Aristotle and the Appearances

Paul Nieuwenburg. Journal of the History of Philosophy. Volume 37, Issue 4. October 1999.


G. E. L. Owen’s influential article Tithenai ta phainomena has had a very special efficacy in converting long-standing suspicions into the certainty of what one might call, without exaggeration, an orthodoxy. One of Owen’s arguments is widely thought to remove, in a quite definite way, all doubt surrounding the interpretation of the ambiguous term ta phainomena, usually rendered ‘the appearances,’ as it figures in one of Aristotle’s reflections on ‘the methods of ethics.’ And indeed, as the pun on Sidgwick’s The Methods of Ethics suggests, the argument paves the way for extracting from Aristotle’s instructions a telos not unlike that of a certain modern conception of ethics-a coherent body of propositions wrought from our common-sense intuitions about moral matters.

Point of departure is a quotation of the methodological reflection from the Nicomachean Ethics (EN 1145b2-7), the subject of the argument in question. The reflection sketches the procedure to be followed in the treatment of akrasia, or the phenomenon of, to put it in a rather slipshod manner, doing what one knows to be wrong. It will be helpful to comply with standard practice and to give Owen’s own translation of the bulk of this passage (b2-6), supplementing it with the missing final clause (b6-7).

Here as in other cases we must set down [tithentas] the phainomena and begin by considering the difficulties [diaporesantas], and so go on to vindicate if possible all the common conceptions [ta endoxa] about these states of mind [peri tauta ta pathe], or at any rate most of them and the most important; for when both the difficulties are solved and the endoxa are left, it would have been proven sufficiently [dedeigmenon hikanos]. (EN 1145b2-7)

Owen argues that W. D. Ross’ ‘empiricist’ construal of the term phainomena (‘observed facts’) cannot be consistently maintained in a dialectical examination of beliefs on moral matters-and we shall presently see why this should be so. Indeed, the term must be systematically ambiguous, referring to ‘hard’ empirical data in one context (notably APr 46a20) and to endoxa (or legomena [EN 1145b20]), i.e., ‘common conceptions’ in the other.

As a consequence, the method outlined in this passage would look like this. First, you have to assemble and present a collection of beliefs (tithenai ta phainomena) on the relevant subject-matter. Since it is most likely that tensions will disfigure the set of endoxa you have garnered, you must, second, by way of preliminary exercise ‘go through the perplexities’ (diaporein) they raise, in order to pinpoint the inconsistencies and infelicities barring the way to endoxic coherence. The removal of these frictions will result in a ‘maximal consistent subset’ of endoxa on the matter, namely, when the perplexities (aporiai), or difficulties (duskhere), are solved. Sufficient proof has been given when all, most, or the most important (kuriotata) endoxa reappear in this final set. “And then? And then nothing: your philosophical task is over.”

The intersubstitutability of ‘appearances’ (phainomena) and ‘reputable beliefs’ (as I prefer to render endoxa) is buttressed by a fairly general consensus in contemporary Aristotelian scholarship. Many authors assume that in the text under consideration and, in its wake, in certain other ethical contexts, both terms have the same reference, without, however, being synonymous. Henceforth, I shall label this assumption (I) (for ‘intersubstitutability assumption’), covering both Owen’s conclusion and the argument leading up to it, on the obvious supposition that espousing the conclusion entails a commitment to the validity of the argument. Against all appearances, however, (I) is not as foolproof as its widespread acclaim suggests. Although extensions of (I) to other methodological contexts in Aristotle’s ethical writings (in particular EE 1214b28-1215a) prejudice a due understanding of Aristotelian ethics, the point is, I believe, worth arguing in its own right.

In what follows, I shall first lay bare a ‘perplexity’ latent in (I). It will emerge that its premises do not as firmly support the conclusion as it has seemed to many writers. Therefore, I shall not side with those who take for granted that it has been conclusively shown that the phainomena cannot possibly be ‘observed’ or ‘evident facts’; which means that I shall not endorse (I). Since (I) is levelled at Ross’ ‘observed facts,’ I pass on the onus of proof to its adherents.

To fill the gap opening up between the attributed and actual force of the argument I shall subsequently expound an interpretative counter-proposal for reading the crucial texts. It is not my aim to give a full-scale interpretation of Aristotelian akrasia, nor to defend, on Aristotle’s behalf, the ‘observableness’ of his appearances. I merely want to show that we can make sense of the method sketched at EN 1145b2-7 if we remain as faithful as possible to the Wortlaut of Aristotle’s Greek and go back in the unfashionable direction signposted by ‘observed facts.’ ‘Appearances,’ let so much be conceded, is, as a translation of phainomena, more faithful to the Greek and has the advantage that it does not conjure up the Baconian protocol. Nevertheless, the term is rather non-committal, and I shall try to give it more substance by upholding, in essentials, its empirical orientation.


(I) goes on two legs. First, taking the verb tithenai in the expression which gives the title to his article to mean ‘to set out,’ Owen claims that what Aristotle actually proceeds to ‘set out’ are endoxa (= legomena 1145b20]) on the subject; and these “turn out as so often to be partly matters of linguistic usage or, if you prefer, of the conceptual structure revealed by language.” This pleads for the intersubstitutability of phainomena and endoxa (= legomena). As the bracketing makes clear, my chief grudge concerns the referential equation of the first two terms, and I shall leave the third in peace.

Second, both the fact that Socrates’ position, to the effect that “those who act against their own conviction of what is best do so in ignorance,” ‘plainly contradicts’ (amphisbetein) the phainomena (1145b23-8) and that Aristotle eventually retains this position (1147b13-7) conjointly show that the phainomena cannot be ‘observed facts,’ but must be “what would commonly be said on the subject.”

