KOOKY OBJECTS REVISITED:
ARISTOTLE’S ONTOLOGY
S. MARC COHEN
Abstract: This is an investigation of Aristotle’s conception of accidental com-
pounds (or ‘‘kooky objects,’’ as Gareth Matthews has called them)Fentities such
as the pale man and the musical man. I begin with Matthews’s pioneering work
into kooky objects, and argue that they are not so far removed from our ordinary
thinking as is commonly supposed. I go on to assess their utility in solving some
familiar puzzles involving substitutivity in epistemic contexts, and compare the
kooky object approach to more modern approaches involving the notion of
referential opacity. I conclude by proposing that Aristotle provides an implicit
role for kooky objects in such metaphysical contexts as the Categories and
Metaphysics.
Keywords: accident, Aristotle, ontology, sameness, substitution.
In the cou rse of my forty-year friendship and colla boration with Gary
Matthews, he has taught me many things, more than I have space to
enumerate here. So I won’t try to enumerate them. Instead, I am going to
focus on just one of those thingsFthe idea that it is worth taking
seriously a certain aspect of Aristotle’s ontology that seems at least
quaint, if not weird, to our contemporary ears. Its quaintness, or
weirdness, consists in the fact that it is a finer-grained ontology than we
are accustomed toFit distinguishes between entities that we tend not to
distinguish, and hence gives ontological status to entities that philoso-
phers nowadays are not entirely comfortable with.
Gary is not alone , of course, in his interest in this aspect of Aristot le’s
thought. Other scholars have given it different labels: Frank Lewis talks
about Aristotle’s ‘‘accidental compound theory’’ (Lewis 1982, 29 n. 11),
and Nicholas White writes of Aristotle’s doctrine of ‘‘modal individua-
tion’’ (White 1986, 480). But the catchiest label for this view is Gary’sF
he calls it Aristotle’s theory of kooky objects (Matthews 1982 and
1992)Fand the label seems to have stuck. So that is how I’ll refer to it
here. I am going to talk about kooky objects: what they are, where to
find them in Aristotle’s texts, and what roles they play in his metaphysics
and epistemology.
r 2008 The Author
Journal compilation r 2008 Metaphilosophy LLC and Blackwell Publishing Ltd
Published by Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK, and
350 Main Street, Malden, MA 02148, USA
METAPHILOSOPHY
Vol. 39, No. 1, January 2008
0026-1068
r 2008 The Author
Journal compilation r 2008 Metaphilosophy LLC and Blackwell Publishing Ltd
I
Kooky objects make their most obvious appearance in Aristotle’s
discussion of change in Phys. I.7, where he is discussing the case in which
an unmusical man becomes musical. In such a case, he says, the man
survives but the unmusical does not. It might seem that it is only the
attribute of unmusicality, or perhaps the man’s own peculiar bit of
unmusicality, that Aristotle is supp osing to perish when the man becomes
musical, but he quickly dispels that impression: ‘‘The man survives, but
the unmusical does not surviv e, nor does the compound of the two, namely,
the unmusical man’’ (190
a
19–21).
1
This compound, the unmusical man, is
an example of a kooky object; it goes out of existence when the man
becomes musical, and it comes into existence when the man loses his
musicality.
On this view, the man and the unmusical man are not identi cal, since
one survi ves the onset of musicality, while the other does not. Thus, the
man, the unmusical man, and the musical man are distinct things . What,
then, one wonders, is the relation between the man and the musical man,
if it is not identity? Aristotle’s answer is that they are accidentally the
same
2
or accidentally one.
3
That is, they are distinct entities that may, for
a time at least, coincide; in the sense they may be, for a time at least, one in
number.
Here we have reached our first stumbling block. For it seems decidedly
odd to say that two things can be tempor arily one. Aristotle’s response,
I think, would be to say that our ability to distinguish the entities in
question does not make them two. For the manFlet’s call him Coriscus
(as Aristotle does at 1015
b
17)Fand the musical man are not two
menFindeed, they are not two of anything (Matthews 1982, 226). That
they are distinct entities (onta) does not, in Aristotle’s view, make them
two. For ‘‘entity’’ has a variety of senses (as Aristotle puts it, ‘‘entity is
said in many ways’’),
4
and there is no one sense of ‘‘entity’’ in which both
Coriscus and the musical man are both said to be entities. This leads us to
a preliminary answer to the question of what kind of things these are that
are accidentally the same. Entity, in its primary sense, applies to things in
the category of substan ce, independently existing things such as a man or
1
All translations from Aristotle’s works are my own.
