Professional Documents
Culture Documents
Phenomenological Research
Philosophy and Phenomenological Research
Vol. XC No. 3, May 2015
doi: 10.1111/phpr.12184
© 2015 Philosophy and Phenomenological Research, LLC
Although the comments that follow are largely critical, I want to express at
the outset my great admiration for Marko Malink’s excellent book. It makes
a major contribution to our understanding of Aristotle’s logic and the
broader philosophical commitments that underlie it. It is also a model of ex-
egetical clarity and philosophical rigor the rest of us working in this field
should strive to emulate.
I want to explore some of the implications of Malink’s interpretation of the
modal syllogistic in the Prior Analytics for Aristotle’s theory of demonstration
in the Posterior Analytics. I’ll do so by focusing on Malink’s account of the
connection between essence and necessity. I’ll argue that if Malink’s account of
the modal syllogistic is correct, and in particular if his account of the connection
between essence and necessity is correct, then a broad range of demonstrations
Aristotle presents in the Posterior Analytics cannot be modal syllogisms. Ma-
link may happily accept this as a consequence of his view. Still, it seems worth
showing that this is a consequence of his view. I’ll also argue that Malink’s
account of the connection between essence and necessity risks committing
Aristotle to an unattractive form of essentialism.
1
Aristotle’s Modal Syllogistic (Cambridge, MA and London: Harvard University Press,
2013), 125–6. All page references are to this book.
2
See Posterior Analytics 1.7, 75b1, 1.10, 76b6–7, b13–15, 1.22, 83b19–20, 2.13, 96b20,
b23–4. See also Alexander in Top. 50.6–51.5, cited and discussed by Malink (125).
2. Barbara NXN
S11 plays an important role in a central part of Malink’s book: his account
of the validity of Barbara NXN (114–67).
A necessarily belongs to all B (AaNB)
B belongs to all C (BaXC)
A necessarily belongs to all C (AaNC)
3
A is aN-predicated of B iff A necessarily belongs to all B.
4
To avoid confusion in what follows, I have used “D” in place of Malink’s “C.”
Michael Pakaluk has objected that Malink “render[s] Barbara NXN valid by
reducing it to Barbara NNN” and that on Malink’s interpretation the second pre-
mise of Barbara NXN is or amounts to a necessary premise (what Pakaluk calls
“a kind of shadow N-premise”), which Pakaluk claims is false.5 However, Pak-
aluk’s objections are mistaken. As Jacob Rosen points out, “‘B necessarily
applies to all C’ is not a further assumption or ‘shadow premise’ in Malink’s
explanation.”6 Rather, it follows from “A necessarily belongs to all B” and “B
belongs to all C,” together with the relevant claims about essence and necessity
(S9–S11) to which Malink argues (with considerable textual evidence and
philosophical sophistication) Aristotle is committed. In the same way, Malink’s
interpretation does not “reduce” Barbara NXN to Barbara NNN. As I explained
above, his view is that the premise pair of Barabara NXN, together with the rel-
evant claims about essence and necessity (S9–S11), entails the premise pair of
Barbara NNN. To argue that p and q entail r is not to reduce p (or q) to r. And to
argue that p and q entail r is not to introduce r as a “shadow premise.”
5
Michael Pakaluk, review of Aristotle’s Modal Syllogistic by Marko Malink, Bryn Mawr
Classical Review (2014.03.04).
6
Jacob Rosen, review of Aristotle’s Modal Syllogistic by Marko Malink, Notre Dame
Philosophical Reviews (2014.08.33).
unlike the modal syllogistic, the Posterior Analytics is not primarily con-
cerned with modally qualified propositions such as ‘A necessarily belongs
to all B’. Rather, it is concerned with assertoric propositions such as ‘A
belongs to all B’ which are true by necessity. (164–5, n24)
On the other hand, in what follows I’ll argue that if Malink’s S11 is true, then a
broad range of demonstrations Aristotle presents in the Posterior Analytics can-
not be apodeictic syllogisms. So it turns out that on Malink’s view (i) the theory
of demonstration is not part of the modal syllogistic, and yet (ii) some demon-
strations must be apodeictic syllogisms, while (iii) others cannot be. These
claims are logically consistent with each other. However, there is something a
bit surprising about this picture. We might have thought that if some demonstra-
tions are apodeictic syllogisms, then demonstrations in general are and the the-
ory of demonstration is part of the modal syllogistic.
In what follows I’ll focus on (iii). I’ll suggest a different picture according to
which a broader range of demonstrations than Malink allows to be apodeictic syl-
logisms are such. I’ll conclude by discussing in more detail the account of the con-
nection between essence and necessity that emerges from Malink’s account.
7
See especially Posterior Analytics 1.6.
8
Aristotle on Knowledge and Learning: the Posterior Analytics (Oxford University Press,
forthcoming).
9
See Posterior Analytics 1.4, 73b31–2, 1.5, 74a30–4, 2.13, 96b15–25. See also the
account of necessary accidents above.
10
See Posterior Analytics 2.2 and 2.17 (99a21–2), where Aristotle says that “the middle
term is an account (logos) of the major term, which is why all the sciences come about
through definition (di’ horismou).” Later in the passage (at 99a27–9) it becomes clear
that by “logos” Aristotle means “definition.” He says that the middle term in the demon-
stration of “leaf-shedding” (the major term in the example demonstration he is consider-
ing in this part of the text) is what leaf-shedding is (ti esti), i.e., its essence.
