¹ See Barnes, Jonathan, ‘Aristotle's Concept of Mind’, Proceedings of the Aristotelian Society, lxxii (1971–1972), 101–10Google Scholar; reprinted in Articles on Aristotle, ed. Jonathan Barnes et al. (London, 1975–9), iv.32–41. All references are to the latter pagination.

² Hicks, R. D., Aristotle: De Anima (Cambridge, 1908), p. 275Google Scholar, offers the received understanding of Aristotle's concern: ‘We are dealing with certain ἒργα και πρζεισ, or πθη και εργα We have to determine what is the υποκειμενον or logical subject to which they should be attributed.’ For similar interpretations, see Sorabji, Richard, ‘Soul and Body in Aristotle', Philosophy 49 (1974), 55–79CrossRefGoogle ScholarPubMed (see esp. p. 73), reprinted in Articles on Aristotle, iv.42–64; Modrak, D. K. W., Aristotle: The Power of Perception (Chicago, 1987), pp. 115, 153Google Scholar; and Owens, Joseph, ‘Aristotelian Ethics, Medicine, and the Changing Nature of Man’, The Concept Papers of Joseph Owens, ed., Catan, J. R. (Albany, 1981), p. 174Google Scholar, and ‘Aristotle's Definition of Soul’, ibid., p. 211 n. 34.

³ Many other passages violate Aristotle's stricture. See, e.g., 418a14, 427a20.

⁴ Hamlyn, D. W., Aristotle's De Anima Books II and III (Oxford, 1968), p. 81Google Scholar.

⁵ See Fine, Gail, ‘Plato and Aristotle on Form and Substance’, Proceedings of the Cambridge Philological Society 209 (1983), 23–47CrossRefGoogle Scholar.

⁶ Ross, W. D., Aristotle's De Anima (Oxford, 1961), pp. 212–13Google Scholar.

⁷ For an excellent discussion of Aristotle's criteria for ουσια, see Fine, art. cit. 26–30.

⁸ Of course, the evidence that the soul is a substance is not limited to the DA and Met.; but most of it will be found in those works.

⁹ See Ross, W. D., Aristotle: Metaphysics (Oxford, 1924), ii.213Google Scholar.

¹⁰ Or so most commentators suppose. See e.g. Ross, , Metaphysics, ii.164, 209Google Scholar. If ζῷον is the form, then this passage presents no problem for my view. I argue on the supposition that ζῷον compound in order to show that even then there would be no problem for my view that forms are υποκειμενα.

¹¹ This is strongly confirmed by Aristotle's summary in Met. VIII 1: ‘There is substance as υποκειμενον, in one way as matter (by matter I mean that which, while not a τóδε τι in actuality, is a τóδε τι in potentiality), and in another as λóγοσ or shape, which being a τóδε τι is separate’ (1042a26–9). Here Aristotle clearly notes that there are two distinct types of υποκειμενον and marks the difference in terms of their being (or failing to be) τóδε τι in actuality. Because the form is a τóδε τι in actuality, it is a υποκειμενον₂.

¹² Bonitz, Hermann, Aristotelische Studien (Hildesheim, 1969), p. 301Google Scholar.

¹³ Here I disagree with Ross, Metaphysics ii.164, 208. He paraphrases the beginning of Met. VII 3 as: ‘We showed [substratum] to underlie in two ways, as the thing underlies accidents and as the matter underlies the actuality.’ I do not see that the πθη of 1038b6 must all be accidental.

¹⁴ Cf. Brunschwig, Jacques, ‘La Forme, Prédicat de la matière?’, Études sur la Métaphysique d'Aristote, ed. Aubenque, Pierre (Paris, 1979), pp. 131–58Google Scholar. When confronted with a problem similar to the inconsistent triad I have constructed, Brunschwig argues that Aristotle has not two types of υποκειμενα in mind, but rather two types of predication. He suggests that form is not predicated of matter in the way the properties are predicated of substances, and so that in the former case Aristotle uses ‘predication’ in an extended sense. But 1029a23ff. requires one sense of ‘predication’ for both types of υποκειμενα.

¹⁵ Others have recognized that forms count as υποκειμενα for Aristotle. M. Frede, for example, has suggested that it is ‘most puzzling…that there is a way in which substantial forms might be construed as the ultimate subjects and, hence, as the real things as opposed to mere properties of things’ (‘Substance in Aristotle's Metaphysics’, in his Essays in Ancient Philosophy [Minneapolis, 1987], 75)Google Scholar. He resolves this puzzle by claiming that form can be understood as ‘ultimate subject’ since ‘the form [is] the centerpiece of the cluster of entities that constitute the concrete object’ (p. 77). But Frede evidently does not believe that Aristotle recognizes two distinct types of υποκειμενα Although he acknowledges that ‘1029a23–4 suggests that matter is the ultimate subject by being the subject of the substance in question which, in turn, is the subject of the non-substantial entities’ (75), he concludes that form is the ultimate subject (p. 77) only to wonder how forms can be both ultimate subjects and universals, since ‘it is of the very nature of ultimate subjects that they cannot be predicated and, hence, cannot be universal’ (p.77). He rightly concludes that this is a pseudo-problem, since Aristotelian substantial forms are particulars, and not universals; but he assimilates what I have called υποκειμενα to υποκειμενα in claiming that ultimate subjects cannot be predicated. Met. 1029a21–5 requires that substantial forms be predicated of matter; we have seen numerous other passages which require that they nevertheless be υποκειμενα Frede's final analysis, then, is inconsistent with Met. 1029a21–5, in so far as it presupposes a univocal account of theυποκειμενον.

