Logic: A Brief Introduction

Ronald L. Hall, Stetson University

PART II - Categorical Logic

Chapter 5 - Categorical Propositions
5.1 Introduction
It was Aristotle (384-322 BCE), the most famous student of Plato (428-347 BCE), who was the first to develop a
formal system of logic. Aristotles followers gathered his writings on logic and compiled them into what they
called the Organon, which is a word that means "instrument." And this is exactly how Aristotle conceived of logic,
as an instrument for the scientific and philosophical investigation of reality.
For Aristotle, reality is organized in categories or classes. Such classes have members and Aristotle thought
knowledge of reality consisted of true propositions (assertions) about these categories and their members. Such a
proposition would be, for example, All human beings are mortal.
What makes this system of logic formal is that, according to Aristotle, every proposition, regardless of its content,
can be expressed in one of a small number of basic propositional forms. He was aware of the fact if we could
reduce all propositions to these basic forms or types, then it would be much easier to manage the task of
comprehending the ways in which these propositions are related in arguments. For example, consider the
following classic valid argument:
All human beings are mortal.
Socrates is a human being.
Therefore Socrates is mortal.

Aristotle realized that the conclusion of this argument is related to its premises formally, that is, without respect
to the fact that the propositions are about mortality, human beings, and Socrates. That is, he came to see that all
arguments of this form are equally valid. To see this, consistently substitute anything you like for the categories of
human beings, mortal and Socrates. (You might find it odd to think of Socrates as a category, but for Aristotle it
was indeed a class, although it has only one member. We will discuss this issue more fully later.) Consider the
following argument that has exactly the same form that the argument above has, and certainly is valid.
All dogs are mammals.
Fluffy is a dog.
Therefore, Fluffy is a mammal.

64

Now if there are only a few propositional forms, then there will be a small number of possible combinations of
such propositions, that is, a small number of argument forms. Hence it will be easy to see by a process of
substitution which of the thousands of arguments with their various and sundry contents are valid and which are
not, for what makes the valid ones valid has nothing to do with their content. Rather what makes an argument
valid or not is simply a matter of the formal relation that exists between its premises and its conclusion. (Recall,
an argument is valid if it is impossible for the conclusion to be false if its premises were true, and invalid
otherwise.)
What an enormously important breakthrough in the study of logic. With Aristotle's ground-breaking insight, it
became easier to distinguish between good reasoning and bad reasoning. To honor Aristotles contribution, this
system of logic is often referred to as Aristotelian Logic.
In this chapter, and in the next, we will explore Aristotle's system of categorical logic in detail. What we will
discover is that while it is a powerful system of logic, it does have some limitations. We will begin to see this
especially when we come to the task of translating ordinary language into categorical propositions. As we will find
out, translating from ordinary language into categorical form is not always easy to do. Indeed, sometimes it seems
as though putting ordinary language into categorical form requires too much forcing, and perhaps even some
distortion of meaning. This is so because our ordinary language is richer in its capacity to mark distinctions than
the forms of categorical language can capture. It is this fact that finally gave rise to the development of different
and more powerful logical systems. We will study these systems later when we get to sentential logic and
predicate logic.
Now, we must turn to the first issue in our exploration of categorical logic, namely, to the formulation of the basic
propositions in this formal system. In the next chapter, we will turn to the study of categorical arguments. Along
the way, we will deal with the issue of translation, that is, with the issue of translating ordinary language into the
language of categorical logic. Our discussion of translation in this chapter, and in the next, should prepare us for
exploring more sophisticated formal systems.

5.2 Standard-Form Propositions

In the system of categorical logic, there are four standard-form propositions, and only four. You mean everything
that we can assert can be translated into one of these four forms? Thats right; there are only four kinds of
propositions, only four. Now dont forget, there are thousands of different propositions, if we differentiate them
in terms of their content, but only four if we distinguish them in terms of their form.
So what are these four standard-form categorical propositions? Well, here they are:
1. All S is P
2. No S is P.
3. Some S is P
4. Some S is not P

In the statements of these propositional forms, the S stands for the subject of the assertion, and the P stands
for the predicate of the assertion. In case your grammar is a little rusty, let me refresh your memory about what
65

subjects and predicates are. It might help to say that a predicate is what is said of a subject and the subject is
what something is said of. Clearly the two terms are defined in relation to each other. To be more precise,
however, we must state that relation in terms of categories or classes. So lets be clear that both the subject and
the predicate terms in standard-form categorical propositions are class terms (but notice again, classes may have
only one member; and to complicate this a little more, a class may have no members at all; but we will get to this
presently). In all of these propositional forms an assertion is made to the effect that the class or members of the
class, designated by the subject class term is either included in or excluded from the class, or members of the
class, that is symbolized by the predicate term.
Clearly then, the proposition, All human beings are mortal, can be translated into the first standard-form
categorical proposition, All S is P. As we might state this: All members of the class of human beings are also
members of the class of mortal beings. This statement captures the sense of the original proposition, as does its
symbolic expression, All S is P.
Before we go on to consider the other standard-form propositions, let's adopt the standard designation of each of
these propositions. This will make it easier to refer to them. We will now designate each proposition with a capital
letter. These letters are A, E, I, and O. The four standard-form propositions then are as follows:
A: All S is P"
E: No S is P"
I: Some S is P
O: Some S is not P

