Review Copy

A BRIEF INTRODUCTION TO ARGUMENTS 5

A BRIEF INTRODUCTION TO ARGUMENTS

Evaluating Arguments rope. On the other hand, it is very easy to give appall-
ingly bad arguments for true conclusions: Every sen-
The main tool of philosophy is the argument. An ar- tence beginning with the letter c is true;Chickens lay
gument is any sequence of statements intended eggs begins with the letter c; Therefore, chickens lay
to establishor at least to make plausiblesome eggs.
particular claim. For example, if I say that Vancouver But there is a deeper reason why the evaluation of
is a better place to live than Toronto because it has arguments doesnt begin by assessing the truth of the
a beautiful setting between the mountains and the conclusion. The whole point of making arguments
ocean, is less congested, and has a lower cost of liv- is to establish whether or not some particular claim
ing, then I am making an argument. The claim which is true or false. An argument works by starting from
is being defended is called the conclusion, and the some claims which, ideally, everyone is willing to ac-
statements which together are supposed to show cept as truethe premisesand then showing that
that the conclusion is (likely to be) true are called the something interestingsomething newfollows
premises. Often arguments will be strung together in from them: i.e., an argument tells you that if you be-
a sequence, with the conclusions of earlier arguments lieve these premises, then you should also believe this
featuring as premises of the later ones. For example, I conclusion. In general, it would be unfair, therefore, to
might go on to argue that since Vancouver is a better simply reject the conclusion and suppose that the ar-
place to live than Toronto, and since ones living con- gument must be a bad onein fact, it would often be
ditions are a big part of what determines ones hap- intellectually dishonest. If the argument were a good
piness, then the people who live in Vancouver must, one, then it would show you that you might be wrong
in general, be happier than those living in Toronto. in supposing its conclusion to be false; and to refuse
Usually, a piece of philosophy is primarily made up of to accept this is not to respond to the argument but
chains of argumentation: good philosophy consists simply to ignore it.2
of good arguments; bad philosophy contains bad It follows that there are exactly two reasonable
arguments. ways to criticize an argument: the first is to question
What makes the difference between a good and a the truth of the premises; and the second is to
bad argument? Its important to notice, first of all, that question the claim that if the premises are true then
the difference is not that good arguments have true the conclusion is true as wellthat is, one can critique
conclusions and bad arguments have false ones. A the strength of the argument. Querying the truth of
perfectly good argument might, unluckily, happen to
have a conclusion that is false. For example, you might  2IFRXUVHRFFDVLRQDOO\\RXPLJKWOHJLWLPDWHO\NQRZ
argue that you know this rope will bear my weight be- IRUVXUHWKDWWKHFRQFOXVLRQLVIDOVHDQGWKHQ\RXFRXOG
cause you know that the ropes rating is greater than VDIHO\LJQRUHDUJXPHQWVZKLFKWU\WRVKRZLWLVWUXH
my weight, you know that the ropes manufacturer is IRUH[DPSOHDIWHUWKHURSHEUHDNV,FRXOGGLVPLVV\RXU
a reliable one, you have a good understanding of the DUJXPHQWWKDWLWLVVDIHDJDLQWKRXJKWKLVZRXOGQRW
safety standards which are imposed on rope makers VKRZWKDW\RXUDUJXPHQWZDVEDGMXVWWKDW,QHHGQRW
and vendors, and you have carefully inspected this rope EH SHUVXDGHG WKDW WKH FRQFOXVLRQ LV WUXH +RZHYHU
for flaws. Nevertheless, it still might be the case that this WKLVZLOOQRWGRIRUSKLORVRSKLFDODUJXPHQWVDOOLQWHU
rope is the one in 50 million which has a hidden defect HVWLQJSKLORVRSK\GHDOVZLWKLVVXHVZKHUHWKRXJKZH
causing it to snap. If so, that makes me unlucky, but it PD\KDYHUPRSLQLRQVZHFDQQRWMXVWLQVLVWWKDWZH
doesnt suddenly make your argument a bad onewe NQRZDOOWKHDQVZHUVDQGFDQWKHUHIRUHDIIRUGWRLJQRUH
were still being quite reasonable when we trusted the UHOHYDQWDUJXPHQWV
6 PHILOSOPHY

the premises (i.e., asking whether its really true that then the conclusion necessarily must be, or they are
Vancouver is less congested or cheaper than Toronto) not. Strong deductive arguments are called valid;
is fairly straightforward. The thing to bear in mind is otherwise, they are called invalid. The main thing to
that you will usually be working backwards down notice about validity is that its definition is an if
a chain of argumentation: that is, each premise of a then statement: if the premises were true, then the
philosophers main argument will often be supported conclusion would be. For example, an argument can
by sub-arguments, and the controversial premises in be valid even if its premises and its conclusion are not
these sub-arguments might be defended by further true: all that matters is that if the premises had been
arguments, and so on. Normally it is not enough to true, the conclusion necessarily would have been as
merely demand to know whether some particular well. This is an example of a valid argument:
premise is true: one must look for why the arguer 1. Either bees are rodents or they are birds.
