Non-Contradiction as an Ontological Principle:

An Interpretation of Aristotle’s Metaphysics IV.3-4

I. Introduction

In his defense of the principle of non-contradiction (PNC) in Metaphysics IV,

Aristotle notes the danger that a petitio principii presents for his project (Met. 1006a17).

Nevertheless commentators have occasionally suggested that Aristotle does in fact beg the

question. Whereas Lukasiewicz claims that modus tollens presupposes PNC1 and Wedin has

noted that Aristotle uses modus tollens in his defense of the principle,2 Priest contends that

Aristotle’s proof of PNC only forces the opponent to abandon the denial of PNC if he is

already committed to the principle.3 These worries, however, emerge from a misguided

tendency prevalent among more recent commentators to assume that Aristotle is defending

the modern logical version of PNC, -(p and –p).4 In this paper, I will endeavor to show that

the demonstrandum of Aristotle’s defense is not the modern logical version of PNC (PNC-

logical), but rather a more limited, ontological version of the principle (PNC-ontological)

which states that it is impossible for a property (F) to belong and not to belong to the same

subject (X) at the same time in the same respect. The reason why Aristotle does not beg the

question is that even though he uses argumentation strategies that might presuppose PNC-

logical, he is not trying to prove PNC-logical. Instead, he is trying to prove PNC-

1 Lukasiewicz (1993, 86).

2 Wedin (2000, 155). Lukasiewicz, however, does not recognize this point.

3 Priest (1998, 105). See Whitaker (1996, 199) for an attempt to deal with this problem. Of course, this is not

an instance of a petitio principii if Aristotle’s task is merely to prove PNC. However, it is a petitio principii if it is

his task to prove the falsity of not-PNC.

4 Gottlieb (1992), Lukasiewicz (1993), Priest (1998), and Wedin (2000) all see Aristotle as defending the modern

logical version. Cresswell (2003) claims that Aristotle is not.

1
ontological, and because his argument does not presuppose PNC-ontological, he does not

beg the question.

In the first part of the paper, I examine Lukasiewicz’s claim, still current today, that

Aristotle is not only defending PNC-ontological, but also a logical and a psychological version of

the principle. Here I will reject this claim, contending that the psychological version is really

a claim about PNC-ontological and that while the logical version Lukasiewicz claims to find is

about propositions and their truth-values, it is nevertheless not the modern logical principle,

-(p and –p). Indeed, we will see that, for Aristotle, the fact that two contradictory statements

cannot be true at the same time, what I will call PNC-logical(A), actually depends on PNC-

ontological, and this is why Aristotle needs to defend the latter in order to secure the former.

In the second part of the paper, I will analyze Aristotle’s elenctic proof of the principle and

identify three potential areas in which his argument might beg the question if we assume he

is trying to prove PNC-logical. In my concluding remarks, I will contend that we should

abandon this latter assumption not only because there is no textual evidence for it, but also

because once we do, worries about a petitio principii dissolve.

II. The Various Formulations of PNC in Metaphysics IV

The ontological thrust of Aristotle’s defense of PNC has been noted by a number of

commentators. Indeed, Aristotle first formulates the principle in terms of objects and their

properties: “It is impossible for the same attribute at once to belong and not to belong to

the same thing and in the same relation” (1005b19-20).5 This, however, should not be a

surprise as Aristotle’s defense is found in the opening stages of a work which proposes to

study being qua being (1003a32). The fact that Aristotle’s book is about being suggests that his

defense of PNC should have something to do with being as well. Nevertheless, Aristotle

5 Trans. Tredennick.

2
quickly moves on to claim that someone cannot believe that something both is and is-not (F)6

(1005b23-24). This has led some commentators to conclude that Aristotle wants to defend

an additional, psychological formulation of the principle. Indeed, Lukasiewicz has argued that

Aristotle wants to defend three versions of the principle, the ontological, psychological, and

logical.7 Whereas evidence for the first two versions of PNC can be found in the opening

stages of Aristotle’s argument, the best evidence for the logical version appears in Aristotle’s

summary of what his argument has established. It reads: “the most certain of all beliefs is

that opposite assertions are not both true at the same time” (1011b14).8 While Lukasiewicz

notes that the logical and ontological formulations are logically equivalent, they nevertheless

have different meanings.9 Whereas the ontological version is about objects and their

properties, the logical version is about assertions and their truth-values.