2
Cf. Metaph. D.9, 1017
b
31: ‘‘Both the man and the musical are said to be the same [in the
accidental sense] as the musical man.’’ Cf. also Top. 103
a
30.
3
Cf. Metaph. D.6, 1015
b
17: ‘‘Examples of things that are accidentally one are Coriscus
and the musical and musical Coriscus.’’ Clearly, this is not a list of three accidental unities,
since Coriscus is one per se and not per accidens. Rather, Coriscus and the musical are
accidentally one; Coriscus and musical Coriscus are accidentally one.
4
Cf. Metaph. G.2 (esp. 1003
a
33–
b
5), E.2 (esp. 1026
a
34–
b
26), Z.1 (esp. 1028
a
10–30), K.3
(1060
b
32–1061
a
10).
r 2008 The Author
Journal compilation r 2008 Metaphilosophy LLC and Blackwell Publishing Ltd
S. MARC COHEN
4
a horse. Coriscus is an item in the category of substance, but the musical
man with whom he coincides, at least temporarily, is not.
5
So Coriscus and the musical man with whom he coincides are distinct
but are one in number. This way of doing busines s requires us to be very
careful when we are counting, lest we count things that are one in number
twice. This may seem a daunting prospect, but we are actually able to
handle it pretty well. For example, if we count up the number of U.S.
presidents, we get forty-three (George W. Bush is the forty-third pres-
ident). But if we count up the number of persons who have been
president, we get forty-two. This oddity, as everyone knows, is due to
Grover Cleveland, who was both the twenty-second and the twenty-
fourth president. There have been forty-three presidents, but Cleveland
was two of them. To put it another way, we’ve had forty-three presidents,
but only forty-two people who have been president. From Aristotle’s
perspective, it is easy to account for this. When we are counting people
who have been president, we are counting persons (substances), but when
we are counting presidents, we are counting kooky objects. (This is
intended to be an ontological, not a political, comment.) And it turns out
that two of those presidents (the twenty-second and the tw enty-fourth)
coincide with just one person. Of course, they do not coincide with one
another. So the relation of coinciding with is not transitive.
The system we actually use to number the presidents makes sense
provided that we are comfortable with kooky objects. An ontology that
lacks kooky objects, on the other hand, would have to take a different
tack here. One such tack would be to claim that we have had, in fact, only
forty-two presidents (namely, the forty-two men who have held the
office), but then we would get into trouble if we try to retain our current
numbering system. For we would have to say that George W. Bush is the
forty-third of our forty-two presidents. This does not seem like a happy
solution.
So we would be better off to change our numbering system. We could
insist that Grover Cleveland was the twenty-second president but not the
twenty-fourthFinstead, when Cleveland was elected to succeed Benjamin
Harrison in 1892, the twenty-second president succeeded the twenty-
third. And then we would have to say that when Cleveland’s second term
ended in 1896, the twenty-second president was succeeded by the twenty-
fourth. On this proposal, the twenty-fou rth president turns out not to be
Cleveland at all but rather the man who succeeded himF William
McKinleyFthe man who counts, on the numbering system we actual ly
use, as the twenty-fifth president. On this solution, the numbers add up
(Bush is the forty-second of our forty-two presidents), but the ordering
of presidents becomes strange.
5
Precisely how the musical man fits into Aristotle’s categorial scheme is a complicated
question, which we will postpone for now.
r 2008 The Author
Journal compilation r 2008 Metaphilosophy LLC and Blackwell Publishing Ltd
KOOKY OBJECTS REVISITED
5
All this to save the doctrine that the description ‘‘the twenty-second
president’’ denotes a substance! By contrast, the Aristotelian treatment of
that expression as denoting a kooky object should strike us as eminently
sensible. And it is not only sensible, it is the ont ology that makes the most
sense of our actual practice of counting presidents.
6
There is a residual problem , of course, concerning identity statements.
Because Grover Cleveland was both the twenty-second and the twenty-
fourth president, we might be inclined to express this by means of these
two identity claims.
Grover Cleveland 5 the twenty-second president
Grover Cleveland 5 the twenty-fourth president
But then we would be forced by the transitivity of identity to accept:
The twenty-second president 5 the twenty-fourth president
But this cannot be right, since the former’s term of office preceded that of
Benjamin Harrison, the twenty-thi rd presiden t, whereas the latter’s did
not. The twenty-second and twenty-fourth presidents are distinct (kooky)
objects. Indeed, they are two different presidents. Grover Cleveland and
the twenty-second president, on the other hand, are distinct objects, but
they are not two of anything. The fir st is a substance that coincides with
the latter.