So there you have it, categorical logic has four, and only four, standard-form propositions. And it is a matter of
convention, that is, a matter of convenience, that we refer to them with the letters, A, E, I, and O respectively.
From this, it should be clear that our example above, namely, All human beings are mortal, is an A proposition.
And it is easy to think of examples of other A propositions.
Now let us consider the E proposition: No human beings are pigs. Clearly this means (literally speaking; we will
leave metaphor out of the picture at this point) that no member of the class of human beings is also a member of
the class of pigs. As opposed to the A proposition, which is an assertion that the class designated by its S-term is
included within the class designated by its P-term, the E proposition is an assertion that the class designated by its
S-term is excluded from the class designated its P-term.
When a proposition asserts a relation of class inclusion (actually it is speakers who assert things with propositions,
but for convenience, we will adopt this shorthand way of referring to propositions as asserting something) we will
say that it has an affirmative quality. By the same token, when a proposition asserts a relation of class exclusion,
we will say that it has a negative quality. Accordingly, the A proposition has an affirmative quality while the E
proposition has a negative quality.
In categorical logic, we must sharply distinguish the quality of a proposition from its quantity. Just as there are
two qualities that propositions can have, namely, affirmative and negative, they may also have two different
quantities. These two quantities are commonly designated with the terms universal and particular. Every
66

standard-form categorical proposition is said to be either universal or particular in quantity. Both all and no
are universal quantifiers since they refer to all of the members of the class of S in the A and E propositions: All S
is P and in No S is P.
OK, lets re-group. So far it should be clear that of the four Standard-form categorical propositions, A and E
propositions are both universal in quantity and the A proposition is affirmative in quality while the E proposition
is negative in quality. What makes them both universal is that in both cases, the S-term of each proposition is
talking about every member of the class it designates. What makes the A proposition affirmative is the fact that it
asserts that the class designated by its S-term, in part or as a whole, is included within the class designated by its
P-term. What makes the E proposition negative is that the class designated by its S-term, in part or as a whole, is
excluded from the class designated by its P-term. (It is probably the case that the A designation derives from the
Latin term for I affirm, just as the E designation derives from the Latin term for I deny, which when
pronounced sounds like NEE-go, hence the E.)
In our two examples above, the S-term is the class of human beings. In the first case, the A proposition, All
human beings are mortal is a universal affirmative proposition because every member of the class designated by
the S-term, human beings is asserted to be included within the class designated by the term mortal (beings).
In the second case, the E proposition, No human beings are pigs is a universal negative proposition because
every member of the class designated by the S-term, human beings, is asserted as excluded from the class
designated by the P-term, pigs.
So much then for the A and the E propositions. Lets turn to the two others, namely, the I proposition and the O
propositions. Like their cousins the A and the E, these two propositions have a common quantity and a different
quality. Consider the quality of the two propositions first. The I proposition is affirmative, the O proposition is
negative. The same reasoning holds with regard to the quality of these two propositions as in the case of their
universal counterparts. The I proposition, Some S is P is affirmative in quality because it asserts that the class
designated by its S-term, in part or as a whole, is included within the class designated by its P-term. Similarly, the
quality of the O proposition, Some S is not P is negative because it asserts that the class designated by its Sterm, in part or as a whole, is excluded from the class designated by its P-term. (It is probably the case that the I
designation derives from the Latin term for I affirm, which has an I in it, just as the O designation derives
from the Latin term for I deny, which has an O in it.)
If we say that a proposition is universal in quantity because its S-term refers to all of the members of the class that
it designates, we are going to need a term to capture the case in which the S-term does not refer to all of its
members. That is, we need a term to express the case in which the S-term refers to only some (at least one) of the
members of the class it designates. The term some logicians have adopted to capture the quantity of some is
particular. Accordingly, it is said that the I proposition and the O proposition are both particular in quantity.
Often, however, the term existential is used instead of the term particular. I prefer the term existential to
particular since to say that Some S is P or that Some S is not P is just to say that there exists at least one
member of the class S.
In sum then, here is where we are. I and O propositions are both existential in quantity and the I proposition is
affirmative in quality and the O proposition is negative in quality. Every standard-form categorical proposition has
either a universal or an existential quantifier and is either affirmative or negative in quality.
67

The text box below shows the various combinations of qualities and quantities that characterized the four
standard-form categorical propositions.
Quantity
Universal: A E; Existential: I O
Quality:
Affirmative: A I; Negative: E O

Categorical logic lends itself very nicely to visual representation. This is so because it involves class inclusion and
exclusion. Classes can easily be represented with Venn Diagrams. We can represent all of the standard-form
categorical propositions with two intersecting circles, each representing a separate class, one the S-class and the
other the P-Class.
DIAGRAMING A, E, I, and O PROPOSITIONS:
In the case of the A proposition, we represent the assertion that all of S is included within P ("All S is P") by
emptying that section of the circle representing the S-class that is outside of the circle representing the P-class.
That is, we empty the section of the interlocking circles that is excluded from P by shading it in black, as indicated
below. To blacken a section is to erase it. The diagram shows that all of the class of S that is left (after the
blackening process) is contained wholly within the class P.

THE A PROPOSITION

We represent the E proposition, No S is P in a similar way. To represent the assertion that all of S is excluded
from all of P, we empty the section of the interlocking circles that is common by shading it in black, as indicated
68

below. Again, to blacken a section of the diagram is to erase it. Now clearly all of S that remains is excluded from
P.
THE E PROPOSITION

We represent the I and O propositions differently. We represent the I proposition, that is, the assertion that
Some S is P by showing that the class S has at least one member and that this member of S is also a member of
the class of P. Remember, Some S is P is to be read as: There eXists and S We can show this by placing an X
in the area that is common to both S and P as follows:

THE I PROPOSITION

Finally, we represent an O proposition, that is, the assertion that Some S is not P by showing that the class S has
at least one member and that this member is not a member of the class of P. We do this by placing an x the
area of S that is not in the area of P as follows:
THE O PROPOSITION
69