Before I show that Aristotle’s logical version of PNC is not, as Lukasiewicz believes,

the modern logical version, -(p and –p), and that PNC-logical(A) ultimately depends on

PNC-ontological, I first want to argue that Lukasiewicz’s psychological version is not a

version of PNC at all, but rather a claim about PNC-ontological.10 It is a claim about PNC-

ontological because Aristotle is arguing that PNC-ontological is the most certain of all

6 I have adopted Whitaker’s (1996) convention of distinguishing between “is-not F” and “is not-F”. Whereas

the former is a negation which states that a property F does not belong to a subject, the latter affirms that the

property “not-F” belongs to a subject.

7 Gottlieb (1992, 185f.) also identifies three versions, but refers to them differently: ontological, doxastic

(psychological), and semantic (logical). Brinkmann (1992, 205ff.) also distinguishes three different principles,

but claims that they are all expressions of what he calls the principle of determinacy

8 Trans. Tredennick. It should be noted, however, that Aristotle speaks of contradictory statements and their

truth values at 1007b18.

9 Lukasiewicz (1979, 51f.).

10 For an extended analysis of this issue that largely agrees with my own, see Wedin (2004).

3
principles (1005b22), and Aristotle uses PNC-ontological to establish this point. We know

that the argument uses PNC-ontological because Aristotle tells us this at the beginning of

Metaphysics IV.4: “We have just assumed that it is impossible at once to be and not to be,

and by this means we have proved that this is the most certain of all principles” (1006a3-5).11

While this much is clear, there is some ambiguity about the argument itself. On the standard

interpretation,12 the argument is supposed to work by applying PNC-ontological to the

knowing subject. Specifically, a knowing subject who believes that something both is and is-

not (F) has contrary beliefs, a point supposedly secured in chapter fourteen of De

Interpretatione. Because beliefs are properties of a thinking thing, a thinking thing that has

contrary beliefs has contrary properties. Having contrary properties, however, violates

PNC-ontological. This is because if an object has one (white) of two contrary properties

(white and black), it cannot have the second (black) of the contrary pair, a fact that follows

from the nature of contrary properties. Thus, if white belongs to an object, black does not

belong to an object and vice versa. Therefore, if white and black both belonged to the same

object, white and black would also not belong to that object. Thus, we would have an object

that both is and is-not white and is and is-not black, a two-fold violation of PNC-

ontological. Therefore, if PNC-ontological is true and believing that X is and is-not F is an

instance of a knowing subject having contrary properties, it is impossible for a knowing

subject to believe that X is and is-not F.

While there are a number of issues that arise from this standard reading, the

important point for our purposes is that Lukasiewicz’s so-called psychological formulation

of the principle is really a claim about PNC-ontological. If this argument is correct, then

11 Trans. Tredennick.

12 Here I follow Whitaker (1996, 185). Lukasiewicz (1979, 52f.) and Wedin (2000, 115f.) provide similar

treatments.

4
there are only two formulations of PNC remaining, the logical and ontological. The next

step in our argument is to deal with the passages that have suggested to Lukasiewicz and

others that Aristotle is interested in defending the modern logical version of the principle, –

(p and –p). This will be important because while Cresswell has argued that Aristotle is not

defending PNC-logical, he does not provide an alternative explanation for the passages

which suggest that Aristotle is.13 In what follows, I will argue that although Aristotle does

argue for PNC-logical(A), which states that two contradictory statements cannot be true at

the same time, he is not arguing for the modern logical notion, -(p and –p). This distinction

is important because Aristotle’s logical version of the principle, unlike the modern logical

version, ultimately depends on PNC-ontological.

To understand why this is so, we first need to recognize the difference between

Aristotle’s understanding of contradictory statements and our own. For many contemporary

logicians, contradictory statements are defined as statements for which it is logically

impossible for both to be true or false at the same time. 14 For Aristotle, however, there is

nothing in the definition of contradictory statements that guarantees that they cannot both

be true or false at the same time. To grasp this point, we need to get clear about Aristotle’s

understanding of negation.15 Specifically, Aristotle does not think in terms of negating

propositions, where the negation is external to the proposition and implies the falsity of the

embedded proposition. Instead, for Aristotle, negation is a second kind of assertion. When

we negate, we simply assert that a property (F) is separate from a given subject (X).