If we want to get clearer about the meaning of phainomena, (I) suggests a natural yet self-defeating course of interpretation. For ex hypothesi, prying deeper into endoxa should yield more clarity about phainomena. Since the former constitute the raw material of dialectical reasoning, the logical step is to turn to the Topics, Aristotle’s manual or compendium of dialectic. The aim of this treatise is to find a method of engendering in its students a capacity to construct arguments about every proposed ‘problem’ (problema), which is a translation of an aporia into a question of the form ‘p or ~p?’ (Top 101b28-34). Such dialectical arguments are composed ‘from endoxa’ (Top 101a18-21; 101b5-10; SE 183a37-8; Rh 1354a1-6), which term is elucidated as follows:

Endoxa are the things which are believed [ta dokounta] by either all, or most, or the wise [tois sophois], and by all, most, or the most familiar [gnorimon] and endoxoi of those [i.e., the wise]. (Top 100b21-3).

Whether this be a characterization or a definition in the strict sense, as it stands it furnishes extensional criteria sufficiently determinate to brand Socrates’ argument an endoxon.) It is at this juncture that the troubles begin to assert themselves for (I). But before we point these out, let us explore some considerations pertaining to the (doubtful) claim, essential to the second leg, that Socrates’ position is one of the endoxa to be salvaged (I shall appropriate the scholastic notation SOR to symbolize ‘Socrates’ argument.’)

On the one hand, inclusion of SOR in the reference of endoxa throws doubt on the translation we started with. Even if we, with writers like Sidgwick, gloss ‘common beliefs’ as the beliefs of ‘reflective’ or ‘educated’ persons, we should point out that ‘common beliefs’ or ‘common conceptions’ are misnomers in terms of the Topics characterization. For this perfectly allows for counterintuitive and uncommon beliefs to be endoxa, and SOR, therefore, need not meet with general acceptance, not even among the intellectual elite, to achieve this status.

Socrates’ reputation in itself seems to warrant examination of his views on ethical matters. It should be recorded that Aristotle conceives of Socrates mainly as a moral philosopher (Met 987b1-2; 1078b17-8; PA 642a28-30). The ‘wise’ (sophoi) figuring in the characterization of endoxa need not be those who have sophia in the sense of ‘first philosophy’ as familiar from the Metaphysics. They may be the experts, the most outstanding and sagacious representatives of their disciplines, who are ‘wise’ in a particular respect (kata meros). ‘Wisdom’ in this sense is what Aristotle calls the ‘excellence of art’ (arete tekhnes) (EN 1141a9-10; Top 104a14-5; 33-7), but it emerges that this qualification also extends to the model sciences of mathematics and geometry (EN 1142a12-8). So if one may be wise with respect to mathematics and sculpture, why not with respect to ethics?

One might argue that SOR is not endoxon, precisely because of its collision with the phainomena (which are, ex hypothesi, common beliefs), in the following way. Aristotle condemns the attempt to cram the plurality of styles of friendship into one definition as “forcing the phainomena,” which has the unwelcome consequence of stating paradoxa (EE 1236b22-3). The term paradoxos is glossed, much in the way we know it from Hume, as ‘contrary to the beliefs of the many’ (enantion tais ton pollon doxais) (Top 104a11-2; b24-5). Paradoxic positions sustained by well-known philosophers, such as the Eleatic ‘what is is one,’ are the material from which dialectical theseis are constructed (Top 104b24-8; Ph 185a1-3). SOR satisfies these criteria, and this explains why it is not included among the phainomena-endoxa. On the other hand, the discussion of a thesis does not properly belong to an inquiry conducted in the context of some particular discipline (EN 1095b31-1096a2; Ph 185a5-7; 14-8; DC 306a11-5), but to dialectic proper (Ph 185a2-3; DC 298b20). In ethics, then, Aristotle could never consistently attempt to vindicate a position which is patently paradoxic. Therefore, SOR must be endoxic.

Such considerations might be adduced to bolster the claim, vital to (I), that SOR is an endoxon. Actually, however, (I) relies on grounds internal to the akrasia discussion of EN H to argue that it is. Doesn’t Aristotle say that if not all endoxa, then most or the most important or authoritative of them must be shown to be true (EN 1145b4-6)? And isn’t SOR among the latter? Isn’t it, eventually, ‘left standing,’ that is to say, doesn’t it stand the dialectical test?

More disturbingly, however, the claim tends to tear (I) apart. For if, ex hypothesi, SOR is an endoxon, then it is a phainomenon; still, it is not found in the initial set of phainomena-endoxa ‘set out’ or ‘down’ by Aristotle. On (I), this is anomalous. One cannot consistently maintain that Aristotle ‘sets down’ phainomena-endoxa, that these are ‘common beliefs,’ that SOR contradicts these common beliefs, and that SOR, not being itself a common belief, is among the endoxa left standing. One thing, such a set of claims entails is that the phainomena, in the sense of ‘common beliefs,’ can only be a subset of the relevant endoxa. But this means that endoxa and phainomena cannot have the same reference-at least not in the passage under scrutiny. We might try to find a way out by making the phainomena that subclass of reputable beliefs which we call ‘common beliefs.’ In view of the considerations just invoked, we need not have any misgivings in saving SOR, which simply is an expert endoxon. But then again, this conclusion is not supported by the two legs of (I).

I doubt, therefore, that the argument ‘sets it beyond doubt’ that phainomena and endoxa, in this particular context, have the same reference. In point of fact, I doubt that the phrase tithenai ta phainomena means something like ‘to present a catalogue of common beliefs.’ It is all too readily taken for granted that the polysemic verb tithenai means something like ‘to present’ or ‘to list’ the items denoted by the term phainomena. This, I shall argue, is not quite its sense here. Admittedly, the phrase tithenai ta Phainomena might, literally, mean ‘to lay down the appearances’-yet this translation does not, in itself, take us much further.

I shall now embark on the more constructive part of the journey, the purpose of which is, as noted, to show that keeping up an ‘empiricist’ construction of the ‘appearances’ does not reduce the akrasia program to nonsense, but is perfectly consistent with general Aristotelian methodology.


