Something Nothing Estoicismo

OXFORQ STUDIES
IN ANCIENT
PHILOSOPHY
SOMETHING AND NOTHING:

THE STOICS ON CONCEPTS
AND UNIVERSALS
VICTOR CASTON
EDITOR: DAVID SEDLEY
VOLUME XVII
. it 51 TL); it 8'
au
iL),' OK,dS' ovap

Pindar, Pythian 8. 95
1999
I T is commonplace to find the Stoics described as nominalists. I As
early as the fifth century CE they are described as holding that 'there
are only particulars' and, a century later, that 'common entities are
~
Victor Caston 1999

This paper began its life long ago in Julia Annas's seminar on Hellenistic Philosophies of Mind in the spring of 1989, and was subsequently presented at the winter
meetings of the American Philosophical Association, where Wolfgang Mann commented on it; I have profited greatly from their criticisms. I would also like to thank
very warmly Jamie Dreier, Stephen Menn, and David Sedley for extensive discussions and written comments in the interim, as well as Tad Brennan, Jaegwon Kim,
Baron Reed, Ernie S088, and Jim Van Cleve for challenging criticisms. Each saved
me from numerous errors; the remainder, of course, are solely my own.
0 Y "FORD
UNIVERSITY PRESS
I e.g. E. Zeller, Die Philosophie der Griechen in ihrer geschichtlichen Entwicklung

[Philosophie der Greichen] , 5th edn. (3 vols. in 6 parts; Leipzig, 1923; 1st edn.
1844-52), iii/I. 80-1 (cf. ii/I. 295-6); C. Prantl, Geschichte der Logik im Abendlande
[Geschichte] (4 vols.; Leipzig, 1855-70), i. 420, 427-33; L. Stein, Die Erkenntnistheorie der Stoa [Erkenntnistheorie] (Berlin, 1888),276-300; V. Brochard, 'La logique
des Stoi'ciens', in his Etudes de philosophie ancienne et de philosophie moderne [Etudes]
(Paris, 1926),220-38, esp. 221-4, 235-8, and his 'La logique des stoi'ciens (Deuxieme
etude)', in his Etudes, 239-51, esp. 242-6; P. Couissin, 'La critique du realisme
des concepts chez Sextus Empiricus' ['Critique du realisme'], Re'Vue d'histoire de
la philosophie, I (1927), 377-405 at 390; J. B. Gould, The Philosophy of Chrysippus
(Leiden, 1970),43,60,106, 119,159,201,203,208-9; A. Graeser, Zenon 'Von Kition:
Positionen und Probleme [Zeno] (Berlin, 1975), 84; G. Verbeke, 'Der Nominalismus
der stoischen Logik', Allgemeine Zeitschrijt fur Philosophie, 2 (1977), 36-55; M.
Schofield, 'Preconception, Argument, and God', in M. Schofield, M. Burnyeat, and
J. Barnes (eds.), Doubt and Dogmatism: Studies in Hellenistic Epistemology (Oxford,
1980), 283-308 at 286; M. Forschner, Die stoische Ethik: Uber den Zusammenhang
'Von Natur-, Sprach- und Moralphilosophie im altstoischen System (Stuttgart, 1981),
74,81. For more qualified endorsements see A.-J. Voelke, L'ldee de 'Volonte dans Ie
stoi'cisme (Paris, 1973), 28; B. Inwood, Ethics and Human Action in Early Stoicism
(Oxford; 1985),28,37-8,83,204.
Victor Caston
Something and Nothing
nothing'.2 More recently, however, it has been argued that the Stoics
are best understood as 'conceptualists', after the manner of British
empiricists such as John Locke, rather than nominalists in the strict
sense. J There are, unfortunately, texts to support both positions.
Worse, there are even texts that suggest a kind of realism: 4 the Stoics
are committed to the existence of qualities such as wisdom and
moderation that make an individual, for example, wise or moderate.
One might well wonder whether there is a coherent position here
at all and, if there is, just what sort of position it might be.
The process of solving this puzzle has been thwarted by a second
and more troubling one. According to virtually every modern treatment of the subject, (I) the Stoics identify Platonic Forms with concepts, while (2) denying that a concept is something, even though
(3) they regard the genus Something as the highest genus, comprehending both what exists and what does not. 5 It is hard to un-
derstand why the Stoics would have introduced concepts, only to

banish them in this way; or what it could mean for items to fail to
be something and fail to be either existent or non-existent. Such a
position appears difficult to motivate and even unintelligible. Something seems to have gone badly wrong.
A fresh look at the evidence suggests a more charitable reconstruction. First, our evidence presents not one Stoic theory, but
several (Section I). Concepts are mentioned only by the founder
of the Stoa, Zeno of Citium, and his successor Cleanthes (Sections
2-'7); beginning with Cleanthes' successor, Chrysippus, they cease
to be mentioned. Later Stoics appeal to linguistic conventions instead, retaining only certain broad features of the earlier theory,
while correcting for some of its more egregious consequences (Sections 8-<). If this is right, the appeal to concepts is not part of the
'canonical' form of Stoicism, which coalesced under Chrysippus
and his followers. Both theories, on the other hand, acknowledge
qualities, but in neither case do they imply realism: Stoic qualities are similar to what are sometimes called 'quality instances' or
'tropes'. In a scheme that already appeals to concepts or language,
such items might seem otiose. But in fact they constitute one of
the more philosophically insightful manreuvres the Stoics make in
responding to Plato.
Second, the Stoics never identify Platonic Forms with concepts
or, for that matter, anything else. Their strategy is eliminativist,
not reductivist. On their view, there are no such things as Platonic Forms-they are literally nothing at all. Concepts, however,
are something. They are precisely what replace Forms; they actually perform functions that Forms were supposed to. Something
is thus not only the highest genus for the Stoics, but fully com-
I Syr. In Metaph. 104. 21 Kroll and Simp!. In Categ. 105. I I Kalbfleisch, respectively. On the translation of oVTlVa see sect. 4. I.
, D. N. Sedley, 'The Stoic Theory of Universals' ['Stoic Universals'], Southern
Journal of Philosophy, 23 (1985), supp!. 87-92; A. A. Long and D. N. Sedley, The
Hellenistic Philosophers (2 vols.; Cambridge, 1987), i. 181-3. An earlier denial of
nominalism can also be found in V. Goldschmidt, Le Systeme stoicien et l'idee de
temps [Le Systeme stoiCien], 4th edn. (Paris, 1989; 1st edn., 1953), 166-'7.
C. H. Kahn, 'Stoic Logic and Stoic LOGOS' Archiv fur Geschichte der Philosophie, 51 (1969), 158-72 at 165-6.
The best statement of this position is to be found in David Sedley's succinct
but forceful 'Stoic Universals', and at full length in Jacques Brunschwig's magisterial 'La theorie stoi'cienne du genre supreme et I'ontologie platonicienne' ['Genre
supreme'], in ]. Barnes and M. Mignucci (eds.), Matter and Metaphysics (Naples,
1988), 21-127. But their position is in fact the culmination of a long tradition:
H. Ritter, Geschichte der Philosophie [Geschichte] (4 vols.; Hamburg, 1829-34), iii.
548-9; Pranti, Geschichte, i. 420, 427; Stein, Erkenntnistheorie, 292-8; A. C. Pearson, The Fragments of Zeno and Cleanthes (London, 1891), 72-5; L. Robin, La
Pensee grecque et les origines de l'esprit scientifique [Pensee grecque] (Paris, 1923),415;
Couissin, 'Critique du realisme', esp. 401-5; O. Rieth, Grundbegriffe der stoischen
Ethik: Eine traditionsgeschichtliche Untersuchung [Grundbegriffe] {Berlin, 1933),90I, 173-5, 180; A. Orbe, En los albores de la exegesis iohannea {Rome, 1955),280-6;
P. Hadot, Porphyre et Victorinus (2 vols.; Paris, 1968), i. 158-62, 174-6; K. Wurm,
Substanz und Qualitat: Ein Beitrag zur Interpretation der plotinischen Traktate VI
I, 2, und 3 {Berlin, 1973),169, 175-81; A. A. Long, Hellenistic Philosophy: Stoics,
Epicureans, Sceptics, 2nd edn. (Berkeley, 1986; 1st edn. 1974), 141, 147; Graeser,
Zeno, 69-'78; M. Frede, 'The Origins of Traditional Grammar' ['Origins of Grammar'], in R. E. Butts and]. Hintikka (eds.), Historical and Philosophical Dimensions
of Logic, Methodology, and Philosophy of Science (Dordrecht, 1987), 51-'79, repro in
his Essays in Ancient Philosophy (Minneapolis, 1987), 338-59 at 349; D. E. Hahm,
The Origins of Stoic Cosmology (Columbus, 1977), 8; P. Pasquino, 'Le statut ontologique des incorporels dans I'ancien sto'icisme' ['Statut ontologique'], in J. Brunschwig (ed.), Les Stoiciens et leur logique (Paris, 1978), 375-86 at 378; Long and
147
. Sedley, The Hellenistic Philosophers, i. 163-6, 181-3; P. Aubenque, 'Une occasion

manquee: La genese avortee de la distinction entre I' "etant" et Ie "quelque chose"',
'in P. Aubenque (ed.), Etudes sur Ie Sophiste de Platon (Naples, 1991), 365-85 at
379
The only exceptions to this tradition are partial and confused, because they continue to identify concepts with Platonic Forms: E. Elorduy, Die Sozialphilosophie
der Stoa [Sozialphilosophie] (Leipzig, 1936), 85, 88 (though cf. 124-5); 'La logica de
la Estoa', Revista deftlosofia (Madrid), 3 (1944), 7-65 and 221-65 at 229-31; and EI
estoicismo (2 vola.; Madrid, 1972), i. 248 If.; M. Pohlenz, Die Stoa: Geschichte einer
geistigen Bewegung, 4th edn. (2 vols.; Gottingen, 1970; 1st edn. 1943-'7), i. 64-5, 295;
J. M. Rist, Stoic Philosophy (Cambridge, 1969), 156, 165, and 'Zeno and the Origins
of Stoic Logic' ['Zeno'], in J. Brunschwig (ed.), Les Stoi'ciens et leur logique (Paris,
1978), 387-400 at 394-5; C. Atherton, The Stoics on Ambiguity [Stoic Ambiguity]
(Cambridge, 1993),76-'7, 144,256.
Victor Caston
prehensive as well: it includes every item, whatever its status,

that is acknowledged by what they take to be the correct ontology."
I shall finish by drawing some general conclusions about 'the
problem of universals' and in particular the difference between the
ancient and modern debates (Section 10). This appears to be yet
one more area where we have much to gain from our predecessors.
main a mystery: nothing else is known about Longinus' and Antoninus' position, not to mention Cornutus' position, which Syrianus discusses later in the passage (106. 7-x3). But one point is
clear. Beginning with Chrysippus, 'most of the Stoics' did not appeal to concepts when responding to Plato, but conventions involving names. In fact, what distinguishes Antoninus is precisely
his reintroduction of concepts into the discussion and the syncretism in which it results. This strongly suggests that Cleanthes
was the last mainstream Stoic to appeal to concepts in responding to
Plato.
The position does not seem to have originated with Cleanthes,
though. We know that Zeno severely criticized Plato in an effort to
strike at Arcesilaus--a strategy as misdirected, N umenius tells us,
as when Cephisodorus criticized Plato's Forms in order to attack
Aristotle (ap. Euseb. Praep. Evang. 14. 6. 9-1 I, ii. 275. 6-276. 7
Mras). Numenius further claims that in making this critique, Zeno
'wickedly and shamelessly introduced' alterations to Platonic doctrine (ii. 276. 5 Mras). But he fails to tell us exactly what these
innovations were. Other testimonia suggest it might have been the
same theory as Cleanthes': 'The Stoics who follow Zeno said the
Ideas are our concepts." The same position is elaborated in parallel passages of Diogenes Laertius and Stobaeus that constitute our
most important evidence for the early Stoic theory. Because it has
attracted more discussion, we shall begin with the later report from
Stobaeus:
I.
Stoic responses to Plato's theory of Forms
Modern treatments generally speak as if all the early Stoics agreed

that a solution to the problem of universals involves an appeal to
concepts. One late source does collapse the position in this way,
claiming simply that 'the Stoics think the Ideas are our concepts'
(Ps.-Galen, De hist. philos. 25 =Dox. Gr. 615. x6 Diels). But our
other sources do not. The most nuanced account is given by Syrianus, who distinguishes a variety of approaches:
The Forms, then, were not introduced by these divine men Esc. Socrates,
Plato, the Parmenideans, and the Pythagoreans) for the use of conventions
concerning names [-n}v Xpr,GLV n/, TWV ovo!-'aTwv GIJV,)8ta,), as Chrysippus,

Archedemus, and most of the Stoics later thought, since in themselves the
Forms differ in many ways from what is said by convention. Nor do
they supervene on the understanding, analogous to the much-talked-about
expressibles [AKTa) , as Longinus recommended, given that what supervenes is insubstantial-how could the same thing both be an intelligible
and supervene? Nor, according to these philosophers, are Ideas concepts
[Wo~!-,aTa), as Cleanthes later said. Nor do they supervene on the understanding in the way conceptual ideas do, as Antoninus said, combining the
views of Longinus and Cleanthes. (In Metaph. 15.21-3)
The historical sketch from which this excerpt is taken is not intended to be chronological. Syrianus has instead organized the
positions systematically, with special attention to how they deviate from the orthodox Platonic position. 7 Several of them re For the sake of convenience, I shall forgo etymological scruples and use 'ontology' and 'entity' to designate, respectively, the domain of a metaphysical theory
and what falls within that domain, including what does not exist or have being of any
sort (should there be such according to the theory in question).
7 This should answer Brunschwig's concern about the reliability of Syrianus'
report ('Genre supreme', 78). Chronologically, it is a jumble: immediately after
149
Zeno's position. (I) They say that concepts are neither something nor
something qualified, but (2) as if they were something and as if they
were something qualified, being apparitions of the soul. (3) They were
called 'Ideas' by past philosophers. For Ideas are of things that fall under
concepts, such as men, horses, and (to speak more generally) all animals
and as many other things they say there are Ideas of. But the Stoics say
mentioning Antoninus, Syrianus goes back to consider the views of Socrates; he
then skips forward to middle Platonists (Plutarch, Atticus, and Democritus); and
then finally he goes back again to earlier moderates such as the Peripatetic Boethus
and the Stoic Cornutus. But it is extremely unlikely that Syrianus would be confused about Socrates' dates. In any event, IJ"""povat 105. 29 does not pose a problem:
it is to be taken with 1T(J.P' (J.v.,.oi, in the previous line to refer to 'these divine men'
at 105. 20-1, just as IJ".,..pov does at 105. 22-3, a point already recognized by Zeller
(Philosophie der Griechen, ii/I. 81 n. 4).
Ps.-Plut. Epit. I. 10.5 (=Dox. Gr. 309.9-10); Euseb. Praep. Evang. 15 45 4,
ii. 413. 8 Mras. Ps.-Galen's report mentioned above (p. 148) clearly derives from
these.
Victor Caston
that the latter are non-existent and that while we participate in concepts,
we bear those cases they call 'common nouns'. (Eel. 1. 136. 21-137. 6
Wachsmuth)
There is a great deal in this passage, and we shall return to it often.
For ease of reference the text can be subdivided according to the
following rubrics: (I) what concepts are not; (2) what concepts
are; and (3) the relation between concepts and Platonic Forms.
The first topic immediately leads us into a thicket of metaphysical
difficulties, though, which must be cleared away (Sections 2-4)
if we are to understand properly the second and third (Sections
5-7)
2.
Something
The first clause of the passage claims that concepts are 'neither
something nor something qualified'. How on earth, though, could
concepts fail to be something? A natural response would be the following: only if they were nothing at all, only if there were no such
things as concepts. But it is hard to see how that would help matters.
One might intelligibly claim that Platonic Forms were fictions or,
alternatively, that concepts were. But it is not clear what could be
gained by introducing one fiction and identifying it with another.
How can concepts pull the load Forms are supposed to, if they are
just as illusory?
Scholarly opinion has been more sanguine, though. In fact, it is
one of those rare moments when scholars have been virtually unanimous. The opening clause, they believe, must be taken as referring
to the Stoics' highest genus, Something (T6 TL), so that by denying concepts are something, Zeno places them outside this genus,
banishing them to a 'logical and metaphysical limbo' .9 Such claims
require us to reflect on the very foundations of Stoic ontology. In
this section, therefore, we shall consider what it is to be something
according to the Stoics, before evaluating in Section 3 what it might
be not to be.
The phrase is Sedley's ('Stoic Universals', 89; cf. Long and Sedley, The Hellenistic Philosophers, i. 181).

2. I.
Being vs. being something
In making Something the highest genus, the Stoics are rejecting the
view,that Being or What Is (TO ov) constitutes the highest genus. 'O
According to Plutarch (Comm. not. 1073 D-E), the Stoics claim
that
(NE) There are some things which are not.
Put in this way, the position sounds contradictory (as Plutarch
clearly intended it to). But it needn't, once it is taken in its proper
sense. For the Stoics reject the Platonic identification of being with
being something. liOn their view, to speak of something as 'a being'
(ov) is precisely to mark an item's ontological status, in contrast with
other uses of 'to be', which are unmarked. By disambiguating these
two uses of 'to be', the Stoics can thus avoid contradiction-indeed,
nothing short of this will work. To remove the sound of paradox,
10
Alex. Aphr. In Top. 301. 19-27, 359. 12-18 Wallies; Philo, Leg. alleg. 2. 86, 3.
175, Quod det. pot. 118; S.E. PH 2. 86, 223, M. 10. 234; Scholia in Arist. Categ.
34b 8-II Brandis; Plot. 6. I. 25. 3-6, 26, cf. 6. 2. I. 22 Henry-Schwyzer. According to an older tradition, beginning with Ritter (Geschichte, iii. 553-5) and Zeller
(Philosophie der Griechen, iii/I. 94-5 n. 2), the earliest Stoics accept What Is as
the highest category, while Something is elevated to that status only by Chrysippus or a later Stoic in order to accommodate incorporeals. But the evidence for
this hypothesis is slight and can be satisfactorily explained away. (I) Although
as a Stoic example of the highin most printed editions D.L. 7. 61 cites 'Ta
est genus, this reading does not occur in any of the manuscripts, but stems from
the Latin translation of Aldobrandini. For a judicious discussion see Brunschwig,
'Genre supr~me', SO-I. (2) The 'quod est' of Sen. Ep. 58. 7-14, 16-22 is not
Stoic, but middle Platonic: see the detailed analysis of j. Mansfeld, 'Substance,
Being and Division in Middle Platonist and Later Aristotelian Contexts' ['Substance, Being and Division'], in his Heresiography in Context: Hippolytus' Elenchos
as a Source for Greek Philosophy (Leiden, 1992), 78-109 at 84-5 n. 22, where he
argues that Seneca's use of the first person does not in context entail endorsement (against Brunschwig, 'Genre supr~me', 51-60). (3) As for 'Toli y,v'Kw'T,hou 'TOU
OV'TOS at S.E. M. 8. 32-6, Brunschwig (,Genre supreme', 46-'7) plausibly follows
W. Heintz, who obelizes 'Toli c'$V'TOS, despite his own acceptance of Zeller's thesis
(Studien zu Sextus Empiricus [Sextus Empiricus] (Halle, 1932), 151-3)--d. the parallel passage, PH 2. 86-'7, which correctly declares Something to be the highest
genus.
II For the classic discussion of this view see G. E. L. Owen in 'Aristotle on the
Snares of Ontology', in R. Bambrough (ed.), New Essays on Plato and Aristotle
(London, 11:)65), 69-95, repro in his Logic, Science and Dialectic: Collected Papers
in Greek Philosophy (Ithaca, 1986), 259-'78 at 260 and 264-5; although see his later
qualifications in 'Plato on Not-Being' ['Not-Being'] in G. Vlastos (ed.), Plato, i.
Metaphysics and Epistemology (Garden City, N], 1970), 223-67, repro in his Logic,
Science and Dialectic, 104-37 at 135-6. For an excellent recent discussion see L.
Brown, 'Being in the Sophist: A Syntactical Enquiry', Oxford Studies in Ancient
Philosophy, 4 (1986),49-'7.
ov
Victor Caston
we can use a distinct verb, 'to exist', for the ontologically marked
use, while reserving 'to be' for the copula and the particular quantifier 'there is/are'. tl The Stoics' claim can then be reconstrued as
follows:
But we should be more careful. Meinong does distinguish bestehen and existieren in a way that closely parallels this distinction, but
it is not the distinction for which he is famous. The objects that
subsist on his view are abstract objects, such as numbers, properties,
and states of affairs. IS They thus differ from a centaur, say, or the
object of my secret desire, which would be concrete if they were to
exist; but seem to lack being (Sein) entirely-such things neither
exist nor subsist. What distinguishes Meinongian objects, therefore, is not a certain kind of being, but rather their independence
from being of any kind, whether existence, subsistence, or some
other kind (should there be any). Their independence consists in
just this: objects have their nature or character (Sosein) independent
of being (Sein), independent of whether they are or are not; taken in
themselves, they are 'beyond being and not-being' (jenseits von Sein
und Nichtsein). Thus, when Meinong speaks of the 'AufJersein of the
pure object', he is precisely not assigning some new type of being
to objects, but rather considering them apart from being altogether
(literally, aufJer Sein).'6A true Meinongian is thus prepared to say
that objects have certain attributes or characteristics, even if they
lack being entirely. In fact, it is only this last thesis that has radical
implications. Someone who accepts abstract objects need not agree
to it, even if he thinks abstract objects do not exist, since he might
insist that attributes belong only to objects that have some sort of
being or other, even if less than existence in the full sense. Thus,
while there are non-existents on such a view, the logic remains fundamentally the same: the difference is merely verbal, with 'being'
in place of 'existence'. For the Meinongian, though, the difference
(NE') There are some things which do not exist.

