avicenna on existence

A v i c e n n a on Exis tence

A L L A N B A C K

I AM GOING TO PRESENT an Aristotelian view of predication at odds with many current interpretat ions o f Aristotle and views of predication, x Yet the advantage o f this view is that it vindicates, or at least renders intelligible, certain doctrines and texts in Aristotle that appear baffling or careless f rom a modern standpoint. One such case is the fact that Aristotle seems to use an existential import assumption in his syllogistic without formulat ing that assumption and using it as a premise. Thus, his proof that a particular affirmative (I) proposit ion follows f rom a universal affirmative (A) proposition seems to be invalid. Again, Aristotle claims that, al though there are ten categories for all the things "that there are," there still is a science of being qua being, that transcends all these categories. On the face of it, what is real about being qua being should be in one of the categories. Aristotle's discov- ery of a focal meaning (to pros hen) of being, beyond the categories, seems a bit forced. ~ I shall show how such problems are resolved, on the basis of a certain view about predication. Tha t view appears in full force in works of Ibn Sina (Avicenna); there are traces of it earlier, back to Aristotle himself. I am going to sketch out that view, and show its merits as an interpretation of

Those who hold such an interpretation of Aristotle's views on predication include: Wil- liam and Martha Kneale, The Development of Logic (Oxford, 1962), 58-59, 64-66; J. L. Ackrill, Aristotle's Categories and De Interpretatione (Oxford, 1963), a23; Manley Thompson, "On Aristo- tle's Square of Opposition", in Aristotle, ed. J. Moravcsik (Garden City, 1967), 52-53, 7o; Jan Lukasiewicz, Aristotle's Syllogistic (Oxford, 195 0, 13o. All these hold the view, which I also attribute to al-Farabi in section 2, that, al though 'is' in 'S is' makes an existence claim, 'is' in 'S is P' indicates only the relation of predication, between subject and predicate. ! shall discuss below some of the modern interpretations that partly agree with the view that I shall be discussing.

On this view of Aristotle's metaphysical enterprise, cf. G. E. L. Owen, "Logic and Meta- physics in some Earlier Works of Aristotle", in Aristotle and Plato in Mid-Fourth Century, ed. Diiring & Owen (C~teborg, a96o), 185, where Owen claims that Aristotle in his early works, including On Interpretation, holds that "to be is to be either a substance of some sort or a relation or a quality or a member of some other category. There is no sense to the claim that something exists over and above one of the particular senses." Owen thinks that Aristotle had changed his mind when he proposed a science of being qua being.

[351]

35 ~ JOURNAL OF THE HISTORY OF PHILOSOPHY 25:3 JULY 1987

Aristotelian doctr ine. I shall not be primarily concerned here with the question, whether or not Aristotle himself held this view. 3

1.

Before relating this Aristotelian view of predicat ion to the works o f Ibn Sina and al-Farabi, I would like to sketch its basic outline. In brief, it claims that the s t ructure o f a simple categorical proposit ion, 'S is P' or 'S P's', is: 'S is (existent) as a P'. T o a certain point, this view is clearly Aristotle's: sentences like 'Socrates walks' or 'Socrates is walking' are composed of a subject term, a predicate term, and a copula, be it explicit or implicit. 4 One perhaps novel fea ture is how the syntax o f such a proposi t ion is to be s tructured: with the usual subject t e rm ('S'), the copula ('is') is taken as the verb, and the predicate ('P') is taken as a de te rmina t ion of the copula, in an accusative of respect or in some o the r grammatical construction; S is, in respect of being P. Here the copula always makes the existence claim, that the subject S exists, regard- less o f whe ther or not there is a predicate besides the copula. On this view, then, 'S is P', as it means 'S is existent as a P', implies 'S is', that is, 'S exists'. T h e predicate, if used, gives a de terminat ion of the respect in which the subject is: S is ex i s t en t - -How?- -a s a P.

T h e r e seem to have been two main features that led to the development , or, at any rate, the explicit formula t ion o f this view. First, there are texts in Aristotle that suggest it. For example, in the Categories, Aristotle says that 'Socrates is healthy' and 'Socrates is sick' are both false when Socrates does not exist. 5 So Aristotle seems to be saying that both these proposit ions have as a t ru th condit ion the claim that Socrates exists. Hence, he suggests that, in general, simple categorical proposit ions have as a t ru th condit ion the existence of the subject. Again, Aristotle gives forms like 'S is' and 'S is not ' as the most general forms o f propositions. 6 Again, in the many places where Aristotle considers the fallacy o f secundrum quid ad simpliciter (converse accident), he discusses the relations between proposit ions o f the form, 'S is P', and those o f the form, 'S is'Y He seems to sanction the inference f rom 'S is P'

3 I intend to discuss this issue fully elsewhere. 4 On Interpretation, x9b19-9o; 2oa3- 5. 5 Categories, 13b18- 9. 6 On Interpretation, 19b14- 9. As G. E. L. Owen remarks, "Aristotle on the Snares of Ontol-

ogy", in New Essays on Plato and Aristotle, ed. Bambrough (New York, 1965), 86, modern commentators find this passage difficult to explain.

Aristotle also remarks, at 16b 14-5, that indefinite verbs apply equally to what is and to what is not, and so hints that determinate verbs apply only to what is existent.

7 E.g., On Interpretation, ~ la25-33; Sophistical Refutations, 167al- 4.

AVICENNA ON EXISTENCE 353

tO 'S is', while seeking to exclude those instances that lead to fallacy. Texts such as these might suggest that the predicate 'P' is something added onto, or predicated additionally, with respect to the simpler assertion, 'S is'.