For a start, we should carefully attend to the phrasing of Aristotle’s reproduction of SOR and its subsequent rebuke. Let us divide the passage in two parts (at the expense of elegance I adopt a more ‘Boethian’ translation technique, thus trying to preserve as much of the original word order as possible):

Part (1): Now someone could be perplexed as to in what way someone who behaves akratically supposes rightly [pos hupolambanonta orthos]. That he should behave so when he has understanding [epistamenon], some say it is impossible; for it would be upsetting-so Socrates thought-that if understanding was in a man something else could master it and ‘drag it about like a slave.’ For Socrates was entirely opposed to the view in question, holding that there is no such thing as akrasia; no one, he said, acts against what he supposes [hupolambanonta] best-but through ignorance [di’ agnoian]. (EN 1145b21-7)

It is this argument (logos) which, as is usually translated, ‘plainly conflicts with’ or ‘plainly contradicts’ the phainomena (amphisbetei tois phainomenois enargos), and, as Aristotle immediately continues,

Part (2): it is necessary to inquire about what happens [deon zetein peri to pathos], if through ignorance, what mode [tropos] of ignorance comes into being [ginetai]. For it is evident [phaneron] that he who behaves akratically [ho akrateuomenos] does not mean [ouk oietai] [to do so] before getting in the affection [prin en toi pathei genesthai. (EN 1145b28-31)

The aporia raised in part (1) (EN 1145b21) is not, as is often thought, about the possibility of akrasia itself (pos modifying the whole clause); it rather asks for a specification of the epistemic mode involved in it (pos modifying hupolambanon orthos). This explains the use of hupolambanein in the formulation of the aporia; it is the verb corresponding to his generic term of art for epistemic types in general, hupolepsis, the species of which are understanding (episteme), belief (doxa), practical wisdom (phronesis), and their contraries (DA 427b24-6). It is clear that these ‘contraries,’ to which the term ‘ignorance’ (agnoia) points, are preemptively excluded by the very terms in which the aporia is couched. The perplexity, then, may be taken to ask for a suitable interpretation of the term eidos of the endoxon which says that the akratic, “while knowing [eidos] that the things he does are base, does them on account of affection [dia pathos]” (EN 1145b12-3). This seems to imply that Aristotle takes some interpretation of this claim for granted-which indeed emerges from occasional remarks in the preceding parts of the Nicomachean Ethics (1095a6-11; 1102b14-25; 1111b13-4; 1136a31-4).

The fact that Aristotle casts the aporia in his own technical idiolect leads me to the imposition of a constraint on the interpretation to be developed. We should, to the greatest extent possible, construe the passages under consideration in the spirit of the distinctions reflected in Aristotle’s choice of words. As we shall see, this will have important consequences for the interpretation of that part of SOR which Aristotle reproduces or paraphrases in his own terms (and which Owen inaccurately renders “those who act against their own conviction of what is best do so in ignorance”).

Now that we have cast a glance at the text containing the second instance of phainomena, let me rise to the occasion and introduce an additional constraint on interpretation.

It has been felt that Aristotle’s appearances, across different contexts, need inoculation to shield them against semantic infection. Of course, since Owen’s interpretation of EN 1145b2-7, which renders phainomena ambiguous, is its nidus, a successful frontal assault on (I) will dissipate the urgency of such an intervention. Although there is a spot of dogmatism to Owen’s requirement that the two instances of phainomena in our texts (EN 145b3; b28) should be given the same sense, on pain of ‘inconsistency,’ I shall adopt it as a further constraint upon interpretation. We shall, with Ross, attempt to maintain an empirical interpretation of ‘appearances’ in both cases.

Back to the surface of the text. That SOR is at loggerheads with common belief is twice commemorated only in the relevant section of Plato’s Protagoras (Prot 352b2-c2; d4-e4), so that we may indeed speak of a plain conflict in this connection. Up to now I have tacitly assumed that the adverb ‘plainly’ (enargos) qualifies the main verb amphisbetein. But this translation, venerable as its pedigree may be, is not the only, and not even the most likely alternative. The adverb may modify phainomenois rather than amphisbetein. On this account, SOR, instead of ‘plainly contradicts what appears,’ ‘contradicts what plainly appears.’

If one presumes phainomena to refer to ‘common beliefs,’ the prevalent interpretation appears inescapable. Admittedly, the adjective enargos may apply to linguistic items, meaning something like ‘clear’ or ‘distinct.’ Such a construal, however, fails to make sense. Aristotle cannot possibly have in mind something like ‘clearly formulated’ common beliefs, for this plainly conflicts with the elaborate operations of disambiguation performed during the discussion of the aporia about the exact way in which the akratic are said to act knowingly (EN 1145b-12-3; 21-2; 1146b8-9; 1147b18). If Aristotle’s procedure shows anything, then it is that the endoxa are not stated clearly at all.

Against this, it might be argued that the adverb is deployed exactly as a pointer to the observational thrust of the term phainomena. For enargos and its derivatives have close ties with the sense-modality of vision. If this is so, a conflict with ‘plainly observed facts,’ despite the cluster of associations which comes with it, seems more fitting than a ‘plain contradiction of common beliefs.’ Let us briefly survey some circumstantial evidence.

First, we employ the Greek equivalent of the sceptical ‘looks’ locution, phainetai, when something which we do not perceive clearly (enargos) looks like, for instance, a man (DA 428a 12-5). Further, while composing, the tragedian is advised to ‘place before the eyes’ (pro ommaton tithesthai) the scenes from which the plot is concocted. Visualizing the actions will make him see them as vividly (enargestata horon) as when he would himself partake of them (Poet 1455a22-34; Rh 1386a32-b4; 1410b33-5). Finally, Anaxagoras, when asked by a layman what sort of person he considered truly happy, is reported by Aristotle to have briskly rebuffed his interrogator: “None of whom you think; he would appear [phaneie] a strange fellow to you” (EE 1215b6-8). The many (hoi polloi), or, as they are still called by the time of Sidgwick’s Methods and Bradley’s Ethical Studies, ‘the vulgar,’ allocate the title according to external goods like wealth, these being the only goods perceived by them (aisthanomenoi) (EN 1179a13-6). And wealth, it should be recorded, is one of the ‘manifest and evident’ goods (enargon ti kai phaneron) dismissed as qualifiers for ‘well-being’ (eudaimonia) at an early stage of the inquiry (EN 1095a22-3; 1096a6-7).