The Stoics thus reject half of the Platonic position. Both accept
that if an item is-that is, if it exists-then it is something. But
Platonists also insist on the converse; and it is this that the Stoics
reject. An item may be something even though it does not exist. The
genus Something is thus wider than What Is, or as we might now
say, What Exists.
Such a position inevitably conjures up thoughts of Meinong and
his Gegenstandstheorie. This comparison has been thought to gain
support from a further distinction drawn in canonical Stoicism,
between what exists ('7'0 ov) and what subsists ('7'0 Vc/>C1'7'OS: Galen
Meth. med. 10. 155. 1-8)." Although some things do not exist,
they nevertheless subsist and even on occasion obtain (Vc/>C1'7'WC1L Kat
V1TapXOVeJlv: Pluto Adv. Colot. I I 16 B-C). Here we have a distinction
clearly associated with Meinong's name ever since Russell.'
II A striking example of the unmarked use of the quantifier occurs in Chrysippus,
who uses it to posit entities that, on his theory, are non-existents: 'given that there are
also such expressibles .. .' (OIl1'WV li( Ka.' 1'OLOV1'WV AfK1'WV: Quaest. log. 3, fro 3, col. 8.
15-16 von Arnim). To speak of an 'existential' quantifier in this context thus begs
the question against positions like the Stoics'. For a discussion of the larger issues
involved see A. Orenstein, Existence and the Particular Quantifier (Philadelphia,
1978), esp. 28.
11 For a defence of this distinction in connection with Meinong see A. A. Long,
'Language and Thought in Stoicism' ('Language and Thought'], in A. A. Long
(ed.), Problems in Stoidsm (London, 1971), 75-113 at 89""90, and later adopted
by Long and Sedley, The Hellenistic Philosophers, i. 164; but cf. G. Watson, The
Stoic Theory of Knowledge [Stoic Knowledge) (Belfast, 1966), 92-6, who more correctly draws a comparison with the early Russell (see n. 14 below). There has
been extensive controversy over the Stoics' terminology here: see R. E. Witt,
"Y1TO<71'a.OLS" in H. G. Wood (ed.), Amicitiae Corolla (London, 1933), 319-43;
H. Dorrie, "Y1TOO1'a.OLS, Wort- und Bedeutungsgeschichte', Nachricht. d. Akad.
d. Wiss. in Gott., Philol.-hist. Kl. (1955), 35""92; P. Hadot, 'Zur Vorgeschichte
des Begriffs "Existenz": v"a.PXfLV bei den Stoikern', ArchifJ fur Begriffgeschichte,
13 (19 69), "5-27; A. Graeser, 'A propos v"a.PXfLV bei den Stoikern', Archi'V fur
Begriffgeschichte, IS (1971), 299-305; V. Goldschmidt, 'v"a.pXfLV et txpf01'a.Va.L dans la
philosophie sto'icienne', Re'Vue des etudes grecques, 85 (1972),331-44; and A. Schubert, '''v"a.PXfLv'' und "txpt01'a.08a.L" bei den Stoikern', ch. 5 of his Untersuchungen
zur stoischen Bedeutungslehre (Gottingen, 1994).
.. For the popular interpretation of Meinong see B. Russell, 'Meinong's Theory
of Complexes and Assumptions', Mind, 13 (1904), 204-19, 336-54, and 509-24,
repro in Douglas Lackey (ed.), Essays in Analysis [Essays] (New York, 1973),21-'7 6
IS3
at 21, 36; Introduction to Mathematical Philosophy (London, 1919), 169; and 'My
Mental Development', in P. A. Schilpp (ed.), The Philosophy of Bertrand Russell
(2 vols.; Evanston, Ill., 1944), i. 1-20 at 13. At points Russell recognizes that this
is not Meinong's position: see his review of Meinong's Untersuchungen :mr Gegenstandstheorie und Psychologie ['Review of Meinong'], Mind, 14 (1905), 530-8, repro
in Lackey, Essays, 77-88 at 78; as well as his 'On Denoting', Mind, 14 (1905), 47993, repro in Lackey, Essays, 103-19 at '09. In fact, the position Russell ascribes
to Meinong and criticizes was actually Russell's own only a few years earlier: The
Principles of Mathematics, 2nd edn. (London, 1937; 1st edn. 1903),43,71.45-1.
" A. Meinong, 'Ober Gegenstandstheorie' ('Gegenstandstheorie'l, in A. Meinong (ed.), Untersuchungen:mr Gegenstandstheorie und Psychologie (Leipzig, 1904),
I-SO, repro in the Alexius Meinong Gesamtausgabe (Gesamtausgabe], ed. R. Haller
and R. Kindiger (7 vols.; Graz, 1968-78), ii. 481-535 at 486-8 .
" Meinong, 'Gegenstandstheorie', 489"'94, esp. 492-3; the expression 'jenseits'Von
Sein und Nichtsein' occurs at 494.
Victor Caston
is not verbal. The question is whether having an attribute presupposes any kind of being at all-whether 'Pegasus is a winged horse'
can be true, for example, even if 'Pegasus' fails to denote any sort
of being. I shall refer to a theory as 'Meinongian' if it rejects such a
presupposition, even if it departs from Meinong's theory on other
points. 11
If we now reconsider the four types of subsistent acknowledged
by canonical Stoicism, we find that they are all abstract objects:
place, time, void, and what can be expressed by language. The Stoics insist that such objects are not bodies and cannot act or be acted
upon. On their view, bodies are alone capable of this and as such
constitute the sole existents (see p. 208). Such characterizations are
very much in line with what Meinong has to say about subsistents.
But they are neither here nor there as regards 'Meinongianism' in
the sense I have described. For these later Stoics could still insist
that attributes belong only to entities with some form of being or
other, that is, which either exist or subsist. 18
centaurs and giants, objects that would be concrete were they to

have being. What counts as something, moreover, is characterized
quite broadly. Anything we can think of falls 'within the nature of
things' and thus is something, whether or not it has being of any
kind-all that matters is that we can think of it. Our thoughts,
therefore, not only have a content on every occasion, but an object:
even if a thought fails to correspond to an existent object, there still
will be something of which it is a thought. And, as a Meinongian view
requires, we can truly say that such objects are centaurs and giants,
regardless of whether they exist. On its most obvious construal, the
principle underlying their position guarantees not only that there
is an object of which we are thinking, but also that it is the sort of
thing we are thinking of-if I am thinking of an F, then there is
an x such that I am thinking of x and x is an F. (For more on this
principle see Section S. 2.)
This new characterization of the genus Something is more nuanced than it at first appears. At first, one might think it amounts
to,the following claim:
154
2.2
Thought and object
That having been said, it appears that some Stoics were committed
to the more well-known Meinongian position:
To some Stoics it seems that Something is the first genus, and I shall explain
why. In the nature of things, they say, some things exist and some do not.
In fact, the nature of things includes even those things that do not exist
but come to mind, such as centaurs, giants, or anything else that, having
been made up by a false thought, takes on an appearance, even though it
does not have reality. ,. (Sen. Ep. 58. 15; cf. S.E. M. 9. 49)
As on the view just considered, what exists does not exhaust all
there is: there are things that do not exist. But on the present view
they are not assigned any other type of being-they are simply
said not to exist. The examples are quite different: they include
" I am thus using the term in a somewhat more stringent sense than R. M.
Dancy (,Ancient Non-Beings: Speusippus and Others', in his Two Studies in the
Early Academy (Albany, NY, 1991),63-119, esp. 63-4 and 72-6), who characterizes
the acceptance of any kind of non-existent (including subsistents) as 'Meinongian'.
II As Watson, for example, takes the Stoics to do, in line with the early Russell
(Stoic Knowledge, 95).
'Substantia' is the Latin for ouota and so stands for a real object; 'substantiam
habere' can be used as an equivalent to the marked use of 'esse' (as e.g. at Sen. Ep.
I 13. 4). See p. 169 below.
I.
ISS
(TS) 'ix (It is possible to think of x~x is something).

But put in this way, the claim is nothing remarkable-every ontology is committed to (TS). The universal quantifier in (TS) is
objectual and so ranges over the objects in the domain of an ontology. But, according to any ontology, every object in its domain is
something in the broad sense in question here, no matter what the
ontology. The consequent of (TS) is thus trivially satisfied in every
ontology, and so (TS) will be as well.
The Stoic claim, however, is meant to be controversial, precisely
because the range of what we can think seems to outstrip most
ontologies. We generally assent to sentences of the form
It is possible to think of x,
where 'x' is replaced by a singular noun phrase such as 'Chiron'

or 'a centaur', even if nothing in our ontology corresponds to that
phrase. And that is just where the Stoics' principle kicks in. For
whenever we assent to such a sentence, they claim, it will also be
true that
x is something,
where 'x' is replaced by the same phrase-a result most of us ob-
157
Victor Caston
viously cannot accept, since we think there are no such things as

centaurs, for example, or giants. In accepting every instance of t.he
following schema (where 'x' is replaced by a singular noun phrase)
The view that locates the summum genus in the object thus turns out to
be justified. Everything there is is an object of a possible presentation;
everything there is is something ... In short, everything which is not
nothing, but is in some sense 'something', is an object. 12
15 6
(TS') It is possible to think of x~x is something

the Stoics are thus making a radical claim. Substitutional quantification shows how capacious their highest genus is really meant to
be.
Such a principle need not weaken their sense of reality. For these
Stoics do not accept every instance of the following schema (where
'x' is replaced by a singular noun phrase):
(TE) It is possible to think of x~x exists.
Consider 'Chiron' again. Here the consequent is as false for the
Stoics as it is for anyone else; hence, since the antecedent is true,
the conditional as a whole will be false. These Stoics are in fact quite
restrictive about what exists, reserving it exclusively for those things
that can act or be acted upon. Thus, while they reject the modern
dogma that 'Everything exists', 20 they nevertheless maintain a more
stringent, causal conception of reality than many philosophers do.
Perhaps the most tantalizing aspect of Seneca's report is the suggested connection between thought and object. If these Stoics were
also to accept the converse of (TS'), that is, the thesis that everything can be thought, by accepting every instance of the following
schema
(ST') x is
something~it
is possible to think of x
Twardowski's position is a natural extension of the one Seneca reports. What there is is the same as what there is to be thought-a
Parmenidean formula (DK 28 B 3), even if taken in a very unParmenidean spirit. But the Stoics need not have taken this step.
They might have held, like Meinong's student Ernst Mally, that
being something is independent not only of actually being thought
(as on Twardowski's view), but of even the possibility of being
thought. On Mally's view, the genus Something extends even beyond the range of thought-some things cannot be thought. 23 Or,
alternatively, the Stoics might not have taken any position at all.
There does not appear to be sufficient evidence on this question.
As Seneca does not identify the Stoics who hold this doctrine,
we do not know when it originated H or even whether it is related to
the other division of the highest genus into existent and subsistent
objects. The two characterizations are compatible; but they are not
equivalent, as the examples of centaurs and giants show. Thus,
while it is possible that some Stoics combined the two,25 it is by no
means certain-they may not belong to the same stage of theorizing
or even to the same theory. Without further evidence, we simply
far as we have a presentation of it, in contrast to us and our presentational activity
concerning it.'
Twardowski, Inhalt und Gegenstand, 37-8.
E. Mally, 'Ober die Unabhiingigkeit der Gegenstiinde vom Denken', Zeitschrzjt
fur Philosophie und philosophische Kritik, 155 (1914),37-52. For a discussion of these
issues see D. Jacquette, 'The Origins of Gegenstandstheorie: Immanent and Transcendent [ntentional Objects in Brentano, Twardowski, and Meinong'. Brentano
Studien, 3 (1990-1), 177-202 at 188-</0.
,. Brunschwig argues that the view Seneca reports is late, because he thinks this
position is an extension of Seneca's own (Ep. 58. 8-16, cf. 22), which is itself hetero
dox and late ('Genre supreme', 56-7; cf. Pohlenz, Die Stoa, ii. 37). But Mansfeld
has questioned whether the heterodox view is Seneca's own ('S~bsta~ce, Bein~ and
Division', 84-5 n. 22); and we may question whether the Memonglan vIew IS ItS
extension: nothing precludes it from being earlier than the orthodox, Chrysippean
one.
" Mansfeld does combine the two, taking Seneca's summary to be 'hurried and
incomplete' ('Substance, Being and Division', 101). But this assumption leads him
to worry whether Seneca confuses 'unqualifiedly non-existent' objects with those
which are only 'qualifiedly non-existent', viz. incorporeal subsistent abstracta; or
whether Seneca has only chosen the examples for rhetorical reasons (101-2). Such
worries evaporate if we are dealing with two different stages of Stoic theory.
II
lJ
they would be in a position to accept the following intentional

definition of 'something': to be something is just to be a possible
object of thought. The highest genus is described injust these terms
by Kasimir Twardowski, who, like Meinong, was a student of
Brentano's and advocated a 'Meinongian' theory of non-existent
objects. Having argued that every presentation has not only a content, but also an object,21 Twardowski continues:
'0 e.g. W. V. O. Quine, 'On What There [s', in his From a Logical Point of View:
9 Logico-Philosophical Essays (Cambridge, Mass., 1961), 1-19, esp. at I.
11 Zur Lehre ~'om Inhalt und Gegenstand der Vorstellungen: Eine psychologische Untersuchung [Inhalt und Gegenstand) (Vienna, 1894), 6, 29-34. Cf. also 36: 'Something is presented in each presentation, whether it exists or not, whether it is represented as independent of us and impressing itself on our perception or whether
it is formed in imagination by ourselves; whatever else it might be, it stands, in so
15 8
Victor Caston
cannot assume that both are in play in a given context. This a

question to which we shall return shortly (see pp. 175-6).
gument have failed, however;19 and so it is still an open question

how exactly Chrysippus used this fallacy. I postpone analysis of it
until our discussion of Chrysippus (see Section 8.3).
3. 'Not-somethings'?
Yet it is hard to see how such doctrines help. They only seem to
make our initial worry worse: if concepts aren't even something,
when 'something' applies so broadly, then they can't be anything
at all. But that, it seems, is just tantamount to excluding concepts
from Stoic ontology altogether.
Commentators are unanimous in not excluding concepts from
Stoic ontology, though. If concepts are not something, they maintain, they are not nothing either; instead, they are 'not-somethings' .16 This neologism is supposed to be the Stoics' own:
according to Simplicius, the Stoics declared common entities to
be ovnva (In Categ. 105. I I), a word ordinarily translated 'nothing' when in the singular, but etymologically composed from 'not'
(ou) and 'something' (n)-hence, in the plural, 'no-things' or 'notsomethings'. Simplicius mentions this term, moreover, in connection with an argument of Chrysippus' known as the 'No One' argument (0 '\6yoS" OonS"), which, it is thought, constitutes 'the formal
Stoic proof' of their anti-Platonic position.1 7 The argument runs
as follows: 'If someone is in Megara, he is not in Athens; but man
is in Megara; therefore, man is not in Athens.'lg The mistake, it is
argued, is to treat man as someone (nS"), or in general to treat any
common entity as something (n); they are 'not-somethings' instead.
Attempts to extract a criterion for 'not-somethings' from this arI . Prant!, Geschichte, i. 420, 427; Robin, Pensee grecque, 415; Couissin, 'Critique
du realisme', 403; Hadot, Porphyre et Victorinus, i. 159-62; Wurm, Substanz und
Qualitiit, 169; Pasquino, 'Statut ontologique', 378; Sedley, 'Stoic Universals', 87;
Long and Sedley, The Hellenistic Philosophers, i. 164; Brunschwig, 'Genre supreme',
passim.
17 The phrase is Sedley's ('Stoic Universals', 87). But Brunschwig also thinks that
the argument plays a central role ('Genre supreme', 80-6, cf. 92-1) and believes,
like Rist ('Zeno', 394-5). that it passed through Zeno's hands first, because of the
related argument offered by his teacher Stilpo (D.L. 2. 19). See sect. 8.3.
" In what follows I shall use the unfashionable 'man' as an unmarked term for all
human beings, because of a peculiarity of English usage which allows 'man' (but not
'human') to occur in statements without an article or quantifier preceding it--e.g.
'Man is a rational animal' or 'Man has reached the Moon'. This usage preserves
an ambiguity present in the Greek that proves crucial for understanding the Stoic
analyses of such sentences.
159
3. I. The place of concepts
It is surely right to keep concepts within Stoic ontology. The first

reason is that the Stoics seem willing to quantify over concepts,
since they recognize a plurality of them, including the concepts of
horse, man, and animal. 30 These are not chance examples. For the
Stoics posit a hierarchy of genera and species, which are defined in
terms of concepts:
A conjunction of many concepts that cannot be removed is a genus, e.g.
Animal, since this comprehends the different animals ... That which is
contained by a genus is a species, as Man is contained by Animal. That
which is a genus but does not have a genus, such as Something [olov T6 Tt),
is the most generic;" that which is a species but does not have a species,
like Socrates, is the most specific. (D.L. 7. 60-1)
The first definition commits the Stoics to a plurality of concepts:

every genus is a conjunction (av,\,\rJl!ILS") of many concepts, which
essentially constitute it and cannot be separated from it. 32 The
,. Of those who take the argument to provide such a criterion, Brunschwig has
been the most conscientious, not only in trying to formulate it precisely, but also
in testing it ('Genre supreme', 92-1). He demonstrates that it will not work with
mass terms such as 'water' or 'earth' (85): the sophism seems to behave in these
cases in the same way that it does for 'man'; and yet such terms clearly designate
something. Nor is the counter-example idle. For the same reasons, Brunschwig
points out (97-8), the 'OiJ.rI~ test' will be 'powerless' to justify the Stoics' inclusion of
void in the category of Something, since 'void' is a mass term-despite the fact that
the 'test' was supposed to distinguish precisely a non-existent something like void
from what is not even something. We can inoculate the test against this counterexample, Brunschwig argues, by restricting it to extracosmic void (99). But the need
for ad hoc restrictions just shows that the 'test' itself does not constitute a genuine
criterion after all.
lO Stob. Eel. I. 137. 1-3; D.L. 7. 60-1; S.E. M. 1 I. 9; Origen, In loan. 2. 13 93,
68. 29 Preuschen.
" Following EgJi's emendation (printed by Hiilser); the principal manuscripts
have oToy TOU (B) and oToy TOIJ (P), or omit the phrase entirely (F). See n. 10 above.
n The passage offers no further clues as to what av)J."I""~ means in this context.
But the only other logical use I have found supports the above interpretation, though
not itself Stoic: 'a species is not a differentia, but the conjunction of differentiae with
the genus' (Simp!. In Phys. 282. 14-15). This suggests that when the Stoics speak
of the genus as a conjunction of concepts, the concepts in question are not concepts
of the different species falling under the genus, but rather the general concepts
expressed in the definition of the genus and 'inseparable' from it in so far as it must
Victor Caston
distinction between genus and species, of course, is a relative one:

species will also be a genus if it contains other species beneath it,
~ust as a genus will be a species if it is contained by a higher genus;
m fact, all but the highest genus and the lowest species will be
both genus and species. But then-provided that a conjunction of
concepts is itself a concept-every genus and every species (above the
lowest) will be a concept, a point confirmed by Sextus Empiricus (PH
2.219). Concepts can differ both in extension, depending on what
falls under them (Stob. Ecl. I. 136.24-137.3), and in intension as
well, since e~ch will be constitutedfrom different concepts. Concepts
of nat~ral kmds are. not the only concepts, moreover. According
to Plotmus, the StOICS recognize concepts for each of the different
numbers, starting with I, which are distinct from the existent units
pairs, triplets, and so on, that fall under them (6. 6. 12. 13-29). (Fo;
more on genera and species see Section 7.2.)
The second reason is that the Stoics are quite happy to assign robust e~~stemologic~l a~d ps~chological roles to concepts. Concepts
are cntlcal for StOIC dIalectIc: each of the many kinds of division
the Stoics employ is defined in terms of genera and species (D.L. 7.
61-2), which, as we have just seen, are defined in terms of concepts.
Division is not simply a mainstay of Stoic argument. It is also the
~eans by which we articulate the structure and order of concepts
mto a system, which guides enquiry and ultimately the acquisition
of knowledge. Concepts thus provide a crucial underpinning for
rational investigation. Nor can this be a merefafon de parler. As we
shall see below, the Stoics offer a psychological account of concepts
and how they are related to our mental states, and they are willing
to speak of particulars as 'falling under' and 'participating in' concepts. All of these positions underscore the genuine role concepts
play within Stoic theory.
How could such a role be played by what is not something? David
Sedley is unusual in having candidly confronted this difficulty. 33 Although identifying concepts as 'not-somethings' relegates concepts,
he says,
160
~
ap~ly onlY,to those instances that actually fall under it. For species concepts to be
at Issue, au>J.T/""~ would have to signify disjunction; and this conflicts with its use
In grammatical and embryological contexts. It would also be at odds with the other
Stoic ~efinitions, since a lowe~t g~~us would be a disjunction of the species falling
under It as. well; yet these are ~ndlvlduals, such as Socrates, and not concepts. This
has led MIchael Frede to conjecture that the Stoics are committed to 'individual
conce~ts' as well ("~he Stoic Notion of a Grammatical Case' ['Grammatical Case'],
Bulletl~ of the Institute of Classical Studies, 39 (1994), 13-24 at 19-20). But given
the avallablity of a 'conjunctive' reading, such speculation is unnecessary.
to a logical and metaphysical limbo, the Stoics also give [them] a key role
in the business of dialectical analysis ... Is this a contradiction? I doubt
it. The logical and metaphysical outlawing of concepts is not a denial of
their epistemological value. It is a warning to us not to follow Plato's path
of hypostatising them. ('Stoic Universals', 89)
This response might be construed in either of two ways. (I) We