The other feature that may have encouraged the development of this Aristotelian view concerns Arabic. When Aristotle's works came to be translated into the Arabic language, there was difficulty and controversy about how to provide for an explicit copula in Arabic. Normally a copula is implicit, in both nominal and verbal Arabic sentences. The solution generally adopted was to use 'mawj~d' ('existent' or 'present') to represent the Greek copula 'estin', and to put the normal predicate in an accusative of respect. 8 But then a sentence like 'Socrates is white' would read literally as 'Socrates is existent as white'. So, on the one hand, such a translation would encourage the development of the view of predication that we are considering. But, on the other, and more interestingly philosophically, that translation may have been encouraged by what were perceived to have been Aristotle's views of the structure of categorical propositions.

2 .

I now shall turn to the works of Ibn Sina where this view of predication is developed and presented. I shall suggest that, despite Ibn Sina's admirat ion and reiteration of the views of al-Farabi, he departed from al-Farabi in advancing this view of predication.

A1-Farabi has a view of predication that accords with the standard interpretation o f predication in Aristotle. On that interpretation, al though 'is' in 'S is' makes an existence claim, 'is' in 'S is P' merely indicates the relation of predication, between subject and predicateY Comment ing on On Interpreta-

tion 19b19-eo, where Aristotle considers sentences of the form, 'S is P', and where, he says, the copula 'is' ( 'mawjad' in Arabic) is "predicated in addition (proskat~goroumenon)," al-Farabi says: "The hyparctic verb (sc., the copula) is not itself predicated, and not principally in tended to be predicated, but is predicated only for the sake of something else. '' '~ Thus, al-Farabi seems to be taking 'proskat~goroumenon' to mean 'incidentally predicated', as op- posed to 'additionally predicated' . By this distinction, I mean that a term is

s Cf. F. W. Zimmermann, AI-Farabi's Commentary and Short Treatise on Aristotle's De Interpreta- tione (London, 198 0, pp. xliv-v; E. M. Goichon, La Distinction De L'essence et De L'existence D'aprks Avicenne (Paris, 1937), 14- 5.

0 Cf. n. 1. Nicholas Rescher, "A1-Farabi: Is Existence a Predicate?", in Studies in the History of Arabic Logic (Pittsburgh, 1963), 4 o, cites a text where al-Farabi denies that'S is' makes a "real" predicational claim.

'~ Translated by Zimmermann, AI-Farabi's Commentary, lox; W. Kutsch & S. Morrow, eds., Al-Farabi's Commentary on Aristotle's Peri Hermeneias (Beirut, 196o), lo 5, 15-16.

3 5 4 J O U R N A L OF THE H I S T O R Y OF P H I L O S O P H Y ~ 5 : 3 J U L Y 1987

additionally predicated when it is in a complex predicate and may be truly predicated of the subject separately. E.g., 'white' in 'Socrates is a white man' is additionally predicated, as it follows from that proposition that Socrates is white. A term is incidentally predicated when it is in a complex predicate but may not be truly predicated of the subject separately. E.g., 'good' in 'Socrates is a good cobbler' is incidentally but not additionally predicated, as it does not follow from that proposition that Socrates is good. So, for al-Farabi, in 'Socrates is just', 'is' is not additionally predicated, by itself, of Socrates. Thus it need not follow from 'Socrates is just' that Socrates is (existent).

A1-Farabi gives truth conditions for the various forms of simple categorical propositions consistent with this analysis. To be sure, 'mawjfid', the Arabic word used to represent the copula ('estin'), normally means 'existent'. Yet al-Farabi does not say that the presence of 'mawjfld' in a tripartite sentence, that is, one of form, 'S is P', requires there to be an existence claim. For he does not give a statement about the existence of subject or predicate as a truth condition for any form of categorical proposition. Indeed, Zim- mermann, anxious to stress that the copula in a tripartite sentence is not used existentially for al-Farabi, uses 'hyparctic' instead of 'existential' to translate 'mawj~d' above."

In several texts Ibn Sina appears to agree with, if not repeat, this view of predication, given by al-Farabi. Following the latter's discussion of On Inter- pretation 19b19-2o, Ibn Sina remarks that Arabic does not have an explicit copula, as Greek does, and notes three ways in which the copula may be expressed in Arabic. '3 Citing "the learned," among whom al-Farabi is no doubt pre-eminent, Ibn Sina claims that in 'S is P' 'is' does not make an existential determination of the subject. Rather, in such cases the copula is not itself predicated, but instead indicates the relation of predication, between 'S' and ,p,,4 The copula, then, is predicated incidentally and not additionally. Again, Ibn Sina tends to characterize 'every S is P' as 'everything that is described as being S is described as being p,.,5 Here too the copula seems to be taken as indicating only the relation of predicate to subject. So, in these texts, Ibn Sina, like al-Farabi, takes 'S is' to be an

H For such a use of 'proskat~goroumenon', cf. Alexander of Aphrodisias, In Aristotelis Analyticorum Priorum Librum I Commentaria, ed. Wallies (Berlin, 1883), 369, 34-37 o, 6. Also cf. the use of 'pross~mainon' at On Interpretation, 16b6, and Ammonius, In Aristotelis De Interpreta- tione Commentarium, ed. Busse (Berlin, 1897), 47, aoff.

'* Zimmermann, A1-Farabi's Commentary xlv, Ix n. t . ,3 Ibn Sina, AI-BurM~n, ed. Affifi & Madkhour (Cairo, 1956), 77, 3-9- ~4 Ibid, 77, 4 and 12-13. ,5 Ibn Sina, AI-Q~y6a, ed. Madkhour (Cairo, 1972 ), 2t , 13ff; Kit~b al-Najat, ed. Kfirdi (Cairo,

1938), 13; Mantiq al-Mashr~q~yfln (Cairo, t973), 24.