In view of these facts of usage, the collocation of to phainomena enargos (EN 1145b28) with phaneron (EN 1145b31) may be meaningful to the extent that the latter is an explicative echo (gar) of the former. And indeed, in the Physics we find phanera in formulations suspiciously close to the one under consideration, where it would be equally reasonable to expect phainomena. Seen from this angle, Ross’ attempt to bring the Ethics’ reflection on method into line with more empirical passages is far from wayward. Oddly enough, Ross himself translates ‘plainly contradicts’ and thus fails to capitalize on this opportunity to strengthen his reading of the passage.


One of the objections to ‘observed facts’ already alluded to is that it caters to the recruitment of ethical inquiry to Baconian empiricism. However, to claim that this objection misfires is not tantamount to denying all kinds of differences in method between, say natural science and ethics:

Most [principles] of a particular discipline are peculiar [diho] to it. That is why [diho] it belongs to experience [empeirias] to deliver the principles about a particular thing, I mean for instance astronomical experience in astronomical science (for after the phainomena had been grasped sufficiently this was the way the astronomical demonstrations were found), and the same goes for any other craft and science whatsoever; hence, when the facts [huparkhonta] about each particular thing are grasped, our next task is to set out the demonstrations readily. For if, according to our inquiry nothing of the facts that truly hold of the things [ton alethos huparkhonton tois pragmasin] is left out, we shall be able to find and demonstrate about everything of which there is demonstration, but of what there is by nature [pephuken] no demonstration, to make this clear. (APr 46a17-27)

Aristotle explicitly extends the scope of these remarks to all sciences and crafts whatsoever-and he does so more than once (APr 43a21-2; 46a3-4; 53a2-3). If we are prepared to take such generalizing phrases seriously, we shall have to allow for empirical survey of the facts (phainomena=huparkhonta) in ethical inquiry. And indeed, there is nothing unusual to the idea that the principles (arkhai) appropriate to the more descriptive parts of ethics include facts accessible to observation. Nevertheless, this does not mean there are no differences between the methods of the several inquiries.

When we investigate place (topos) in natural science, the removal of endoxic inconsistencies seems ultimately vouched for by the explanatory definition at which the enquiry aims:

What, then, place is [Ti esti] might become clear in the following way. Let us assume [labomen] about it which things are truly believed [dokei alethos] to hold of it in itself [kath’ hauto huparkhonta] … These things laid down [hupokeimenon], we must study the remaining parts. One ought to try to conduct the inquiry in such a way that the what it is [to ti estin] will be rendered, so that both the things one is perplexed about are solved [ta aporoumena luesthai] and the things which are believed to belong [ta dokounta huparkhein] to place will belong-and furthermore, the cause of the difficulty and of the perplexities about it will be evident [cf. EN 1154a22-61; for in this way one would give the best proof [an kallista deiknuoito]. (Ph 210b32-211a11)

Commentators generally point out the striking similarities with our Ethics passage. Nevertheless, from the point of view of (I), it is noteworthy, in view of the universal purport of ‘as in all other cases’ (EN 1145b3), that Aristotle does not refer to the ‘setting down’ of reputable beliefs as an independent and integral feature of the procedure to be followed. This, however, is not the only difference with the akrasia program. For whereas the latter settles for ‘sufficient’ proof (dedeigmenon hikanos) (EN 1145b6-7), the Physics program aspires to nothing short of the ‘best’ proof.

The adverbial modifier of degree ‘sufficiently’ (hikanos) refers to the methodological consequences of the metaphysical presuppositions of ethical inquiry. Aristotle more than once stresses the intricate affinities between the ontological foibles of, literally, the ‘subject-matter’ of an inquiry (he hupokeimene hule) and the degree of ‘precision’ (akribeia) or scientific rigour of exposition it is able to cope with (EN 1094b11-27; 1098a26-b4; 1103b34-1104a5):

We should also remember what has been said before, and not in like manner look for precision [akribeian] in all things, but in each kind of thing according to the underlying matter [kata ten hupokeimenin hulen] and to such an extent as [epi tosouton eph’ hoson] is appropriate to the inquiry [oikeion tei methodei. For a carpenter and a geometer look for the right angle in different ways; the former as much as it [i.e., the right angle] is useful with a view to his work [pros to ergon], the latter [inquires] what it is [ti estin] and what sort of thing it is; for he is a spectator of truth. In the same way it should also be done in other matters, that subordinate matters do not outgrow the actual works. And the explanation [ten aitian] should not be demanded in all cases alike, but in some cases proving the ‘that’ [to hoti] well is sufficient [hikanon], just as [it is the case] also about the principles [peri tas arkhas]; the ‘that’ is primary [proton] and a principle [arkhe]. Some principles are studied by induction [epagogel, others by perception [aisthesei], again others by a kind of habituation [ethismoi tint], that is, different [principles] in different ways [kai allai d’ allos]. (EN 1098a26-b4)

‘Sufficient proof’ in ethics, apparently, is ‘rightly proving the that’-which is a redescription hardly less arcane than the one we set out to clarify. ‘The that’ (to hoti), as opposed to ‘the because’ (to dihoti), refers, in Aristotelian jargon, to something’s being the case, or the ‘fact,’ in contrast with the explanation (aitia) sought for it (APo 89b23-35). Such an explanation, which is the answer to the ‘what-is-it’ question (ti esti) (Ph 210b32; 211a8), brings out the essence, or the formal cause, of the explanandum (APo 90a15-21; 31-2; 93a4). Whereas natural science aims at explanation, in ethics, ‘the that’ is accorded pride of place:

Presumably, then, we should begin from things which are familiar to us. That is why it is necessary that he who is going to listen sufficiently about things noble and just and political in general has been brought up nobly by means of habits. For the ‘that’ is a principle [arkhe] and if this sufficiently appears [phainoito] [to someone], he will not need the ‘because’ [tou dihoti] in addition. (EN 1095b3-7)

Since the nature of its subject-matter (he tou pragmatos phusis) is, as Aristotle repeatedly stresses, unsympathetic to explanations, these are not to be expected in ethics (EN 1094b25-7; APr 46a27; cf EN 1098b5; 1103b34-1104a5; 1137b14-9). (Whence do we obtain this kind of information?) ‘Sufficient proof’ in ethics, then, must be given about first principles, and ‘the that,’ or the fact of the matter, constitutes some such principle (proton, cf. Top 101a36-b4). ‘The that,’ Aristotle continues (EN 1098b3-4), is studied in its appropriate way: I take the arkhai he mentions here to cover only factual first principles, and the modes of principle-acquisition listed as those relevant to ethical inquiry.