might understand it as advocating a form of fictionalism, or even
instrumentalism, about concepts: that while there are no concepts
strictly speaking, talk of 'concepts' is nevertheless a useful or convenient way of speaking, whose truth or usefulness is to be explained
hygienically, without reference to Forms or concepts. H But ontologically speaking, this just puts us back at square one, by excluding
concepts from Stoic ontology. Alternatively, (2) we might take the
warning against hypostatization at face value. Plato's mistake, on
this reading, does not consist in assigning Forms an ontological
status, but in assigning them an elevated ontological status. The
Stoics, in contrast, only give them their due: as not-somethings,
concepts possess an ontological status, though just barely. Unlike
the fictionalist reading, this is not an anti-realist reading. However
etiolated their status, concepts still fall within the Stoics' ontological scheme: they belong to the class of not-somethings. As Sedley
puts it: 'if a centaur is "something", this is not opposed to "nothing at all" in Stoic usage, but to "not-something" (outi), ('Stoic
Universals', 87).
Between these two readings, most commentators decidedly fall in
the second camp, although the former, more sophisticated, reading
" Stein too, we should note, is aware of the difficulty; but he attempts to solve it
by fabricating a distinction between VO~JLClTCl, which are valuable for acquiring knowledge, and Ivvo~JLClTCl, which are not (Erkenntnistheorie, 294-6). Such a distinction
would, ma~e nonsense of the reports concerning genera and species, which are said
to be VVOT/JLClTCl.
J4 This is suggested by Sedley's frequent reference to concepts as 'convenient
fictions' ('Stoic Universals', 87), 'mere projections', and even 'putative objects'
(,Stoic Universals', 89), not to mention his claim that statements about universals are
only 'shorthand for conditionals with indefinite subject terms' (,Stoic Universals',
88; cf. Long and Sedley, The Hellenistic Philosophers, i. 182) and so contain nothing
which denotes a concept. If this is the thrust of Sedley's interpretation, though, it
is misleading to describe the Stoic position as 'conceptualism'.
162
Victor Caston
may represent Sedley's considered view. 35 If so, his interpretation

will ~iffer less from the one I shall defend, at least as regards the ontological status of 'not-somethings'; there are also elements which
I think correctly characterize Chrysippus' position (see below, pp.
196-7). But it would still be mistaken as regards the nature and role
o.f co~cepts, as well as Zeno's and Cleanthes' overall strategy. FictlOnahsm about concepts adds nothing to fictionalism about Forms.
If fictional ism works at all, it will work straight away with Forms: we
can construe ~ll such tal.k as fictional discourse, without appealing
t? further fictions. The Introduction of concepts is entirely unmotivated on such a view. But the Stoics do assign a rich and positive
role to concepts, as we have just seen; and this suggests that they
adopted .a more realistic attitude. Not-somethings cannot play the
roles asslg.ned to concepts, if 'not-something' refers to nothing at
all. For this reason, I shall stalk the vulgate interpretation instead
according to which not-somethings form a discrete class of entities:
3.2. Against not-somethings
Despite its avowed intentions, on the vulgate reading the Stoics'

strategy is not genuinely deflationary-on the contrary, they posit
a whole new class of entities. And acknowledging such entities has
peculiar consequences, whatever their status. I shall mention three
such difficulties, before reconsidering whether concepts are notsomethings for the Stoics after all.
(I) If there are not-somethings, by definition they fall outside
the genus Something. But then either there will be a superordinate genu~, which i~cludes both somethings and not-somethings,
or there will not. Neither result is acceptable to the Stoics.
(a) If there is a superordinate genus (as in Diagram I), Something
would no longer be what many of our sources attest (see n. 10
abo,:,e), namely, the highest genus. Some commentators are willing
to bite the bullet here and simply reject this evidence, preferring to
speculate as to what the superordinate genus might be. 36 But the
" Sedley has said (in personal correspondence) that he never intended to ascribe
~ clas.s of not~so~ethi~gs to the Stoics when he contrasted 'not-something' with
not~mg at all . HIS claim was rather about the expression 'not-something', namely,
that It ~oes.not mean what the word 'nothing' does, even though both have the same
extensIOn (I.e. the empty, s~t); instead,. 'not-something' means what is not something
and therefore (on Sedley s mterpretatlOn) what is not a particular.

J. The candidate usually put forward (Rieth, Grundbegriffe, 90; Hadot, Porphyre

X?
..
Something
,.
Not-something
'"
What does not Exist
What Exists
DIAGRAM I
price seems prohibitive. It is true that Alexander of Aphrodisias

argues that there must be a genus higher than Something, such as
What is One (TO lv: In Top. 359.12-16). But he offers this precisely
as a refutation of the Stoic view. His attack would completely misfire
if th~ Stoics thought there were a superordinate genus; to interpret
him charitably, we should assume he has not gone so far wrong.
Something
Not-something
'"
What Exists
What does not Exist

DIAGRAM 2
(b) If there isn't a superordinate genus, on the other hand, then

Something can retain its status as highest genus. But it will no
longer be a fully comprehensive genus. There will be entities-notsomethings-which fall outside of it (see Diagram 2).37 If so, the
status of being highest genus is much less significant ontologically:
it does not tell us about all there is. It is 'highest' only in the sense
that there is no genus that subsumes it. In fact, it need not even
be unique. For Not-something will also constitute a 'highest' genus
in just the same sense, given that ex hypothesi it is not subsumed
by a higher genus. The Stoics seem to have had higher aspirations,
though. Plotinus says that the Stoics posited Something in order
to 'comprehend everything within a single genus' (6. I. 25 3-5 lVL
et Victorinus, i. 162) is What is Thought, TO voo"/-"VOV, a word that occurs in several
Stoic contexts: e.g. D.L. ,. 52 and S.E. M. 3. 40-50; 9 393-411. Such a conjecture
is incompatible with Seneca's testimony, however, which assigns anything that can
be thought to the genus Something and so would require us to posit a separate
group of Stoics who disagree with Seneca's Stoics, piling one conjecture on top of
another. I cannot, however, accept Brunschwig's objection to this conjecture-that
Ta voo"/-,,va are always particulars and therefore cannot include concepts which are
universal ('Genre supreme', 41)-for two reasons: (I) he assumes that concepts
are not-somethings, a position I reject below (see sect. 4); and (2) it would be
paradoxical for the Stoics to claim that concepts are not 'something thought' (see p.
165).
" This is the option Brunschwig endorses ('Genre supreme', 40).
Victor Caston
YJI' 7Tp,;>"al-'fjelJlOJlTa, Tel 7TelJlTa), 'ranking first a single genus of what
exists and the rest' (6. r. 25. 26-7 TOU ~JlTO, KaL TeVJI ~;>";>"WJI EJI n YJlO,
7TPOTelTTlJI).38 A number of criticisms, moreover, turn precisely on
the assumption that the highest genus is the genus 'of everything'
(7TelJlTWJI).3. In making Something the highest genus, the Stoics seem
to have thought it was also fully comprehensive and unique. But if
so, there cannot be any not-somethings.
(2) The claim that there are not-somethings-that there are items
which can be quantified over, even though they neither exist nor
fail to exist-is in itself problematic. One might have thought this
shouldn't trouble philosophers accustomed to Meinongian heights;
it shouldn't be any more vertiginous than allowing non-existent
objects of thought. If there were Stoics willing to countenance
non-existent objects and even non-subsistent objects, why not notsomethings as well? It seems a little late to feel scruples over a
burgeoning ontology.
But in fact the vulgate interpretation is much more radical. It
seems to commit the Stoics to the claim that there is something
which is not something, a position which is straightforwardly contradictory. The sound of paradox can be avoided, of course, by
speaking of 'items' or 'objects' rather than 'something' when using
the particular quantifier. But at this point that is mere word-play.
We can use the particular quantifier without presupposing that an
item in the domain of an ontology exists, subsists, or has any other
kind of ontological status. But we cannot use the quantifier without presupposing that such items are something (cf. Plato, Parm.
r60 E 2--7)-which is just to say that 'something' must be a fully
comprehensive predicate. Everything is something.
(3) The most decisive objection, however, is that if concepts were
not-somethings, they could not even be thought. Sextus Empiricus
comments that 'It is not possible to be taught by means of notsomethings [Sui Tci)JI ovnJlwJI]; for according to the Stoics they are
non-subsistent for thought [avv7ToaTaTa Til S,aJlo{<;tJ' (M. r. r7). It
JI We should perhaps understand along similar lines the claim that something
is the 'highest genus of existents' (yvllcwTaTov TWV OVTWV: Philo, Leg. alleg. 3. 175)
and the genus 'of all existents' (1TC1VTwv TWV OVTWV: Scholia in Arist. Categ. 34b8-1 I).
These reports are not false-a comprehensive genus would trivially be the genus of
all existents-but so understood. they would not have the force they were clearly
intended to have. (Brunschwig is less charitable: cf. 'Genre supreme', 44-5.)
,. See esp. Alex. Aphr. In Top. 301. 25. 359. 12-16; S.E. PH 2.86-'7; cf. also Philo.
Leg. alleg. 2. 86 (rather than the more careless version at 3. 175; see n. 38 above).
r65
is not clear whether 'non-subsistent' is to be understood here in

the technical Stoic sense or not. But to say that not-somethings
are 'non-subsistent for thought' means, at the very least, that they
are unavailable to thought-that we cannot think of such items.
No other reading will make sense of Sextus' argument about the
impossibility of learning. But it also fits hand in glove with what
Seneca reports; in fact, the Stoic principle (TS') is just its contrapositive. Given that these Stoics are committed to every instance
of the schema
(TS') It is possible to think of x~x is something,
they will also be committed to every instance of the following
schema:
(TS") -,(x is
something)~-,(it
is possible to think of x).
But that is just to say in more careful terms what Sextus does: if a
sentence of the form 'x is not something' is ever true, then it must
also be true to say that' I t is not possible to think of x', for the same
value of 'x'. Or as Sextus puts it, not-somethings are non-subsistent
for thought.
Yet such a result, when applied to concepts, is intolerable. For
a concept (w0'YJl-'a) cannot function as a concept if it cannot be
thought-it cannot function in divisions and definitions or perform
any of the epistemological and psychological roles the Stoics assign
to them. To attribute to any Stoic the view that concepts are unthinkable is, therefore, uncharitable in the extreme. But if concepts
do fall within thought's reach, then by the same token they will fall
within the 'nature of things' for these Stoics and so will be something
after all, against the vulgate reading. We ought, then, to reconsider
the evidence for concepts as not-somethings.
4. The ontological status of concepts

4. I. Much ado
There are several reasons for doubting whether the Stoics ever
spoke of concepts as 'not-somethings'. To begin with, the word
oun is never explicitly applied to concepts. In fact, the term is quite
rare. Apart from the text of Sextus just quoted, there are only two
Victor Caston
other occurrences in Stoic testimonia. The first is in Simplicius, in

connection with Chrysippus' analysis of the 'No One' argument:
common entities (KoLva), he claims, are said by the Stoics to be 'notsomethings' (ounva: In Categ. 105. II). For a Platonist like Simplicius, though, common entities are just Forms; and so it would
follow that concepts are not-somethings only if the Stoics identify
them with Forms. A similar problem confronts the second text.
Origen compares the view that evil is nothing (OUSEV) to that of 'cer. tain Greeks [who] say that among not-somethings (TWV ou nvwv) are
genera and species, such as animal and man' (In loan. 2. 13.93,68.
28-30 )-items that Origen not only takes to be Forms himself,40 but
believes were Forms for these unnamed Greeks, too! But without
such identifications, neither text provides incontrovertible evidence
that the Stoics said concepts were 'not-somethings'.
One should also question whether oun is a technical term. In the
singular, it simply means no one (ounS') or nothing (oun), depending
on gender!l And, though rare, the plural does occur: in each case,
it has the same meaning as the singular, serving only as a slightly
more emphatic variant!] In fact, this is precisely how Origen understands it in the text just cited, where it is taken to be equivalent
to 'nothing' (ovSEv). To speak of not-somethings or no-things, then, is
just to speak of nothing at all. 44 As the Cyclops Polyphemus learnt,

'no one' does not name anyone, not even a 'not-someone'. Rather,
it signifies the absence of any such person. In just the same way,
'nothing' doesn't name anything either, not even a putative 'notsomething'. Far from designating an item belonging to a special
metaphysical category, it signifies the complete absence of any such
thing.
Thus understood, the only options for the Stoics are something
and nothing-tertium non datur. And, of course, nothing isn't anything at all. Therefore, everything in the domain of Stoic ontology
is something:
166
e.g. Prine. I. 4. 5, 67. 20-68. IS (esp. 68. 10); cf. I. 2. 2,30. 2-8; 2. 1. 4, 110.
7-10 and I I I. 7-9; 3. 6. 8, 289. 33-290. I Koetschau; cf. In loan. 1. 19. 113-15. See
Henri Crouzei, Origene et Plotin: Comparaisons doctrinales (Paris, 1992),457-63 .
\ At Prine. 2. 3.6 (cf. also I. 2. 2,28. 13-17) Origen rejects a view according to
which the exemplars that constitute the higher world are mere objects of thought:
'For that reason, we have just said that an exposition of this world is difficult, lest
it occasion in some the belief that we posit certain apparitions [imagines] which the
Greeks call i~'o.,; but nothing could be further from our reasoning than to say that
the incorporeal world consists in the mind's merely being appeared to [in sola mentis
fantasia] or in a stream of thoughts [vel cogitationum lubrico].' The language here,
though slightly jumbled, echoes our evidence from Stobaeus and Seneca-even
Wolfson, who links 'imagines' with Philo's EiKOV'~ ';OWf4o.,.o, (conveniently ignoring
'sola' in the second half of the line), is forced to cite Zeno's views when he comes to
the 'vel' clause at the end (The Philosophy of the Church Fathers (3 vols.; Cambridge,
1956), i. 270-4).
41 Pohlenz seems to have understood it this way: 'Dagegen wurde vielleicht erst in
der Spatzeit dem Etwas als logische Antithese ein 'Nichts' entgegengestellt (OV,.,vo.),
das natiirlich nur theoretische Bedeutung hatte' (Pohlenz, Die Stoa, i. 295).
" Apart from the texts cited from Sextus, Simplicius, and Origen, we have the
following plurals: Homer, Od. 6. 279; Aesch. Ag. 1099; Callim. frr. 254. 8, 255. 9
(21) Pfeiffer; Eustathius In Od. I. 249. 13 Stallbaum; Anth. Gr. 5. 8. I Beckby. (II.
7. 196 and Od. 2. 199 could be construed as neuter plurals; but they can also be read
as being in the masculine singular.)
.0
(S) '<;Ix (x is something).

The genus Something thus has the status of a unique and fully
comprehensive genus after all, the one genus that comprehends
everything there is-'something' is just opposed to 'nothing' on
the Stoic view. We should therefore reconstrue reports allegedly
about 'not-somethings' more simply as follows:
(A) Nothing cannot be thought!' (S.E. M. I. 17)
(B) Common entities are nothing. (Simp!. In Categ.
gen In loan. 2.13.93)
105. II;
Ori-
Both theses are familiar. As we saw in the last section, (A) is just
another way of saying what Seneca's Stoics say: what isn't anything
can't be thought!6 We have encountered a version of (B) before as
It is instructive to compare the excellent discussion of o,;Ov and f471OV in A. P. D.
Mourelatos, '''Nothing'' as "Not-Being": Some Literary Contexts that Bear on
Plato', in G. W. Bowersock, W. Burkert, and M. C. J. Putnam (eds.), Arktouros
(Berlin, 1979), 319-29. His main thesis might at first seem to support the introduction of not-somethings: he shows that in addition to an 'existential use', where
o,;~'v and f471~'v function like the negation of an 'existential' quantifieJ'-i.e. where
they signify that there is no such thin~they sometimes have a 'characterizing use',
where they mean 'not a something' or 'not a thing' in the sense of being of no account
or worth, like the Homeric o';",~o.vo~ (319-22). Nevertheless, Mourelatos points out
that 'The Homeric indefinite pronoun ov,..~, from which o';",~o.vo~ is derived, has
only the existential sense of 'nobody', never the characterizing sense' (322, my emphasis;
cf. esp. n. 6 on Od. 9. 364 ff.), citing in turn A. C. Moorhouse, 'A Use of ouS." and
1'''15.,,', Classical Quarterly, NS 15 (1965), 31-40 (cf. esp. 32).
Fabricius' gloss on M. I. 17 is thus correct: 'secundum Stoicos quos potissimum
Sextus oppugnat, nihili null am in mente conceptionem formare possumus' (Sexti
Empirici Opera Graece et Latine (2 vols.; Leipzig, 1840-1; 1st edn. 1718), ii. 12
note e; my emphasis).
Again, compare Twardowski: '''Nothing'' is not to be conceived as the name
of objects of possible presentation, but rather as a syncategorematic expression:
168
Victor Caston
well. If there aren't any common entities, it follows that there are
only particulars for the Stoics, just as Syrianus claims (In Metaph.
104. 21 ).
Syrianus' report has added significance. Had the Stoics identified Forms with concepts and included them within their ontology
as not-somethings, Syrianus' global claim would be false-there
would be common entities as well as particulars. But if there are
not, then the Stoics are not admitting Forms when they acknowledge concepts. The two are on a different footing. It does not follow, then, that concepts can be characterized as ouTtlla just because
Forms are.
4.2. Concepts as something
The only text, in fact, which suggests that a concept is not something
is our original passage from Stobaeus, who clearly states that concepts are 'neither something nor something qualified [j.L~T Ttlla. tvaL
j.L~T 7ToLa]'. But he omits a crucial word, a word we find in the parallel report by Diogenes Laertius. This has been overlooked because
virtually every commentator has understood Diogenes' report in
the same way as Stobaeus', which the Greek certainly allows. 47 But

it also allows for a different construal, which for our concerns
pivotal:
IS
A concept is an apparition of thinking, which is neither something existent

nor a qualified thing [oih. '1'1 !!. OUT. 7TOI()V]; but it is as if it were something
existent and as if it were a qualified thing [waav.l SI TI !!. Kal waavEI 7TOl6v],
as for example an image of a horse occurs even when none is present. (D.L.
7. 61 )"
On this reading, the Stoics do not banish concepts from the highest
genus at all, They just deny that concepts are anything existent, and
that, as we know, is still compatible with their being something. In
one stroke, this reading dissolves all of our earlier worries. Unlike
Platonic Forms, which are nothing, concepts are something for
Zeno and Cleanthes-they are non-existent objects.
Reading the text in this way further allows us to understand the
dual denial, otherwise not well explained, 49 that concepts are neither
something existent nor something qualified. What Zeno is denying
is that concepts belong to either of the first two Stoic categories.
Socrates, for example, is both
(I) a real subject (1J7TOK{j.L.IIOII) and an existent (511, ens) in virtue
of his existence
(ova{a,
essentia, substantia), and
"Nothing" indicates the limits of presentation, where it ceases to be presentation ...

