4215 - An Introduction To Formal Logic

Topic Subtopic
Philosophy Modern Philosophy
An Introduction
to Formal Logic
Course Guidebook
Dr. Steven Gimbel
Gettysburg College
PUBLISHED BY:
THE GREAT COURSES

Corporate Headquarters
4840 Westfields Boulevard, Suite 500
Chantilly, Virginia 20151-2299
Phone: 1-800-832-2412
Fax: 703-378-3819
www.thegreatcourses.com
Copyright The Teaching Company, 2016
Printed in the United States of America
This book is in copyright. All rights reserved.
Without limiting the rights under copyright reserved above,

no part of this publication may be reproduced, stored in
or introduced into a retrieval system, or transmitted,
in any form, or by any means
(electronic, mechanical, photocopying, recording, or otherwise),
without the prior written permission of
The Teaching Company.
Steven Gimbel,
Ph.D.
Professor of Philosophy
Gettysburg College
D r. Steven Gimbel received his

Ph.D. from Johns Hopkins
University before joining the faculty
at Gettysburg College, where he is a Professor of Philosophy. At
Gettysburg, he has received the Luther W. and Bernice L. Thompson
Distinguished Teaching Award and was named to the Edwin T.
and Cynthia Shearer Johnson Distinguished Teaching Chair in the
Humanities. He also serves as Chair of the Philosophy Department.
Dr. Gimbels research focuses on the philosophy of science, exploring

the nature of scientific reasoning and the ways in which science and
culture interact. He has published numerous articles and four books:
Defending Einstein: Hans Reichenbachs Writings on Space, Time
and Motion; Exploring the Scientific Method: Cases and Questions;
Einsteins Jewish Science: Physics at the Intersection of Politics and
Religion; and Einstein: His Space and Times.
Dr. Gimbels previous Great Course is Redefining Reality: The

Intellectual Implications of Modern Science.
Table of Contents
Introduction
Professor Biography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i
Course Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Lecture Guides
Lecture 1
Why Study Logic? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Lecture 2
Introduction to Logical Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Lecture 3
Informal Logic and Fallacies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
Lecture 4
Fallacies of Faulty Authority . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Lecture 5
Fallacies of Cause and Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Lecture 6
Fallacies of Irrelevance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Lecture 7
Inductive Reasoning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Lecture 8
Induction in Polls and Science . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Lecture 9
Introduction to Formal Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Lecture 10
Truth-Functional Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
Lecture 11
Truth Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
Lecture 12
Truth Tables and Validity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Lecture 13
Natural Deduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Lecture 14
Logical Proofs with Equivalences . . . . . . . . . . . . . . . . . . . . . . . . . 133
Lecture 15
Conditional and Indirect Proofs . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
Lecture 16
First-Order Predicate Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
Lecture 17
Validity in First-Order Predicate Logic . . . . . . . . . . . . . . . . . . . . . . 162
Lecture 18
Demonstrating Invalidity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
Lecture 19
Relational Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
Table of Contents iii

Lecture 20
Introducing Logical Identity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
Lecture 21
Logic and Mathematics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
Lecture 22
Proof and Paradox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
Lecture 23
Modal Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
Lecture 24
Three-Valued and Fuzzy Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
Supplemental Material
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
Image Credits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
iv Table of Contents
Course Scope
An Introduction to
Formal Logic
L ogic is the study of rational argumentation. A belief is rational if

we have good reason to believe that it is true. An argument is a
set of sentences such that one sentence, the conclusion, is claimed
to follow from the other sentences, the premises. That means that
logic is the study of the support we need to have good reason to
believe the things we believe.
We are not wired to be logical. All kinds of bad reasons for believing
things seem to us like perfectly good reasons. But the situation is not
hopeless. It turns out that logic is a skill that can be learned. We can
be trained to spot the reasoning errors, look for the proper kind of
support, and analyze arguments we hear from others to assess their
strength and weaknesses.
Well begin by examining the human mind and seeing the ways in
which logic is and is not a natural part of the way we think. Well look
at some of our cognitive biases, ways in which social psychologists
have demonstrated that the brain naturally works against good
inferences. Humans can be rational beings, but it takes work to
realize the pitfalls we need to avoid.
Then, well introduce a wide range of logical concepts. We will

rigorously introduce the notion of an argument and examine both the
types of argumentsdeductive and inductiveand the criteria by
which we assess an argumentvalidity and well-groundedness. We
will learn that arguments have two parts: conclusions (that which is
being argued for) and premises (the support given for the conclusion).
Next, well focus on informal logicthat is, considerations of well-
groundedness, the criterion of assessment that considers the truth
of an arguments premises. Well learn to spot common fallacies,
reasoning errors that sound good to the ear but that actually
undermine the support for the conclusion.
Inductive arguments are those that begin with observation and lead
to broader, generalized conclusions. Well study the basis for strong
inductive arguments and the ways in which inductive arguments can
go wrong, and then well examine the use of inductive arguments in
science and polls.
Formal deductive logic is the part of logic concerned with the forms
of deductive arguments. An argument is deductive if the content
of the conclusion is contained in the content of the premises. If a
deductive argument has good form, regardless of the truth or falsity
of its premises, we say that the argument is valid.
To handle different types of sentences, we will develop two logical

languages. The first, truth-functional logic, looks at the logical
behavior of truth-functional connectiveswords such as not, and,
or, and ifwhich produce sentences that we can determine
the truth or falsity of simply by knowing the truth or falsity of the
sentences combined. We will learn how to determine whether or
not such arguments are valid using both truth tables and natural
deduction proofs.
The second logical language is first-order predicate logic, which

builds on top of the elements of truth-functional logic in a way that
allows us to account for the logical content within sentences as well
as between them. In this way, a more powerful logical system can
be constructed that can handle everything in truth-functional logic,
everything in Aristotles logic, and more.
The pinnacle of first-order predicate logic for us will be the ability to

account for sentences that contain quantities. This means that we can
2 Course Scope
begin to account for mathematical truths in our logical system. The
possibility of giving logical justifications for mathematical propositions
is a major element in the history of the development of logic. Well
discuss the rise and fall of logicism, the view that logic provides the
ultimate foundation for all of mathematics.
Well examine modal logic, which deals with the concepts of

possibility and necessity, distinguishing between those sentences
that just happen to be true and those sentences that must be true.
We will see how modal logic was reinterpreted to develop a logic for
ethical claims.
Finally, well explore three-valued logic and fuzzy logic. The basis for
all of the logical considerations is that all sentences must have one
of two truth values: true and false. But what if we include a third truth
valueor even a continuum, ranging from definitely true to definitely
false, possibly taking any value in between?
In this course, well see that logic is a tool for helping us think in a
clearer, more rigorous fashion. Logic allows us to distinguish between
proper and improper forms of reasoning but never leads us to
conclusions on its own. The content of our arguments come from our
experiences, beliefs, and hypotheses. Logic does not control life, but
it is an important part of a well-lived life.
Course Scope 3
Lecture 1
Why Study Logic?
S ince the times of classical Greece, the definition of human

being has hinged on the notion of people being rational. Our
brain is what sets us apart from the rest of the animal kingdom, and
the nature of that brain is reason. The claim has been that we are built
for logic. But contemporary psychology and sociology have shown
otherwise. We have built in cognitive biases, and we have internalized
social facts from our cultural background, and these both often
lead us to unwarranted conclusions. We cannot rely on our natural
instincts, but must train ourselves in the way of rational thought.
Innate Human Irrationality

We have arranged our entire civil society around the idea that we
are logical beings naturally capable of rational thought. But the
world is complex. In the same way that there are optical illusions,
images that will trick our senses into thinking they perceive
something they do not, there are cognitive illusionsthat is, ways
of thinking that convince us that we are right about something we
believe when we are, in fact, wrong.
Contrary to our beliefs, we are not wired to be completely rational.

Indeed, there are multiple factors that lead us to be irrational.
Some of these are psychological, some are sociological, and
some are cultural.
But understanding that they are out there is the first step to
realizing that we need to think carefully about thinking carefully
and to teach ourselves how to think in ways that are the most
likely to avoid errors.
One of the more well-known examples of being led to irrational
belief comes from psychologist Solomon Aschs experiments in
the early 1950s on conformity. The subject of the experiment would
sit at a table with several other people, who he or she was told
were also participating in the experiment but unbeknownst to the
subject were really confederates working with the experimenter.
It was explained that this was a

study in perception and that they
would be shown a series of charts
with line segments of different length
(labeled A, B, and C) and that they
just needed to say which one was
the same length as a fourth line. Reference A B C
The table was arranged so that the subject would answer last,
after the others said aloud which line they thought was of the
same size as the comparison line. The first chart was shown, and
all of the confederates answered correctly, and so did the test
subject. The same thing happened with the next chart.
Then, on the third try, the confederates all gave what was
obviously the wrong answer. C was the same size as the
comparison line, but they all responded that A was. When it got
to the test subjects turn, 12 out of 18 times in the original study
(and it has been reconfirmed many times since), the test subject
gave the wrong answer so as not to stick out.
Not conforming with the group is something we fear, and we

do things that we know are wrong in order to get along with the
majority. What Asch shows is interesting: that we will act in such a
way as to do something contrary to our reasonable beliefs.
Interviewed after the experiment, most subjects said that they

knew the answer was wrong, but they didnt want to embarrass
themselves or mess up the researchers data by being different.
Lecture 1Why Study Logic? 5

More interestingly, others said that because the task was so easy
and because the others answered so quickly and confidently,
they started to doubt themselves.
Aschs study was about what people would dowhether they

would be willing to act in a way that they knew was wrong. But
what came out was that, in some cases, it was not just about their
actions, but also about their thoughts, about the way in which
humans will change what we believe in order to fit in with those in
our environment.
This is what psychologist Irving Janis called groupthink, which

occurs when we sacrifice critical capacities of thought in order to
adopt a consensus belief. All other things being equal, we would
not have come to hold this belief, but because we are in a group
that believes it, we do, too. It is not just that we act as if we believe
itwe do believe it.
Janis, Asch, and a few generations of researchers following them

have documented all kinds of factors that amplify or diminish
groupthink. If those around you are unified in expressing the
belief and do so passionately, you are more likely to get swept
up in it.
But if even one person expresses skepticism, you are more likely
to be skeptical yourself.
If there is a strong, charismatic leader espousing the view, people

are more likely to come to believe it.
If there are rewards for believing and punishments for disbelieving,

people will respond.
Our beliefs are affected by the beliefs, actions, and personalities

of those around us, even if they violate what we otherwise know
to be good reasoning.
6 An Introduction to Formal Logic

Documenting this has become one task of social psychology,
and these researchers have documented many ways that we
naturally reason badly, called cognitive biases.
Many cognitive biases have been discovered and replicated in

multiple experiments. Some involve behaviorseffects such as
hyperbolic discounting, where we will irrationally choose less of a
reward now instead of delaying gratification for more of a reward
later, even if the greater reward is certain to be granted.
There is also what has come to be known as irrational escalation

effect, where we will judge an action less risky than we know it is
because we have already invested in it.
Humans are not purely rational beings. Our ultimate interest in

cognitive biases involves the ways in which we are innately wired
to acquire false beliefs.
Rational Actions and Thoughts

We know that we act contrary to what we know to be rational, but
just as interestingor even more soare the ways in which we
are regularly wrong about what is rational.
Researchers Rolf Reber and Norbert Schwarz have shown that

the font a sentence is written in will have a measurable effect on
how likely we are to think it is true. The clearer the words are
visually, the more stock we put in their truth.
Psychologists define the halo effect as follows: If someone is

successful or has a virtue in one wayfor example, the person
is successful in businessthen we will take that as evidence that
the person is an authority in whatever he or she does or says.
Furthermore, if someone we think is attractive tells us something
and someone unattractive tells us the same thing, we are much

more likely to believe it when told to us by the good-looking
person, even with the same or less evidence.
We have a positive outcome biasthat is, we think that we are

more likely to succeed at tasks, win at games, or have good things
happen to us than is actually the case. We imagine ourselves
after the positive outcome, and because we can envision it so
clearly, we come to think that it is more probable than it is.
We overestimate our abilities. The overconfidence effect is the

irrational degree of belief we put in our own choices, even when
we know they are random. Interestingly, this effect is amplified
when the person is the most ignorant. We are most overconfident
about our abilities to do that which we know the least about.
The Dunning-Kruger effect is the name given to the fact that

the less people know about an area or how to do something,
the more likely they are to overestimate their ability to do it or
understand it. The more ignorant we are, the more brazen we are
in our belief about our abilities and knowledge.
In the early 1980s, psychologist Benjamin Libet ran a series of

experiments that showed that the parts of our brain that determine
our actions are sometimes triggered before the parts of our brain
that make decisions. We act first, and then we use our brain to
concoct a story that justifies our belief that we acted because we
consciously and rationally decided to, even if that justification has
nothing to do with the real reason for the action.
We find this kind of backfilling a justification behavior on different

cognitive levels. We suffer greatly from what social psychologists
call confirmation biasthat is, when we hold a particular belief,
we search out that which we believe supports the belief and
explain away or outright ignore that which undermines rational
belief in the proposition.

When faced with a small amount of supporting evidence and a
huge amount of disconfirming or falsifying evidence, we focus on
that which backs us up and use it to swamp the overwhelming
evidence against usespecially when the belief is core to our
worldview. We see this often in the world of politics.
We will do whatever we need to in order to save our preexisting

beliefs. Where do these come from? That is the purview of
sociology. We are acculturated into a belief system, and the
worldview it gives usthe basic categories and presuppositions
that come with itis notoriously difficult to change.
Emile Durkheim, one of the founding fathers of sociology,

discussed what he called social facts, which we acquire from
being part of a society. Social facts are ways of thinking or acting
that originate outside the individual, are enforced by the society,
and become a part of the individual.
These are beliefs that become invisible to us because they are

the lens through which we see the world, and questioning their
truth strikes us as absurd, if not dangerous.
Where are our culturally generated blind spots? What do we

believe is necessarily true that is, in fact, completely false? We
dont know, but figuring it out is the job of philosophers and
cultural critics, and among the most important tools they employ
is logic.
Logic and Reasoning

We are not rational animals by nature, but that does not mean
that we cannot be or should not be. Cognitive biases from our
psychology and social facts from our cultural upbringing and
political context might give us false beliefs, but we are capable of
excavating and rigorously analyzing them.

We can question what we believe, even our most deeply held
beliefs, and objectively determine whether there are legitimate
grounds for rational belief. For this, we need logic.
It might turn out that there is good reason to believe what we

believe. In this case, our belief will be strengthened. If it turns out
that there is good reason to disbelieve it, then we will have saved
ourselves from having a false belief, which could not only affect
other beliefs, but also cause us to act in ways that have negative
consequences for us and those around us.
When we are aware of our psychological and cultural biases, we

are able to step away from them and begin to think rationally. But
it does not happen automatically. We might be aware of some
cognitive biases and still fall into them.
Logical thinking is not an innate talent, but it is a skill that one can
develop. It takes training and practice. You must see what makes
some forms of reasoning effective at determining what is likely to
be true and what makes other forms of reasoning likely to lead
you into error, even though it seems attractive to us.
Readings
Durkheim, The Rules of Sociological Method.

Fine, A Mind of Its Own.
Sutherland, Irrationality.

Questions
1.
If the human brain is the result of evolution, which selects for
properties that have an advantage in survival, then why do we have
these cognitive biases built into us?
2.
How do we best go about finding the cultural blind spots that we
have? If there are assumptions we have been given our whole lives,
how do we find them?

Lecture 2
Introduction to
Logical Concepts
T he word logic means rational argumentation. A belief is rational

if we have good reason to believe it, and an argument is a set
of sentences such that one sentence, the conclusion, is claimed to
follow from the other sentences, the premises. The first skill we have
to master is finding the parts of an argumentthat is, recognizing
what is the conclusion and what are the premises. Then, we need
to figure out what type of argument we have: deductive or inductive.
Finally, we can begin to analyze the argument. For this, there are two
criteria that have to be satisfied: validity and well-groundedness.
The Definition of Logic

For the purposes of this course, logic is the study of rational
argumentation. By rational, we mean that which we have
grounds to show is likely true. A belief is rational if we have good
reason to believe it is at least probably the case in reality.
The other word, argumentation, is the important one. The central

term in all of logic is argument. For logicians, an argument is
a set of sentences such that one sentence, the conclusion, is
claimed to follow from the other sentences, the premises.
Arguments have two parts: a conclusion and premises. The

conclusion is the point of the argument. It is the thing being
argued forthat which we are trying to convince ourselves
or others of. We give arguments in order to provide legitimate
reasons to believe the conclusion. The premises are those
reasons. The premises are the grounds that are being proposed
to support rational belief in the conclusion.
Every argument has one and only one conclusion. Premises, on

the other hand, can come in any number. There are mathematical
arguments with an infinite number of premises and weird logical
arguments that have no premises.
Consider an example: All men are mortal. Socrates is a man;

therefore, Socrates is mortal. In this case, the conclusion is
Socrates is mortal. That is what the argument is trying to
convince us of. Why should we believe that Socrates is mortal?
Because he is a man, and all men are mortal.
The first thing you do when you approach an argument is to

find the conclusion and then set out the premises. It is crucially
important that you do this correctly.
Consider what would happen if we misidentified the conclusion

in this case. If it is true that all men are mortal and that Socrates
is one of these men, then it turns out that it is absolutely the case
that Socrates is mortal. But suppose we wrongly thought that
the conclusion is all men are mortal and that the premises are
Socrates is a man and Socrates is mortal. Just because one
man is mortal, it doesnt necessary follow that they all are.
By misidentifying the conclusion and premises, we have taken

a good argument, one that gives us good reason to believe
something, and turned it into a flawed argument that does not.
So, the first tasks that are necessary for us to develop are
figuring out when we have an argument and determining what
the conclusion is and what the premises are. We often have help
with these taskscalled indicator words. There are certain words
we use to point out conclusions, and there are certain words we
use to point out premises.
Lecture 2Introduction to Logical Concepts 13

A conclusion is supposed to be the thing that is established
by the argument, so we use words that indicate this. The most
obvious one is therefore. My heart is beating; therefore, I am
alive. We use other words for this function as well, including
thus, hence, so, and the Latin ergo. But be careful: Not
every use of these words indicates a conclusion.
We also use words and phrases that indicate premises, such

as because, since, and given that. Again, not every use of
these words is an indication of a premise in an argument.
Indicator words are the easiest way to determine whether we

have an argument and, if so, what the conclusion is and what the
premises are. But we dont always have indicator words. How,
then, do we determine if we are looking at an argument?
The easiest way is to try to insert your own indicator words

for example, therefore or because. If you look at a passage
and the word therefore can be naturally inserted in a way that
maintains the meaning of the passage, you are probably looking
at an argument, and what immediately follows therefore is your
conclusion.
Similarly, if you can insert because into a passage without

changing the meaning, you are likely looking at an argument,
and what comes right after your inserted because is probably
a premise.
Types of Arguments
The first two steps are to figure out if a passage contains an
argument and then to identify the structurethat is, to pick out
its conclusion and fully lay out its premises (both stated and
implied). The third step is to determine what kind of argument it is.

Arguments come in two kinds: deductive and inductive. An
argument is deductive if its conclusion is no broader than its
premisesthat is, if the conclusion only refers to that which
is mentioned in the premises. The technical term we use is to
say that deductive inferences are non-ampliativethat is, the
conclusion is contained within the content of the premises.
Reconsider the following argument: All men are mortal. Socrates

is a man. Therefore, Socrates is mortal. In the premises, we
have all men being mortal, but then move in the conclusion
to just one of these men being mortal. Deduction moves from
broad to narrow.
Inductive arguments are

ampliativethat is, they do
have a conclusion that is
broader than the premises.
For example, the Scottish
philosopher David Hume
famously considers the
following argument: The
sun has risen every other
morning, so the sun will rise
tomorrow as well.
Notice that the premises

contain a lot of data: that the
sun has risen every day in the
past. But the conclusion is
not about one of those days;
it is about a day outside of the
data set. We are using facts
about millions of other days to David Hume
expand our reasonable belief (17111776)
to one new one.

In the case of the Socrates being mortal argument, we had a
claim about all men and Socrates was one of them, but here we
have a collection of data and the day in the conclusion is not one
of them. It is a further instance not contained in the content of the
premises.
With deduction, we are arguing from broad to narrow, but with

induction, we are arguing from narrow to broad.
The important result of this difference is in the degree of belief

we can reasonably put in the conclusions of these different types
of arguments. Because the content of a deductive argument is
already contained in the premises, then in a successful deductive
argument, the conclusion will be absolutely certain. If all men are
indeed mortal and Socrates is one of those men, then there is no
other alternative but that Socrates is mortal.
But with an inductive argument, because the conclusion outruns

the premises, we have no guarantees. Just because the sun
has always risen, there is no guarantee that it will do so again. It
probably will, but not definitely. Successful inductive arguments,
because they are ampliative, only give us high probability, not
absolute certainty.
With an inductive argument, because the conclusion lies outside

the scope of the premises, there is a risk that even a very good
inductive argument might have a false conclusion. This is not to
say that we should not believe it. We should believe that which is
probably true. It is wonderful to achieve the absolute certainty we
get from deduction, but in most real-life cases, we are restricted
to the high probability of induction.
It will turn out that when we evaluate argumentswhen we

determine whether an argument is successful or not, whether it
has flaws or notwe will need different logical tools for deductive
and inductive arguments.

Criteria for Acceptable Arguments
Identifying the parts of arguments and the types of arguments
are the initial steps, but the whole point is to determine which
ones are good arguments. Which ones give you good reason to
believe in the truth of the conclusion?
To evaluate arguments, we use two criteria: validity and well-

groundedness. An argument is valid if and only if, assuming the
truth of the premises for the sake of argument, the conclusion
follows from them.
The important thing to notice here is that we are assuming the

premises are true for the sake of argument. Maybe they are true;
maybe they are false. We dont care. Validity does not concern
the content of the premises.
All we are looking at is whether the premises, if true, would lead

you to the conclusion. Validity is not about the content of the
argument, but about the form of the argument. Validity looks at
the skeleton of the argument and determines if it is strong enough
to support the weight of the conclusion.
That we are assuming the truth of the premises in the first

criterionvalidityshould bother you a little bit. After all, that is a
huge assumption to make. What justifies our ability to make such
an assumption is the second criterion: well-groundedness.
An argument is well-grounded if and only if all of its premises are

true. Well-grounded arguments have true premises. Maybe the
conclusion is true, or maybe it is false, but what is important for
us in looking at the well-groundedness of an argument is just the
truth or falsity of the premises.

An argument that satisfies both of the criteriaan argument that
is both valid and well-groundedis considered to be sound. A
sound argument gives us good reason to believe its conclusion.
In order to determine which arguments are sound, we need to

develop tests for validity and well-groundedness. Validity looks
at the structural elements of the argument, and its study is
called formal logic, because it is an examination of the form of
arguments. Validity for deductive and inductive arguments are
completely different matters, and we need different tools.
Well-groundedness concerns look at the acceptability of the

argument other than the form and are called informal logic. We
will begin there as we work to build a complete account of the
ways we determine what arguments give us good reason to
believe their conclusions.
Readings
Barker, The Elements of Logic, chap. 1.

Copi, Introduction to Logic, chap. 1.
Kahane, Logic and Philosophy, chap. 1.
Questions
1.
Which of the following p
assages contains an argument?
a Groundhogs burrow in the winter to protect themselves from

the cold. They acquire a layer of fat through the summer
and fall months, which allows them to stay warm and nourish
themselves as they hibernate. They flourish in areas where
food is plentiful, making it easier for them to survive the winter.

b The groundhog population here is likely to die off in the near
future because residential construction is eliminating easy
access to food. Without the food, they will not have sufficient
fat for them to make it through the winter, and the population
will decrease until it can no longer sustain itself.
2.
Find the conclusion and premises in the following arguments.
a You should buy the most expensive model of this computer.

I know that it is a lot of money, but the speed and increased
functionality are worth it. If you skimp on the model, then it will
not be that much better than your current computer, and that
would make it an even bigger waste of money.
b I dont like Indian food. You cant eat the gluten that is in most
Italian food. The Persian restaurant is very expensive. The only
other restaurant on this block is the Irish pub, which we both
like and is affordable. Lets eat there.
3.
Are the following arguments deductive or inductive?
a If you dont water your plants, they will die. If we dont drink
water, we will die. If you take a fish out of water, it will die. It
seems that it is true of everything living that it needs water.
b That recipe calls for a tablespoon of sugar. It says that it serves

12. Theres only four of us, so if I scale back the recipe, Ill only
use a third of a tablespoon of sugar. Thats just a teaspoon.

Answers
1.
a his is not an argument; its simply a collection of facts about
T
groundhogs.
b This is an argument, because it is providing reasons why one

should believe that the groundhog population in a specific
region will die off.
2.
a remise: The most expensive model of computer has increased
P
speed and functionality, which make it worth the extra money.
Premise: The less expensive model is not much better than

your current computer, so buying it would be a waste of money.
Conclusion: You should buy the most expensive model of

computer.
b Premise: I dont like Indian food.
Premise: You cannot eat at the Italian restaurant.
Premise: The Persian restaurant is very expensive.
Premise: The Irish pub is the only other restaurant on the block
and is affordable.
Conclusion: We should eat at the Irish pub.

3.
a his is inductive, because it generalizes from plants, humans,
T
and fish to all living things.
b This is deductive, because four is one-third of 12, and one-

third of a tablespoon is a teaspoon, so it necessarily follows
that the scaled-back version of the recipe will call for a single
teaspoon of sugar.

Lecture 3
Informal Logic and Fallacies
A n argument is well-grounded if all of its premises are true. In

assessing the well-groundedness of arguments, we need
to be on the lookout for a number of standard reasoning errors,
called fallacies. One category of fallacy concerns what logicians
call begging the question. These are unwarranted assumptions that
depend on accepting the conclusion, which is a problem because the
point of the argument is to provide independent warrant for belief in
the conclusion. We can commit this error by creating various fallacies,
such as circular argument, begging the question, equivocation, and
distinction without a difference.
Fallacies as Reasoning Errors

There are common reasoning errors that create unsound
arguments but that are nonetheless attractive to our minds.
They sound like good arguments, but in fact they are unsound
arguments. These reasoning errors are called fallacies. A fallacy
is an identifiable category of argument that does not support its
conclusion.
One of the most well-known fallacies is circular argument. A

circular argument is one in which the conclusion is identical to
the premise. Consider the following argument: My name is Steve;
therefore, my name is Steve. Is it valid? If we assume the truth of
the premise, are we led to also accept the truth of the conclusion?
Yes. It is a perfectly valid argument. But is it a good argument?

Does the premise give us independent warrant for rational belief
in the conclusion? No.
The point of an argument is to provide independent support for
the conclusion because the conclusion is in doubt. But if the
conclusion is in doubt, then so is the premise, because they are
one and the same proposition.
If we do not have independent reason to think that the premise

is true, then the well-groundedness of the argument is in doubt,
and the argument cannot be said to be sound, and we do not
have good reason to believe the conclusion. It is a bad argument
logically.
Unfortunately, it is an effective argument psychologically. It works

well, if done right. If we say something clearly, slowly, loudly,
confidently, or forcefully enough, people will believe it. Repeat it
and it seems more likely to be the case. Circular argument is a
fallacy, but it is effective.
Does the fact that it is logically flawed mean we have reason to

reject the conclusion of a circular argument? If we have a valid,
well-grounded argument, then we have good reason to accept
the likely truth of the conclusion. But if we find a flaw, a fallacy
committed in the argumentif the argument is invalid or not well-
groundeddo we have reason to think it is likely false? No.
Just because someone makes a bad argument for a conclusion

does not give us rational justification for thinking that the
conclusion is not the case. Remember that an argument provides
us good reason to believe in the likely truth of the conclusion. If
an argument fails, then it means that those specific premises do
not give us reason to believe that conclusion.
But it doesnt mean that there can be no other set of premises

that does. There could be a good argument for that conclusion,
and this one just isnt it. Every true proposition can be made the
conclusion of a terrible argument.
Lecture 3Informal Logic and Fallacies 23

If we find a fallacyif an argument we are analyzing turns out
to be flawedwhat do we know about the truth or falsity of the
conclusion? Nothing. Might it be true? Yes, for other reasons.
Might it be false? Maybe. So, what do we know about it? Nothing.
Are we justified in rejecting it? No. We have to suspend belief.
Keep in mind that a lack of presentation of a good argument is
not a refutation.
Circular Argument
An argument is circular when the premise and the conclusion are
the same proposition. Notice the use of the term proposition
here and not the term sentence. In philosophy, a sentence
is a grammatically correct string of words. A proposition is the
content of the sentence.
Different sentences can be used to express the same proposition.

The easier-to-spot version of circular argument is where the
premise and the conclusion are the same sentence, as in the
example, My name is Steve because my name is Steve.
But more often, the premise and conclusion will be different

sentences that express the same proposition, thereby
camouflaging the fact that the argument is circular. For example,
My name is Steve because Steve is what I am called.
While this example might seem contrived, we find examples often

in real life. One place you will frequently come across the easier-
to-identify version is when you question a deeply held belief. We
will often feel the need to justify our more peripheral beliefs, but
certain things we take as foundational and either beyond the
need for evidence or so central to the way we see the world that
we never thought to question them.

Why are people of your class deserving of benefits not given to
those beneath? Because we are. Why is gold more valuable than
silver? Because it is. These circular arguments will often be given
with a sense of disbelief that anyone could even think to request
justification for the belief.
Frequently, the premise and the conclusion will be different

sentences expressing the same proposition. Because they
sound so different to the ear, the fact that they say the same thing
is overlooked, and the argument is wrongly thought to be sound.
Consider a nave version of a standard ethical argument for

vegetarianism that we call the argument from sentience: It is wrong
to kill animals because it is wrong to kill anything that can feel pain.
What makes this circular is that the only things that can feel pain
are animals. To feel pain, you need a central nervous system. But
anything that has a central nervous system would be an animal.
So, the argument really just says that it is wrong to kill animals
because it is wrong to kill animals. It restates the conclusion as
the premise in a fashion that is not obvious at first glance.
Begging the Question

Circular arguments are one instance of a larger class of fallacies
called begging the question. This is a phrase that is often misused.
One will frequently hear someone say, That begs the question
that when they mean, That leads one to ask the question.
But what begging the question really means is arguing unfairly in

a way that tries to use the conclusion in support of itself. Anytime
we try to get the conclusion to pull itself up by its own logical
bootstraps instead of giving independent reasons for belief,
getting its support from propositions distinct from itself, we are
looking at begging the question.