When I have talked so far about whether a term refers to all or to only some (at least one) member of the class it
designates, that is, when I have talked about whether a proposition is affirmative or negative, I have only
mentioned the range of the S-term. (It can range over the whole or over a part of the class it designates.) I have
not mentioned the range of the P-term. We need to do this and to make it clear that both the S-term and the Pterm in a standard-form categorical proposition can refer to either all or part of a class. But first I need to
introduce the concept of distribution and let you know what distribution means in this context. With this
definition, we will be able to see that the every S-term and every P-term in every standard-form categorical
proposition is either distributed or undistributed
So here is our definition of distribution: Any S-term or P-term is distributed if and only if it refers to all of the
members of the class that it designates, and any S-term or P-term is undistributed if and only if it refers to only
some (at least one) member of the class that it designates.
It should be obvious that in both the universal propositions, A and E, the S-term is distributed. Clearly in the
propositions all human beings... and no human beings... the S-term leaves no member out of the class it
designates.
Similarly, in both the existential propositions, I and O, (Some S are... and "Some S are not...") the S-term is
referring to only some (at least one) members of the class it designates. Hence, in the I and O propositions, the Sterm is undistributed.
So where does this leave us with regard to the P-terms of our Standard-form propositions? Lets take the universal
propositions first. In the A proposition, All human beings are mortal the assertion is not intended to refer to all
of the members of the class designated by its P-term, mortal beings. We say then that the P-term in an A
proposition is undistributed.
In the E proposition, No human beings are pigs the intention is to exclude every member of the class of human
beings from the entire class (pigs) designated by the P-term of this proposition. Accordingly, we say that the Pterm in an E proposition is distributed.
In the case of the particular propositions I and O, both of which have undistributed S-terms, what do we say of
their respective P-terms? In the case of an I proposition, for example, Some cats are pets we are clearly not
talking about the entire class of pets. Hence the P-term in this proposition, which designates the class of pets, is
70

not intended to refer to every member of that class. The whole class of cats can fit into the class of pets
without exhausting the class of pets. There is room for dogs in that class, for example. So we say that the P-term
in an I proposition is undistributed.
In the case of an O proposition, for example, Some cats are not pets we are clearly talking about the entire class
of pets. That is, some cats, feral and other kinds, fall completely outside the entire class of pets. There is, in other
words no room in the class of pets for such creatures. So the P-term in an O proposition is distributed.
Now lets notice some parallels in summary. First, both universal propositions, A and E, always distribute their Sterms. The S-terms in both particular propositions, I and O, are undistributed. The P-terms in both affirmative
propositions, A and I, are undistributed, and in both negative propositions, E and O, the P-term is distributed.
Here is a more graphic summary:

Distribution Table
A: S-term distributed; P-term undistributed (DU)
E: both terms distributed (DD)
I: both terms undistributed (UU)
O: S-term undistributed; P-term distributed (UD)

5.3 More Complex Translations

It is quite obvious that in ordinary modes of conversation, we do not tend to talk or write in Standard-form
categorical propositions. Indeed, there is something almost unnatural, or at least stilted, about such Standardform categorical expression. This is a problem for categorical logical analysis; however, since our method of
assessing the validity of categorical arguments is going to depend on how accurately we translate ordinary
expressions into Standard-form categorical propositions.
Fortunately, many, perhaps even most, but certainly not all, of our ordinary assertions can be reformulated into
Standard-form categorical propositions. A few hints will help in this process. For example, some propositions do
not have explicit quantifiers, or else they are expressed in non-standard ways. When we find such expressions, we
must interpret which quantifier is intended and supply it. Let's see how this might work.
Consider the following proposition: Cheetahs run fast. Clearly, this assertion is not in standard categorical form.
A little reflection, however, will make it clear that the implicit quantifier here is universal. That is, the proper
translation of this proposition should be "All C are R."
Matters would be quite different if the proposition had been: Cheetahs were recently added to the zoos
collection. In this case it is implausible to think that the zoo now has every Cheetah in its collection. So clearly the
proper translation of this expression should be: Some Cheetahs are animals recently added to the zoos
collection.

71

The same sort of distinction exists between the following two propositions: (1) Roses have thorns and (2) Roses
are taking over my flower bed. In the first case, we are clearly talking about all roses whereas in the second we
are referring to some roses.
In determining the quantifier to use in cases where no quantifier is specified, do not be misled into thinking that
the articles the or a or an settle the issue. They do not. For example, The Cheetah is endangered refers to
all Cheetahs, and hence is universal in quantity. Just as clearly, The Cheetah escaped refers to only one, and
hence is particular in quantity. And similarly, A mind is a terrible thing to waste refers to all minds, and hence
is universal in quantity, just as a great mind was lost when he died refers to only one mind, hence is particular in
quantity.
Sometimes qualifiers and quantifiers are embedded in an ordinary expression and this may pose a challenge for
translation. Consider this example: My students are not all well motivated. Even though this sentence has all
in it, there is no plausible way to translate it as an A proposition. The reason for this is that its quality is clearly
negative, and the quality of an A proposition is affirmative. As well, the sense of the assertion seems not to be
universal, since what seems to be intended in the claim that not all like not every is an alternative way of
expressing some are not... For example, the proposition Not all Americans vote clearly means: Some
Americans do not vote. Accordingly the translation of our proposition above should be the following O
proposition: Some students in my class are not well motivated.
And consider one more thing regarding negative qualifiers. Had the expression above been not any instead of
not every or not all the translation would have required a different quantifier. Clearly not any is a universal
claim, and could best be translated as: No student in my class is well motivated.
We might formulate a general guideline regarding quantifier and quantifier translation as follows: absent
qualifiers and quantifiers should be supplied and non-standard qualifiers and quantifiers should be translated as
standard quantifiers. "A few," for example, should be translated as "some," and "not everyone" should be
translated as "some are not," and "anyone" should be translated as "all," and so forth.
Now there are other such general guidelines for good translation. Below are a few. Keep in mind, however, that
good translations must preserve the sense of the original propositions that are being translated. Nevertheless,
even the most creative art can make use of some guidelines. So, again, here are a few:
1. Translate an assertion that contains exclusive terms such as "only" or "only if," "none but," "except," "none
except," and so forth, regardless of where these terms of exclusion occur in the sentence, as a universal
affirmative (A) proposition and make what follows the term of exclusion the predicate term of that proposition.
For example, translate, "None but the brave are rewarded" as "All R (people rewarded) is B."
2. Translate an assertion that contains "the only" as an A proposition. For example, translate "The only survivors
were children," as "All S is C."
3. Translate an assertion that contains "if," regardless of where it occurs in the sentence, as a universal
(affirmative or negative) proposition, and make what follows the "if" the subject term of that proposition, e.g.
translate "You can go if you study" as "All S (times you study) are G (times you can go) or "You cannot go if you
sleep late" as "No S (times you sleep late) are G (times you can go." Remember, there is a very big difference
72