Whoever, therefore, says he has a presentation of nothing has no presentation whatsoever; whoever has a presentation has a presentation of something, of an object'
(Inhalt und Gegenstand, 35).
., Rist (Stoic Philosophy, 156; 'Zeno', 394) is the lone exception, although he does
not draw any further conclusions from it. The only commentator to notice that there
is an ambiguity in the Greek is Brunschwig, who concedes that Diogenes' words
can be read either way, viz. as denying that concepts are somethings or as denying
that they are existents ('Genre supreme', 79-80). He rejects the latter reading on
the grounds that the OV in Diogenes' version can be understood as representing the
.tval in Stobaeus' version, where it functions unambiguously as a copula ('is not
something'). But, of course, this argument can be run exactly in reverse: the .tval
in Stobaeus' version can just as easily be seen as representing the OV in Diogenes'
version, which Stobaeus has construed copulatively, even though it can also be taken
as a predicate complement ('is not some existent'). There is no reason to think
that Stobaeus' phrasing derives from a more authoritative source than Diogenes'
and consequently no independent reason to favour the copulative reading. We do,
however, have independent reason to reject it: as I have argued above in sect. 3.2, it
leads to incoherence and is therefore a highly uncharitable reading.
Interestingly, Victor Goldschmidt takes Stobaeus' text only to deny that concepts
are existents, in line with the reading of Diogenes' text I have suggested-'il est
notable que Tlva presente ici Ie sens de "substance'" (Le Systeme storcien, 18 n. 5;
see also Wurm, Substanz und Qualitiit, 179-80). A similar thought presumably lies
behind the otherwise unfounded claim, made by both F. Ravaisson (,Memoire sur
Ie Sto'icisme', Memoires de I'Institut Imperial de France, 21 (1857),37-8 n. 4) and

Stein (Erkenntnistheorie, 293 n. 656), that Tlva and 'Troia represent the first two Stoic
categories.
Wolfgang Mann has argued (in his comment on an ancestor of this paper) that
Diogenes' words should have occurred in the reverse order to be construed this way:
that is, it should have been OV TI instead of TI ov. But parallel uses of TI OV can be found
with the required sense: Arist. Phys. I. 8, 191b24; Metaph. 5, 1010'20, 1084'30;
Alex. Aphr. Defato 176. 19; In Metaph. 265. 39, 309. II; S.E. PH 3. 148.
Long and Sedley (The Hellenistic Philosophers, ii. 181-2) offer three possible
interpretations of the denial that concepts are 'something qualified': either (I) it is
an unnecessary a fortiori conclusion (they are not even something, let alone something qualified): or (2) it is to emphasize that concepts are neither corporeal nor
incorporeal; or (3) it is to deny that concepts are either 'discrete entities' or qualities
(corresponding to the Platonic and Aristotelian positions, respectively). The first
two of these construals, though, do not seem especially apt in context, while the
last reading (which they favour) would make even more sense if we translate Diogenes as I have suggested: what is distinctive about Plato is not that he made Forms
something, but rather that he made them beings (otla{al) and the things that really are
(OVTWS oVTa). But once we translate the text in this way, the relevance of the Stoic
categories surely takes precedence.
Victor Caston
(2) something qualified (7TOL6v, quale) in virtue of a quality (7TOL6T'T/<;, qualitas). 50
These distinctions are clearly attested for Zeno, most notably in his
treatise 011 Existence (llpl ov(]ta<;, D. L. 7. 134.). There he takes the
cosmos, itself a material object, to be constituted by two ultimate
principles, matter and God, which are identified respectively as
qualityless existence (a7ToLo<; ov(]ta) and rational order ('\6yo<;). The
former is described as that in virtue of which things exist (Calcid.
In Tim. 290 'quod tam his quam ceteris ut sint causa est'; Plot. 6.
I. 25. 21-2 7Tapd. T7i<; iJ,\'T/<; . . . TOt<; aA'\OL<; TO tvaL V7T(lPXLV). The
latter is explained as a quality that permeates all matter and blends
with it totally, making the qualified things in the cosmos distinct
from one another. 51 Zeno clearly intends these distinctions to apply
at both the cosmic and the individual level. Thus, each qualified
thing is analysed as a total blending of qualities with existence
(w<; Td.<; 7TOL6T'T/Ta<; OV-rW Kat Td.<; ov(]ta<; cSL' o'\ov Kp&'wV(]OaL),51 while
the cosmos as a whole is considered an individual, 'the peculiarly
qualified thing constituted from the whole of existence' (TOV (I( T7i<;
a.7T6.(]'T/> ov(]ta<; lUw<; 7TOL6v).53 Zeno even finds an allegorical basis for
this doctrine in the Titan Coeus (Koto<;), whose name isjust a dialect
variant of 'qualified thing' (7TOL6<;). But according to the scholia on
Hesiod's Theogony, Zeno identified Coeus with quality (7TOL6T7J<;):54
as Cornutus explains, it is 'in virtue of [Coeus] that existing things
are certain qualified things' (ND 17, 30. 11-13 Lang KaO' OV 7TOL6.
nva Td. oVTa ((]Tt).
The remaining two Stoic categories, in contrast, are not attested

for Zeno. If these are a later development, as seems likely, we can
.0 For these distinctions see Stephen Menn's excellent reconstruction in 'The
Stoic Theory of Categories' (this volume).
" Calcid. In Tim. 289""'93; Themist. In De an. 35. 32-4 Hein2e; Hippo!. Philos.
21 (=Dox. Gr. 57!. 7-10); Epiphan. Adv. haeres. 3. 36 (=Dox. Gr. 592. 24). For the
Stoic doctrine more generally cf. Origen Orat. 2, 368. 1-19 Koetschau; S.E. M. 9.
75-6, 10.312; Galen Qualit. ineorp. xix. 478 Kuhn.
" Galen, In Hipp. De humor. I, xvi. 32 Kuhn, Nat.fae. !. 2, ii. 5 Kuhn; cf. S.E.
PH 3. 56-62; Plot. 2. 7. !. 8-12. For other relevant doctrines attested for Zeno cf.
Themist. In Phys. 104. 14-18 Schenkl; Hippo!. Philos. 21 (=Dox. Gr. 57!. 7-10,
23-5); Stob. Eel. !. 153.4-6. On Zeno's theory of total blending see J. Mansfeld,
'Zeno and Aristotle on Mixture', Mnemosyne, 36 (1984),306-12 at 308-10.
u D.L. 7.137 together with 148; Clem. Strom. 5.14,104.1-105. I Stiihlin; Euseb.
Praep. Evang. 15. 15. 1-3, ii. 379.9 Mras.
54 Seholia in Theog. ad 134,29. 17-30. 18 Di Gregorio; cf. ad 404,70.2, as well
as Etym. magnum and Etym. genuinum, s. v. Koros.

understand Zeno's characterization of concepts in a new light. He is
denying that concepts belong to any of the categories he recognizes
in the analysis of bodies. One clear reason is that each category
implies existence. Nothing can be a qualified thing unless it has a
quality blended with matter, and ultimately prime matter, without
which nothing can be a subject. But prime matter is identified as
existence. Therefore, concepts cannot fall in either category: they
are neither something existent nor something qualified and so not
bodies at all. But they may still be something.
5. The nature of concepts

The first part of Zeno's denial, then, is not problematic after all.
Much like Seneca's Stoics, Zeno is simply concerned with nonexistent objects of thought. On the reading I have proposed, it is
the second part of his denial that raises questions. For even if we
understand 'something qualified' in a technical sense, as applying
only to existents, there is still something startling about the claim
that concepts are not something qualified. It seems at odds with
the Meinongian doctrine we find in Seneca, which acknowledges
centaurs and giants. Worse, it seems as though concepts might
lack every quality whatsoever, in which case statements about them
could not be true in virtue of some feature they possess.
What Zeno actually says is more nuanced, though. It is true that
the concept of man is not literally a man, nor the concept of horse a
horse. But each is 'as if' it were an object of the relevant sort (waavi
7TOL6v). The analogy he offers is to an image of a horse that appears
in the absence of any horse. Since ex hypothesi there are no horses,
the image cannot be a horse either. Nevertheless, it is as if it were
a horse, in so far as it is an image or representation (avuTlJ7TWfLu) of
one (D.L. 7. 61)." Concepts are like Macbeth's dagger. There is
something which Macbeth sees, but it is neither a genuine dagger
nor anything existent; it is only as if it were an existent and as if it
.. The word aVaTlJ7Tw/"a is a hapax legomenon. But the verb it derives from is not:
it is frequently used of imagining and dreaming, especially when the object does not
presently exist: Philop. In De an. 261. 3 Hayduck; David, Pro/ego phil. 58. 13 Busse;
Ps.-Alex. In Metaph. 587. 8 Hayduck; Philostr. Imag. 2. 17.9; Ongen Adn. In Ex.
(=PG 17, 16. 36 Migne); Simpl. In Ench. 20. IS; Eustr. In Eth. Nic. 260. 4; cf. Philo
fro 19 Lewy (Sit::. d. preufJ. Akad. d. Wiss. ::. Berlin, phil.-hist. Kl., (932); Photo
Bibl. 280'38, b 32.
Victor Caston
were a dagger.56 To be sure, it is has some connection with being a

dagger and being, say, razor-sharp--otherwise Macbeth would not
be tempted to think it was a dagger. But it cannot literally instantiate
these qualities, since non-existent objects cannot literally cut or
pierce or, more generally, possess causal powers. 57
An apparition is thus not a mental state, but an intentional object

towards which certain mental states are directed. I t is the appearing
that is the modification of the soul (1Ta80s Y TiJ I/IvxiJ). 61
Not every apparition is a concept, though. 62 In particular, no
image is a concept, since concepts are not perceptual-they are
apparitions of thought (SLUYO{US):J the intentional objects of conceivings (EWOlaL)." But they are analogous to images in so far as
we can mistakenly attribute to them both reality and the various
qualities they seem to have. For example, one might think that the
sentence 'Man is an animal' is true because there is such a thing as
Man himself and he is literally an animal. But on Zeno's account,
Man is a concept, an apparition of the intellect, and so cannot genuinely be what he seems to be, with all the relevant causal properties,
such as the ability to breathe or reproduce. At most, it is as If he
were an animal. Still, this shows that concepts bear some sort of
internal relation to the qualities in question: what makes a concept
the concept of P, say, and not of G, is that it is as if it were P, but
not as if it were G. The relevant qualities determine the nature of
their respective concepts, even though they are not literally exemplified by them. And they determine them in such a way as to tempt
5.1. Apparitions
. I t is in this sense, no doubt, that Zeno defines a concept as an
apparition (r/>av-raO/-Lu) , as what appears (r/>U{YTUL) when we have a
non-veridical experience, something that seems to thought to be
the case (S6KT)OLS SLUYO{US: D.L. 7. 50).58 Normally, when we are
appeared to, there is something apparent (r/>uYTaoT6y) that both produces the experience and appears to us as a result. 59 But sometimes
we only have an appearing (r/>UYTUOTLK6y), which is not of things
as they actually are. Such an experience is still directed towards
something, but not what produces the experience-in such cases,
cause and object diverge. The Stoics describe the experience as 'an
empty attraction' (SLaKYOS AKVO/-L6s):
An appearing is an empty attraction, an affect in the soul which does not
arise from anything apparent, just as when someone fights with a shadow
or beats his hands against the air; for being appeared to is anchored in what
is apparent, but an appearing is not anchored in anything. An apparition
is that towards which we are attracted in virtue of an appearing, that is, an
empty attraction. It occurs in both the atrabilious and the mad. o
,. For a contemporary treatment along these lines see Don Locke, Perception and
Our Knowledge of the External World (London, 1967), esp. 16-19 and 95-8.
" Against Elorduy, who claims that according to the Stoics centaurs and giants
have bodies and causal powers (Sozialphilosophie, 87; cf. EI estoicismo, i. 250).
II D.L. 7. 61; Stob. Eel. I. 136. 23; PS.-Plut. Epit. 4. I I. 4-5 (=Dox. Gr. 400.
26-401. 5). Long and Sedley translate </>a.vTaal-'a as figment, following Sedley ('Stoic
Universals', 89), who argues: 'The force of this stricture must not be underestimated. For "figments" (phantasmata) are defined as the putative objects of delusory
impressions, and exemplified by the things falsely imagined by dreamers and madmen.' But the Greek is not so pejorative. All that is wrong with the impressions of
dreamers and madmen is that what they imagine does not in fact exist, which would
hardly make them 'nothing' in the Stoics' eyes. On the contrary, it is in virtue of the
fact that they can be thought of-by dreamers, madmen, logicians, or anyone at all,
including God (if Ps.-Plutarch is.right; see n. 62)-that they are held to be something
and so can genuinely be referred to.
,. Ps.-Plut. Epit. 4.12.3 (=Dox. Gr. 402. 6-<); Ps.-Galen, Hist. phil. 93 (=Dox.
Gr. 636. 12-13); cf. Nemes. Nat. hom. 6, 55. 19-20 Morani.
Ps.-Plut. Epit. 4. 12. 4-5 (=Dox. Gr. 402. 10-20); Ps.-Galen, Hist. phil. 93
(=Dox. Gr. 636. 14-16); Nemes. Nat. hom. 6, 55. 20-2.
.0
I73
., Plotinus and Proclus thus speak incorrectly when they describe concepts as 'a
posterior affection in us arising from things' (vaTfpov a7T() TWV 1TpaYI-'a.Twv YfyovVat EV
~I-'iv 1Ta.IJ'1l-'a: Plot. 6. 6. 12. 18-19) and 'activities arising from cases of being appeared
to' Vfpy~l-'aTa a1TO TWV </>aVTaatWV aVfYEtp0l-'fva: Procl. In Parm. 896b12-17 Cousin'),
unless the -I-'a endings are meant to indicate a kind of mental entity produced by,
and hence distinct from, a mental affection or activity.
.. According to Ps.-Plutarch, the Stoics distinguish the completely sensuous apparitions of non-rational animals from those of humans and gods, which are concepts
(Epit.4. I I. 4-5 =Dox. Gr. 40.26-401. 5). Long and Sedley reject this text as 'completely out of step with all the other evidence on .wO~l-'aTa', on the grounds that
not every apparition in a rational soul is a concept (The Hellenistic Philosophers,
ii. 185). While the last claim sounds reasonable, there is no direct evidence for it;
and given that the Stoics notoriously think that all cases of being appeared to in a
rational animal are rational (D. L. 7. 5 I), they might also hold that every apparition
is a concept. But nothing here turns on this part of the report.
., Stobaeus describes it as an apparition of the soul (.pIJX*: Eel. I. 136. 23), rather
than of thought, as in Diogenes' version (7. 61). But the fuller report in Ps.Plutarch
agrees with Diogenes' version (Epit. 4. I 1.4-5 =Dox. Gr. 400. 26-401. 5)
Following Sedley's extremely plausible conjecture, based on the parallel construction of 'apparitions' (</>aVTa.al-'aTa) and 'being appeared to' (</>avTaa{at): Sedley,
'Stoic Universals', 8~; Long and Sedley, The Hellenistic Philosophers, i. 182; see
also Atherton, Stoic Ambiguity, 144, 256. His suggestion is supported by a further
paral1el: just as concepts are considered a species of apparition, conceptions are considered a species of being appeared to (Plut. Comm. not. 1084 F). It would be strange
if the relation between the genera were not preserved in the species.
Victor Caston
mistaken attributions-they must, in some way, be 'like' the objects

they are of.b' Iconic representation is one such case. An image of a
horse is not only of a horse; it also looks like a horse. Conceptual
representation is analogous: the concept of a horse is not only of a
horse, but it seems like a horse as well.
thing different. The type of predication involved depends upon the

status of the object.
Meinong himself, of course, is a thoroughgoing literalist. Notoriously, he believes there are such things as round squares, objects
that are literally both round and square--in fact, that is one reason
why they cannot exist. But impossible objects are not the worst
of it. Consider a principle such as the following (where F is to be
replaced with any predicate complement):
174
5.2. Non-existent objects and their attributes
Since Zeno is committed to non-existents that are characterized by

attributes, at least in an 'as if' sort of way, he can be seen as accepting
a form of Meinongianism, recently described as a 'dual copula' theory, on account of the two ways it construes predications. bb When a
true predication involves an existent, such as 'Clinton is an animal',
it is true because Clinton genuinely is an animal; he literally exemplifies the property. But when a predication involves a non-existent,
say, 'Mickey Mouse is an animal'-or, following Zeno, 'Man is an
animal'-matters are different. Such sentences are true (when they
are true) because the objects in question stand in a different relation
to these properties: they are determined, or better, characterized by
such properties. b7 Although these predications are grammatically
of the same form as predications about existents, they signify some.S To my knowledge. Rist is the only commentator to have paid any attention to
this feature of concepts (Stoic Philosophy, 156-'7). But talk of likeness and analogy
might wrongly suggest that while the concept of F does not literally exemplify F, it
does exemplify qualities similar to F. The difference between a concept and what it
is a concept of is not, however, a matter of which qualities they exemplify, but rather
the relation they bear to exactly the same quality: the concept of F does not exemplify
F, but is determined by it.
The phrase 'dual copula' is due to K. Fine, 'Critical Review of Parsons' Nonexistent Objects', Philosophical Studies, 45 (1984),95-142 at 97, although the distinction
itself is usually traced back to E. Mally, Gegenstandstheoretische Grundlagen der Logik
und Logistik (Leipzig, 1912), 63-4, 75-'7; for a contemporary exploration see E. N.
Zalta, Intensional Logic and the Metaphysics of Intentionality (Cambridge, Mass.,
1988), ch. 2. Mally and Zalta are not the best examples, however, since they use this
approach precisely to deny the Principle of Independence (see E. N. Zalta, 'Lambert,
Mally, and the Principle of Independence', in R. Haller (ed.), Non-Existence and
Predication (Amsterdam, 1986),447-59) and so reject what I am calling a 'Meinongian' theory-in this respect, their theories are closer to Chrysippus' than Zeno's.
For an example of a strictlyMeinongian dual copula theory see W. J Rapaport,
'Meinongian Theories and a Russellian Paradox', Nous, I2 (1978), 153-80, esp.
158-65
., Exemplification and characterization need not be mutually exclusive: it might
be the case that anything which exemplifies an attribute F is also characterized by it.
I t is just the converse that is denied here: not everything that is characterized by F
exemplifies it.
175
(TF) It is possible to think of F ~3x(x is F).

If interpreted naively, (TF) quickly leads to paradox. For it seems
possible to think of an existent round square, to use Russell's famous
counter-example;b8 but then some round squares would be existent
after all and so exist, despite being impossible. Such results would
be intolerable, even for Meinong. One way to avoid the difficulty is
to restrict the range of F, so as to exclude problematic substituends
like 'existent' (a strategy sometimes referred to as the 'dual property', or as I would prefer 'dual predicate', approach). bQ Another is
to keep the range unrestricted, but abandon literalism and interpret
'x is F' differently in different cases (the dual copula approach).
Seneca's Stoics are clearly committed to (TF). For they accept
(TS'), the thesis that whatever we can think of is included in the
nature of things; and on its most obvious construal, this implies not
only (I) that there is some thing of which we are thinking, but also
(2) that it is the sort of thing we are thinking of (see p. 155 above).
But they need not interpret (TF) naively. On the contrary, there
is reason to think that they favoured a dual copula approach like
Zeno's. For they claim that the nature of things includes, in addition
to centaurs and giants, 'anything else that, having been made up by
a false thought, takes on an appearance [imaginem], even though it
does not have reality [substantiam], (Sen. Ep. 58. IS). The thoughts
., 'Review of Meinong', 8 I.
Meino~~'s initial response was to object to the inference from 'is an existent'
to 'exists' (Uber die Stellung der Gegenstandstheorie im System der Wissenschaften
(Leipzig, 1907), repro in his Gesamtausgabe, v. 16-18); but he later abandoned this in
favour of a 'dual property' approach, by restricting schemata like (TF) to 'constitutive properties', thus excluding 'extraconstitutive' ones like 'existent' or 'exists' (A.
Meinong, Ober Miiglichkeit und Wahrscheinlichkeit: Beitriige zur Gegenstandstheorie
und Erkenntnistheorie [Ober MiiglichkeitJ (Leipzig, 1915), repro in his Gesamtausgabe,
vi. 176-7). The phrase 'dual property approach' is again Fine's (see n. 66 above);
and the distinction can, once again, be traced back to Mally. For a contemporary
version see T. Parsons, Nonexistent Objects (New Haven, 1980).
Victor Caston
in question have objects that only appear like the genuine articles:
it is not just that they appear to exist; they appear to be the same
sorts of things. We are thus dealing with something very much like
Zeno's apparitions. Though it cannot be proven with certainty,
the doctrines are sufficiently close for one to think that the Stoics
Seneca has in mind just are Zeno and Cleanthes-it is without a
doubt the simplest hypothesis. Far from reporting the renegade
doctrines of some later and otherwise unknown faction, Seneca
'would then be concerned with the doctrine of the founders of the
school, a doctrine soon abandoned by 'Chrysippus, Archedemus,
and most of the other Stoics'.
Forms and particulars (Phaedo 100 D 6; for more on participation

see below, pp. 183-4).
The Stoics are not trying to preserve Plato's Ideas by identifying
or reducing them to concepts and so keep them within their ontology. They are trying to do without them, by replacing them with
something that actually does the job. According to the Stoics, when
the Platonists made claims about Forms, they spoke incorrectlytheir theory cannot be salvaged. The most that can be done is to
address the concerns that led Platonists to posit Forms; and this,
the Stoics believe, can be done without positing anything more
than concepts. They are eliminativist about Platonic Forms, not
reductivist. '
I t would therefore be a mistake to transfer what the Stoics say
about Platonic Forms to concepts, in particular their claim that
common entities are nothing. Suppose someone were to say about
witches, 'Look, it's all a fiction. There aren't any witches-they
don't exist. Witches are just women who don't conform to the prevailing conventions of their day.' If this last remark were taken as
an identity statement, it would follow that there aren't any nonconformist women, contrary to what the speaker obviously intends.
Such women and such behaviour are quite real-it is witches, properly speaking, that don't exist. Far from including witches within
our ontology, the speaker is trying to explain the phenomena that
lead people incorrectly to speak of 'witches'. The women they refer
to happen to be nonconformist, and this helps to explain why they
are called 'witches'. But no one can actually refer to witches for the
simple reason that there aren't any, just nonconformist women.
If the Stoics say that common entities are nothing, then they
mean just that, namely, that there are no Platonic Forms-they are
nothing at all. There are concepts, on the other hand, even if they
are not full-blown existents. They are possible objects of thought,
whose differences can be articulated and intelligently used in reasoning. By distinguishing concepts from each other and the objects
each stands in relation to, the Stoics show their commitment to
them. Concepts have a secure place within their ontological scheme.
6. The rejection of Platonic Forms