AVICENNA ON EXISTENCE 3 5 5

existential j udgmen t , but claims that, in 'S is P', 'is' is predicated only incidentally. In this way, ' H o m e r is a poet ' can bc true, even though ' H o m e r is (existent)' is false. So the c o m m o n claim made about Ibn Sina, that he was a faithful follower o f al-Farabi, appears to be vindicated. '6

Yet, strangely, af ter giving an account o f predicat ion in line with the doctr ine o f al-Farabi, Ibn Sina at times has discussions at odds with this account. These discrepancies need not point to inconsistencies in Ibn Sina's works. As is of ten remarked , Ibn Sina hirnsclf states that in his encyclopedia works (the Shif t ' and related summaries) he is merely giving the doctr ine o f the "Westerners ," that is, o f the Aristotclians. ̀ 7 In his Logic of the Orientals, Ibn Sina p roposed to bc rnorc original. Yet that work seems in general merely to repea t the doctr ines o f the encyclopedic works. What I am proposing is the hypothesis that in the encyclopedic works Ibn Sina at times gives his own, novel view, af ter giving the "standard," Aristotelian view, which for him would natural ly be the one presented by al-Farabi. '8 In any case, there are texts in Ibn Sina that present a view of predicat ion di f ferent f rom the one o f f e r ed by al-Farabi. I shall now tu rn to those texts.

A. In present ing tables o f opposites for simple categorical propositions, Ibn Sina follows al-Farabi in giving t ru th conditions for each proposi t ion in those tables. These tables o f opposites are constructed on the principle that the grammatical subject remains constant for a table, while the predicate may bc af f i rmed or denied, and may bc positive, metathctic, or privativc. An example o f a positive prcdicatc is ~just'; a metathetic predicate is 'not-just' or 'non-just ' ; a privativc predicate is 'unjust ' or 'tyrannical'. T h e distinction between the rnetathetic and the privative predicate, though present in the Prior Analytics, is not p rominen t in Aristotle, but is so in Theophras tus . '9 Being unjust is a more specific claim than being not-just, since a stone is not-just but is not unjust. So 'unjust ' entails 'not-just', but not vice versa.

T h e general s t ruc ture o f a table o f opposites is as follows:

,6 On this claim, cf. Majid Fakhry, A History of Islamic Philosophy (New York, :970), 147. '~ Ibid, 15~-53. ,s In his autobiography, Ibn Sina makes explicit his great respect of al-Farabi as an inter-

preter of Aristode. Cf. The Life oflbn Sina, ed. and trans. Gohlman (Albany, 1974), 3~,7-34,3. It should be noted that the Logic of the Orientals merely summarizes the discussions in the

Shif(~' on predication, and, indeed, on most else. The hypothesis that I am proposing offers a way to reconcile the unoriginal character of Ibn Sina's "Oriental" writings with his claim that he has doctrines different from those of the "Western" Aristotelians. Note that one need not accept this historical hypothesis to accept the logical analysis that I give below. Still, cf., A. M. Goichon, La Philosophie D'Avicenne (Paris, 1950, 17-18.

, 9 Cf. Zimmermann, Al-Farabi's Commentary, lxiii. But also see Ammonius, 161,18; Prior Analytics 1.46; Ackrill, 143. Also cf. al-Nafftt, 15,13ff.

356 J O U R N A L OF T H E H I S T O R Y OF P H I L O S O P H Y ~5:3 JULY 1987 --is P --is not P --is not not-P --is not-P --is not P* --is P*

P and P* are contraries. Each horizontal pair are contradictories, and any i tem in one co lumn is o p p o s e d (at least contrary) to any i tem in the other , except in the tables for indefinite and par t icular proposit ions. In the co lumn on the left, proposi t ions below follow f r o m proposi t ions above, but not vice versa; in the co lumn on the right, proposi t ions above follow f r o m propositions below, but not vice versa. ( 'Not ' may a p p e a r before the subject term.) Ibn Sina gives tables for the following subject terms: a singular name ( 'Zayd'), an indefinite n a m e ( 'man') , a par t icular ( 'sonle man') , and a universal ( 'every m a n ' ) ? ~

Yet Ibn Sina gives m o r e t ru th condit ions than al-Farabi does for propositions in those tables. I bn Sina states that his basic principle is: " the reality of the a f f i rmat ion is the j u d g e m e n t o f the presence (wujfad) o f the predicate to the subject. ''~' T h u s far, I bn Sina seems jus t to be following al-Farabi and On

Interpretation, as c o m m o n l y in te rpre ted . '~ But then Ibn Sina immediate ly concludes: "so every subject o f a proposi t ion is existent e i ther in individuals or in the intellect. ' '~ Th i s view too may be implicit in al-Farabi 's work. But Ibn Sina at least makes explicit the claim that one t ru th condit ion for 'S is P' is that S exists somehow; that is, Ibn Sina says that the subject mus t exist in re

or in intellectu. This view accords with his doctr ine o f the threefo ld distinction o f quiddity, accord ing to which quiddities may exist e i ther in individuals o r in the mind. ~4 Moreover , as we shall see, this view will help Ibn Sina explain why ' H o m e r is a poet ' is t rue, a l though ' H o m e r exists' is false. In general , I bn Sina holds that claiming that S is P entails claiming that S exists somehow.

Ibn Sina adds this existence condit ion systematically to the t ru th conditions o f the tables o f opposi tes given by al-Farabi. Following Aristotle and

,o In the Commentary on De Interpretatione and in ALBurh~n respectively. Also cf. Ackrill, 145. It should be noted here that I have given the tables as in Ibn Sina. Ibn Sina gives tables in a different order than al-Farabi. This difference does not bear on the present topic.