thinks it is true, and then engage with that argument. 2. Bees are not birds.
Understanding and critiquing the strength of 3. Therefore bees are rodents.
an argument (either your own or someone elses) If the first premise were true, then (since the second
is somewhat more complex. In fact, this is the main premise is already true), the conclusion would have to
subject of most books and courses in introductory be truethats what makes this argument valid. This
logic. When dealing with the strength of an argument, example makes it clear that validity, though a highly
it is usual to divide arguments into two classes: desirable property in an argument, is not enough all
deductive arguments and inductive arguments. Good by itself to make a good argument: good deductive
deductive arguments are the strongest possible kind arguments are both valid and have true premises.
of argument: if their premises are true, then their When arguments are good in this way they are called
conclusion must necessarily be true. For example, if sound: sound arguments have the attractive feature
all bandicoots are rat-like marsupials, and if Billy is that they necessarily have true conclusions. To show
a bandicoot, then it cannot possibly be false that that an argument is unsound, it is enough to show
Billy is a rat-like marsupial. On the other hand, good that it is either invalid or has a false premise.
inductive arguments establish that, if the premises It bears emphasizing that even arguments which
are true, then the conclusion is highly likely (but not have true premises and a true conclusion can be
absolutely certain) to be true as well. For example, I unsound. For example:
may notice that the first bandicoot I see is rat-like, and 1. Only US citizens can become the President of
the second one is, and the third, and so on; eventually, America.
I might reasonably conclude that all bandicoots 2. George W. Bush is a US citizen.
are rat-like. This is a good argument for a probable 3. Therefore, George W. Bush was elected Presi-
conclusion, but nevertheless the conclusion can never dent of America.
be shown to be necessarily true. Perhaps a non-rat-like This argument is not valid, and therefore it should not
bandicoot once existed before I was born, or perhaps convince anyone who does not already believe the
there is one living now in an obscure corner of New conclusion to start believing it. It is not valid because
Guinea, or perhaps no bandicoot so far has ever been the conclusion could have been false even though
non-rat-like but at some point, in the future, a mutant the premises were true: Bush could have lost to Gore
bandicoot will be born that in no way resembles a rat, in 2000, for example. The question to ask, in thinking
and so on. about the validity of arguments is this: Is there a
coherent possible world, which I can even imagine, in
which the premises are true and the conclusion false?
Deductive Arguments and Validity
If there is, then the argument is invalid.
The strength of deductive arguments is an on/off When assessing the deductive arguments that you
affair, rather than a matter of degree. Either these encounter in philosophical work, it is often useful to
arguments are such that if the premises are true try to lay out, as clearly as possible, their structure. A
 Review Copy
A BRIEF INTRODUCTION TO ARGUMENTS 7

standard and fairly simple way to do this is simply to which appears in the previous argument about God
pull out the logical connecting phrases and to replace, and evilis modus tollens: if P then Q; not-Q; therefore
with letters, the sentences they connect. Five of the not-P. A disjunctive syllogism works as follows: either
most common and important logical operators are P or Q; not-P; therefore Q. A hypothetical syllogism
and, or, it is not the case that, if then , and if and only has the structure: if P then Q; if Q then R; therefore if
if. For example, consider the following argument: If P then R. Finally, a slightly more complicated but still
God is perfectly powerful (omnipotent) and perfectly common argument structure is sometimes called
good, then no evil would exist. But evil does exist. a constructive dilemma: either P or Q; if P then R; if Q
Therefore, God cannot be both omnipotent and then R; therefore R.