Negation, of course, stands in contrast to affirmation, where we assert that a property (F)

belongs to a given subject (X). A contradictory pair of assertions therefore consists of two

13 Cresswell (2003, 170).

14 For a list of the various definitions of contradiction, see Grim (2004, 51ff.).

15 For more on this, see Englebretsen (1981).

5
assertions which affirm (“X is F”) and negate (“X is-not F”) the same thing (F) of the same

thing (X) (De Int. 17a30-35). Thus, there is nothing in the definitions of negation and

contradiction to ensure that two contradictory assertions cannot both be true at the same

time (PNC-logical(A)). Nevertheless, Aristotle does hold that, in most cases (see discussion

below), two contradictory assertions cannot be true at the same time (PNC-logical(A)). His

reasoning, however, is not based on the definitions of contradiction and negation, but rather

on certain facts about the world.

To understand how PNC-logical(A) depends on certain facts about the world, we

first need to turn to De Interpretatione, where Aristotle is discussing exceptions to what

Whitaker has called the rule of contradictory pairs (RCP). This rule states that for “every

contradictory pair [of assertions], one member is true and the other false.”16 Of course, RCP

is simply a combination of PNC-logical(A) and what is known as Aristotle’s law of excluded

middle (LEM). Whereas PNC-logical(A) states that two contradictory assertions cannot

both be true at the same time, LEM states that two contradictory assertions cannot both be

false at the same time. Taken together, we have RCP.

Aristotle’s discussion in De Interpretatione is also important for our purposes because it

highlights the way in which the rules governing contradictory assertions depend on the

ontological status of their subject matter. Indeed, they admit of exceptions precisely because

of this dependency. Thus, in chapter seven of De Interpretatione, Aristotle identifies a class of

contradictory assertions, namely non-universal assertions about universal subjects such as

“man is pale” and “man is-not pale,” that can be true at the same time (De Int. 17b29),

thereby violating RCP and PNC-logical(A). In the next chapter, he identifies an additional

exception to RCP, arguing that, in the case of contradictory pairs in which a name signifies

16 Whitaker (1996, 78).

6
more than one thing, it is possible for both to be false, thus violating LEM (18a27).17 In

chapter nine, we find Aristotle’s much discussed analysis of future contingent assertions,

where he argues that “it is necessary for one or the other of the contradictories to be true or

false—not, however, this one or that one, but as chance has it; or for one to be true rather

than the other, yet not already true” (19a36f.).18

While there is much debate about whether Aristotle is suspending RCP in the case of

future contingents or the principle of bivalence (PB),19 which states that any given assertion

must be either true or false (17a1-3), the important point for our purposes is the fact that

Aristotle allows for exceptions to the rules governing contradictory assertions because of the

way things are. On one reading, Aristotle’s concern in chapter nine is that if RCP holds for

future contingents,20 then fatalism will be true. Fatalism, however, cuts against the common-

sense view that we can influence the future. To block this conclusion, Aristotle argues that

the rules typically governing contradictory assertions do not apply to future contingents, and

17 This is an interesting instance which reinforces the point about Aristotle’s understanding of negation. It

occurs when a term “cloak” signifies two things, such as horse and man. “Cloak is pale” and “cloak is-not

pale” can both be false when man is pale and horse is-not pale. While modern logicians would say that “cloak

is pale” is false because one of the subjects which “cloak” signifies is not pale, they would contend that “cloak

is not pale” is true because it is not the case that pale can be attributed to both of the subjects of which pale is

predicated. For Aristotle, however, to assert that the “cloak is-not pale” is to assert that the predicate pale does

not belong to horse and does not belong to man. However, since pale does belong to man, it is false to say

that “cloak is-not pale”.

18 Trans. Ackrill.

19 A third possibility is that he is rejecting the inference that true or false statements about the future entail the

necessity of the future. See Whitaker (1996, 129ff.) for a brief summary of the secondary literature.

20 This, in any case, is Whitaker’s (1996) interpretation. Others have argued that Aristotle’s real concern is the

principle of bivalence, the view that every assertion must be true or false.