*' Ibn Sina, Al-Burh~n, 82,15ff; On Interpretation, 19b 5. Cf. AI-Farabi's Commentary, ed. Kutsch & Morrow, xo3,3-23; trans. Zimmermann, 98-99 .

"~ On Interpretation, a9b5; al-Farabi, Commentary, ed. Kutsch & Morrow, p. lol,3; trans. Zimmermann, 95.

*s Ibn Sina, Al-Burh~n, 8~,15ff. Cf. Al-Maq~l~t, ed. E1-Khodeiri et al. (Cairo, 1959), ~ , 9 - 1 0 .

,4 Cf. AI-Burh~n, 82,a5ff., and Al-Madkhal, ed. Madkour et al. (Cairo, 195~ ), x5,1-15. The doctrine of the threefold distinction of quiddity states that quiddities have three respects: in themselves, in individuals, and in the mind. They are "real" in themselves, but "exist" only in individuals or in the mind.


al-Farabi, he distinguishes the forms 'S is not P' (i.e., 'S is not existent as P') and the metathetic 'S is not-P' ('S is existent as not-P'). 25 He likewise takes into account cases in which the subject S has the potential for being P but is not yet in a state to realize that potential, and others, where S does not have the potential o f ever being P. The remaining cases of conditions of t ruth and falsity dist inguished here by Ibn Sina concern whether or not the predicate 'P' has contraries as well as a contradictory. Thus, for the truth of 'Zayd is not just ' , any of the following conditions suffices: (t) Zayd is non-existent; (2) Zayd is unjust, i.e., a tyrant; (3) Zayd is in a state intermediate between justice and injustice; (4) Zayd has the unrealized potential o f becoming just or unjust (as when Zayd is a child); (5) Zayd is not the sort o f thing that could ever be jus t o r unjus t J 6 For the truth of 'Zayd is not-just', it suffices that Zayd be both existent and not just; i.e., he must be existent, and must be not jus t for any o f the following reasons: (x) he is unjust; (2) he is in a state intermediate be tween justice and injustice; (3) he has the unrealized potential o f being jus t or unjust; (4) he is not the sort o f thing that could ever be jus t o r u n j u s t . 27 More schematically, these truth conditions are as follows: the singular proposit ion, 'S is not (existent as) P' means that it is not the case that S is both existent and is P; hence, either S does not exist or S is not P, for one reason or another. 'S is (existent as) not-P' requires both that S be existent and that S not be P, for one reason or another.

The scheme that Ibn Sina is following appears to assume that when the negation forms an indefinite predicate, o f form 'not-P', the copula is af- firmed, as well as the indefinite predicate, and that when the negative ap- plies to the copula the conjunctive claim, that the copula 'is' and the predicate 'P' are predicated o f the subject, is denied. Thus, 'S is not-P' may be read as 'S exists and is not-P', whereas 'S is not P' may be read as 'it is not the case that S exists and is P'.

The truth conditions that Ibn Sina gives for singular categorical affirmative proposit ions are singularly uninformative: 'Zayd is just ' is true only when Zayd is just, and is false otherwise? 8 It is clear f rom the rest o f this table and f rom what Ibn Sina says elsewhere that these truth conditions could also be put as: Zayd exists, and Zayd is just. In tables for indefinite and quantified categorical proposit ions Ibn Sina is a bit more explicit for the affirmative form. E.g., he says that 'man is just ' is true when all or only some

�9 5 Al-Burhan, 77,8ff. Also cf. Categories, l~b26ff.; On Interpretation, 19b~4-6; A1-Farabi, Commentary, ed. Kutsch & Morrow, lo9,14-~4; trans. Zimmermann, xo5-6.

,6 Al-Burh~n, 83, 7-8; 84,6-xo. �9 7 Ibid, p. 83,8-xo; 8 4 0 1 - 3. �9 s Ibid, 84,5- 7.

358 J O U R N A L OF T H E H I S T O R Y OF P H I L O S O P H Y ~5:3 JULY 1987

men are just, and is false when men do not exist? 9 Yet, in general, Ibn Sina states the existence condi t ion more explicitly for the negative forms. Indeed , this d i f fe rence is not surprising, given that in such cases there of ten tend to arise disjunctive lists o f conditions, each o f which by itself suffices for the t ru th or falsity o f the proposi t ions at issue.

Ibn Sina analyzes indefinite and quantif ied simple categorical propositions along the same lines. As illustrated above, the indefinite proposit ions are r educed to the quantif ied ( 'man' means 'some man' or 'every man'), and so are o f no especial interest. T h e quantif ied propositions, however, present the problem, how Ibn Sina proposes to handle quantifiers, on his view of the copula. He appears to pu t the quantif ier with the predicational claim, but not with the existence claim. Thus , he says that 'every man is not just ' is false when men do not exist, or when some o f them are just, or when all o f them are just. 3~ So 'every S is P' claims that (1) S exists and (~) every S that exists is P. Elsewhere, Ibn Sina analyzes 'every S is P' as "everything that is described as being S is described as being P, as long as its essence is existent. ''al This o ther explication gives a clue for how the first condition, that S exists, is to be unders tood . S is the essence, and comes into and passes out o f existence. So the condit ion, that S exists, means that the quiddity or essence S is instantiated. T h e second condit ion requires that all such instantiations be P.

B. Again, in comment ing on On Interpretation 11, Ibn Sina follows al- Farabi in explaining why ' H o m e r is' does not follow f rom ' H o m e r is a poet'. In ' H o m e r is a poet ' , the copula is predicated only incidentally, to signify the predicat ion o f 'poet ' o f H o m e r ; 'is' is not predicated in its own right. 3.