6. I. Elimination vs. reduction
Against the traditional interpretation, which identifies concepts
with Platonic Forms, I have argued that the Stoics treat them differently: though mere 'apparitions', concepts are something, not
nothing. This reasoning is borne out by Stobaeus' fuller report.
It doesn't say that Platonic Forms are concepts for Zeno and his
followers, or that the Stoics call Forms 'concepts'. They claim that
concepts are what the'Platonists were in the habit of calling 'Ideas'
(Eel. I. 136. 23-4 Taii-ra O U7TO TWV dpXa{wv lUas 7Tpoaayopl)f.a8a,).
For the Stoics, concepts are the only things the Platonists could
possibly be trying to describe, since (I) there are no such things as
Forms (dVU7TCfpKTOUS: 137.4-6) and (2) concepts play some of the
roles Forms were meant to play. In particular, they are supposed to
stand in similar relations. Both types of entity are said to possess
extensions: what each Idea is supposedly of in fact 'falls under' a
concept (136. 24-137. I TWV yap KaTa TO. VVO~J-taTa U7T07Tt7TTOVTWV
elva, TaS- lOEas-), such as men, horses, and 'whatever else they say
there are I deas of' (137. 2-3 Kat TWV ci'\'\wv o7Toawv Myoua,v lOEaselva,). This relation is said to be the reason why (yap: 136. 24) the
Platonists must have been describing concepts, in spite of themselves. To underscore the parallel, the Stoics appropriate the word
'participation' (/-l.TEXHv: 137. 5) to describe the relation between a
concept and what falls under it, the exact word that Plato uses for
177
'0 The only commentator who approaches an eliminativist reading is]. Mansfeld:
he claims that while Ideas do not exist, 'within ourselves there exist concepts' ('Zeno
of Citium' ['Zeno'], Mnemosyne, 31 (1978), 134-'78 at 156). This seems just a loose
way of saying that there are concepts, on Zeno's view; for, as we have seen, the Stoics
deny that concepts exist, given their definition as t/>o.v-ro.of.'o.Ta.
Victor Caston
One might object, however, that someone who accepts (TS')

cannot consistently be an eliminativist about Forms. For given
as he couches his rejection in the formal mode, in terms of the

expressions the Platonists use, since even nonsensical expressions
exist and can be referred to. Significantly, this is how Stobaeus first
introduces their position-concepts are what the 'ancients used to
call Ideas'-and it is the only strictly hygienic way of stating the
claim. The presence of such a nuance in our testimonia may be seen
as further confirmation of the eliminativist reading.
(TS') It is possible to think of x~x is something,

it follows that if it is possible to think about Platonic Forms, Forms
will be something after all, against what the Stoics claim. This is
where close attention to the logical form of (TS') pays off. For
although Zeno accepts every instance of (TS'), he needn't accept
. the antecedent in each case. By maintaining that the sentence 'It
is possible to think of a Platonic Form' is not true, he can avoid
being committed to the conclusion that they are something, without
altering his commitment to (TS '). 71 This response is analogous to
Aristotle's denial that Heraclitus accepts contradictions: even if
Heraclitus sincerely says he can, he can't actually do it-it's simply
not possible to believe a contradiction (Metaph. r 3, IOOSb 23 -6).
So, too, when people claim they are thinking about Forms, Zeno can
respond that they are confused. What they are actually doing, to the
extent they are doing anything, is thinking about the corresponding
concepts.
A neat trick, one might think; perhaps too neat. For why couldn't
it be applied in the case of centaurs and giants as well, or any
other putative non-existent? Zeno can respond that the cases are
not on a par, though. To deny that we can think of centaurs or
giants is highly counter-intuitive. But with regard to theoretical
entities, there is a serious question as to whether our words express
a coherent idea at all or just confusion. In such cases, our own
offhand impressions may not be enough to settle what we actually
succeed in thinking. And in the case of Platonic Forms, Zeno can
urge, we do not in fact think what we claim to. The closest we get
are thoughts of concepts.
Still, Zeno should be careful in how he phrases the eliminativist
denial. If he claims, in his own voice, that Platonic Forms are nonexistent-as Stobaeus at one point puts it (dVU7TllPK'TOU,: 137.4-6 ) Zeno will be guilty of the same confusion he accuses the Platonists
of, or else he will arguably have conceded that thought about Forms
is possible and so run foul of his own (TS '). But he is safe so long
" In a similar way, Zeno could avoid any commitment to impossible objects, such
as the round square, by denying that it is possible to think such a thing. But to
my knowledge we do not have evidence as to whether he would have accepted or
rejected impossible objects.
179
6.2. Zeno and Plato's Parmenides
The only direct evidence we have as to why Zeno responded to

Plato the way he did is Stobaeus' suggestion that concepts actually
possess the extensions Forms are alleged to. But this is not sufficient to explain the resulting account of concepts, nor the choice
of concepts over, say, words. The answer, I believe, can be found in
Plato's Parmenides, in the famous passage where Socrates suggests
that a Form might be a thought (vO"lfka) that occurs 'nowhere else
but in souls' (ouoafkov ... aAAoBL ~ V rpux ai,), in an effort to escape the 'Third Man' regress (132 B-C).'2 Here, as elsewhere,'J the
Stoics can be seen responding directly to a Platonic argument, by
isolating problematic premisses and rejecting them, thus evading
its conclusions.
Parmenides gets Socrates to agree that if each thought is one
thing, it is of something (TLvo,) rather than nothing (ouOvo,); and
if of something, then of something that is (OV'TO,) rather than of
something that is not (OUK OVTO,). But then, he argues, if each such
thought is of some one thing that is set over all the instances, then it
is the object of this thought rather than the thought itself which has
a better claim to being a Form. That is, Parmenides has Socrates
agree to the following three theses:
(I) Whatever a thought might be of is something. (132 B 7-1 I)
(2) Whatever a thought might be of exists. (132 C 1-2)
" This connection was first suggested by David Sedley, who used the passage
to establish the Stoic distinction between IwoLaL and EVV01/p.a-ra as their intentional
objects (,Stoic Universals', 88). But if I am right in what follows, the connection is
even more thoroughgoing.
7J For an analysis of the Stoic response to Plato's Sophist see Brunschwig, 'Genre
supreme', 66-73, which conclusively demonstrates how the Stoic distinction between bodies and incorporeals rests on a very precise response to the arguments in
the battle between the Giants and the Friends of the Forms.
180
Victor Caston
(3) Every thought about a kind F must be of some unique thing.

(132 c 3-7)7.
181
onto logically. But it can be so without being an existent, much less

a Form.
From Zeno's perspective, Socrates should not have allowed Parmenides to lead him down the garden path. Plato, of course, was
happy to go down that path. What impresses him about Parmenides' argument is precisely (2). Elsewhere he reaffirms the coextension of what is (ov)-understood both existentially and predicatively-with what is something (n). 76 What impresses Zeno, in contrast, is intentionality and the distinction between act and object,
(I). As his own position shows, (I) does not entail (2): we can accept
that there is always something a thought is of without accepting that
it is always something that exists. But if so, then Parmenides' argument goes awry. Far from establishing the existence of Forms, it
can be used to introduce non-existent objects of thought. In short,
Zeno emends Socrates' proposal: to evade Parmenides' criticisms,
he should appeal not to thoughts (vo~fka7'a), but immanent objects
of thought (tvvo~fa7'a)-non-existent objects which are as if they
were certain kinds of things. In this way, the Third Man regress
can successfully be disarmed. For the concept does not literally
exemplify the same characteristics as the objects that fall under it,
even though it is in some sense 'like' them; and it therefore does
not generate a 'new' man above them.
If we have been correct so far about the outlines of Zeno's theory,
the closest historical comparison is not to Locke, who explains
universals in terms of existent ideas, but rather William of Ockham
(at least in his early fie tum theory), Thomas Reid, and Kasimir
Twardowski." The central and most striking feature of all their
theories is the appeal to non-existent objects of thought. Twardoski
is perhaps the most succinct:
But if Parmenides' argument is to succeed, he must establish the

existence of a certain kind of object towards which our thoughts
are directed, and that requires that he understand (I) not merely as
the claim that every thought has a content, but that every thought
has an object; (2) and (3) should therefore be construed similarly.75
All general thoughts concerning a given kind will thus be about
a single, existent object, which Parmenides can then argue is the
Form itself.
Zeno is sympathetic to the main lines of this analysis. He agrees
entirely that we must distinguish the object from the act; and that
we should be more concerned with the object than the act. What
he disputes is the ontological status of the object and its title to
be a Form. To forestall Parmenides' conclusion, Zeno must reject
at least one of the premisses; and he is arguably committed to
both (I) and (3). (I), of course, is just (TS'), the doctrine which
Seneca reports and I have argued stems from Zeno. As for (3), different thoughts of a given kind F will involve the same object of
thought, given that (a) such thoughts are directed towards concepts
and (b) concepts are jointly individuated by their extension and
their composition from other concepts (as suggested by the Stoic
definition of a genus; see pp. 159-60). If you and I are both thinking
of the kind Man, there is a single thing we are both thinking of, the
concept Man, whatever else we might think about it.
Rejecting (2) is enough to do the trick, though. Although our
general thoughts are always of some single thing, it is not something
existent on Zeno's view, but something non-existent that is literally
in thought (;v + v67)fka)-to use Brentano's quite similar terminology,
a concept is an 'immanent' object, an object that is in some sense
contained in the thought. Such an object is distinct from the mental
act which is directed upon it; and it may not even depend on it
Plato's Ideas are nothing other than the objects of general presentations.
Plato attributed existence to them. Today we no longer do this: the object of
a general presentation is presented by us, but it does not exist-at most one
" Cf. Arist. Metaph. A 9, 990b24-6; Alex. Aphr. In Metaph. 88. 7-13.
" Against R. E. Allen (Plato's Parmenides (Minneapolis, 1983), 150-1, 152-3),
who defends a content interpretation. He argues that an object interpretation would
make the subsequent dilemma in Plato's text (132 c 9-1 I) 'impossible' and so is
precluded by it (154, 157). But this is unconvincing given his own reading of the
dilemma, which slides from one equivocation to another. On the contrary, an object
interpretation is perfectly compatible with the dilemma as stated in Plato's text,
against Allen's claims.
,. Soph. 237 c 10-D 4; Parm. 164 A 7-8 2, 166 A 4-6.

" On William of Ockham see M. M. Adams, 'Ockham's Nominalism and Unreal Entities', Philosophical Review, 86 (1977), 144-76; P. Boehner, 'The Realistic
Conceptualism of William of Ockham', Traditio, 4 (1946), 307-35. On Reid see
K. Lehrer, 'Reid on Conception and Nonbeing', in Haller (ed.), Non-existence and
Predication, 573-83, esp. 579-83; M. David, 'Nonexistence and Reid's Conception
of Conceiving', in Nonexistence and Predication, 585-99; R. D. Gallie, Thomas Reid
and 'The Way of Ideas' (Dordrecht, 1989), 107-29. For Twardowski's view see his
Inhalt und Gegenstand, 102-1 I.
Victor Caston
can speak of its existence in the sense that it manifests itself in the objects
of the corresponding particular presentations, in a form that has been
somewhat altered by the individual modifications of that object. (lnhalt
cases or terms (noli 1TTWaWII), namely, those which later Stoics call
'common nouns' (1TpoaTJyoptas: 137.5-6).79 Simplicius confirms this
terminology: concepts are called 'participable' (J-I-BKTa) because
they are participated in (a1T() TOU J-I-TExwBaL); cases are called 'bearable' (TwKTas-) because they are borne (a1To TOU TuyxallaBaL); and
attributes are called 'properties' (aUJ-I-fJaJ-l-aTa) because they are proprietary or belong to something (a1To TOU aUJ-I-fJfJTJKEllaL). Certain
Academics followed a similar pattern, calling qualities 'possessible'
(KTa) because they are possessed (a1To TOU xaBaL: In Categ. 209.
10-14; cf. 214. 24-7).
The appeal to 'participation' here is not crypto-Platonism. Zeno
is using the term simply as a place-holder for whatever primitive
relation actually holds between concept and object, much as Plato
himself does (Phaedo 100 D 6 01TTI S~ Kai 01TWS 1TpoaayopwOJ-l-EII'f}).
With one crucial difference. For Plato, the relation is a 'causal'
or, more precisely, an explanatory one. It is that because of which
(SL' on) an individual is, for example, admirable (100 D I): what is
admirable itself makes admirable things admirable (7TOLi: 100 D 5)
and is thus their explanans or 'cause' (aiTta: 100 C 6)."0 It therefore
constitutes their being (ouata) as admirable things-literally, their
being admirable--because it alone truly is admirable. What is not
(J-I-~ ovaL), in contrast, cannot be the ground of anything's having
that characteristic (H.Ma. 287 c).
Zeno can agree with a surprising amount of this. Concepts are
not responsible for the way things are. But that does not imply
that nothing is. On the contrary, Zeno endorses a 'simple-minded
theory of causation' much like Socrates' in the Phaedo. For any F,
what makes something F is just the quality of F -ness. It is wisdom,
for example, that makes someone possess the attribute of being wise
and moderation the attribute of being moderate (Stob. Eel.!. 138.
14-22). But, again, with one crucial difference. In every case, Zeno
takes the explanans to be, not a Form or a concept, but a body. 81 The
quality of F-ness is thus something immanent within the object,
und Gegenstand, 106)
Reid speaks similarly about Plato's Forms, identifying them as nonexistent objects of thought." Taken at face value, such statements
retain Forms within the ontological scheme and so appear reduc_tionist, rather than eliminativist. If so, Zeno takes a more careful
stand, by acknowledging the gulf that lies between concepts and
Forms.
7. Concept and object

The only features of Forms that Zeno retains, then, are (1) their
role as intentional objects of thought and (2) their possessing an
extension. Everything else he rejects, especially the features most
important to Plato. It is not simply that concepts do not exist.
Concepts are not paradigmatic instances or exemplars either: they
do not literally have the attributes their instances all possess, but
merely in an 'as if' sort of way. It follows from this that concepts
lack the most distinctive feature of Platonic Forms, their role as the
'causes' or explanantia (atrtaL) of why particulars have the characteristics they have. To use Plato's terminology, concepts are copies, not
models-it is the concrete individual which is the original. Concepts
are posterior to individuals in every way: onto logically, epistemically, and causally. They cannot therefore reasonably be considered
a new type of Form. They are simply the only thing that comes even
close to what Socrates and Parmenides are describing, namely, the
one item that 'stands over all the instances' of a kind.
7. I. The relation of concepts to objects

The most significant relation concepts hear to objects is that they
possess extensions. According to Stobaeus, objects such as humans
or horses fall under (V1T01TL1TT6I1TWII: 1. 137. I) the relevant concepts. Individuals such as ourselves can even be said to participate (J-I-TXLY ~J-I-as) in concepts, just as we bear (TtryXallLII) certain
11 T. Reid, Essays on the Intellectual Powers of Man (1785; new edn. Cambridge,
Mass., 1969), 5. 5 507-8.
,. The 'Stoic philosophers' in Stobaeus' report who 'call cases common nouns'
belong to a later stage, since Zeno distinguishes only four parts of speech: names,
verbs, articles, and conjunctions. The canonical five are due to the additional distinction between proper and common nouns, introduced only later, presumably by
Chrysippus (D.H. Demosth. diet. 48, 232. 19-233. 2; Compo verb. 2, 6. 17-7 13
U sener-Radermacher).
10 Phaedo 100 D-I02 E; H. Ma. 287 C-D, 289 D, 294 D, 299 E; cf. Euthph. 6 D.
.. Stob. Eel. l. 138. IS; cf. Simp!. In Categ. 209. 2-3, 217. 32-218. 2.
Victor Caston
that permeates its existence or matter, in virtue of which the object

is the sort of thing it is (0,' <> Ylveral 7"1: 1. 138.14,18). It is impossible
for an object in which a quality is present to lack the corresponding
attribute and so fail to fall under the corresponding concept (I. 138.
philosophers (221. 14,221. 34-222. 4, 223. 12). But the remaining

description is entirely Stoic. Simplicius immediately begins to pick
at each formulation, phrase by phrase, leaving no doubt that he has
taken it verbatim from a Stoic source (222.33-223. I I).
Each clause of the Stoic characterization has clearly been weighed
and the antitheses balanced. According to (I), existing things can
be grouped according to genuine differentiae. But if we push further and ask what is common to, say, Dion's wisdom and Theon's
wisdom, the answer will be (2a), that there is no such thing that can
be isolated in itself. If we were to divide each man up into his constituents, there would be no single thing that was in both of them.
Of course, it is still true in some sense to say that both 'have the
same quality', namely, wisdom. But (2b) there is nothing more to
this than the distinctive characteristics each possesses and the concepts they fall under-analysis literally terminates (ci7T6ATjyouaav)
with these.A3 In speaking of a peculiarity (loL6TTj,), the Stoics draw
a clear contrast with commonality (KOLV6TTj,): a peculiarity is not
something that different objects share, but rather that by which one
differs from the others (cf. 238. 10-20). Each peculiarity, moreover,
has (3b) an intrinsic character or suchness, literally from within itself (lg aVT17,)-it does not acquire this from some transcendent
quality, as Socrates seems to suggest in the Phaedo about the large
'in us' and the large itself (102 A-B). The point may be significant,
since Plato never says that the large in us is a 'cause', only the large
itself-he may be anxious about just the sort of independent role
Zeno is suggesting. For if the large in us suffices to make us large,
what need is there for another explanans?
On Zeno's view, then, what makes something the sort of thing it
is is just the peculiarity present in it and in it alone-there is no
single thing that makes all the individuals of a certain kind that
kind of thing. There is a single thing they all fall under, of course,
namely, a concept, and it is by means of this that we can think
of the suchness we find in each quality instance. But this hardly
implies that a such ness exists isolated in any way. Concepts are
merely intentional objects-they are completely without being and
so cannot provide the explanantia the Platonist seeks .
16-17,21-2).
The labour Plato assigns to Forms is thus divided. Concepts are

that which particulars fall under or 'participate' in, by means of
. which we can classify them and draw inferences about them; but it
is qualities that make objects what they are. This might tempt one to
think that in spite of the emphasis given to concepts in our sources,
Zeno's response to Plato is actually to be found in his teachings on
qualities (see n. 4). But qualities do not do all the work. Because
they are themselves bodies, they occupy discrete locations: each
is present in a single, contiguous, unified body (Simp!. In Categ.
214. 26-37). Thus, if two non-overlapping individuals, Dion and
Theon, are both wise, there will be a body present in each of them,
the quality of wisdom, that makes each wise. But the wisdom in
Dion and the wisdom in Theon are not the same body. When Dion
dies, his wisdom perishes with him, but not Theon's (S.E. PH 2.
228). They are what are sometimes called 'particularized qualities',
or 'quality instances', or 'tropes'. What they have in common is
simply the fact that they both fall under a single concept.
This division oflabour is confirmed by Simplicius, in a discussion
of whether anything is common to the different types of quality
mentioned by Aristotle in the Categories:
What is common to quality in bodies, the Stoics say, is '(x) a differentia
of existence that (2a) cannot be isolated by itself [OUK (bo8u~A1)7Tn}v KaB'
EauT~v], but rather (2b) is exhausted by a concept and a peculiarity [d,
lVVOWLa Kat l810T1)Ta ci7TOA~rOuuav]; and is characterized (3a) not by duration or strength, but rather (3b) by its intrinsic suchness [Tji l, aUT7),
TOIOUTOT1)TI], in virtue of which a coming-to-be of something qualified subsists [KaB'1/v 7TOLOU uq,tuTaTal riVEDIS]'. (In Categ. 222. 30-3)"
The opening phrase 'what is common to quality' (TO KO'VOV TTJ>

7TO'OTTjTO,) does not stem from the Stoics, but Simplicius, who uses
it in the larger context to discuss the positions of quite disparate
., Following Petersen's emendation of EVVOT/lJ.U, with Kalbfleisch, Hiilser, and
Long and Sedley-for the Iv vOTJfJ.U of the manuscripts (defended by Zeller, Phi/osophie der Griechen, iii/I. 98 n. I, and accepted by von Arnim). As Kalbfleisch notes in
his critical apparatus, the subsequent gloss EVVolq. at 223. 6 clearly militates in favour
of Petersen's reading.
18s
IJ d1l'oA~y .. v and KUTaA~y .. V Ei, often signify a process of analysis into ultimate
elements. Mereological and geometrical analyses are the most common--e.g. Pluto
Comm. not. I0781!-but Sextus Empiricus uses it for etymological derivations as
well (M. I. 242, 244).
Victor Caston
Simplicius' criticisms are also revealing. He is no longer reporting

the Stoics verbatim: what he attributes to them here is tangled with
his own point of view. For he begins by suggesting that the Stoics
claim that qualities do not subsist:
position-I suspect it is Cornutus'-in so far as it posits subordinate common entities, which are prior to individual characteristics
(In Metaph. 28. 11-40).8'
For just these reasons, though, we might suspect that the original
Stoic position was stronger and.more radical, namely, Simplicius'
first option, (a). On this view, there is no room for any genuinely
common entity, whether transcendent or immanent. For the Stoics,
common entities are 'nothings' (ou'TLva), as Simplicius elsewhere reports (In Categ. 105. I I). The only qualities there are are qualities
in the Stoic sense, which are peculiar to the individual to which they
belong. Any 'commonality' they have is due to their each falling under a single concept, which is entirely post rem. If so, their position
should not be confused with a more moderate Aristotelianism.
186
Still more absurd is to claim that (4) qualities lack subsistence [u.rroo1'aaLv],
but terminate in a concept, unless these [words] mean not (a) that it is
exhausted by a concept and a peculiarity on the grounds that the quality
is non-subsistent [wS' aVV1TOO1'llTOtJ oUaT/S' 1'TJS' 1TOL01'7)1'OS'], but rather (b) that
it cannot in itself be isolated in the way that substance [ouota] is, instead
being separable by thought and by a peculiarity. (In Categ. 223. 2-6)
I t certainly would be absurd for the Stoics to craft such an elaborate characterization of quality and, in the same breath, deny that
qualities even subsist, as (4) alleges. But (4) is easily explained if it
is phrased in the way Simplicius understands their position, that is,
as a denial of what Simplicius takes a quality to be, that is, quality
in the Platonic, not the Stoic, sense. Then the absurdity he envisages, (a), would just be the denial that anything common to Stoic
qualities subsists-a denial absurd to a Platonist, perhaps, but not
an incoherent position outright.
The fact that Simplicius offers two different interpretations suggests that it was not fully evident from the context which position
the Stoics took. Simplicius attempts to domesticate the Stoic position, by assimilating it to an Aristotelian one, (b). On this view, there
is something genuinely common to distinct objects; it simply denies that there are transcendent universals (ante rem), existing separately from the object. It is thus not eliminativist about common
entities (1'<1 KOLVa) at all, but embraces immanent universals (in re).
In another passage Simplicius describes such a position, without
attribution, in the following way: 'Those who destroy the nature of
common entities, while taking them to subsist in particulars alone,
do not consider them to be anywhere themselves by themselves' (In
Categ. 69. 19-2 I). Simplicius finds this sort of position less objectionable, since even if it does away with the true nature of common
entities as he conceives of it, nevertheless a kind of commonality is retained as an inseparable part of concrete objects. He thus
allows that such philosophers 'speak correctly about subordinate
commonality' (rijs Ka'Ta'T'TaYfJ-v'Y/S KOLVO'T'Y/1'OS), even if the unsubordinated commonality (dKa'Ta'TaK'TOS) or transcendent Form is not
given its due (69. 21-3). Syrianus similarly tolerates a later Stoic
7.2. Concepts as generic objects

