Introduction to Research

K P Mohanan and Tara Mohanan

1 April 2021

CHAPTER 10
TYPES OF REASONING

10.1 Looking Back and Looking Forward

10.2 Inductive Reasoning
10.2.1 Sources of Knowledge: Experience and Testimonies
10.2.2 Examples of Deductive and Inductive Reasoning
10.2.3 A Rule of Inference in Inductive Logic
10.2.4 The Issue of Representativeness
10.3 Abductive Reasoning
10.3.1 Deductive vs. Abductive Reasoning
10.3.2 Defeasibility in Abductive Reasoning
10.3.3 A Rule of Inference in Abductive Logic
10.3.4 Acknowledging Fallibility
10.4 Wrapping up
The Research Gym

10.1 Looking Back and Looking Forward

The previous chapters have pointed out how:
~ reasoning is central to just about every component of research in all domains, and
~ the concept of validity underlies the distinction between sound and flawed reasoning.
In Chapter 8, we explored the concepts of conceptual and formal validity, and in
Chapter 9, we followed it up by examining the conceptual and formal elements of
predicate calculus, including rules of inference and formulae; propositions,
predicates, and arguments; implication, negation, conjunction and disjunction.
We then defined validity of reasoning in a system of formal logic in terms of the steps
that are sanctioned by that system.

www.ThinQ.education
In Chapter 9, we also showed how predicate
calculus is an extension of propositional The concept of predicate in
calculus. Propositional calculus takes the formal logic translates as the
concept of proposition as an atomic symbol. concept of function in
Predicate calculus decomposes that concept mathematics and computer
into the components of predicate and its science, where it expresses
arguments (represented as F (x,y...)), and the attributes, relations, and
quantifiers of the argument variables, processes.
∀ (for all x) and ∃ (there is some x…).

A Note on Two Lineages of Formal Logics

If we examine the evolution of Formal Logic as a branch of academic knowledge,
we find at least two lineages. One of these is the search for what we might call the
foundations project in mathematics, pursued by both mathematicians and
philosophers of mathematics. In this lineage, predicate logic is called First Order
Logic, and is seen as the foundation that unifies all research in mathematics into
what is shared by logic and set theory. This is the only logic that this lineage
acknowledges — the logic that supplies the concepts and formalisms for the
expression of axioms, definitions, and proofs in mathematics.
In another lineage, First Order Logic is not the only formal logic, it is just one of
several formal logics. In this lineage, logic needs to function as the foundation not
only for mathematics, but also for the other domains in academic knowledge. It is
not sufficient for formal logic to tell us how to check the validity of reasoning in
mathematical proofs. It has to also tell us how to check the validity in the
reasoning from sample to population; in deriving predictions of scientific theories;
in arguments that support or refute scientific and philosophical theories; and also,
for instance, in the reasoning that medical doctors engage in when diagnosing an
illness on the basis of its symptoms; in the reasoning that lawyers in the
courtroom use to argue in support of or against an accused; and the reasoning that
CEOs and ministers use when thinking through problems of policies and products.

In Chapter 10, we will explore the types of reasoning appropriate for the different
functions of reasoning in different components of research, making these points:
A. We need three broad types of reasoning (and hence three systems of formal logic),
with further sub-cateogorisations within each. The three types of reasoning are:
1. Deductive Reasoning: to derive logical consequences (theorems/predictions)
from the premises of a theory.
2. Inductive Reasoning: to derive observational generalisations on a population,
based on observational reports (data points) on a
sample from that population.
3. Abductive Reasoning: to justify the postulation of theoretical statements and
theoretical concepts.
B. Given the concept of validity outlined in Chapter 8 in terms of the rules of inference
in a given system of formal reasoning, what is valid in one form of reasoning need
not be valid in another.