What is a question? Is it anything of a certain grammatical form
where our voice goes up at the end? No. There are non-questions
that look like questions and sound like questions but are not
questions. A question is a request for information. A pseudo-
question, on the other hand, seems like a request for information
but really isnt. For example, Youre not going wear that, are you?
A so-called leading question is a question that is not fairly asked

to elicit an honest response from the listener, but rather a sentence
that looks like a question but is designed to lead the listener to a
particular desired response. Does this make my butt look big? is
a famous leading question that one might receive from a significant
other. It is not a question; there is only one correct answer.
So, one way to beg the question is to use questions that are not
questions. Another way is to use the connotative power of language
to frame questions unfairly. Words have both denotationthat is,
they pick out certain thingsand connotationthat is, they lend
an emotional weight to those things.
Companies vying for our business do this all the time: Looking
to relocate to another part of the country, but you dont have a
lot of money? Try Budget Movers. Are they cheaper than other
companies? We have no evidence, but the name implies it. The
words are chosen because of their ability to sway the customer.
Equivocation
The basis of the fallacy known as equivocation is ambiguity.
Words can mean more than one thing. This is not a logical
problem, just a feature of language. The fallacy of equivocation
occurs when we change the meaning of a word in the middle of
an argument. Consider the following argument.

Tables are furniture.
My statistics book has tables in it.
Therefore, there is furniture in my statistics book.
It certainly looks valid, doesnt it? And all of the premises are true.
So, it must be sound, right? We have good reason to believe the
conclusionexcept, of course, that the conclusion is absurd.
Whats the flaw?
The meaning of the word table has changed from a raised

flat surface on which to place things to a rectangular array of
numbers. There is nothing inherently wrong with the ambiguity as
long as the meaning of operative ambiguous terms is maintained
throughout the argument, but in this case, the meaning has
changed, and the result is the absurdity of thinking that there is
furniture inside of a book.
Unfortunately, not all examples of equivocation are this clear-cut.

The usual cases are trickier because many ambiguous terms will
have distinct meanings that are related. As such, the shift from
one to the other is subtler and, as a result, easier to overlook.
Consider the following example.
We have a right to vote.

One should always do what is right.
Therefore, one should always vote.
In this case, the word right is being equivocated upon. In one

case, we have one sense of right, meaning a legally protected
action, and in the second premise, the same word means an act
that is morally necessary. These are two different meanings, but
ones that can be confused if you are not careful.

Distinction without a Difference
A distinction is a linguistic separation of two concepts that are
different. Distinction without a difference is a fallacy that occurs
when we try to draw a distinction between two things that are not, in
fact, distinct. I didnt steal it; I just didnt ask before I borrowed it.
One place we will often find people trying to draw distinctions

without a difference is when they are trying to justify holding
views they know they should not.
We will often try to justify our misdeeds or our problematic views

by trying to distance them from the categories to which we know
they really belong. Im not racist; I just think that members of that
particular minority are less intelligent.
Notice the difference between distinction without a difference and

circular argument: In the case of distinction without a difference,
the form is Its not A; its A, whereas in a circular argument, we
are saying, A because A.
In distinction without a difference, we are trying to say that A is

both A and not A, whereas in a circular argument, A certainly is
A, and we should believe A because of itself. Both repeat A, but
in making very different logical mistakes.
Readings

Damer, Attacking Faulty Reasoning, chap. 5.
Kahane, Logic and Contemporary Rhetoric, chap. 3.

Questions
Identify the fallacies in the following passages from the following

list: circular argument, question-begging language, equivocation,
distinction without a difference.
1.
That dish cant be too spicy; after all, she said she made it with chili
peppers. Chilly is the opposite of hot, and she didnt say she used
hot peppers.
2.
Do you want to go to the same old boring beach this year for vacation,
or should we be adventurous and try something new, such as the
mountains?
3.
Its not that I dont care about you; Im just not concerned with your
life choices.
4.
Fried foods are bad for you because they are not part of a healthy
diet.

Answers
1.
equivocation
2.
question-begging language
3.
distinction without a difference
4.
circular argument

Lecture 4
Fallacies of Faulty Authority
N o one knows everything. In many situations, we have to rely on

others to learn what we need, or we have to infer something from
other things we know. These are both fine ways of acquiring rational
beliefs, if done properly. In this lecture, you will continue learning
about ways that arguments can go wrong by examining five new
reasoning errors that are associated with flawed appeals to authority
and flawed inferences.
Appeal to Authority
When we want to know something we dont know, it is perfectly
rational to ask someone who does. We call this an appeal to
authority, and arguing from authority is a legitimate means of
reasoning. We do it all the time, and we should. When you are
sick, hopefully it is your physician you seek out, because he or
she is the expert you need.
What makes someone a legitimate authority, someone whose

answer to a question we would have good reason to believe?
There are three requirements. First, the purported authority must
have actual physical presence in the material world. The authority
must exist.
Surprisingly, we hear arguments from authority that violate

this simple criterion all the time. Someone who says, I heard
somewhere that or I read somewhere that is making an
argument from authority, but doing so in a way that does not tell you
the name of the authority. Whenever someone cites an authority,
you have every right to demand to know the name of the authority.
The second step in a successful argument from authority is to
make sure that the person cited as an authority is, in fact, an
expertthat is, someone we have good reason to believe would
know the correct answer to the relevant question.
What makes someone an expert? It depends on the case. On

one hand, if the question were about basic general science,
for example, then someone who has a basic education in the
area would be enough to qualify. On the other hand, if it were
an intricate question about a cutting-edge aspect of science,
then an expert would be someone with more education who is
involved in that particular subfield.
The third criterion in considering someone a legitimate authority is

that the person must not only be in a position to know the answer
to the question, but the person must also not have a stake in your
believing one way or the other. The expert must be objective or
disinterested.
If you are buying a used car from Joe, someone you know to
be a mechanic, should you take his word on the condition of the
engine? Joe is the sort of person who would know, but he also
stands to profit if he convinces you that the car is in better shape
than it is. You should not consider Joe to be an authority in this
case, even though he is indeed an expert. You should take the
car to another mechanic to give you an objective appraisal.
If someone meets all three of these requirements, that person

can be considered an authority, and his or her word can be
reasonably taken. If someone fails to meet one or more of these,
then the use of that persons word in an argument from authority
would be a reasoning error.

Appeal to Common Opinion
A related fallacy is called appeal to common opinion, also known
by the Latin, ad populum. Simply because many people believe
something does not make it true. It was widely believed that the
world was flat and that slavery was morally acceptable.
At some level, we all know that simply because something is a

common belief, that is not sufficient for rational belief, but what
we are dealing with here is not mere intellectual laziness but a
deeper cognitive bias: groupthink.
We like to think of ourselves as independent minded, but we are

deeply influenced by the views of those around us. We dont
like to stick out, and when we believe something the majority of
others do not, it will often lead to doubt and insecurity. When we
surround ourselves with others of like mind, it is comforting to us,
leading us to be surer of our views than is rationally warranted.
In a deep way, we are programmed to commit the fallacy of

appeal to common opinion. To be rational, we need to learn to
keep this proclivity in check. This does not mean that it is always
rational to reject common opinion; sometimes appealing to
common opinion is a good inference.
Suppose that you are attending a performance at a venue where

you have never been before, and after the show is over, you need
to find your way to your car, which is in the garage where many of
those in the theater have also parked. If you just follow the crowd,
you will probably end up exactly where you need to be. In this
case, there is nothing wrong with believing that what everyone
else believes is likely true. But this is not the usual case.
Lecture 4Fallacies of Faulty Authority 33

Appeal to Tradition
Widely held beliefs might be true, or they might be false. We
need independent reason to believe them rationally, and simply
appealing to the fact that everyone else thinks so is not enough.
But there might be reasons given why everyone else thinks so.
Some of these might be good, but others are not, despite the fact
that they are often cited. One of these problematic justifications
is called an appeal to tradition: But weve always done it that
way. Some traditions are good; some traditions are not.
If a belief, or belief system, or way of doing things has been

honed and crafted over generations, there might be good reason
to keep doing something in a certain way.
We learn from mistakes, both our own and those of others. And it
is certainly true that being an apprentice to someone with much
more experience can be a wonderful way of learningthat is the
argument from authority. But it requires evidence that the way we
have been doing it is the right way, or the thing we have always
believed is, in fact, true.
Intellectual inertia is not rational justification. Just because it has

always been believed or always been done a certain way does
not make that belief or that method justified.
An interesting version we find uses an appeal to other peoples

traditions. It is an ancient Chinese cure is something one hears
as an attempt to justify approaches to alternative medicine. Are
ancient Chinese medical practices effective? This may or may not
be true, and we need independent evidence. Merely appealing
to tradition is not sufficient.

The Fallacy of Novelty
In the same way that looking backward in time does not give
us legitimate warrant for rational belief, neither does looking
forward. The converse of the fallacy of appeal to tradition is the
fallacy of novelty. Just because something is the latest does not
necessarily make it the greatest. There is a reason we say new
and improved; just because it is new does not entail that it is
improved.
We regularly see this fallacy in advertising. Manufacturers of

products have a problem: They want you to be satisfied with the
product you have bought from them so that you will be loyal to
their brand, but they also want you to be dissatisfied so that you
will look to buy a replacement for it.
One way of making you want something you have already bought
is to convince you that while what you have is good, there is a
new one that is even better. If it is betterif it has capabilities the
old one does not that would be helpful, or if its new design makes
it more efficient or easier to usethen these might be legitimate
reasons to decide to purchase a new one.
But manufacturers try fallacious versions of this approach as

well. Just think of the terms that are often used for new versions:
upgrades or updates. Theyre explicitly saying that new is
better just because it is new. This may or may not be true and
thus requires evidence, not just assertions of its status as the
most recent addition.
Arguing by Analogy
Arguing by analogy is like faulty authority in that its a flawed
version of a good form of reasoning. It is a perfectly fine way to
arrive at reasonable beliefs.

We use arguments from analogy in science all the time. We
use computer models to predict the weather. Over the past
few decades, these computer analogues of the actual weather
systems have become increasingly accurate.
What makes arguments from analogy successful is that the

systems selected to be analogues do share certain structural
similarities to the system being modeled. The fallacy of faulty
analogy occurs when we argue by analogy using a flawed
analogy, where the system and the analogue are not alike in the
ways used to draw the inference.
Successful analogies involve structural similarities between two

systems. An analogy can go wrong if that structural similarity is
missing.
Some political candidates cite business experience as a central

reason to suppose them competent for public office. In both
cases, they would hold positions of authority in an organization.
But the goal of running a business is to defeat competition
in a marketplace and create monetary profits for only your
shareholders. By contrast, the goal in politics is to create laws
in a way that brings more than monetary benefit to all of society.
These are radically different tasks, and one cannot infer

competence in one from success in the other. The analogy
between the two does not hold because the structures of the two
do not share the requisite commonalities.
But even if there is a commonality, that is not sufficient. For

example, someone might argue that having a drink of alcohol is
like shooting up heroin: Both alter the brain, both are addictive,
and lives have been shattered by both. While that is certainly
true, the effects of a single glass of wine are different from heroin
use. The analogy fails despite the similarity because of a radical
difference in degree.

Readings

Questions
Identify the fallacies in the following passages from the following list:
faulty authority, appeal to common opinion, faulty analogy, fallacy of
novelty.
1.
My GPS says that we should take route 70 to route 97. You think we
should take route 85 to Buckminster Road after looking at a map.
Im thinking that the GPS directions are better because the GPS is
equipped with the latest route-finding algorithm.
2.
Look at the line for that new movie! It is out of the theater door, down
the street, and around the corner. That film must be great.
3.
I was reading the expert reviews for this product on its website. All of
them were outstanding. I think that this could be the product weve
been waiting for.
4.
She must wear dentures. He said that her teeth were like the stars,
and we know that the stars come out at night.

Answers
1.
fallacy of novelty
2.
appeal to common opinion
3.
faulty authority
4.
faulty analogy

Lecture 5
Fallacies of Cause and Effect
M uch of the reasoning we do involves the relation of cause and

effect. What will happen if you take a new medication while
also taking your pill for high blood pressure? But cause-and-effect
reasoning, called causal reasoning, can be quite difficult. As the old
dictum goes, correlation does not imply causation. Just because
two events are frequently seen together does not mean that one
necessarily causes the other. As you will learn in this lecture, there
are several regularly recurring mistakes that people make when
asserting cause and effect.
Cause-and-Effect Reasoning
Some of the most interesting and important claims we make
are about cause and effect. So much of what we want to know
about the world, about each other, and about ourselves involves
the kinds of why questions that assert cause-and-effect
relations. Unfortunately, they are also among the most difficult to
establish logically.
Most people have heard the old logical dictum that correlation
does not entail causationthat is, that simply because you can
find two things occurring together, it does not mean that we can
assert with any certainty that one caused the other.
It may be the case, but simply identifying a correlation does

not itself establish in any way that there is a cause-and-effect
relation. We need to understand the mechanism by which event A
brings about event B if we want to be justified in asserting that A
causes B.
We will look at different ways that this kind of erroneous inference
can be madethat is, different ways in which arguments that
attempt to establish that event A caused event B miss their mark.
It is certainly true that for event A to cause event B, A must

precede B in time. Causes come before their effects. First, we
observe the cause and then the effect. But just because we saw
event A before event B does not itself give us good reason to
infer that A caused B.
The Post Hoc Fallacy

Causes come before their effects, but time order is not
sufficient warrant to assert cause-and-effect relations. To do
so is to commit the error known by the Latin phrase post hoc,
ergo propter hoc, which means after this, therefore because
of it. Just because it comes after, it does not mean that it
happened because of it. Most logicians abbreviate this as the
post hoc fallacy.
We often make post hoc arguments around individual events

when the effect is something unusual or significant. Your car,
which has run beautifully for 10 years with no need of repairs, is
suddenly making strange noises, and you lent it to Ben just last
week. What did he do to your car?
When something odd occurs, we try to determine why. And if

we see something different in the antecedent contextif there
was something else unusual beforehandwe might jump to the
conclusion that whatever was different before must be the cause
of what was different after. But that prior difference might be
completely unrelated to the posterior difference.

Unbeknownst to you, your timing belt was on its last legs, a part
that usually only lasts 10 years. Bens driving had nothing to do
with it.
The post hoc inferences that we cling to the strongest are those
that do not correlate single events, but when we notice that most
of the time when we observe events of type A, we also tend to
observe thereafter events of type B. Seeing the repetition of the
correlation strengthens our belief in the causation.
Neglect of a Common Cause

Correlation does not necessarily mean causation. Again, if we
want to assert a cause-and-effect relation between A and B, we
need more than time order. We need the mechanism. Making the
inference based only on time order, only on correlation, can lead
to a flawed inference in a few different ways.
By taking correlation to imply causation, we commit the fallacy

called neglect of a common cause. Just because whenever we
see A, we also see B does not mean that A causes B or that B
causes A; there might be a third thing, C, that causes both A
and B.
It might be the case that whenever you see your normally mild-
mannered coworker taking an aspirin, he is uncharacteristically
cranky and short-tempered. This does not mean that aspirin has
a negative side effect that has psychological ramifications, nor
does it mean that he takes the pills because of his change in
disposition. Indeed, there might a third factorfor example, a
headache or a toothachethat causes both the foul mood and
the taking of the medicine.
Lecture 5Fallacies of Cause and Effect 41

Causal Oversimplification
Real-world phenomena are complex and often brought about by
a convergence of a multiplicity of factors. By misconstruing or
ignoring this complexity, we can commit one of two cause-and-
effect reasoning errors. The first is called causal oversimplification,
which involves picking out one part of a complex causal web and
ignoring the web.
We see this fallacy committed often when we are dealing with

complex social issues. The claim has been made that the reason
the divorce rate in America is so high is because of the womens
rights movement.
It wasnt until feminism removed women from their traditional

gender roles that the divorce rate started climbing significantly. The
feminist movement caused the breakdown of American marriage.
There is certainly no doubt that the advance of womens rights is a

part of the story about the increased number and rate of divorce.
Given the social context of most women before the advance of

feminism, many women would not have had the financial means
to live on their own, because of a lack of educational opportunities
or well-paying jobs, and as a result, some women were forced to
stay in unsatisfying or abusive marriages.
But once a greater range of possibilities became normalized

for women, such circumstances were less common, and bad
marriages dissolved more frequently.
But the complete story is more than just desertion of traditional

gender roles. There are many other aspects of the changing
culture that are intricately wrapped up in the story. To simply pick
out one element, one causal factor, and say it is the cause is to
commit the fallacy of causal oversimplification.

Confusion of a Necessary with
a Sufficient Condition
An error that is related to causal oversimplification is the
confusion of a necessary with a sufficient condition. A condition
A is necessary for B if you cannot have B without first having had
A. In other words, A is necessary for B if A is required for even
the possibility of B. A doesnt bring about B by itself, but if there
is no A, there is no B.
Oxygen, for example, is necessary for fire. If there is no oxygen,

there can be no fire. This doesnt mean that everywhere there is
oxygen there will also be fire, but take away the oxygen and you
remove the chance for fire. Oxygen is necessary for fire.
A condition A is sufficient for B if A, by itself, is enough to bring

about B. Winning a high-stakes lottery, for example, is sufficient to
become a millionaire. But while it is sufficientthat is, it is by itself
enough to make you a millionaireit is not necessary. There are
other ways to become a millionaire. You could start a tech company
that gets bought by Google, or you could be born or married into it.
While thinking that a necessary condition is sufficient is the

most common version of this fallacy, the converse can be found
occasionally as well. Sometimes we take a sufficient condition
and wrongly assert it to be necessary.
This will often be the result of focusing on a particular way of doing

something that has become habitual for us, and we allow ourselves
to be blinded to other ways of accomplishing the same task.
For example, you might ask your child why a dirty dish that
was just used is sitting in the sink. Your child might answer,
The dishwasher is running. While putting a dirty plate in the
dishwasher is sufficient for cleaning it, it is not necessary. There
are other ways to clean it, such as washing it by hand.

The Slippery Slope Fallacy
The final causal fallacy is perhaps the most famous: the slippery
slope fallacy. There is no doubt that there are causal chains
that is, an event A causes B, but then the effect of B becomes the
cause of C, which in turn causes D.
We can have causally related chains of events. This is what leads

to an alternate name one will sometimes see for the slippery
slope fallacy, the domino fallacy, which refers to the tipping of
dominoes. They are arranged until the first one goes, and then it
hits the second, which hits the third, and so on.
While such chains of cause-and-effect relations exist, the fallacy

is where one asserts the existence of such a chain without giving
full causal arguments for each step in the chain. Arguing for
cause-and-effect relations is tricky.
We need to show the underlying mechanism at work, and in the

complex world of intervening causes, often A would bring about
B, all other things being equal, but when rubber meets road, not
everything is always equal.
When warning people of an act we think is imprudent, we will

often neglect to do all of the logical heavy lifting and simply
assert a causal chain or the likelihood of it and leave it at that.
I wouldnt take that first sip of beer. It always starts with beer, but
then it goes to wine and then hard liquor, which paves the way
for marijuana, and then addictive drugs like cocaine and heroin.
That little sip might seem harmless, but it is the first step on a
slippery slope to addiction, losing your house and your family
everything will be gone.

Of course, there are stories of people who have followed this
unfortunate path to ruin. But the question for us is whether there
is good reason to believe that each of these steps down the
slippery slope awaits the particular person warned. If so, make
the case for each step. If you cannot, then the argument fails.
Readings

Kahane, Logic and Contemporary Rhetoric, chaps. 2 and 4.
Questions
post hoc fallacy, neglect of a common cause, causal oversimplification,
confusion of a necessary and sufficient condition.
1.
People who drive nice cars also tend to have large homes. I guess
that if you have a car that nice, you dont want to park it in front of a
small house.
2.
Zydeco music always has an accordion in it. This polka band has an
accordion player, so I guess they play Zydeco.
3.
I had eggs for breakfast, and then I played the best round of golf in
my entire life. Its eggs every Saturday for me. I want to bring down
my handicap.

4.
The reason drug use is down is because they have been showing
those just say no commercials on television. The message must
really be getting through to people.
Answers
1.
neglect of a common cause
2.
confusion of a necessary and sufficient condition
3.
post hoc fallacy
4.
causal oversimplification

Lecture 6
Fallacies of Irrelevance
W hen we disagree with others, the process is not merely logical,

but emotional. When someone thinks that we are wrong about
something, we feel attacked and often feel justified in attacking in
return. As you will learn in this lecture, the fallacies of irrelevance
frequently occur when our emotions lead us to focus on some aspect
of the disagreement other than the actual disagreement. We have to
make sure that we are on guard at all times to avoid committingand
getting sidetracked bythese diversionary fallacies.
Fallacies of Irrelevance
One of the most difficult aspects of engaging in passionate
discourse is keeping the discussion focused on the question
at hand. When someone is disagreeing with us, especially
if it involves a proposition we take to be important, we can
feel attacked.
The result is that the fight-or-flight portion of the brain becomes

engaged, and this can overtake the functioning of the part of the
brain associated with our rational faculties. We feel that we have
to defend ourselves through any means possible, not necessarily
the ones that will lead to open-minded consideration.
The outcome is often logically unfortunate. Frequently, the

conversation gets hijacked by irrelevant appeals that cause the
discussion to veer off in directions that do not serve the central
point, but serve only to obscure it.
Ad Hominem
One of the most common diversionary fallacies is where instead
of attacking the argument, we instead focus on attacking the
arguer. This fallacy is known by its Latin name, ad hominem,
which translates as to the man. The idea is that we are focused
on the person instead of the case the person is making.
Arguments are acceptable if they are soundthat is, if their form

is valid and their premises are well-grounded. If an argument is
valid and has true premises, then that is true regardless of whose
mouth it comes out of.
Arguments stand or fall on their own merits. Whose mouth it

comes out of is irrelevant. The argument is valid because of its
logical structure and is well-grounded because of the truth of its
premises. The identity, background, or motivation of the speaker
has nothing to do with the satisfaction of either criterion.
To argue that we should not acceptor, indeed, even consider

the argument because of the source of the argument is to commit
an ad hominem fallacy.
Ad hominem attacks tend to come in three general categories.

The first is the youre a jerk version. There are horrible, immoral
people in this world who do nothing to make the world a better
place and who often serve their own petty desires at the cost of
the well-being of others. But if that person makes an argument,
we need to analyze it objectively. We need to evaluate the validity
of the argument and assess the likelihood of the premises truth.
The second version is a form of guilt by association where we

discount an argument not for objective reasons, but because the
person offering it belongs to some identifiable group. Dont listen
to her; shes a feminist. You cant take his argument seriously;
hes a conservative.

A common variation of this kind is to point out that the speaker
is not among those who follow the advice the speaker is giving.
Known by its Latin name, tu quoque, the but you do it, too
objection is just an illegitimate ad hominem attack. It might be
true that the person telling you not to drink is an alcoholic, but
that doesnt mean it is not good advice.
The third class of ad hominem attacks is where we focus on the

motivations of the speaker. Of course, youd say that. You stand
to profit if its true. Again, maybe that is correct, or maybe it isnt,
but the argument stands or falls on its own merits, regardless of
who, where, when, or why the argument is made.
Attacking a Straw Man

Another diversionary tactic that we must be on guard against
is called attacking a straw man. The strange name comes from
the fact that it is easier to beat the stuffing out of a scarecrow
than it is to take on an actual human being. It is a metaphor for
arguments that do not address the actual argument made but
rather a weaker, easier-to-refute version.
Logicians have something called the principle of charity,

according to which, when one analyzes an argument, one must
assess the strongest-possible version of that argument. To defeat
a weak version does nothing in terms of demonstrating the given
argument to be unsound. It can only be rejected as not providing
legitimate grounds for rational belief in its conclusion if the
strongest version, the best understanding, is seen to be flawed.
Think of prize fighting. If a particular boxer is the reigning

champion, then he or she has to take on all challengers. One
cannot keep the title of heavyweight champion of the world and
fight only, for example, 12 year olds. While there are surely some
tough preteens out there, the point of being the champion of the
Lecture 6Fallacies of Irrelevance 49

world is that you are the top of the topthat you can defeat the
toughest competition anywhere.
It is the same thing with argumentation. If we are to refute an

argumentfind a flaw in it that leads us to reject it as providing
good reason to believethen, like the heavyweight champion of
the world, we as critical thinkers need to take on the strongest
version of the argument. To take on a weaker version and then
assert that we have done anything is to attack a straw man.
There are two main varieties of attacking a straw man. One

version is to alter the scope of the premises offered, making
them broader or narrower than the ones offered to weaken the
argument while keeping the rest of the premises intact. The hint
that this is what you are hearing is the phrase Oh, so what you
are saying is.
The other kind of straw man argument is more radical. It is where

the interlocutor replaces all of the premises wholesale. When you
hear the phrase the real reason, you are likely looking at a
straw man argument.
Why would someone say the real reason? Because what

they are doing is replacing the original reasonsthat is, the
premiseswith new premises, and odds are that these new
premises are going to be a whole lot easier to undermine. But in
undermining them, the interlocutor has done nothing with regard
to the soundness of the original argument, because the original
argument is gone.
Red Herring
Where attacking a straw man is the error wherein we replace
the premises of an offered argument, the fallacy known as a red
herring is where we change the conclusion.

When we replace premises with those that are easier to attack
but maintain the conclusion, we are still talking about the same
thing, only talking about it differently. But when we change the
conclusion, we are completely changing the topic of conversation.
That is a red herringthe ultimate in argumentative diversion.
Anyone who has ever been in a serious interpersonal relationship

knows all about red herrings.
You really need to clean those dishes in the sink. You make
yourself a snack and just clutter the kitchen and leave it for me.
That is not respectful or fair to me.
Well, if you want to talk about messes and respect, what
about the fact that you never pick up your dirty clothes in the
bathroom? You just throw them on the floor before you get in
the shower and leave them there.
This began as a discussion about dishes. The first partner made

the following argument.
One should clean up ones own messes, because not to do so

is disrespectful and unfair.
The dishes in the sink are your mess that is not cleaned up.
Therefore, you should do those dishes out of respect and
fairness.
That seems like a sound argument. The conclusion follows from

premises that certainly seem to be true. How does the other
partner respond to this argument? Not by showing that the
argument is flawed, but by giving a new argument.
One should clean up ones own messes, because not to do so

is disrespectful and unfair.
The clothes on the bathroom floor are your mess that is not
cleaned up.

Therefore, you should pick up the clothes on the bathroom
floor out of respect and fairness.
Notice that while the form of these two arguments is the same
and there is some overlap in content, the conclusions are different
propositionsthat is, they are completely different arguments.
Both are worth assessing, but they need to be considered one
at a time.
What we have here is a combination of a red herring and tu

quoque. Well, you do it, too, or something so much like it that you
cant criticize me for doing what you do.
If I do it, then I should be criticized and I should change my ways,

but that is a different question from what we are talking about,
which is the stack of dirty dishes in the sink.
The thing about people is that while we certainly have trainable

rational capabilities, we are also bundles of insecurities and
dedicated to agendas of our own which we take to be crucially
important. Arguments between two people in a relationship can
display insecurities.
Agenda-based red herrings are often seen in political

discussions. Consider a conversation like the following.
Your gun control proposal is an affront to gun owner rights.

We are talking about liberty being seized by an overinvasive
government here.
Oh, thats funny coming from the person who proposed such
draconian abortion regulations. If you want to talk about rights
and liberty being stripped by an overinvasive government,
there is example A.
Notice what happened. We started with a conversation about

the political benefits and flaws of a proposed piece of legislation

about firearms, and instead of evaluating the argument, we shifted
to a completely different topic: the permissibility of abortion.
Both are important issues. We should give both careful, thoughtful

attention. But we need to do so one at a time. I understand that
abortion rights is an important issue to you, and we will give it our
due attention, but right now we are talking about gun control.
When there is an issue that is important to us, we will often see

traces of it everywhere. Just because something reminds you of
a topic you want to discuss doesnt mean that we cannot first
finish the discussion we have started.
Readings

Questions
ad hominem, tu quoque, attacking a straw man, red herring.
1.
ont listen to him. He cant even speak proper English, so you know
D
his argument is also nonsense.
2.
You say that the changes to the tax code would promote fairness, but
the real reason you are in favor of it is that you want to punish the rich.

3.
You say that we need to help the homeless, but what about the
working poor who have a place to live? Do you think we should just
ignore them?
4.
You know that famous celebrity who is always going on and on about
the need to care about the environment and leave a small carbon
footprint? It turns out that she has a mansion, and you know that
thing uses a ton of electricity in the summer when she runs the air
conditioner. So, if she can use a lot of power, so can I.
Answers
1.
ad hominem
2.
attacking a straw man
3.
red herring
4.
tu quoque

Lecture 7
Inductive Reasoning
T his lecture begins the move from thinking about well-

groundedness to validity. Each of the two different types of
arguments, inductive and deductive, require different means of
determining validity. Inductive arguments start with a set of observed
instances and use that information to infer beyond it. Inductive
inferences do not give us the certainty of deduction, but in the
messiness of the real world, they are the inferences we most often
make. Because their conclusions outrun the content of their premises,
inductions are incapable of giving us complete confidence in their
conclusions, but they give us high probability, and for reasonable
belief, that is good enough.
Inductive Arguments
Arguments come in two different kinds: deductive and inductive.
A deductive argument is one in which the scope is non-
ampliativethat is, the scope of the conclusion is no broader
than the scope of the premises.
In other words, deductive arguments go from broad to narrow;

the conclusion doesnt talk about anything that wasnt already
covered in the premises. Because there is no new information
in the conclusionbecause it is just milking a specific result out
of the premisesone nice thing about deductive arguments is
that if they are sound, or well-grounded, then their conclusions
must be true.
Inductive arguments, on the other hand, are ampliativethat is,
their conclusions do move beyond the scope of the premises to
give us rational belief about something we have not yet observed.
Induction is ampliative in that it amplifies our rational beliefs; it
takes us from narrow to broad.
Inductive arguments are wonderful because they give us new

knowledge about the world. They take what we already know and
give us logical permission to believe new things that we did not
know before.
Deduction only rearranges our previous knowledge into new

forms we may or may not have considered, but induction actually
generates completely novel beliefs about the world. This does not
come without a fee, and the cost of this growing of our stockpile
of rational beliefs is certainty.
Because deductive inferences are only rearranging what we

knew before, if what we knew before was known to be true, then
the conclusions that come from deduction will also be true. But
with induction, because we are making a logical leap beyond the
content of the premise set, there is no absolute guarantee that
the result must be true.
The best we get from induction is likely truth. If you have a good
inductive argument, the conclusion is probably true. While
probably true is less desirable than definitely true, probably
true is sufficient for rational belief. We should believe that which
is probably true.
Deductive certainty in all of our beliefs would be wonderful, but

it is not available to us. We need inductive inferences because in
most real-life situations, it is all we have. And it is good enough.
We make inductive inferences all the time, and we should. High
probability is all we need for rational belief.

Types of Inductive Argument
While there are many different kinds of inductive arguments, there
are three forms that are the most important because they are
the most frequently used. In one form, called inductive analogy,
we move from a set of data showing universal adherence of a
particular property in a particular population and assert it of a
future individual.
An inductive analogy is an inductive argument of the following form.
P1 has the property A.


Pn has the property A.
I have only seen n instances of P.
Therefore, Pn+1 has the property A.
I have seen some number n different Ps, and every one of them
has had the property A; therefore, I believe that the next P I see
will also have the property A.
We make these kinds of inferences all the time. No, I dont want
to go see that scary movie, because itll give me nightmares like
all the other ones.
When we apply what weve learned from all other instances in the
past to one in the present or future, inductive analogy is the form
of the argument we are using.
A related but stronger inductive inference is the universal

generalization. It has the following form.