11
Other Model 2 demonstrations in the Posterior Analytics include: “eclipse belongs to the
moon” through the middle term “screening of the sun by the earth,” “thunder belongs to
the clouds” through the middle term “fire extinguishing,” and “leaf-shedding belongs to
broad-leafed plants” through the middle term “coagulation of sap.”
12
2R = interior angles equal to two right angles.
13
In the Posterior Analytics, “2R” is both an example of an attribute that is demonstrated
to belong to its subject (triangle) through the subject’s essence (Model 1) (1.4, 73b31–2,
1.5, 74a30–4) (see below) and an example of an attribute that is demonstrated to belong
to its subject through the attribute’s (2R’s) essence (Model 2) (2.2, 90a33). I argue else-
where (see above n8) that Aristotle’s account is coherent. It’s not that there are two
independent causes of 2R, its essence and triangle’s essence. It’s rather that there is a
chain of explanation in which 2R’s essence and triangle’s essence both feature as causes.
The main difficulty with Aristotle’s account is that he does not indicate what the essence
of 2R is, nor does he explain how 2R follows from triangle’s essence.
Take the major premise. According to S11, since “essence of triangle” is the
subject of aN-predication, it is the subject of essential predication. Malink
argues that some terms are essentially self-predicating (139–40). If this is true
of “essence of triangle,” then the major premise is consistent with S11. Now
take the minor premise. S11 says that “triangle” is the subject of essential predi-
cation, which, the minor premise tells us, is true. So S11 is consistent with
14
See the texts cited in n10 above.
5. Interim Conclusion
The question I am urging we ask is, are any Model 2 demonstrations apo-
deictic syllogisms? I’ve argued that if Malink’s account is right, and in par-
ticular if S11 is true, then the answer is “no.”
Perhaps this should incline us to S11*, which permits (without of course
requiring) Model 2 demonstrations to be apodeictic syllogisms. The problem
now is that if we substitute S11* for S11 in the argument for the validity of Bar-
bara NXN, the argument is spoiled. Suppose we follow the second of S11*’s
two disjuncts, according to which B, which is the subject of aN-predication, is
aN-predicated of a term that is the subject of essential predication. Now S9 is no
longer relevant. For S9 relies on the thought that B is the subject of essential
predication, which the second of S11*’s two disjuncts does not assert. And if
S9 is not relevant, then we cannot prove BaNC from AaNB and BaXC.
6. Aristotle’s Essentialism
I want to conclude by suggesting that Malink’s interpretation risks committing
Aristotle to an unattractive form of essentialism. In particular, it risks commit-
ting Aristotle to an unattractive account of necessary accidents, especially con-
dition (iii), according to which if P is a necessary accident of S, then P
necessarily belongs to all S because of S’s essence. My worry is that on Ma-
link’s account all the necessary accidents of a given subject are explained
(demonstrated) from the subject’s essence alone. For example, Malink quotes
approvingly David Charles’ claim that “‘the essence is the one cause of all the
kind’s derived necessary properties.’ (Charles 2000: 203)” (126; my empha-
sis) That is, Malink seems to commit Aristotle to:
Strong Essentialism: The essence of a subject is the sole (one and
only) cause of all of that subject’s necessary
accidents. (Cf. 126)
There are two possibilities: either the middle term is D or it’s some other
term, E. Suppose it’s D:
AaND
DaNB
AaNB
The ellipsis “. . .” indicates that the argument needs to be filled in with addi-
tional premises. These additional premises must eventually yield AaND, for
this is required as the major premise in any demonstration whose minor
premise is DaNB and whose conclusion is AaNB. The most direct way to
From S11 and AaNE it follows that E is the subject of essential predication.
From S9, AaNE, and EaND it follows that E is predicated essentially of D. From
EaND and DaNB, EaNB follows. Since E is the subject of essential predication
(S11) and is predicated essentially of whatever it is predicated (S9), E is predi-
cated essentially of B. Now D is also predicated essentially of B. So two terms
are predicated essentially of B: D and E. In addition, A necessarily belongs to
all B (AaNB) because of D and E alone. This is Strong Essentialism, which is
entailed by S11 together with S9 and S10: if A necessarily belongs to all B and
AaNB is demonstrable, then A necessarily belongs to all B because of B’s
essence alone (which in this case comprises two terms: D and E).
The same result can be obtained even if we don’t assume EaND straight
off but derive AaND from a longer series of aN-predications. The only dif-
ference is that more terms will be packed into B’s essence. And herein lies
the problem. The essence of a subject is the sole cause of all of that sub-
ject’s necessary accidents only if we pack a sufficient number of terms into
the subject’s essence to account for all the necessary accidents. This threat-
ens the unity of the essence. It also makes Strong Essentialism seem ad
hoc: it’s guaranteed to be true because the essence of B is just that set of
terms needed to explain all of B’s necessary accidents. But this gets things
the wrong way around: it’s not that D, E, etc. are B’s essence because they
explain all of B’s necessary accidents; it’s rather that D, E, etc. explain all
of B’s necessary accidents because they are B’s essence.
Now suppose that S11* is true. In that case, there is no guarantee that all of
B’s necessary accidents will follow from B’s essence alone. Rather, S11*
(unlike S11) is consistent with the view that B’s essence is a cause of all of its
necessary accidents, but not the only cause. For the essences of the necessary
accidents themselves are also needed to explain why they belong to B. I sug-
gest that it is this weaker form of essentialism that Aristotle is committed to.