7
Aristotle can and has to do this because the rules governing contradictory assertions are

ultimately related to “how the actual things are” (19a33).21

In Metaphysics IV, Aristotle’s task is not to identify certain classes of assertions that

constitute exceptions to the basic rules governing contradictory assertions, but rather to

argue that the world is structured such that, for present singular assertions, two contradictory

assertions cannot be true at the same time (PNC-logical(A)) (Met. 1011b14). In other words,

Aristotle’s task in Metaphysics IV is to reject theories that deny PNC-ontological. This is

because the possibility that the same property can belong and not belong to the same object

at the same time and in the same respect is the one and only way for two contradictory

present singular assertions to be true at the same time. This is because, for Aristotle, “X is

F” is true iff X is F, and “X is-not F” is true iff X is-not F (1011b26-28).22 So both “X is F”

and “X is-not F” can be true at the same time (not PNC-logical(A)) iff it is possible that X is

and is-not F (not PNC-ontological). Thus, in order to secure PNC-logical(A) with respect to

present singular assertions, Aristotle must provide a defense for PNC-ontological.

Indeed, the situation would be devastating for Aristotle’s program of scientific

demonstration if PNC-ontological did not hold. Not only would PNC-logical(A) fail, but

also LEM and PB. Since “X is F” is false when X is-not F and “X is-not F” is false when X

is F, both “X is F” and “X is-not F” would be false if X is and is-not F. Thus, LEM would

fail. However, we have just shown that “X is F” and “X is-not F” are both true if X is and

is-not F. As a result, both assertions would be true and false if PNC-ontological did not

hold. Thus, PB would fail.23 Finally, since LEM and PNC-logical(A) fail and RCP is the

21 Trans. Ackrill.

22 Cf. 1027b17ff.

23 This, of course, assumes that PB is a principle based on an exclusive, rather than inclusive, “or”—that an

assertion must be true or false means that it must be one or the other and not both.

8
combination of LEM and PNC-logical(A), RCP will also fail. Since RCP, PNC-logical(A),

LEM, and PB will all fail if PNC-ontological does not hold, we can see why Aristotle thinks

that PNC-ontological is the “ultimate belief” for anyone who is demonstrating anything and

why it is the starting-point for all other axioms (Met. 1005b33f.).24 Indeed, as Aristotle’s

defense of PNC in Metaphysics IV.4 reveals, if PNC-ontological does not hold, not only will

demonstration be impossible, but even significant speech.

III. Aristotle’s Elenctic Proof and the Threat of a Petitio Principii

Thus far, we have argued that the primary object of Aristotle’s proof is PNC-

ontological and that while PNC-ontological secures PNC-logical(A) for present singular

assertions, Aristotle’s logical version of PNC is not the modern logical version, –(p and –p).

In this section, we will unpack Aristotle’s elenctic proof of PNC-ontological and explore the

way in which he employs rules of inference such as modus ponens and modus tollens to make his

case. Because these rules, especially modus tollens, seem to presuppose the modern logical

version of PNC, Aristotle would run the risk of a petitio principii if he were arguing for PNC-

logical. However, he is not arguing for PNC-logical, but rather PNC-ontological.

Therefore, Aristotle only begs the question if his proof presupposes, in some way, that it is

impossible for some attribute to belong and not belong to the same object at the same time

and in the same respect. I will conclude by contending that it does not.

As we have seen, Aristotle argues, at the end of Metaphysics IV.3, that PNC-

ontological is the most certain of all principles by presupposing the truth of PNC-

24 With Wedin (2004, 258ff.), I defend Aristotle against Lukasiewicz (1979) who rejects Aristotle’s claim that

PNC is the highest principle of all demonstrations. However, I differ significantly from Wedin in my

explanation of why Aristotle thinks that PNC is an ultimate belief for anyone demonstrating anything.

Specifically, Wedin contends while PNC is not used in all deductions, the validity of the principles used in

deductive reasoning all depend on –(p and -p).

