Deduction and Induction

by Keith Vore

Introduction. Let’s start with a story. A young man has taken a position in an airport
responsible for helping arriving travelers. It is a busy job. During the morning rush, he
notices a beautiful 35-year old woman enter the terminal, and notes a moderate bulge in
her stomach. In a rush to make acquaintance, he smiles and says, “Madam, you have
a glow about you. When is your baby due?” The woman’s glowering answer, of
course: “I am not expecting…”

Who hasn’t made this terrible mistake, or at least almost made it, only to have a spouse
or significant other kick us in the shins beneath the table, before we opened our mouth
to reveal our ignorance?

When we perceive our environment, we often have to interpret those perceptions—we

have to make sense of them.1 The airport story above begs a question: how do we
infer reliable conclusions and prevent this sort of thing from happening? We must be
able to draw accurate inferences, since the complete picture is often missing. In
drawing those inferences, however, we must be able to distinguish between what is
believable from what is not believable. This paper is designed to help you better
understand the processes of deduction and induction, and the pitfalls associated with
each. We first discuss deduction, followed by induction and its forms, before turning to
why this information is important for future red teamers.

Deduction. Deduction consists of taking two or more premises—assertions—and from

those premises, extracting a conclusion which already exists within the premises.2 Here
is an example:

 Premise: All men are mortal

 Premise: Socrates is a man
 Conclusion: Socrates is mortal

The example reflects a premise that all men will someday expire, and that Socrates is
an example of one of these men. The logical conclusion is that, as an example of the
population all of whom will eventually die, Socrates will eventually die. This example is
a syllogism, a specific form of deductive reasoning. Note that this deduction does not
reflect any “new” knowledge; instead, it merely reflects specific knowledge already
inherent in the first and second premise.

1
Here is another example:

 Premise: All politicians are liars

 Premise: All used car salesmen are politicians
 Conclusions: All used car salesmen are liars

Regardless of your past experience in purchasing automobiles, you no doubt note that
something is amiss in this example. At this juncture, we make two points:

1) One or both of the premises may not, in fact, be true; and

2) This is nevertheless a valid deductive argument.

By valid, we only mean that the conclusion necessarily follows from the premises. This
makes the syllogism “technically” correct to a logician, however ridiculous the
conclusion may be in real life. Where the deduction falls short is that one (or more) of
the premises is questionable. For a deductive argument to be practical, it must not only
be valid—that is, that the conclusion necessarily follows from its premises—the
deduction must also have premises which are true. A deductive argument that is both
valid and contains true premises is called a sound argument.3

A method to test the soundness of a deductive argument is to search for

counterexamples. If a deductive argument is sound, then the premises are true and
the conclusion must necessarily follow from them. If, however, you can think of an
example that disproves the conclusion, then you have shown that the deduction is
unsound. From the example above, do you know any used car salesmen who are not
liars? All kidding aside, perhaps you do—or someone you know does. Once you’ve
found a counterexample, you can then more closely examine the premises themselves
to see whether one or more are untrue. In this case, both premises are untrue. Using a
counterexample to assess the soundness of an argument is an example of
disconfirming it: refuting it by finding evidence which disconfirms the assertion.

The premise “All politicians are liars” is fairly easy to assess as false. Other premises,
however, may contain hidden assumptions that are much more challenging to spot. For
example: “Canadian fishermen have the exclusive right to harvest [fish in Canada],
because the fish are coming to Canadian rivers to spawn.”4 On the face of it, this might
appear to be a sound argument. But within the first part of the argument is the word
“exclusive,” as well as the notion that Canadians are the only fishermen allowed to
harvest fish in Canada. Perhaps Canadian law asserts that claim, but it is highly
unlikely; many Americans fish annually in Canada, with legitimate fishing licenses
issued by Canadian authorities. Vigilance is required in assessing whether premises
contain hidden—implicit—assumptions.

2
Red teamers should therefore ensure that deductive reasoning is sound—valid, with
true premises. “A sound argument is the most compelling reason one can come up with
to convince others to agree with her belief or position. A sound argument is a
proof…Refusing to accept the conclusion of a sound argument would be illogical.” 5

Deduction, however, is not limited to the syllogisms. Deductive reasoning can be much
more expansive in terms of the number of premises used in arriving at a conclusion. 6
Dr. Watson credited the great detective Sherlock Holmes with powerful deductive
abilities; Watson felt that Holmes “could see the logical consequences of apparently
disconnected reasons, the number and complexity of which left others at a loss.”7
Watson is essentially saying that Sherlock Holmes had a superior ability to perceive and
“connect the dots” than his peers—but that the perceptions Holmes made were out in
the open, for all to see. Deduction plays “an indispensable, though limited, part in both
scientific thinking and everyday life.”8

In almost every scientific investigation, deductive reasoning is used to set up the basic hypothesis
to be investigated and to extract the implications of the findings. When we perform analytical and
computational acts such as those involved in working out a family budget or filling out an income-
tax report, we are using deductive processes. When we play bridge, chess, or even ticktacktoe,
we reason deductively. When we are jurors or plaintiffs or defendants, we make deductive use of
both the law and the evidence in judging, complaining, or defending. When we talk or plead or
argue with each other, or grapple with the everyday problems in our lives, much of our reasoning
consists of inferences which, valid or invalid, are deductive—that is, logically inferred from what
9
we already know, or think we know.