In essentials, then, Aristotle, at EN 1145b2-7, by talking about sufficient proof, is merely restating the methodological digressions from book A. To ‘prove the that sufficiently’ is describing akrasia by testing the endoxic approximations about it, not merely against each other, but also against the facts. In the methodological reflection Aristotle prepares us for the fact that in the disquisition on akrasia explanations are not to be expected-and to do so is incumbent on him in virtue of the universal import of the Prior Analytics remarks on method (HA 491a5-14; PA 639b5- 10; 640a13-5).


The parallel passage from the Magna moralia (which is, if not a genuinely Aristotelian work, genuinely Aristotelian in spirit) is quite unequivocal as to the factual status of akrasia:

About akrasia and continence one should first state the perplexities [ta aporoumena] and the arguments contrary to [enantioumenous] the phainomena, so that, having considered the matter together from the point of view of the perplexities and the contrary arguments and having examined these, we shall see the truth about these things as far as possible [eis to endekhomenon; cf. hikanos]; for it will be easier to see what is true in this way. (MM 1200b20-4)

Interestingly, the author neither announces nor presents a catalogue of common beliefs; he immediately goes on to wipe away the absurd conclusion of Socrates’ premises:

For it is absurd that we, being convinced by such an argument, annihilate [anairein] what plausibly occurs [to pithanos ginomenon]; for there are akratic people, and, while knowing [eidotes] [that the things they are about to do are] base, they do them all the same. (MM 1200b30-2)

That there is akratic behaviour, then, is a fact which it would be absurd to argue against; it is there for us to perceive. Similarly, in natural science it cannot be proven that there is such a thing as nature. This fact is established by an empirical survey of cases (epagoge): some natural beings can simply be seen to be in motion, and it can therefore be ‘laid down’ (hupokeistho) that they are (Ph 185a12-4; 254a15-b4). To prove the existence of nature would be to prove the evident by the non-evident (phanera dia ton aphanon deiknunai), and the particular example chosen by Aristotle, of a person blind from birth attempting to prove the existence of colours, could not be more to the point (Ph 185a12-4; 193a3-9).

There is some irony in Aristotle’s claim that the Eleatic philosopher Parmenides, who is the author of a treatise About nature (Peri phuseos), is not theorizing ‘about nature’ (peri phuseos) (Ph 184b25-185a1; cf. DC 298b14-8; PA ‘642a13-7). Eleatic monism (‘what is is one and unmoved’) is incompatible with the logical behavior of the term ‘principle,’ which presupposes a plurality of beings: a principle is always a principle of something. Nature, according to the fundamental ‘supposition’ (hypothesis) of natural science, is a principle of change (kinesis) (Ph 185a1-13; 253b5-6). Parmenides, then, ‘annihilates’ (anairein) this principle. And in the more particular context of Aristotle’s discussion of the problem of generation, he is said to ‘completely annihilate’ (holds anairein) the existence of generation and corruption (Ph 185a2; DC 298b1420; cf MM 1200b25).

As to the plurality of principles, there is a remarkable tension in Parmenides’ thinking. On the one hand, a dogged confidence in the selfsufficiency of argument leads him to his radical monism. On the other hand, Parmenides, ‘looking more’ or ‘better’ (mallon blepon) than, for instance, Melissus, is ‘forced (anankazomenos) to follow the phainomena’; which compel him to ‘posit’ (tithenai) a duality of principles (‘hot’ and ‘cold’) ‘according to perception’ (kata ten aisthesin) (Met 986b27-34; cf. 984b9-11; Ph 188b29-0; PA 642a18-20; EE 1236b21-2). The deliverances of perception, then, push Parmenides some way towards a plurality of principles or causes-that is to say, towards the truth as Aristotle conceives of it.

The Eleatics fail to adopt the ‘appropriate’ method (Ph 254a30-b1; GC 325a13-5) Their accounts take on self-defeating proportions in that they rest on a mistaken transference of the stability of epistemic states to perceptible substances, which are the only ones assumed to exist; their being in flux is therefore thought to be merely apparent (DC 298b21-4). But, one could imagine Aristotle demur, if the only things there are are perceptible, why not look and conclude, first, that there is a plurality of principles of change, and second, that perceptible things do change. By ‘overlooking’ sense-perception, the Eleatics exhibit a ‘weakness of thought’ which does not very favourably emerge from a comparison with a lunatic, who certainly would not go so far as to believe that fire and ice are ‘one’ (Ph 253a32-4; GC 325a16-24). The reason why atomists like Democritus and Leucippus do not annihilate change and the plurality of beings is that they proceed according to the method of natural science: they take just what by nature exists’ for a principle or starting-point (GC 324b35-325a2).

Let us take a closer look at the relevant elements of the nomenclature in which Aristotle casts his discussions of these matters. Sometimes the terms hupothesis and arkhe stand in for each other (EE 1235b25; 30-1); moreover, in Aristotelian idiom it is also possible, perhaps somewhat redundantly, to ‘suppose principles’ (hupotithesthai arkhas). For instance, the atomists, being more experienced with the bearing of natural phenomena, are better able to do this than those who, without looking about them, revel in long-winded argument, and that is why the theories of the former are ‘appropriate’ or germane to natural science (oikeioi logoi) (GC 316a6-14; EE 121 8a8-9; 1227a8-9; b23-5; b28-30; 1235b25-31; EN 1151a16-9). In other words, they ‘agree with the phainomena’ (GC 325a23-6), which, as we have seen, SOR does not.