Because concepts are themselves characterized by the attributes
their objects instantiate, they appear like all the objects that fall
under them. Therefore, even if they are not common entities, they
do seem to be general in some sense; and it is this which allows them
to serve as genera and species.
In an intriguing, but unparalleled, report (M. 7. 241 ff.), Sextus Empiricus refers to 'being appeared to generically' (YVLKaL
c/>aV'TaataL: 7. 246),85 as part of an extended catalogue of the different ways in which the Stoics say we can be appeared to. Although Sextus presents the distinctions as if they all belong to a
single, elaborate division, it appears to be a conglomerate of at
least two different divisions. 86 The division which interests us is
.. The position cannot antedate Chrysippus, as it depends on the distinction
between common qualities and peculiar qualities. But the priority of the common
over the peculiar is an unusual position for a Stoic to take: as Syrian us notes, it is
one of the things that makes Cornutus' position distinctive and more moderate (In
Metaph. 106.7-13).
" The context shows that this expression specifies a type of appearance, and not
the 'generic' case of being appeared to, against Heintz (Sextus Empiricus, 116-17).
He emends the plural y.vIKa{ to the singular on the grounds that the expression
'Y'VIKO~ X' always signifies the genus X. But this is refuted by the plurals .llilKa' and
y,vIKai ''71'~a.. ~ at PH I. 188. (The other arguments he offers for emendation are
equally nugatory.)
The division consists of three subdivisions: the first is based on whether one
is appeared to convincingly (ml1avai c/>avTaa{al); the second on whether one is appeared to truly (ci>''7I1.;~ cf>aVTaa{al); the third on whether one is appeared to securely
(KaTa>''71rTIKal cf>avTaa{al). The third is a natural subdivision of the second and so
188
Victor Caston
the second, according to which we can be appeared to (I) truly,

(2) falsely, (3) both truly and falsely, and (4) neither truly nor
man-since 'x is a man' is true of every individual man-he cannot

be found in any of the subdivisions in which any other one can. He
is a very peculiar man indeed.
The problem is more serious than just a commitment to kooky
objects, though-that much the proponents of this view had surely
bargained for. As Sextus construes it, the position seems dangerously close to violating the Principle of the Excluded Middle (where
'p' can be replaced by any declarative sentence),
falsely. Being appeared to generically is offered as an example

of (4).
To think of man generically--or as the Stoics also seem willing
to say, of 'the generic man' (0 YEI"KOS aI'8pw1Tos)-is not to think of
an average man so much as an arbitrary man: it is a way of focusing
on all and only those characteristics that every man has, in contrast
with average features, such as having 2.5 children, which might not
be possessed by any man. 87 More precisely, anything that can truly
be said of every individual man will also be true of the generic man,
but anything that fails to hold of even one will not. It will thus be
true that
(I) Man is an animal
since it seems as though neither alternative obtains in the case in

question. The only difference is that Sextus has a privative statement like (3) instead of the external negation of (2), namely,
(4) -.(Man is Greek).
since 'x is an animal' applies to any man you like. But it will not be
true that
(2) Man is Greek
since 'x is Greek' does not apply to every individual man. Nor will
it be true that
(3) Man is non-Greek
(PEM) p v -.p,
(~a.p~apo<;)
since 'x is non-Greek' does not apply to every man either. Sextus
elaborates this in the material mode: the generic man is neither
Greek nor non-Greek and, more generally, 'the genera of those
items whose species are of such-and-such a sort or such-and-such
a sort are neither of such-and-such a sort nor of such-and-such a
sort' (M. 7. 246). Thus, while the generic man is, by hypothesis, a
the two seem to go together, while neither seems to belong with the first. Long and
Sedley argue for a similar breakdown of the passage (The Hellenistic Philosophers,
ii. 242), but inexplicably print the second subdivision with the first (39 G) and not
the third (40 E). The first division is unlikely to be earlier than Chrysippus, the first
Stoic to whom the technical term 'convincing' (meavav) can be attributed: it occurs
prominently in the titles of some of his treatises (D.L. 7.190,199; cf. 200), but also,
significantly, to characterize certain ways in which we are appeared to: Galen, PHP
5 5 19~ 320. 16-18 De Lacy (cf. D.L. 7.89); Plut. Stoic. repugn. 1055 F-1056 A (cf.
1057 B-<).
" For a sophisticated and in-depth treatment of arbitrary objects see K. Fine,
Reasoning with Arbitrary Objects [Arbitrary Objects] (Oxford, 1985), esp. ch. I. For
a discussion of the principle of generic attribution and the alleged paradoxes it gives
rise to see 9-14; for a comparison with Meinongian objects see 44-5.
But it is clear, by just the same reasoning, that (4) is not true either,
since 'it is not the case that x is Greek' is not true of every individual
man. And if we construe this result in the material mode, as Sextus
does , then it follows that it is neither the case that Man is Greek
nor not the case,
(5) -.[Man is Greekv...,(Man is Greek)],
thus violating (PEM).
But the original Stoic formulation is more carefully stated in
the formal mode, in terms of truth: some cases of being appeared
to are 'neither true nor false' (OUTE aA'T/8ELS OUTE tjJWSELS). And that
makes sense. For (5) would be true only if 'it is neither the case
that x is Greek nor not the case that x is Greek' were true of every
individual man. But in fact it isn't true of any of them--on the
contrary, what is true of every individual man is that 'either x is
Greek or it is not the case that x is Greek'. That is, it is not (5),
but
(6) Man is Greek v -. (Man is Greek)
which is true. Far from violating the Principle of the Excluded
Middle, then, the original Stoic formulation is committed to it."
.. As one might expect given the Stoic commitment to t~e rul~ of Double Negation (D.L. 7. 69), since (~~p-+p) entails (pv~p). Certain StOICS, moreover, are
known to have criticized a position, possibly Epicurean, for violating (PEM): cf.
Plut. Comm. not. 1080c.
Victor Caston
I t does violate the Principle of Bivalence, however, since this is

also framed in the formal mode:
This violation of the Principle of Bivalence is not like other cases

where the principle is allegedly violated, moreover. It is not, for
example, due to a predicate's being inappropriate to an object: although we might hesitate to say whether 'x is green' can be applied
to numbers, no such problem is _posed in the present case by 'x
is Greek', since many humans are Greek. Nor is there any real
problem of fuzzy boundaries or vagueness, despite the fact that a
predicate like 'is Greek' depends on national affiliations. Precisely
the same problem arises for 'the triangle is right-angled', where the
boundaries are perfectly precise. For if this sentence is about the
generic triangle, it is neither true nor false, for reasons similar to
those outlined above.
Indeterminacy has a different ground here. The generic man is
what Meinong called an 'incomplete' object:9J he is determinable,
but not determinate. Like each individual man, it is true to say
of the generic man that he is either Greek or not Greek. But unlike them, it is not true to say of him either that he is Greek or
to say he is not. Typically, the distinction between determinates
and determinables is used to draw a contrast between objects and
attributes. But the generic man is quite different from attributes.
Consider the attribute expressed by the predicate 'is a man'. It
is simply false to say of this attribute that it is a man or an animal. In contrast, it is true to say of the generic man that he is
a man and an animal, and even to say that he is made of flesh
and blood and extended in space and time. Such predicates are
as true of him as they are of Socrates and Plato--they share this
in common. 9 The generic man differs from them not because he
(PB) Any declarative sentence 'p' (in the object language) must
be evaluated either as true or as false.
But on the view we have been considering, 'Man is Greek' turns
out to be neither true nor false. For suppose (PB) did hold. Since
(2) is not true, it will follow that it is false that 'Man is Greek'.
But then, given a classic conception of falsehood (which the Stoics
share), S.
(F) Any sentence 'p' (in the object language) is evaluated as
false just in case' -,p' is evaluated as true,
it follows that
(4) -,(Man is Greek)
will be true. But (4), by hypothesis, is not true. Therefore, even
though (2) is not true, it is not false either, thus violating (PB).90
The report that we are sometimes appeared to 'neither truly nor
falsely' is accurate: t on this view, some general thoughts will be
neither true nor false.92
S.E. M. 8. 103,323; D.L. 7. 65, 78.
.0 One could maintain that such cases are 'neither true nor false' without violating (PB), if one were to restrict (PB) to propositions (as Chrysippus does: see p.
193 below) and hold that sentences such as 'Man is Greek' do not express 'wellformed propositions' (Long and Sedley, The Hellenistic Philosophers, ii. 184). But
nothing is amiss with the syntax of such sentences; on the contrary, sentences of
the form 'F is C' include definitions and other true generalities, which do express
propositions. If there is a problem, then, it must be with the semantics. But unlike Liar sentences-the only type of declarative sentence for which we have evidence that a proposition fails to be expressed according to Chrysippus (see n. 100
below)-specifying the truth conditions for 'Man is Greek' does not involve contradiction. The issue is simply whether there is anything that verifies or falsifies
it.
., Being appeared to 'neither truly nor falsely' thus differs from the third class
of cases in the Stoic division, of being appeared to 'both truly and falsely', which
merely appear to violate the Principle of Non-Contradiction (PNe), but in fact do
not. The explanation Sextus offers (M. 7. 245) is entirely deflationary: while such
mental states are true in so far as part of what they claim is true, 'they are false
in so far as another part of what they claim is false, which clearly does not violate
PNC.
., Once again, care must be taken not to confuse statements in the material mode
with those in the formal mode (see p. 189 above). For if (PB) is framed in the object
language as
(PB') It is true that pv It is false that p,
it follows from (PEM) trivially, given an object-language version of (F) and Convention T,
(T) It is true that p +-+ p,
a convention that plausibly should govern the use of any truth predicate. In fact,
once we have added such a predicate, we can argue even more directly that
(.) I t is true that Man is Greek v it is false that Man is Greek
is true in the object language, since 'it is true that x is Greek vit is false that x is
Greek' is true of every individual man, and hence of the generic man. But notice
that (PB) is still violated on the level of the metalanguage: for while (.) is true, neither
disjunct will be evaluated as true or as false, thus reinstating the position at a higher
level.
., Meinong, ObeT Moglichkeit, 168-81.
The predicate is equally true of the generic man and individual men, even
though it holds in each case for different reasons. On Zeno's theory, the generic
man does not exemplify the attribute of being an animal, but is instead character-
Victor Caston
lacks these attributes, but because he lacks the further determinations by which they differ from one another. Thus, although it
is true to say he is spatially extended, it is neither true nor false
to say, for example, that he is six feet tall. And what we think
about does at times seem incomplete in this way: for I am able
to think of a man, without thinking of a man of any particular
height.'
On this view, genera and species are not attributes, or sets or
classes, or the extensions of these. They are objects of a certain
sort, characterized by exactly those attributes that determine their
extension. Viewed from this perspective, it is no longer odd to consider the individual Socrates as a species. He is not an exception
at all: he is an object, like all other species. To be sure, he differs
from others in so far as he is existent and corporeal, and not a
concept. But such differences are accidental to the example, since
there are other lowest species that are incorporeal, non-existent
concepts: those under the genus Concept, for example. What distinguishes all genera from lower species is that genera are less
determinate-they stand above the differences that separate the
objects falling under them. It is this feature that gives them their
'generality' .
self and so not a position any later Stoic could have held without
abandoning orthodoxy.96 Chrysippus steadfastly maintained that
there are no exceptions to the Principle of Bivalence-every proposition (dgtw/La) is either true or false 97-not just with regard to
future contingents,98 but even when confronting the Sorites,9 the
Liar,lOo and Democritus' cone paradox. 101
8. Chrysippus' critique
8. I. Problems with generic objects
However natural such a view might seem for Zeno, it would wreak
havoc on Chrysippus' theory. Because it violates the Principle of
Bivalence, it is not a position Chrysippus could have accepted himized by it (see pp. 173-5 above). One might object that the distinction provides
a counter-example: for 'x exemplifies being an animal' is true of every individual
man, but not of the generic man. But a dual-copula theorist should deny this last
claim: since 'exemplifies being an animal' is true of all individual men, it will be
true of the generic man as well. But in his case it will be true because he is characterized by the second-order attribute of exemplifying being an animal, unlike individual men, who exemplify it along with the first-order attribute of being an
animal. The same strategy can obviously be applied again to still higher-order attributes. For a dual-predicate response to this objection see Fine, Arbitrary Objects,
13-14.
.s Cf. G. E. M. Anscombe, 'The Intentionality of Sensation: A Grammatical Feature', in R. J. Butler (ed.), Analytical Philosophy, 2nd ser. (Oxford, 1981), 156-80
at 161.
193
It is unclear whether Cleanthes, falling between Zeno and Chrysippus, could

have tolerated violations of (PB), given his own response to the Master argument. I
do not know of any direct evidence either way; but it is worth noting that, according
to R. Gaskin's exhaustive survey, Cleanthes' rejection of the first premiss of the
Master argument can be understood as tantamount to an objection to (PB), on one
of the two main construals of this premiss, the broadly 'Priorean' one (The Sea
Battle and the Master Argument: Aristotle and Diodorus Cronus on the Metaphysics
of the Future (Berlin, 1995),297-8). Gaskin himself thinks that, even if this construal
were right, Cleanthes would have avoided rejecting (PB) in favour of Ockhamism,
because he was a Stoic; but that would beg the very question at issue here.
., Cic. De fa to 21, 38; Luc. 95; Tusc. I. 14; Pluto Comm. not. 1066 E, Defato 574 F;
Aul. Gell. 16.8. 8; Simpl. In Categ. 406. 34-407. 5. Cf. D.L. 7. 65-6; S.E. M. 8.
73-4; Stob. Eel. I. 621. 12-13 Hense; Suda I. 255 Adler. For further discussion see
M. Frede, Die stoische Logik [Stoische Logik] (Gottingen, 1974),40-1.
Cic. Defato 21, 38; Plut. Defato 574 F; Simp!. In Categ. 406.34-407.5.
Chrysippus notoriously held that, rather than admit indeterminacy, at a certain point one should not answer and simply 'be silent' in response to the question whether a predicate applies or not. For the most recent reconstruction of
Chrysippus' solution to the Sorites paradox (with references to the literature) see
M. Mignucci, 'The Stoic Analysis of the Sorites', Proceedings of the Aristotelian
Society, 93 (1993), 23 1-45.
100 Our testimonia concerning Chrysippus' solution to the Liar paradox are problematic and in conflict with one another. He knew himself that it was prone to
misinterpretation: in addition to his many responses to other solutions, he found it
necessary to write two introductions to his solution (D. L. 7. 196), as Riistow has
rightly pointed out (Der Lugner: Theorie, Geschichte und Auflosung (diss. ErIangen,
1910),63-5). With Riistow (80, 82 if.), I am inclined to think that Chrysippus' solution was that Liar sentences do not express propositions and so do not represent
exceptions to the Principle of Bivalence, which the Stoics explicitly frame in terms
of propositions. The failure of such sentences to express propositions is stated explicitly by Chrysippus in his Quaest.log. 1Il in fro 3, col. 10. 16-18, a treatise which,
even in its fragmentary state, is manifestly concerned with the lack of any simple
correspondence between language and what it expresses (e.g. co Is. 12. 13-19, 13.
30-3, 15. 12-14; cf. co!. 8. 12-22). This solution is confirmed by Alex. Aphr. In Top.
188. 19-28 (cf. 189. 13-15).
The absence of any proposition corresponding to Liar sentences, we should note,
also makes good sense of Chrysippus' refusal to accept or reject the validity of the
Liar argument (Cic. Luc. 96 'hoc negas te posse nec adprobare nee inprobare'), as well
as his denial that a disjunction composed of the Liar sentence and its negation is false
(Plut. Comm. not. 1059 D)-in the absence of determinate truth-values, both refusals
are entirely proper. Plutarch's further claim, however, that Chrysippus held that
such arguments are valid and have true premisses, while the contradictories of their
conclusions are true (Comm. not. 1059 D-E), must be dismissed. If correct, Plutarch '8
report would imply that Chrysippus abandoned the Law of Non-Contradiction in
Victor Caston
Worse still, it seems that the principle underlying generic objects

leads to absurdity and even paradox, especially when we consider
the highest genus Something, which is comprehensive and so includes everything within its extension. As a generic object, it cannot be characterized by any of the differences that characterize the
various objects that fall under it. Given that there are incorporeal
objects as well as corporeal ones in its extension, for example, it will
not be true to say either that the concept Something is incorporeal
or that it is corporeal, a criticism already made by several of the
Stoics' opponents. I02 Such a result would presumably be acceptable to people who posited so generic an object; and, in any case,
it clearly does not result in the absurdity that it is nothing at all or
that nothing applies to it (S.E. PH 2. 224; M. 10. 236), since every
universal predicate will still apply to it. Much worse does follow,
though. Since some things exist and other things do not, it will not
be true either to say that the concept of Something exists or to say
that it does not exist (Plot. 6. 1. 25. 8-10). In fact, since it includes
objects which are not concepts as well as concepts, and individuals as well as genera, it will not be true to say that Something is a
concept or even a genus.
Such considerations would appear to be decisive. Without seemingly ad hoc restrictions on the.principles underlying generic objects, paradox is unavoidable. But there is also reason to think that
these criticisms originated from within the Stoic school and were
directed at an earlier Stoic theory. The critique occurs most extensively in Sextus, where it ends with several arguments that clearly
derive from later Stoic material: they argue that general terms like
'man' and 'sees' need not refer to something shared by many individuals, but can be applied to one without applying to another and
so designate a distinct quality in each case (PH 2. 227-8). Moreover, the terminology-a 'common noun', for example, is said to
be 'introduced into the syntax of a proposition'-suggests that the
material is no earlier than Chrysippus. When Alexander claims that
'the Stoics say' that the concept of Something is neither corporeal
nor incorporeal (In Top. 359.14-16),103 therefore, he need not be
reporting anything from Zeno or Cleanthes themselves, but rather
something from their later Stoic critics. It is thus conceivable that
Zeno and Cleanthes were never aware of the difficulties in their
theory, and that it was only subsequent criticisms that paved the
way towards new developments.
194
order to solve the Liar Paradox, an option Chrysippus explicitly rejects (Quaest.
log. [[[ in fro 3, col. 10. 12-13). A simpler solution would be the following: what
Plutarch reports is not from a passage where Chrysippus is speaking in propria
persona, but rather a dialectical one in which he traces out the consequences of
a view he will reject, namely, that Liar sentences express propositions and so are
truth-evaluable.
For an excellent discussion, with a reconstruction along different (Kripkean)
lines, see W. Cavini, 'Chrysippus on Speaking Truly and the Liar', in K. Doring
and T. Ebert (eds.), Dialektiker und Stoiker: Zur Logik der Stoa und ihrer Vorliiufer
{Stuttgart, 1993),85-19. Like the solution above, however, Cavini's interpretation
preserves bivalence.
'0' Chrysippus' response--namely, that the surface is 'neither equal nor unequal'--would violate the Principle of the Excluded Middle, if what is not equal
is always unequal (as Plutarch explicitly assumes at Comm. not. 1079 c). But the
Stoics deny exactly that: according to Plutarch, they held that the proposition
If something is not equal, then that thing is unequal

is false (1080 c). The Stoics can therefore consistently uphold PEM, by accepting
x is
equalv~(x
is equal)
while denying
x is equal vx is unequal.
This appears to treat 'equal' and 'unequal' as mere contraries rather than contradictories, thus leaving the Stoics open to Plutarch's complaint that they never explain
what this 'middle' could be (1080 B). But this does not seem fatal, as there are
many answers open to them (e.g. the one Long and Sedley provide: The Hellenistic
Philosophers, i. 302).
,0, S.E. PH 2.223-5, M. 10.234-6; Alex. Aphr. In Top. 359. 14-16; cf. Plot. 6.
1. 25. 6-,].
195
8.2. Linguistic conventions