Lecture 7Inductive Reasoning 57


Pn has the property A.
I have only seen n instances of P.
Therefore, all Ps have the property A.
While both inductive analogy and universal generalization have

the same premises, notice the difference in the conclusions. One
is an analogy in that it says that a future instance will be like all
past instances, whereas the other makes a much broader claim,
saying something about all members of the observed population.
Universal generalizations are also common in everyday life.

Every time Ive tried something with cilantro in it, I have thought
its disgusting. I dont like cilantro. Notice that in this example, we
are making a generalization about an entire category of things,
not just a prediction about a single upcoming instance.
The third form of inductive inference is called statistical

generalization. Its form is as follows.
X percent of all observed Ps have the property A.

Therefore, X percent of all Ps have the property A.
Like universal generalization, we are generalizing over the entire set

of Ps from some limited sample of Ps, but here we are not attributing
the property to all of them, but to some percentageeither an
explicit percentage or a vaguer amountof the population.
Four out of five dentists surveyed recommend sugarless gum for

their patients who chew gum, so likely the overwhelming majority
of dentists overall agree.
In some cases, we are generalizing a more specific statistic,

such as a percentage of the population as a whole; sometimes
we are taking a specific statistic, such as 80 percent of dentists

surveyed, and generalizing it broadly; and sometimes we are
taking a vague sense, such as a lot of the time or most people
surveyed, and making equally vague generalizations.
We need to be careful that we dont fall prey to one of the

standard fallacies we see with inductive inferences: exaggerated
accuracy. If a certain basketball player has made 75 percent of
his free throws for the year, that could mean that we think he will
hit most of his free throws in the next gamethats a perfectly
fine statistical generalization.
But it is not good reason to think he will make three out of four.
There is a lot of mathematical machinery in the study of statistics
that tells you exactly how big a sample you need and what
percentage needs to have the property in order to have a 90
percent, 95 percent, or 99 percent degree of confidence that your
statistic can be accurately generalized to the entire population.
When you have a statistic you want to generalize, you are almost
always going to have to soften your generalized conclusion a bit.
Inductive Fallacies
One of the fallacies associated with inductive reasoning is
exaggerated accuracy. There are several more we need to be
aware of. One involves premises that appear in the first two
inductive argument forms and implicitly in the third.
Note the last premise in inductive analogy and universal

generalization: I have only seen n instances of P, and note the
form of the premise in statistical generalization: X percent of all
observed Ps have the property A. In both cases, we see what is
called requirement of complete information.

Induction is unlike deduction in that deduction has, but induction
lacks, the property logicians call monotonicity. An inference is
monotonic if adding more premises will not turn a valid argument
into an invalid one. This is true for deductive arguments.
As long as the argument is valid to begin with, new premisesno

matter what they arewill always keep the argument valid. While
this is true for all deductive arguments, it is not true for inductive
arguments. Inductive arguments are non-monotonic because
adding a new premise can turn a perfectly good inductive
argument into a bad one. One new piece of information can
completely destroy an argument in a way that cannot happen for
deductive arguments.
So, for an inductive inference to be valid, we need a guarantee

that the evidence given in the premises is all of the relevant
observations we have. We need complete information. To fail to
do so is called cherry-picking.
This is a common error where when someone wants to support a

point, he or she will present lots of evidenceso much evidence
that we cant help but believe the conclusion it leads to. But if the
person was careful to select only the evidence that supports his
or her position and excludes the counterevidence, we will be led
to believe something that is not well supported.
The key to avoiding this error is in the selection procedure for the
sample we use to collect our observation. The sample is the set
of n individuals mentioned in the premises. How do we go about
getting our data from which to make the inductive inference?
There are two keys to having an acceptable sample for a good

inductive inference. The first is sample size. We need the sample
to be large enough to support our inference. How large is large
enough?

This is something that statisticians have studied in great detail
and have tables dedicated to showing for varying degrees of
confidence in our inferences. For the purposes of this course,
one should be on the lookout for absurdly small samples, called
anecdotal evidence.
If someone is generalizing on the basis of just a few examples

or often just a single experience, then we have the fallacy of
insufficient sample. If this error is committed while making
a universal generalization, we sometimes call it a hasty
generalization, but the term insufficient sample covers all of the
forms of inductive reasoning from too small of a sample.
The size of the sample isnt the only thing we need to be

concerned about. Bad samples could be quite large. To make
a good induction, samples have to be representative of the
population over which we are making the inference.
The sample needs to look like the population in miniature

because there might be aspects of different subgroups within
the population that affect the distribution of property we are
examining. If you have a poorly distributed sample, you commit
the fallacy of unrepresentative data.
Good samples model the heterogeneity of the population. If

your population is homogeneous, then you dont need to worry
so much, but the key is that you cover the relevant subgroups
in proportion to their general representation. This assumes, of
course, that you know what subgroups are relevant and their
proportion of the general population beforehand.
But sometimes you dont. In those cases, the key is a random

sample. The idea is that if we pull enough individuals out of the
population without a bias toward or away from any subgroup,
then relevant subgroups will show up in the sample roughly the
proportion they occupy in the whole.

The key to the random sample, then, is to make sure that your
sample is large enough that small subgroups will appear and
that your selection procedure does not accidentally bias your
selection toward or away from such subgroups.
The gamblers fallacy is one in which we try to make an inductive

inference from data that bear no cause-and-effect relationship
to each other, where there is not a probabilistic relation between
the members of the sample and the property being observed.
Arguing that past experience in any way affects the next instance
is to commit the gamblers fallacy.
Readings

Questions
1.
If inductive arguments do not give us conclusions with absolute
certainty, why should we believe the conclusion of a sound inductive
argument? What inductive arguments do you accept in day-to-day life?
2.
Identify the fallacies in the following passages from the following
list: cherry-picking, insufficient sample, unrepresentative data,
gamblers fallacy.

a Al Capone: Italian and mafia. Lucky Luciano: Italian and mafia.
John Gotti: Italian and mafia. Sammy the Bull: Italian and mafia.
I could go on and on, naming hundreds of Italian mobsters. So,
it must be the case that all Italians are connected.
b I always hold my breath when crossing a bridge to make sure

that nothing bad happens.
c My Ford has 300,000 miles on it and has never needed a

single major repair. Ford makes quality cars.
d A survey conducted by Christian televangelist Pat Robertson

polled more than a thousand of his followers and overwhelmingly
showed that people want creationism instead of evolution taught
in public schools.
Answers
1.
Answers will vary.
2.
a cherry-picking
b gamblers fallacy
c insufficient sample
d unrepresentative data

Lecture 8
Induction in Polls and Science
M ost of what we believe traces back to inductive reasoning. We

learn from experience, and the reason that works is because of
the successfulness of induction. Inductive arguments are everywhere
in life. There are two places, however, where we conspicuously
find formalized inductive inferences that deserve extra discussion:
Inductive inferences are made whenever a poll is taken and reported,
and induction is a crucial element of scientific reasoning. In this
lecture, you will learn about induction in these two contexts. Youll
learn how to read and understand what information a poll is giving
and how to understand the roles of induction in scientific results.
Inductive Inferences in Polls

We have looked at the basic forms of inductive reasoning and
the ways in which some inductive arguments can go wrong. With
respect to polls, those are precisely the errors we need to be on
the lookout for, with some interesting twists and a few additions.
Virtually all polls taken are of the form of the inductive argument
called statistical generalization.
X percent of all observed Ps have the property A.

Therefore, X percent of all Ps have the property A.
We are generalizing a statistic we find in our observed sample to

the entire population. What we want to examine with respect to
polling is what makes for a successful statistical generalization
and how much reason to believe is achieved as a result.
Inductive arguments, because they are ampliativethat is,
because their conclusions are broader in scope than their
premisescome with risk. Even the best inductive arguments do
not provide us with the certainty of deductive arguments.
Successful inductions give us high probability, a level of

reasonable belief. We want to see, then, how a single poll or
collection of polls should affect what we believe and how deeply.
Several of the problems that can plague inductive arguments come

from flawed samples. This is the case with polls, too. Indeed, one
of the biggest concerns of legitimate pollsters is sampling.
For a good sample, we need both size and distributionthat is,

we need for the poll to have asked enough people, and we need
for those people to resemble the population being generalized
in miniature. All of the relevant subgroups must be present in the
sample in roughly the proportion in which you find them in the
general population.
With respect to size, we looked at the fallacy of insufficient sample

in its simplest form: anecdotal evidence, where we generalized
from just one or two experiences. But in a poll, where we are
trying to generalize over an entire electorate or an entire national
population, how big is big enough?
It is actually surprising how small a sample can be and still

be of an acceptable size to make meaningful claims about so
large a population. About 1,000 respondents is sufficient for a
nationwide poll in the United States, which has more than 150
million registered voters.
The other concern, once we have enough people, is that the sample
is well distributedthat it looks like the population as a whole in
miniature. With polls, we have two concerns: that we are sampling
the right population and that the sample is properly representative.
Lecture 8Induction in Polls and Science 65

But even when we have determined the proper population and
can develop filters to screen out those who do not belong, we
have the question of creating a well-distributed sample.
There are a few different ways of trying to develop a well-

distributed sample. First, we could identify all of the important
demographic groups and their proportions in the population
beforehand and then make sure that our sample is shaped to
resemble it.
Second, we could take a random sample. If we select enough

people at random from a large group, the sample will come to
resemble the population as long as there were not selection
biases toward or against given subgroups.
Clearly, the first option is preferable. If we know what properties

are germane, we can find people with the desired profiles in the
desired proportions. But how do we know this?
Often, it is done inductively. We look at past polls and see how

accurate their predictions came out and see what their sample
looked like in terms of different subgroups. Pollsters call this
making a model of the population. They collect their data but
then have to decide how to weight the contributions of various
parts of the population.
Once the sample worries are behind us, there is a new concern:
Can we trust what people tell the pollster? There are two worries
here. The first is that people sometimes intentionally mislead
pollsters.
Especially on questions that have a moral element, people dont

want to be seen as being on the unpopular side of the question
and will tell pollsters what they think will lead the pollster to think
better of their character, even if it does not reflect the way they
will actually vote or what they actually believe.

The name for this is the social desirability bias and with respect
to polls is often referred to as the Bradley effect, named for Los
Angeles mayor Tom Bradley, who when running for governor of
California had a sizeable lead in the polls but lost the election.
Some have contended that this was the result of people falsely
telling pollsters that they would vote for Bradley when they would
not, so that they would not appear to be racist.
Second, it turns out that very different results can be achieved

by asking the same question in different ways. How the pollster
chooses to ask the question will affect how people answer it.
This is a cognitive bias that social psychologists call the framing
effect, and it relates to the fallacy of begging the question.
Words not only denotethat is, pick out objects to refer tobut
also have connotative power; they convey emotional or value-
laden judgments as well. Questions framed differently will allow
the connotative power of the language employed to steer the
listener toward or away from particular viewpoints in ways that are
subtler than leading questions.
Polls not only report on the public mood and beliefs, but also
influence them. This means that we need to be cautious.
Questions might have been asked intentionally or unintentionally
in a way that biases them.
Ones own ideological filter can lead one to frame a question in

a leading way that seems perfectly fair to you. The key, then, to
making good inferences from polling data is to take a step back.
Polls themselves are inductive argumentsthat is, they take data
and extrapolate a general result.
But we are now able to use the polls themselves as data in an

inductive argument at a higher level. We can now take polls of
polls. There are websites that work as poll aggregators, collecting

all of the polls on a particular topic and displaying their results,
the size of their samples, etc. By bringing all the polls together,
we can form a rational basis for reasonable belief based on them.
Induction in Science
Induction is used in two different ways in science: in supporting
hypotheses and in testing theories. We need to keep these two
contexts separate.
A hypothesis is a proposed statement of purported fact. Scientific

theories, by contrast, are sets of general axioms, which together
form a system of thought that provides a full picture of the
workings of some part of nature.
Hypotheses are proposed individual statements of possible truth;

they are more specific than the axioms, and we get evidence
for them individually. The axioms work together as a group, and
what we test in that very different context is the theory as a whole.
We might be able to derive hypotheses when working within the
theory, but the parts of the theory are not themselves hypotheses.
There are different inductive processes for hypotheses and

theories. The philosopher Karl Popper pointed out that a hypothesis
is scientific only if it is falsifiablethat is, only if there are observable
circumstances that would render the statement false.
How do we go about our scientific inductive inferences? The first

step is to identify the independent and the dependent variables.
The independent variable is the thing we adjust or administer
the thing under our control. The dependent variable is what we
measure. It is the thing not under our control, and it may or may
not change as a result of our adjusting the independent variable.

For example, to determine whether high doses of vitamin C
reduce the symptoms of the common cold, we can give someone
high doses of vitamin Cthe independent variableand we can
then check to see if they receive relief from their cold symptoms
the dependent variable.
One problem is that the world is a complex, messy place. While

we are giving someone vitamin C, they might also be eating
something else that is decreasing their symptoms, and we will
wrongly attribute the effect to our independent variable. So, we
need to conduct the experiment in a way that controls for other
independent variablesthat is, we need to do our best to create a
fixed environment in which we can screen off intervening causes.
When we get the data, we analyze it to see if the results are

statistically significant. If so, we have reason to think that the
results are consistent with the hypothesis. What that means is that
the hypothesis has not been
falsified, but we also do not
yet have enough evidence
to think it is probably true.
Where do we see induction

in science with respect to
theories? The philosopher of
science Hans Reichenbach
drew a distinction between
the context of discovery and
the context of justification.
What this distinction has
come to mean is the context
in which scientists come
up with their theories and
the context in which they Hans Reichenbach
provide good reason to (18911953)
believe they are true.

The context of discovery is generally thought to be freethat is,
there is no specific logic of discovery, no turn-the-crank method
of coming up with scientific theories. But while there is no set
method, there is induction, because scientists are working from
their experiences and the data.
They have a question about how a certain system works, and

they consider what they know and make inductive leaps. They
look for modelsanalogies where the system could be thought
to work like a different system that is better understood.
The most important place in scientific reasoning that we find

induction is in the context of justification. Once a theory has been
proposed, why should we believe it? Theories are testable; they
have effects, results, and predictions that come from them. These
observable results of a theory are determined deductivelythat
is, if the theory is true, then in this given situation, a particular
observable consequence should result.
We go into the lab, set up the situation, and see if we observe

or measure the result as expected. If not, then the theory has
failed and, as it stands, is not acceptable. It will either have to be
rejected or fixed.
But if the theory says to expect a particular result and we observe

it, now we have evidence in favor of the theory. That evidence is
inductive. It might be that a particular theory predicts the result,
but there will also be other theories that are different from the first
one but are also supported by the result.
As such, none of the theories are certain, in the way that deductive
inferences are, but rather they receive inductive support. They
are more likely to be true than they had been. The probability of
truth has been increased.

To go from supporting evidence, which makes a theory more
likely, to conclusive evidence, which makes a theory probably
true, we need lots of evidence as well as evidence of different
types.
It is good for a theory if it accounts for everything we already

knew. We call this retrodiction. This is especially true if what
we already knew was previously unexplained. But better than
explaining what we already knew, prediction is taken as strong
evidence.
The best evidence brings about what scientist William Whewell

termed consilience, which is when a theory designed to
account for phenomena of type A turns out also to account for
phenomena of type B. If you set out to explain one thing and also
are able to explain something completely different, that is strong
evidence that your theory is probably true.
Readings
Bradburn, Sudman, and Wansink, Asking Questions.

Gimbel, Exploring the Scientific Method.
Kelley, The Art of Reasoning, chaps. 1517.
Questions
1.
Push polls are fake polls in which people believe that they are being
asked for their opinions by a reputable pollster but are actually being
lobbied by an interested party that is trying to change their opinion.
Political campaigns, for example, will have people pose as pollsters
and ask biased questions designed to influence the respondents
answer. Because the person believes the pollster to be legitimate, the
respondent thinks that the questions are not skewed and therefore

have their own thoughts unfairly affected. Are all polls, in some sense,
push polls, or can a fair poll be conducted that takes the temperature
of a population without affecting it?
2.
Because science is inductive in the inference it requires for evidence,
science proves nothing. If the results of science are never proven
if scientific progress might require us to surrender even the most
deeply held current beliefsthen is it ever rational to believe the
results of science?

Lecture 9
Introduction to Formal Logic
Y ou have learned that there are two criteria for acceptable

arguments: well-groundedness and validity. You have also seen
that there are two types of arguments: inductive and deductive. So far,
you have examined some aspects of informal logicthat is, analysis
of the well-groundedness of argumentsand you have looked at
validity concerns for inductive arguments. What remains is deductive
validity, the aspect of logic that has historically commanded a
great amount of attention from logicians. In this lecture, you will be
introduced to formal logic.
Formal Logic
In part because of its use in
mathematics and in part because
it is the study of what we can
know with absolute certainty,
there is a long trail of work in
formal logic. The first figure in
the history of thought to give us
a functional formal account of
reasoning was Aristotle.
Possibly the smartest person ever Aristotle

(384 B.C.E.322 B.C.E.)
to have lived, Aristotle provided us
with the starting point of virtually
every academic discipline, from
economics and physics to literary theory and ethics. But for this
course, it is his work on categorical logic that is important.
Categorical Propositions
Aristotles logic begins with a little bit of grammar. Formal logic
examines what propositions necessarily follow from what other
propositions because of their forms.
Propositions need to be expressed as declarative sentences, and

declarative sentences, Aristotle contends, have a specific form.
All declarative sentences that we will consider have a subject
and a predicate.
The subject is what the sentence is about, and the predicate is

what we are asserting about the subject. A sentence is true if
and only if the subject does have the property asserted by the
predicate, and it is false if the subject does not.
These categorical sentences come in four types. First is the

universal affirmative, sentences of the following form: All As are B.
All people have noses.
Second is the universal negative: No As are B.
No circles have angles.
Third is the particular affirmative: Some As are B.
Some people are blonde.
Finally, there is the particular negative: Some As are not B.
Some wines are not expensive.
We arrange these sentences into what is called the square of

opposition.

Universal
All As No As
are B are B
Affirmative Negative
Some As Some As
are B are not B
Particular
The top line is for universal sentences. The bottom line is for
particular sentences. The left side is for affirmative sentences.
The right side is for negative sentences.
There are abbreviations for each. Universal affirmative sentences

are called A sentences. Universal negations are called E
sentences. Particular affirmative sentences are I sentences, and
particular negative sentences are called O sentences.
Universal
All As No As
are B are B
Affirmative Negative
Some As Some As
are B are not B
Particular
Lecture 9Introduction to Formal Logic 75

The reason we call this the square of opposition is because
opposite cornerssentences that are connected diagonally on
the squareare negations of each other. That is, an A sentence
is true if and only if the O sentence is false.
All Boy Scouts are boys if and only if it is not true that some Boy
Scouts are not boys. Similarly, if it is false that all Boy Scouts are
boys, then there must be some Boy Scout that isnt a boy. A and
O sentences will always have different truth-values.
And it is the same for E and I sentences. If it is true that no clowns

are happy, then it is false that some clowns are happy. And if it
is true that some clowns are happy, then it must be false that no
clowns are happy. E and I sentences must have different truth-
values.
The sentences on opposite corners are called contradictories

because they contradict each other: One and only one can be
true at a time. But this is not the case with A and E or I and O
sentences.
A and E sentences are called contraries. They cannot both be

true, but they can both be false. It is false that all paintings use
the color blue and false that no paintings use the color blue.
I and O are called subcontraries. They can both be true, but they
cannot both be false. It is true that some people are having a
birthday today and some people arent. But if it is false that some
As are B, then it must be true that some As are not B.
In these I and O sentences, the word some means at least one,

maybe all. By using the word some we are not saying that only
some, but not all. It might be simultaneously true that all people
have mothers and that some people have mothers.

The word all is slightly more complex. It means every member
of the category. But what about the peculiar categories that are
empty? Consider the following A sentence: All unicorns have
a horn. Is this sentence true or false? It seems to be true by
definitionto have a horn is part of what it is to be a unicorn. But
at the same time, there are no unicorns for the sentence to be
true of.
How do we make sense of this? The answer is that we have two

different meanings: the hypothetical viewpoint and the existential
viewpoint.
In the hypothetical viewpoint, the sentence All unicorns have a

horn is true, because it means that all unicorns, if there are any
(and there might not be), have a horn.
From the existential viewpoint, the sentence All unicorns have

a horn is false, because it now means that there are unicorns
and all of them have a horn. We need to know whether we are
working with the hypothetical or existential viewpoint before we
start doing our logical manipulations.
Notice one important difference. From the existential viewpoint, A

sentences imply I sentences and E sentences imply O sentences.
If no square has five sides, then some squares do not have five
sides.
As long as we know that the subject exists, then if the predicate

holds for all, it must hold for some. But from the hypothetical
viewpoint, just because an A sentence is true does not mean that
the corresponding I sentence will be true.
In the hypothetical viewpoint, it is true that all griffins have the

body of a lion, but that does not mean that some griffins do,
because the word some means that there is at least one, and
the griffin is a mythical beast. There arent any.

So, with vacuous subjects, we can have true A sentences and
false I sentences or true E sentences and false O sentences.
Both viewpoints are legitimate, but for claritys sake, lets assume
the hypothetical viewpoint unless otherwise noted.
Categorical Syllogisms
Recall that our interest here is deductive validity. We want to
know when we have to believe a conclusion if we also believe the
premises. For Aristotle, the key to reasoning is a type of argument
called a syllogism, which is an argument with two premises.
Categorical syllogisms are arguments with a categorical sentence

as a conclusion and two categorical sentences as premises.
All humans are mortal.

All Greeks are human.
Therefore, all Greeks are mortal.
The conclusion has two terms: a subject and a predicate. The

subject of the conclusion is called the minor term (S). The
predicate of the conclusion is called the major term (P).
In the example, Greek is the minor term, and mortal is the

major term. There is another term that appears in both premises,
but not in the conclusion. This is called the middle term (M).
Human is the middle term in the example.
The premise with the minor term and the middle term is called the
minor premise, and the one with the middle term and the major
term is called the major premise. We always write out the major
premise first.

The example has a major premise that is an A sentence, a minor
premise that is an A sentence, and a conclusion that is yet another
A sentence. Reading top to bottom, we say that this argument has
the mood AAA. AAA syllogisms are nice because if they turn out
to be invalid, you can call for a free tow into the nearest garage.
A All humans are mortal. Major Premise
A All Greeks are human. Minor Premise
A Therefore, all Greeks are mortal. Conclusion
But notice that we can have different variations of AAA. Compare

our example with a new one.
All Greeks are human.

All humans are mortal.
Therefore, all mortals are Greeks.
These two arguments both have A sentences for premises,

but because the terms are switched around, they are different
arguments. Indeed, the first example is valid, but the second is
not. The point, though, is that the order of the terms matters.
There are four possible arrangements of the terms. Each

arrangement is called a figure.
The first figure has a major premise that starts with the middle
term and a minor premise that starts with the minor term.
The second figure has a major premise starting with the major
term and a minor premise starting with the minor term.
The third figure starts both premises with the middle term.
The fourth figure starts the major premise with the major term
and the minor premise with the middle term.

This is a complete catalogue of possible figures.
1 2 3 4
M P P M M P P M Major Premise
S M S M M S M S Minor Premise
S P S P S P S P Conclusion
Combining mood with figure, we get a complete list of possible

forms of categorical syllogisms. For example, Some dogs have
four legs, and some four-legged beings are not cats; therefore,
no cats are dogs.
The major premise is an I sentence: Some dogs have four legs.

The minor premise is an O sentence: Some four-legged beings
are not cats. And the conclusion is an E sentence: No cats are
dogs.
So, the mood is IOE. The middle term appears second in the
major premise and first in the minor premise, so it is of the fourth
figure. We have an argument that is IOE-4.
The first example argument has a mood of AAA, and the middle
term is second in the major premise and first in the minor premise,
so it is an AAA-1.
There are four possibilities for the major premiseA, E, I, O

and the same four possibilities for the minor premise and the
conclusion. That gives us 4 4 4, or 64 possible moods.
We then have four figures for each mood, giving us 64 4, or
256 categorical syllogisms whose validity status we need to
determine. How do we do that?

Aristotles Five Rules
There are five simple rules that detect the 15 valid forms from
the hypothetical viewpoint. But before we work with the rules, we
need to understand one more concept: a distributed term.
A term in a categorical sentence is distributed if that sentence

says something about the entire category the term refers to.
So, in an A sentence like All Greeks are human, the subject is
distributed because it tells us something about all Greeks.
Likewise, in an E sentence like No cowboys are werewolves, we

are told something about the entire class of cowboys, so again
the subject is distributed.
In an I sentence like Some people are allergic to peanuts,

we are not told anything about any entire class, so no term is
distributed.
In an O sentence like Some computers are not made by

Apple, it seems intuitively that, like with I sentences, neither
term is distributed. But, in fact, the predicate of an O sentence is
distributed. We know of the entire category of products made by
Apple that it fails to include some computers.
So, O sentences distribute the predicate, A and E sentences

distribute their subjects, and I sentences distribute nothing.
Having this concept, we can now set out the five rules.
1 In all valid syllogisms, the middle term is distributed in at least

one of the premises.
2 In all valid syllogisms, any term distributed in the conclusion is
also distributed in the premises.
3 In all valid syllogisms, at least one of the premises must be
affirmative.

4 In all valid syllogisms, if the conclusion is negative, one premise
must be negative.
5 In all valid syllogisms, if the conclusion is particular, at least
one of the premises must be particular. (This one is for the
hypothetical viewpoint only; for the existential viewpoint, just
use the first four.)
A syllogism is valid if and only if it satisfies all five of these rules. If

it violates even one, its an invalid argument.
Venn Diagrams
These rules work, but they are not terribly intuitive. We can
determine which categorical syllogisms are valid and which are
not, but we come away without a sense of why. It would be nice to
have a method that allows us to see why a categorical syllogism
is or is not valid.
Fortunately, we have one. It was provided by the 19th-century

English logician John Venn. He came up with a way of visually
representing the content of each type of categorical propositions
in diagrams that forever bear his name: Venn diagrams.
We start with a pair of overlapping circles, with the one on the

left representing the subject class and the one on the right
representing the predicate class. The area in the overlap is the
part of the subject class that also is in the predicate class.
The parts of the circles outside of the overlap are for members of
the classes that belong only to one or the other class. We indicate
that a region is empty by shading it in, and we represent that an
area is not empty by putting an X in it.

For categorical syllogisms, we just enter the information from the
premises into a larger diagram that has three circles, one for the
minor term on the bottom
left, one for the major
term on the bottom right, Venn Diagram
and one for the middle
term, which is placed
above and appropriately
in the middle. Middle
Term
We enter in the premises
and see if what comes out
contains the information
within the conclusion. If it Minor Major
Term Term
is, then the argument is
valid. If not, it is invalid.
Readings

Copi, Introduction to Logic, chaps. 5 and 6.
Questions
1.
Identify the form of the following categorical syllogisms.
a No dogs are blue. Some blue things are not fruit. Therefore,
some fruit are dogs.
b All desserts are sweet. Some children are sweet. Therefore,

some children are desserts.

2.
Use Aristotles rules to determine if the following syllogism is valid.
Some movies are not comedies.

Some movies are not romances.
Therefore, some romances are not comedies.
3.
Use Aristotles rules to determine if the following syllogism is valid.
No food is poisonous.
Some mushrooms are poisonous.
Therefore, some mushrooms are not food.
4.
Use Venn diagrams to show whether the two categorical syllogisms
in questions 3 and 4 are valid.
Answers
1.
a EOI-4
b AII-2
2.
In OOO-1, the predicate of each sentence is distributed. The first rule
requires that the middle term be distributed in one of the premises. O
sentences distribute the predicate, but the middle term is the subject
in both premises and therefore is undistributed. Because the middle
term is not distributed, by the first rule, the argument is invalid.

3.
In EIO-1, the minor premise is an E sentence, so both terms, food
and poisonous, are distributed. The major premise is an I sentence,
and nothing is distributed in an I sentence. The conclusion is an O
sentence, which distributes the predicatein this case, food. The
first rule requires that the middle term be distributed in one of the
sentences, and it is in the minor premise. The second rule requires
that any term distributed in the conclusion be distributed in the
premises. The term food is distributed in the conclusion and also
in the minor premise. The third rule requires that at least one of the
premises be affirmative, and the major premise satisfies that one. The
fourth rule requires that if the conclusion is negative, then one of the
premises is, and we do have a negative conclusion and a negative
minor premise. Finally, according to the fifth rule, if the conclusion is
particularwhich it isthen one of the premises must be particular,
and the major premise is. So, the argument is valid.

Lecture 10
Truth-Functional Logic
A ristotles categorical logic was the first attempt at creating

a rigorous calculus of human thought, a surefire means of
determining when a deductive argument is valid. The problem with
Aristotles system is that it only works for sentences of categorical
form; for non-categorical sentences, the tool kit of Aristotelian logic
will not let us do what we need to do. We need a different system
one that can handle a wider range of propositions that we use in
deductive arguments. This system is called truth-functional logic.
The Need for Truth-Functional Logic

At the end of the 19th century and beginning of the 20th century,
truth-functional logic was developed by philosophers who found
a need for a new language. Philosophy has to be done with
words. But the words we use in everyday, ordinary language are
ambiguous and vague. They are not defined precisely enough to
do the work that philosophers need.
Instead of doing philosophy in ordinary language, they thought, they

would develop a new artificial language, one that the exactitude
they required was built into its very grammatical structure.
It would resemble spoken language enough that we could

translate our philosophical questions and intuitions into it, but
it would be strict enough that we could finally answer some of
these questions.
We could see whether they were real questions or merely pseudo-

questions, and if they were real questions, we could determine
their truth conditionsthat is, what we would have to look for
to know if they were true or false. The German mathematician/
logician/philosopher Gottlob Frege called his attempt at framing
such a language Begriffschrift, or concept writing. It ultimately
became truth-functional logic.
In the phrase truth-functional, we use the word function in the

same way that mathematicians do. Mathematicians look at an
algebraic equationfor example, y = x + 2and say that y is a
function of x. What this means is that for every value of x, there is
a single, unique value for y. For example, if x is 1, then y is 3.
A function takes in something and then spits out a well-defined,

completely determined something else. Numerical functions take
in numbers or pairs of numbers and spit out a single number.
Instead of numbers, truth-functions take in truth-values and spit
out a single truth-value.
In classical logic, there are two truth-values: true and false. Every
sentence has one and only one truth-value. If a sentence is true, it
is not false. If it is not true, then it has to be false. Truth-functional
logic is a two-valued system, and every sentence has one or the
other of these values.
For now, it doesnt matter how we know which it is or whether

there are some sentences whose truth-values we dont know how
to ascertain. All we are concerned with is if these sentences are
true, does that mean some other sentence also has to be true?
In fact, the content of the sentence will be completely irrelevant. All

we are concerned with here is the form. When we translate spoken
language into this language, we will remove everything about the
content of the propositions and strip it down to the bare skeleton
the logical structure. Whether one sentence follows from another
depends only on its form, so content will go away for us.
Lecture 10Truth-Functional Logic 87

The Elements of Truth-Functional
Language
Truth-functional language has two elements: atomic sentences
and truth-functional connectives. Atomic sentences are simple
declarative sentences that are either true or false.
The sky is blue is atomic.
This is a simple sentence that has a truth-value. We may or may not

know the truth-value, but again, that is of no consequence. In our
language, we use lowercase letters to represent atomic sentences.
To our atomic sentences, we add truth-functional connectives.