9
ontological. In Metaphysics IV.4, Aristotle’s task is to defend PNC-ontological itself. The

problem that he immediately confronts is that the principle cannot be positively

demonstrated. This is because PNC-ontological is the most fundamental principle of all

demonstration. If PNC-ontological were a conclusion of a demonstration, it would no

longer be the starting-point of all other axioms. This is because Aristotle stipulates that any

demonstration must proceed from premises that are prior to and better known than the

conclusion (An. post. 71b22). Therefore, if some other principle were to be used to prove

PNC-ontological, that premise would be the starting-point of all other axioms. Of course,

even if some other principle dethroned PNC-ontological, the opponent could again demand

a demonstration of this most fundamental principle and such a demonstration would, in

turn, require a principle more fundamental than it. This, however, would lead to an infinite

regress, and this is why it is want of education to demand a positive demonstration of the

most fundamental principle.

While we are told that there cannot be a positive demonstration of PNC-ontological,

Aristotle believes that he can provide an elenctic demonstration (Met. 1006a13). Although

commentators like Lukasiewicz have been baffled by this distinction, it seems quite clear

what Aristotle has in mind and why it is significant for his argument. Specifically, Aristotle is

claiming that he can successfully refute anyone who holds that something both is and is-not

(F). This is different from a straightforward demonstration because the interlocutor is now

responsible for making an argument (logos). If he does not agree to such a procedure,

Aristotle claims that he has no argument, and if he has no argument, then there is no reason

to debate with him. In the end, he is no better than a plant (1006a13-16).

It is at this point that Aristotle invites his interlocutor to meet the most minimal of

demands, namely to say “something significant both to himself and to another” (1006a22-

23). Of course, once the interlocutor answers this demand, Aristotle thinks that he can

10
successfully refute him. The reason why Aristotle thinks he can refute his interlocutor once

he says something significant is that significant speech entails the acceptance of PNC.

Indeed, we are told in Metaphysics XI that this is precisely the strategy used in Metaphysics IV:

“But he who wants to prove to the asserter of opposites that he is wrong must get from him

an admission which shall be identical with the principle that the same thing cannot both be

and not be at one and the same time, but shall not seem to be identical” (1062a5-9).25 What

seems to be said here is that significant speech entails PNC-ontological, and therefore if he

can get the interlocutor to accept significant speech, he can force him to accept PNC-

ontological.

It is here that we get a sense of the elenctic structure of Aristotle’s proof.

Specifically, the form of Aristotle’s argument is the same form that Gregory Vlastos finds in

Plato’s elenctic dialogues.26 There, Socrates combats an interlocutor’s belief that p first by

getting him to commit to some additional premise q and then showing that q entails not-p. In

other words, Socrates’ strategy is to convict the interlocutor of being committed to an

inconsistent premise-set. In the Metaphysics, Aristotle’s strategy is to combat the

interlocutor’s claim that p, i.e. not PNC-ontological, by getting his assent to q, i.e. significant

speech, and then showing that q entails not-p.

There are, however, two worries that have emerged in the secondary literature about

this strategy. The first is that Aristotle’s negative refutation seems to depend on the

interlocutor’s commitment to significant speech. That is, the interlocutor might assent to

the force of Aristotle’s demonstration that significant speech is inconsistent with the denial

of PNC-ontological, but still be able to deny PNC-ontological by then abandoning his

25 Trans. Ross.

26 Vlastos (1999).

11
commitment to significant speech. It is this apparent loophole in Aristotle’s strategy that has

led some commentators to think that Aristotle’s elenctic refutation is really a transcendental

argument on behalf of PNC-ontological.27 That is, Aristotle is arguing from the possibility

of significant speech to the truth of PNC-ontological. The problem with this reasoning is

twofold. First, there does not seem to be any reason why the interlocutor needs to assent to

the possibility of significant speech in order for the proof to work. That is, Aristotle could

simply by-pass the interlocutor, claim that significant speech is possible, and then prove

PNC-ontological. The problem with this is that it transforms Aristotle’s elenctic proof into

a straightforward demonstration. The second reason strikes at the heart of transcendental

arguments as such, and it recalls the dictum that one person’s modus ponens is another’s modus

tollens. That is, while Aristotle, on this reading, uses the premise “if significant speech is

possible, then PNC-ontological” to prove PNC-ontological by affirming the antecedent, his

interlocutor could just as well deny the possibility of significant speech by denying the

consequent. That is, he could reject PNC-ontological and join Cratylus in trying to purge

philosophy of linguistic utterances (1010a13).