¹⁶ Hicks, , op. cit., p. 275Google Scholar.

¹⁷ All of the examples of predicates Aristotle provides in this passage are what Strawson calls P-predicates. While not necessarily ascribing states of consciousness, P-predicates ‘imply the possession of consciousness on the part of that to which they are ascribed’ (Strawson, P. F., Individuals [London, 1979], p. 105)Google Scholar. M-predicates, e.g. weighing 108 pounds, by contrast, imply that their subject are bodies. There are, then, two classes of P-predicates: those which beyond entailing the possession of consciousness entail the possession of a body as well (P2-predicates), and those which do not (P-predicates). Using this notation, a precise formulation of Hicks’ first premise would be: an entity can have Pj-predicates only if it has P2-predicates. For our purposes, however, it will be simpler to ignore this complication and to read Hicks’ first premise as equivalent to the claim that nothing can have P-predicates if it does not also have M-predicates, and this will not compromise the criticism I offer of his interpretation.

¹⁸ See, e.g., Tye, Michael, ‘On the Possibility of Disembodied Existence’, Australasian Journal of Philosophy 61 (1983), 275–82CrossRefGoogle Scholar.

¹⁹ It may be objected that there are salient disanalogies between souls, as characterized in the: De Anima, and the god of Met. XII, and that consequently the evidence I have cited against Hicks’ analysis is irrelevant. This objection is not compelling: (a) premise (1) of Hicks' argument remains false, whether or not it was recognized to be so; and (b) the evidence of Met. XII demonstrates that Aristotle indeed recognizes its falsity. It would be wrong, of course, to; presume that souls have many characteristics in common with Aristotle's god (although cf. my ‘Soul and Body in Aristotle’, Oxford Studies in Ancient Philosophy vi [1988])Google Scholar; but this; presumption is not required for the present argument.

²⁰ Some commentators will also raise reasonable questions regarding the second premise., Why should it be absurd to regard weaving and housebuilding as actions performed by the soul? Some will argue that this sort of claim will obtain only if souls or Aristotelian forms generally are immaterial; in such a case, it might reasonably be suggested that souls alone cannot perform such feats. But the assumption that souls and other particular forms are immaterial in Aristotle has been widely doubted. Many commentators argue that particular forms in Aristotle are constituted by matter, especially when this is construed as proximate matter, and so are themselves material. See e.g. Wiggins, David, Sameness and Substance (Cambridge, MA, 1980)Google Scholar, and Loux, Michael J., ‘Form, Species and Predication in Metaphysics Z, H ανδ π’, Mind (1979), 11 n. 2Google Scholar.

²¹ This theory was evidently advanced by Xenocrates; for evidence, see Rodier, ii. 138–9. Zeller remarks, ‘Again, it [the soul] is not a number that moves itself, for it is not a number, and, if it were a number it certainly could not do so.’ See Zeller, Eduard, Aristotle and the Earlier Peripatetics, trans. Costelle, B. C. F. and Muirhead, J. H. (New York, 1962), ii.4Google Scholar.

²² Hicks supposes that 408a29–b29 is essentially a digression: ‘At this point [408a29] comes a pause in the criticism and refutation. The two theories of harmony and circular motion are dismissed, but, before A. proceeds to refute the self-moving number of Xenocrates, he stops to consider once more if motion can be attributed to the soul and, if so, in what sense this is possible’ (273). He adds, more directly, ‘The whole passage 408a34–b29… is more or less of a digression’ (p. 279). On my analysis, the passage does represent a pause in the refutation, but is not a digression; rather, Aristotle means to guard himself against a dialectical backlash resulting from his criticisms.

²³ See Hicks, , op. cit., p. 263Google Scholar.

²⁴ I regard ὡσ εἰπειν as qualifying πονμουσι, as opposed to πντεσ as is claimed by Hicks, p. 266 and Rodier, G., Aristote: Traité de Lâme (Paris, 1900), ii.124Google Scholar and presupposed in the Oxford translation of Smith, πονμω is not a word we find elsewhere in the DA, nor even very frequently in other works of Aristotle. The point would seem to be that everyone agrees that motion is to be explained by reference to the soul, even though, according to Aristotle, the soul itself does not move καθ' αυτ This more circumspect way of speaking frees him from claiming directly that the soul moves καθ' αυτ.