However , then, Ibn Sina proceeds to discuss more critically the opinions o f those preced ing him. His main concern is to explain how it is t rue that H o m e r is a poet. For Ibn Sina this is o f course a problem, given his t ru th conditions for singular categorical propositions. According to what he says, ' H o m e r is not a poet ' ough t to be true, and ' H o m e r is a poet ' and ' H o m e r is a not-poet ' false, since H o m e r does not exist. T h e r e is also the basis for this problem that I men t ioned above, namely that Aristotle says in the Categories

,9 Ibid, 85,14- 5. Also cf. Al-Maqfd~t, 25,4- 5. 30 AI-Burh~n, 89,1-2. s, Al-Q~y~, 2x,6 and 13ff. As stated above, the phrase, 'as long as its essence is existent',

suggests that there need be no instantiation of an essence for a universal proposition to be true. But cf. ibid, ~ 1,6-t 2 and 24,1, where Ibn Sina appears to require instantiations, in the imagina- tion at least, and my discussion of phantasm below. From the sophisms that Ibn Sina resolves, that phrase appears to have the function of limiting the proposition to the times at which there are existent subjects, as in 'everyone on the road to Baghdad stops at Quarimiyas'. Cf. ibid, p. 23,~7ff, al-Naj~t, 24,x2-5; ~5,8.

3, Al-Burh~n, lO9,2-11. Cf. On Interpretation, 21a~5-8, and Al-Farabi's Commentary, ed. Kutsch & Morrow, ~6o,23-7; trans. Zimmermann, 155. Also cf. Ammonius, 186,15.


that when the subject does not exist, nei ther an aff irmation nor a denial or a predicative o f it is true. But then proposit ions like 'Ho mer is a poet ' ought to be false.

Ibn Sina first admits that, on the normal reading, ' H o m e r is a poet ' is false:

Besides all this we have learned from them that nothing is predicated of the non-existent. And we know that when we say that Homer is a poet it is not true as meaning that Homer is a thing described as being a poet, but that the phantasm which is from Homer is described as being the phantasm fancied of Homer, and it is true to connect to it the concept, 'is a poet', i.e., he is an existent phantasm which has the property that when the phantasm of past time is connected to him, and the concept of poet is connected with him, (that concept) is true of him. 33

' H o m e r is a poet ' is false, since H o m e r is not a thing, i.e., since the individual, Homer , does not exist in re.

However , Ibn Sina says, there is a way to unders tand ' H o m e r is a poet ' to be true. On this reading, the subject term, 'Homer ' , refers to "an existent phantasm." T o that phan tasm there are connected, so the statement claims, two items: the phan tasm o f past t ime and the concept o f poet. Those con- nections do hold, and so the s ta tement is true.

Ibn Sina does not elaborate on this explanat ion much, even when he is concerned with the d i f fe ren t mental faculties in his commenta ry on On The

Soul. In parallel, my remarks will be sketchy and somewhat tentative. For Ibn Sina, ' H o m e r is a poet ' may be unders tood to claim that there once was a person named H o m e r who was a poet, though that person does not presently exist. So, at one time, H o m e r was a thing that existed in re, and, at that time, people were acquainted with H o m e r by sensation by direct acquain- tance, if you will. Now, however, ou r knowledge o f H o m e r is not via sensation. Rather, we are able to create a phantasm, i.e., fo rm a picture of Homer , on the basis o f what we have r e m e m b e r e d o f Homer . (We may r e m e m b e r both ou r own sense exper ience o f H o m e r and what we have been told o f Homer . ) In the same way, we may have a phantasm of past time, constructed f rom what we de te rmine to be general features o f o u r past experience, as contrasted with ou r present experience. A phantasm, then, is a p roduc t o f the fancy, const ructed out o f sense data, be it presently being acquired or be it s tored in the memory . As we may be aware o f and think about a phantasm now, a phantasm has existence in inteUectu. Note that the phantasm itself exists now, in the mind, and so there is a way to talk o f H o m e r at present and still satisfy Ibn Sina's existence condition. Fur ther , the phantasm H o m e r has certain propert ies , such as being a poet; more precisely, it has the pro-

s3 AI-Burhan, lo9,1~-xlo,1.

360 J O U R N A L OF THE HISTORY OF PHILOSOPHY 2 5 : 3 JULY 1987 per ty o f having been a poet. Ibn Sina notes this more precise sense when he says that the phantasm o f past t ime is connected to Homer . 'Poet' does not deno te a phan tasm but ra the r a concept. We know what a poet is f rom present exper ience and f rom ou r intuit ion o f quiddities in themselves (if, indeed, there be a quiddity o f poet). So concepts, which are the quiddities in the mind, always have existence in mtellectu. Along these lines, then, Ibn Sina is able to give a read ing o f how ' H o m e r is a poet ' can be true, on his theory o f predicat ionP 4

Ibn Sina is thus able to admit t rue statements about things that are not real at p resent because he recognizes two kinds o f existence, in re and /n intellectu. Normally, ou r discourse is about real things, that have present existence both in re and in intellectu. But, at times, ou r discourse concerns objects that have present existence only in intellectu. Still, any assertion about objects, such as phantasms, will be false, if they are taken to be about things existent in re. 3~ So, for Ibn Sina, every categorical proposi t ion makes an existence claim for its subject, unless the copula is negated. T h e claim is that the subject t e rm is instantiated, or, more precisely, that the quiddity of the subject has existence at present . Normally in ou r discourse that existence is p resupposed to be real existence. But sometimes that existence will concern a quiddity existing in the mind, and be based on a phantasm. But that phantasm too must be thought o f at present , and must be based on the real existence o f things in the past, or, perhaps , in the future .

.