A connective is a word that joins atomic sentences together to
form a more complex molecular sentence. The word and, for
example, is a connective.
We can take the sentence The sky is blue and the sentence
I am 12 feet tall and use the connective and to create a
whole new sentence: The sky is blue and I am twelve feet tall.
Connectives just join atomic sentences to form new sentences.
A connective is truth-functional if and only if the truth-value of

the molecular sentence is completely and uniquely determined,
knowing only the truth-values of the constituent atomic sentences
and the definition of the connective. And is truth-functional.
We define truth-functional connectives in terms of what is called a

truth table. In a truth table, we start by setting out all of the atomic
sentences involved, listing all of the possible combinations of
truth-values beneath.
And is a dyadic connectivethat is, it joins two sentences

togetherso we need two sentences (it doesnt matter what they
say, just that theyre atomic): Lets call them p and q. We start our

truth table by writing down a column for p and p q
a column next to it for q.
T T
We then list all of the combinations of truth-
values they could have. Because we have two T F
sentences, each of which could have one of F T
two truth-values, there are four possibilities
(shown at right). F F
After our atomic sentences are all entered in our truth table,
we add a new column in which appears nothing but what has
come before in the table and one new connective. We will use
the symbol & as our symbol for and, although some logicians
prefer the wedge () or a dot ().
Next is the important step: filling in the value for the molecular
sentence. In truth-functional logic, what we mean by and
is and.
Think about how we use the word and. p q p&q

Suppose that you have a friend Bob
over at the coffee machine and you say, T T T
Bob has sugar in his coffee and Bob
has cream in his coffee. When is that T F F
sentence true, and when is it false? All F T F
the cases are already in our truth table,
so lets take them one by one. F F F
The first case has both sentences true. If it is true that Bob has
sugar in his coffee and it is true that Bob has cream in his
coffee, then what do we know about Bob has sugar in his coffee
and Bob has cream in his coffee? We know it is true.
And sentences are true when both sentences joined are true. So,
in the first row of the third column in our truth table, we put a T.

Suppose that he does have sugar but doesnt have cream? What
do we say about the sentence Bob has sugar in his coffee and
Bob has cream in his coffee? It is false.
Similarly, if he has cream but no sugar, for an and sentence to

be true, both constituent atomic sentences must be true. And if
neither are true, then its false.
And is truth-functional because it is possible to construct this

truth table; that is, knowing just the truth-values of p and q, we
can uniquely determine the value for
p&q in every case.
p q p>q
Not all connectives are truth-functional. T T
Consider the word because. It is a
connective. But it is not truth-functional. T F
We cannot construct a truth table for it. F T
Lets try. Well use a greater than (>)
symbol to mean because. F F
What goes in the first row of the third column? If p is true and q
is true, do we know if p because q is true? Let p be The sky is
blue and let q be Albert Einstein was a physicist.
Both p and q are true. Is it also true that the sky is blue because
Albert Einstein was a physicist? No. To determine the truth-
value of because sentences, we need to know more than
just the truth-values of the constituent atomic sentences. So,
because will have no place in our truth-functional language as
a connective.
We will, in fact, have four truth-functional connectives. The first

is negation, which is just our way of saying not. It is our only
monadic, or one-place, connective.

p p
We will use the minus sign () to represent it,
although others use the tilde (~) or a minus T F
sign with a little nib on the end (). Its truth
table is shown at right. F T
Negation flips the truth-value of the sentence it is applied to.
In spoken language, in addition to the simple word not, we

also use phrases like its not the case that or it is false that to
denote negation.
p q p&q
The second is conjunction, which is
logic terminology for and. We have T T T
already seen its truth table.
T F F
In spoken language, we use the F T F
word and as well as as well as, in
F F F
addition to, and also. One that fools
a lot of people is but, which in many
contexts is equivalent to and. I intended to call you back, but
I got distracted. In this sentence, but just means and. The
sentence is true if and only if both atomic sentences are true.
The third connective is the disjunction, or. For the disjunction,

we will use the universal symbol , which comes from the word
vel, which means or in Latin.
Lets see if we can construct the truth p q pq

table. It is another dyadic connective,
so the table starts the same. T T
Lets see if we can fill in the last T F

column. Well start with the middle two F T
rows because they are the easiest.
Lets think about our friend Bob at the F F
coffee machine.

Suppose that the sentence we want to consider is Bob has
sugar or cream in his coffee. Sentence p is Bob has sugar in
his coffee, sentence q is Bob has cream in his coffee, and the
disjunction is Bob has sugar in his coffee or Bob has cream in
his coffee.
Lets take the second case, in which he does have sugar but has
no cream. What do we know about Bob has sugar or cream in
his coffee? It is true. Similarly, if he has cream, but no sugar, the
disjunction Bob has sugar or cream in his coffee is true. If one
or the other is true, then the or sentence is true.
What about the last row? Suppose that Bob has neither sugar nor
cream in his coffee. What do we know about the sentence Bob
has sugar or cream in his coffee? It is false.
For the top row, suppose that Bob has both sugar and cream in
his coffee and you say, Bob has sugar or cream in his coffee.
Are you right or wrong? Is the sentence true or false?
We use the word or to mean both one, the other, but not both
and one, the other, or both. If you are having coffee, you might
ask, Sugar or cream? and your friend could reasonably reply,
Yes, both please. But if your friend is having tea and you offer,
Lemon or cream? then clearly you mean which of the two,
not both.
One is what we call the inclusive sense of or, and the other is the
exclusive or. Which one do we use? It is arbitrary, because once
we pick one, we can use the other connectives to build the other.
It doesnt go away; it just becomes a little more cumbersome to

write. So, purely as a convention, we will choose the inclusive or
because we prefer to be inclusive. Logic is for everyone.

So, the truth table for disjunction, then, p q pq
is as shown at right.
T T T
In spoken language, we often include T F T
the term either when we are signaling
to our audience that we mean the F T T
exclusive or and use just or to mean F F F
the inclusive.
We should either go all the way or not start at all implies that
there are two mutually exclusive possibilities and we have to
select only one. If you like that shade of purple, paint the upstairs
or the downstairs powder room in that color is inclusive, leaving
open the chance to do both.
Another way of saying or is unless. Just like or, we have

inclusive and exclusive senses of unless. When the call comes
in, page me unless Im in a meeting is exclusive. Ill have just
salad, unless they have that soup I love is inclusive. But in both
cases, unless is or.
The last connective is the conditional, which is our if, then

sentences. The sentence that is the if clause is called the
antecedent, and the sentence that is the then clause is the
consequent. We will write the conditional using the right-pointing
arrow (); others use the sideways horseshoe ().
In this truth table, we have a dyadic

p q p q
connectivetwo atomic sentences are
being joinedso it starts off like the T T
last two.
T F
Lets address the last column. It will F T
help to think of this in terms of an
example. Suppose that you make a F F
claim to your friend Jane in the form of

a conditionalfor example, If you beat me in backgammon, I
will give you my car. When do you lie to Jane? When can she
take you to court for failing to uphold a contract?
In the first case, the antecedent p and the consequent q are both
true. Jane did beat you at backgammon, and you did give her your
car. Did you lie? No. So, the first row of the third column gets a T.
For the other easy case, skip down to the bottom line. The
antecedent and the consequent are both false. Jane did not beat
you at backgammon, and you did not give her your car. Did you
lie? No. In two-valued logic, any sentence that is not false must
be true. Because this is not false, it is thus true.
Consider the third line. Jane did not beat you at backgammon,
but you gave her your car anyway. Did you lie? No. You never said
that beating you at backgammon was the only way to get your
car. Maybe Jane bought it from you, or maybe you were feeling
generous and just gave it to her. Regardless of the how and why,
Jane did not beat you at backgammon but did get the car.
You did not violate your agreement, so your if, then sentence is
not false and therefore must be true.
Consider the second line. Here, Jane did beat you at

backgammon, and you refused to give her your car. Does she
have warrant for complaint? Absolutely.
This is the only case in which the p q p q

conditional is false: when the antecedent
is true and the consequent is false. T T T
In spoken language, we have a number T F F

of ways to express the conditional. If F T T
you put your hand on a hot stove, then
you will feel pain: sp. We drop the F F T

word then: If you put your hand on a hot stove, you will feel
pain: sp. We change the order: You will feel pain if you put
your hand on a hot stove: still sp.
The word if picks out the antecedent wherever it is: in front, in

the middle, etc.; wherever you see the word if, you have found
the antecedent.
It is crucial when translating a conditional to correctly identify

the antecedent and consequent. Notice a difference in the truth
tables of the conjunction and disjunction, on the one hand, and
the conditional on the other.
Notice that for and and or, they are symmetric. Paper or
plastic; plastic or paper. Logically, there is no difference. The
truth-value wont change, regardless of the order. But this is not
so with the conditional, where the order does matter.
Consider the following two sentences: If you get shot, then you
bleed and If you bleed, then you get shot. These are two very
different sentences. The first is true, and second is false. So, for a
conditional, unlike for a conjunction or disjunction, order matters.
So, we look for the word if to pick out our antecedent. There are
other stylistic variants for if. When sometimes means if. Yes,
I will go to dinner and a movie with youwhen pigs fly. In other
words, if pigs fly, feel free to make a reservation for two. Given
that and on the condition that are other ways of saying if.
The interesting one is only if. We said that wherever you find the
word if, it picks out your antecedent. The only exception is when
the word if is paired with only. Only if always picks out the
consequent.
Think about the difference between the sentences. There is fire

only if there is oxygen. This means that if there is fire, then there

is oxygen: fo. It does not mean that if there is oxygen, then
there is fire.
The sentence is true; there is oxygen around you but no fire. So, it
could not be translated as of. So, if picks out the antecedent
wherever it is, unless it is only if, in which case it picks out the
consequent.
Readings

Questions
1.
Consider the connective it is the case that, for which we can use the
symbol =. = is a one-place connectivethat is, it works on a single
sentence (for example, =p). Is = truth-functional?
2.
Translate the following into truth-functional logic.
a If I am here and you are here, then all is well in the world.
b Unless it is raining or too hot, we will go to the park and have

a picnic.
c Because I am hungry and you are tired, things will go badly.

If we go home, I will not be hungry and you will not be tired.
Things will not go badly this time. We are going home.

Answers
1.
We can construct a truth-table for =, as shown at right. p =p
That means that we can determine whether =p is true T T

knowing only whether p is true or false. So, = is a truth- F F
functional connective (albeit not a terribly interesting one).
2.
a Let i represent I am here, u represent You are here, and a
represent All is well in the world: (i&u)a.
b Let r represent It is raining, h represent It is too hot, p

represent We will go to the park, and e represent We will
have a picnic: (rh) (p&e).
c Let h represent I am hungry, t represent You are tired, b

represent Things go badly, and m represent We go home:
(h&t)b
m(h&t)
b
Therefore, m.

Lecture 11
Truth Tables
T ruth-functional logic is an artificial logical language designed to

show us what sentences follow from what other sentences. And
as you will discover in this lecture, truth tables are amazing logical
instruments. We can use them to answer a wide range of questions
that we might have about truth-functional sentences. And the answers
they give are absolute and rigorous. With truth tables, we have an
algorithmic method for determining logical properties.
Truth Tables for Connectives

Atomic sentences are basic sentences that are either true or
false, such as The sky is blue. Connectives are words that
connect atomic sentences together into larger, more complex
molecular sentences.
Connectives are said to be truth-functional if and only if the truth-

value of the molecular sentence created using the connective
can be completely and uniquely determined knowing nothing
more than the truth-values of the constituent atomic sentences
and the definition of the connective.
This is equivalent to saying that we can construct a truth table

for the connectives. A truth table contains a row for each of the
possible arrangements of truth-values for all of the constituent
atomic sentences and a listing of the resulting truth-values for
molecular sentences.
It is very nice that we can define our connectives using truth
tables, but this is actually their most trivial use. We can construct
truth tables for any truth-functional sentence, and when we do,
they tell us the truth conditions for that sentencemeaning what
must be the case in the world for the sentence to be true and
what can be the case in the world that renders the sentence false.
For example, consider the following sentence in our truth-

functional language: (pq)&p. The first step in constructing
a truth table is to list all of the atomic sentences in the object
sentence. For this example, its just p and q.
Each new column must contain only what already appears in

the truth table plus one new truth-functional connective. We do
this by finding the main connective and building up the parts it
connects.
In our sentence, (pq)&p, it is the conjunction that applies to

the rest of the sentence. The first conjunct is (pq). It is not in the
table, but p is and q is, so by adding one new connective, , we
can make it.
Once one side of the conjunction is in the table, we can work on

the other side, p, which is not in the table, but p is, so by adding
one connective, , we can build it.
We now have what is on the left side of the and and what is on
the right side of the and in our table, so we can add a column
that joins them.
Then, we can add the possible combinations of truth-values for p

and q: T-T, T-F, F-T, F-F.
We know that pq is true when p is true, q is true, or both are

true. And p always has the opposite truth-value of p.
Lecture 11Truth Tables 99

Finally, we can fill in the values for the last column using the
values from the second and third columns.
p q pq p (pq)&p
T T T F F
T F T F F
F T T T T
F F F T F
It is pq and p, and we know that and sentences are only

true when both conjoined sentences are true. So, we see that the
only time the sentence (pq)&p is true is when p is false and
q is true.
Types of Truth-Functional Sentences

By determining the truth conditions of a truth-functional sentence,
we can tell what type of sentence it is. Truth-functional sentences
come in three categories: tautologies, contradictions, and
contingencies.
A tautology is a sentence that is always true. For example, It

is raining or it is not raining. That is a sentence that is true no
matter what the weather is doing. We can tell when a sentence is
a tautology using truth tables by looking at the column beneath it
and seeing if it is a complete line of nothing but Ts.
By contrast, a contradiction is a sentence that is always false.

For example, It is raining and it is not raining. Because an
and sentence is only true when both conjoined sentences are
true and because the negation of a sentence always has the
opposite truth-value, It is raining and it is not raining must be

false, because when one side is true, the other will be false. In a
truth table, this is seen when the column beneath the sentence is
nothing but Fs.
A contingency is a sentence whose truth-value is contingent on

how the world is. Sometimes it is true; sometimes it is false. A
sentence is a contingency if it is neither a contradiction nor a
tautologythat is, if there is at least one set of truth-values for
the constituent atomic sentences that renders the sentence true
and at least one set of truth-values for the constituent atomic
sentences that renders the sentence false.
The sentence we just used as an example, (pq)&p, is a

contingency because its column has at least one T and at least
one F.
Constructing Truth Tables for Multiple

Sentences
We can construct truth tables for any sentence that allows us to
determine the sentences truth conditions, and this tells us what
type of sentence it is: tautology, contradiction, or contingency.
But we can construct truth tables that include multiple sentences
as well. This allows us to use truth tables to determine whether
certain relations between sentences hold.
One such relation is truth-functional equivalence. Two truth-

functional sentences are logically equivalent if and only if they
have the same truth conditions; that is, they always have the
same truth-valueboth are true or both are falseno matter the
state of the world.
We test for this by constructing a truth table that includes both

sentences and see if the arrangement of Ts and Fs in the two
columns are exactly the same.

Is the sentence a&b equivalent to (a&b)? Using algebra,
can you distribute the negation through the sentence? We need a
truth table that includes both sentences.
First, list the atomic sentences: a and b.
Next, lets build our first sentence. The main connective in a&b
is the conjunction. It conjoins a with b, neither of which is in the
table, but both of which can be made by adding one connective
to something that is already in the table.
Once we have a and b, we can make a&b. That is one of

our sentences. Now, lets build the other. The main connective
in (a&b) is the negation. It applies to the rest of the sentence.
So, we need a&b. It isnt in our table, but a is and b is, so we can
make it.
Once we have a&b, we just negate it in the last column and we

have both sentences we are looking to compare.
Next, we enter the truth-values, just like last time: T-T, T-F, F-T, F-F.
Because a is the negation of a, and b is the negation of b, the

next two columns are as follows.
a: F, F, T, T.
b: F, T, F, T.
The next column is the conjunction of the previous two. Note that
it is only true when both a and b are true: F, F, F, T.
The next column is just the conjunction of a and b: T, F, F, F.
The last column is just the negation of the values we just

determined: F, T, T, T.

a b a b a&b a&b (a&b)
T T F F F T F
T F F T F F T
F T T F F F T
F F T T T F T
So, to see if the two sentences are equivalent, lets see if they
have exactly the same arrangement of Ts and Fs beneath them.
Look at the second and third rows of a&b and (a&b). They
are not equivalent.
But what would happen if instead of (a&b), we compare

a&b to (ab)? Lets change the conjunction from & to in
the appropriate columns. We know what ab looks like. Its only
false when both a and b are false. And the last column is just the
negation of the one we just determined.
a b a b a&b ab (ab)
T T F F F T F
T F F T F T F
F T T F F T F
F F T T T F T
This time, we compare the two columns, a&b and (ab),

and they are exactly alike. Therefore, a&b is truth-functionally
equivalent to (ab).
Intuitively, this makes sense. If we say, Bob doesnt have sugar

and Bob doesnt have cream in his coffee, that is synonymous to
saying, Bob doesnt have sugar or cream in his coffee.

This is exactly what we want. The results of truth-functional logic
match those of our natural way of reasoning, but we have a
precise method of demonstrating it.
In addition to equivalence, another relation between sentences

is consistency. Two sentences are consistent if it is possible for
them both to be true at the same timethat is, if there is some
arrangement of the constituent atomic sentences such that both
sentences are rendered true. Notice that this is an incredibly weak
relation: The sky is blue is consistent with I am a warthog.
We can see this using a truth table. The sky is blue is an atomic
sentence. Lets use s to abbreviate it. I am a warthog is also
atomic, and well use w for it.
The truth table is trivial to set up. There are two atomic sentences,
and we put them in the table.
Next, we add the truth-values.
Because the first line is a case in which both sentences are true,
they are consistent. The sky is blue is also consistent with I am
not a warthog.
s w w
Next, we need an additional column for
w. We know that w has the opposite T T F
truth-value as w, which gives us what is
shown at right. T F T
F T F
In the second row, s is true and w is
true. So, the two are also consistent. F F T
Indeed, The sky is blue is consistent with pretty much

everything except The sky is not blue.

To tell if two truth-functional sentences are consistent, we construct
a truth table with both of them in it and see if there is any rowwe
only need onein which both have the truth-value T.
There is a third relation between sentences called implication. A

sentence S1 truth-functionally implies sentence S 2 if and only if
whenever S1 is true, S2 is also truethat is, there is no case in
which S1 is true and S 2 is false. This does not mean that they both
have all of the same truth-valuesthat is, equivalence. Rather, S1
implies S 2 when S1s truth is enough to guarantee S 2s truth.
Implication is unlike equivalence and consistency in that it is not

necessarily symmetricthat is, just because S1 implies S 2 does not
mean that S 2 implies S1. Indeed, the only time we will have mutual
implication is when we have two sentences that are equivalent.
For example, consider the following two sentences: If today is

Sunday, then Ill visit my parents, unless today is not Sunday and
I will visit my parents. Today is Sunday, or if I visit my parents,
then it is not Sunday.
The first sentence is composed of two atomic sentences: Today

is Sunday and I will visit my parents. Lets use x for the first and
y for the second: If x then y, unless it is not the case that x and
y. The comma shows us that the main connective is unless
which is another way of saying or. So, we have the following.
(if x, then y) (it is not the case that x and y)
The first disjunct is just a conditional.
(xy) (it is not the case that x and y)
The second disjunct is a conjunction.
(xy) (it is not the case that x&y)

It is not the case that x is just x. So, the first sentence is as follows.
(xy) (x&y)
The second sentence is Today is Sunday, or if I visit my parents,

then it is not Sunday. Substitute in the x and y while leaving the
logical terms in place: x, or if y, then not x.
The main connective is the disjunction.
x (if y then not x)
The second disjunct is a conditional.
x (ynot-x)
The consequent is a negation, giving us the following.
x (yx)
So there are our sentences: (xy) (x&y) and x (yx).
What type of sentence is each? Are they equivalent? Are they

consistent? And does either imply the other?
First, we need a truth table for two atomic sentences, x and y.
The first sentence has the disjunction, the or, as its main
connective. On the left side is xy. The right side is x&y. We
have y, but not x, but we do have x, so we can make x.
Next, we can make x&y.
By adding the , we can complete our first sentence:

(xy) (x&y).

In the second sentence, x (yx), the main connective is the
or. The left side is just x, which is in the table already, and the
right side is the conditional yx, and we already have y and x
in the table. So, we need one more column for that, as well as a
final column for the second sentence.
The next step is to add the truth-values.
We know what the conditional (xy) looks like.
The fourth column is just the negation of the first.
The fifth column is just the fourth conjoined with the second.
The sixth column, the first sentence, is the third column or the
fifth column.
The seventh column has the second as the antecedent and the
fourth as the consequent.
The last column, the second sentence, is the first column or the
seventh column.
x y x y x x&y (x y) (x&y) y x x (y x)
T T T F F T F T
T F F F F F T T
F T T T T T T T
F F T T F T T T
When we look at the first sentence, we see that underneath it is

T, F, T, T.

The second sentence has nothing but Ts beneath it. It is a
tautology. Those two columns are not the same, so the two
sentences are not equivalent.
In the first, third, and fourth rows, both sentences have the truth-
value T; remember that we only need one, so they are consistent.
Lets think carefully about implication: Is there a case in which

the first sentence is true and the second sentence is false? No,
because the second sentence is never false.
The first sentence does imply the second. Indeed, if the second
sentence is a tautology, then any sentence would imply it,
because it would be impossible for the first sentence to be true in
a case where the second sentence is false, given that the second
sentence is never false.
What about the other direction? Is there a case in which the

second sentence is true and the first one is false? Yes, this occurs
in the second row. So, the second sentence does not imply the
first.
Readings

Questions
1.
Use a truth table to determine whether the sentence p[q (qp)]
is a tautology, contradiction, or contingency.

2.
Translate the following sentence, construct a truth table for it, and
determine if it is a tautology, contradiction, or contingency.
If you get a pie, then pick up ice cream, but if you dont get ice
cream, dont get a pie.
3.
Consider the following two sentences:
I will pick up the kids from school, you will pick up the kids from
school, or we will both pick up the kids from school
and
If you dont pick up the kids from school, I will.
Are they consistent? Are they equivalent? Does one imply the other?
4.
Consider the following two sentences:
If you are not in love, dont get married
and
If you are in love, get married.
Are they consistent? Are they equivalent? Does one imply the other?

Answers
1.
tautology
p q q q p q (q p) p [q (q p)]
T T F T T T
T F T T T T
F T F T T T
F F T F F T
2.
contingency
(pi)&(ip)
p i p i i p i p (p i)&(i p)
T T T F F T T
T F F T F F F
F T T F T T T
F F T T T T T

3.
The sentences are consistent, equivalent, and imply each other.
(iu)&(i&u) (this could also be translated simply as iu), ui.
i u iu i&u (iu) (i&u) u u i
T T T T T F T
T F T F T T T
F T T F T F T
F F F F F T F
4.
The sentences are consistent. They are not equivalent, and there is
no implication.
lm, lm
l m l m l m l m
T T F F T T
T F F T T F
F T T F F T
F F T T T T

Lecture 12
Truth Tables and Validity
T ruth-functional logic provides us with the tools to assess

arguments surrounding many of the things we believe. Does
the evidence we cite really lead us with absolute certainty to the
conclusions we hold? To find out whether they do, we can use
truth tables. As you know, they can be used to test for properties
of individual sentences and relations between sentences. But our
ultimate goal in logic is to test for validitythat is, to see when we are
logically obligated to believe an arguments conclusion based on the
information contained in the arguments premises. As you will learn in
this lecture, truth tables are able to do this, too.
Validity and Invalidity for

Deductive Arguments
In order to understand the means of testing for validity and
invalidity, recall a relation between truth-functional sentences that
you have already learned how to test for: implication.
A sentence S1 implies a sentence S 2 if and only if in every case

in which S1 is true, S 2 is also truethat is, there is no case
in which S1 is true and S 2 is false. We test for implication by
constructing a truth table containing both S1 and S 2. We then
look at every row in which S1 is true and see if S 2 is also true in
every case.
Validity is just a generalization of implication. An argument is

valid if and only if its premise set implies its conclusion. As such,
we can just generalize our truth table test to turn it into a validity
testand, interestingly, an invalidity test.
It might seem obvious that we should be able to use a single test
to demonstrate that an argument is valid if it passes and invalid if
it fails, but there is another validity test that is not an invalidity test.
But truth tables are able to do both. How?
1 Construct a big truth table that includes all of the premises and
the conclusion.
2 Fill in all of the truth-values.
3 Look for every casethat is, every rowin which all of the
premises are true.
4 See if the conclusion is true in all of those cases. If it is, then
the argument is valid. If there exists even one case in which all
of the premises are true and the conclusion is false, then the
argument is invalid.
Modus Ponens
Consider the famous argument called modus ponens, which is
Latin for the way that affirms. It is any argument of the form
mn; m, therefore, n.
For example, consider the following argument: If you are

drunk, then your cognitive abilities are impaired. You are drunk.
Therefore, your cognitive abilities are impaired.
Let m be the sentence you are drunk and n be the sentence

your cognitive abilities are impaired.
We need a truth table that includes the premises and the

conclusion. The first step in building a truth table is to list all of the
atomic sentences we will need. In this case, it is m and n.
Next, lets look at the first premise: mn. Its not in our table, but
we have both m and n, so by adding just one connective, we can
make it.
Lecture 12Truth Tables and Validity 113

We look at the second premise, m, and see that it is already in
the table, as is the conclusion, n.
Next, add the truth-values. This is just

the truth table for the conditional. m n m n
Once we have the complete truth T T T

table, we can test the argument for T F F
validity. In which rows are both of the
premises true? It is only the first row. Is F T T
the conclusion also true in this case? F F T
Yes. So, the argument is valid.
If it is true that you are drunk and if it is true that being drunk
leads to impaired cognitive abilities, then it must also be true that
your cognitive abilities are diminished.
The Fallacy of Affirming the Consequent

Modus ponenss evil twin is known to logicians as the fallacy of
affirming the consequent. It is any argument of the form mn; n,
therefore m.
Using the same sentences as the previous example, the argument

would be as follows: If you are drunk, then your cognitive abilities
are impaired. Your cognitive abilities are impaired. Therefore, you
are drunk.
The name probably gives away the validity status, but lets see
whether this so-called fallacy really is an invalid argument form.
We need a truth table that includes both of the premises and

the conclusion. This one is easy for us because it is just the
truth table we constructed for the previous example. The

only difference is that we have m n m n
switched the conclusion and one
of the premises. T T T
T F F
To test for validity, lets find every
row in which both of the premises F T T
are true. In both of the first and
F F T
the third rows, the conditional if
you are drunk, then your cognitive
abilities are decreased is true, and in both of these cases you
do have decreased cognitive capacities.
In the first case, you are drunk, but in the third case, you are not
drunk but have decreased cognitive capacities for some other
reasonperhaps lack of sleep, for example.
Is the conclusion also true in both of those cases? It is for the first
row, but not for the third row.
Note what this means: If we are presented with this argument and
we know for a fact that both of the premises are trueit is the
case that being drunk diminishes your cognitive abilities, and you
do, in fact, have diminished cognitive abilitiesdo we know with
absolute certainty whether you are drunk? That is, do we know if
the conclusion is also true?
No, because we dont know if our world is the first case or the
third case. In both of these rows, the premises are true, but this is
not enough for us to know which one is our case and, therefore,
not enough to know whether the conclusion is true or false.
This argument is invalid. It does not give us reason to either

believe or deny the conclusion. This argument gives us reason
to believe absolutely nothing. The fallacy of affirming the
consequent lives up to its name.

Three Atomic Sentences
The following example is a bit more complicated than the previous
two. If there is precipitation and the temperature is below freezing,
then there is snow. There is not snow or the temperature is below
freezing. Therefore, there is precipitation or there is snow.
Let a be the sentence there is precipitation. Let b be the

sentence the temperature is below freezing. And let c be the
sentence there is snow.
This gives us the argument form (a&b)c; cb, therefore ac.

There are three atomic sentences. How do we build a truth table
for this? The same way as we do with two. The first step is to list
all of the atomic sentences.
Next, lets work on the first premise. The main connective is the
conditional. The antecedent is a&b. We already have both a and
b in the table, so we can make a&b by adding one connective.
We have c, the consequent, already in the table, so by adding

the arrow to what is already in the table, we can make the first
premise: (a&b)c.
In the second premise, the main connective is the or. On the left
side, we see c. We have c, so one connective added makes it.
Once we have the left side and the right side of the disjunction in
the table, we can make the second premise: cb.
Finally, we have a and c already in the table, so we can add one

connective to make the conclusion: ac.
Once we set up the columns, we can add the truth-values. We

have three atomic sentences that can take two possible truth-

values, so the number of combinations is 2 2 2 = 8. There
are 8 rows in each column. The first three columns are as follows.
a b c a&b (a&b) c c cb ac
T T T
T T F
T F T
T F F
F T T
F T F
F F T
F F F
The fourth column is just the conjunction of the first two.
The fifth column is a little tricky. Remember that the only time
the conditional is false is when the antecedent is true and the
consequent is false. But notice that in this case, the antecedent is
listed after the consequent in the table, so we need to be mindful.
For the sixth column, c is just the negation of c.
The seventh column is a disjunction of the sixth and second

columns.
Finally, the conclusion in the eighth column is the disjunction of

the first and third columns.

a b c a&b (a&b) c c cb ac
T T T T T F T T
T T F T F T T T
T F T F T F F T
T F F F T T T T
F T T F T F T T
F T F F T T T F
F F T F T F F T
F F F F T T T F
Now that our truth table is filled in, lets find every case in which
both of the premises are true. It is rows 1, 4, 5, 6, and 8. Is the
conclusion true in all of these cases? Lines 6 and 8 show us
cases in which the premises are both true and the conclusion is
false. So, this is an invalid argument.
Readings


Questions
1.
Is the following argument valid?
pq. p q. Therefore, pq.
2.
Determine whether the following argument is valid using a truth table.
Your mother will stay with us only if we kennel the dog and get all
of the carpets in the entire house steam cleaned. I am not going
through the whole hassle of getting the house prepared for steam
cleaning again. So, tell your mother she is not staying with us.
Answers
1.
invalid (row 4)
p q p q p q p q pq
T T T F F F T
T F F F T T T
F T T T F T T
F F T T T T F

2.
valid
m(k&c). c. Therefore, m.
m k c k&c m (k&c) c m
T T T T T F F
T T F F F T F
T F T F F F F
T F F F F T F
F T T T T F T
F T F F T T T
F F T F T F T
F F F F T T T

Lecture 13
Natural Deduction
L ogic determines what we have good reason to believe. We have

legitimate warrant to accept the conclusion of an argument if that
argument is both well-grounded and valid. To protect ourselves, we
need a validity test. For arguments in our truth-functional language,
we have truth tables. In this lecture, you will be introduced to a
different validity test: natural deduction proofs. Why introduce a
second validity test when we already have a perfectly good one?
Truth tables are not always user-friendly, and our truth-functional
language is insufficient for all of the logical work we want to do.
Natural Deduction Proofs

Think of a natural deduction proof as a game. First, look at the
playing board. Then, learn the goal we need to accomplish in
order to win and the rules we have to follow in seeking that goal.
Finally, learn strategies for successful playing.
When it comes to the playing board, a natural deduction proof

has three columns. The first column is comprised of positive
integers in ascending order. Every line has a number, which is
the name of the line.
The second column includes nothing but true sentences written

in our truth-functional language. If a single false sentence gets
into the second column, it could create logical havoc. The second
column is, therefore, a very exclusive logical club.
For a sentence to move past the velvet rope and gain admittance
to the second column, it needs an identification card that vouches
for its undeniable truth. That identification card is a justification,
and it is placed in the third column.
So, the game of natural deduction proof is played on a board that

has numbered and justified true sentences written in our truth-
functional language. First comes the number, then the sentence,
and then the justification.
There are several categories of justification, but the first one is

premise.
Natural deduction proof is a validity test, and an argument is valid

if and only if, assuming the truth of the premises for the sake of
argument, the conclusion must follow from them.
We are trying to show that if the premises are true, then the
conclusion has to be, too. So, we begin by admitting the
premises to the second column. Being a premise is like being
born to a wealthy family: You get privileges you did not have to
earn, unlike others.
All other sentences will need to earn the justification that will
let them into the second column, but the premises are simply
granted justification by virtue of the argument whose validity is in
question.
The game begins when all of the premises are entered onto the
board as numbered lines, justified by premise. The game ends
when the conclusion appears as a justified line of the proof.
Because only true sentences are allowed in the second column,

if we start with the premises and show that the conclusion follows,
then we will have, indeed, proven the validity of the argument.