²⁵ Two additional points are relevant to our assessment of this passage. First, Aristotle evidently expresses some caution in saying that it is ‘perhaps’ (ἲσωσ better to speak in this way. The source of his reservation is unclear, but my own view is that he is concerned that he will be construed (as indeed he has been) as denying that the soul is a υποκειμενον, something he cannot afford to do given its substantiality. He may also express caution because one might equally say that the soul itself perceives an d so forth, but thataisthêsis does not count as a species of motion (that is, one can deny the second as well as the first premise of the dialectical argument under consideration). Indeed, this would seem to be Aristotle's considered view. H e is quite careful throughout the DA to avoid saying that αἲσθησισ involves an affection of the soul; he prefers to say that it is affected in a certain way (πσχειν rather than merely πσχειν). Importantly, at DA 417b14–16 he suggests ‘either one should not say that it [viz. αἲσθησισ is alteration, or that there are two types of alteration’. Cf. Furley, D., ‘Self Movers’, in Lloyd, G. E. R. and Owen, G. E. L. (edd.), Aristotle on the Mind and Senses (Cambridge, 1978), pp. 176–7Google Scholar. Aristotle's point in DA I prefigures the ambivalence to be found in later books, an d should no t be received as evidence of a contradiction.

Second, not only does Aristotle refrain from denying that the soul is a υποκειμενον₂ in DA I, he employs locutions which seem to presuppose that it is: he says that recollections originate from the soul (π' κεινησ [sc. ψυχσ] DA 408b 15–16), that the body is moved υρò τσ psi;υχσ at DA 406a34, and even grants that cetain actions are effected υρò τσ psi;υχσ at DA 408b7. These locutions suggest that the soul originates motion without moving itself; an d in each case it seems plain that the soul is a υποκειμενον₂ of an action.

²⁶ The dative locution τῇ ψυχῇ has misled some commentators and inclined them toward an unacceptable reading, because they have presupposed that it must be taken instrumentally. Beyond landing Aristotle in a contradiction, the prevalent instrumental reading disregards the cautionary note which follows his remark. After claiming that a man performs various actions τῇ ψυχῇ, Aristotle adds, ‘but this not in the sense of there being motion in it [i.e. the soul], but [in the sense of motion's] sometimes extending to it or [originating] from it’ (DA 408b15–16). That the soul be in a certain state is causally basic to there being an instance of αἲσθησισ or νμνηστσ; ψυχῇ should for this reason be taken causally. A rough paraphrase will then be: we should say that a man perceives or thinks because his soul is in a certain state — but this doesn't mean that it moves καθ αυτó in being in this state; rather the body moves καθ αυτó and the soul κατ συμβεβηκóσ. Aristotle's remark is then similar to saying, ‘Typing involves the use of a typewriter, but one types because of some activity of the brain. Of course, we should not say that one's brain types, but that an individual types because of some brain activity'. The instrumental reading would treat the ψυχ as an instrument of the composite, much in the same way that a typewriter is an instrument of a typist. But this way of reading τῇ ψυχ makes Aristotle's cautionary note perverse and, more importantly, contradicts his conviction that the body is the organ of the soul and not vice versa. The activity of the soul is explanatorily basic in describing the organization of the body and so of the compound (DA 415b19; cf. DA 407b25 and De Partibus Animalium 642a 11).

²⁷ This explains Aristotle's comment ‘but this not in the sense of there being motion in it [i.e. the soul], but [in the sense of motion's] sometimes extending to it or [originating] from it (DA,. 408b15—16). He means to point out that though the compound moves καθ' αυτó, the ψυχ does not so move. It is unclear why he would make such a remark if his primary, or even subsidiary, k. concern were with the soul's ability to be a υποκειμενον.

²⁸ This interpretation does not attempt to justify or even characterize Aristotle's claim that the soul is not a μγεθοσ, or his related suggestion that the soul cannot be moved καθ' αυτó, even though these premises are obviously crucial to his argument. See my ‘Soul and Body in Aristotle’ (above [note 19]) for a more detailed analysis of these claims. But for the present I can suggest that my interpretation explains a feature of Aristotle's analysis which Furley (art. cit., 177) finds unmotivated: ‘The point is that external objects are not in themselves sufficient causes for the voluntary movements of animals. But they do have some effect on the soul, and it would be obstinate of Aristotle to deny that the effect can be called a movement.’ It will not be obstinate or in any sense capricious of Aristotle to hold this view if it is entailed by his ontological commitments rather than by his account of motion as such; and the analysis I have given suggests that this is so.

²⁹ See n. 2 above. Some of the ancient commentators came much nearer the mark in discussing DA 408b11—15. See especially Sophonias, , In Libros Aristotelis De Anima Paraphrasis, ed. Hayduck, M. (Berlin, 1882), 27.29–28.3Google Scholar. See also Alexander, , De an. lib. alt. 104.38Google Scholar:‘πασαι γρ αἱ λεγóμεναι τσ ψνχσ κινσεισ του συναμϕοτερου του ζωντóσ εισιν.’

³⁰ I am pleased to thank T. Irwin, G. Fine, P. Mitsis, J. Ackrill, and the Editors for excellent comments on portions of earlier drafts of this work.