between "if" and "only if." When an assertion contains only if take this as stating a necessary condition and
translate it as follows; translate this by making the only if statement into an A proposition with what immediately
follows the only if as the predicate term of this A proposition. For example: You may go to the movies only if
you do your homework should be translated as All M (times you can go to the moves) are H (times you do your
homework).
4. When "not" is contained in an "all" proposition, as in "All S are not P," translate the assertion as an "E"
proposition ("No S is P.")
5. Translate assertions with non-standard verbs into the verb "to be." For example, "Good logic students study
hard," would be translated as "All G (good logic students) are S (students who study hard).
6. When a term in an assertion does not contain a noun, then it must be supplied. For example, "Some people
smoke" should be translated as "Some P (people) are S (smokers)."
7. Assertions with singular proper names are complicated. For now we will say that such propositions should be
translated as universal propositions. For example, "Socrates is mortal" is translated as the A proposition "All S are
M." As well, Socrates is not mortal should be translated as the E proposition No S are M. (Admittedly, most,
but not all, singular propositions seem to carry existential import so this practice of translating singular
propositions into universal propositions needs qualification when it comes to assessing the validity of arguments
containing singular propositions. This is discussed in the next chapter.)
8. Assertions of existence, such as there is should be translated with the use of the existential quantifier
some. For example, there is ice in the refrigerator should be translated as Some I is R.
In the next chapter, we will return to this issue of translation. As we will see, the problem of translating ordinary
language into Standard-form categorical propositions is one of the biggest difficulties in the Aristotelian formal
system, and one of the reasons that new systems of logic were developed later.

5.4 The Traditional Square of Opposition

Hopefully it is obvious to you that these four Standard-form categorical propositions bear some relation to one
another regarding their truth-values, at least when the class terms in these propositions are common. For
instance, if we are talking about the class of human beings and the class of pigs, the fact that the E proposition
involving these two classes, e.g., No human beings are pigs is true, means that we can immediately infer that its
corresponding universal proposition, namely, the A proposition, All human beings are pigs, cannot also be true,
that is, must be false. As well, if the I proposition Some human beings are pigs is false, then the corresponding
particular, namely, the O proposition, Some human beings are not pigs cannot also be false, that is, must be
true.
We can organize these various immediate inferences in terms of what has been called a system of oppositions.
There are three basic categories of these oppositions. The first category is that of contradiction. No doubt you
recognize that people sometimes contradict themselves. (Actually the term literally means, to speak against.)
Perhaps, however, you do not know exactly what the charge that we have contradicted ourselves amounts to.
Well in the language of categorical logic, we can now define the notion of contradiction more precisely. We will
say that two propositions are contradictories if and only if they both cannot be true and they both cannot be
73

false. This being true, then of course if a given proposition is true its corresponding contradictory proposition
must be false, and if false, its corresponding contradictory proposition must be true.
In categorical logic A and O propositions are contradictories and E and I propositions are also contradictories.
Knowing this allows us to make some very useful immediate inferences. If I know that Some human beings are
not pigs is true, the proposition All human beings are pigs must be false. As well, if Some cats are pets is true,
then its corresponding contradictory proposition, No cats are pets must be false. And we can go in the other
direction also. If No people are pigs is true, then its corresponding contradictory proposition, Some people are
pigs, must be false. In general then, if we know that a given proposition is true then we know that its
contradictory proposition is false, and vice versa.
As we might put this, immediate inferences of this sort constitute valid arguments. That is, if we assume an A
proposition is true, then it must be the case that its corresponding proposition, the O proposition, is false. This is
called an immediate inference because it is not mediated by any other premise. Immediate inferences involve
only one premise and only one conclusion.
Contradiction, however, is not the only relation of opposition. Propositions can also be related to one another as
contraries. Two propositions are contraries if and only if they cannot both be true, though they could both be
false. This relation exists only between the two universal propositions, the affirmative A proposition and the
negative E proposition. Given that A and E propositions are contraries, if we know that an A proposition is true,
then we also know that its corresponding E proposition must be false. And if we know that an E proposition is
true, then we know that its contrary, the A proposition must be false. Contraries, to repeat, cannot both be true.
But since they can both be false, knowing that an A or an E proposition is false, tells us nothing about the truthvalue of its contrary.
The same relation does not hold between the corresponding particular affirmative and negative propositions,
namely, the I and the O propositions. These two particular propositions are said to be sub-contraries. Two
propositions are sub-contraries if both propositions cannot be false, even though both can be true. So, if we
know that Some cats are not pets is false, then we can immediately and validly infer that Some cats are pets is
true. Again, however, the fact that the proposition Some cats are pets is true, does not imply that the
proposition Some cats are not pets is false, for clearly both of these propositions are in fact true. (Please take
note of the difference between the terms infer and imply. The premises in a valid argument imply its
conclusion, but only people can infer, as well as imply something, that is, only people can make inferences.
Premises do not infer their conclusions; they imply them, if valid. Persons make inferences, valid and invalid.
There are no invalid implications.)
There is one more such opposition to consider. This is the relation called sub-alternation. This is a relation
between a universal proposition and the corresponding particular proposition with the same quality (affirmative
or negative). The relation of sub-alternation exists accordingly between an A and an I proposition and between an
E and an O proposition. It is really quite an obvious relation when you think about it. If an A proposition is true, for
example, All human beings are mortal then surely it must be true that Some (at least one) human being is
mortal. And if No human beings are pigs is true, then it must be true that Some (at least one) human being is
not a pig is also true. However, if we know that an A or an E proposition is false, we cannot immediately infer
anything about its corresponding particular proposition. If No cats are pets is false, we do not immediately know
74

