, should be used in the demonstration of a conclusion.41 Soth ere can be a ch ain of demonstrat ions. But episteme should ultimate ly depend ondemonstrat ion from premises th at are primit ive a nd immediate. Aristot leprobably does not consider this complication for the ease of his initial charac-terization.

Given this initi al picture, the al lusion to nous almost right a fter the ch arac-terizat ion ofepisteme should come as no surprise. For 71b913 defines only the de-monstrative scientific knowledge which ultimately depends on non-demonstrable

41 Reasoning is demonstration when it pr oceeds from premises which a re true and primi-tive or of such a kind that we have derived our original knowledge of them through premiseswhich are primitive and true (Top I 1 100a27 30). See also APo I 10 76b10 and Rhetorica I2 1357a 7. The whole structure of the discussion in APo I 3 to the effect that infinite regressin demonstration is inadmissable also makes it clear that Aristotle allows for finite chains ofdemonstrations.


23/49

MURAT AYDEDE 22

but somehow known premises. It is only when th is defin ition is construed as a (or,the)general definition of what it is to have knowledge that the possibility of

having non-demonstrable knowledge is prone to become an epistemological puz-zle. Th is, of course, does not remove, at le ast in our eyes, the existence of certa inprima facie epistemological problems in Aristotles conception of scientific knowl-edge, but seems to suggest that they are not, at least in the Posterior Analytics,within the scope of Aristotles philosophical worries, at least not directly any-way. When viewed in th is way, Ar istotles discussion and rejection of circularand infinitely regressive demonstration in the subsequent chapter becomes po-lemica l in its purpose. I believe Aris totle uses th is polemical d iscussion for ex-pository purposes, in order to better convey what he thinks scientific knowledgeinvolves. Indeed he says, neit her of these views is true or logical ly unavoid-able (72b89). His primary a im is not to show against the skeptic t he possibil -it y of scientif ic knowledge in general. In this respect, that Ar istotle does noteven attempt to give an argument for hi s cla im th at not all knowledge is demon-strative (72b19) has always been a source of puzzlement.42 For how could Aris-totle so lightly pass over such an important point, which is the very point atissue?! But on my reading th is is not a puzzle at a ll, because th is is not Aris-totle s worry.43 He simply gives an explicative account of a certain sort of knowl-edge, namely, the kind of knowledge we have when we have a demonstration.That is all he wants to clar ify .

Let us examine the properties Aristotle cites that the premises of a properdemonstration should possess; namely, that they must be true, primitive, imme-dia te, bet ter known than, prior to, and explanatory of t hei r conclusion. That thepremises should be true needs, I th ink, hardly commenting on. But as a bare

minimum, we can say th at episteme is a kind of knowledge, what is known must betrue, and since episteme is a sort of derived knowledge what it is derivedfrom must also be known, hence must be true.

Wha t about immediacy and primit iveness of the premises? In general, ac-

42 Again, see, for instance, the quotes from Grene (1963) and Morrow (1970) above.43 Not that this cannot constitute a legitimate philosophical worry in general, but it is not

the worry Aristotle is attending to there. Apart from the Posterior Analytics, Aristotle,throughout all his writings, is indeed known to be notorious in balking at skeptical ques-tions. He just does not seem take the skeptic seriously at all. See, for instance, Met IV 6where he writes: There are, both among those who have these convictions and among thosewho merely profess these views, some who raise a difficulty by asking, who is to be the judgeof the healthy man, and in general who is likely to judge rightly on each class of questions.

But such inquiries are like puzzling over the question whether we are now asleep or aw ake....These people demand that a reason shall be given for everything; for they seek a starting-point, and they seek to get this by demonstration, while it is obvious from their actions thatthey have no conviction. But their mistake is what we have stated it to be; they seek a reasonfor things for which no reason can be given; for the starting-point of demonstration is notdemonstration (1011a214). Also, cf. Physica II 1 193aff.