If this is right, we should expect Chrysippus to reject objects like
the generic man and to offer an alternative account of the intuitions
that lie behind it. That is precisely what we find. When introducing
Chrysippus' objections to Platonic Forms, Simplicius warns that
one must recall
the Stoics' conventions [aIlV~e!Lav) about generically qualified things [nov
how, in their view, terms are uttered and how common
entities [KOLVa) are said by them to be nothing at all [ot/nva]. (In Categ. 105.
Y(VLKWV 1TOLWV):
9-1 I)
'OJ That Alexander is speaking of the concept of Something, rather than concepts
in general, is clear from his use of the singular 'the concept' (TO ~vv6T)f'oa), in contrast
to the plurals 'bodies' and 'incorporeals' (aWf'oclTWv Kal aaWf'oclTwv). Alexander needs
only one case to show that the genus One includes the genus Something and not vice
versa; and the counter-example is the genus Something itself, which, unlike One,
does not fall under itself.
Victor Caston
The appeal to conventions seems to have the following point. Realists often take the use of general terms to have metaphysical implications. To cite the most notorious example, Socrates in the Republic
claims that he and his companions are 'in the habit of positing a
single Form for each of the pluralities to which they apply a single
name' (596 A). And Zeno seems to operate on a similar principle,
except that he posits non-existent intentional objects where the
~latonist posits Forms. Both posit generic objects, with all their attendant difficulties. To avoid them, then, Chrysippus must eschew
not this or that aspect, but the semantic principle that underlies
both. And that is just what Simplicius reports. Strictly, there are
no generically qualified things; therefore all talk ostensibly about
them, by means of common nouns, must be reconstrued. Syrianus'
report is similar: Chrysippus, Archedemus, and 'most of the Stoics'
thought that Plato's Forms were originally introduced 'for the use
of conventions involving names' (ovvTJO{a<;: In Metaph. 105. 22-3).
Chrysippus is an eliminativist about all generic entities, whatever
their ontological status.
We actually have some evidence as to how Chrysippus' analysis
of such conventions went. A definition such as
M. 8. 97) linked anaphorically, similar to our use of variables bound
(7) Man is a rational mortal animal

might easily be misconstrued as a statement about a generic object.
But against such a construal, Chrysippus argues that (7) has the
same meaning (8vva(-k' 'TOY alhoy oY'Ta) as the following universal
generalization (KaOoA'KoY: S.E. M. I I. 8): 104
(8) If something (n) is a man, then that thing (KLYO) is a rational
mortal animal.
Such generalizations have three significant features: (a) unlike definitions, they are conditional, rather than categorical, in form; (b) the
common noun has been shifted from subject position to predicate
position (in the antecedent); and (c) the subjects in the antecedent
and consequent are a pair of 'indefinite' pronouns (D.L. 7. 70; S.E.
104 Although Sextus begins by citing the writers of handbooks, Chrysippus is
explicity named as the source of this doctrine at II. I I (cf. Epict. Diu. 2. 20. 23); and this may even have roots in Zeno, who wrote a treatise entitled Ka9o),I/<&.
(D.L. 7. 4). On the meaning of ot TEXVoypo.r/>o, see Long and Sedley, The Hellenistic
Philosophers, ii. 18S; for its use in connection with rhetorical handbooks, see R. Bett,
Sextus Empiricus: Against the Ethicists (Adversus Mathematicos XI) [Against the
Ethicists] (Oxford, 1997),55.
197
by a quantifier, in this case a universal one: 105

(8') '<Ix (x is a man~x is a rational mortal animal).
Such sentences, according to Chrysippus, differ from definitions

'only in syntax' (",'Ail 'TiJ ovv'Tag,: S.E. M. I 1.8--9) and so signify the
same proposition. 106 But then there is no need to take a definition as
referring to a generic object. For (8) is clearly not a statement about
a generic object, but a generalization that applies to any individual
object that falls within the genus-that is precisely the point of
using singular pronouns in the subject position and shifting the
common noun to predicate position. And, since (7) does not differ from (8) in meaning, it should not carry any such implication
either. (~) is clearly intended as the analysis of (7), which lays bare
the underlying significance of our ordinary conventions involving
names.
The conditional form is also important, in so far as it removes
any hint of ontological import from the use of general terms. The
definition of centaur, for example, would (given Chrysippus' interpretation of conditionals)107 merely state that necessarily, if something is a centaur, then it will have certain features-if it didn't, it
wouldn't be a centaur. But there needn't be any such thing for the
conditional to be true; it just has to be the case that if there were, it
would have such features. Unlike Seneca's Stoics, then, Chrysippus
needn't be committed to this sort of non-existent in order to speak
intelligibly about centaurs.
With this strategy for avoiding reference to generic objects, Chrysippus can easily maintain the Principle of Bivalence. For although
the sentences we considered earlier,
(2) Man is Greek
10' For further discussion of Chrysippus' analysis of definitions as universalized
conditionals see P. Crivelli's careful 'Indefinite Propositions and Anaphora in Stoic
Logic', Phronesis, 39 (1994), 187-206, esp. 199-201; also Bett, Against the Ethicists,
54-6. For an attempt to delineate the differences between the Stoic use of anaphora
and contemporary predicate calculus see U. Egli, 'The Stoic Concept of Anaphora',
in R. Bauerle, U. Egli, and A. von Stechow (eds.), Semantics/rom Different Points of
View (Berlin, 1979), 266-83.
106 A point on which Frede agrees (Stoische Logik, 64), even though this undermines the hypothesis that subsentential expressions stand in a one-<lne correlation
to what they signify.
'07 S.E. PH 2. 110-12; D.L. 7. 73. See also e.g. B. Mates, Stoic Logic (Berkeley,
1953),47-51; Frede, Stoische Logik, 80-93.
Victor Caston
(3) Man is non-Greek,
each violate the Principle of Bivalence when taken as statements

about a generic object, once we construe them as generalizations,
(2') If something is a man, then that thing is Greek
(3') If something is a man, then that thing is non-Greek,
the problem evaporates. For (2') and (3') have determinate truth.values: both are false, as neither generalization holds universally.
Not all humans are Greeks, nor are all humans non-Greeks; there
are some of each. Such generalizations therefore do not pose a threat
to (PB).
By the same token, such generalizations do not force him to recognize incomplete objects. For the falsehood of (2') and (3') is
compatible with the truth of
(9) If something is a man, then [that thing is Greek v that thing
is non-Greek],
which Chrysippus surely accepts as true-on his account, this
would be the proper construal of the claim that 'man is either Greek
or non-Greek'. But from (9) it follows that there is no generic man.
For suppose there were such a thing. Then it would satisfy the antecedent of (9), since by hypothesis the generic man is a man. But
it would then follow that
(10) Man is GreekvMan is non-Greek,
which is false for Chrysippus, since he construes its disjuncts as (2')

and (3') respectively, both of which are false. lOS Nothing can be a
man and yet fail to be Greek and fail to be non-Greek, as the generic
man is alleged to. Consequently, there is no such thing as the generic
man, whether this is taken to be a Platonic Form or a Zenon ian
concept. For Chrysippus, talk of 'genera' and 'species' amounts to
nothing more than the relation between certain generalized (,indefi-
10. According to the Stoic definition, a 'disjunction' (S,,'.vy!'-'vov) is true, strictly

speaking, just in case exactly one of the disjuncts obtains and the disjuncts 'completely conflict'-it is impossible either that both be true or both false (S.E. PH
2.162,191; Galen, Inst.log. 9.17-10.2, II. 22-12. 18 Kalbfleisch; cf. Au!, Gel!.
16.8. 12-14). Objects that satisfy the antecedent in (9) will in general satisfy these
conditions, but not 'Man'. In this case the consequent is false.
199
nite') propositions and the singular ('definite') propositions that are

'subordinated' to them (v7TOT<h7'fLV).109
This strategy, of appealing to generalizations over individuals
rather than to special objects, is confirmed by other evidence.
Chrysippus is known to have compared Platonic Forms to certain
kinds of geometrical theorems:
Chrysippus compared such theorems, Geminus says, to Ideas. For just as
[Ideas] comprehend the occurrence of unlimited [objects] v.rithin defined
limits, so too in these theorems the comprehension of unlimited [figures)
occurs within defined loci. (Produs, In Eucl. 395. 13-18 Friedlein)
The context of this remark shows that the kind of theorem at issue
is one that does not lend itself easily to hypostatization, because it
holds for an indefinitely wide range of cases, circumscribed only
by a locus or general description: 'parallelograms with the same
base and along the same parallel lines have the same area' (Euclitl I. 35). Here we are not even tempted to take the theorem as
referring to a generic object such as 'the parallelogram', or even
several such objects, which are indeterminate in all other respects:
the perimeter of such parallelograms can grow indefinitely large,
in fact, while still bounding the same area. llo The theorem thus
applies to an indefinite number of cases along an indefinite range,
each of which is perfectly determinate and different from the others in some respects-the theorem states only those features which
they all must have in common. And the theorem itself, of course, is
not an object sharing these features, but a proposition. It restricts
which objects fall under it by expressing certain specific conditions:
for example, 'if something is a parallelogram of height h and base
h, then that has an area a.'
The implication for Platonic Ideas is evident. We should not look
for generic objects in this case either, but only the 'defined limits',
that is, the specifications that circumscribe the relevant range of
cases, from which generalizations can be made. No commitment is
necessary here other than to particulars and their relation to what
is expressible in language.
10. Following the very suggestive observation about the term 'subordination' in
Crivelli, 'Indefinite Propositions', 193-4.
110 I would like to thank Stephen Menn for having insisted on this point.
Victor Caston
200
8.3. The No One argument

Chrysippus uses a related argument to reject Platonic Ideas called
the 'No One argument' (Simpl. In Categ. 105. 7-20; D.L. 7. 187,
198, cf. 72). Despite some minor variations in our reports, the
correct version appears to be that found in Philoponus (In Categ.
72 app. crit.):'"
(I) If someone is in Athens, he is not in Megara.
(2) Man is in Athens.
(3) Man is not in Megara.

We are not told precisely how Chrysippus interpreted these statements or how he thought they constituted an argument against
Platonic Ideas. But the Neoplatonic commentators who report it
all locate the trouble in the use of the pronoun 'someone' (nS') and
the assumption that
(0) Man is someone,
from which the argument derived its name." 2 For the upshot of
Chrysippus' discussion-namely, that the universal man is no one
(oun,)-is an allusion to Odysseus' famous ruse. When the Cyclops Polyphemus asks who has blinded him, Odysseus replies 'No
One' (Dun,: Od. 9.366-460, esp. 366-7,409-1 I). To his chagrin,
Polyphemus learns that 'no one' is not a proper name, but a syncategorematic term: his neighbours do not rush to his defence when he
complains that 'No one is killing me', since in normal usage these
words signify that there isn't any malefactor, even if Polyphemus
takes them to express something else. Just so, the common noun
'man' does not function referentially like a proper name in sentences like (2) and (3), even if Platonists think it names an object.
Normal usage does not require us to posit such entities. To claim
that man is no one or, more exactly, nothing {oun)-the masculine gender of the indefinite pronoun n, being added here only for
colour--is precisely not to hold that 'man' names a shadowy 'not'" Following Michael Frede's plausible and interesting reconstruction (Stoische
Logik, 56-8), which is based on a conjecture about the significance of a book-title
listed between two books on the No One argument at D.L. 7. 198. His reconstruction
is confirmed not only by Marc. 217, which is printed in the app. crit. of Philop. In
Categ. 72, but also by Elias, In Categ. 178.3-5, as Mansfeld has convincingly shown
(,Versions of the Nobody', Mnemosyne, 37 (1984), 445-7 at 445-6).
'" Simpl. In Categ. 105. 14-16; Ps.-Philop. In Categ. 72 app. crit. (e cod. Marc.
217),11. 38-41 (counting from the top of the note); Elias, In Categ. 178. 5-7.
201
something'. It is rather to deny that 'man' names anything at all.

Like the crafty Odysseus, Chrysippus knows better. 'Man' signifies
something here as part of a more complex expression. 113
To show that the Platonists must reject (0), Chrysippus must aim
his reductio ad absurdum against the Platonists-it must be directed
ad hominem. The sophism must therefore be read in such a way
that (a) on Platonist assumptions, it constitutes a valid argument;
(b) they cannot accept the conclusion; and (c) they cannot abandon
any of the stated premisses. Only then can Chrysippus demand that
they give up the unstated assumption, (0), and so abandon Platonic
Ideas. The trick is to find a construal of propositions (1)-(3) that
will allow them to play this kind of role.
From Chrysippus' perspective, the only acceptable readings of
(2) and (3) render the sophism harmless. As they stand, (2) and
(3) are ambiguous. Both use the common noun 'man' (avBpw1To,)
as a grammatical subject without an article or quantifier. As we
have already seen, this can signify the genus, as in the definition
'man is a rational animal' (av8pw1To, fan '!iov AOYLKOV BVTjT6v: S.E.
M. I I. 8). But it also can signify a member of that genus, in a socalled 'intermediate' proposition, e.g. '[a] man is sitting' (avBpw1To,
KaBTjTaL: S. E. M. 8. 97), which is true just in case there is some man
and he is sitting (Alex. InAn.pr. 402.12-18; cf. Ps.-Apul. Int. 177.
17-3 I). If (2) and (3) are both read as intermediate propositions,
the argument runs:
(I ') If someone is in Athens, then it is not the case that he is
in Megara.
(2/) Someone is a man and he is in Athens.
" (3 ') Someone is a man and it is not the case that he IS In
Megara.
So understood, the argument is not only valid, but sound: both

premisses are clearly true. But the conclusion is no longer objectionable: of course the man in Athens is not in Megara. If, on the
other hand, (2) and (3) are both read as universal propositions, the
argun;tent runs:
'" The allusion may just possibly have greater significance. In the Sophist Plato
compaped his materialist opponents to giants (246 Aff.), materialists who offer a
causal criterion of existence much like the Stoics' (see Brunschwig, 'Genre supreme',
66-73). Here Chrysippus returns the compliment. The Platonist is like a blinded
Cyclops, someone who is semanticalIy one-eyed, so to speak, because he thinks
nouns always name an object, even when a noun applies to many things.
202
Victor Caston
(I') If someone is in Athens, then it is not the case that he is
in Megara.
(2") If someone is a man, he is in Athens.
(3") If someone is a man, it is not the case that he is in Megara.
Although the argument is again valid, this time the conclusion

is clearly fals~it implies that no one is in Megara, which, we
are assuming, is a populated town. This is plainly the outrageous
.conclusion the sophism is meant to suggest. But on this reading,
the second premiss is also false: it maintains that every man is in
Athens; yet not all of us are so fortunate. The argument, therefore, is unsound and harmless. For Chrysippus, the only way to
get the shocking conclusion with true premisses is to combine the
intermediate (2') with the universal (3"). But then the argument is
obviously invalid.
Chrysippus, then, can handle the term 'man' in several ways
without untoward results: it does not produce absurdity or impose metaphysical costs. The No One argument is meant rather
for someone who wants more from claims like 'Man is in Athens'
than (2') or (2")-someone, in particular, who mistakes 'man' for a
name, just as Polyphemus mistook 'no one' to be Odysseus' name.
Suppose there is such a thing as the Platonic Form of man and the
noun 'Man' names it:
(o'*') Man is someone/something.
Then the argument can be construed as follows:
(I') If someone is in Athens, then it is not the case that he is
in Megara.
(2011<) Man is in Athens.
.. (3'*') It is not the case that Man is in Megara.
So understood, the argument is valid. Moreover, the Platonist has
strong reasons to accept (I') and (2'*') and to reject (3). Common
sense stands behind (I '); and given that Man is instantiated in
Athens, the Platonist has good reason to endorse (2). But by the
same reasoning, (3) must be denied, since it implies that Man is
not instantiated in Megara, and we have just as much reason to
suppose that Man is instantiated in Megara as in Athens. Yet if the
validity of the argument cannot be called into question, or the truth
of (I') or (2'*'), then the Platonist has no choice but to abandon the
20 3
implicit assumption, (0), namely, that there is something named

by 'Man'.
Faced with this result, a Platonist might reconsider the commonsense intuition behind (I '). The Neoplatonic commentators, for
example, argue that if (I) is unrestricted, it is false; in fact, the
common entity, Man, is precisely a counter-example to it, since
(they claim) he can be in both places at once. (I) is true only if it is
restricted to individuals, i. e .
(I) If something is an individual and it is in Athens, then it
is not the case that it is in Megara.
But all that follows from (I) and the claim that Man is both in
Athens and in Megara is that Man is not an individual or a 'this
something' (T68t: n: Simp!. In Categ. 105. 11-12), something a
Platonist can obviously accept. From this perspective, the No One
argument is little different from the 'Sailcloth' objection in Plato's
Parmenides (I3I B): both crucially assume that nothing can be in
two distinct places as a whole at once. But that is precisely what the
Platonist denies.
The inspiration for the No One argument is often thought to
derive from a similar argument used by the dialectician Stilpo-a teacher of Zeno's (D.L. 2. 114, 120; 7. 2, 24)-to 'destroy the
Forms' (dv?7PL Kui TO. t:i87J). I I ' The argument runs as follows:
A person who says man exists [says] no one [exists];'" for he is neither
this one here nor that one there--for why should he be this one here rather
than that one there?-therefore, he is not even this one. Or again: lettuce is
not what is being pointed to, since there was lettuce a thousand years ago,
so this is not lettuce. (D. L. 2. 119)
Stilpo's argument is quite different. In the first place, it is an indifference argument ('no more this than that') and so differs in logical
form from the No One argument. But it also has a more radical
conclusion. I t is not simply that 'man' does not name any particular
man (as it is often taken to show)-if that were all, the conclusion
would be compatible with Platonism. Rather, Stilpo denies that
'man' can be predicated of any individual, just as he claims that
the green leafy substance before him is not in fact lettuce. The
"4 Thus e.g. Rist, 'Zeno', 394; Brunschwig, 'Genre supreme', 80-4.
I"
Taking >.<yov-ra and .tva, as implicit with !-'TJSlva. For a discussion of possible
emendations see G. Giannantoni, Socratis et Socraticorum reliquiae (4 vols.; Naples,
1990), iv. 105-6 n. 9.
Victor Caston
astonishing claim Stilpo makes at the beginning of the reportnamely, that 'no one exists'-is thus quite deliberate. Common
nouns can never be truly predicated on his view (Plut. Adv. Colot.
II20A-B; cf. Euseb. Praep. Evang. 14. 17. I, ii. 303.16-18 Mras).
The only expressions that can are proper names. 116
But the Stoics reject such sophistic reasoning, as it would force
radical and large-scale revisions in our ordinary speech. Zeno and
Chrysippus have no intention of disrupting such practices. Their
worry, after all, is what makes such predications true. This is clear,
moreover, from the two readings of the No One argument that
Chrysippus can accept: each exchanges 'man' as a grammatical
subject for 'man' as a predicate complement. Neither Zeno nor
Chrysippus thinks this commits us to Platonic Forms; and Chrysippus doesn't even believe it commits us to generic objects. The Stoics
thus attempt to steer a middle course between Stilpo's revisionism
and Platonic profligacy. The No One argument plays a distinctive
role in Chrysippus' theory.
key assumptions. First, Chrysippus must assume a certain semantic

asymmetry between the subject and predicate of a sentence. Otherwise, his paraphrases are pointless. Second, he must believe that
predicates are significant without committing us to the objectionable entities subjects would, were paraphrase unavailable. A proper
defence of these claims would take us deep into the intricacies of
the Stoic theory of 'expressibles' ('\K'Ta), a large and controversial
topic. So as not to try the reader's patience further, I shall leave
a detailed investigation for another occasion 110 and simply outline
what I take to be the salient features of Chrysippus' response.
20 4
9. Chrysippus' response
20 5
9. I. Attributes and expressibles
The Stoics begin from a distinction between naming and attribution, which takes its origin from Plato's Sophist (261 E-263 D). But
the Stoics mark this difference in strongly ontological terms. What
we name are bodies. But in speaking about bodies, we do something
quite different. We say of Cato, for example, that he walks or that
he knows-we attribute (Ka'TTJyopLa8aL) something to him. But that
which we attribute to him-that is, the attribute (Ka'TTJy6pTJp.a) I 2-is
not a body, even though it is supposed to belong to a body (Sen. Ep.
117. 13). There is, however, a close connection between attributes
and bodies. An attribute like being wise belongs to someone because
of the presence of a bodily quality, wisdom, which is its 'cause' or
Chrysippus' strategy, then, is to analyse our ordinary ways of speaking by means of a logically hygienic paraphrase that shows its commitments perspicuously. In particular, he relies on a technique of
shifting common nouns from the subject position of a sentence,
a position where they might appear to refer to Forms or other
generic objects, to predicate position, where they are supposed to
be harmless. A similar technique is used in the analysis of singular sentences, this time with proper nouns, making the sentence's
existential commitments fully explicit. 111
As familiar as this ploy is to contemporary philosophers-one
need only mention Quine in this connection 11M-it rests on several
1960), 37-8, 176-86; Philosophy of Logic, 2nd edn. (Cambridge, Mass., 1986;
1st edn. 1970), 25-{i. Mario Mignucci rejects any strict identification of the Stoics'
technique with Quine's in his superb 'Sur la logique modale des Stolciens' ['Logique
modale'], in. Brunschwig (ed.), Les Stoiciens et leur logique, 317-46, on the grounds
that the Stoics do not allow names to function as predicates (322-3). But then,
strictly speaking, neither does Quine, who modifies them to produce predicates.
Names become an inseparable part of a predicate, and this is very much like what
the Stoics do.
". I am much indebted here to Nicholas Denyer's reconstruction in Language,