The other reason for rejecting the transcendental reading is that we can make sense

of Aristotle’s elenctic refutation by noting one more facet of his argument. Specifically,

Aristotle thinks he can refute anyone who asserts an instance of not-PNC-ontological, such

as Socrates is and is-not a man, because he thinks that significant speech is a necessary

condition for asserting anything at all. Thus, if the interlocutor wants to assert an instance of

not-PNC-ontological, he must say something significant. Indeed, he does not even have to

make an assertion, i.e. a sentence capable of being true or false. Instead, all he has to do is

utter a significant word, and in so doing, he will have committed himself to the very

27 See Gottlieb (2007).

12
proposition he wants to deny, namely PNC-ontological. Thus, while Aristotle’s elenctic

refutation does not provide positive proof of PNC-ontological, it does show that it is

impossible to assert an instance of not-PNC-ontological. Therefore, the interlocutor is left

with a choice, either accept PNC-ontological or refrain from asserting anything at all, even

an instance of not-PNC-ontological. Of course, if he refrains from asserting anything at all,

he will be, again, no better than a vegetable (1006a15).

As noted above, a second problem with Aristotle’s proof has emerged in the

secondary literature, and it is particularly worrisome for those who read Aristotle as

defending the modern logical version of PNC, –(p and –p). Specifically, his elenctic proof

seems to presuppose PNC-logical at three possible junctures. In the first case, the problem

has to do with the conclusion of the elenctic proof. If we assume that Aristotle is arguing

with someone who denies PNC-logical, his elenctic demonstration will not force the

interlocutor to abandon his position. While Aristotle will have shown that significant speech

entails the acceptance of PNC-logical, -(p and –p), the opponent’s acceptance of PNC-

logical will only entail the rejection of not-PNC-logical, (p and –p), if the opponent is already

committed to PNC-logical. That is, having accepted PNC-logical, the opponent, qua

proponent of not-PNC-logical, is free to accept both PNC-logical and its negation. While

Whitaker has argued that once the interlocutor accepts PNC-logical he will not be able to

consider the possibility of accepting both PNC-logical and its negation,28 Priest has

contended that Whitaker’s strategy simply presupposes PNC-logical.29 That is, the only

reason why the acceptance of PNC-logical, -(p and –p), entails that abandonment of its

negation, (p and –p), is because failing to abandon not-PNC-logical would be contradictory.

28 Whitaker (1996, 199).

29 Priest (1998, 105).

13
That is, it would commit the interlocutor to holding both (p and –p) and –(p and –p), and

this is only problematic for someone already committed to PNC-logical.

For those who read Aristotle as defending PNC-logical, the structure of Aristotle’s

elenctic proof also seems to presuppose PNC-logical. Specifically, Aristotle’s proof takes the

form of a modus ponens, and one might worry that modus ponens presupposes PNC-logical.

While Lukasiewicz has argued that modus ponens, in contrast to modus tollens, does not

presuppose PNC-logical,30 the difficulty with Lukasiewicz’s position is that it threatens

Aristotle’s claim that PNC is the starting point of all axioms (1005b35). That is, on this

reading, one cannot help Aristotle avoid a petitio principii by showing that modus ponens

operates independently of PNC and can therefore be used in the proof of PNC because

PNC, as the ultimate belief, is supposed to be secure modus ponens. Indeed, Wedin has noted

this problem and tried to resolve it by arguing that while the validity of modus ponens depends

on PNC-logical (thereby preserving the ultimacy of PNC-logical), a modus ponens argument

does not use PNC-logical as a premise in the proof (thereby avoiding a petitio principii).31

While the final instance of a possible petitio principii is less obvious, it is more

threatening than the second. It is less obvious because it is embedded in Aristotle’s dense

argument for the claim that “man” and “not-man” do not signify the same thing (1006b13-

28).32 It is more threatening because Aristotle employs modus tollens rather than modus ponens.

Specifically, he claims that if (a) “man” and “not-man” did not signify something different,

(b) “being man” would not signify something different from “not being a man” (1006b23-

25). Aristotle argues, however, that it has been proven that (not-b) “being man” and “not

being man” signify something different (1006b28). Therefore, Aristotle concludes that (not-

30 Cf. Lukasiewicz (1993, 69 and 86ff.).

31 Wedin (2000, 116ff. and 131).

32 Indeed, Wedin (2000, 155) is the one who points this out. Lukasiewicz misses it.

14
a) “man” and “not-man” signify something different, i.e. do not signify the same thing.

