Z.

6: The Identity Thesis

The problem

At 1032a5, Aristotle claims that “in the case of primary things, those spoken of in their
own right, a thing and its essence are one and the same.” What does this mean?
Primary things spoken of in their own right would appear to be substances (as
opposed to accidental compounds like pale man). Abbreviating “the essence of x” as
E(x), the claim seems to be this:

If x is a substance, then x = E(x)

We’ll call this the Z.6 identity thesis. Since Socrates and Callias are substances, this
would seem to yield these two instances of the Z.6 identity thesis:

1. Socrates = E(Socrates)
2. Callias = E(Callias)

But Z.4 seemed to tell us that it was species that have essences. A natural way to
understand this is as follows: if x has an essence, then either x is a species, or (if x is
not a species) the essence of x is the essence of the species that x belongs to.

Since Socrates is not a species, E(Socrates) must be the essence of Socrates’ species,
E(human). Similarly, it follows that E(Callias) is the essence of Callias’ species,
E(human). That is:

3. E(Socrates) = E(human)
4. E(Callias) = E(human)

But from (3) and (4) it follows (substituting equals for equals) that:

5. E(Socrates) = E(Callias)

And now the identity thesis gives us an intolerable result. For if we substitute into (5)
the equals given by (1) and (2), we get:

6. Socrates = Callias

But Socrates and Callias are supposed to be two different substances. So how can they
have the same essence (as Z.4 seems to say) and yet each substance be identical to its
essence (as Z.6 seems to say)?

Copyright © 2008, S. Marc Cohen 1 Revised 2/26/08

Some alternative solutions

1. “… is one and the same as…” does not mean identity.

2. The substances under discussion are not individuals like Socrates and Callias.

3. The essences of individuals are not shared, i.e., E(Socrates) ≠ E(Callias)

A type 2 solution: the Z.6 Identity thesis pertains to proper definables

The best solution I know of is that of Code (“Aristotle: Essence and Accident”). These
are the general lines.

Essence and Definition

An essence is the ontological correlate of the definiens in a definition. Hence, the

essence must be the essence of something that is definable. This tilts us toward
something at the species level (man, horse), rather than the individual (Callias,
Nashua).

The Z.6 thesis should be understood as the object language counterpart of the
straightforward semantic thesis that in a fully analyzed definition, the definiens and
the definiendum denote one and the same entity. (The definiendum denotes the
thing and the definiens denotes the essence of that thing.) Hence, for the definiens
and definiendum to denote the same thing is for the thing to be identical to its
essence.

Instances of the identity thesis

Man = E(man)

Horse = E(horse)

Support for this interpretation

The way to test this interpretation is to see whether it fits together with its
surrounding text, and whether it can explain some passages that would otherwise be
difficult to understand. This interpretation does very well in that regard:

The regress argument in Z.6

At 1031b28, Aristotle gives an infinite regress argument in favor of the identity

thesis. If a thing is not identical to its essence, we are led on an infinite regress:

Copyright © 2008, S. Marc Cohen 2 Revised 2/26/08

 We can also see that it is absurd <for something not to be the same as its essence>, if we
give a name to each essence; for apart from that essence there will be another essence as
well—for instance, another essence will be the essence of the essence of horse.

A few lines later (1032a3), he makes clear that this is an infinite regress:
Moreover, if there is another essence, the essences will go on to infinity.

His way of breaking the back of the regress (back to 1031b31) is to assert the
identity thesis:
But why not let some things be essences at once, going no further, since essence is
substance?

The account of definitions and their parts in Z.10

The claim that a (definable) thing and its essence are identical has consequences
for definitions. Aristotle has just licensed us to talk about the essence of an
essence (although he claims, of course, that they are identical). This means that
not only is a substance definable, but its essence is also definable. That means
that we should be able to treat the definiens of a definition as a definiendum,
and ask how to define it?

So Aristotle needs an account of definitions that will guarantee that when we go

to define the definiens of a definition, the definiens of the definiens will be the
same as the original definiens.

This is exactly what we find in Z.10. And since Z.7-9 are generally agreed to be
a later interpolation (in Z.11 1037a22, Aristotle gives a summary of Z.4-10 that
makes no mention of anything in Z.7-9) it seems likely that Aristotle turned
immediately from stating the Z.6 identity thesis to providing for its defense.

Z.10: definitions and their parts

Z.10 begins with a consideration of definitions and their parts. Its main question is
this (1034b22): “… whether or not the account of the parts must be present in the
account of the whole.” The remainder of the chapter is devoted to a careful
distinction between those parts of a thing that do get mentioned in its definition and
those parts that do not.