The Rules of Inference
To get us from the premises to the conclusion, we need some
moves. The first set is rules of inference, which are simple valid
argument forms. If we have lines in the proof that have the form of
the premises, then we can write a new line in the proof of the form
of the conclusion of this valid argument form.
Because the form is valid, we know that if the premises are true,
the conclusion must be and therefore can be entered into the
second column.
We have already seen one of these rules of inference in the

previous lecture. Recall the argument form modus ponensthat
is, ab; a, therefore b.
If we have a conditional that we know is true and we also know

that the antecedent is true, then we can conclude that the
consequent must be true as well.
We can use modus ponens if we have a conditional on some

line of the proof and on any other line of the proof we have the
antecedent of that conditional.
If these two lines appear, then on a subsequent line, we can write

the consequent of the conditional in the second column and
next to it, in the third column, write MP for modus ponens and
write the numbers of the two lines on which the conditional and
antecedent are found.
Lets work out a proof for the following argument.
1 ab
2 bc
3 cd
4 de
Lecture 13Natural Deduction 123

5 ef
6 fg
7 a therefore, g
We start by listing all of the premises as numbered lines of the

proof, justified by premise. The proof therefore begins as follows.
1 ab premise
2 bc premise
3 cd premise
4 de premise
5 ef premise
6 fg premise
7 a premise
We look at line 1, and we have a condition ab, and then we see

on line 7 that we have a. If a, then b, and we know a, so on line 8
we can write the following.
8 b MP 1,7
The justification must include the name of the rule of inference

used and the lines on which it is being used. This says that we
used modus ponens on lines 1 and 7 in order to get this line, line 8.
But look at line 2. It says that if we have b, we get c, and our new
line 8 gives us b. So, we can add the following.
9 c MP 2,8
And this gives us lines 10 through 13.
10 d MP 3,9
11 e MP 4, 10
12 f MP 5, 11
13 g MP 6, 12

At this point, lets look at what we have. The conclusion g showed
up as a justified line of the proof. This means that the proof is
complete. We win the game. We declare the argument valid by
writing beneath the proof QED, which stands for quod erat
demonstrandum, or which was to be demonstrated. With this
proof, we have proven the validity of the argument.
The less optimistic sibling of modus ponensthe way of

affirmingis modus tollens, or the way of denying. If we have,
on any line of a proof, a conditional, and if on any other line of
the proof, we have the negation of the consequent, then we may
write on a further line of the proof the negation of the antecedent,
justified by modus tollens (MT). In other words, we are using the
valid argument form.
ab
b
Therefore, a
If a is true, then b must be true. But we know that b is not true, so

a cannot have been, or else b would be. This is valid.
The other rule for sentences that are conditionals is the

hypothetical syllogism (HS). Recall that a syllogism is an argument
with two premises. A hypothetical syllogism is an argument with
two conditionals as premises.
ab
bc
Therefore, ac
If we have two conditionals in a proof and the antecedent of one

is the consequent of the other, we can combine them, eliminating
the middle term. We justify the new line with HS and the line
numbers of the operative conditionals.

This gives us another way of producing a proof from the original
example.
1 ab premise
2 bc premise
3 cd premise
4 de premise
5 ef premise
6 fg premise
7 a premise
8 ac HS 1,2
So, we took the conditional on line 1 and the conditional on line

2 and squished them together. If we have a, then we get b. But b
gives us c. So, a gives us c.
But this can be done again and again.
1 ab premise
2 bc premise
3 cd premise
4 de premise
5 ef premise
6 fg premise
7 a premise
8 ac HS 1,2
9 ad HS 3,8
10 ae HS 4,9
11 af HS 5,10
12 ag HS 6,11
But we now have line 12 that tells us that if a, then g, and line 7
that gives us a. So, by modus ponens, we get the following
13 g MP 7,12

This is not uncommon. Many arguments will have multiple ways
of constructing a proof.
So far, all of the examples have used atomic sentences as the

antecedent and consequent, but this is not necessary. This was
only done to make the examples easier.
You can use modus ponens, modus tollens, and the hypothetical
syllogism on lines that are as complex as you want, as long as
the main connective is the conditional and the form holds.
Modus ponens just requires a conditional on one line and the

antecedent on another. They dont have to be atomic sentences
but could be molecular sentences of any complexity.
With the hypothetical syllogism, we match up the middle term,

and the antecedent stays the antecedent and the consequent
stays the consequent, even if the terms are complex molecular
sentences.
If we have some sentence on line m and another on line n, then

we know that both are true because all of the sentences in
the second column are true. But if they are both true, then the
conjunction of the two must also be true.
So, we can take the sentences on any two lines and put an and
between them, justifying the move with the rule of inference
called conjunction (Conj). Suppose that we have the following.
8 (dh) (e&t)
9 i&(fg)
Then, we can make the following line.
10 [(dh) (e&t)]&[i&(fg)] Conj 8,9

The conjunction rule allows you to join any two lines with an and.
Conjunction introduces an and. To get rid of one, you use the

rule called simplification (Simp).
We know that the only time a conjunction a&b is true is when a

is true and b is true. So, if you have a conjunction in your proof,
you can break it down and write both conjuncts on their own lines,
justified by simplification. Suppose that we are given the following.
8 (sr)&(xt)
Then, we can write the following.
9 sr Simp 8
10 xt Simp 8
Use simplification anytime you can. You will often be able to do

things with the parts that you cannot do with the whole. Break it
down and free up as much logical fodder as you can.
Be careful, however: We can only use simplification if the

conjunction is the main connective. You cannot simplify parts of
lines, only entire lines. In general, rules of inference must only be
used on entire linesthat is, operate on the main connective of
the sentences.
We also have rules for introducing and removing disjunctions. To

introduce one, we use the rule called addition (Add). Given any
line, we can turn it into a disjunction with any sentence. Suppose
that we are given the following.
23 t

We can form the following.
24 tm Add 23
Where did m come from? Why m? If you dont like m, you can
make it w or even (c&f ) (q&j). You can add any sentence to
any other sentence just because you want to.
But we said that the sentences in the second column have to be

true or else really bad things happen. How can we just add any
sentence?
Think about disjunctions: Or sentences are true when one or

the other (or both) disjuncts are true.
While we can introduce any sentence we wanttrue or falseit

will be stuck in the disjunction. It will not be able to be freed and
appear on its own.
To get a sentence out of a disjunction, we use the disjunctive

syllogism (DS). If we know that Bob has sugar or cream in his
coffee and we know he doesnt have cream, then we know he
has sugar. If we know that Bob has sugar or cream in his coffee
and we know he doesnt have sugar, then we know he has cream.
That is the disjunctive syllogism.
If we have a line whose main connective is an or, and we know

that one of the disjuncts is false, then the other one has to be
true. It looks like the following.
5 (c&d)f
6 f
7 c&d DS 5,6
You can use the disjunctive syllogism if either disjunct is negated.

The order does not matter.

The last rule has
three premises and is The Rules
called the constructive of Inference
dilemma (CD). It is
an argument of the
following form. MP ab; a, therefore b
ab MT ab; b, therefore a
cd
ac HS ab; bc, therefore
Therefore, bd ac
The idea is that a takes Conj a, b, therefore a&b

you to b, and c takes
you to d. You know Simp a&b, therefore a,
that either a or c is therefore b
true, which means that
you will be led, then, Add a, therefore ab
to either b or d. So, if
you have two unrelated DS ab, a, therefore b or
conditionals, see if ab, b, therefore a
there is a disjunction
of the antecedents. CD ab, cd, ac,
If so, you can use the therefore bd
constructive dilemma.
Readings

Hurley, Logic, chap. 7.

Questions
1.
Provide a proof for the following argument.
p (r&q). p&(qs). Therefore, r&s.
2.
Translate the following argument into truth-functional logic and
construct a proof to show that it is valid.
If you have an identification card showing that you are older than
21, then either you are of legal age or this ID is a fake. I see your
card, but I know that you are not 21. Hence, this must be a fake ID.
Answers
1.
1 p (r&q) premise
2 p&(qs) premise
3 p Simp 2
4 qs Simp 2
5 r&q DS 1,3
6 r Simp 5
7 q Simp 5
8 s MP 4,7
9 r&s Conj 6,8

2.
c: You have an identification card showing that you are older than 21.
o: You are older than 21.
f: The identification card is fake.
If you have an identification card showing that you are older than 21,
then either you are of legal age or this ID is a fake.
c(of )
I see your card, but I know that you are not 21.
c&o
Hence, this must be a fake ID.
Therefore, f.
1 c(of) premise
2 c&o premise
3 c Simp 1
4 o Simp 1
5 of MP 1,3
6 f DS 4,5

Lecture 14
Logical Proofs
with Equivalences
T he system of proof we have is sound, but it is not complete. A

system of proof is sound if and only if you can only construct a
proof for valid arguments. Our system as it is so far must be sound
because the only moves we can use are rules of inference, which
are themselves valid arguments. But the system is not complete.
A system of proof is complete if and only if every valid argument
formable in the language can be proven in the system. As the system
stands now, there are valid arguments that we are unable to prove.
We need more moves. We call these equivalences.
The Equivalences
Equivalences are sentence forms in our truth-functional language
that are truth-functionally equivalentthat is, sentences that must
always have the same truth-value. For any of these, if you doubt
that they are really equivalent, work out a truth table, and you will
see that they are.
The first equivalence is called double negation (DN). It asserts

that a sentence p and the double-negated sentence p are
equivalent. It is not the case that you didnt call your mother
means that you did call your mother.
Two negations yield an affirmation. Negation changes the truth-

value of a sentence from true to false or false to true. So, doing
it twice just brings you back to the original truth-value. Therefore,
you could always add two negations or remove a pair of adjacent
negations without changing the truth-value of the original
sentence. For us, this makes the two equivalent.
Suppose that we have a proof with the following lines.
6 ch
7 h
We have a conditional and the negation of the consequent, so we

can use the rule of inference modus tollens to get the negation of
the antecedent in the proof. The antecedent is c, so we get the
following.
8 c MT 6,7
But if the conclusion we are looking for is c, then we would need

one more step.
9 c DN 8
Because c and c are equivalent, if c appears in the second

column, then it must be true, and therefore c must be true
and thus can also be entered in the second column. If there
are two negations directly next to each other, not separated by
parentheses, we can wipe them away.
However, in some circumstances, we will need them. In these

cases, we can use DN to add them. Suppose that we have a
disjunction in a proof.
12 b d
Suppose that we also have the sentence d in the proof.
13 d

This means that we know that b is true or d is false. But we know
that d cannot be false because line 13 says that it is true. We
would like to use the disjunctive syllogism here, but to use the
rule DS, we need a disjunction and the negation of one of the
disjuncts. We do not have that.
But by using DN, we can manufacture it.
14 d DN 13
Next, we do have the negation of one of the disjuncts. One is b

and the other is d. The negation of d is d. And that is what
we have on line 14. So, we can now do the following.
15 b DS 12,14
So, double negation can be used to add or take away two

negations that are directly next to each other anywhere in a proof.
There are two important differences between the rules of

inference and the equivalences. First, equivalences work in both
directions. We can substitute c for c, or c for c. Modus
ponens, for example, is an inference, so it only goes one way.
The second difference is that we can use equivalences on parts

of lines. Rules of inference can only be used on entire lines.
The reason for this is that while rules of inference are truth-
preservingthat is, they always take you from a true sentence to
another true sentenceequivalences always maintain the truth-
value of the sentence, true or false.
Truth-functional sentences derive their truth-values solely from

the values of the constituent parts. Some of these parts may be
true; some may be false. Equivalences always leave the truth-
value unchanged and therefore can be used on parts of lines,
where rules of inference cannot.
Lecture 14Logical Proofs with Equivalences 135

One similarity is that although we will be using atomic letters like a
and b to show the equivalences, they may be used on molecular
sentences or any level of complexity. We could take any line in a
proof, atomic or not, and apply any of our equivalences.
The second equivalence is a pair of relations called De Morgans

theorem (DeM), named for the 19th-century logical pioneer
Augustus De Morgan. It says that the negation of a disjunction is
a conjunction of the disjuncts and the negation of a conjunction is
a disjunction of the conjuncts.
(ab)::a&b
(a&b)::a b
When you negate an or, you get an and, and when you negate
an and, you get an or.
Think of this in terms of Bob at the coffee machine. If we say that

Bob has sugar or cream in his coffee, how could we be wrong?
The only way we are wrong about Bob having sugar or cream is
when he has neitherthat is, when he doesnt have sugar and he
doesnt have cream.
If we say that Bob has sugar and cream in his coffee and again
are wrong, what could be the case? We would be wrong about
his having both sugar and cream in his coffee if he didnt have
sugar or he didnt have cream, or he had neither. This is De
Morgans theorem.
We cannot distribute a negation, but we do have an equivalence

called distribution where we can distribute an and or an or.
a (b&c)::(ab)&(ac)
a&(bc)::(a&b) (a&c)
We can distribute the a or the a& through the other sentence.

Notice an important aspect of this equivalence. One side is
a disjunction and the other is a conjunction. A helpful hint
when doing proofs is that anytime you have the negation of a
disjunction, use De Morgans theorem as quickly as possible,
because it results in a conjunction upon which we can use the
rule of inference simplification and get new pieces we might be
able to use.
Distribution has two sibling equivalences, association and

commutation. Where distribution is used when you have mixed
conjunctions and disjunctions, association works where you have
two ands or two ors.
a&(b&c)::(a&b)&c
a (bc)::(ab)c
This is association (Assoc). For an and sentence, the only time

it is true is when both of the conjuncts are true. So, a&b is true
when a is true and b is true, and (a&b)&c is true only when (a&b)
is true and c is truethat is, when all three are true.
Similarly, when we regroup them, the only time the sentence is

true is when all three are true. They are equivalent. In the case
of the disjunction, the only time the sentence is false is when a,
b, and c are all false, no matter how they are grouped. So, if we
have two ands or two ors, we are free to shift the parentheses
how we see fit.
The last of the trio is commutation (Comm), where we are taking

the terms of a conjunction or disjunction and moving them
somewhere else.
ab::ba
a&b::b&a

Remember that in basic truth tables, conjunction and disjunction
are symmetric. This is just a result of that fact.
But remember that the conditional was not. The only time a
conditional is false is when the antecedent is true and the
consequent is false.
So, lets suppose that we have two sentences, a, which is true,

and b, which is false: ab is true then false, and thus is false.
But ba gives us false then true, and that case is true. So, if we
try to take the antecedent and the consequent and swap their
places, we change the truth-value: a then b and b then a are not
equivalent.
We cant switch around the antecedent and the consequent

and maintain the truth-value of the resulting sentence, so while
commutation holds for conjunction and disjunction, it does not for
the conditional.
The conditional gets its own version of this, called contraposition

(Contra), as in contrary (a change in truth-value) and position
(a change in location).
ab::ba
If we know that ab is true, then if you get a, b must follow.

So, if there is no b to be found, there could not have been an
a in the first place. If there is fire, then there is oxygen. In other
words, if there is no oxygen, then there is no fire. We can swap
the antecedent for the consequent in a conditional as long as we
negate both in the move.
The conditional also has its own version of distribution, which is

called exportation (Exp).
(a&b)c::a(bc)

Suppose that we need two things to happen to bring about a
thirdfor example, we need for the temperature to be below
freezing and for there to be precipitation in order to get snow.
That means that if the temperature is below freezing, then if there
is precipitation, it will snow.
So, we can shift those parentheses the way we do with

conjunctions and disjunctions using the rule association, but only
if we change the first arrow to a conjunction or vice versa.
That is the equivalence that the conditional and the conjunction

share.
The disjunction and the conditional share one as well. It is called

implication (Impl).
ab::ab
If you say to one of your teenage children, If you are going to

go to that party, you will complete your chores, then what you
are saying is that the teenagers chores will be complete or the
teenager will not be attending that party.
The key here is to think of the truth table for the conditional. The
only time a conditional is false is when the antecedent is true and
the consequent is false. That means that a conditional is true if
that case is not the casein other words, if the antecedent is
false or the consequent is true.
And that is just what implication says. This means that we can
switch back and forth between conditionals and disjunctions as
long as we negate the first term.

The last equivalence is given the misleading name tautology
(Taut).
aa::a
a&a::a
It is fairly obvious that a or a or a and a is the same thing

as saying a: Tonight, lets have either Indian food or Indian food.
Actually, lets have both Indian food and Indian food.
So, anytime you have a conjunction or a disjunction of the same

term, you can collapse it.
Of course, aa and a &a are not tautologies. They are

contingencies. If a is true, then a&a and aa are true. If a is
false, then a&a (false and false) and aa (false or false) are both
false. On the other hand, aa is not equivalent to a. Why?
Because it is a tautology. If we have a sentence a, then aa

will always be true, regardless of the truth-value of a. If a is true,
then aa is true, then true, and that is true. If a is false, then
aa becomes false, then false, and this is also true. The only
time a conditional is false is when the antecedent is true and the
consequent is false.
If the antecedent and the consequent are the same sentence,

then this could never be the case. As such, aa and a&a are
equivalent to a, and using the equivalence tautology, we can
substitute one for the other at will.
But a and aa are not equivalent because a could be false and

aa would be true. So, the equivalence only holds for and
and or.

The Equivalences
DN a::a
DeM (ab)::a&b; (a&b)::a b
Dist a (b&c)::(ab)&(ac); a&(bc)::(a&b) (a&c)
Assoc a (bc)::(ab)c; a&(b&c)::(a&b)&c
Comm ab::ba; a&b::b&a
Contra ab::ba
Exp (a&b)c::a(bc)
Impl ab::ab
Taut aa::a; a&a::a
Readings

Questions
1.
Construct a proof for the following argument.
(p q)r, p&r. Therefore, qs.

2.
Translate the following argument and construct a proof to show that
it is valid.
I need to be in Atlanta for work, but we could also visit Boston or

Chicago. If we fly through Denver, then it makes sense to go to
Chicago. I suggest that we give Chicago a miss this time. So, well
go to Atlanta and Boston and not fly through Denver.
Answers
1.
1 (p q)r premise
2 p&r premise
3 p Simp 2
4 r Simp 2
5 p (qr) Assoc 1
6 qr DS 3,5
7 q DS 4,6
8 qs Add 7
9 qs Impl 8
2.
I need to be in Atlanta for work, but we could also visit Boston or
Chicago.
a&(bc)
If we fly through Denver, then it makes sense to go to Chicago.
dc

I suggest we give Chicago a miss this time.
So, well go to Atlanta and Boston and not fly through Denver.
Therefore, a&(b&d).
1 &(bc) premise
2 dc premise
3 c premise
4 d MT 2,3
5 (a&b) (a&c) DeM1
6 c a Add 3
7 a c Comm 6
8 (a&c) DeM 7
9 a&b DS 5,8
10 (a&b)&d Conj 4,9
11 a&(b&d) Assoc 10

Lecture 15
Conditional and
Indirect Proofs
W e have been developing a system of natural deduction. We

want a system that is both sound and complete. A system of
proof is sound if every argument for which there is a proof is valid.
Our system so far is sound. A system of proof is complete if every
valid argument has a proof in the system. Our system is not yet
complete. There are valid arguments that our rules of inference and
our equivalences are not enough to complete a proof. We need two
new proof strategies and a new justification to reach that ultimate
goal. The justification is assumption.
Assumption as a Justification
At this point in the course, you should be flabbergasted at the
thought of assumption being a justification. How can we enter a
sentence in the second column that is a mere assumption?
After all the time we spent discussing why it was essential to only
put sentences in the second column that we know to be absolutely
true, do you mean to tell us that now we can just enter any
sentence in the second column and justify it as an assumption?
Surely, there are limits to what can be assumed, and those limits
are based on rational inferences about what we can know.
No. You are free to assume any sentence you want and enter
it into the second column of the proof. We did say clearly and
explicitly that if even one false sentence shows up in the second
column, logical chaos could result.
But it is okay. We have protection. We have boxes. When we
introduce an assumption (Assumpt) into a proof, we put it in a
box, and that box is logical quarantine. Nothing inside of the box
is allowed to come out into the general population.
Anything can be brought into the box, but once a sentence has
been in the box with the assumption, anything that inferred from it
is to be deemed to be contagious in that it could be infected with
the possible falsity of the assumption.
When we open a box and insert an assumption, we create a new

logical worldone that resembles our logical world in certain
ways, so we can infer true propositions in the real world from
what we observe in the hypothetical box world. One way to draw
such inferences is called conditional proof.
Conditional Proof
We use conditional proof when we want to prove a conditional.
We use this form of reasoning all the time, especially if we have
children. There are two ways one can learn lessons in life: the
easy way and the hard way. The sentence You should not put
your hand on a hot stove can be learned the hard way by putting
your hand on the stove.
The easy way is conditional proof. We say to the person who

has never yet put a hand on the hot stove, Before you do, lets
think about this. Lets start by assuming that you put your hand
on the hot stove. Dont really do it in the real world, but create a
hypothetical world in which you did. What would be the result?
The temperature of the stove is significantly higher than that of

your hand. By the first law of thermodynamics, heat would flow
rapidly from the stove into your hand. Your hand heats up very
quicklyso quickly that it would cause damage to the skin.
Lecture 15Conditional and Indirect Proofs 145

You would suffer burns. The nerves in your hand would send
message of those burns to your brain, which would register it as
incredible amounts of pain.
What do we know from this thought experiment? We know that

if you put your hand on a hot stove, then you will experience
incredible amounts of pain. By assuming the antecedent
and then deriving the truth of the consequent, we can assert
the conditional If you put your hand on a hot stove, you will
feel incredible amounts of pain. We learned the truth of the
conditional the easy way.
This is how conditional proof works. To demonstrate a

conditional by conditional proof, the first step is to enter all
of the premises into the proof. The second step is to open a
box and insert the antecedent into the box as an assumption
justified by Assumpt.
Next, pull your premises into the box as needed. Then, use rules
of inference and equivalences. Proceed as if everything were
normal inside of the box, until such time as the consequent of the
conditional appears as a justified line inside of the box.
At that point, we know that if the antecedent is true, then

the consequent has to be true. So, close the box, and in the
general proof, write down the conditional justified by CP m,n.,
where m is the first line inside the box and n is the last line
inside the box.
Indirect Proof
One use of assumptions is in conditional proof. The other is
called indirect proof.

When we think of proofs in mathematics, we think of Euclidean
geometry, which are proofs of the type we started with, where
you assume the premises and demonstrate the conclusion. This
is called direct proof.
But most mathematical proofs do not take this form. Most are
indirect proofs, or to use the Latin name, reductio ad absurdum,
or reduce to absurdity.
The absurdity is a contradiction. Recall how much we worry

about contradictions. If even one contradiction is true, then
everything is true, and truth goes away. We fear contradictions.
But, like early humans who learned to harness the power of
otherwise dangerous fire for cooking, we, too, will learn to control
the power of contradictions.
The fundamental principle behind traditional logic is the law of

the excluded middlethat all sentences are either true or false.
A sentence that isnt true is false, and a sentence that isnt false
is true.
So, if we want to prove that a sentence has to be true, that is the

same thing as proving that it cant be false. This is the trick. We
want to prove that p cannot be false. But how?
We know that contradictions are always false, so if we can show

that the negation of p, in conjunction with the premises we are
asserting to be true, necessarily leads to a contradiction, then we
have logical grounds on which to reject the negation of p.
But if p is false, then p must be true. If we can show that denying

p leads to absurdity, then we have no choice but to accept p.
That is indirect proof.

It begins when we enter the premises into the proof, then open
a box and insert the negation of what we are trying to prove into
the box. We use our premises, the assumption, and our rules of
inference and equivalences until a contradiction, any sentence of
the form a&a, appears in the box.
It does not have to be a contradiction involving the assumption.

Any contradiction will do, because, as you have learned, if you
grant even a single contradiction, all sentencesincluding all
contradictionsare true.
At this point, we have shown that the negation of the conclusion,

when added to the premises, necessarily results in contradiction,
so we close the box, and in the general proof, we write down the
conclusion, justified by IP with the line numbers from the opening
to the closing of the box.
In a conditional proof, the box was a hypothetical logical world in

which we posit something that might not be true in order to see
what else would have to be true.
In an indirect proof, inside the box is a bizarre logical world in

which we put propositions together that we believe cannot be put
together in order to observe the nonsense that results.
If you are particularly clever, you might have realized that

it might not be the case that the negation of the conclusion is
responsible for the contradiction. All we have shown is that the
combined set of the premises and the negated conclusion lead
to a contradiction.
How can we assert that the negation of the conclusion is

necessarily false when it might not be the conclusion that is
to blame for the derived contradiction? If the fault is with the
premises and not the negation of the conclusion, what gives us
the right to assert the conclusion?

The answer is that if it is not the assumption that is to blame for
the appearance of the contradiction, then it has to arise from the
premises alone. That means that the premises are inconsistent
that they cannot all be true at the same time.
An argument with inconsistent premises must be valid, because

it would be impossible for all of its premises to be true and its
conclusion to be false. So, on a technicality, we know that the
argument is valid. So, whether the contradiction comes from the
assumption or not, the derivation of the contradiction guarantees
that the argument is valid.
Readings

Questions
1.
Use conditional proof to show that the following argument is valid.
s(y&z). (wq)(y&x). Therefore, (sq)(yw).
2.
Use indirect proof to show that the following argument is valid.
(al)&(am). a(t&l). Therefore, (tm) l.

3.
Translate the following argument and construct both a conditional
and an indirect proof to demonstrate its validity.
I cant eat turkey or pasta without overeating. So, if I eat turkey, I will
eat turkey and overeat.
Answers
1.
1 s(y&z) premise
2 (wq)(y&x) premise
3 sq Assumpt
4 (sq)w Add 3
5 s (qw) Assoc 4
6 s (wq) Comm 5
7 (y&z) (y&x) CD 1,2,6
8 y&(zx) Dist 7
9 y Simp 8
10 yw Add 9
11 (sq)(yw) CP 310
2.
1 ln premise
2 (n&t)m premise
3 [(tm) l] Assumpt
4 (tm)&l DeM 3
5 (tm) Simp 4
6 l Simp 4
7 t&m DeM 5
8 t Simp 7
9 m Simp 7
10 n MP 1,6

11 n&t Conj 8,10
12 m MP 2,11
13 m&m Conj 9,12
14 (tm) l IP 313
3.
t: I eat turkey; p: I eat pasta; o: I overeat.
(tp)o; therefore, t(t&o).
Conditional proof
1 (tp)o premise
2 t Assumpt
3 tp Add 2
4 o MP 1,3
5 t&o Conj 2,4
6 t(t&o) CP 25
Indirect proof
1 (tp)o premise
2 [t(t&o)] Assumpt
3 [t (t&o)] Impl 2
4 t&(t&o) DeM 3
5 t&(t&o) DN 4
6 t Simp 5
7 (t&o) Simp 5
8 t o DeM7
9 tp Add 6
10 o MP 1,9
11 t&o Conj 6,10
12 (t&o)&(t&o) Conj 7,11
13 t(t&o) IP 212

Lecture 16
First-Order Predicate Logic
T ruth-functional logic allowed us to account for arguments where

the relevant logical structure occurred between sentences.
The truth-functional connectives showed us the truth conditions
of complex combinations of sentences. But some of the logically
relevant elements of our arguments occur inside of the sentences.
Truth-functional logic is too broad to get at this. We need a new,
sharper logical language that builds on truth-functional logic but
allows us to pull the logical content from within sentences. This new
language is first-order predicate logic.
The Need for First-Order Predicate Logic

Think about the early lectures about logic. The standard argument
we used as an example was the following classic argument.
All men are mortal.