perfectly good, so either God is not all-powerful or he
is not perfectly good. The structure of this argument
Inductive Arguments and Inductive
could be laid bare as follows:
1. If (O and G) then not-E. Strength
2. E. I noted above that the validity of deductive argu-
3. Therefore not-(O and G). ments is a yes/no affairthat a deductive argument
4. Therefore either not-O or not-G. is either extremely strong or it is hopelessly weak.
Revealing the structure in this way can make it easier This is not true for inductive arguments. The strength
to see whether or not the argument is valid. And in of an inductive argumentthe amount of support
this case, it is valid. In fact, no matter what O, G, and the premises give to the conclusionis a matter of
E stand forno matter how we fill in the blanks degree, and there is no clear dividing line between
any argument of this form must be valid. You could the strong inductive arguments and the weak ones.
try it yourselfinvent random sentences to fill in for Nevertheless, some inductive arguments are obvious-
O, G, and E, and no matter how hard you try, you will ly much stronger than others, and it is useful to think
never produce an argument with all true premises a little bit about what factors make a difference.
and a false conclusion.3 What this shows is that There are lots of different types and structures
validity is often a property of the form or structure of of inductive arguments; here I will briefly describe
an argument. (This is why deductive logic is known
four which are fairly representative and commonly
as formal logic. It is not formal in the sense that it is
encountered in philosophy. The first is inductive
stiff and ceremonious, but because it has to do with
generalization. This type of argument is the prototype
argument forms.)
inductive argumentindeed, it is often what people
Using this kind of shorthand, therefore, it is possible
mean when they use the term inductionand it has
to describe certain general argument forms which are
the following form:
invariably valid and whichsince they are often used
1. x per cent of observed Fs are G.
in philosophical writingit can be handy to look out
2. Therefore x per cent of all Fs are G.
for. For example, a very common and valuable form
That is, inductive generalizations work by inferring a
of argument looks like this: if P then Q; P; therefore
claim about an entire population of objects from data
Q. This form is often called modus ponens. Another
about a sample of those objects. For example:
(a) Every swan I have ever seen is white, so all
 6LQFH WKH DUJXPHQW DERXW *RG DQG HYLO LV YDOLG WKHQ
swans (in the past and future, and on every part
ZH DUH OHIW ZLWK RQO\ WZR SRVVLELOLWLHV (LWKHU DOO LWV
of the planet) are white.
SUHPLVHV DUH WUXH DQG WKHQ LW LV VRXQG DQG LWV FRQFOX
VLRQ PXVW LQHVFDSDEO\ EH WUXH 2U RQH RI LWV SUHPLVHV (b) Every swan I have ever seen is white, so probably
LV IDOVH LQ ZKLFK FDVH WKH FRQFOXVLRQ PLJKW EH IDOVH all the swans around here are white.
WKRXJK ZH ZRXOG VWLOO QRW KDYH VKRZQ WKDW LW LV IDOVH (c) 800 of the 1,000 rocks we have taken from the
7KH RQO\ ZD\ WR HIIHFWLYHO\ FULWLTXH WKLV DUJXPHQW Moon contain silicon, so probably around 80%
WKHUHIRUH LV WR DUJXH DJDLQVW RQH RI WKH FODLPV  DQG  of the Moons surface contains silicon.
 Review Copy
8 PHILOSOPHY

(d) We have tested two very pure samples of cop- (c) Eagles and robins are alike in having wings,
per in the lab and found that each sample has feathers, claws, and beaks. Eagles kill and eat
a boiling point of 2,567C; we conclude that sheep. Therefore, robins kill and eat sheep.