The key idea is that parts of the form get mentioned, but material parts (except in
special circumstances) do not.

The idea seems to be this. Suppose we have a definition, of the form:

x =df ABC
Copyright © 2008, S. Marc Cohen 3 Revised 2/26/08
 Further suppose that each of A, B, and C is definable, and that this is a proper
definition (i.e., A, B, and C are parts of the form of x). That means the definitions of
each of A, B, and C are (implicitly) in the original definition.

So to make the original definition more explicit, we should replace each term, A, in
the original definition with its definition, D(A):

x =df D(A) D(B) D(C)

It may be that these terms, D(A) etc., are further definable, say by DD(A) etc. In that
case, we should make those replacements, as well.

Example: man is defined as rational animal. But animal is also definable, e.g., as
sensitive living thing. So we can rewrite the original definition: man is defined as
rational sensitive living thing. If rational, sensitive, and living are also definable (as
they presumably are) we should plug in their definitions, as well.

Can this go one forever? Aristotle says no (APst. I.22)—definitions cannot be

infinitely long. Hence we will ultimately arrive at a fully analyzed definition, in
which every definable term has been replaced by its definition, until nothing further
definable remains.

Now, let y be such an ultimate part (not further definable), and ask this question:
what is the essence of y? E(y) is whatever is denoted by D(y). But y is not further
definable, so D(y) is just y all over again. Hence E(y) is just y itself: y = E(y)

Z.10 therefore provides exactly the account of definitions that the Z.6 identity thesis
requires.

The regress argument in Z.6 reconsidered

We can now apply what we’ve learned from Z.10 to the regress argument in Z.6.
Aristotle argues that unless we let at least one thing be identical to its essence, we
will get an infinite regress.

His example concerns horse and E(horse). Suppose they are not the same, that is,
suppose that horse ≠ E(horse). Now he wants to ask about the essence of E(horse).
(He seems to think we will find this question more tolerable or intelligible if we first
give E(horse) a name. So let’s call E(horse) “Charlie” and ask: what is the essence
of Charlie? Presumably, whatever drove us to deny the identity of horse and
E(horse)—i.e., to deny that horse = Charlie—will also drive us to deny the identity
of Charlie and E(Charlie). (Presumably, we should also give E(Charlie) its own
name—say, “Jim.”) That leaves us with this infinite sequence of distinct objects:

Copyright © 2008, S. Marc Cohen 4 Revised 2/26/08

 horse, Charlie, Jim, E(Jim), …

which we can rewrite as:

horse, E(horse), EE(horse), EEE(horse), EEEE(horse), …

Aristotle insists that this sequence cannot be infinitely long. But how can we be sure
that it doesn’t go on for quite a ways before stopping? If it did, we would be stuck
with horse ≠ E(horse) and no infinite regress.

But what we learned from Z.10 shows us that this cannot happen. That is, if the
regress breaks down anywhere, it breaks down everywhere—i.e., it can’t get
started. Here’s why:

According to Aristotle, definable things are individuated by their definitions: if x

and y have the same definition, they are the same thing, x = y. In Z.13 1038b14, he
states the ontological counterpart of this claim: things with the same essence are
themselves one <and the same>.

E(x) = E(y) → x = y

So if the chain above comes to an end, it will be because at some point we have a
true identity statement, such as:

EEE(horse) = EEEE(horse)

But if these two essences are the same, then so are the things of which they are the
essences:

EE(horse) = EEE(horse)

Repeating the same move yields:

E(horse) = EE(horse)

and ultimately:

horse = E(horse)

Hence Aristotle’s conclusion: “… why not let some things be essences at once,
going no further …?”

Copyright © 2008, S. Marc Cohen 5 Revised 2/26/08

 Why does Aristotle hold this thesis?

Aristotle’s reasons for maintaining this identity thesis seem to be mainly

epistemological and anti-Platonic. He spends a lot of time in the beginning of Z.6
talking about Platonic Forms and the dangers of separating them from their essences
(as he thinks Plato did). The worry is that if everything is distinct from its essence,
no one will ever know anything, since “knowing a thing is knowing its essence”
(1031b20).

The argument for this is straightforward enough. Suppose everything is distinct

from its essence, but that we do know something, say x. It would seem to follow that
we know infinitely many things if we know anything at all. For to know x is to
know E(x), and to know E(x) one must know EE(x), and to know EE(x) one must
know EEE(x), etc., ad infinitum. But one can’t know infinitely many things. So one
can’t know any thing at all.

Copyright © 2008, S. Marc Cohen 6 Revised 2/26/08