While Morton Hunt, quoted above, asserts that in many real world situations we reason
deductively, he doesn’t say we do so in a sound manner. We often reason deductively
without realizing it, or realizing the pitfalls of unsound deductive thought. Note his
reference to much of our reasoning consisting of inferences which are “valid or invalid.”
Sound deductive reasoning, by necessity, deals in certainty. False premises lead to
unsound deductive reasoning: away from what is certain, and toward what is probable.
Probability lays within the much wider realm of induction.

Induction. While deduction is the process of extracting a conclusion already extant

within its premises, induction is the process of inferring a conclusion that, at best, is
probable. And the probability of the conclusion’s truth varies directly with the degree of
likelihood that its premises are true.

Here is an example: I once asked my then-four year old grandson what color (sweet)
corn is. His answer: “yellow.” He knows nothing of the concept of deduction, but based
upon his few years on earth, his subconscious and instantaneous “deduction” probably
went like this: “Every cob of corn I’ve ever seen is yellow…so all corn is yellow.” Of

3
course, this is a false premise—there is white corn, blue corn, Indian corn, etc. But he
doesn’t know that yet—he was only four years old. His exposure to the variety of corn,
to date, had been limited. (An office colleague has had much the same reaction with his
own children, although in a counterintuitive sense. In their happy household, “apples
are only green…while pears are only red.”)

My grandson’s mistake—the false premise of all corn being yellow—is common. We

often “deduce” in spite of limited exposure and experience, not realizing that unseen,
unconsidered alternatives exist. We should say, instead, that some, a lot, or most corn
is yellow. There is, therefore, a certain percentage of corn that is one color or another;
the act of assessing the conclusion is whether or not the stated percentage is accurate.

My grandson’s conclusion was an example of inductive reasoning. Rather than

asserting certainty, an inductive conclusion maintains that something, at best, is
probable. Here are two examples:

 Premise: Some dogs have fleas

 Premise: Bowser is a dog
 Conclusion: [There is a probability that] Bowser has fleas

 Premise: 98% of snails are slimy

 Premise: There’s a snail in my garden
 Conclusion: [There is a high probability] that the snail in my garden is slimy

Notice that in both examples, the premises do not say “all” dogs have fleas, or “all”
snails are slimy. Anything less than 100% precludes absolute certainty.

At this point, you may feel confused: you’ve been told previously that while deduction
moves from the general to the specific, induction moves in the opposite direction, from a
specific observation to a generalization. Both are broad misconceptions. While
inductive logic does include the process of making generalizations, there is more
involved. Perhaps a better working definition for induction by Jamie Watson and Robert
Arp is the following: “In an inductive argument, a conclusion follows from a set of
premises with some degree of likelihood or probability so that, if the degree is high
enough and if all of the premises are true, then the conclusion is likely or probably is
true.”10 Philip Johnson-Laird defines it as “The process of deriving plausible
conclusions that go beyond the information in the premises.”11 Again, induction deals
with probability; deduction with certainty. As such, an inductive argument could have a
false conclusion even though its premises are true. With induction, there is no
guarantee of certainty.

4
Based upon the degree of likelihood contained within the premises, inductive arguments
are assessed as strong or weak. Watson and Arp state, “If the [inductive] arguer
achieves her goal and the conclusion follows from the premises with a high degree of
likelihood, the argument is strong. On the other hand, if the conclusion does not follow
with a high degree of probability, the argument is weak.”12 Inductive strength or
weakness is, of course, a matter of degree. Assessing the strength or weakness of
inductive arguments includes making assessments of the population and sample used.
A population is an entire group of some type of object observed: all dogs, or all snails.
The sample is a subset of the population examined: some dogs, or 98% of snails.
Some strength/weakness guidelines from Gary Jason’s Critical Thinking are as follows:

 The larger the sample, the more representative it is [of the population], so the
stronger the inference is.
 The more the sample matches the population in relevant respects, the more
representative it is, so the stronger the inference is.
 The more random the sample, the more representative it is, so the stronger the
inference is.13 (Bold font and italics added.)