The verb corresponding to hupothesis, hupotithenai, sometimes retains its sense despite the amputation of its prefix. So when Aristotle, in the Posterior Analytics, analyzes the notion of demonstrative science, he lists as one of its parts “what it posits [tithetai]” to exist, being “the genus of which it studies the essential attributes,” Yet he sees no reason why scientific practice should not ignore “positing [hupotithesthai] that the genus exists, when it is evident [phaneron] that it does (for it is not likewise clear that number exists and that cold and hot do)” (APo 76b11-22). Apparently, the existence of hot and cold is not so evident to minds enchanted by the charms of logic as it is to the mentally deranged. By the same token, in the case of akrasia Socrates is lured into denying Aristotle’s description of the facts of the matter.

By now an alternative, more technical, or more specifically methodological sense of tithenai can be descried: to ‘posit,’ or to assume, what is ‘truly believed’ to hold of something (Ph 210b33), what ‘evidently exists’ (APo 76b17-8), the facts (huparkhonta) (Mem 451a19-20), or ‘the phainomena’ (EN 1145b3). Since we have already witnessed Aristotle happily replacing phainomena by huparkhonta in our quotation of a key methodological text (APr 46a20-3), we may conclude that tithenai ta phainomena has a quite legitimate sense besides ‘presenting a catalogue of common beliefs,’ viz., ‘positing (in the sense of assuming) what is evidently the case’ (Insom 459a11-2). Let us now for a moment switch back to the passage we set out with.


The methodological reflection of EN 1145b2-7 not only announces the program for the following chapters. Its ties with the immediately preceding remarks are organic, and I claim that they need to be taken into account if we want to make sense of the procedure Aristotle proposes. We may do so by giving tithenai the sense we wound up with in the last section. But before we do this, we should take a look at the customary way of forging the connection.

There is, so it is generally acknowledged, at least a syntactic linkage to be found in the demonstrative tauta in the expression ta endoxa peri tauta ta pathe. Its antecedent is usually sought in the states (hexeis) or dispositions (diatheseis) mentioned a couple of lines before (EN 1145a33-b2): akrasia, weakness (malakia), and effeminacy (truphe) on the one hand, continence (enkrateia) and perseverance (karteria) on the other.

However, our constraints on interpretation impel us to heed the punctiliousness Aristotle exhibits when, while working towards a specification of virtue (EN B 5), he distinguishes among mental attributes, in particular among the episodic affection (pathos) and the permanent state (hexis). One of the differences cited lies in the fact that moral praise and blame apply to states, not affections (EN 1105b28-106a2; 1101b12-4). Akrasia, as the first endoxon set out has it, is susceptible of blame, and must, conseqently, be a state (EN 1145b8-10; 1148b5-6; cf. 1105b28-30 with 1146a13-6). This does not sit well with the stringent interpretation we have to impose on the expression ‘reputable beliefs about these pathe.’

In view of the fact that in the introductory remarks leading up to our methodological reflection there has been only talk of states and dispositions (1145a25; 33; b1), two interdependent questions, which are customarily suppressed, conjointly urge themselves upon us. The first concerns the way to account for this sudden switch from states to affections, the second regards the antecedent of the demonstrative pronoun tauta. It is in dwelling on these questions that an adequate appraisal of the role of SOR gradually begins to take shape.

Syntactically, I submit, ta phainomena has the most recommendable credentials for performing the part of antecedent of tauta. It is not mandatory to construe pathos as ‘state of mind’ (Owen); it might likewise mean ‘happening,’ ‘what happens,’ or ‘incident.’ There is a perfectly natural way (pace Davidson) for the English equivalents of both Greek terms to refer to the same items; as when, for instance, Anthony Kenny holds that “the facts, or what happens, sets the standard by which the utterance is judged and found true or false” (cf. Met 1051b2- 9). If ta endoxa peri tauta tauter to pathe referentially (but not synonymously) equals ta endoxa per ta phainomena, then an alternative interpretation looms up. SOR, though an endoxon, is not, strictly speaking, an endoxon about the relevant phainomena. On the contrary, the argument even ‘contradicts’ them. If this interpretation holds, then Aristotle’s remark about SOR’s conflict with the phainomena reverberates with the methodological injunction compressed into the formula ta endoxa peri tauta ta pathe the endoxa should be ‘about’ the phainomena, or the ‘facts,’ and not, as Socrates (like Parmenides) turns out to do, ignore and deny them. This is a regular feature of Aristotelian method across different fields of inquiry and we find avowals of it scattered throughout the corpus.

Admittedly, this reading of the text seems, at first sight, somewhat hard to swallow. By dissolving the aforementioned grammatical links it seems to undo the claim that there is a connection of the methodological program with the remarks preceding it; we shall have to make up for this. I shall devote the remainder of this article to addressing these matters and endeavor to render the sketchy proposal just made more palatable.


Just before he discloses his methodological intentions, Aristotle warns us against conflating, among other things, the states of akrasia and vice (kakia, mokhtheria) (EN 1145a35-b2; cf. a15-7). This, I claim, is just what Aristotle takes SOR to boil down to. The translation of the argument into his own idiom shows that he construes the argument as a sacrifice of akrasia on the altar of vice, in particular, the vice of intemperance (akolasia). There is thus no place for akrasia in Socrates’ taxonomy of moral life.

Let us first tackle part (1). The endoxon of which one of the key terms is raised an aporia about tells a motivational story markedly different from that told by SOR. Akratic action is done ‘knowingly’ (eidos), but ‘through affection’ (dia topathos) (EN 1145b12-4; cf 1134a20-1; 1135b20). As already noted, the very phrasing of the aporia presupposes a commitment to the truth of this story. The words hupolambanon orthos, a more technical reformulation of the unspecific eidos, seem to confirm this. To see why this should be so, it might be illuminating to trace the genealogy of the aporia.