Aristotle’s use of modus tollens is more threatening because Lukasiewicz has argued that modus

tollens, in contrast to modus ponens, depends on PNC, and since modus tollens depends on PNC,

Aristotle cannot use modus tollens to prove PNC without begging the question. For if the

consequent is denied (not-b) and the antecedent is not denied (a), the affirmed antecedent (a)

would entail the consequent (b), and that would leave us with a contradiction (b and not-b).

This then is why Lukasiewicz claims that using modus tollens to prove PNC begs the question.

At this point, we might try to save Aristotle from his blunder by introducing Wedin-

type distinctions or even reworking Aristotle’s argument so that it does not use modus tollens.

Here, however, we must recall that these difficulties only emerge because commentators

have assumed that Aristotle is defending PNC-logical and that PNC-logical is supposed to

be the starting point for axioms and even inferential rules such as modus ponens and modus

tollens. As we have seen, however, Aristotle is defending PNC-ontological, not PNC-logical,

and it is this distinction that explains why Aristotle does not beg the question.

In responding to worries about the relationship between PNC and modus ponens and

modus tollens, we need to make two points. The first is that while there might be some

question as to whether modus ponens and, especially, modus tollens depend on PNC-logical, this

is not an issue that the interpreter of Metaphysics IV needs to resolve. Since Aristotle is

defending PNC-ontological, the real question is whether modus ponens and modus tollens

presuppose PNC-ontological. I hold that they do not for the simple reason that whereas

modus ponens and modus tollens are rules of inference, PNC-ontological is a claim about being

qua being. Whether PNC-ontological holds of property bearing subjects has nothing to do

with whether certain rules of inference govern the relationship between the premises and the

conclusion of a given argument. Thus, Aristotle can use modus ponens and modus tollens in his

attempt to prove PNC-ontological without begging the question.

15
 As noted above, the problem with detaching PNC-ontological from modus ponens and

modus tollens is that it seems to undermine Aristotle’s claim that PNC-ontological is the

starting point of all axioms. Such worries, however, are unfounded, and they are unfounded

because they presuppose that Aristotle considers modus ponens and modus tollens to be axioms

of this sort. Here, there are two good reasons for thinking that Aristotle does not hold this

position. The first is that Aristotle never calls modus ponens and modus tollens axioms. Indeed,

when Aristotle does provide us with examples of axioms, which unfortunately is not very

often, he mentions three: PNC-logical(A), LEM, and the principle that if equals are

subtracted from equals the remainders are equals.33 In response, one might argue that while

Aristotle does not explicitly call modus ponens and modus tollens axioms, he still thinks of them

as such. The major objection to this response is that Aristotle does not include these rules

of inference in his system of syllogistic demonstration.34 This is only a later development.35

Of course, this does not mean that Aristotle does not or cannot use what we now recognize

as modus ponens and modus tollens to secure a given conclusion. Indeed, we have seen that he

does precisely this in his defense of PNC-ontological. What it does mean, however, is that

we should not worry about whether PNC-ontological secures the validity of modus ponens and

33 McKirahan (1992, 68ff.). Regarding PNC, McKirahan does not distinguish between PNC-ontological and

PNC-logical(A). Citing PNC-logical(A), rather than PNC, is therefore my modification. Also, I am tempted to

add RCP to McKirahan’s list. Indeed, An Post. 71a14 and 77a23, which state that everything must be either

asserted or denied, supports RCP if we read the “or” as exclusive. Of course, the inclusive reading of “or”

would support LEM.

34 Striker (1998, 215ff.) notes that Aristotle did not recognize complex sentences formed by logical connectives

such as conjunctions and conditionals.

35 See Bobzien (2002).

16
modus tollens in order to make good on Aristotle’s claim that PNC-ontological is the starting-

point of all axioms.