2
10.2 Inductive Reasoning
10.2.1 Sources of Knowledge: Experience and Testimonies
In Chapter 8, we made a distinction between learning from experience and learning
from the statements formulated by ourselves or others. So our knowledge about the
sourness of lemons comes from experience; most of us learn this by tasting a slice of
lemon or lemon juice. In contrast, our knowledge that a cobra bite can be fatal
comes from what others tell us; we do not have the experience of being bitten by a
cobra and dying as a result.
This distinction between two sources of learning underlies two broad forms of
reasoning: one form where it is based on inferences from our experience, and the
other form that ultimately rests on propositions expressed in sentences of a natural
language or formal language.
When we have to justify mathematical conjecturess (e.g., the conjecture which,
when proved, is known as the ‘angle
sum theorem’), we begin with This form of reasoning is also used in
premises that we postulate as our scientific and philosophical theories to
starting point. We then show that derive the logical consequences of
the statements follow logically from theoretical premises.
those premises. The form of reasoning In science, these consequences are
used for such derivations is deductive called predictions.
reasoning.
In mathematics, theories, unlike conjectures, are not expected to be justified.
However, scientific, philosophical, ethical and aesthetic theories do have to be
justified. When we are asked to justify such theories, we draw upon and articulate
our experience in the form of sentences of a language. (e.g., “I ate a slice of lemon
yesterday, and it was very sour,” or “I saw a boy bullying a younger child yesterday;
what he was doing was bad.”) In such cases, the justification takes the form of non-
deductive reasoning.
Non-deductive forms of reasoning fall into two broad categories: inductive
reasoning and abductive reasoning. Let us take a look at how inductive reasoning
differs from deductive reasoning, before moving on to abductive reasoning.

10.2.2 Examples of Deductive and Inductive Reasoning

If someone were to tell us that Zeno has a beak, we would conclude that Zeno is a
bird. If someone were to tell us that Xetus has four legs, we would conclude that
Xetus is not a bird. What is the rational justification for these conclusions?
We may articulate the justification as follows:
Example 1
All living organisms that have beaks are birds. (We know this.)
Zeno has a beak. (We are told this.)
Hence, it is reasonable to conclude that Zeno is a bird.
Example 2
No bird has four legs. (We know this.)
(= All living four-legged organisms are non-birds.)
Xetus has four legs. (We are told this.)
Hence, it is reasonable to conclude that Xetus is not a bird.

3
The propositions that all creatures that have beaks are birds, and no bird has four
legs, are premises that we are presenting in support of our conclusions, as part of
our rational justification. But suppose someone were to ask: “Why do you believe
that creatures that have beaks are birds?” or: “Why do you believe that no bird has
four legs?” What is the rational justification for these beliefs?
Our responses to the skeptic would be as follows:
Example 3
We have examined a large sample of birds.
We have noted that every bird in our sample has a beak.
Hence, until we find evidence to the contrary,
it is reasonable for us to conclude that all birds have beaks.
Example 4
We have examined a large sample of birds.
We have noted that not a single bird in our sample has four legs.
Hence, until we find evidence to the contrary,
it is reasonable to conclude that no bird has four legs.
In sum, Ex. 1 and 2 illustrate what is called deductive reasoning; and Ex 3 and 4
illustrate what is called inductive reasoning.
A cautionary note for those who
To get an initial handle on the two types of
have already come across the
reasoning, notice that in Ex. 1 and 2, the first terms ‘deduction’ and ‘induction’
premise is a generalisation on a in textbooks and popular
population (the population of living treatments: these terms have
organisms), and the second is statement on a been associated with various
particular sample of the population, in this meanings in published
case a sample of one, namely, Zeno in Ex. 1, literature. This could be
and Xetus in Ex. 2. The conclusion is on that confusing, so it would be useful
sample. to go by what concepts we
assign to these terms in the
In contrast, the premises in Ex. 3 and 4 context of this course, and not
assert a pattern in a particular sample of a bring to them your prior
population. The conclusion is a general one preconceptions.
on the population as a whole.
This gives us a glimpse of the distinction between deductive and inductive
reasoning, but we are not ready to define them yet. Let us begin by noting that:
~ propositional calculus and predicate calculus are both formalisations of
deductive reasoning; and hence
~ all the examples of reasoning given in Chapter 9 are examples of deductive
reasoning.
Some of the examples below are repeated from Chapter 9:
Example 5
If Zeno is a lion, then Plato ate peanuts yesterday.
Zeno is a lion.
Therefore, Plato ate peanuts yesterday.
Example 6
Athena is taller than Zeno.
Zeno is taller than Apollo,
Therefore, Athena is taller than Apollo.

4
Example 7
Leda’s room is a rectangle.
It is 15 ft long.
It is 10 ft wide.
The area of a rectangle is the product of its length and breadth.
The product of 15 and 10 is 150.
Therefore, the area of Leda’s room is 150 sq ft.