Socrates is a man.
Therefore, Socrates is mortal.
We said that it is valid. Can we use our truth-functional apparatus

to demonstrate this? Let a represent All men are mortal, b
represent the sentence Socrates is a man, and c stand for
Socrates is mortal. Can we give a proof from the premises a
and b to the conclusion c? No.
Truth-functional logic is also known as sentential logic because it

is the logic of sentences and the connectives that connect them.
It is, in a certain sense, a very blunt weapon, because it can only
deal with the logical relations between sentences.
What we see in the Socrates example is the need to apply
logical machinery inside of sentences. We do need to be able
to evaluate the relations between sentences, but we also need
the much finer work done of illuminating the logical relationships
within sentences.
Think about Aristotles categorical logic, in which we have

inferences that concern the content inside of the grammatical
structure of the individual sentences. Truth-functional logic
cannot get at the aspects that Aristotle did.
So, for a period, there was a running argument as to which is the

real logic: Aristotles categorical approach or the truth-functional
approach.
That debate ended when we formulated a new logical language

that combined the two together. This new hybrid language is
first-order predicate logic. It contains all of the elements of truth-
functional logicall of our connectives, exactly as we have
been using thembut adds some new elements that allow us
to investigate the logical content of the grammar of individual
sentences instead of simply giving them a truth-value.
Elements of the First-Order

Predicate Language
We begin our new language with the four truth-functional
connectives: , &, , and . But now, instead of sentences being
represented by a single lowercase letter, individual sentences
require a subset of three new categories of elements.
As we said when we were starting Aristotelian logic, sentences

have a subject and a predicate. Sentences say something about
something else. The thing that we are talking about we will call the
individual. What we are saying about it we will call the predicate.
Lecture 16First-Order Predicate Logic 153

We will start with the easiest version in which our predicates
are properties, such as is tall, is blue, or is yummy. We will
represent these properties in our language with uppercase letters,
such as T, B, or Y. So, uppercase letters are our properties.
One restriction on properties is that they must be world referring,

such as being tall, blue, or yummy. They cannot be logical
properties, such as is true, is valid, or is logically equivalent
to. This is why we call this language first-order logic. The
first-order language talks about things, and the second-order
language talks about the first-order language.
The subject, that which is being said to have the property, will
be represented by lowercase letters. But we need to be careful
because there are two different types of individuals we must
distinguish between. Borrowing terms from mathematics, there
are constants and variables.
A constant is a specific thing, such as Bob, the sky, or a particular

sandwich. Constants are not classes of things, but particular
things. For these, we use the lowercase letters in the first part of
the alphabetthe letters a through t.
So, if we want to say in our new language that Bob is tall, we select
an uppercase letter for the property is tallfor example, T. And
we need a lowercase letter to represent Bobfor example, b.
In this language, we put the predicate first and the individual

second. So, Bob is tall is written Tb. For Larry is tall, we use
l to represent Larry, and we get Tl. Bob and Larry are both tall.
That is a truth-functional combination of the sentences Bob is
tall and Larry is tall, giving us Tb&Tl. We cannot write Tbl,
because T is a property and only applies to one thing.
What about If Bob is tall, youre a monkeys uncle? Here, again,

is a truth-functional combination of predicate sentences. We have

seen that Bob is tall is Tb, and now we need a capital letter
for the property of being a monkeys uncle. Lets use M. And for
you, well use the lowercase letter y. So, If Bob is tall, youre a
monkeys uncle gets written TbMy.
We need to discuss sentences that are not about specific entities

that we can point to. Think of the end of a mystery novel. All of
the suspects have been gathered in the study and the detective
says, Someone in this room is a murderer. That someone is a
particular individual, but we dont know whom. So, we cant use
a constant.
For unknown individuals or for all of the individuals in some

category, we need something else in our language. We need
variables. Variables will be lowercase letters at the end of the
alphabetthe letters u through z. These will represent arbitrary
members of some class or unknown members of a class.
Suppose that we want to write the title of the song Everything

Is Awesome in our new language. How would we do that? You
might think that it would beusing A for awesomeAx. That
is some x has the property A. But then how would we write
Something is awesome? Wouldnt we also write it Ax?
But these are clearly different propositions, and the whole point
of our artificial logical language is to avoid the ambiguity we
find in ordinary spoken language. We need to do something
about this.
That something is a new pair of elements, called quantifiers. We

have two of them. The first is the existential quantifier (). We read
it as for some.
Quantifiers have to quantify over a variable name. So, we always

pair a quantifier with a variable. x reads for some x. We cannot
quantify a constant: b makes no sense in our language. For

some Bob? There is no need for some Bob; there is simply
Bob. Whenever you use a quantifier, you need to use a variable
and only a variable.
To say, Something is awesome, we would write xBx.
Suppose that Bob is quite distinguished looking, but not beautiful.

We might want to say, Something is beautiful, but it isnt Bob.
Here is a truth-functional combination of sentences. Remember
that but is a way of saying and, so we have the following.
Something is beautiful, and it is not the case that Bob is

beautiful.
That would translate into our new language as follows.
xBx&Bb
That is the existential quantifier. It is called this because it asserts

the existence of something.
The other quantifier is the universal quantifier (), which is read

for all. If we want to say, Everything is beautiful, we pick a
variable to quantify overfor example, yand we assert B of all
ythat is, yBy.
Suppose that we want to say, If Bob is not beautiful, then not

everything is beautiful. We have a conditional, as follows.
If Bob is not beautiful, then it is not the case that all things are
beautiful.
Substituting in our truth-functional symbols of the connectives:
Bob is beautifulall things are beautiful

We now know how to translate these first-order pieces.
BbyBy
Notice again that all quantifiers have to quantify over something,

and that something has to be a variable. All quantifiers need a
variable.
The relation goes in the other direction as well. All variables

need a quantifier. If T is the property is tall, what does Tz
mean? It doesnt mean something is tall or everything is tall,
because those both need quantifiers. In our language, it does
not mean anything. It is a violation of the grammar of first-order
predicate logic.
Fundamental Forms
Instead of the sentence letters of truth-functional logic, we now
have a more complex representation that allows us to get at the
content contained inside of sentences the way that Aristotelian
logic did. Indeed, we can subsume Aristotles categorical logic
within this broader approach. To do that, we need to translate the
four categorical forms into our new first-order language.
Lets start with A sentences, sentences of the form All As are

B, such as All people have noses. We need one capital letter
to represent the property of being humanfor example, H
and another for having a nosefor example, N. It is saying that
everything that is in the H group is also in the N group, so we
are looking at a universal quantification: It is all somethings are
something.
So, we need a universal quantifier and a variable name. It is

going to be x(Hx?Nx), where the question mark will be one of our
truth-functional connectives. Which one?

Some people intuitively think that it should be and. But it
isnt. Note what the sentence says if we include a conjunction:
x(Hx&Nx) says that everything is both human and has a nose.
That is not what we are trying to say. We are trying to say that
if something were to be a human, then it would be a thing with
a nose. So, we need the connective to be a conditional: All
humans have noses is written x(HxNx) in our language.
What about E sentences, sentences of the form No As are B?
The flawed intuition many people have in this case is to slap a

negation in front of the A sentence form. But remember the
square of opposition: The negation of an A sentence is an O
sentence, not an E sentence. Just because it is false that all
people love cilantro does not mean that no people love cilantro,
just that some people dont.
Remember that an E sentence is a universal sentence. As such,

we will need a universal quantifier. If we say, No people live on
Mars, we are saying something about all peoplethat they are
not Martians. What we are saying when we say that no people
are living on Mars is that, for all things, if that thing is a person,
we know that that thing does not live on Mars. In symbols,
x(HxMx).
So, the negation does not go in front of the sentence, but rather in
front of the second property. An E sentence, No As are B, is still
a claim about the entire set of things in the A class, but instead of
saying that they are all in the B class, we are saying that they are
not in the B class.
What about the particular sentences? Lets start with I sentences:

Some As are B. Clearly, this is an existential claim. We are
saying that there is something and that thing is both A and B.

We translate I sentences as z(Az&Bz). Unlike with the universal
sentences, we do use the conjunction in the particular. We are
not saying that there is something that if it were an A, it would be
a B. Instead, we are saying that there is something, and it is both
A and B.
What about O sentences? Lets think about sentences of the form

Some As are not B. Again, it is making an existential claim.
We are saying that there is at least one thing, and it has certain
properties. In this case, the properties are being A and not
being B. So, translating it fully, we get x(Ax&Bx); there is a
thing, and it is A and not B.
We now have all four of our categorical sentence forms translated

into our new first-order predicate language.
A: x(AxBx) E: x(AxBx)
I: x(Ax&Bx) O: x(Ax&Bx)
Add to these our categorical forms, the five basic forms, and we
have the building blocks to translate most sentences into our first-
order predicate language.
The sky is blue Bs

Everything is blue xBx
Nothing is blue xBx or xBx
Something is blue xBx
Something is not blue xBx
We will call these nine sentence forms our fundamental forms.

Many of the sentences we come across in normal conversation
are either one of these fundamental forms or a truth-functional
combination of them.

When translating, then, the key is to go step by step.
First, look at the sentence and figure out if it is one of the nine
fundamental forms or if it is a truth-functional combination of
the fundamental forms.
If it is a truth-functional combination, find the main connective
and insert the symbol for it, while leaving the rest of the
sentence in spoken language.
Look at the parts connected by the connectives and again
ask whether these are fundamental forms or truth-functional
combinations. Keep going until you are left with nothing but
truth-functional symbols and spoken language versions of
fundamental forms.
At this point, or if the sentence was not a truth-functional
combination, ask whether the fundamental form is categorical
or basic. Once you know that, figure out which form, and the
translation from there is straightforward.
Readings

Questions
Translate the following sentences into first-order predicate logic.

Use the following abbreviations: e = Emmett Kelley, C = is a clown,
A = is an artist, P = is a person, L = loves clowns, F = is afraid
of clowns.

1.
Emmett Kelley was a clown and an artist.
2.
All the world loves a clown.
3.
Some people love clowns, but some people are afraid of them.
4.
Clowns are artists, but only some artists are clowns.
Answers
1.
Ce&Ae
2.
x(PxLx)
3.
x(Px&Lx)& y(Py&Fy)
4.
x(CxAx)& y(Ay&Cy)

Lecture 17
Validity in First-Order
Predicate Logic
B ecause the universal quantifier potentially applies to an infinite

number of things, truth tables are not an option for us as a
validity test for first-order predicate logic. We can, however, extend
our system of natural deduction proofs to work for this stronger
language. To do so, we need to add four new rules and one new
equivalenceall of which you will learn about in this lecture.
New Rules of Inference

In first-order predicate logic, we need our other method of
demonstrating validity, natural deduction proof. We will, however,
need to augment it in order to deal with quantifiers. This will mean
four new rules of inference and one new equivalence.
The four new rules of inference will include two rules for taking
away quantifiers and two rules for adding them on. The idea is
to strip the quantifiers away and reduce things back to truth-
functional sentences. Then, once we have a sentence of the form
we want, we add the quantifiers back in to get the first-order
predicate sentence we were looking for. In essence, these rules
allow us to reduce things to what we already know how to do.
These rules have personalities. Two are laid-back, relaxed rules

that just do their job, no matter what is going on around them.
The other two are finicky, picky rules that refuse to do their job
unless very specific demands are met.
The first rule, universal instantiation (UI), allows us to remove a
universal quantifier that applies to an entire line and replace the
quantified variable with either a free variable or any constant.
Consider the sentence x(Bx&Cx); everything is both B and C. If

everything is B and C, what do we know about your favorite pair
of socks? It has to be B and C. If everything is B and C, we know
that the sentence formed by replacing the quantified variable x
with any constant must also be true. If we know it is true, then it
can appear in the second column of a proof.
But we can also replace the quantified variable with an unquantified,

or free, variable. That is, suppose that we had the following.
12 x(Bx&Cx)
Then, we could write the following.
13 By&Cy UI 12
Previously in this course, we made the seemingly unequivocal

claim that free variables are not allowed in our languagethat
By&Cy has no meaning. It doesnt mean that some y are B&C
and it didnt mean that all y are B&C. It means that it doesnt have
a meaning in our languagethat it is grammatically incorrect.
This is true, outside the context of a proof. In a proof, the use of

UI gives the sentence By&Cy a meaning. It means that for any
arbitrary thingfor example, yy is B&C. We can reason about
a universal property of the whole group if we have reason to think
that it holds for each member of the group.
While universal instantiation allows us to take away a universal

quantifier that applies to an entire line in a proof and replace the
quantified variable with either a constant or any free variable
name, existential instantiation (EI) allows us to remove an
Lecture 17Validity in First-Order Predicate Logic 163

existential quantifier and replace it with a free variable, but only
with a free variable that appears free nowhere earlier in the proof.
There are two restrictions on EI that we need to be clear about.

First, we cannot use EI to instantiate a constant. Consider the
sentence zBz. It says that something is B. What is that thing?
We dont know what exact thing it is, only that somethingthat is,
at least one thing, maybe all thingsis B.
If we used a constant to instantiate an existential, that is saying

that we do know what the thing is, and it is c. Maybe it is c;
maybe it isnt. If all we know is that it is something, then asserting
that some constant is the thing is to say more than we know, and
this means that we could be wrong.
If a sentence could be wrong, then it cannot be entered in the

second column of the proof. So, EI cannot be used to take away
a quantifier and substitute a constant for the quantified variable.
It has to be a variable that gets put in.
Again, when we use EI we have a sentence with a free variable,

something that we generally dont allow. But this free variable
means something slightly different from the free variable that results
from UI, in which the free variable means any arbitrary thing.
By contrast, EI means the thing whose identity we do not yet,

and may never, know. Think of the victim of a murder who is
discovered with no identifying information. If the victim is male,
we use the term John Doe. This name is a placeholder. It refers
to the person who was killed, even though we dont know who
the person is. The free variable instantiated with EI is the logical
equivalent of John Doe.
Suppose that we have the following.
8 xBx

Then, we can use EI to get the followingas long as z does not
appear free earlier in the proof.
9 Bz EI 8
In this sentence, the free variable z means whatever the thing is

that is z. From line 8, we know that it exists; we just have no clue
about its identity, so we will use z to refer to whatever that thing is
that is B.
The reason that we have to instantiate with a new variable name

is that all we know about the thing that is B is that it is the thing
that is B. If we reused a variable that appeared free earlier in the
proof, then we would be saying that the same thing has several
propertiesmaybe it does, and maybe it doesnt, but we cannot
be sure. For example, suppose the following.
10 xBx
11 xCx
If we used EI on both of these lines with the same variable name,

we would get the following.
12 By EI 10
13 Cy EI 11
And then we could form the following.
14 By&Cy Conj 12,13
Maybe there is a thing that is both B and C, but we dont know

that. Maybe one thing is B and a completely different thing is C.
It is true that some numbers are even and that some numbers
are odd, but if we make the mistake of re-instantiating with the
same free variable, we will be proving that some numbers are
both even and odd.

It is therefore crucial that every time we use EI, we introduce a
new free variable name.
While the first two rules are instantiation rules that let us remove
quantifiers, the other two are generalization rules that let us
reinsert them. And just as one came with no restrictions and the
other had constraints, it is the same with generalization rules. The
only difference is that it is the existential that is the easy one.
Existential generalization (EG) is the rule by which we add an

existential quantifier to a line. We can always take a line with a
free variable or a constant in it and add an existential quantifier
to bind that variable. If the line has a constant, then we know
that the property asserted of the constant is true of something
namely, that thing.
We might know that Jane is happy. If this is true, then it must

necessarily be true that something is happy. Because we are
being less specific, we are guaranteed that the move will be
truth-preserving, and therefore, we can write the new existentially
quantified sentence in the second column of the proof.
The same is true if we use EG to generalize a free variable,

regardless of whether it was instantiated universally or
existentially. If it is true that all books have pages, then it is true
that some books have pages. You can existentially generalize
any free variable anytime and anywhere.
The same is not true of universal generalization (UG), which

cannot be used on a constant. Just because Johns mom let
him go to the party doesnt mean that all moms will. Similarly, we
cannot use UG to generalize a variable that appears free in a line
justified by EG. Just because some men are jerks does not mean
that we can conclude that all are.

We can only use UG to generalize a free variable that was
introduced into the proof using UI and that has never appeared
free in a line on which EI has been used. EI stamps the
sentences it operates on with the quantificational version of guilt
by association.
Suppose that we have the sentence Everything has mass, but

if that thing is heavier than the earth, then there is something
else that orbits it. It is a truth-functional combination. The main
connective is and.
(Everything has mass)&(If that thing is heavier than the earth,

then there is something that orbits it)
It is important to see that both conjuncts in this sentence refer

to the same thing. In the first conjunct, we are saying that any
given thing you choose has a propertyhaving mass. In the
second conjunct, we are saying that the thing we picked for the
first conjunct also has the property of being orbited by something
else if it has a mass greater than that of the earth. Because both
conjuncts refer to the same thing, the quantifier we use must
have both sentences in its scope. So, we say the following.
x[(x has mass)&(if x is heavier than the earth, then there is

something that orbits x)]
Having mass is just a property, so lets use M to represent that.
x[Mx&(if x is heavier than the earth, then there is something

that orbits x)]
The second conjunct is clearly a conditional.
x[Mx&(x is heavier than the earththere is something that

orbits x)]

Being heavier than the earth is a property, and so is orbits x.
Lets use H and O for those. So, what we are saying is that there
is some thing y such that if x is heavier than the earth, then y
orbits x. This gives us the following.
x[Mx& y(HxOy)]
If we had that as a premise in an argument, we could use UI on it

because the scope of the universal quantifier is the entire rest of
the sentence.
1 x[Mx& y(HxOy)] Premise

2 Mx& y(HxOy) UI 1
Next, we can simplify.
3 Mx Simp 2
4 y(HxOy) Simp 2
On line 4, we see a sentence with an existential quantifier whose

scope is the entire rest of the line. So, we can use EI. We cannot
instantiate it with the free variable x because it already appears
free previously in the proof.
If we did use x again, we would have something orbiting itself.

Clearly, thats not what we want. But we could use any other free
variable name for this thingwhatever it isthat is orbiting x.
Lets keep it as y.
5 HxOy EI 4
Now we are in the situation we wanted to illustrate. We introduced

the free variable named x into the proof using UI in line 2. So, you
would think that we would be free to generalize it using UG. But
you would be wrong.

Because of line 5, x appears free in a line justified by EI, and
this means that x can no longer be generalized using EG,
even though it was introduced via a universal quantifier. By
simply being free on a line justified by EI, even though it was
not being instantiated, that is enough to disqualify it from being
generalizable using UG.
A New Equivalence
Finally, we have one new equivalence: quantifier negation (QN). If
we have a negation next to a quantifier, we can reverse the order
if we switch the quantifier.
:: and ::
If it is false that some people are 12 feet tall, then it is true that
all people are not 12 feet tall. If it is false that all grass is green,
then it is true that some grass is not green. We can always move
a negation from one side of a quantifier to another if we switch
the quantifier.
These are the only adjustments necessary to have a validity

test for first-order predicate logic. We take an argument in our
first-order predicate language, enter the premises into the proof,
use our original eight rules of inference, our new four rules of
inference, our original nine equivalences, our new equivalence,
and our proof strategies until such time as the conclusion
appears as a justified line in the proof, and we have a complete
demonstration of the validity.

Readings

Questions
1.
Show that the following argument is valid by constructing a proof.
x(Fx&Gx). x(GxHx). Therefore, x(Fx&Hx).
2.
Translate the following argument and show that it is valid by
constructing a proof.
All magical unicorns are immortal. There are no immortal beings.

Therefore, no unicorns are magical.
Answers
1.
1 x(Fx&Gx) premise
2 x(GxHx) premise
3 Fx&Gx EI 1
4 GxHx UI 2
5 Fx Simp 3
6 Gx Simp 3
7 Hx MP 4,6
8 Fx&Hx Conj 5,7
9 x(Fx&Hx) EG 8

2.
All magical unicorns are immortal.
x[(Mx&Ux)Ix]
There are no immortal beings.
xIx
Therefore, no unicorns are magical.
x(UxMx)
1 x[(Mx&Ux)Ix] premise
2 xIx premise
3 Ux Assumpt
4 (Mx&Ux)Ix UI 1
5 Ix UI 2
6 (Mx&Ux) MT 4,5
7 Mx Ux DeM 6
8 Ux DN 3
9 Mx DS 7,8
10 UxMx CP 39
11 x(UxMx) UG 10

Lecture 18
Demonstrating Invalidity
O ne of the wonderful aspects of truth tables is that they are both

validity and invalidity tests. But in first-order predicate logic,
we cannot construct truth tables. Natural deduction proof is our only
validity test. Successful construction of a proof guarantees that the
argument is valid, but an inability to find a proof does not establish an
arguments invalidity. So, well need a different way of demonstrating
invalidity. We will formulate two different ways of demonstrating that
an argument in first-order predicate logic is invalid: the method of
counterexample and the method of expansion. Both will show that an
argument is invalid by introducing semantic elements into our purely
syntactic approach.
Semantics, Syntax, and Pragmatics

There are three levels to language: syntax, semantics, and
pragmatics. Syntax deals with form and structure. Questions of
grammar are syntactic, as are word order and internal structure.
Semantics looks at meaning. Questions of what a word refers to

is a semantic concern. Semantic issues are important issues that
are complex and philosophically interesting.
Syntax looks at what we say, semantics looks at what it means,

and pragmatics deals with how we say it and how that can
change the meaning. Issues of tone or inference from that which
is unsaid are pragmatic aspects.
Our logical languages are thin because they are purely syntactic
languages. In a deep sense, they are languages with no
meaning. When we translate spoken language arguments into
truth-functional or first-order predicate logic, we remove the
semantic content.
Why are we ignoring semantics with our logical languages if it is

important? Because our interest is determining deductive validity,
which is a purely syntactic concept.
Whether a deductive argument is valid has nothing to do with what

the argument is about. It only matters whether the conclusion
follows from the premises. As a result, our artificial languages
remove the flesh from the spoken language arguments and
rigorously display the inner skeleton, because that is sufficient to
answer our questions about validity.
But in demonstrating invalidity, we are going to restore semantic

elements. The idea is that, in one sense, these semantic aspects
are irrelevant. Consider the invalid argument affirming the
consequent: If a, then b; b, therefore a.
This is a flawed argument form, regardless of what we fill in for

a and b. We could say If you are a dog, you have a nose. You
have a nose. Therefore, you are a dog. Or we could say, If you
won the lottery, you would be happy. You are happy. Therefore,
you won the lottery.
Whether you are a dog or have won the lottery is irrelevant here.
The content does not matter because it is the underlying form
of the arguments that is to blame. But the flaw in the structure
becomes clearer to us when we add some semantic content to
that structure.
Lecture 18Demonstrating Invalidity 173

We are not declaring individual arguments to be invalid, but
argument forms. When we add semantic content to the argument
forms, we are not merely saying that just the filled-in version of the
argument is invalid; rather, we are saying that the skeleton itself is
flawed. We are saying that any argument with that same form is
invalid because we are showing that it is possible to construct a
bad argument with that form.
The Method of Counterexample

An argument is invalid if it is possible for its premises to be true
while its conclusion is false. In truth-functional logic, we could
construct a truth table in which every possible combination of
truth-values for the constituent sentences is considered.
But another way is to simply come up with an example of an

argument that has the same form but has true premises and a
false conclusion. This is called the method of counterexample,
and because we dont have the ability to construct truth tables for
first-order predicate logic, we will have to use it.
In order to give semantic content to an argument form in first-

order predicate logic, there are three things we need to do.
1 We need to give meaning to the quantifiers. (When we say all

x or some x, all whats or some whats are we talking about?)
2 We need to say which specific elements of this universe are
picked out by our constants.
3 We need to give meaning to the properties.
These combine to give what logicians call an interpretation. An

interpreted argument has the semantic flesh put back on the
syntactic skeleton, and thus we can talk about the truth or falsity
of the interpreted sentences.

The goal is to generate an interpretation of our argument that
results in the premises all being true and the conclusion false.
This will conclusively show that the argument form is invalid.
If we want to construct an interpretation, we need first to give

meaning to the quantifiers. We do this by specifying what is
called a domain of discourse, or a universe.
When we use a universal quantifier to say all y, what is the all

we are talking about? It can be any set. It could be the set of all
politicians or the set of all brown things. It could be the set of all
things whatsoeverwhat is called an unrestricted domain.
Often, we will pick sets of numbers to be our domain of discourse

because their properties are well defined and well behaved.
Next, we assign a specific member of the universe to each

constant. If the domain of discourse is the set of positive integers,
each constant becomes a particular number. If the domain is
brown things, then each constant becomes a specific brown
thingnot, for example, mud, but rather a particular mud stain
on a particular shirt.
Each constant gets mapped onto a particular member of the

universe, but two different constants can get mapped onto the
same thing. People have multiple names. Some might call you
by your name, while others might call you mother or father, for
example. The same can be true for constants.
Finally, we take each property and assign it a meaning by

selecting a subset of our domain of discourse. This can be done
in two ways. We can simply choose members and put them
together to form a subset. The property is then being a member
of that subset. Orand this is the more common waywe
specify an actual property.

Note that the empty set, the set with no members, is a subset
of every set. That means that it is acceptable to have empty
categories. Indeed, sometimes it will be quite advantageous. For
example, we can use being a unicorn as a property.
Once we have specified a domain of discourse, chosen a member

for each constant, and assigned subsets to our properties, we
can then translate our formal propositions into spoken language
and determine if the premises of our argument are true and if our
conclusion is false on this interpretation. If so, we have found a
counterexample, and our argument is invalid.
The Method of Expansion

In addition to the method of counterexample, the other invalidity
test is the method of expansion. Coming up with counterexamples
requires that we fully endow our argument with semantic content.
Expansions only require some semantic content, but not full
meaning.
We do this by starting with a small universe that contains only

two things: a and b, for example. We then see if we can make the
premises true and the conclusion false for this small universe.
We do this by eliminating the quantifiers, reducing the problem

to truth-functional logic. If there are only two things, then our
universal quantifier becomes a conjunction.
The sentence xFx says that everything is F, but everything

in our small universe is just a and b. So, xFx is equivalent to
Fa&Fb. Similarly, xFx says that something is F. That thing has
to be either a, b, or both, because that is all there is. So, xFx
is equivalent to FaFb. This holds for any sentence in our first-
order predicate language.

Consider the sentence x[(DxNx)Jx]. We see that we
have a universal quantifier, so we start by writing down a
conjunction. On the left side, we replace every instance of
the quantified variablein this case, xwith a. On the right
side, we replace every instance of x with b. This gives us
[(DaNa)Ja]&[(DbNb)Jb].
Suppose that we have a complex existential quantification

for example, y[(Gy&Fy) (Gy&Fy)]. It is an existential
quantification, so we start by writing down a disjunction, and
on the left side, we replace every instance of the quantified
variablein this case, ywith a. On the right side, we replace
every instance of y with b. This gives us the following.
[(Ga&Fa) (Ga&Fa)] [(Gb&Fb) (Gb&Fb)]
But what if it is not a quantification, but a truth-functional

combination of quantified propositions? We can handle that, too.
Recall that the scope of a quantifier ends at a truth-functional
connective that is not shielded within parentheses. So, what we
do is keep all of the truth-functional connectives in place and
individually expand the quantifications.
How is this an invalidity test? When we expand the sentences,

the quantifiers are gone. All we have are sentences that can be
true or false and truth-functional connectives.
We have successfully reduced a first-order predicate question

to a truth-functional question, and we know how to answer the
truth-functional question of determining whether an argument
is invalid.
We could construct a truth table and find a row in which all the
premises are true and the conclusion is false. It turns out that the
truth tables will quickly become unwieldy and that the preferred

approach is to plug truth-values into the atomic sentence to
see if we could arrange them to give us true premises and a
false conclusion.
Readings

Questions
1.
Give an interpretation to show that the following argument is invalid.
x(FxGx). x(FxHx). Therefore, x[(Gx&Hx)Fx].
2.
Create an expansion to show that the following argument is invalid.
x(Fx&Gx). x(FxHx). Therefore, x(Fx&Hx).
Answers
1.
Domain = animals; F = human, G = warm-blooded, H = has a heart.
x(FxGx) = All humans are warm-blooded. (true)

x(FxHx) = All humans have a heart. (true)
x([(Gx&Hx)Fx] = All warm-blooded animals with a heart are
humans. (false)

2.
Domain = {a,b; Fa = T, Fb = F, Ga = T, Gb = F, Ha = T, Hb = F.
x(Fx&Gx) = (Fa&Ga) (Fb&Gb) = (T&T) (F&F) = T F = T.

x(FxHx) = (FaHa)&(FbGb) = (TT)&(FF) = T&T = T.
x(Fx&Hx) = (Fa&Ha)&(Fb&Hb) = (T&T)&(F&F) = T&F = F.

Lecture 19
Relational Logic
F irst-order predicate logic exposes the logical structure within

sentences. But the language as we have developed it so far
can only handle certain kinds of internal logical connections: the
attributions of properties to individuals. In order to be able to more
fully express the kind of content we put in our arguments, we need
to add another piece of machinery: relations between individuals.
In doing this, translation becomes significantly trickier because
when we need to assert a relation between variables, we will need a
different quantifier to bind each variable. Multiple quantifiers as well
as mixtures of properties and relations combine to make translating
into our extended first-order language more difficult.
Relations with Constants

Just as we did for properties, we will use capital letters to represent
relations. The only difference is that the number of lowercase
letters that follow the capital letter will increase. We translated Phil
is tall into our first-order predicate language as Tp, where T is the
property of being tall and p is the constant for Phil.
If we take T to be the relation taller than and j to be the constant

for Jose, then we will translate Phil is taller than Jose as Tpj.
Relations will look like properties, except that they will have more
constants or variables.
We can combine these to create truth-functional combinations,

as in the sentence Phil is tall, but he is not taller than Jose. In
this case, the main connective is and. Remember that but is
just another way of saying and.
So, we have the following.
(Phil is tall)&(It is not the case that Phil is taller than Jose)
Phil is tall is just Tp.
Tp&(It is not the case that Phil is taller than Jose)
It is not the case that Phil is taller than Jose is a truth-functional

combination with the main connective not.
Tp&(Phil is taller than Jose)
Phil is taller than Jose is Tpj, so the final formulation of the

sentence in our first-order language is as follows.
Tp&Tpj
You might think that we should have translated the sentence

as follows, because Jose is taller than Phil if Phil is not taller
than Jose.
Tp&Tjp
But this is problematic. Suppose that Jose and Phil are the same
height; then, neither is taller than the other. The negation of Txy is
not necessarily Tyx.
Some relations are symmetricthat is, if x bears the relation

to y, then y bears the relation to x. Consider the relation is
the sibling of. If you are Bobs sibling, then he is yours. It is a
symmetric relation.
Other relations are asymmetric. If x bears the relation to y, then y

does not bear it to x. Think of the relation is the parent of. If x is
the parent of y, then y is not the parent of x.
Lecture 19Relational Logic 181

Yet other relations are nonsymmetric. If x bears the relation to
y, then y might or might not bear the relation to x. Think of the
relation is in love with. When x loves y and y loves x, life can be
happy, but just because x loves y does not mean that y loves x.
To return to our example, when we translate, we need to translate

the sentence itself. To take Phil is not taller than Jose and turn it
into Jose is taller than Phil is to import external knowledge about
the is taller than relation, knowledge that is not in the structure
of the sentencethat is, not part of the syntax.
But, rather, this new knowledge is in the semantic content. We

know what is taller than means and that (with the rare exception
of people of the same height) it is antisymmetric. But when we
look at the validity of a relational argument, we need to ask what
the syntax alone tells us. If the relation is antisymmetric, then we
will be able to formulate that as a sentence that will be a premise
in the argument.
We will be able to say that if one person is taller than a second,

then the second is not taller than the first. That is content we will
need to make explicit as a premise. It cannot be simply assumed
in the translating because it is not part of the syntactic structure
of the sentence being translated.
So far, we have looked only at relations between two individuals,

but we are in no way limited to them. Think of the relation
between. It requires three things, one to be in the middle and
two to be on either side. We could write Kim is between Sharleen
and Gregory as Bksg. This is the usual way of writing itthat we
put the subject first.
The objection is that it is a between-ness relation, yet the thing

between is not visually between in the representation in our
language. One could define the symbol for between-ness such
that Kim is between Sharleen and Gregory is represented Bskg.