2,567C is the boiling point for copper. (d) Anselms ontological argument has the same
(e) Every intricate system I have seen created (such argumentative form as Gaunilos perfect island
as houses and watches) has been the product argument. But Gaunilos argument is a patently
of intelligent design, so therefore all intricate bad argument. So there must be something
systems (including, for example, frogs and wrong with the ontological argument.
volcanoes) must be the product of intelligent (e) An eye and a watch are both complex systems
design. in which all of the parts are inter-dependent
The two main considerations when assessing the and where any small mis-adjustment could
strength of inductive generalizations are the follow- lead to a complete failure of the whole. A watch
ing. First, ask how representative is the sample? How is the product of intelligent design. Therefore,
likely is it that whatever is true of the sample will also the eye must also be the product of intelligent
be true of the population as a whole? For instance, al- design (i.e., God exists).
though the sample size in argument (c) is much larger The strength of an argument from analogy depends
than that in argument (d), it is much more likely to be mostly on two things: first, the degree of positive
biased: we know that pure copper is very uniform, so relevance that the noted similarities (F, G, H ) have
a small sample will do; but the surface of the Moon to the target property X; and second, the absence of
might well be highly variable, and so data about the relevant dissimilaritiesproperties which A has but
areas around moon landings may not be representa- B does not, which make it less likely that A is X. For
tive of the surface as a whole. Second, it is important example, the similarity (brain size) between humans
to gauge how cautious and accurate the conclusion is, and dolphins cited in argument (a) is much more
given the datahow far beyond the evidence does relevant to the target property (planning) than are
it go? The conclusion to argument (a) is a much more the similarities cited in argument (b). This, of course,
radical inference from the data than that in argument makes (a) a much stronger argument than (b). The
(b); consequently, though less exciting, the conclu- primary problem with argument (c), on the other
sion of argument (b) is much better supported by the hand, is that we know that robins are much smaller
premise. and weaker than eagles and this dissimilarity makes it
A second type of inductive argument is an far less likely that they kill sheep.
argument from analogy. It most commonly has the A third form of inductive argument is often
following form: called inference to the best explanation or sometimes
1. Object (or objects) A and object (or objects) B abduction. This kind of argument works in the
are alike in having features F, G, H, following way. Suppose we have a certain quantity of
2. B has feature X. data to explain (such as the behavior of light in various
3. Therefore A has feature X as well. media, or facts about the complexity of biological
These examples illustrate arguments from analogy: organisms, or a set of ethical claims). Suppose also
(a) Human brains and dolphin brains are large, that we have a number of theories which account for
compared to body size. Humans are capable of this data in different ways (e.g., the theory that light
planning for the future. So, dolphins must also is a particle, or the theory that light is a wave, or the
be capable of planning for the future. theory that it is somehow both). One way of arguing
(b) Humans and dolphins are both mammals and for the truth of one of these theories, over the others,
often grow to more than five feet long. Humans is to show that one theory provides a much better
are capable of planning for the future. So, dol- explanation of the data than the others. What counts
phins must also be capable of planning for the as making a theory a better explanation can be a bit
future. tricky, but some basic criteria would be:
 A BRIEF INTRODUCTION TO ARGUMENTS 9

1. The theory predicts all the data we know to be around us. Therefore, the theory of evolution is
true. true.
2. The theory explains all this data in the most The final type of inductive argument that I want to
economical and theoretically satisfying way mention here is usually called reductio ad absurdum,
(scientists and mathematicians often call this which means reduction to absurdity. It is always a
the most beautiful theory). negative argument, and has this structure:
3. The theory predicts some new phenomena 1. Suppose (for the sake of argument) that posi-
which turn out to exist and which would be a tion p were true.
big surprise if one of the competing theories 2. If p were true then something else, q, would
were true. (For example, one of the clinchers for also have to be true.
Einsteins theory of relativity was the observa- 3. However q is absurdit cant possibly be true.
tion that starlight is bent by the suns gravity. 4. Therefore p cant be true either.