that Some cats are not pets must also be false (or must be true), although as a matter of fact it happens to be
true. And if we know that All cats are pets is false, we do not also immediately know that Some cats are pets
must be false (or must be true), although as a matter of fact it happens to be true.
If we turn the relation of sub-alternation around, we can make a couple of more immediate inferences. If an I
proposition is false, then surely its corresponding A proposition must be false. If it is false that Some human
beings are pigs then surely it must be false that All human beings are pigs. And if Some cats are not pets is
false, then surely, No cats are pets must also be false. However, nothing follows about the truth or falsity of
their corresponding universal propositions if we know only that the particular proposition is true. From the fact
that the I proposition Some cats are pets is true, nothing follows about the truth or falsity of the A proposition,
All cats are pets which as a matter of fact is false. We could, however think of a case in which an I proposition is
true and its corresponding A proposition is also, as a matter of fact, true. For example: Some cats are mammals
is true, and so is All cats are mammals.
And the same can be said about the relation between the O and the E propositions. If the O proposition is false,
then the corresponding E must also be false. However, if the O proposition is true, nothing follows about the truth
or falsity of its corresponding E proposition, that is, it may be true or it may be false. Think of some examples that
demonstrate this relation of sub-alternation.
We see then that:

Contraries: cannot both be true, but both may be false

Sub-Contraries: cannot both be false, but both may be true
Contradictories: cannot both be false, cannot both be true
Sub-Alternations: If the super is true, the sub must be true; if sub is false, super must be false

75

The image below represents the various relations of immediate inference in the traditional square of opposition.

The boxes below summarize the various immediate inferences in the traditional square of opposition:
If A is true, then E is false, I is true, and O is false.
If A is false, then O is true, and E and I are unknown.

If E is true, then A is false, I is false and O is true.

If E is false, then I is true and A and O are unknown.

If O is true, then A is false and E and I are unknown.

If O is false, then A is true, E is false and I is true.

If I is true, then E is false and A and O are unknown.

If I is false, then A is false, E is true, and O is true.

5.5 The Existential Assumption

However brilliant Aristotle was, he did overlook one important thing in the formulation of his system of
categorical logic. What he did not consider is that classes can be empty. Aristotle assumed that every class has at
least one member. We call this assumption, the existential assumption. Recognizing that this assumption is not
necessarily true has enormous consequences for categorical logic, especially with regard to the immediate
inferences that we have just been discussing.
Before we can say what the consequences are of making the existential assumption, we must say a word about
the difference between the universal and the particular quantifiers. As we have said, the particular quantifier
some is to be read as asserting, there exists at least one. So, for example, the proposition Some S is P is to
be read as asserting that its subject class has at least one member. Some flowers are roses asserts that there is
at least one member of the class of flowers and that member is included also in the class of roses. Given the fact
that particular propositions make an existential assertion, and given that classes can be empty, we must draw the
76

following conclusion: Every particular proposition that has an S-term that designates an empty class is false. For
example, consider the proposition, Some Martians are kind. If we can grant that the class of Martians is empty,
the assertion that there exists at least one member of that class that is kind must be false, for no members of that
class exist. And the same can be said for the corresponding O proposition, Some Martians are not kind. Both of
the propositions are false, since they both assert that there are members of their respective subject classes, when
these classes are empty.
Recall that in Aristotles system the I and the O propositions are sub-contraries. Recall also that sub-contraries
cannot both be false. As Aristotle reckoned, if we know that a given particular proposition is false we can validly
and immediately infer that its corresponding sub-contrary is true. We now recognize that this immediate
inference is valid only if we make the existential assumption. If we do not, that is, if the class designated by the
subject term of a particular proposition is empty, both of these propositions are false, as we just pointed out in
the case of the propositions about Martians.
The moral here is this: the immediate inference from the falsity of a given particular proposition to the truth of its
corresponding sub-contrary does not hold unless the existential assumption is made.
Again, since it is obvious that some classes are empty, we must again augment the Aristotelian system of logic.
Accordingly, we will say that the immediate inference from the falsity of a given I or O proposition to the truth of
its corresponding sub-contrary is not a valid immediate inference, unless the existential assumption is made. Since
that assumption is not normally made, we will say that normally any immediate inference from a given particular
proposition to its sub-contrary is not valid.
When we turn to universal propositions, things are a little more puzzling. Consider the propositions, All Martians
are kind and No Martians are kind. Given that the class of Martians is empty are these two propositions true or
false? Is there any assertion of existence in these two universal propositions? Logicians have agreed that the
answer to this question is no. If you have trouble seeing this, think about the following two propositions: All
the money in my wallet is yours and No money in my wallet is yours. Now suppose, as it is not hard to imagine,
that I have no money in my wallet. In this case, that is, if the subject term here, namely the class of money in my
wallet is empty, then both propositions seem to be true. If I give you all the money in my (empty) wallet and if I
give you none of the money in my (empty) wallet, I will have given you the same amount of money, namely, no
money.
The moral of these considerations is that universal propositions do not assert the existence of members of their
subject classes. As we just demonstrated in the case of the class of money in my wallet when that class is
empty, both an A proposition and its corresponding E proposition can be true. In Aristotles system the A and the
E propositions are contraries, which means that both cannot be true. We now see that the inference from the
truth of a given universal proposition to the falsity of its corresponding contrary is not valid unless we make the
existential assumption. So again, the immediate inference from the truth of a given A or E proposition to the
falsity of its corresponding contrary is not a valid immediate inference, unless the existential assumption is made.
Since that assumption is not normally made, we will say that normally any immediate inference from a given
particular proposition to its contrary is not valid.