Finally, similar to refuting deductive conclusions, counterexamples can refute inductive

conclusions as well.

Forms of Induction. Inductive reasoning, absent the certainty of deduction, can lead
to horrendous mistakes. On March 6, 1987, the skipper of a car ferry in Zeebrugge,
Belgium made a plausible induction that his crew had closed the ferry doors—they had
always done so in the past. The chief officer and the bosun made the same mistake.
But the assistant bosun, responsible for closing the doors, was fast asleep. The car
ferry capsized, drowning 188 people.14 This is a form of induction, wherein the skipper
generalized based upon patterns and pattern matching. “The basis of induction is
repeated observation, so that the facts about similar experiences accumulate to the
point where one sees a repetitive pattern and can draw a conclusion about it.”15 The
skipper of the ferry had gotten underway numerous times, and each time the assistant
bosun had closed the bow doors. Who in their right mind wouldn’t have closed those
doors, knowing disaster would otherwise follow? This instance shows the inherent
danger in inductive reasoning: just because something has always happened in the
past, it is not guaranteed to happen in the future. It’s all about probability.

In addition to patterns and pattern matching, other forms of induction include the
following:

5
  Inductive generalization: reasoning from a sample to population. (“All corn is
yellow...”)
 Inductive instantiation: reasoning from a population to a sample. (“Some dogs
have fleas, so Bowser…”)
 Analogies and metaphors—reasoning from one sample to another sample
 Testimony—accepting a conclusion based on what people report observing
 Causal inference—reasoning about perceived cause and effect16
 Abduction—inferring the reason “why” something occurred

Analogies and metaphors are forms of inductive reasoning. Analogies are a

comparison of two things, one of which is better understood than the other. Then,
based on the similarities (or dissimilarities, as the case may be), new information is
inferred about the lesser known object. Inductive arguments based upon analogy tend
to be weak, since there are often more dissimilarities than similarities; we often tend to
focus on the similarities, which are the first things that come to mind. That being said, if
properly analyzed, analogy can make for a strong inductive argument, if the similarities
are relevant to the inferred conclusion and outweigh the dissimilarities.17 Analogies
tend to be expressed explicitly as similes (using like or as in making the comparison),
and implicitly as metaphors.18

Testimony (such as that from an “expert witness” in a court of law) is an example of

inductive reasoning based upon selective observation. Whether we choose to believe
testimony should depend upon a number of things—sample and population size,
expertise and credentials of the observer, test parameters, interests, values and biases
of the observer, what the observer may be omitting, etc. Causal inference—reasoning
about cause and effect—is also highly subject to misinterpretation. In both of these
types of inductive reasoning, it is prudent to be cautious about what one may
reasonably conclude.

Philip Johnson-Laird defines abduction as “an inductive process yielding an

explanation of an event or set of events.”19 Charles Sanders Peirce, an American
logician and semiotician, coined the term “to refer to the inferential process that yields
causal explanations,” since explanation strengthens our belief in inductive reasoning.20
When we infer a conclusion by induction, then, we are abducing if we extend the
inference to include the reasons, or motives, of why something happened. This is tricky
business, and prudence suggests that we should be careful as to how much of an
explanation we abduce. While abductions may in fact strengthen our own beliefs, they
may just as well lead us down a primrose path. For example, remember back to the
opening story. Our harried airport employee induced that the woman he noticed was
pregnant. He could easily have abduced that she was pregnant because she appeared

6
to be 35 years old, and that “the biological clock is ticking.” His abduction imputes
motive, whether or not that motive exists.

In spite of the uncertainties and mistakes associated with inductive reasoning, however,
it has helped mankind by extending its knowledge.”21 By extrapolation, we are able to
gain insight beyond what we previously knew. Using this process, humans have made
valuable discoveries. Sir Alexander Fleming, a Scottish biologist, once noticed that
while some of his laboratory cultures of staphylococci had been infected and killed by a
fungus, other cultures were not. Without intending to do so, Fleming induced that by
nurturing cultures of this mold he could kill staphylococci (and other bacteria). He
eventually called the product of this mold “penicillin.”22 “The inductive mind, shaping
experience into a likeness of the world, fumblingly but regularly arrives at good enough
approximations of reality to ensure humankind’s survival.” 23

So What? Why is this relevant to red teamers?

Think back to the story of the “pregnant lady” in the Introduction. Was that a
deduction? An induction? Or did that thought even occur to the fool? He thought that
because he’d seen similar-looking females in the past, all females who looked like that
were with child. But did he consciously consider whether this specific female fit the
population? Probably not. Instead, he jumped to a subconscious “deductive”
conclusion of certainty based upon his patterns of experience. The severity of his faux
pas only smacked him when he was forced to admit “Well, I assumed…,” yet he never
intentionally set out to assume anything. Instead of deducing, he induced—which
brings no guarantee of certainty. And it all happened in the blink of an eye.