This brings us to our final point, where we must respond to Priest’s charge that

Aristotle’s conclusion begs the question. Here, we can say that while Aristotle’s argument

does depend on the principle that something (X is and is-not F) cannot both be possible and

impossible at the same time and in the same respect, a principle which forces the

interlocutor to reject the possibility of something on the grounds that it has been shown to

be impossible, Aristotle nevertheless does not beg the question because he is not trying to

prove that something cannot both be possible and impossible at the same time and in the

same respect. Instead, he is simply trying to show that it is impossible for X to be and not-

be F. While Priest rightly holds that the aforementioned principle is an instance of PNC, it

is only an instance of the modern logical version of PNC, not Aristotle’s limited, ontological

version of PNC, and since Aristotle is providing us with a negative demonstration of the

latter and not the former, I hold, again, that Aristotle escapes the charge of a petitio principii.

17
 Works Cited

Translations and Commentaries of Aristotle’s Works:

Ackrill, J. L. 1984. Translation of Aristotle’s De Interpretatione in The Complete Works of

Aristotle, ed. Jonathan Barnes, Vol. 1. Princeton, NJ: Princeton University Press.

Kirwan, Christopher. 1993. Translation with commentary. Aristotle Metaphysics: Books

Gamma, Delta, and Epsilon. Oxford: Clarendon Press.

Ross, W.D. 1984. Translation of Aristotle’s Metaphysics in The Complete Works of Aristotle, ed.

Jonathan Barnes, Vol. 2. Princeton, NJ: Princeton University Press.

Tredennick, Hugh. 2003. Greek and English text. Aristotle: Metaphysics I-IX. Cambridge,

MA: Harvard University Press.

Secondary Literature:

Bobzien, Susanne. 2002. “The Development of Modus Ponens in Antiquity: From Aristotle

to the 2nd Century AD.” Phronesis 47(4): 359-394.

Brinkmann, Klaus. 1992. “Commentary on Gottlieb.” Proceedings of the Boston Area

Colloquium in Philosophy 8: 199-209.

Cresswell, M.J. 2003. “Non-Contradiction and Substantial Predication.” Theoria 69: 166-

183.

Englebretsen, George. 1981. Logical Negation. Assen, The Netherlands: Van Gorcum.

Gottlieb, Paula. 1992. “The Principle of non-Contradiction and Protagoras: The Strategy of

Aristotle’s Metaphysics IV 4.” Proceedings of the Boston Area Colloquium in Philosophy 8: 183-

198.

------. 2007. “Aristotle on Non-Contradiction.” Stanford Encyclopedia of Philosophy. Web

address: http://plato.stanford.edu/entries/aristotle-noncontradiction

18
Grim, Patrick. 2004. “What is a Contradiction?” In The Law of Non-Contradiciton: New

Philosophical Essays, eds. Graham Priest, J.C. Beall, and Bradley Armour-Garb. Oxford:

Clarendon Press.

Lukasiewicz, Jan. 1979. “Aristotle on the Law of Contradiction.” In Articles on Aristotle,

Vol. 3, eds. J. Barnes, M. Schofield, R. Sorabji, 50-62. New York: St. Martin’s Press.

------. 1993. Ueber den Satz des Widerspruchs bei Aristoteles. Trans. Jacek Barski. Vol. 5 of Zur

modernen Deutung der Aristotelischen Logik, ed. Niels Oeffenberger. Hildesheim: Georg

Olms Verlag.

McKirahan, Richard D. 1992. Principles and Proofs: Aristotle’s Theory of Demonstrative Science.

Princeton, NJ: Princeton University Press.

Priest, Graham. 1998. “To be and not to be—That is the Answer. Aristotle on the Law of

Non-Contradiction.” Philosophiegeschichte und Logische Analyse 1: 91-130.

Striker, Gisela. 1998. “Aristotle on the Uses of Logic.” In Method in Ancient Philosophy, ed.

Jyl Gentzler, 209-226. Oxford: Clarendon Press.

Vlastos, Gregory. 1999. “The Socratic Elenchus.” In Metaphysics and Epistemology, ed. Gail

Fine, 36-63. Oxford: Oxford University Press.

Wedin, Michael V. 2000. “Some Logical Problems in Metaphysics Gamma.” In Oxford Studies

in Ancient Philosophy, ed. David Sedley, 113-162.

------. 2004. “Aristotle on the Firmness of the Principle of Non-Contradiction.” Phronesis

49(3): 225-265.

Whitaker, C.W.A. 1996. Aristotle’s de Interpretatione. Oxford: Clarendon Press.

19