That is fine, but its not how we often do it. We will have to be clear
how we define the relations in this extension of our language.
Relations with Variables

Relations with constants are straightforward. Define the relation,
and part of that definition will be which place after the capital
letter stands for which place in the relation. But aside from that,
it is simple. This is not the case once we start in with variables
and quantifiers.
Suppose that we want to say, Phil is taller than something or

other. We need a constant for Phil but a variable for the unknown
thing that Phil is taller than. It is something, not everything, so the
quantifier for that variable will be existential.
xTpx
This, of course, is different than something is taller than Phil,

which would be the following.
xTxp
The order matters.
The move to the universal quantifier should be just as simple. Phil

is taller than everything: xTpx. Everything is taller than Phil: xTxp.
The next move is the Pandoras box of first-order logic. What

about something is taller than something? We need multiple
quantifiers. We are saying that there is a thing and there is
another thing, and the first thing bears the taller than relation to
the second thing.
xyTxy

Notice that this is exactly the same sentence as the following.
yxTyx
Remember that bound variablesthat is, variables that are in

the scope of a quantifierare dummy variables. They are just
placeholders that show us which quantifier they are connected
to. With properties, this never really came into play, but with
relations, it will be a relevant fact with great regularity.
For intuitions sake, lets switch relations to the loving relation and
restrict ourselves to talking about people. Suppose that we want
to say, Someone loves someone or other. We know that this is
the following.
xyLxy
There exists a thing such that there is another thing, and the first
thing loves the second one.
Suppose that we want to say, Someone loves everyone. Now,

we are saying that there is a thing, and that thing bears the loving
relation to all other things.
x yLxy
This is a different sentence from Everyone loves someone or

other. That is saying that for all things, there is a thing that the
first thing loves.
xyLxy
We need to be careful here, because there are two different

propositions we need to keep separate in our mind. One is
Everyone loves someone or other. This means that if you picked
any given person, that person loves some person.

This is different from the sentence There is a person everyone
loves. Now, we are saying that there is a thing, and that thing is
loved by all other things.
y xLxy
Notice how changing the order of the quantifiers has a radical

effect on the meaning of the sentence. xyLxy means that for
each and every person, there is some person out there whom
that person loves. But switch around the quantifiers and we get
y xLxy, which says that there is some person who is universally
loved. That is not something that follows from the first sentence.
The order of the quantifiers matters just as much as the order of
the variables.
Instead of just the two quantified basic forms that we had with
properties, we now have eight.
1 xyLxy Someone loves someone or other.

2 yxLxy Someone is loved by someone or other.
3 x yLxy Everyone loves everyone.
4 y xLxy Everyone is loved by everyone.
5 x yLxy Someone loves everyone.
6 xyLxy Everyone loves someone or other.
7 y xLxy There is someone who is loved by everyone.
8 yxLxy Everyone is loved by someone or other.
There is one more. All of these presume that the one loving and
the one loved could be different. But what about the sentences
Someone loves himself/herself and Everyone loves himself/
herself? For these, the one doing the loving, the first variable, and
the beloved, the second variable, are the same thing. As such,
they get the same variable and are bound by the same quantifier.
xLxx Something loves itself.

xLxx Everything loves itself.

So, when we go through the process of asking whether something
is a truth-functional combination of quantified propositions or is a
fundamental form, and we find that it is fundamental, and then
ask whether it is categorical or basic, and we answer basic, we
have 10 new basic forms we need to know.
With the categorical forms, we move up another notch of

complexity. We will keep the basic form of the A, E, I, and O
sentences, but the placement of quantifiers and the placement of
parentheses will become very important.
Consider the A sentence All who love someone are loved by

someone or other. The sentence makes clear that if person A
loves person B, then there is a person C, who might not be A or
B, who loves A.
We know that this is an A sentence, so it will have the general form

x(AxBx), but it will need to be made more intricate to handle
the relations. The key is to go step by step, partially translating as
we go. So, we start with the following.
All people who love someone or other are loved by someone.
We identify it as an A sentence and add in the structure, while

keeping the content in place not related to the general form.
x(x loves someone or othersomeone loves x)
Because the person who x loves in the antecedent might be

different from the person who loves x in the consequent, they
will need different quantifiers. And because neither appears in
both terms, the quantifiers will be a part of each term. Because
x appears in both, the quantifier for x must appear outside of the
parentheses in the order that both terms lie within its scope.
x[y(x loves y)z(z loves x)]

x(yLxyzLzx)
Readings

Questions
1.
Translate the following sentences into first-order relational logic. Use
Hx = x is human, Mxy = x and y are married, Pxy = x is a parent of y,
Gxy = x is a grandparent of y, j = John, m = Mary, r = Roberta.
a John and Mary are married and are the parents of Roberta.
b Everyone has a person who is his/her parent.
c Anyone who is the parent of ones parent is ones grandparent.
2.
Translate the following argument into first-order relational logic and
construct a proof for it to show that it is valid.
Everyone has a person who is his/her parent. Anyone who is the

parent of ones parent is ones grandparent. Therefore, everyone
has a grandparent.

Answers
1.
a Mjm&(Pjr&Pmr)
b x[Hxy(Hy&Pyx)]
c x y z[(Pyx&Pzy)Gzx]
2.
Everyone has a person who is his/her parent.
x[Hxy(Hy&Pyx)]
Anyone who is the parent of ones parent is ones grandparent.
x y z[(Pyx&Pzy)Gzx]
Therefore, everyone has a grandparent.
x(HxyGyx)
1 x[Hxy(Hy&Pyx)] premise
2 x y z[(Pyx&Pzy)Gzx] premise
3 Hx Assumpt
4 Hx yPyx UI 1
5 yPyx MP 3,4
6 Hw&Pwx EI 5
7 Hw Simp 6
8 Pwx Simp 6
9 Hwy(Hy&Pyw) UI 1
10 y(Hy&Pyw) MP 7,9
11 Hz&Pzw EI 11
12 Hz Simp 11
13 Pzw Simp 11

14 Pwx&Pzw Conj 12,13
15 y z[(Pyx&Pzy)Gzx] UI 2
16 z[(Pwx&Pzw)Gzx] UI 15
17 (Pwx&Pzw)Gzx UI 16
18 Gzx MP 14,17
19 yGyx EG 18
20 HxyGyx CP 319
21 x(HxyGyx) UG 20

Lecture 20
Introducing Logical Identity
T here is one final addition that needs to be made to our first-

order relational logical language, a new relation called identity.
Two things are identical only if they are the same thingthat is, two
individuals are identical if and only if they are different names for the
same thing. Identity is an equivalence relationthat is, it is reflexive,
symmetric, and transitive. Once we have it, we can start to translate
sentences that include quantities such as I have two arms and I
would like at least three slices of pizza.
Identity as an Equivalence Relation

Identity is an equivalence relation. We talked about truth-
functional equivalence when we were examining everything that
could be done with truth tables. We saw that two sentences were
truth-functionally equivalent if and only if they always shared the
same truth-valuethat is, the arrangement of Ts and Fs in the
columns in a truth table were identical.
We made use of this notion to give us our nine equivalences

in our system of natural deduction proof. This provided us with
new tools that we could use to substitute one sentence form for
another in a fashion that still guaranteed truth. These tools greatly
enlarged the number of arguments we could construct proofs for.
But just as we expanded everything else in first-order predicate

logic so that we could move beyond just talking about truth-
functional relations between sentences, when we talk about identity
in first-order logic, we will move from the notion of equivalent
sentences and place our equivalence relation inside of sentences.
With identity, we now want to talk about the equivalence of
individuals. We want to be able to say that two things are the
same thing. This is a different notion of equivalence. Two
sentences can be truth-functionally equivalent and say different
thingshave different meanings. So, we need a new, different,
more robust concept of equivalence.
Equivalence relations are relations that have three important

properties: r, s, and t. Equivalence is reflexive. That means that a
thing is always equivalent to itself. Think of reflexive as reflective:
When you look in a mirror, you never have to wonder whom you
are seeing. Your reflection is always you. That is r.
Next is s, which stands for symmetric: If a is equivalent to b, then

b is equivalent to a. Some relations, such as is the sibling of, are
symmetric. Some, such as is the parent of, are antisymmetric.
And others, such as is the brother of, are nonsymmetric. For
equivalence, the relation must be symmetric.
Finally, we have t, which stands for transitive. If a is equivalent

to b and b is equivalent to c, then we know that a is equivalent
to c. This should remind us of the kind of reasoning behind the
hypothetical syllogism, the rule of inference in natural deduction
proofs where if we have ab, and bc, we can infer ac.
If you have all three of these properties, then you get the most
important of all logical relations: If you have a relation that satisfies
r, s, and tthat is, reflexive, symmetric, and transitivethen you
have equivalence.
Translating with Identity

When we apply this relation to individuals, we have identity. For
identity, we will use in our language the universal symbol for
equivalence, the equal sign (=). It will not be a truth-functional
Lecture 20Introducing Logical Identity 191

connective that applies to sentences, but rather a relation within
sentences that shows the equivalence between individualsthat
is, constants and variables.
What makes identity the most important of all logical relations is

that it allows us to translate into our logical language all kinds of
sentences that we could not translate before.
Consider the following two sentences: Dave is taller than everyone

and Dave is taller than everyone else. The first sentence is false.
Dave is someone. So, if it is true that Dave is taller than everyone,
then it would have to be true that Dave is taller than Dave.
But we know that the is taller than relation is anti-reflexive

that is, because everything is always the same size as itself,
nothing can be taller than itself. It is not a contradiction because
this knowledge about the is taller than relation is something
semantic that we are importing. It is false because of the meaning
of is taller than, not because of the form of the sentence. But
with the interpretation we have given that form, we know that the
sentence is always false.
The other sentence, Dave is taller than everyone else, however,

could be true. It says that Dave is taller than everyone who is not
Dave. To know whether that sentence is true or false, we need to
know not just the form of the sentence and the nature of the is taller
than relation, but now we need to know how tall Dave is and how
tall everyone else is. These are completely different sentences.
The first sentence could be translated into our language already:

Dave is taller than everyone. It is of the fundamental form, and
it is categorical. It is an A sentence. All people are things that
Dave is taller than. We know that an A sentence, then, takes the
following logical form.
x(x is a personDave is taller than x)

We need a letter for the property is a person. We use H for
that. We need a letter for the is taller than relation. Lets use
T. And we need a constant for Dave; d would be the obvious
candidate. This, then, gives us the following.
x(HxTdx)
Without equivalence, we cannot translate the second sentence

without some philosophical weirdness. The sentence is Dave is
taller than everyone else. It is of the fundamental form, and it is
categorical. It is still an A sentence, but the slight difference is in
the antecedent.
x(x is a person other than DaveDave is taller than x)
To say that x is a person other than Dave is to assert a

conjunction; we are saying that x is a person, and x is not Dave.
x[(x is a person&x is not Dave)Dave is taller than x]
We know how to do most of this already.
x[(Hx&x is not Dave)Tdx]
But what do we do with x is not Dave? The intuitive move many

people make is to make a property of Dave-ness and assert that
x does not have this property.
x[(Hx&Dx)Tdx]
Making a property out of an entity, though, opens some strange

philosophical problems. How do we define Dave-ness? Dave
is always changing. Are there essential properties that belong
always and only to Dave?

Give us any set of properties and we can usually point to either
other things that have them and are not Dave, or we can point to
times of Daves life or after his death when Dave would have to be
categorized as not-Dave. But Dave is always Dave.
But even if you could come up with a satisfactory set of

properties that cluster to form Dave-ness, we still want to be able
to meaningfully talk about hypothetical cases in which Dave is
still Dave but lacks certain aspects that Dave has.
Would Dave have been so smart and successful if he had different

parents or if he had gone to different schools? We could have
interesting discussions about these types of questions and come
up with rational arguments that seem to be sensible. But what
we are doing is adding and subtracting properties from Dave in
order to do it. As such, taking Dave not only to be an individual
but also a property seems to be something we want to avoid.
Recall the purpose of developing this artificial language: to have

a tool that allows us to reason clearly and rigorously without
the ambiguity and sloppiness of ordinary spoken language. By
allowing individuals to become properties in this way, we are
inviting exactly the kinds of problems we are trying to avoid. To
escape this, we use identity.
That is not to say that there are no problems with self-identity.

What does it mean to be the same thing over time? But, while
interesting, these are not new problems we are introducing
through our language.
Using equivalence, then, we can finish our translation.
x([Hx&(x=d)]Tdx)
Dave is taller than everyone else.

Perhaps the most important sentences we can now translate are
ones involving quantities. Consider the sentence Steves class
has at least two students registered for it. Previously, we might
have wrongly thought that this is an I sentence: Some students
are registered for Steves class. We know how to translate an I
sentence. In the following, S is the property of being a student,
and R is the property of being registered for Steves class.
x(Sx&Rx)
This says that there is at least one student registered for Steves
class. So, we might have wrongly thought that this is a hidden
truth-functional combination and that it really is a conjunction of
two I sentences.
x(Sx&Rx)& y(Sy&Ry)
That says, There is one student x who is registered for Steves

class and a student y who is also registered for Steves class.
The problem here is that we do not know who x or y is. All we

know from these sentences is that x is some member of our
domain of discourse and y is some member of our domain of
discourse.
Could both of these be true and x and y be the same thing?

Absolutely. What we need is an additional claim, an assurance
that x and y refer to different members of the universe. And for
that, we need equivalence.
xy{[(Sx&Rx)&(Sy&Ry)]&(x=y)}
Now that says, There are at least two students registered for
Steves class.

Notice what we had to do with the quantifiers, grouping them at
the front of the sentence. The reason is that the equivalence (or,
in this case, the negation of the equivalence) relates x to y, so
the equivalence needs to be within the scope of both existential
quantifiers.
The jurisdiction of a quantifier ends at a truth-functional

connective. So, in the second formulation of the sentence,
x(Sx&Rx)& y(Sy&Ry), the x logically disappears once we hit
the central conjunction.
In order to have the equivalence within the scope of both

quantifiers, we need to put the entire rest of the sentences inside
the braces and put the quantifiers outside so that they reach the
equivalence portion.
Equivalence Substitution
We can now translate sentences that require equivalence into our
language, and that means that we can translate arguments. Can
we determine if they are valid?
We will need one new rule for our proof structure: equivalence
substitution (ES). It is, perhaps, the most obvious rule. If, on
some line of a proof, we have, as the entire content of the line, a
statement of equivalence between two individuals, then on any
subsequent line in which either individual appears in a sentence,
we can substitute the other equivalent individual.
Note that this only applies to free variable namesthat is,

variables that are not in the scope of a quantifier. We cannot
change a bound variable namethat is, a quantified variable
without changing the meaning of the sentence.

So, we can use ES on free variables and constants. When we
say that Mark Twain is Samuel Clemens, we are asserting the
equivalence of two names. We can equate two variables, as in
cases in which we claim that whoever is stealing your lunch is
also the person who is spray painting graffiti on the walls. And
when we say that the butler did it, we are equating a named
individual, Jeeves, with a free variable, the previously unnamed
murderer. We can use ES on an equivalence that connects any
free variable(s) or constant(s).
Readings

Questions
1.
Translate the following sentences, using Ax = x is an apple; Hxy = x
has y.
James has an apple, and Florence has a different apple.
2.
Translate the following argument and construct a proof to show that
it is valid.
James has at least two apples. Therefore, James has an apple.

Answers
1.
xy{[(Ax&Hjx)&(Ay&Hfy)]&(x=y)}.
2.
xy{[(Ax&Hjx)&(Ay&Hjy)]&(x=y)}; therefore, x(Ax&Hjx).
1 xy{[(Ax&Hjx)&(Ay&Hjy)]&(x=y)} premise
2 y{[(Ax&Hjx)&(Ay&Hjy)]&(x=y)} EI 1
3 [(Ax&Hjx)&(Ay&Hjy)]&(x=y) EI 2
4 (Ax&Hjx)&(Ay&Hjy) Simp 3
5 Ax&Hjx Simp 4
6 x(Ax&Hjx) EG 5

Lecture 21
Logic and Mathematics
A s you learned in the previous lecture, equivalence relation allows

us to translate into our language numerical sentences. This
means that we can now ask questions about the form of mathematical
reasoning. This is the question that launched the study of deductive
logic in the first place. If our study of deductive logic looked like work
you have seen in mathematics classes, thats because the two studies
are inextricable. We cannot discuss mathematics without discussing
logic, and we cannot discuss logic without discussing mathematics.
Logic in Euclidean Geometry

Aristotle was the first to develop a systematic approach to formal
reasoning with his categorical logic, but the hallmark of formal
reasoning, the book that brought deduction to the forefront of the
Western intellectual tradition, was written about 50 years later. It
was Euclids The Elements.
Euclids use of deductive proofs started with five axioms and five
postulates that were all seemingly self-justifyingthat is, they
were so obvious that no one could possibly deny their truth. He
starts with general mathematical ideas, such as equals added to
equals yields equals.
But what was amazing is where he went from there. He used

these basic truths, and then added some definitions. Definitions
have to be true because they are true by definition. We are
free to define any term in any way we want. So, Euclid gives us
definitions of point, line, circle, and other geometric terms.
But then he combines all of this and, using strict deductive
reasoning, demonstrates the necessary truth of hundreds of
theoremsthat is, propositions he derives from the axioms and
postulates. These are not obvious. Some of them are incredibly
intricate, and some of them are counterintuitive. But after reading
Euclids work, one is convinced that each must be true.
What Euclid did that made his book one of the most important
in human history is his organization of the theorems of plane
geometryhis use of a strict deductive system to derive them all
from first truths.
Euclids use of deduction was so elegant and powerful that it

made thinkers who came after him long to do for every aspect of
human thought what he did for geometry. So, Euclids Elements
reigned as the unchallenged epitome of rigorous thoughtthat
is, until some tried to improve it.
Mathematicians love elegance: doing the most with the least, not
making any more assumptions than are absolutely necessary.
Euclid had made 10 such assumptionsthat is, he had five
axioms and five postulates.
But one of the postulates bothered mathematiciansthe fifth

one. It seemed different. The following are the five postulates.
1 Aline segmentcanbedrawn connecting any two points.

2 Alinesegmentcanbeextended indefinitely far in either direction.
3 A circle of any radius can be drawn around any point.
4 All right angles are equal to one another.
5 Take any two straight lines and have a third line fall across
them. If, on one side of the crossing line, the other two lines are
approaching each other, they will eventually meet on that side of
the crossing line.

The fifth postulate seems much more like one of the many
theorems that are proven in the book than the other postulates,
which are more about what can be constructed. So, some
thought, maybe it is a theorem. If it is, then it could be proven from
the definitions, the five axioms, and the other four postulates.
If you could create a deductive argument with the fifth postulate

as a conclusion, then it would not need to be assumedit would
be omitted from the list of postulates. Diminishing this list would
create a more elegant system, and that would be astounding.
Over the centuries, many mathematicians tried. All failed.
Despite the universal failure, the nagging sense persisted.

So, some changed tactics. Instead of constructing what we
have called a direct proof, maybe they could instead proceed
indirectly, in which one sets out the premises and the negation of
the conclusion and reasons until a contradiction rises.
Many attempts were made, but no one could show that this process
led to explicit contradictions, and this is what is needed to conclude
an indirect proof. After a while, several mathematicians thought that
perhaps the contradictions would never arise. If that were true,
then what they were playing with was not a self-contradictory set of
premises, but rather propositions that were consistent.
The Discovery of Non-Euclidean Geometry

But if that were true, what they had in their hands was a new
geometrya geometry other than that of Euclid, a non-Euclidean
geometry. While a handful of mathematicians nearly simultaneously
came to this conclusion, the name most associated with this move
to asserting the existence of a geometry different from that of
Euclid is the Russian mathematician Nicolay Lobachevsky, and the
system is now widely referred to as Lobachevskian geometry.
Lecture 21Logic and Mathematics 201

The introduction of Lobachevskian geometry set off alarm bells in
the world of mathematics. Non-Euclidean geometry seemed like
a contradiction in terms. Euclid had used deduction. Deductive
reasoning guarantees the truth of its conclusions. How could
there be alternative mathematical truths?
Mathematical truths are not like physical truths. Things in the

world could be otherwise, but mathematical truths seem to be
necessary. They have to be true.
So, we have one geometry with one set of axioms and a second
geometry with a different set. Which one is true? Which set of
axioms should we believe?
Surely, nearly everyone thought, it must be Euclid. How could

we possibly reject Euclid? So, the push continued to find the
contradiction that lay hidden among the results of Lobachevskian
geometry. If a single contradiction could be found, the whole
non-Euclidean threat could be rejected.
But it was a pointless search. This fact was proven by three different
mathematicians: the German Felix Klein, the Frenchman Henri
Poincar, and the Italian Eugenio Beltrami. Each constructed what
we call a relative consistency proof for non-Euclidean geometry.
A consistency proof is an argument that establishes that the

axioms of a system are non-contradictory. A full-out consistency
proof for non-Euclidean geometry would have been wonderful
and established beyond any doubt that non-Euclidean
geometries were non-contradictory. But we got something only
half as good: a relative consistency proof.
Klein, Beltrami, and Poincar took the language of the non-

Euclidean universe and erased all of the meanings. They then
gave the non-Euclidean terms a new meaning in the Euclidean
world and showed that the relations among the reinterpreted

axioms are true in the Euclidean universe. They made Euclidean
models of non-Euclidean geometry.
The radical result is that the only way the axioms could form an
inconsistent setthat is, the only way non-Euclidean geometry
would ever produce a contradictionis if Euclidean geometry did.
Because we could show a formal relation between the sentence

forms and because consistency and inconsistency are formal
properties, we could prove that the two systems are relatively
consistent. We did not show that non-Euclidean geometry
is consistent, but rather that it is only inconsistent if Euclidean
geometry is inconsistent.
But is Euclidean geometry consistent? Mathematicians thought,

if it is, we should be able to prove it. So, how can we prove that
Euclidean geometry is consistent?
We had been working with it for centuries, and no one had found
a single contradiction yet. But just because a contradiction had
not been found yet does not mean that one will not surface
eventually. We needed a proof.
But before we could prove Euclids geometry consistent, a small

detail needed to taken care of. It was long known that Euclid had
played fast and loose with some of his proofs and had made use
of some assumptions that he had not accounted for in his axioms
and postulates. If we were going prove that Euclidean geometry
was consistent, then we first needed to clean up Euclids work.
Logic in Mathematics
We needed a complete re-axiomatization of Euclidean geometry
and a rigorous formalization of the logic to be used. It was a big
project that would require a big brain and intellectual courage.

The man for the job was
one of the greatest minds in
the history of mathematics,
David Hilbert.
Unlike some mathematicians

of his time, Hilbert saw logic
as the most powerful tool
of the mathematician. He
embraced it and the problems
that came from it. Contrary to
constructivismwhich held
that logical proofs were not to
be admitted in mathematics
Hilbert became the founding
father of formalism, the view David Hilbert
that mathematics was about (18621943)
the forms of equations and
mathematical entities.
This formalist project needed a formal methodology. With

his student Wilhelm Ackermann, Hilbert wrote a book called
Principles of Mathematical Logic. Logic was to be the method,
and it needed to be made rigorous, so Hilbert developed what
we have come to know as natural deduction proof.
It was Hilbert who first developed the system we have been

studying, although his version is a bit different. Hilberts system
has many fewer rules, but philosophically, the foundation is the
same. We have a formal process that demonstrates validity, a
formal property using purely formal means.
He took this view and attacked many of the major mathematical

problems of his dayamong them, the question of geometry.

Poincar, Beltrami, and Klein had produced relative consistency
proofs. They had shown that the only way that the non-
Euclidean geometry of Lobachevsky could be inconsistent, self-
contradictory, is if Euclidean geometry was as well.
No one doubted the consistency of Euclid, but no one had a proof

of it either. And if we are mathematicians, we have warrant for
rational belief only if we have a proof. So, is Euclidean geometry
consistent? We need a proof.
But before we could prove Euclids system consistent, we

needed a reformulation of Euclid. We needed for all and only the
assumptions, the real axioms, to be set out. We needed a strict
means of logical processnatural deduction.
And we needed the theorems of Euclid to be re-derived. Hilbert

did exactly this in his book The Foundations of Geometry,
published in 1899. In it, he replaces Euclids 30 definitions, five
postulates, and five axioms with 20 axioms. He then proceeds to
do the unthinkablehe improves Euclid. Hilbert had created a
completely logically precise version of Euclidean geometry.
Saving Geometry by Reducing It to

Arithmetic
The question could now be approached. Is Euclid consistent?
For that, Hilbert could not produce a proof. But he could take the
next step down the road begun by Klein. Hilbert could produce
a relative consistency proof for Euclid, showing that the only way
that Euclidean geometry held inconsistencies is if arithmetic did.
Math is about the internal structure and relations of the concepts

within the axioms, and Hilbert showed how we could produce
a coherent interpretation of the Euclidean axioms inside the
language of arithmetic.

We could take the sentence forms of the axioms he used to
reformulate Euclid and find their equivalences inside of the results
of arithmetic. That means that the consistency of all of geometry
now relied on the consistency of arithmetic.
But is arithmetic consistent? Mathematicians doubted this less

than they doubted the consistency of Euclid. But this certainty is
unwarranted, or at least ungrounded, without a proof. Could we
prove arithmetic consistent?
In the year 1900, Hilbert gave the keynote address at the Second
International Congress of Mathematics, the most prestigious
gathering of mathematicians from all around the world. Hilberts
address laid out the 23 problems that required the combined
efforts of the mathematical world in the 20 th century. It is one
of the most important addresses ever given in the history
of mathematics.
Of those 23 problems, about 12 were solved in that century

there is still debate about a few of them. For this course, the
important one is the second problem: to investigate the
consistency of the arithmetic axioms. Geometry now rested on
arithmetic, and it was incumbent on mathematicians to prove a
logical property of arithmetic in order to provide mathematics
with the firm foundation it required.
Logic was not only a part of mathematics, but the truth of all
mathematics itself now rested on the ability to produce a proof
of a logical property. Logic and mathematics had become
inextricably entwined.

Readings
Euclid, The Elements.

Gray, The Hilbert Challenge.
Kramer, The Nature and Growth of Modern Mathematics, chaps. 2932.
Yandell, The Honors Class.
Questions
1.
David Hilberts logical approach to mathematics upset mathematicians
like Paul Gordan because the purpose of mathematics, they argued,
is to produce what you prove. By proving that something cannot not
exist, they contended, is not really showing us the thing, and the thing
is what we want. Was Gordan correct that Hilbert was not really doing
mathematics, or did they just have an old-fashioned idea of what
mathematics is? Does mathematics have revolutions that change
how we think of the mathematical realm like scientific fields do?
2.
The results of Euclidean geometry are easy to visualize, and they
seem to resemble our actual observations. Could observations be
used as a foundation for mathematical truth? Doing so would mean
that they would be approximate and not necessary truths. Does this
diminish mathematics? Does the fact that we have a difficult time
envisioning non-Euclidean geometry mean that it is or has to be
false? What role does and should our ability to envision something
play in determining when a mathematical proposition is true or false?

Lecture 22
Proof and Paradox
A t the turn of the 20th century, the foundations of mathematical truth

hinged on logic. The question of the consistency of Euclidean
geometry had been reduced to the question of the consistency
of arithmetic. So, at the beginning of the 20th century, we were left
with the logical question of the consistency of arithmetic. Logic and
mathematics had become entangled, but to a group of thinkers,
including Gottlob Frege and Bertrand Russell, the connection would
run even deeper than others had previously suspected.
Arithmetic in the 20 th Century

Traditionally, mathematics was thought to have two separate
realms: geometry (shapes in space) and arithmetic (numbers and
equations). Non-Euclidean geometry had the mathematical world
in a tizzy, but it would be wrong to say that life was peaceful in
the arithmetic realm.
The consistency of arithmetic, the ultimate grounding of numerical

truths, was something that was considered indubitable. But at the
end of the 19th century, nothing was beyond doubt.
The great German mathematician Richard Dedekind had

considered that while non-Euclidean geometry had been
reduced to Euclidean geometry and Euclidean geometry had
been reduced to arithmetic, perhaps the can needed to be
kicked one more step down the road and arithmetic needed to
be reduced to set theory.
Think about numbers as conceptual representations of sets and
addition, subtraction, multiplication, and division as operations on
setsfor example, merging sets or looking for members that sets
have in common. This way, we would just need a complete and
consistent theory of sets, and that could produce for us the long-
sought foundation for mathematical truth.
Sharing Dedekinds interest was a younger German

mathematician, Georg Cantor, whose results are among the
strangest in mathematics. Cantor thought about numbers as the
size of sets. Two sets are of the same size if we can draw a line
from each member of one set and have it end at a member of the
other set, and there is no member of either set that is not paired.
We can, therefore, talk about the size of sets without counting. If

everyone who buys a concert ticket sits in one seat and we want
to see if the show is sold out, undersold, or oversold, we dont
need to count the seats and count the tickets sold, we just need
to see if there are empty seats, if there are people without seats,
or if every person has a seat and every seat has a person.
We call this mapping. A mapping is one-to-one and onto, what

mathematicians call a homeomorphism, if all of the members
of each set are mapped onto one member of the other set.
Homeomorphic sets are the same size.
The ability of being able to say that two sets have the same
number of members without having to count means that we can
talk about infinite sets. Some infinite sets are the same size even
though one might be a proper subset of the otherfor example,
even numbers and the positive integers. Some infinite sets are
larger than other infinite setsfor example, real numbers and
rational numbers.
Lecture 22Proof and Paradox 209

Mathematicians became suspicious of thinking about infinity. We
cannot construct infinite sets, so we should not be able to speak
of them. They are not part of mathematics.
The use of logic and the use of sets in mathematics were joined
not only in the minds of opponents, but also in the minds of
those who saw it as the path to ultimate basis for certainty in
mathematics.
Questioning the Basis of

Mathematical Truth
The question of the consistency of non-Euclidean geometry
was reduced to the consistency of Euclidean geometry and that
reduced to the consistency
of arithmetic, which was
in turn reduced to the
consistency of set theory.
Was this a never-ending
and therefore ultimately
unjustifiedpath? What is
the intellectual bedrock on
which it all ultimately rests?
Some philosophers, such as

Plato, thought that numbers
are mystical, metaphysical
things, and that is where the
ultimate justification resides.
Other philosophers, such as
Immanuel Kant, argued that Immanuel Kant
it was in the human mind, in (17241804)
our own intuitions, that we
find the justification.