The Socratic assumptions, to the effect that virtue is understanding (episteme) (EN 1116b4-5; MM 1182a15-23; EE 1216b2-8) and that understanding cannot be ousted by other types of mental states, entail that wrong action must be caused by ignorance (di’agnoian) (EN 1145b26-7). In his discusson of involuntary action in EN 3.1 Aristotle stipulates that an action is involuntary when, among other reasons, it is caused by (one type of) ignorance (di’agnoian) (EN 1110b18-1111a21). The phrase ‘through ignorance’ is dissected according to the structure of the practical syllogism, yielding two interpretations for the ignorance of SOR. The source of error may be pinpointed either in the universal major premise, typically cast in prescriptive form (dei(n)) and representing deliberate choice (he en tei prohairesei agnoia) (EN 1110b31-3; 1150a20; 24-5; 1152a5-6); or in the particular minor premise concerning the concrete circumstances of the action (EN 1110b31 – 1111a1). To exploit one of Aristotle’s own examples: the major ‘one ought to taste everything sweet’ may express prohairetic ignorance or ignorance of the universal; the minor ‘this here is sweet,’ as I call it, ‘situational ignorance’ (EN 1147a29-31; cf. 1142a20-3).

Now by Aristotle’s own distinctions ignorance and affection mutually exclude one another as causes of action. If one acts ‘through affection’ (dia pathos) (e.g., anger), this means that one does not act ‘through,’ but ‘in ignorance’ (agnoon) (EN 1110b24-7). If the akratic, then, acts ‘through affection,’ as the endoxon has it, this entails that he does not act through ignorance, but in ignorance. It now becomes clear why Owen’s paraphrase of SOR (‘in ignorance’) is inaccurate: it is crucial for an adequate understanding of the passage that we render ‘through ignorance,’ thereby excluding, in Aristotelian terms, the motivational efficacy of affection in SOR. ‘In ignorance,’ on the other hand, is perfectly compatible with an affective cause (EN 1136a6-9).

However, in so far as the akratic acts through affection, this also means that he does not act in knowledge (ouk eidos) (EN1110b26-7). The problem defined by these terms, therefore, consists in finding an interpretation of the eidos of the reputable belief, which is reconcilable with its motivational story. This is why the aporia is formulated the way it is. On the other hand, sheer logic forces Socrates to maintain that ignorance must be the cause, or principle (arkhe) (EN 1134b20- 1), of morally reprehensible action. So far for part (1).


Part (2) latches onto just this aspect of SOR. The vicious person, according to Aristotle, “is ignorant [agnoei] of what he ought [dei] to do and to leave” (EN 1110b28-9). According to Aristotle, it is exactly this sort of ignorance (i.e., that concerning the universal) SOR is pointing to. What Aristotle tries to convey by the explanatory remark (gar) concluding part (2) (EN 1145b30- 1) is that the facts do not support a reduction of akrasia to intemperance. For this would, on Aristotelian premises, require that the akratic is motivated by a deliberate choice.

Akrasia and intemperance are indistinguishable in virtue of their respective objects, bodily pleasure (EN 1118a28-b8; 1146b19-23; 1148a4-8; 1149a21-2; 150a9-11), but they can be prised apart as motivational properties of the subject (EN 1146b22-4; 1148a16-7; 1150a23-7; 1151a20-6; 1152a4-6). It is evident (phaneron) that akrasia and vice are not identical, because the latter involves action by deliberate choice (prohairesis), whereas the former issues in a similar action, yet in defiance of deliberate choice (EN 1151a5-7). As Aristotle puts it in one place, the intemperate “is led [agetai] while deliberately choosing [prohairoumenos], holding that one always ought [aei dein] to pursue the object of pleasure immediately present; but the akratic does not mean [ouk oietai] [to do it], but pursues it” (EN 1146b22-4; cf. 1110b28-33; 1114a31-6; b5).

Since someone who is acting akratically evidently (phaneron) does not mean (ouk oietai) to behave in the way he does before (prin) he is overtaken by appetite, this cannot be ignorance of the universal expressed in a premeditated choice (EN 1135b8- 11). (During the entire discussion of akrasia, the verb oiomai [which I render ‘to mean’] acts as a stand-in for prohaireisthai [‘to deliberately choose’]; they are, however, not synonymous verbs [EN 1146b23-4; 1148a17; 1151a4-6; 1152a5-6]). This explanatory remark, I claim, answers the foregoing directive to ‘inquire about what happens’ (deon zitein peri to pathos). What exactly happens to a person currently manifesting akratic behaviour (ho akrateuomenos)?

It is crucial to read the immediately following indirect question ‘dynamically’; Aristotle asks what mode (tropos) of ignorance comes into being (ginetai) (cf. EN 1128b22). Since the term tropos can be taken to refer to a premise of the practical syllogism (EN 1146b35-1147a1; 8; cf. 1152a15-6), Aristotle might be taken to be gesturing towards the twofold analysis to which the expression di’agnoian is subject. If it is designed, by its very terms, to rule out a pre-existing ignorance in the akratic, the question is to be construed as rhetorical; and since only situational ignorance can be at stake, this rhetorical question is supposed to be fatal to SOR.

To paraphrase part (2). If we attentively study akratic behaviour, it is more to the point to ask (if it is at all the case, as Socrates says, that this behaviour is caused by ignorance) what form of ignorance pops into existence at the moment the subject is overtaken by occurrent akrasia. For if there is any motivational role for ignorance, it cannot be the form of ignorance Socrates has in mind; since, for the Socratic reduction to work, this must be prohairetic ignorance, which requires that the pursuit of pleasure be deliberately chosenwhich is evidently not the case. The facts, then, go so far as to refute Socrates’ self-consistent reduction of akrasia to vice. And that is all part (2) aspires to; for all it needs to show, to dispense with the ‘counterfactual’ conclusion of SOR, is that the akratic is, in fact, not moved by prohairetic ignorance.

At the same time, however, the question leaves some scope for the other form of ignorance, situational ignorance, to play a part in the analysis of akrasia-and it turns out that it does play such a part (EN 1147b9- 17). But then again, it is not the motivational part, because this, on Aristotle’s terms, is occupied by appetite (epithumia) (EN 1147a35-b3). Yet to assign the cause to appetite is quite compatible with a situational ignorance as a concomitant. Still, if akrasia, in some way or other, involves ignorance, then we should specify its role by attending to the facts, and not, as Socrates, be guided by unjustified assumptions.