77

The immediate inferences from the truth of A and O propositions to the truth of their corresponding subalterns
meets with a similar fate if the existential assumption is not made. It should be obvious that if All Martians are
kind is true because its subject class is empty, it does not follow that its subaltern, Some Martians are kind
would also have to be true. In this case, because the quantifier some asserts that some Martians exist, the
subaltern of the A proposition, its corresponding I proposition, can be false even when the corresponding A
proposition is true. The same holds for the relation between the E proposition and its corresponding subaltern,
the O proposition. Again the moral: the immediate inference from the truth of a given A or E proposition to the
truth of its corresponding subaltern is not a valid immediate inference, unless the existential assumption is made.
Since that assumption is not normally made, we will say that normally any immediate inference from a given
particular proposition to its subaltern is not valid.
So what is left of Aristotles system of immediate inferences once we abandon the existential assumption? With
regard to the inferences involving contraries, sub-contraries, and subalterns, the answer is nothing. But this is
not the whole story. Even if we abandon the existential assumption, some immediate inferences in the
Aristotelian system of categorical logic do remain valid. These are the valid immediate inferences that are
involved in the opposition called contradiction. Given that the proposition No Martians are kind is true because
there are no Martians, it is still the case that the corresponding contradictory I proposition, Some Martians are
kind must be false since it asserts that the class of Martians has members when it does not. The same holds for
the relation between the A and the O propositions. So we again arrive at a moral: the immediate inference from
the truth of a given proposition to the falsity of its corresponding contradictory proposition, or from the falsity of
a given proposition to the truth of its corresponding contradictory proposition, are valid immediate inferences,
even if the existential assumption is not made.
It is almost time to put this material to the test. The exercises below should help you on your way in sorting out
these various immediate inferences and in grasping firmly the traditional square of opposition and the import of
making and not making the existential assumption. For convenience, we will call the square of opposition that
does make the existential assumption "The Traditional Square of Opposition" and the square of opposition that
does not make the existential assumption, "The Modern Square of Opposition." As well, inferences that are valid
only if the existential assumption is made we will say are traditionally valid.

5.6 Other Immediate Inferences

There are three other important kinds of immediate inferences that we need to consider. The first one,
conversion, is the most obvious, for it is quite common to hear such expressions as, All college teachers have
college degrees but not conversely. What this means is that it is false that All people who have college degrees
are college teachers. Notice that second proposition simply turns the S and the P terms around. In general, we
say that the converse of a given proposition is obtained by exchanging the S-term and the P-term. In the case
we just mentioned, the converse of All S is P is All P are S. And this is true of all of the Standard-form
categorical propositions. The converse of the E proposition No S is P is No P are S, the converse of the I
proposition Some S is P is Some P are S, and the converse of the O proposition Some S is not P is Some P
are not S. We can also reverse, or double this procedure, e.g., the converse No S is P is No P are S, and the
converse of No P are S is No S is P. Or as we might put this, the converse of the converse of No S is P is No S
is P. The same operations apply to all of the Standard-form categorical propositions.
78

OK, so far this is easy enough. The slightly more difficult question that arises, however, has to do with which of
these conversions are valid immediate inferences and which are not. But before we say which are and which are
not, let us just pause and make sure we are clear about what an immediate inference is, and especially what a
valid one is.
Since inferences can be valid (or invalid), and since these terms apply only to arguments, it follows that an
immediate inference is an argument. (Did you notice that what we just said was itself an argument, a deductive
one, and valid to boot?) Well, what makes such inferences immediate is that they are not mediated by some
other premise. In general we say that an immediate inference is one in which some given proposition implies
some other corresponding proposition without the intervention of any other proposition. We have seen this
already in, e.g. contrary propositions. That is, from the fact that an A proposition is true, it follows immediately
that its corresponding contrary E proposition must be false. Such an immediate inference is a valid argument form
because it accords with the following general rule that applies to contraries: If a given proposition (either an A or
an E proposition) is true, then its corresponding contrary must be false. (We call such an inference a valid
argument form, since the relation between the two propositions holds regardless of their particular contents.)
With this said, lets now see if we can determine which Standard-form categorical propositions can be validly
converted and which cannot. We have already noted, an A proposition does not validly imply its converse. If we
assume that an A proposition is true, we cannot validly infer that its converse is true. We have already seen that
this is an invalid inference in the example above. Clearly, if it were true that All college teachers have college
degrees is true (and of course even though most do, it is obviously true that some do not), the converse of this
proposition would not have to be true as well. Hence we say that an A proposition does not validly imply its
converse, or if you will, the immediate inference from an A proposition to its converse is invalid.
But quickly, we must qualify this observation. Intuitively, it should be obvious that from the truth of the
proposition, All college teachers have college degrees, it does follow that Some people who have college
degrees are college teachers must also be true. This is a valid immediate inference. But notice that the second
proposition not only reverses the S and P terms of the original A proposition, it also limits the quantity of the
original A proposition, that is, it changes the quantity from universal to existential, from all to some. If we
make this additional change, then the immediate inference from the truth of its limited converse is valid. We can
generalize this as follows: the inference from an A proposition to its limited converse is valid, and without this
limitation, the inference is invalid. But this inference is only traditionally valid (or valid by limitation since the
inference is from a universal proposition (which does assert existence). If the S-class were empty this inference
would not hold.
Converting the other propositions is less qualified. The immediate inference from an E proposition to its converse
is valid, and so is the immediate inference from an I proposition to its converse, but the immediate inference from
an O proposition to its converse is not valid. This should be clear if you simply make some substitutions into our
propositional forms. Clearly if the E proposition No human beings are pigs is true, then it must be true that No
pigs are human beings, and vice versa. Just as obviously, from the truth of the I proposition, Some roses are
flowers it follows that Some flowers are roses, is also, and must be, true. Just as surely, it should be obvious
that if an O proposition, Some cats are not pets is true, it does not follow that Some pets are not cats must
also be true.
79