24/49


cording to Aristotle, a proposition of the subject-predicate form, all S is P (le t usabbreviate t his into SaP) ,44 is immediate just in case there is no middle term,M,

such that MaP, SaM | SaP is a demonstrat ion. This way of cha racter izing im-mediacy is of course circular i f we are inquiring, as we are, into wha t makes a sy l-logism a demonstration. Notice th at in the above quotat ion, where Aristot lebriefly g losses eac h of the ch aracterist ics af ter hav ing cited them, immedi ateis replaced by non-demonstrable . Perha ps th is is why he later on does not giveth e obvious definit ion of immediate:

To argue from primitive premises is to argue from appropriate principles; for I callthe same thing primitive and principle. A principle of a demonstration is an immedi-ate proposition, and an immediate proposition is one to which there is no otherprior. (APo I 2 72a69)

This passage, and indeed the whole chapter, gives th e impression that Aristot leuses immediate, primit ive, and non-demonstrable in the same sense. IfP isprimiti ve, t hen there is no Q prior to P such that P is derivable from Q (where Qmay be a set of true propositions). But, of course, for th is cha racteriza tion towork appropria tely, i.e . in t he sense Aristotle intends primit iveness to be a fea-ture of premises in a demonstration, the sense of derivation should be specified.And the obvious and intuitive candidate for it is demonstrat ive deriv ation. Foras Aristotle is well aware, there can be many sound deductive arguments estab-lishing P, where P happens to be in fact demonstrable , that are not demonstrat ionin the strict sense. Hence we have again the same sort of circular ity . So we arestill in dark about what demonstration is, in the strict sense.

In fact, Aristotle gives h is first h int in the direct ion of breaking the circular-

ity when he says that the premises of a demonstration must be prior as much asth ey a re explana tory. Indeed, as we wil l see more clearl y in a moment, explana-tion from naturally prior and necessary premises is th e key idea for Aristoteliandemonstration. At 71b3572a5 Aristotle writes:

Things are prior and better known [gnor imotera] in two ways; for i t is not the sameto be prior and better known by nature and prior and better know in relation to us.I call prior and better known in relation to us what is nearer to perception, priorand better known absolutely [haplos] what is further away .What is most universal is furthest away, and the particulars are nearest; and theseare opposite to each other.

44 It is not necessary for an immediate premise to be of a universal affirmative form. Aris-totle allows them to assume universal negative form, see APo I 15. Aristotle in many of hisexamples uses also singular statements where it is clear that he intends them to be taken asimmediate. But Aristotle is quite explicit that the proper form for scientific discourse is uni-versal affirmative premises and demonstrations based on them, see APo I 14.


25/49

MURAT AYDEDE 24

Although Aristot le can natural ly be taken to be ta lk ing about premises, h is dis-tinct ion is not restricted only to propositions. In fact, i t is clear from other pas-

sages tha t the dist inction finds its natura l pl ace among concepts. A concept isprior by nature to another in so far as i t is more general and abstract. For exam-ple, animal is naturally prior to and better known than man, animate be-ing is prior to and better known by nature th en animal , and so on. Accordingly,premises, at least in some rough and intuit ive way, can be so graded according toth eir general it y and the abstract leve l of the terms they contain, as long as theyare true and universal . But even if much of th is is intuiti vely clear , we are sti l lin dark about why what is thus further awa y from perception is cal led prior bynature and wha t i ts relevance is to demonstrat ion.

In Chapter 2 Aristotle does not el aborate on the requirement th at th e prem-ises of a demonstrat ion should be explanatory of its conclusion. And th is is nosurprise. For in a certa in sense, th is is going to be one of th e main focuses of Aris-totle in the remainder of h is treat ise. In fact Aristotle needs to address the ne-cessity requirement in the definition ofepisteme in order to elaborate on explana-tion. And th is is exactly wha t he does beginning in Chap ter 4.

- IV -

After quickly sta ting at 73a214 th at demonstrat ion is syl logism from wha tcannot be otherwise, and hence, implying that necessity ofepisteme is transferredfrom the premises of demonstration, he says at 73a256 that therefore we mustcomprehend the nature and ch aracter of th e premises from which demonstrat ionsproceed as if t h is was not what he has been examining so far. But in a sense,

th is is not surprising: we ha ve seen so far from our brief examination of wha t Ar-istotle says about the first conjunct in the definition ofepisteme, t hat his ch arac-teriz at ions of the fea tures of demonstrat ive premises are eit her impl ici tly circu-lar or not genuinely i lluminating. Indeed, th at the real issue begins here and th eforegoing discussion was a kind of warm-up becomes immediately clear fromwha t Aristotle embarks on in the remaining part of the chap ter. Wh at he doesis to introduce three technical terms: predicated of all (kata pantos), per se(kathauto), and universal (katholou).