In mathematical inquiry, the form of reasoning used in making calculations and

proving theorems is deductive reasoning. Mathematical inquiry does not permit
any other form of reasoning.
In scientific inquiry, the form of reasoning used to derive the logical consequences
(predictions) is also deductive reasoning. However, in observational research, the
form of reasoning used to arrive at observational generalizations from the sample
of data points is inductive reasoning.
Given that the sample-to-population reasoning in Ex. 3 and 4 comes under
inductive reasoning, and all the other examples given above come under deductive
reasoning, perhaps it would make sense at this stage to say that:
Deductive reasoning is the form of reasoning used to derive the logical
consequences (theorems, predictions) of the premises of a theory, and is
formalised as predicate calculus.
Inductive reasoning is the form of reasoning used in observational research
(in science) to arrive at observational generalisations on a population
from a pattern in a sample of the population.
Given these definitions, we may view inductive reasoning as reasoning from sample
to population. In the diagram below, the larger ellipse represents the population,
and the smaller ellipse inside it represents the sample:

How about reasoning in the other direction? We may view deductive reasoning as
reasoning from the population to a sample:

10.2.3 A Rule of Inference in Inductive Logic

We are now ready to characterise the essential difference between deductive
reasoning and inductive reasoning in terms of their rules of inference. Let us go
back to the rule of Modus Ponens discussed in Chapter 8:
Rule of Inference in Deductive Logic: Modus Ponens
Given premises of the form If P then Q, (Premise 1)
and P (Premise 2)
it is legitimate to conclude that Q (Conclusion)

5
Clearly, the reasoning in Ex. 3 and 4 does not derive its validity from this rule.
Instead, we formulate the rule of inference for these examples as follows:

Rule of Inference in Inductive Logic

Given premises of the form Pattern Pn is true of
sample S of population P (Premise 1)
and S is representative of P (Premise 2)
until we have evidence to the contrary,
it is legitimate to conclude that Pn is true of P (Conclusion)

EXERCISE 1
Scenario Question
Simi is an adult human being. Someone asks:
“How many hearts does Simi have?” You say, For each claim in italics:
“Exactly one.” What is the rational justification
“Every adult human has exactly one heart.” for the position taken?

Synergy Stadium is circular in shape, and its Did you use deductive reasoning
diameter is 1000 meters. or inductive reasoning for the
What is its circumference? justification?
“Approximately 3141 meters.”

10.2.4 The Issue of Representativeness

Suppose you take a random sample of a hundred adult humans. And you find the
following patterns in the sample:
Pattern 1: Everyone in the sample has exactly one heart.
Pattern 2: Everyone in the sample has the heart slightly to the left of center.
Pattern 3: Everyone in the sample has exactly ten fingers.
Pattern 4: Everyone in the sample has dark skin.
Pattern 5: No one in the sample is more than 250cm (8.2 ft) tall.
Pattern 6: No one in the sample is less than 100cm (3.3 ft) tall.
How many of these patterns would you be willing to take as an observational
generalisation on the adult human population?
Chances are that you would be more than willing to say that Pattern 1 is true of the
adult human population. Would you be equally sure about the other patterns? Let us
take each of them in turn.
There is a congenital condition called Situs Inverses in which the positions of the
major internal organs in the human body are mirrored from their normal positions
(https://en.wikipedia.org/wiki/Situs_inversus), so that the heart is on the right side, and
the liver on the left. The condition is found in about 0.01% of the human population.
So, if you took a random sample of even a thousand adult humans, you may not find
even one person with the heart on the right. If you are aware of this phenomenon, you
would reject Pattern 2 as a generalisation on the human population.

6
Likewise with Pattern 3. It is likely that you have come across someone with eleven
fingers (an extra digit on one hand), or even twelve fingeres (an extra digit on each
hand), in which case you would not generalise Pattern 3 either.
Based on your prior experience, and even systematic observation, you would reject
Pattern 4 as well. As for Patterns 5 and 6, you might be doubtful, because even
though you may not have seen anyone more than 250cm or less than 100cm in height,
you may not want to rule out the possibility of such people somewhere in the world
outside your experience.
In each of the cases where you either reject the pattern as a generalisation, or are
doubtful about it, the reason is that the sample is not representative of the
population. But how do you know that the stated patterns are not representative of
the population of adult humans? The answer would be: because you already know
that there are human beings who have more than ten fingers, or have skin that is not
dark; and there may be adult human beings who are more than 250cm or less than
100cm in height. The relevant concept here is that of representativeness.