Let me indicate briefly how my in terpre ta t ion o f this Aristotelian view of predicat ion differs f rom those that have been proposed recently. Rescher claims that Ibn Sina holds t h a t ' S is P' requires that there really be an S, whereas Ibn Rushd (Averroes) holds t h a t ' S is P' requires only that there be an S, in re or in intellectu, s6 1 have claimed that Ibn Sina in fact holds the view that Rescher attributes to Ibn Rushd alone. T h e texts cited by Rescher do, to be sure, show that Ibn Sina claims t ha t ' S is P' requires that S exist in re. But, as we have seen, there are o ther passages where Ibn Sina endorses mental existence as sufficient for the existence requirement . Rescher also cites, in suppor t o f his view, an a rgumen t between Ibn Sina and Ibn Rushd, on

Cf. lbn Sina's discussion of the griffon, ibid, 11o,2ff and 8~,16-8. On the relation between phantasms and concepts, cf. F~ Al-Nafs, ed. Anawati (Cairo, 1974), 32,7ff.; 147,1ff. In general, phantasms are based on particular experience, whereas concepts are of universals.

s5 Cf. ibid, 1 lo,7-x4. On predication of non-existent objects, cf. Ammonius, 186,15. 36 Nicholas Rescher, "Existence in Arabic Logic and Philosophy," in Studies in Arabic Logic

(Pittsburgh, 1966), 73-75.


whether existence is an a t t r ibuteY But this is a separate question, tied more closely to the issue, what is the ontological significance of this logical analysis of predication. I conclude that Ibn Sina, like Ibn Rushd, recognizes two sorts of existence, each of which, in general, without contextual clues, suffices to satisfy the existence requirement .

Wedin claims that Aristotle holds t h a t ' S is P' has the s t ruc ture 'S exists and that S is P'. (I forego here any remarks on the merits of his view as an interpretat ion of Aristotle.) Thus, far, his position and the one that I have been discussing are identical. These positions, though, differ in two important respects. Wedin claims that 'Homer is a poet' is not of the same logical type as 'Socrates is sick/healthy', even though Homer and Socrates are both supposed to have existed really, and to have 'poet' and 'healthy/sick' as accidents. 39 Thus, 'Homer is a poet' does not require that Homer exist, whereas 'Socrates is sick/healthy' requires that Socrates exist. Ibn Sina, indeed, like the bulk of the Aristotelian tradition, does not consider these propositions to be o f d i f ferent logical types, but instead claims that each of them requires the existence of the subject. The second major difference is that Wedin, like Rescher for Ibn Sina, recognizes only one sort of existence, real existence. 4~ Wedin represents this sort by the existential quantifier. In contrast, Ibn Sina recognizes (at least) two sorts of existence.

.

This view of the structure o f simple categorical propositions offers ready answers to vexing problems of interpret ing Aristotle's texts.

A. For one, it explains how Aristotle can say in On Interpretation that 'Homer is a poet' is true, while 'Homer is' is false, and in the Categories that 'Socrates is sick' and 'Socrates is healthy' are both false if Socrates does not exist. As we have seen, 'Homer is' is true, if the subject is taken to exist in intellectu. But that proposition, like those about Socrates, is false if the subject term is taken to refer to something existing in re. Similarly, 'Socrates is healthy' or 'Socrates is sick' would be true if 'Socrates' were taken to refer to a phantasm existing in intellectu.

B. Again, this view o f predication explains what appears, f rom a modern perspective, to be a great inadequacy of Aristotle's syllogistic: the lack of an explicit existential import assumption. 4' Tha t assumption appears to be cru-

s7 Ibid, 75-76- ss Michael Wedin, "Aristode on the Existential Import of Singular Sentences", Phronesis

23:~ (1978): 18o. s9 Ibid, 186. 40 Ibid, 188. 4~ Cf. I. M. Bochefiski, A History of Formal Logic, trans Thomas (Notre Dame, 197o ), 7 o.

362 J O U R N A L OF THE H I S T O R Y OF P H I L O S O P H Y 2 5 : 3 JULY 198 7

cial for just ifying A to I conversion. (In Aristode, that is the conversion f rom 'every B is A' to 'some A is B'. Given I to I conversion, from 'some A is B' to 'some B is A', and vice versa, a conversion valid in both modern and traditional logic, 'some B is A' would also follow from 'every B is A'.) Aristotle's justification o f A to I conversion is quite brief: "But, if every B is A, then some A is B. For if no A were B, then no B would be A. But we assumed that every B is A. ''4~ Aristotle thus gives an indirect p roof for A to I conversion. From 'every B is A', it follows that some A is B or not [some A is B]. 'Not [some A is B], Aristotle takes to mean 'no A is B'. But, if no A is B, then no B is A, which conflicts with the assumption, 'every B is A'. Therefore , 'some A is B' follows f rom 'every B is A'.

The difficulty with this p roof f rom the modern perspective is the implicit assumption that 'not [some A is B]' means 'no A is B'. But suppose that there is nothing that is A. T h e n it would not be the case that some A is B, even though it could still be false that no A is B. For example, suppose that there are no goat-stags. T h e n 'some goat-stag has hooves' would be false, but still would not imply that no goat-stag has hooves. Indeed, 'every goat-stag has hooves' would still be true: if anything existed that is a goat-stag, it would have hooves. So the difficulty in Aristotle's p roof for A to I conversion is that the negation of 'some A is B' has "two causes of truth," first, that no A that exists is B, and, second, that no A exists. Aristotle seems to ignore the second option. Likewise, the Greek commentators do not mention it, but merely repeat Aristotle's proof. They instead cite objections concerning complex predicates, but these objections are resolved easily, and do not bear on the present problem. 43 So it appears that Aristotelian logic, if it is to be rendered consistent, has an implicit existential import assumption: the subject term, at least in affirmative categorical propositions, must refer. T h e n 'every B is A' would imply that some B exists, and, thereby, the second cause of t ru th ment ioned above for 'not [some A is B]' would be ruled out. The difficulty is that Aristotelians never acknowledge this existential import assumption in their proofs.