This would have been a big surprise under the In fact, this argument style can be either inductive or
older Newtonian theory, but was predicted by deductive, depending on how rigorous the premises
Einsteins theory.) 2 and 3 are. If p logically implies q, and if q is a logical
Here are some examples of inferences to the best contradiction, then it is deductively certain that p
explanation: cant be true (at least, assuming the classical laws of
(a) When I inter-breed my pea plants, I observe logic). On the other hand, if q is merely absurd but not
certain patterns in the properties of the plants literally impossible, then the argument is inductive:
produced (e.g., in the proportion of tall plants, it makes it highly likely that p is false, but does not
or of plants which produce wrinkled peas). If prove it beyond all doubt.
the properties of pea plants were generated Here are a few examples of reductio arguments:
randomly, these patterns would be highly sur- (a) Suppose that gun control were a good idea.
prising. However, if plants pass on packets That would mean its a good idea for the gov-
of information (genes) to their offspring, the ernment to gather information on anything
patterns I have observed would be neatly ex- we own which, in the wrong hands could be
plained. Therefore, genes exist. a lethal weapon, such as kitchen knives and
(b) The biological world is a highly complex and in- baseball bats. But that would be ridiculous. This
ter-dependent system. It is highly unlikely that shows gun control cannot be a good idea.
such a system would have come about (and (b) If you think that foetuses have a right to life
would continue to hang together) from the because they have hearts and fingers and toes,
purely random motions of particles. It would be then you must believe that anything with a
much less surprising if it were the result of con- heart, fingers, and toes has a right to life. But
scious design from a super-intelligent creator. that would be absurd. Therefore, a claim like
Therefore, the biological world was deliberately this about foetuses cannot be a good argu-
created (and therefore, God exists). ment against abortion.
(c) The biological world is a highly complex and in- (c) Suppose, for the sake of argument, that this
ter-dependent system. It is highly unlikely that is not the best possible world. But that would
such a system would have come about (and mean God had either deliberately chosen to
would continue to hang together) from the create a sub-standard world or had failed to
purely random motions of particles. It would be notice that this was not the best of all possible
much less surprising if it were the result of an worlds, and either of these options is absurd.
evolutionary process of natural selection which Therefore, it must be true that this is the best of
mechanically preserves order and eliminates all possible worlds.
randomness, and which (if it existed) would (d) The anti-vitalist says that there is no such thing
produce a world much like the one we see as vital spirit. But this claim is self-refuting. The
 Review Copy
10 PHILOSOPHY

speaker can be taken seriously only if his claim ment. The following are implicit ad hominem argu-
cannot. For if the claim is true, then the speaker ments: You say you want to close down the church?
does not have vital spirit and must be dead. Well, Hitler and Stalin would agree with you! and We
But if he is dead, then his statement is a mean- shouldnt trust the claim, by philosophers such as An-
ingless string of noises, devoid of reason and selm, Aquinas, and Leibniz, that God exists, since they
truth. (If you want more information, see Paul were all Christian philosophers and so of course they
Churchlands Eliminative Materialism and the were biased. Such attacks are fallacious because they
Propositional Attitudes, Journal of Philosophy have nothing at all to do with how reasonable a claim
78 [1981].) is: even if the claim is false, ad hominem attacks do
The critical questions to ask about reductio arguments nothing to show this.
are simply: Does the supposedly absurd consequence Straw man arguments are particularly devious,
follow from the position being attacked? and Is it and this fallacy can be hard to spot (or to avoid com-
really absurd? mitting) unless great care is taken. The straw man
fallacy consists in misrepresenting someone elses
position so that it can be more easily criticized. It is
A Few Common Fallacies like attacking a dummy stuffed with straw instead of
Just as it can be useful to look for common patterns of a real opponent. For example, its not uncommon to
reasoning in philosophical writing, it can also be help- see attacks on pro-choice activists for thinking that
ful to be on guard for a few recurring fallaciesand, abortion is a good thing. However, whatever the mer-
equally importantly, to take care not to commit them its of either position, this objection is clearly unfair
in your own philosophical writing. Here are four com- no serious abortion advocates think it is a positively
mon ones: good thing to have an abortion; at most they claim
Begging the question does not mean, as the me- that (at least in some circumstances) it is a lesser evil
dia would have us believe, stimulating one to ask a than the alternative. Heres an even more familiar ex-
further question; instead, it means to assume as true ample, containing two straw men, one after the other:
(as one of your premises) the very same thing which We should clean out the closets. Theyre getting a
you are supposedly attempting to prove. This fal- bit messy. Why, we just went through those closets
lacy is sometimes called circular reasoning or even last year. Do we have to clean them out every day? I
(the old Latin name) petitio principii. To argue, for ex- never said anything about cleaning them out every
ample, that God exists because (a) it says in the Bible day. You just want to keep all your junk forever, which
that God exists, (b) God wrote the Bible, and (c) God is simply ridiculous.