Let us end our little foray into American political history and return to
Aristotle’s example of Coriscus and the musical man. We have determined
that they are distinct entities, although they are not two of anything. Since
they are not two of anything, one might be tempted (as some of Aristotle’s
readers have been tempted) to propose that they are, after all, identical, but
only contingently so. On this view, the relation of accidental sameness is
construed as contingent identity.
7
Of course, to those who think that
identity holds as a matter of necessity where it holds at all, this is a proposal
that Aristotle ought to reject. And, in fact, I think he does reject it.
Aristotle would surely admit that it is a contingent fact that Coriscus
and the musical man are one and the same. But this is not to say that the
contingent fact is that they are identical. Rather, what is contingent is that
two nonidentical entities coincide. Althou gh many of the things Aristotle
classifies as accidentally the same might be considered contingently
identical by proponents of that notion, Aristotle’s accidental sameness
relation is not contingent identity.
8
For according to him, things that are
accidentally the same are in a way the same, and in a way different. But if
6
Rea 1998 makes a good case for treating many other quite familiar objects (e.g., fists
and statues) as kooky.
7
Barnes 1979 makes this suggestion.
8
A convincing case for this has been made in Matthews 1982, 228–29.
r 2008 The Author
Journal compilation r 2008 Metaphilosophy LLC and Blackwell Publishing Ltd
S. MARC COHEN
6
x is contingently identical to y, x does not stand to y in some relation
weaker than identity. Rather, it is supposed to be a contingent fact that x
stands to y in the full-blooded relation of identity. So Aristotle’s notion of
accidental sameness is not conting ent identity.
9
Let us return to the question of what the relation of accidental
sameness amounts to. An approach that might sound attractive to
contemporary ears is this. Accidental sameness is not some relation
weaker than identity. Rather, what Aristotle has discovered is the
phenomenon of referential opacity. Coriscus and the musical man are
identical, in the strictest sense. However, their names are not interchange-
able in all contexts. Sometimes how you refer to something, and not just
what you are referring to, affects the truth or falsity of what you say about
it. A famous passage in Sophistical Refutations (179
a
33–
b
32) seems to
support this understanding of Aristotle. For there Aristotle describes a
case in which you do not know the masked man, but you do know
Coriscus, even though the masked man is Coriscus. In this case, we might
suppose, whether it is true to say that you know this man depends upon
how we describe him. One and the same object is referred to by both the
expression ‘‘Coriscus’’ and the expression ‘‘the masked man,’’ but it can
still be true that you know who Coriscus is but do not know who the
masked man is. The reason is that the expression ‘‘knows who x is’’
generates a referentially opaque context.
But this can hardly have been Aristotle’s solution to the puzzle of the
Masked Man. Aristotle presents the puzzle as an exampl e of the fallacy of
accident. According to his official account in the Topics, one commits this
fallacy when one claims that somet hing ‘‘belongs in the same way both to
a thing and also to its accident’’ (homoio
ˆ
s...to
ˆ
i pragmati kai to
ˆ
i
sumbebe
ˆ
koti huparkhein, 166
b
29ff.). If one thinks of an accident of x as
an attribute that belongs to x accidentally, one may well suppose that
Aristotle has in mind the bizarre fallacy of attributing to a thing an
attribute of one of its attributes. An example of such a fallacy would be to
infer that Socrates is himself a quality from the fact that pallor, one of his
attributes, is a quality. Is it possible that this is the kind of fallacy
Aristotle has in mind? One might expect Aristotle’s examples to clarify
the issue, but unfortunately they are not at all helpful.
His two examples are these: (1) Coriscus is different from a man, and
Coriscus is a man, therefo re, Coriscus is different from himself; (2)
Coriscus is different from Socrates, and Socrates is a man, therefore,
Coriscus is different from a man. About (2) Aristotle comments tersely
that ‘‘it is an accident of that from which he has be en said to be different
9
A notion of contingent identity like that developed by Stephen Yablo (1987) comes a
lot closer to Aristotle’s idea. But on Yablo’s account, contingently identical things are not,
strictly speaking, identical. So contingent identity for Yablo, like accidental sameness for
Aristotle, is not a species of identity.
r 2008 The Author
Journal compilation r 2008 Metaphilosophy LLC and Blackwell Publishing Ltd
KOOKY OBJECTS REVISITED
7