To summarize once again: The immediate inference to the contrary of an E proposition and to the inference to the
contrary of an I proposition are valid while the inference to the contrary of an O proposition is invalid. With regard
to the A proposition, we say that while the inference to its unqualified contrary is invalid, the inference to its
contrary qualified by limitation is valid.
The second of the three kinds of immediate inference we are looking at is called obversion. This one is easy in this
respect: the inference from any of the four Standard-form categorical propositions to its obverse is valid. What
needs explanation here is the operation for deriving the obverse of a given proposition. As we just saw, the
operation for obtaining the converse of a proposition is simply switching the S and the P terms. The operation for
obtaining the obverse of a given proposition is a little more complicated and requires that we introduce a new
concept, namely, the concept of a complementary class.
We will say that the complement of a given class is the class of everything outside of the given class. The
complement, for example, of the class of cats is the class of everything that is not a cat. But we have to be careful
about how this not is being used. Clearly dogs are not cats, that is, the class of dogs is outside the class of cats.
But the class of dogs is not the complement of the class of cats, since the class of pigs is also outside the class of
cats. To make this clear, we adopt the use of the sometimes grammatically awkward prefix, non. The
complement of the class of cats is accordingly the class of non-cats, dogs, pigs and sealing wax included. As well,
the class of the complement of the class of non-cats is the class of cats, the class of non-non-cats. Just as in the
case of two negatives, two nons or a non-non, cancel each other out.
So now we are ready to say what the operation is for obtaining the obverse of a given proposition. This operation
has two steps. First, change the quality of the given proposition, either from affirmative to negative or vice versa.
That is, the obverse of any given proposition has the opposite quality of the given proposition. For example, the
obverse of the E proposition is an A proposition and vice versa, and the obverse of an I proposition is an O
proposition and vice versa. That is, in making this first step in obtaining the obverse of a given proposition, we
change the quality of the given proposition but leave its quantity intact; if a proposition is universal, its obverse
will still be universal; if its quantity is existential, its obverse remains existential.
The second step in deriving the obverse of a given proposition is to replace the P-term of that given proposition
with its complement. So the obverse of an A proposition All S is P is the E proposition, No S are non-P. It
should be clear that this is a valid immediate inference if you give these propositional forms some content. For
example, it should be clear that the A proposition All roses are flowers is logically equivalent in truth-value to
the E proposition No roses are non-flowers, and vice versa. As well, you should be able to see that Some S is P
logically implies its obverse, Some S is not non-P and that Some S is not P logically implies its obverse, Some S
is non- P.
I need to point out here that it is also true that if a given proposition were false its obverse would also be false. As
we might put this, every Standard-form categorical proposition is logically equivalent to its obverse. Two
propositions are logically equivalent if they have the same truth-values and cannot have different truth-values.
Logically equivalent propositions assert the same thing, and hence one can be substituted for the other without
changing the truth-value of either. Since the immediate inference of the obverse of any given proposition is a
valid inference, we say that any given proposition and its obverse are logically equivalent in truth-value. (This
equivalence relation also applies to the operation of conversion, with the exception of conversion by limitation.
80

For example, No S is P is logically equivalent to its converse, No P are S. This is just another way of saying that
if one is false the other is also false, and if one is true the other is also true. Just as clearly, if Some P are S is true
we do not know that All S is P is also true. Consider this: Some pets are cats is not logically equivalent in truthvalue to the proposition, All cats are pets. In general we have established that E and I propositions are logically
equivalent to their converses; and A, E, I, and O propositions are logically equivalent to their obverses.) When
two propositions are logically equivalent we can substitute one proposition for the other without changing the
truth value or meaning of either proposition.
The last kind of immediate inference we will consider is contraposition. In order to derive what is called the
contrapositive of a given Standard-form categorical proposition, all we have to do is follow two steps. First, we
switch the S-term and the P-term around (that is, the S-term becomes the P-term and the P-term becomes the Sterm.); second, we replace each of these class terms with its complement. (Actually there is no proper order to
these two steps: you can first replace both the S-term and the P-term with their complements and then make the
switch.). For example the contrapositive of the A proposition, All S is P is All non-P are non-S. (Notice that in
deriving this contrapositive we leave the quality and the quantity unaltered.)
As a matter of fact the immediate inference from an A proposition to its contrapositive is valid, as we can see
from the following example: if the proposition All students are wisdom seekers is true (or false), then surely its
corresponding contrapositive, namely, All non-wisdom seekers are non-students would have exactly the same
truth-value. These two propositions are logically equivalent and hence the inference from the truth of one to the
truth of the other, or from the falsity of one to the falsity of the other, is valid.
As well, the inference from an O proposition to its contrapositive is also valid. We can see this clearly from the
following example: from the assumption that Some pets are not cats is true (and as a matter of fact it is true)
surely it follows that Some non-cats are not non-pets must also be true. We have here a valid inference. Again
we leave the quality and the quantity of the derived position the same as they were in the original.
The inference from the I proposition to its contrapositive, however, is not valid. This may not be obvious at first
glance. It may not be clear, for example, that if Some women are firefighters is true, that it is not the case that
its contrapositive Some non-firefighters are non-women must also be true. A different example may help. It
should be clear that if Some human beings are non-professional firefighters it does not follow that Some
professional fire-fighters are non-human beings must also be true; and as a matter of fact it is false.
In order to make it clear that the immediate inference from an I proposition to its contrapositive is not valid, we
resorted to using a proposition with a P-term that is itself a complement term, that is, a term with a non prefix.
Recall that two nons cancel each other out. So following the steps in deriving the contrapositive, we first
replaced the P-term with its complement. But instead of replacing the P-term, namely, non-professional
firefighters with its complement, non-non-professional firefighters we simply replaced it with professional
firefighters. Then we replaced the subject term, human beings with its complement, non-human beings and
then we switched the terms around.
One interesting moral of this is that S and P class terms may be complement class terms. This means that if we
begin with a categorical proposition whose terms are complement classes, we can still make valid inferences from
them. For example, the contraposition of the A proposition All non-S are non-P is the proposition, All P is S.
81