T h e view of predication given by Ibn Sina offers a way of resolving this difficulty. The t ru th conditions for 'every B is A' are: 'B exists and every B that exists is A'. The condit ion that B exists rules out the possibility that it is not the case that some B is A because no B exists. Hence the contradictory of 'some B is A' will be 'no B is A', and the proof given by Aristotle of A to I

4, Prior Analytics, 25a18- 9 (trans. Jenkinson). 4s Alexander of Aphrodisias, 34,23ff., and especially John Philoponus, InAristotelisAnalytica

Priora Commentaria, ed. Wallies (Berlin, t9o5), 48, t2ff.


conversion will be quite intelligible. Moreover, it will be understandable why A to I conversion was generally taken to be obvious by the Greek commentators and Islamic philosophers. Ibn Sina's truth conditions for 'every S is P' and 'some S is P' differ only in that in the A case all instantiations of S must be P, while in the I case only some instantiations of S must be P. The validity of A to I conversion thus will depend merely on a shift from 'every' to 'some', or, in traditional terms, from whole to part. (In modern terms, given that A to I conversion is in the object language, while statements of the truth conditions and their consequences are in the metalanguage, this justification of this conversion will not beg the question.) So this view of predication not only vindicates A to I conversion, and similar inferences, but also explains why Aristotelian logicians would have tended to view such inferences as obvious: given the reading of 'every A is A' as 'every B is existent as an A', the existential import condition is stated baldly.

C. This view of predication also bears on the distinction of essence and existence. At first it might be thought that here is an objection to the claim that this view of predication is Aristotelian, or even one held by Ibn Sina. Aristotle distinguished questions of existence from questions of essence, and Ibn Sina makes that distinction one of the cardinal tenents of his metaphysics. Now a definition gives the essence of something, as in 'man is rational animal'. On the analysis that I have presented, that definition means 'man is existent as rational animal'. But then any definitional statement would appear to imply the existence of the definiendum. But surely there can be definitions of unicorns and goat-stags without there existing unicorns and goat-stags.

This difficulty is only apparent. Note that the view of predication that I have attributed to Ibn Sina allows for the definition of things that do not really exist. Ibn Sina would hold that anything that is to be defined must exist at least in inteUectu. Unicorns and goat-stags, existing in the mind, have definite natures, and the definition of goat-stag states that such a concept is existent as having that nature. Of course, if it is concluded from 'a goat-stag is existent as a thing that is half goat, half stag' that a goat-stag is existent, it must be specified what sort of existence, mental or real, is being asserted. Ibn Sina seems to hold that definitions generally express existence in intellectu. Thus he holds that definition is a concern of logic, and logic concerns quiddities in the mind, which have mental existence. 44 Again, he claims that 'heptagonal house' is a universal term, even if there are really no items of

Cf. Al-Madhhal, 15,16-7; 19,8 et passim. Still, the basis for the truth of real definitions lies in quiddities in themselves.

364 JOURNAL OF THE HISTORY OF PHILOSOPHY 25:3 JULY 198 7

which 'heptagonal house ' can be truly predicated. Still, that te rm is predicated o f many, with respect to individuals in the mind. 45

Even Aristotle makes remarks that suggest that what is def ined must exist. In the very passage where he distinguishes the question o f existence, that it is, f rom the quest ion o f essence, what it is, Aristotle claims that there is no definit ion o f what does not exist: "He who knows what h u m a n - - o r any o t h e r - - n a t u r e is, must know also that man exists; for no one knows the na ture o f what does not exist---one can know the meaning o f the phrase or name 'goat-stag', but not what the essential na ture o f a goat-stag is. ''46 Aristo- tle is evidently dist inguishing real f rom nominal definition. 47 T h e normal definit ion is the real one for Aristotle, and one o f the necessary conditions for a real definit ion is that the subject o f that definition exist. Nominal definitions, where it is st ipulated that one phrase "is identical to" ano ther in meaning, can be given without the subject's existence. In Ibn Sina's scheme, a nominal defini t ion would concern what exists in the mind but not in reality.

Thus , on this view of predication, a definit ion o f form, 'S is P', will have the structure, 'S is existent as a P'. But, without fu r the r knowledge, that existence can be taken only as ex i s tence /n intellectu. Real definitions require the existence o f the subject in re as well as in intellectu. So we can see how this view of predicat ion is consistent with the distinction o f existence and essence in an Aristotelian f ramework.

What light, then, does this view of predicat ion shed on the relation o f essence and existence, so p re -eminen t a fea ture in Ibn Sina's metaphysics? First, essences must be dist inguished into the real and the nominal. A real essence presupposes real existence; a nominal essence presupposes mental existence. Now Ibn Sina separates things that exist into what necessarily exists, namely God, and what possibly exists. Possible existents are such that their essence and existence are separable, in the sense that their essence does not requi re their existence. Hence an external cause, ultimately God, is required, Ibn Sina says, to make what possibly exists actually exist. 48 Accord- ing to what I have claimed for Ibn Sina's view of predication, Ibn Sina holds that things that possibly but not actually exist have mental existence (in

45 Ibn Sina, Al-I~h~yyat, ed. Anwati et al. (Cairo, 196o), vol. 2: 195,6-1o. Note that R. E. Allen, "Plato's Early Theory of Forms", in The Philosophy of Socrates, ed. Vlastos (Garden City, x971), 327, holds that Plato early on held a stronger view wherein a (real) definition of S requires that there are (really) S's.