This is all very well, one might remonstrate, but does not the foregoing merely corroborate the view that SOR conflicts with endoxa, viz., that endoxon which assigns the cause of the akratic action to affection? And does not, therefore, SOR conflict with beliefs, not with facts?

Obviously, there is such a conflict. However, we proposed to give a peculiar weight to the expression ta endoxa peri tauta ta pathe (cf Top 101b1). So the relevant endoxon, being in concert with ‘what happens,’ or the facts, has to be about those facts in the right way. ‘In the right way,’ that is, in the way that someone who is supposed to construct theories ‘about nature’ is to assume its existence and not, like Parmenides, deny it. This is just the implication of the Socratic reduction of akrasia to vice: it denies akrasia as an independent fact of moral life, and, as we have seen, the sentence immediately preceding the methodological reflection admonishes us not to identify akrasia with intemperance.

Hence, when Aristotle implores (deon) us to ‘inquire about the pathos,’ he is conforming to the instructions of EN 1145b2-7: we must (dei) strive for coherence among those endoxa which reflect, more or less accurately, the facts of the matter. SOR’s conflict with the phainomena, then, is a conflict with a particular belief, certainly; but it is so only by courtesy of the fact that that belief is about the phainomena. It presupposes the facts, whereas SOR denies them. This explains why Aristotle does not undertake it to save the endoxon identifying akrasia and intemperance (EN 1145b 14-7); for he has committed himself only to save the endoxa peri tauta ta pathe, and it emerges that this endoxon is not among them (EN 1151b32 -1152a6).

But how to account for the fact, crucial to (I), that SOR is ultimately upheld? Well, this is not a fact at all. Commentators too willingly conflate the argument as a whole (logos) and a single premise. It is the first which contradicts the facts, not the latter. It is only the assumption that the sway of understanding cannot be flouted by any mental phenomenon whatsoever that is, in the end, ‘left standing’ (EN 1147b14-7). That these two should be kept apart is shown by the fact that the preliminary rehearsal of the aporia merely considers the options for the epistemic condition of the akratic. But it is still clearer from the parallel passage from the Magna moralia (MM 1200b25-37). Its author first exposes Socrates’ position as a whole as contrary to the facts, and then, assuming the existence of akrasia, goes on to raise the question as to whether understanding is the likely candidate to substitute for ‘knowing’ (cf. Rh 1357a7-13).

In the end Aristotle makes no commitment whatsoever to those parts of SOR which state that the action is caused by ignorance, that one cannot act wrongly knowingly, or that akrasia does not exist. However, the fact that the assumption about the supremacy of understanding is ‘left standing,’ which makes it an endoxon in the sense of EN 1145b2-7, renders its absence from the initial doxography ‘set out’ on (I) all the more perplexing. I conclude, therefore, that Aristotle’s evaluation of Socrates’ position, insofar as it goes beyond the assumption of the supremacy of understanding (i.e., EN 1145b25-31), should be treated as a parenthesis. The considerations invoked (in section 2) to support the claim that SOR is endoxic, then, apply only to that assumption, and not to the argument in toto.

As one scholar notes, there certainly is a mild irony to Aristotle’s concluding remark that what Socrates sought to establish “seems to sumbainein,” namely, that understanding is not vanquished by affection (EN 1147b14-5). The Greek verb sumbainein does not only have the logical sense of ‘to follow’; it also has the factual sense of ‘to happen.’ What is more, it may convey the coincidental character of the fact it refers to: to take a threadbare example, stumbling upon a treasure while digging a hole to plant a tree (Met 1025a15-9). Consequently, one might construe Aristotle’s remark so as to claim that Socrates ‘coincidentally’ stumbles upon the truth. More interestingly, however, its present participle sumbainonta is sometimes used by Aristotle in the same way as phainomena in its sense of ‘facts’ (cf. MA 701a8-9; DA 403a19-20). The irony, then, is that, in the particular case of Socrates, the assumption of the supremacy of understanding is not at all controverted by those facts which also show up the existence of akrasia as something having certain characteristics in common, though far from being identical, with intemperance.


My counter-suggestion, then, is that the expression tithenai ta phainomena does not refer to assembling or presenting a doxography on a certain subject.

Undeniably, Aristotle does draw up such a catalogue; but he merely does so to structure the subsequent discussion. Instead, the expression refers to ‘laying down’ or assuming the facts defining the parameters of the diaporetic quest for endoxic coherence. The distribution of dialectical pressure presupposed by raising an aporia about ‘knowing’ is nothing but an implicit ‘laying down’ of the phainomena, directing the preliminary rehearsal of the difficulties (and the rest of the discussion). A decision about the assignment of motivational efficacy has already been made-and that it has been made is shown by the fact that Aristotle starts by throwing up a perplexity about ‘in knowledge,’ not about ‘through affection.’ Thus the twofold collocation of phainomena with pathos (EN 1145b3-5; 28-9) seems to me far more significant than the single juxtaposition of phainomena and endoxa.

The remaining task is to give the alternative picture of method sketched by Aristotle. This is quite simple. We only need to amend the picture given in the first section by a statement to the effect that it presupposes an acquaintance or familiarity with the facts of the matter and their ontological behaviour. The state instilled in us by this procedure is called ‘experience’ (empeiria) and it enables us to ‘lay down’ or ‘assume’ the facts, against which the rest of the analysis is to be tested.

I am aware of the fact that my proposal, pointing backwards in the direction of Ross’ largely discredited interpretation, our point of departure, is hopelessly lacking in fashionability. Some may call it dully traditional (in the depreciatory sense, of course), or even reactionary. Yet I tend to think that Ross, in this case, points out to us a better route to Aristotle’s method in ethical inquiry than many others do. In any case, what I hope to have shown is that we can make sense of the program of EN 1145b2-7 without assigning to two of its key terms a shared reference.