But many mathematicians were loath to take these philosophical
paths. They wanted absolute rigor on purely mathematical terms.
For Hilbert, the answer was his view, formalism, in which we just
started with an axiomatized theory that we hypothetically take to
be true.
This meant that all of mathematics was ultimately ungrounded

and that math was just a game like basketball, with arbitrary rules
whose results do not mean anything away from the playing field.
Many mathematicians were scandalized by such trivialization of
their work.
Another view was called logicism. According to this view, there

is an ultimate end to the reduction. The last stop was logic
itself. Logic is necessarily consistent. If we could reduce set
theory to pure logic, then there could be no doubt that it is not
only consistent, but necessarily true because logic is based on
definitions and tautologies.
According to the logicists, all of mathematics is just complex logic

when you get all the way down. Logic is not merely a method to
be used in doing mathematics; logic is the reason for and the
justification of mathematics.
The Logicist Project

The project of coming up with an axiomatized theory of sets that
could be reduced to logic was most famously begun by the Italian
mathematician Giuseppe Peano. He used nine axioms framed in
an early version of formal logic that he developed for the work.
It involved natural numbers and a function called successor. The

idea is that we can posit basic truths about a number and its
successor, the number that comes next. Using this, he showed

how the truths of addition and multiplication could be worked out
within the system, which necessarily obeyed the laws of logic.
Impressed with this work, but having the sense that it needed
to be taken further, was the German mathematician Gottlob
Frege. He wanted to reconceptualize all of mathematics in a
fashion that would be able to show how every mathematical truth
could be traced back to logical concepts via absolutely rigorous
deductions. There would be no gaps in the reasoning, no appeals
to intuition; it would all be strict and formal.
He knew that such rigor in deduction would require a logical

language. He appreciated Peanos work but thought that an
even more powerful logical language would be needed, so he
constructed what he called his Begriffschrift, or concept writing.
It would be his logical language within which his completely

rigorous work could be done. It is a progenitor of our first-order
predicate logic, although one that is much more difficult to work in.
Having his logical language, he began to work on setting out

a version of set theory, which would be a step more abstract
than Peanos. Indeed, he sought to create a logical, conceptual
framework from which Peanos axioms would emerge as
theoremsas results.
For Frege, mathematics must be completely reducible to the

concepts of logic. But what, then, are numbers? Think about
the two seemingly quite different ways we use numbers. On the
one hand, we talk about numbers abstractly. On the other hand,
numbers are amounts of things.
How can we make sense of numbers in such a way that they

are not like all other normal, observable properties? How can
numbers be both in the world and be abstract concepts?

Sets are conceptual things, but we can map abstractly generated
concepts onto concepts we apply to the world. But that mapping
is also conceptual, so mathematics should be reducible to the
underlying logic if the system can be created properly. The
ultimate concept of set would be the basis for everything in
mathematics, and it would be abstract yet applicable.
Frege worked for years to develop this system. He set out

a version of his view in his earlier book, The Foundations of
Arithmetic, and then his full system in his masterwork, Basic Laws
of Arithmetic. He thought he had done it.
Russells Paradox
The first volume of Basic
Laws of Arithmetic came
out in 1893, and he had
continued to work on the
project for the better part
of the following decade. In
1902, he was ready with
the second volume.
But before he could publish

it, he received a letter from
an unknown philosopher
and logician in Britain, the Bertrand Russell
young Bertrand Russell. (18721970)
He was one of the few who
had read Freges work and
admired it to the degree
Frege thought it warranted.
Russell agreed that logic was the basis of all mathematical truth,
and Freges work took us along the path to successfully showing it.

However, Russell came to realize, Freges work could not be
the last word on the matter. Frege thought he had succeeded
in developing a system in which set theory and thus arithmetic
derived from pure logic, thereby guaranteeing that set theory
would be free of contradictions.
But Russell informed him that, in fact, the opposite was true:
Freges system generated a contradiction, what has come to be
known as Russells paradox.
Russell thought that Freges problem was that his axiom for set
construction was far too permissive. It allowed sentences to assert
properties of sentences of its own order. This was a problem. We
need a logical caste system, a hierarchy with strict regimentation
and an absolute rule against semantic fraternization.
This approach was what came to be known as Russells theory

of types, and he thought that it was sufficient to save Freges
logicist project. He thought that this sort of logical segregation
would make the paradoxes disappear while still saving the idea
that all math could be derived from nothing but logic and abstract
concepts added in as definitions.
Russell worked for years developing a system from which an

axiomization of arithmetic would emerge. He joined forces
with one of his former professors, Alfred North Whitehead, and
together they built the first-order logical language that we learned
to work in, and they used it to produce a mammoth work of logic:
Principia Mathematica. It is three huge volumes from which the
basic propositions of arithmetic are painstakingly derived.
Unlike Freges effort, it was a work that was received with great
fanfare. The logicist project seemed to have been carried out. Logic
saved mathematical truth and gave mathematics a firm foundation
from which to emerge. For an equation to be true is for it to be
provable in the system of axioms set out in Principia Mathematica.

Gdels Incompleteness Theorem
In 1931, Austrian mathematician Kurt Gdel did what Russell
thought could not be done. Gdel produced a paradox from the
axiom set of Principia Mathematica, a paradox not unlike the one
Russell had pointed out could be formed in Freges system. This
result is famously known as Gdels incompleteness theorem.
Gdel created a kind of relative consistency proof of his own.

But instead of finding a model of non-Euclidean geometry in
the Euclidean language, he found a model of second-order
mathematical sentences in the first-order language.
In this way, he considers the second-order sentence this

sentence is unprovable. Either it has a Gdel number or it
doesntthat is, either it has a proof or it doesnt.
Russell and the other logicists, to be true is to be provable in

the system of logic and to be false is to be unprovable. If this
sentence is unprovable can be proven, then it is false but has a
proof. This is not allowed. If this sentence is unprovable does
not have a proof, then it is true, but does not have a proof. Either
way, the logicist definition of truth as provability within the system
is flawed.
But the problem is that this flaw is mirrored in the arithmetic itself
through the Gdel numbering. The mirroring connects second-
order truths to arithmetic truths.
So, Russell and Whiteheads axioms of arithmetic necessarily either

let in true sentences that cannot be proven (this makes the system
incomplete) or false sentences that can be proven (this makes the
system unsound). Either way, logicism as Russell worked it out is
dead. Russells attempt to save Freges project fails.

Readings
Coffa, The Semantic Tradition.

Kramer, The Nature and Growth of Modern Mathematics, chaps. 3033.
Nagel and Newman, Gdels Proof.
Questions
1.
Cantors work on trans-infinite numbers requires an axiom of infinity,
an assumption that posits the existence of a first infinite number. From
there, all of the strangeness of his results follows. Mathematicians who
were bothered by his results contended that we should not accept his
initial assumptionthat it was based on a misunderstanding of the
infinite. This seems to be a philosophical objection. On what grounds
should we accept or reject basic assumptions in mathematics? Is
math ultimately philosophical? Is there a logical basis? Is it a matter of
applicability to science? Can we accept any basis we want?
2.
If logicism has failed, which of its competitors is a better foundation for
mathematical truth? Is Plato correct that mathematics is not about this
world, but rather about an ideal world of concepts? Is Kant correct
that mathematics is really a form of psychology that investigates how
the human mind thinks about numbers and shapes? Is Hilbert correct
that mathematics does not give us absolute truths, but only results
of different axiom games we can choose to play if we want? Are
empiricists correct that mathematical truths are really just a kind of
physical truththat one apple and one pear together are two pieces
of fruit and that mathematical truth is no longer exact and necessary
but just another variety of plain old truths?

Lecture 23
Modal Logic
W e hold there to be two different senses of truth: those sentences

that happen to be true and those sentences that must be true.
Our logic so far does not distinguish between garden-variety truth
and necessary truth. Can we develop a logic that does account for
the difference between these kinds of true sentences? This is what we
call modal logic. Modality is the term that philosophers use for the
concepts possible and necessary. Modal logic is an augmented
version of our first-order relational logical language that includes new
operators for possibility and necessity.
Modality
We have been using a particular word repeatedly in these lectures
as if it is clear what it means. The word is true. What does it
mean for a sentence to be true? Philosophers have debated this
for centuries.
One standard definition, what we call the correspondence theory

of truth, holds that a sentence is true if what it says about the
universe is actually the case in the universe.
Philosophers have long realized that there are two different

types of true sentence. One type of true sentence happens to
be true, but the world could be otherwise. These contingent
truths are possibilities.
But then there are sentences we think must be true no matter

how the world is. We call these necessary truths. We have
thought that these necessary truths are sprinkled throughout
human endeavorsin such areas as mathematics, science,
ethics, and metaphysics.
The concept of necessary truth is stronger than mere truth or

that which happens to be true. The term modality is used by
philosophers to refer to the concepts of necessity and possibility.
The logic we have been exploring is only geared to a single

concept of truth. If we want to expand it to be able to look at
arguments involving possible and necessary truths, then we
need to create an augmented system. We need modal logic.
We have seen some necessary truths already. Think back to truth-

functional logic. We saw that there are tautologies, sentences like
I have a brother or I dont have a brother that are always true.
We can extend this to our first-order predicate language and
come up with necessary truths like All dogs are dogs.
There are sentences like these that are necessarily true. These
sentences are necessarily true because of their form. Any
sentence of the form ava or of the form x(DxDx) will be
necessary truths.
There are some examples from mathematics, science, ethics, and

religion that are supposedly necessary truths because of their
content, not because of their form. This means that we will need
some additional machinery to deal with the logic of modal concepts.
We need to add two new operators to our first-order predicate

language. The box (o) will be our symbol for necessary, and
the diamond (u) will be our symbol for possible. They will
function syntactically sort of like quantifiers in that they will be put
in front of the sentences that they are intended to modify.
We might say that the sentence Everything is F is truethat

is, we can assert xFx. But if we want to make a stronger claim

and say that it is no accident of the universe that everything is F,
but rather that it could not be otherwisethat everything must be
Fthen we can write o xFx. This says that the sentence xFx is
a special sentence; it is a necessary truth.
In the same way, we could add a special notation to indicate that

we dont know if xFx is true or not, but it is certainly possible that
all things have the property F. In this case, what we are asserting
is uxFx. This says that the sentence xFx is like a contingency
in truth-functional logic. It is a sentence that could be true.
There is a logical relation between these two. In the same way

that we have the equivalence quantifier negation that allows us to
switch the order of a negation and a quantifierfor example, not
everything is the same as something is notthe same holds
true with our modal operators.
We can employ the equivalence modal negation (MN) and switch

the order of a modal operator and a negation. If we say that it is
not the case that Everything is G is necessary, then we have the
sentence o xGx. But to say that something is not necessarily
true is to say that it is possible that it is false. But this is to say that
uxGx.
Similarly, if we say that it is not possible that everything is F,

then we are asserting the sentence uxGx. But to say that it is
not possible for it to be true, we are saying that it is necessarily
falsethat is, we are asserting the sentence oxGx. If we move
the negation to the other side of the modal operator and switch
the operator, we end up with an entirely equivalent proposition.
One way that philosophers make sense of these new operators

is to give them an interpretation in terms of possible worlds.
By the term possible world, we mean a way the world could
be. Our world, the real world, is just one among the infinite
possible worlds.
Lecture 23Modal Logic 219

When we write a diamond in front of a sentence, we are saying
that the sentence is possibly true. By that, we mean that there
is a possible world (or some set of possible worlds) in which
this sentence is true. When we then use that sentence in valid
arguments, we are seeing what else must also be true in this
particular possible world.
When we put a box in front of a sentence, we are saying that the

sentence is necessarily true. A sentence is necessarily true if it is
true in every possible world. No matter what world you consider,
the necessary truths will be true in them. If we put a box in front of
a negated sentencethat is, we say it is necessarily falsethat
means that it is true in no possible world.
The Mechanics of Modal Logic

The translation of spoken-language modal sentences into our
expanded first-order modal logical language is not different from
our regular translation in our first-order language, except for the
new machinery.
With all of the other additions we have made, we first learned to

translate and then jumped right into how to use them in proofs.
What are the new rules and equivalences associated with the
introduced elements? With modality, however, this is more
complicated because there is not single logical meaning for
them. Rather, there are several possible meanings.
There are multiple viewpoints for modal logic. They are called
modal systems of different strengths because each stronger level
commits us to a new rule, where the weaker ones do not. We
account for different meanings of modality by having different
modal languages that use different additional axioms.

K (named for the philosopher-logician Saul Kripke) is the
weakest modal language, according to which a sentence is
necessary only if we can prove it using no premises. In this
language, the axiom o(PQ)(oPoQ) is true and can be
inserted into a proof for any sentences P and Q.
M or T (named independently by Kurt Gdel and Georg Henrik
von Wright), the next stronger language, adds the claim that if
a sentence is necessarily true, then it must also be truethat
is, oPP.
S3 (where S stands for strength) adds the additional claim
to M or T that if a sentence is necessarily true, then it must also
be possibly truethat is, oPuP.
S4 adds to S3 the additional claim that if a sentence is
necessarily true, then it is necessarily true that it is necessarily
true that is, oPooP.
S5, the strongest modal language, adds to S4 the claim that all
modal claims are necessarythat is, not only are necessary
truths necessarily necessary, but that sentences that are
possible are necessarily possible. In other words, uPouP.
With modal logic, we are left with the question that mathematicians
faced when suddenly they had multiple geometric systems: Which
one is right? Which of these five modal systems is the real one?
The stance that logicians take is similar to the one David Hilbert
took with respect to mathematical systemstake whichever one
you want and then ask the philosophers questions.
Deontic Logic
Philosophers have made another interesting use of modal logic,
turning it into what is called deontic logic. The term deontology
refers to an approach to logic that is based on duties, absolute
rules that must be followed in order to act morally.

Moral theory utilizes three different statuses for possible acts.
An action could be morally necessary, meaning that you have

to do it.
An action could be morally prohibited; these are things that
you should never do.
An action could be morally permissible, meaning that you
could do it if you want or dont do it if you dont wanteither
way is fine.
Deontological ethicists thought that perhaps modal logic could

be modified to become deontic logic. Morally necessary
actions would be represented with the box (o). Instead of oP
representing the necessary truth of some sentence, it would
represent the necessity of a persons undertaking the act P.
Morally impermissible acts would be represented as oP. This
means that it is necessary to avoid doing P. The diamond ()
would be used for morally permissible acts.
The hope was that in adapting modal logic to a moral logic,

we could take the mushiness and hurt feelings that often
accompany arguments surrounding questions of ethics and
replace these feelings with the rigor and objectivity that we see
in logical argumentation.
Unfortunately, the project has not taken off as much as early

advocates had hoped. A few snags have surfaced. Unlike with
the logic of truth in which every sentence is either true or false
and we never have to choose between truths, there can be
situations in which we have to choose between morally required
acts. Additionally, are ethical sentences intrinsically necessary or
only necessary in a context?

Readings
Epstein, Propositional Logics, chap. 6.

Layman, The Power of Logic, chap. 12.
Questions
1.
Is S5 too strong? Is everything necessarily the way it has to be?
Could things be otherwise? Are the alternative possible worlds really
possible? Are things the way they are because that is the way they
have to be?
2.
Some philosophers have argued that to say that a sentence is true is
to say that it describes reality. For example, to say that the sentence
Madagascar is an island is true, it must be the case in reality that
Madagascar is, indeed, an island. So, if we hold that there are true
sentences about possibilityfor example, if we think it is true that
It is possible that Madagascar wouldnt have been an islandthen
there is a possible world in which Madagascar is not an island. But if
to be true is to describe an aspect of reality, then if we hold sentences
about possibility to be true, then the possible worlds must be part of
realityin other words, the possible worlds must all actually exist. Do
we want to give up on having true sentences about possibility? Do
we want to allow for reality to include not only our world but every
possible world, or is there another way to understand the truth of
sentences about possibility?
3.
Is a deontic logic possible? Could there be a logic that helps us
determine what we should do if we want to act ethically?

Lecture 24
Three-Valued
and Fuzzy Logic
A ll of our lectures thus far have elucidated certain elements of

classical logic. What defines classical logic is the adherence to
a central proposition: the law of the excluded middle. This is the claim
that all sentences have one of two truth-values: true or false. But what
happens if we deny this axiom? Because this is the foundation of all
the logical inferences we have examined so far, the fear is that by
eliminating the law of the excluded middle, we make logic impossible.
It turns out that we dont; we just make it more complicated.
Three-Valued Logic
By denying that all sentences must be either true or false, we
create what we call multivalued logic. The simplest of these is
three-valued logic. We are used to having just two truth-values:
T and F. But we are going to augment this set with one new
member. The name of this new truth-value is a completely
arbitrary choice for which logicians have no unified position.
Some use M for middlethat is, it is some middle value between

true and false. Others use I for indeterminate or U for unknown
or N for neither true nor false. Well use M, partially because it
does not presuppose an interpretation for this new truth-value.
One could develop a full-blown first-order predicate logic, but

for simplicity and to see some of the results more clearly, we
will look at what introducing a new truth-value does for truth-
functional logic.
We can still use truth tables, but they will have p p
to be a bit larger because of the new truth-
value. For negation, for example, instead of T F
having two rows, we will now have three. For M M
a sentence p, we have the truth table shown
at right. F T
The only new information is the middle line. The negation of a

sentence with truth-value M remains M. At first glance, this should
strike us as odd. Negation changes truth-values. But if we think of
M as indeterminate or unknown, then if we dont know whether p
is true, then we also dont know if it is false. So, if p has the value
M, then so should p.
What about the other connectives? Lets make one big truth table
for all three remaining connectives.
We know what the columns are. One each for p and q, and
columns for or, and, and if-then. We have three possible
values for each of the constituent sentences, so we will need nine
rows to capture all of the combinations.
Lets start with the disjunction, p or q. All of our truth-values

from two-valued logic remain the same, so we can enter all of
those into the table.
All that is left are the ones that involve the new value M. We know
that a disjunction is true whenever one of the disjuncts is true. So,
lets put in T for T M or MT.
What about the remaining cases: M F, F M, and M M? If,

again, we think of M as indeterminate or unknown, then all three
of these will be unknown as well, because if they turn out to
be true, so will the sentence, but if they turn out to be false,
so will the sentence. So, we can complete the entire column
for disjunction.
Lecture 24Three-Valued and Fuzzy Logic 225

What about the conjunction, p and q? Again, the two-valued
values remain. So, we can add those to the table.
We know that a conjunction is false when one conjunct is false,

so M&F and F&M will both give us F.
True or unknown remains unknown, and unknown or unknown

is quite unknown, so we can now complete the column.
Last is the conditional: if p, then q. Once again, the two-valued

values can be plugged in.
We know that the only time a conditional is false is when the

antecedent is true and the consequent is false. So, if the
antecedent is false or the consequent is true, then we know
that the conditional is true regardless of the truth-value of the
remaining sentence. This was true for two-valued logic and
remains true for three-valued logic.
If a sentence is TM, then there is a chance that it could end up

TF, which would make it false, and it is possible that it could
end up TT, which would have it end up true. So, TM would
remain M. The same kind of reasoning holds for MF. So, we
can plug these values into the table.
The only remaining case is MM. Different three-valued systems

treat this case differently. Some take this to give us a value of
T, while others take it to give us a value of M. For the sake of
being more intuitive, lets go with MM as M. So, we now have a
complete truth table, as follows.

p q pq p&q p q
T T T T T
T M T M M
T F T F F
M T T M T
M M M M M
M F M F M
F T T F T
F M M F T
F F F F T
One consequence of our choice of MM being M is that the

entire row for p=M and q=M has a value of M. If we also think of
negation, p and q with values of M also remain M.
As a result, no matter what combination of truth-functional

connectives we use in creating molecular sentences and no
matter how complex we make those sentences, the row with all
truth-values M will always end up with the truth-value M.
So, there can be no tautologies or contradictions as we

traditionally think of them. We can have modified or pseudo-
tautologies that have no truth-value of F, or we can have modified
or pseudo-contradictions that have no truth-values of T. But
neither of these will function as tautologies or contradictions did
in two-valued truth-functional logic.
We can still use truth tables to determine validity and invalidity

in exactly the same way. We are still making sure that there are

no cases in which the premises are all true and the conclusion is
false. But now, we have a stronger and weaker sense of validity.
Recall that an argument is invalid if there is even a single case

in which all of the premises are true and the conclusion is false.
But now there are two ways to avoid that. The first is validity, in
which for every case in which all of the premises are true, the
conclusion is also true. Well call this strong validity.
But then there will be the arguments in which there are no cases in
which the premises are true and the conclusion is false, but there
are some cases in which the premises are true and the conclusion
is true and some cases in which the premises are true and the
conclusion is middle. We can think of this as weak validity.
But what about the case in which the premises are all true and
the conclusion is sometimes T and sometimes M? We need a
new third value, an M version of validity for these arguments.
What does this M value really mean? We have been saying, for
the sake of intuition, that it means unknown, but unknown is not a
truth-value.
There is a difference between what is and what we know about

what is. Three-valued logic does not seem apt for describing
what is, but it does seem appropriate for talking about our
knowledgeand our lack thereof.
We can be in one of three states with respect to the truth-value

of a sentence: we can know it is true, we can know it is false, or
we can be in a state of doubt, not knowing whether it is true or
false. As such, three-valued logic does seem to be a good model
of our reasoning when we see ourselves as being in a state of
knowledge about some propositions and a state of ignorance
about others.

With three-valued logic, we deny the law of the excluded middle
by positing a third truth-value. But we could add more. We could
create a four-valued logic or an 18-valued logic. It is not clear
what uses we would have for them, but such utilitarian, pragmatic
questions hold little weight with logicians and mathematicians.
But in all of these cases, we are making a particular assumption

that we could deny and create an even more general logic. Why
must these truth-values be discretethat is, why is it an absolute
partitioning of sentences into one of these truth-values? Why
not make a smooth continuum of truth? In this case, we have an
infinite number of truth-values.
Fuzzy Logic
It is tempting to generalize our use of three-valued logic as a
measure of our knowledge or lack thereof and see fuzzy logic
as a measure of our certainty about a sentences truth. But this is
not what fuzzy logic is about. That is induction. Fuzzy logic is not
concerned with likelihood, but rather deals with fuzzy sets.
A non-fuzzy set is one where all members of the universe are

either in or out of the set. Suppose that the domain is the set of
humans, and the set we want to examine is the set of all pregnant
people. This is non-fuzzy; either you are or are not pregnant.
There is a clear and absolute line between in and out, and no one
straddles that line.
But some sets are not like that. Some sets are fuzzy; objects in
the universe can be in the set to varying degrees. Again, take the
domain to be humans, and now consider the property of hairiness.
For example, when men grow a beard, they move further into the
fuzzy set. But over the years, as their hairline recedes, they move
further out of the fuzzy set. Hairiness ranges from zero (hairless)
on one extreme to one on the other extreme (werewolf-like).

All humans have some degree of hairiness. So, while all people
are hairy, some sets are fuzzy as well. Logicians have devised
operations on these sets, called fuzzy connectives, that allow us
to have a fuzzy version of truth-functional logic and even a fuzzy
version of predicate logic. The values plugged in are numbers
between zero and one, and the result of the connectives is a new
value between zero and one.
There are several versions of fuzzy logic, each defining these

connectives in slightly different fashion. Which of the definitions
we use depends on the needs that the fuzzy logic serves.
The standard example of a system that utilizes fuzzy logic

is the heating and cooling system in your house. There is a
thermostat connected to a switch that controls the heating and
cooling system. You ask your system to keep your home at some
particular temperature.
If you have an old system, then this is governed by classical logic.

Let p be the sentence The temperature is the one I desire. If
p is true, then the system does nothing. If p is false, then the
system turns onfor example, the heater is on full blast until the
temperature is reached, at which point it is turned off.
But more modern systems use fuzzy logic to be more energy

efficient. If the temperature is the one desiredif p has a value
of onethen the system does nothing. But if the temperature
deviates from the desired temperature, then it springs to action in
a more intricate fashion.
If it is too cold, it turns on the heat to an appropriate degree,

depending on the degree of deviation from the desired
temperature. If it is way too cold, it cranks the heat way up. If it is
just a little too cold, it gives the heater just a small shot of power
to let it do a little bit of work. This is a different logic governing the
operation of the systemit is a fuzzy logic.

Fuzzy logic is a generalization of multivalued logic that we
created by seeing what would happen if we denied the central
claim of classical logic, the law of the excluded middle. We found
that if we were clever about it, we could find a way of reasoning
that is helpful and useful. And that is what logic is really all about.
Readings
Epstein, Propositional Logics, chap. 8.

Frielberger and McNeill, Fuzzy Logic.
Questions
1.
In three-valued logic, the negation of the new value middle is still
middle. Does it make sense to make the negation of a value the same
value? Is that not the opposite of what we mean by negation?
2.
Fuzzy logic might be usefulthat is, it might be helpful to think of
truth as if it varies smoothlybut does it really? Isnt truth more like
being pregnant, where you are or you arent? Is there really a sense
of in between when it comes to truth?

Bibliography
Barker, Stephen. The Elements of Logic. New York: McGraw-Hill, 1989. A

classic textbook that introduces logical concepts for the non-technician.
Bradburn, Norman, Seymour Sudman, and Brian Wansink. Asking Questions:

The Definitive Guide to Questionnaire Design. Hoboken, NJ: Jossey-Bass,
2004. A clear and comprehensive discussion on how to conduct a poll and
the pitfalls to be avoided.
Coffa, J. Alberto. The Semantic Tradition: From Kant to Carnap. New York:
Cambridge University Press, 1991. A detailed discussion of the emergence
of analytic philosophy from the advances of science, mathematics, and logic
in the 19th and 20th centuries.
Copi, Irving. Introduction to Logic. New York: Macmillan, 1972. One of the
best-selling and longest-lived college-level logic texts out there.
Damer, T. Edward. Attacking Faulty Reasoning. Boston: Wadswoth, 2009. A

guidebook to the fallacies that can be committed in making arguments.
Durkheim, Emile. The Rules of Sociological Method. New York: Free Press,
1966. One of the founding fathers of the field of sociology carefully considers
what is the subject matter of sociology and how sociological research should
be done.
Epstein, Richard. Propositional Logics. Belmont, CA: Wadsworth, 2001.

A college-level text that examines a full range of logical languages, their
structures, and usages.
Euclid. The Elements. New York: Dover, 1952. The classic axiomatization of
plane geometry, where all of the results are derived through deductive proofs.
Fine, Cordelia. A Mind of Its Own: How Your Brain Distorts and Deceives.
New York: Norton, 2006. A popular account of the ways in which our neural
wiring leads us into errors while thinking it is correct.
Frielberger, Paul, and Daniel McNeill. Fuzzy Logic. New York: Simon &
Schuster, 1993. A popularly accessible account of the development of fuzzy
logic and its applications to real-world systems.
Gimbel, Steven. Exploring the Scientific Method. Chicago: University of
Chicago Press, 2011. An investigation into the various accounts that have
been proposed by scientists and philosophers of science to account for the
logic behind the scientific method.
Gray, Jeremy. The Hilbert Challenge. New York: Cambridge University Press,
2001. An examination of the 23 problems David Hilbert charged mathematics
with solving in the 20th century. It discusses the problems, the attempts to
solve them, and whether the attempts were successful.
Hurley, Patrick. Logic. Belmont, CA: Wadsworth, 1996. A college-level

textbook that examines both formal and informal logic.
Kahane, Howard. Logic and Philosophy. Belmont, CA: Wadsworth, 1978. A

college-level textbook that examines deductive logic and the philosophical
questions that gave rise to the interest in the field.
Kelley, David. The Art of Reasoning. New York: W. W. Norton, 1998. A

college-level textbook that covers both inductive and deductive reasoning.
Kramer, Edna. The Nature and Growth of Modern Mathematics. Princeton:

Princeton University Press, 1981. An encyclopedic account of the history
of mathematical thought from ancient times through the first half of the
20th century.
Layman, Stephen. The Power of Logic. Toronto: Mayfield, 1999. A college-level

logic textbook that covers deduction, induction, probability, and modal logic.
Nagel, Ernest, and James Newman. Gdels Proof. New York: New York
University Press, 1958. A popularly accessible account of the development
of Gdels incompleteness theorem and the structure of the argument itself.
Sutherland, Stuart. Irrationality: The Enemy Within. London: Constable and

Company, 1992. A popular book that examines the cognitive biases that
humans possess and the ways in which they lead us into error.
Yandell, Ben. The Honors Class: Hilberts Problems and Their Solvers. Natick,
MA: A K Peters, 2002. An examination of the 23 problems David Hilbert charged
mathematics with solving in the 20th century. It discusses the problems, the
attempts to solve them, and whether the attempts were successful.
Bibliography 233
Image Credits
15. . . . . . . . . . . . . . . . . . . . . . . . . . . GeorgiosArt/iStock/Thinkstock.
69 . . . . . . . . . Vienna museum/Wikimedia Commons/Public Domain.
73. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . denisk0/iStock/Thinkstock.
204 . . . . . . . . . . . Gttingen State and University Library/Manuscripts

and Scholarly Collections/flickr/Public Domain.
210. . . . . . . . . . . . . . . . . . . . . . . . . . . . . wynnter/iStock/Thinkstock.
213. . . . . . . . . . . . Conquistador/Wikimedia Commons/Public Domain.

4215 - An Introduction To Formal Logic

Uploaded by

Copyright:

Available Formats

4215 - An Introduction To Formal Logic

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

4215 - An Introduction To Formal Logic

Uploaded by

Copyright:

Available Formats

What is the topic of the document?

What is the topic of the document?

Who is the author of the document?

Who is the author of the document?

Topic Subtopic

Philosophy Modern Philosophy

THE GREAT COURSES

Copyright The Teaching Company, 2016

Printed in the United States of America

This book is in copyright. All rights reserved.

Without limiting the rights under copyright reserved above,

D r. Steven Gimbel received his

Dr. Gimbels research focuses on the philosophy of science, exploring

Dr. Gimbels previous Great Course is Redefining Reality: The

Table of Contents iii

L ogic is the study of rational argumentation. A belief is rational if

Then, well introduce a wide range of logical concepts. We will

To handle different types of sentences, we will develop two logical

The second logical language is first-order predicate logic, which

The pinnacle of first-order predicate logic for us will be the ability to

Well examine modal logic, which deals with the concepts of

Why Study Logic?

S ince the times of classical Greece, the definition of human

Innate Human Irrationality

Contrary to our beliefs, we are not wired to be completely rational.

It was explained that this was a

Not conforming with the group is something we fear, and we

Interviewed after the experiment, most subjects said that they

Lecture 1Why Study Logic? 5

Aschs study was about what people would dowhether they

This is what psychologist Irving Janis called groupthink, which

Janis, Asch, and a few generations of researchers following them

If there is a strong, charismatic leader espousing the view, people

If there are rewards for believing and punishments for disbelieving,

Our beliefs are affected by the beliefs, actions, and personalities

6 An Introduction to Formal Logic

Many cognitive biases have been discovered and replicated in

There is also what has come to be known as irrational escalation

Humans are not purely rational beings. Our ultimate interest in

Rational Actions and Thoughts

Researchers Rolf Reber and Norbert Schwarz have shown that

Psychologists define the halo effect as follows: If someone is

Lecture 1Why Study Logic? 7

We have a positive outcome biasthat is, we think that we are

We overestimate our abilities. The overconfidence effect is the

The Dunning-Kruger effect is the name given to the fact that

In the early 1980s, psychologist Benjamin Libet ran a series of

We find this kind of backfilling a justification behavior on different

8 An Introduction to Formal Logic

We will do whatever we need to in order to save our preexisting

Emile Durkheim, one of the founding fathers of sociology,

These are beliefs that become invisible to us because they are

Where are our culturally generated blind spots? What do we

Logic and Reasoning

Lecture 1Why Study Logic? 9

It might turn out that there is good reason to believe what we

When we are aware of our psychological and cultural biases, we

Durkheim, The Rules of Sociological Method.

10 An Introduction to Formal Logic