Alt hough Aristotles definition (73a2834) of predicated of al l is sligh tl yambiguous, the general idea i s fa ir ly simple and clea r: the predicate A is predi-cated of al l B if , and only if , th e universal af firmati ve sentence AaB is assertedto be (t imelessly) true. It is clear th at Aristot le here tries to capture wha t we

are committed to when we universal ly quantify over something wit hout any ref-erence to a particular time and place.

For our purposes, the definition of per se predica tion is more important . Infac t, i t is t he key to our reconstruction. Aristotle gives four senses in which onething isper se predicated of another th ing (73a3573b25). However, for us only


26/49


three of t hem are relevant: AaB is a per se predication if, and only if, (1) B isan element in the defini tion of A, or (2) A is an element in the definition of B,

where th e defini tions of A and B state the essences or essential natures of the irdenotations. Al th ough many commentators tend to dismiss the other two specif i-cations of per se as irrelevant to Aristotles purpose, I think that one of them(the forth in Aristotles ordering) is relev ant. Th is has to do, it seems, wi thevents or happenings:

[4] Again, in another way what holds of something because of itself holds of it in it-self [kathauto], and what does not hold because of itself is incidental. E.g. if therewas lightning while he was walking, that was incidental: it was not because of hiswalking that there was lightning that, we say, was incidental. But what holdsbecause of itself holds in itsel f e.g. if something died because of being sacrificed, itdied in the sacrifice s ince it died because of being sacrificed, and it was not inciden-tal that it died while being sacrificed. (73b101 6)

This passage (and many others like it), and Aristotles discussion ofper sepredication in general, make clear where the necessity comes from and how it isrela ted to explanat ion and priority in nature. Aristot le is an essentia li st. Hethinks that certain things and events in nature naturally classify themselvesinto certain necessary kinds and patterns independently of our practical or theo-retica l concerns. On his view, certa in th ings in nature have certa in common cha r-acteris tics th at are the basis for the ir being of the same kind. These characteris-tics are essential to them: they are together what is basically responsible forthem to be the kind of th ings they are . There is a natural and necessary orderamong things in the world. Th is order is objective and naturall y h ier archica l:

someth ing cannot be a man wit hout al so being an animal . Animal hood is essen-ti al for manhood as wel l as, say , for being a horse, but not vice versa. Th at manis an animal is an essential , and thus, aper se predication. For it is part of wha tit is to be a man th at it be an animal. In other words, being an animal is part ofth e natura l and objective essence of being man. But, i f so, th ere is no more funda-mental, natural fact other than that all men are animal that would explain to uswhy al l men are animal. In a sense, that al l men are animal i s self -explana tory,and natural ly and objectively so. It is a basicprinciple, so to speak, that all menare animal . The objective t axonomic order found in nature is wha t determines th eordering of our concepts in terms of priori ty by nature. It is in so far as our conceptscarve nature at i ts joints tha t th ey exhibi t th ese priorit ies. When we look, wh atwe directly see is not animal as such, nor is it man, but Callias the man even

still incidentall y at th at.45

But it is only when we see Call ias in the l ight of t heessentia l at tributes of the natural kind of wh ich he is a member tha t we come tounderstand why Cal l ias exhib its certain features tha t he does. It is only when

45 See De Anima II 6; cf. APo I 31.


27/49

MURAT AYDEDE 26

we relate w ha t is prior in relat ion to us to wh at is prior by nature tha t we come tounderstand why certa in th ings behave in certa in ways. For instance, we may

wonder why these things have four stomachs; we may be answered that it is be-cause they are cows. But what is it about cows th at t hey have four stomachs?Well , because they are ruminants. And presumably th at is th e end of the story.Cows have four stomachs, because they are ruminant animals and all ruminantanimals ha ve four stomachs. Th is is simply but essentia l ly how things are in na-ture. And to find out this is the business of the Aristote l ian scientist .

Simil ar ly wit h events and happenings in the world. Certa in kinds of eventsoccur only after or simultaneously wit h certain kinds of other events. There is aregulari ty and pat tern in the natural phenomena. For instance, we observe t h atthe moon always exhibits phases, and we come to realize that it is not an acci-dent that this is so; there is a principle involved here, something is responsiblefor the phases of the moon. In brief, we rea l ize th at th at the moon exhibi tsphases h as a cause. And when w

ARISTOTLE ON EPISTEME AND NOUS: THE POSTERIOR ANALYTICS

Documents

Transcript of ARISTOTLE ON EPISTEME AND NOUS: THE POSTERIOR ANALYTICS