10.3. Abductive Reasoning

10.3.1 Deductive vs. Abductive Reasoning
Consider the following scenario:
Zeno tells his doctor about what happened the day before. He experienced a sudden
pain in his chest that lasted more than a few minutes. He also felt pain along his
left arm, shortness of breath, and nausea. He also broke out in a cold sweat.
The surgeon concludes that Zeno must have had a heart attack.
This is an example of reasoning from a set of symptoms (the patient’s report) to a CAUSE,
where the symptoms are the EFFECT. This form of reasoning is called abduction.
To understand the nature of abduction, let us compare reasoning from cause to effect
with reasoning from effect to cause:
Example 8 [Reasoning from cause to effect]
A heart attack tends to cause
a sudden pain in the chest that lasts more than a few minutes,
pain along the left arm, shortness of breath, cold sweat, and nausea.
Zeno had a heart attack.
Therefore, it is reasonable to conclude that Zeno must have experienced
a sudden pain in the chest that lasted more than a few minutes,
pain along his left arm, shortness of breath, cold sweat, and nausea.

Example 9 [Reasoning from effect to cause]

A heart attack tends to cause
a sudden pain in the chest that lasts more than a few minutes,
pain along the left arm, shortness of breath, cold sweat, and nausea.
Zeno experienced
a sudden pain in the chest that lasted more than a few minutes,
pain along his left arm, shortness of breath, cold sweat, and nausea.
Therefore, until we find evidence to the contrary or an alternative explanation
for what he experienced,
it is reasonable to conclude that Zeno must have had a heart attack.

7
Ex. 8 illustrates deductive reasoning; and Ex. 9 illustrates abductive reasoning. Here is
another pair of examples:
Example 10 [Reasoning from cause to effect]
Rain causes streets to be wet.
It has just rained.
Therefore, it is reasonable to conclude that the streets must be wet.

Example 11 [Reasoning from effect to cause]

Rain causes streets to be wet.
The streets are wet.
Therefore, until we find evidence to the contrary or an alternative explanation
for the streets being wet,
it is reasonable to conclude that it must have rained.
Ex. 10 illustrates deductive reasoning; and Ex. 11 illustrates abductive reasoning.

10.3.2 Defeasibility in Abductive Reasoning

Compare our earlier examples of abductive reasoning with the one below:
Example 12
When Leda scolds Zeno, Zeno cries.
Zeno is crying now.
Therefore, Leda must have scolded Zeno.
Would you consider this to be an example of valid reasoning? Most likely not. Why not?
Because there could be a number of other causes for Zeno’s crying. Maybe Zeno is crying
because he lost his pet cat; or he didn’t do well in his class test; or his friends didn’t
include him in their game. Unless we rule out these alternative causes, it doesn’t
follow that Leda must have scolded Zeno.
But then, notice that consideration of alternative causes applies to the other examples of
abduction as well, including those that we have accepted as valid.
We accepted the conclusion that it must have rained because we couldn’t think of a
better or equally good explanation. Suppose someone proposes the following causal
explanation for the streets being wet.
When a water truck goes down the street, if the truck is leaking, the street gets wet.
(i.e., A leaking water truck going down the street causes the street to be wet.)
The street is wet.
Therefore, a leaking water truck must have gone down the street.
We now have a competing explanation for Ex. 10. Our earlier conclusion no longer holds.
We must therefore look for further evidence to choose between the two explanations.
Suppose we examine the sidewalk and the lawns near by, and find that they are not
wet. The water truck hypothesis explains why only the street is wet. The rain
hypothesis incorrectly predicts that the sidewalk and the lawns also would be wet.
Hence, we choose the water truck hypothesis. If, on the other hand, we find that the
sidewalk and the lawns and the terraces of houses are also wet, we choose the rain
hypothesis.

8
EXERCISE 2
Suppose you found that the street as well as the lawns are wet, but not the sidewalk.
What would be your causal explanation? We leave that for you to gnaw on.

The above discussion points to an important feature of abductive reasoning that

logicians call defeasibility. It means that a conclusion that we initially judged to be
derived through valid reasoning may turn out to be invalid when either of the following
come to light:
(a) alternative explanations (e.g., a leaking water truck as an alternative to rain);
or
(b) contrary evidence (e.g., the side walks and the lawns not being wet.)