Now what about the immediate inference from an E proposition to its contrapositive? Is this a valid move? It is
not, without an appropriate qualification, or more precisely without what we called above a limitation. Clearly,
No professional firefighters are non-human beings is true, but does it follow that No humans are nonprofessional firefighters must also be true, since it is in fact false. So the immediate inference of contraposition
does not hold for the E proposition. But if we change the quantity of the derived contrapositive from universal to
particular, from No to Some, then the immediate inference is valid, but valid by limitation. If the proposition
"No professional firefighters are non-humans, were true, then surely it follows validly that Some humans are
not professional firefighters is true; and if the first were false the second would be as well.
In summary we have learned the following about the immediate inferences of conversion, obversion, and
contraposition:
1. Conversion: Valid for the E and I propositions, not valid for the O proposition, and valid for the A proposition
only by limitation. Since the E and the I propositions are logically equivalent to their respective converses, either
proposition can be substituted for its converse, and vice versa. The A proposition is not logically equivalent to its
limited converse.
2. Obversion: Valid for the A, E, I, and O propositions. Since all of these propositions are logically equivalent to
their respective obverses, each of them can be substituted for its obverse, and vice versa.
3. Contraposition: Valid for the A and O propositions, not valid for the I proposition, and valid for the E
proposition only by limitation. Since the A and the O propositions are logically equivalent to their respective
contrapositives, either proposition can be substituted for its contrapositive, and vice versa. The E proposition is
not logically equivalent to its limited contrapositive.Conversion:
A: All S is P: Some P are S (limit)
E: No S is P: No P are S
I: Some S is P: Some P are S
O: Some S is not P (not valid)
Obversion:
A: All S is P: No S are non-P
E: No S is P: All S are non-P
I: Some S is P: Some S are not non-P
O: Some S is not P: Some S are non-P
Contraposition:
A: All S is P: All non-P are non-S
E: No S is P: Some non-P are not non- S (limit)
I: Some S is P: (not valid)
O: Some S is not P: Some non-P are not non-S

Conversion

82

Obversion

Contraposition

To repeat: if we do not make the existential assumption, the immediate inferences of conversion, obversion and
contraposition are less affected than in the square of opposition. In the latter, only contradiction survives when
the existential assumption is not made. In the case of the other immediate inferences, there are only two
casualties, the conversion of an A proposition and the contraposition of an E Proposition, both of which require
going from a universal to an existential quantifier to be valid. And if the existential assumption is not made,
neither of these moves from the universal to the existential quantifier is valid. In both cases we are moving from a
proposition that does not assert that its subject class has members, to a proposition that does make this
assertion. Consider this example: the converse (by limitation) of All Martians are bald is Some bald people are
Martians. Certainly the first of these propositions is true because its subject class is empty and the second is false
as a matter of fact. Just as clearly, the contraposition of the E proposition No Martians are bald is true, since its
subject class is empty, while its corresponding contrapositive (by limitation), Some non-bald people are not nonMartians is false.
In summary:

If the existential assumption is not made: only the following immediate inferences remain valid:
contradictions, obversions, and valid conversions and contrapositions that do not involve limitation.
We will say these inferences are valid without qualification.

If the existential assumption is made (Aristotle made it), then the immediate inferences that do involve
limitation are valid, and the inferences of the square of opposition other than contradiction (which is
valid without qualification) are also valid. We will say that these inferences are traditionally valid.

Some immediate inferences are also invalid without qualification, namely, conversions of O
propositions and contrapositions of I propositions.

Well this is all quite a lot. So lets see if we can put what we have learned into practice. Its time for the exercises
to begin.

83

Study Guide for Chapter 5

Standard-Form Categorical Propositions:
E Proposition: No S is P Universal Quantity; Negative Quality
A Proposition: All S is P Universal Quantity; Affirmative Quality
I Proposition: Some S is P Particular Quantity; Affirmative Quality
O Proposition: Some S is not P Particular Quantity; Negative Quality

Distribution Table
A: S-term distributed; P-term undistributed (DU)
E: both terms distributed (DD)
I: both terms undistributed (UU)
O: S-term undistributed; P-term distributed (UD)

Conversion:
A: All S is P: Some P are S (limit)
E: No S is P: No P are S
I: Some S is P: Some P are S
O: Some S is not P (not valid)
Obversion:
A: All S is P: No S are non-P
E: No S is P: All S are non-P
I: Some S is P: Some S are not non-P
O: Some S is not P: Some S are non-P
Contraposition:
A: All S is P: All non-P are non-S
E: No S is P: Some non-P are not non- S (limit)
I: Some S is P: (not valid)
O: Some S is not P: Some non-P are not non-S
A

Conversion

Obversion

Contraposition

84