The ideas in this paper are relevant because you’ve been deducing and inducing your
entire lives, but without knowing the finer details and the associated pitfalls. For
example, when you deduce, do you ascertain whether the deduction was sound—valid
with true premises? In other words, do you ask yourself whether the premises leading
to the conclusion were facts, as opposed to assumptions masquerading as facts? If
not, you may have jumped to a conclusion you later realized was faulty.

If, instead, you induce a conclusion, do you have any idea that your use of pattern
matching, metaphor, analogy, testimony, or cause impacts upon the validity of the
induction? Does the use of metaphor or analogy cause you to self-reflect? Do you
even know what any of these related concepts are? Are you instinctively looking for
counterexamples as a way to disconfirm your inferred conclusions? Do you consciously
assess your inductive inferences as strong or weak? And if you abduce an explanation
as to why something occurred, do you keep your abductive inference to the simplest
possible explanation?

7
Further, do you tend to consciously consider the impact that mental models, frames,
and intuition have in the process of inferring?24 And whether you’ve induced or
deduced, were you inclined to immediately think of counterexamples to refute your
conclusion? Did you attempt to disconfirm before moving on?

Armed with the details provided above, you should begin increasing the process self-
reflection, in terms of assessing whether your inferences are justifiable. Chances are,
you did not realize most of this. We deduce and induce, subconsciously, in the blink of
an eye, without consciously considering the merit of our conclusions.

Finally—and just as importantly—as you increase the quantity and quality of your own
self-reflection, also realize that those for whom you red team may be equally unaware of
the kinds of ideas discussed above…and just as likely fall prey to the pitfalls of
deductive and inductive reasoning. It is for this reason, partly, that red teaming exists.

Conclusion. Hopefully, the ideas contained within this paper will help set you on your
way toward improving the manner in which you infer conclusions. We certainly hope so.
Remember: deduction consists of taking two or more premises and extracting a
conclusion which already exists within those premises. Induction, on the other hand, is
the process of inferring a conclusion that, at best, is probable. The probability of that
conclusion’s truth will directly vary with the likelihood that its premises are true.

The intent is not to turn you into formal logicians, although you will have to do a bit of
that in your red team role. Instead, it’s to awaken you to the fact that you—and “they,”
the staff with whom you will be working—are oftentimes victims of subconscious
deduction and induction, absent self-reflection. Forewarned is forearmed.

And if, in the future, you see a beautiful 35-year old woman at the airport with a
moderate bulge in her stomach, be careful what you infer. Instead, just smile.

And help her with her luggage.

1
Gary Jason, Critical Thinking, p. 231.
2
Sylvan Barnet and Hugo Bedau, Critical Thinking, Reading and Writing, pp. 337-338
3
Gary Jason, p. 71.
4
http://records.viu.ca/~johnstoi/arguments/argument5.htm
5
http://www.butte.edu/~wmwu/iLogic/1.3/iLogic_1_3.html See section 1.3.4.
6
The discussion of deduction within this paper is basic. Entire textbooks have been written about the subject;
entire semesters are required to educate logicians in those texts. Our discussion of deduction is kept to the
simplest terms needed for red teaming, and does not consider truth tables, symbolic logic, Venn diagrams, modus
ponens, modus tollens, affirmation or denial of either the antecedent or the consequent, all of the logic fallacies
related to those topics, etc.
7
Barnet and Bedau, p. 84

8
8
Morton Hunt, The Universe Within, p. 132.
9
Hunt, p. 132.
10
Jamie Carlin Watson and Robert Arp, Critical Thinking, p. 35.
11
Johnson-Laird, p. 427.
12
Watson and Arp, p. 35.
13
Jason, pp. 254-259.
14
Johnson-Laird, p. 176.
15
http://records.viu.ca/~johnstoi/arguments/argument5.htm
16
Jason, pp. 78-80.
17
Watson and Arp, pp. 200-204
18
Jason, pp. 111 and 278.
19
Johnson-Laird, p. 425.
20
Ibid, p. 186.
21
Ibid, p. 4.
22
Ibid, p. 178.
23
Hunt, pp. 157 and 163.
24
Johnson-Laird, pp. 427-428. Mental Model: A representation of the world that is postulated to underlie human
reasoning; a model represents what is true in one possibility, and so far as possible has an iconic structure. Mental
models are the end result of perception and of understanding a description. Those of complex systems are a form
of knowledge-representation in long-term memory. Intuition: An unconscious heuristic process based on
unconscious premises that yields a conscious conclusion.