46 Posterior Analytics, 92b4-8 (trans. Mure). 47 Ibid, 93b29-3 x. 4s Al-II~h~yy~t, Vol. x: 37,1ff.

AVICENNA ON EXISTENCE 365 h u m a n minds or perhaps in the divine mind) and a nominal essence, but nei ther a real existence nor a real essence.

This suggested view of the distinction of essence and existence can also be put in terms of Ibn Sina's threefold distinction of quiddity (triplex status naturae). It might not be too fanciful to identify nominal essences in abstrac- tion f rom mental existence with quiddities in themselves, which Ibn Sina says are real but do not exist either in the mind or in reality. Nominal essences, as they have mental existence, would then be the quiddities in the mind. Real essences, having real existence, would be the quiddities in things. Real essences are not separate f rom real existents; Ibn Sina says that he is not a Platonist: a thing and its fo rm are not separate.

D. This view o f predication also gives an interesting logical perspective on why the study of being qua being is so important for Aristotelians. Ac- cording to it, 'S is P' claims that S is. Now to say that S is (estin) is to say that S has being (einai; to on). But in his early writings, like the Categories, Aristotle divides up beings (ta onta) in several ways: into those said of and those present in a subject, into the individual and the universal, and, especially, into the ten categories. 49 Beings, the things that are, are in di f ferent categories. So to claim that S is would be to claim that S has being in some category.

However, the relation o f this category theory to the structure of predication raises a pressing problem immediately. Aristotle recognizes that there may be valid syllogisms where the terms are f rom different categories. 5~ Let us consider such an example:

Every white (thing) is an animal Every man is white Therefore, every man is an animal

I f we replace each proposit ion with the t ruth conditions given by Ibn Sina for it, we obtain:

White is (existent), and every white (thing) that is is an animal Man is (existent), and every man that is is white Therefore, man is, and every man that is, is an animal

The problem is that, if the sort o f being differs from category to category, the connection between the minor and major terms is rendered suspect, because o f the differences o f claims made by 'man is white' and 'white is

49 Categories, ~a~off.; lb25; On Interpretation, 17a38ff. so At least we may infer that Aristotle considers that syllogisms combining such terms are

not ill-formed and might be valid, from the examples that he uses in his syllogistic. Cf. Prior Analytics, 26b 7. It should be noted, though, that Aristotle uses such combinations mainly to construct counterexamples.

3 6 6 JOURNAL OF THE HISTORY OF PHILOSOPHY ~5:3 JULY 1987

animal'. In other words, if a claim that S is P is to be relegated to a certain category, there is no common ground on which to establish a predication via the linkage of predications in different categories. Yet the validity of syllogisms does not appear to depend on whether the copula, 'is', is in the same or different categories in the premises.

So we can see here a motivation by which Aristotle would have intro- duced first philosophy by saying that "being is said in many ways." That is, the copula, 'is', is said in many ways, as predications are pegged to different categories. (Indeed, it would not be too absurd to translate Aristotle's dictum as: "the (copula) 'is' is said in many ways.") Yet, for the connection of predications, and, moreover, for the unity of science, 'is' must not be equivocal when used of things in different categories. Thus Aristotle says that "there are many senses in which a thing may be said to be, but all that is is related to one central point, one definite kind of thing, and is not said to be by a mere ambiguity. TM Aristotle seeks to discover the focal meaning of 'is', a common ground for the ways in which S is. Likewise, Ibn Sina begins his metaphysics with an investigation of being, where he seeks the common ground of necessary and possible being.

From the perspective of this view of predication, the task of ontology is to find the common ground according to which things are (existent). The task is to find being qua being, not being in one of its special categories. ~2 For Aristotle the pursuit leads to an examination of substance and ultimately to substantial form: that man is, that Socrates is, requires the presence of a certain form in certain matter; that white is, that snub is, requires both the presence of that form in that matter, and the presence of the appropriate substantial form in that matter. In this way, the being of substances does not require the being of accidents, although the claim that an accident is requires that there be a substance. For Ibn Sina, the pursuit of the common ground of being is naturally more explicitly tied to this theory of predication. What he requires is an explication of how a quiddity, or essence, be- comes instantiated. The instantiation is different for different types: the substantial, the materially accidental, the formally accidental. Like Aristotle, Ibn Sina requires an instantiation of the substantial for there to be any instantiation of the other sorts.

Thus, in each case, the metaphysical investigation may be viewed as the attempt to find the focal meaning of the copula and to find out the relations among the different categories of determinations of the copula. 'Is' is said in many ways. Yet some ways require a previous determination of 'is', with

s, Metaphysics, loo3a19-2~. 5, Ibid, I o o 3 a i 9 - 2 2 .


respect to the same subject. The metaphysical task may be characterized as the problem of showing what determinations must be supposed to be appli- cable to the subject, in order that any determination whatever be able to be made of the subject. I f there were no science of being, it would be impossible to have a unified experience: substances, quantities, etc., would just exist in different ways and have no common ground.

.

So this view of predication, explicitly stated by Ibn Sina, offers a useful approach to Aristotelian metaphysics and logic. I am not claiming that this approach is sufficient by itself to explain Aristotelian metaphysics, or even that it was explicitly held before Ibn Sina. But I hope to have shown, despite its bizarre appearances, that it does have its attractions in resolving and illuminating problems in Aristotelian philosophy. 53

Kutztown University

5s An earlier version of this paper was read in September 1984 at the joint meeting of the Society for Ancient Greek Philosophy and the Society for the Study of Islamic Philosophy and Science. I am grateful for the comments of Professor Parviz Morewedge and the referees and editor of the Journal of the History of Philosophy.

avicenna on existence

Documents