would not lie, is to commit a blatant case of begging Arguments from ignorance, finally, are based on the
the question. In this case, of course, one would have assumption that lack of evidence for something is evi-
no reason to accept the premises as true unless one dence that it is false, or that lack of evidence against
already believed the conclusion. Usually, however, something is evidence for its truth. Generally, neither
arguments that beg the question are a little more of these assumptions are reliable. For example, even if
disguised. For example, Adultery is immoral, since we could find no good proof to show that God exists,
sexual relations outside marriage violate ethical prin- this would not, all by itself, suffice to show that God
ciples, or Terrorism is bad, because it encourages does not exist: it would still be possible, for example,
further acts of terrorism, are both instances of circu- that God exists but transcends our limited human
lar reasoning. reason. Consider the following argument by Sena-
Arguing ad hominem means attacking or rejecting tor Joseph McCarthy, about some poor official in the
a position not because the arguments for it are poor, State Department: I do not have much information
but because the person presenting those arguments on this except the general statement of the agency
is unattractive in some way: i.e., an attack is directed that there is nothing in the files to disprove his Com-
at the person (ad hominem) rather than at their argu- munist connections.
 Review Copy
A BRIEF INTRODUCTION TO ARGUMENTS 11

Suggestions for Critical Reflection 2000); an equally brief but highly practical primer
on arguing is Anthony Westons A Rulebook for Argu-
1. Suppose some deductive argument has a ments (Hackett, 2001). There are many books which
premise which is necessarily false. Is it a valid competently lay out the nuts and bolts of formal logic:
argument? Richard Jeffreys Formal Logic: Its Scope and Limits (Mc-
2. Suppose some deductive argument has a con- Graw-Hill, 1991) is short but rigorous and clear; The
clusion which is necessarily true. Is it a valid ar- Logic Book by Bergmann, Moor, and Nelson (McGraw-
gument? From this information alone, can you Hill, 1998), on the other hand, is rather painstaking
tell whether it is sound? but is one of the most complete texts. An interesting
3. Is the following argument form valid: if P then book which explains not only classical formal logic
Q; Q; therefore P? How about: if P then Q; not-P; but also makes accessible some more recently devel-
so not-Q? oped logical languages, such as modal logic and in-
4. No inductive argument is strong enough to tuitionistic logic, is Bell, DeVidi, and Solomons Logical
prove that its conclusion is true: the best it can Options (Broadview Press, 2001). Two somewhat older
do is to show that the conclusion is highly prob- texts, which were used to teach many of the current
able. Does this make inductive arguments bad generation of professional philosophers and are still
or less useful? Why dont we restrict ourselves much used today, are Wilfrid Hodgess Logic (Penguin,
to using only deductive arguments? 1977) and E.J. Lemmons Beginning Logic (Hackett,
5. Formal logic provides mechanical and reliable 1978).
methods for assessing the validity of deductive One of the best introductory texts on inductive
arguments. Do you think there might be some logic is Brian Skyrmss Choice & Chance (Wadsworth,
similar system for evaluating the strength of in- 2000). Other good texts include Copi and Burgess-
ductive arguments? Jacksons Informal Logic (Prentice Hall, 1995), Fogelin
6. I have listed four important fallacies; can you and Sinnott-Armstrongs Understanding Arguments
identify any other common patterns of poor (Harcourt, 2001), and Douglas Waltons Informal Logic:
reasoning? A Handbook for Critical Argumentation (Cambridge
University Press, 1989). Quite a good book on falla-
cies is Attacking Faulty Reasoning by T. Edward Damer
Suggestions for Further Reading
(Wadsworth, 2000), while Darrell Huffs How to Lie with
An entertaining, thought-provoking and brief intro- Statistics (W.W. Norton, 1954) is an entertaining guide
duction to logic can be found in Graham Priests Log- to the tricks that can be played with bad inductive ar-
ic: A Very Short Introduction (Oxford University Press, guments in, for example, advertising.