Thought and Falsehood in Ancient Greek Philosophy [Language, Thought and Falsehood] (London, 1991),33-7.
'" See esp. Alex. Aphr. In An. pro 402. 12-19 and the illuminating discussion of it
in A. C. Lloyd, 'Definite Propositions and the Concept of Reference', in Brunschwig
(ed.), Les Stoiciens et leur logique, 285-<)5.
'" For the general issue see esp. W. V. O. Quine, 'Designation and Existence', The
Journal of Philosophy, 36 (1939), 701-<); 'On What There Is', 7-11. For details on
the technique of eliminating such terms see Methods of Logic, 4th edn. (Cambridge,
Mass., 1982; 1st edn. 1950), 44,278-83, and Word and Object (Cambridge, Mass.,
"0 This investigation forms part of my study of the Stoics in The Problem of
Intentionality in Ancient Greek Philosophy (CUP, forthcoming).
llu The more common translation of KaTTJyopTJp.a as 'predicate' causes unnecessary
confusion. In English, one predicates words of an object, a 'predicate' being itself a
part of speech. But, as modern commentators are often forced to point out, what is
said to be 'predicated' in Greek philosophy is generally not a word, but something
signified by a predicate, whose nature and ontological status are disputed by different theories. The Stoics are no exception: properly speaking, KaTTJyop~p.aTa are
never the predicates of a sentence. The English 'attribute' captures the topic-neutral
significance of the Greek exactly, and it allows us to see the Stoic characterization
of attributes as expressibles (>'EKTa) for the substantive move that it is.
Victor Caston
explanans-that in virtue of which such a person is wise (Stob. Eel.
developed, the Stoics seem to have realized that it could easily be

applied to the problems that gave rise to Plato's theory of Forms;
and if so, then there would no longer be any special need for a
theory of concepts, even when speaking about conceptions ([""OWL)
and preconceptions (7TpoA1]IjlHs)-all of this can be handled much
more simply and elegantly by appealing to expressibles alone. This
hypothesis gains strength if, as I have suggested earlier, there were
Stoic criticisms of the theory of concepts (see Section 8.1), In so
far as expressibles differ from concepts, they may offer a way of
escaping these objections.
Although expressibles are clearly characterized by reference to
language, they are not themselves linguistic items, like utterances or
inscriptions. They are what is signified by linguistic items: 121 speakers and non-speakers of a language can equally hear the utterances
of that language, but only its speakers can grasp what is signified. 124
But speech has intimate connections, both causal and intentional,
with the speaker's mental states; 125 and it is here we should look
for the nature of expressibles. In general, thought is able to voice
in words what appears to us (D.L. 7. 49): the Stoics characterize
an expressible as 'what subsists in virtue of an articulable state of
being appeared to' (TO KUTa >"OYLK~" c/>U"Tua{u" uc/>LaTClf.L"o,,: S.E. M.
8.70, D.L. 7. 63). This and other evidence suggests that expressibles, though distinct from mental states, nevertheless covary with
them. 126 Mental states, according to the Stoics, are capable of acting
and being acted upon, and so are considered existent bodies. 127 Ex-
206
1.
138. 19-22).111
But what is an attribute, if not a body? The standard answer is first

ascribed to Cleanthes, who devoted an entire treatise to attributes
(IhpL KUT7JYOP7Jf.LClTW,,: D.L. 7. 175): an attribute is something that is
or can be expressed, literally an 'expressible' (AKT6,,: Clem. Strom.
8.9. 26. 4). Our sources do not tell us anything more about how he
characterized them. But surely the thought is something like this.
To say that a quality is an explanans of an attribute is simply to say
that it makes a certain attribution true, e.g. to say of Socrates that
he knows. Our attributions are not always correct, of course. But
even in such cases our attributions will express the same thing as
they would if we were correct. The attribute is just that which is
expressed by an attribution. A quality is that in virtue of which the
corresponding attribute belongs to its host object (when it does so
belong).
Such observations hardly amount to a theory, though, and one
may wonder whether Cleanthes' remark involves anything like the
fully developed views about expressibles we find in other testimonia. Whatever he meant by calling attributes 'expressible', he
presumably thought it was compatible with Zeno's theory, since he
also appeals to concepts when responding to Plato (Syr. In Metaph.
105. 28-9). He is unlikely, then, to have advocated a radical departure from Zeno; and on such a theory, attributes do not playa
central role. Concepts and qualities do.
The situation is entirely different with Chrysippus. Syrianus
explicitly distinguishes the response of 'Chrysippus, Archedemus,
and most of the Stoics' from that of Cleanthes: the former appeal
to the use of language, and not concepts, which no longer seem to
get mentioned. 122 Once the theory of expressibles was sufficiently
'" The Stoics typically mark this connection by naming the bodily quality with
(what we would call) an 'abstract' noun formed from the verb that expresses the corresponding attribute: for example, walking (ambulatio), dancing (iipX'JUL,), shoeing
(U1TOS<OL,), greeting (1Tpoouyopw(n,), insulting (>'oLoopiu), and similarly for the names
of the virtues, vices, passions, and mental states in general. Cf. Sen. Ep. 113. 23;
Pluto Comm. not. 1084A-C; Ps.-Plut. Epit. 4.21. 1-3 (=Dox. Gr. 410.25-411. 13).
III Archedemus may deserve special mention here, as someon!! who explicitly
addressed issues surrounding expressibles, attributes, and Plato's theory of Forms
(Clem. Strom. 8. 9. 26. 4; Syr. In Metaph. 105. 21-3). For more on Archedemus
see E. Schmidt, 'Archedemos 5 und 6', Paulys Realencyclopiidie der classischen Altertumswissenschaft, supp!. 12, cols. 1356-92. (There is a brief discussion of his
criticism of Plato on 1364.)
207
III D.L. ,. 43, 62; S.E. M. 8. 11-12.

". S.E. M. 8. 12; cf. I. 145, ISS, PH 2.214,3.267.
115 D.L.7. 49,55;S.E.M.8. 80(Cf. 10.218); Galen,PHP2. 5.15-20,130.29-132.
2,3.7.42,220.16-18; cf. 2. 5.9-13,130.7-19. For the causal connections between
thou~t and speech see]. Barnes, 'Meaning, Saying and Thinking' ['Meaning'), in
Doring and Ebert (eds.), Dialektiker und Stoiker, 47-61 at 57-<).
". Expressibles are said to 'supervene on thought' (1TUpV,pLOTlJ.,,<vov Til oLuvoiq.: S.E.
M. 8. 12; Syr. In Metaph. 105. 25-30) and to subsist 'from' or 'out of'-that is,
as a consequence of-being appeared to (T(;'V lK TOUTWV U,pLOTU"EVWV: D. L. 7. 43);
cf. Origen, Drat. 27. 8, 368. 1-2. The translation 'supervene'-first used, to my
knowledge, by Catherine Atherton (Stoic Ambiguity, 254, 258)-thus seems quite
apt. A. C. Lloyd argues that 1TUPV,p<OTo.VUL signifies that one entity is parasitic or
dependent on another (,Parhypostasis in Produs', in G. Boss and G. Seel (eds.),
Proclus et son influence (Zurich, 1987), 145-57). But Stoic usage of the term may
not require anything stronger than the covariation involved in supervenience. Thus,
the Stoics maintain that place in general supervenes on the whole set of bodies
(7rUPV,p{oTuTa,: Simp!. In Categ. 361. 10-11).
'" Pluto Comm. not. 1084A; cf. Calcid. In Tim. 220; Nemes. Nat. hom. 2,20.
14-17,21. 6-<), 22. 3-6; Tert. De anima 5.3-6.
Victor Caston
pressibles, in contrast, are incorporeal, inefficacious, non-existent,

and yet still something. m They cannot, therefore, be reduced to
mental states or bodies more generally. Most importantly, expressibles do not figure in our mental states by causing them (as sensible
objects do). They are 'grasped' in thought, in so far as they constitute the intentional contents of thoughts. 129 They are what we think
about objects, rather than the objects we in general think about (S.E.
.M. 8. 406, 49-10). A clear distinction is thus drawn between content and object.
propOSItiOns, and the present moment-can even be said to obtain (V7TllPXLII).133 The difference between bodies and incorporeals
is thus much more like the distinction between concrete and abstract
objects, which each have a distinct type of being, rather than a distinction between those objects which have being and those which
have none at all. There is no evidence that Chrysippus accepted beingless objects of any sort. On the contrary, the division of the genus
Something into bodies and incorporeals appears to be exhaustive. 134
Finally, and most importantly, expressibles and concepts differ
with regard to their natures. Expressibles are not 'apparitions': in
general, they do not have the features their subjects have. The
attribute of being wise, for example, belongs to a person who possesses the quality of wisdom; but the attribute itself possesses neither the quality nor the attribute--it is just false to say that the
attribute of being wise is itself wise. But if so, then expressibles
need not pose a threat to the Principle of Bivalence, as generic objects do. Unlike Zenonian concepts, Chrysippean incorporeals can
be fully determinate entities.
Zeno and Chrysippus therefore share the same strategy only
at the most general level. Both seek to eliminate Plato's Forms
by replacing them with non-existent, intentional entities. But the
nature of these entities is quite different. Zeno posits apparitions
which are 'like' the objects that fall under them, generic objects
that possess all and only those characteristics that all of the objects
208
9.2. Content vs. object
Linguistic conventions, as part of meaningful discourse, presuppose something expressible, which constitutes the content of the
corresponding mental states, though not an existent or a body itself
(as mental states are). In answering Plato, then, Chrysippus also
appeals to non-existent, intentional entities, just as Zeno does. But
expressibles differ from concepts in significant ways. First, expressibles are not the only kind of non-existent Chrysippus recognizes.
According to the canonical list, there are four types of incorporealtime, place, void, and expressibles. 130 The other three are clearly
essential to the structure of the universe and its processes, and so
something; but because they neither act nor are acted upon, they
cannot be bodies or existents. l3I His motivations for positing nonexistents are not Meinongian. He is simply trying to be scrupulous
about the ontological presuppositions of physical theory.
Second, Chrysippus assigns incorporeals a positive ontological
status. Although they are non-existent, they are still said to subsist (v<j>{(]7'a(]8aL) , III and some of them-such as properties, true
'"~ Incorporeal: S.E. M. I. 20,7. 38,8. 12,69,75,409-10; PH 2.81,3 52. Inefficacious: S.E. M. 8. 406-10. Non-existent: Plut. Comm. not. 1074 D; Proc. In Tim.
iii. 95. 7-15 Diehl. Something: Plut. Adv. Colot. 1 I 16 B-C; S.E. M. 10. 218.
". S.E. M.8. 12,244; d. I. 20, 25, 28; PH 3.48,52; D.L. 7 51, 52-3.
110 The canonical list: Plut. Adv. Colot. I 116 B-C; S.E. M. 10. 218. Time: D.L. 7
140; Prod. In Tim. iii. 95. 7-15 Diehl. Place: Stob. Ecl.'1. 161. 8-II; S.E. M. 10.
3. Void: Stob. Eel. I. 161. 14-15; S.E. M. 10. 3; Cleomedes, Mot. circul. 8. 10-14
Ziegler. (For expressibles see n. 128 above.)
,,, Only bodies can act or be acted upon: Cic. Varro 39; Plut. Comm. not. 1073 E;
Nemes. Nat. hom. 2, 20. 14-17,21. 6-<) Morani; Tert. De anima 5. 4-5; cf. S.E. M.
8. 26 3. Bodies are the sole existents: Alex. Aphr. In Top. 301. 19-27; Plot, 2. 4 I. 7,
6. I. 25. 22-5,6. I. 28. 7; Plut. Comm. not. 1073 E; Anon. Pro leg. Plat. phd. 9 2-4,
14; Hippol. Philos. 21 (=Dox. Cr. 571. 23)
III Galen, Meth. med. 155. 1-8; Plut. Adv, Colot. I II6 B-C, Comm. not. 1081 F-
20 9
1082 A; Stob. Eel. I. 106. 18-19, 161. 24; Procl. In Tim. iii. 95. 7-15 Diehl. Cf.
Cleomedes, Mot. circul. 8. 10-14 Ziegler.
III Properties: Stob. Eel. I. 106.20-1. A property (UtJl-'j3Ej3T)K6s) is just an attribute
which actually belongs (UtJl-'j3Ej3..JK'll) to the body in question: S.E. M. 8. 100; Stob. Eel.
I. 106. 20-1; cf. Simp!. In Categ. 209. 10-14. True propositions: S.E. M. 8. 10,85.
The present moment: Plut. Comm. not. 1081 F-I082 A; Stob. Eel. I. 106. 18-19. For
a judicious discussion of 'obtaining', with references to earlier controversies, see M.
Schofield, 'The Retrenchable Present', in Barnes and Mignucci (eds.), Matter and
Metaphysics, 330-'74 at 349-58.
114 S.E. PH 2. 223, 3. 48; M. I. 19,8. 35, 10. 218, 234, I I. 224; Plot. 6. I. 25.
6-8. See also Brunschwig, 'Genre supreme', 31-2. We differ only with regard to the
apparent exceptions. On my view, there is no problem about concepts, since they
belong to a different stage of theorizing; for similar reasons, I am inclined to think
that the problem of fictional individuals like Chiron is no longer an issue either, given
Chrysippus' use of paraphrase. The one case that requires a decision is geometrical
limits, and I believe these should be classed with the canonical incorporeals that
supervene on bodies-they are a consequence of bodies and inconceivable without
them, Stob. I. 161. 20-2,25-6; d. also 142. 2-4, Cleomedes, Mot. circul. 16. 2-5,
14. 1-2 Ziegler.
Victor Caston
that fall under them share. Chrysippus wants to avoid such entities
and the difficulties they raise for the Principle of Bivalence, and he
calls in expressibles to fill the gap. Once we understand properly
what utterances express and what they presuppose, he believes, we
shall see that we do not need generic entities of any sort; in fact, we
need nothing more than we are already committed to if we are to
make sense of significant discourse. Thus, when Chrysippus comes
to. treat genera and species, he does so as part of his discussion of
expressibles (D.L. 7. 43, 200).
God exists, for example, unless we know exactly what is meant by

'God'. But definitions of 'universal' are much harder to come by
than one might have imagined. What we find instead is a great deal
of handwaving: either inadequate characterizations, such as being
'repeatable', which apply to entities that are not universals (such as
scattered particulars); or appeals to metaphor and intuition, such
as the same thing being 'in' each particular instance 'as a whole'.
Without clearer criteria, debates over the existence of universals
cannot but be confused and unresolvable.
The ancient and medieval debates are framed differently. Here
we find a relatively clear definition: a universal is an entity whose
nature is such that it can be attributed to more than one thing (Arist.
De into 7,17839-40). To deny that there are universals in this sense,
then, will just be to deny that anything can be attributed to more
than one thing: anything that can be attributed must be attributed
to exactly one thing-only man, for example, can truly be said to be
man and good good, ana so on. Just such a position is ascribed, in
fact, to several ancient philosophers, such as Antisthenes, Menedemus, Diogenes, and Stilpo. IJ5 But it is rejected by virtually everyone
else, and for good reason: it undermines discourse as we know it.
The existence of universals in this sense, therefore, can be questioned, but the answer is trivial. There is an obvious sense in which
virtually all of us are committed to such attributes. The real question concerns their nature: just what sort of thing is an attribute
and what is it to attribute it? On this point we find tremendous divergence. 'Attribute' and 'to attribute' are best viewed as theoretical
terms belonging to our folk semantics, which are accepted by nearly
everyone. What is in dispute concerns what kind of entities actually
play these theoretical roles.
I t is not surprising, then, that Platonists, Aristotelians, and Stoics
each call on their own special entities to play formally analogous
roles-they agree on this precisely because they are trying to answer the same question. What is significant rather is the detailed
differences between these special entities, since this is just where
substantive theories of attribution are to be distinguished. On this
210
10.
'The' problem of universals
A veteran of philosophical debates may be unimpressed by all this

subtlety. After all, both Zeno and Chrysippus commit themselves
to special entities with a special ontological status that stand in a
special one-many relation to concrete objects in their extension,
much as universals do. The appeal to quality instances may not help
either. On the contrary, one might suppose (with Michael Frede)
that this only makes their theory more like Plato's theory:
So it seems that, right down to the terminology, the Stoics have a view
quite like the Platonists considered earlier, except that the Stoics replace
Ideas by concepts ... Thus it seems that a case is something like the Stoic
counterpart of an Aristotelian form or of a Platonist immanent Form, as
opposed to the transcendent form participated in: i.e. it should be something like a common quality or an individual quality. (,The Stoic Notion of
a lekton', in S. Everson (ed.), Language (Cambridge, 1994), 109-28 at 123)
Given such large structural similarities, the differences between
these entities might seem to make little difference. If concepts and
expressibles play exactly the same role as universals, why should
we consider this as a rejection of universals at all, rather than a dIfferent conception of universals? Is there anything more than a verbal
difference here?
The answer to these questions crucially turns on just what 'the'
problem of universals is. Twentieth-century discussions frame it
as an existence problem: whether, that is, a certain sort of entity
exists (or subsists, or has being). The coherence of such an enquiry depends entirely on whether we are given an adequate characterization of the entity in question-we cannot evaluate whether
211
,,, Antisthenes: Arist. Metaph . ..:l 29, 1024b29-34; Simp!. In Categ. 208. 28-3 2,
211. 18-21; Ammon. In Isag. 40.6-10. Menedemus and the Eretrian school: D.L.
2. 134; Simp!. In Phys. 91. 28-33. Diogenes: D.L. 6. 53. Stilpo: Pluto Adv. Col.
1120 A-S; cf. Euseb. Praep. Evang. 14. 17. 1. For a discussion of these views see ch.
3 of Denyer, Language, Thought and Falsehood.
Victor Caston
approach, conceptualism forms a very distinct alternative from both

nominalism and realism, just as different 'nominalisms' do from
each other--those, for example, which appeal to set-membership
or to resemblance relations. The question is not whether these theories recognize a special one-many relation and special relata, but
how good a job their own candidates do in filling that role. The
terms of debate are completely different.
That having been said, the ancients are often concerned with an
existence problem. It is not the existence of universals that concerns
them, however, but the existence of Platonic Forms, the first serious
candidate for the office of attribute. Naturally, Platonists believe
their candidate is the right one for the job and so view attacks on it as
tantamount to a rejection of universals, Simplicius and Origen being
the clearest examples (see pp. 166, 186-7). The Stoics do not reject
qualities, though, especially what they call 'common qualities'; and
they certainly do not reject genera and species. They simply don't
think these entities are what Simplicius and Origen think they
are. The Stoics might, then, be able to accept 'universals' after
all, at least in the sense given above-what they reject would only
be the Platonic account of universals. If so, the title of this paper
would have to be changed to 'Something and Nothing: The Stoics
on Concepts and Forms', since it would only be Forms, properly
speaking, that are not anything at all.
But the Platonists might well object. Forms, they could argue,
are not merely important historically; they have something that any
candidate worthy of the office must have. For Forms are not simply
attributable to many things. They are also the ground or explanans
of things being the way they are-indeed, it is precisely for this
reason that Plato introduces them time and again. But neither Stoic
concepts nor Stoic expressibles are an explanans for things' being
what they are. Qualities are; and yet because these are distinct
bodies, none can be common to more than one thing. A universal,
the Platonist might argue, must be both (a) attributable to many
things and (b) the explanans of those things having the attribute in
question: it must, that is, be both what is the same in each case and
what makes each the same sort of thing. Even twentieth-century
philosophers, such as D. M. Armstrong, lean on each of these legs
at different times. But they are not as consistent as the Platonist in
insisting on both clauses.
The genius of the Stoic response, I believe, consists precisely in
separating these two conditions, by holding different entities responsible for each. What is the same in each case is a concept or,
later, an expressible. But what makes each thing what it is is something corporeal, which belongs only to it. Whether we call concepts
or expressibles 'universals', or whether we reserve that term exclusively for things that satisfy both of the Platonist's conditions, is in
a sense immaterial. For what the Stoics have shown is that what is
identical in each case need not be what makes a thing the sort of
thing it is, and this is a genuine conceptual advance.
212
Brown University
21 3

Something Nothing Estoicismo

Uploaded by

Copyright:

Available Formats

Something Nothing Estoicismo

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Something Nothing Estoicismo

Uploaded by

Copyright:

Available Formats

OXFORQ STUDIES

SOMETHING AND NOTHING:

EDITOR: DAVID SEDLEY

iL),' OK,dS' ovap

Victor Caston 1999

I e.g. E. Zeller, Die Philosophie der Griechen in ihrer geschichtlichen Entwicklung

Something and Nothing

derstand why the Stoics would have introduced concepts, only to

. Sedley, The Hellenistic Philosophers, i. 163-6, 181-3; P. Aubenque, 'Une occasion

Something and Nothing

prehensive as well: it includes every item, whatever its status,

Stoic responses to Plato's theory of Forms

Modern treatments generally speak as if all the early Stoics agreed

concerning names [-n}v Xpr,GLV n/, TWV ovo!-'aTwv GIJV,)8ta,), as Chrysippus,

Something and Nothing

Being vs. being something

Something and Nothing

(NE') There are some things which do not exist.

Something and Nothing

centaurs and giants, objects that would be concrete were they to

Thought and object

(TS) 'ix (It is possible to think of x~x is something).

where 'x' is replaced by a singular noun phrase such as 'Chiron'

where 'x' is replaced by the same phrase-a result most of us ob-

Something and Nothing

viously cannot accept, since we think there are no such things as

(TS') It is possible to think of x~x is something

they would be in a position to accept the following intentional

Something and Nothing

cannot assume that both are in play in a given context. This a

gument have failed, however;19 and so it is still an open question

3. I. The place of concepts

It is surely right to keep concepts within Stoic ontology. The first

The first definition commits the Stoics to a plurality of concepts:

Something and Nothing

distinction between genus and species, of course, is a relative one:

This response might be construed in either of two ways. (I) We

may represent Sedley's considered view. 35 If so, his interpretation

Despite its avowed intentions, on the vulgate reading the Stoics'

and therefore (on Sedley s mterpretatlOn) what is not a particular.

Something and Nothing

What does not Exist

price seems prohibitive. It is true that Alexander of Aphrodisias

What does not Exist

(b) If there isn't a superordinate genus, on the other hand, then

Something and Nothing

is not clear whether 'non-subsistent' is to be understood here in

is possible to think of x).

4. The ontological status of concepts

Something and Nothing

other occurrences in Stoic testimonia. The first is in Simplicius, in

just to speak of nothing at all. 44 As the Cyclops Polyphemus learnt,

(S) '<;Ix (x is something).

Something and Nothing

A concept is an apparition of thinking, which is neither something existent

essentia, substantia), and

"Nothing" indicates the limits of presentation, where it ceases to be presentation ...

Ie Sto'icisme', Memoires de I'Institut Imperial de France, 21 (1857),37-8 n. 4) and

The remaining two Stoic categories, in contrast, are not attested