10.3.3 A Rule of Inference in Abductive Logic

Notice that the causal generalisation below is shared by Ex. 8 and 9:
A heart attack tends to cause
a sudden pain in the chest that lasts more than a few minutes,
pain along the left arm, shortness of breath, cold sweat, and nausea.
And the causal generalisation below is common to Ex. 10 and 11:
Rain causes streets to be wet.
The difference in the modes of reasoning between these two pairs lies in the
direction of inference. To see this, let us restate Modus Ponens as follows. Here, the
rule of inference is within a non-causal logic:
Modus Ponens in Non-Causal Deductive Logic:
Given premises of the form P implies Q (Premise 1)
and P is true (Premise 2)
it is legitimate to conclude that Q is true. (Conclusion)

We can now state the rule of Causal Deductive Logic as:

Modus Ponens in Causal Deductive Logic:
Given premises of the form X causes Y (Premise 1)
and X exists (Premise 2)
it is legitimate to conclude that Y exists. (Conclusion)

Given this rule of inference, we may state Abduction in Causal Reasoning as:
Rule of Inference in Causal Abductive Logic
Given premises of the form X causes Y (Premise 1)
and Y exists (Premise 2)
in the absence of
(a) evidence to the contrary, and
(b) an equally good alternative cause,
it is legitimate to conclude that X exists. (Conclusion)

Let us borrow the arrow notation from Propositional Calculus and adapt it to
represent the cause-effect relation in Causal Calculus as:

X: CAUSER variable
Y: EFFECT variable
Curved arrow: CAUSE

9
Given this notation in which the curved arrow in causal reasoning is the analogue of
the straight arrow in correlational reasoning, we may represent the Modus Ponens of
Deductive Causal Reasoning and Rule of Inference in Causal Abduction as follows:
Modus Ponens in Causal Deductive Logic:

Given premises of the form (Premise 1)

and X (Premise 2)
it is legitimate to conclude that Y (Conclusion)

Given this rule, it becomes obvious that Abduction in Causal Reasoning is:
Rule of Inference in Causal Abductive Logic:

Given premises of the form (Premise 1)

and Y (Premise 2)
in the absence of (a) evidence to the contrary, and
(b) an equally good alternative cause,
it is legitimate to conclude that Y (Conclusion)

This means that whether or not a given example of reasoning is valid depends on the
logic within which the reasoning is based:
Inductive Logic OR
What is valid in need not be valid in First Order Logic.
Abductive Causal Logic
Let us take a step further.
The only logic permitted in the derivation of theorems from the theoretical premises
in Mathematics is that of First Order Logic. The logic for sample to population
reasoning and causal reasoning in science call for Inductive Logic and Abductive
Causal Logic. Hence, what is valid in scientific reasoning need not be valid in
mathematical reasoning.
Any interaction between mathematicians and scientists needs to have as its basis a deep
awareness of what is stated in italics in the above paragraph.

10.3.4 Acknowledging Fallibility

Complete certainty of a conclusion is impossible when we arrive at it through indictive
reasoning, or through abductive reasoning where we infer the cause from the effect.
You would have noticed that in the examples of inductive reasoning, for instance, Ex. 3
and 4, and those of abductive reasoning, such as Ex. 9 and 11, the conclusion includes
the words, until we find evidence to the contrary. This is important to say, because it
signals the fallibility of the conclusion. It is an expression of the tentativeness of the
conclusion, and a willingness to correct it if we find new evidence that shows it to be
wrong.

10
10.4 Wrapping Up
In Chapter 9, we went through the kind of reasoning used for mathematical proofs,
called classical deductive reasoning. Its formalisation, evolved first as propositional
calculus (also called Boolean Algebra), and then enhanced as predicate calculus, is
called First Order Logic.
In this chapter, we expanded our scope of reasoning (and logic) beyond First Order Logic
and mathematics, by considering
Inductive reasoning: the kind of reasoning needed for arriving at observational
generalisations on a population from a sample of observational reports
(called data points);
and
Abductive reasoning: the kind of reasoning needed for arriving at explanations
for observations or observational generalisations.
We also took a brief look at defeasible reasoning.
Classical Deductive reasoning is non-probabilistic and non-defeasible. We discuss
probabilistic deduction and defeasible deduction in the Appendix to this chapter, along
with their equivalents in inductive and abductive resasoning. The Appendix also
discusses two kinds of abduction, one of them being central to rational justification of
scientific theories.
Being familiar with these modes of reasoning helps us become better all-round
researchers. But just knowing these terminologies is not enough. Not even a deep
understanding of these modes of reasoning is enough. To benefit from that
understanding, and to develop core research abilities across all domains of inquiry, it is
important to practice these forms of reasoning over an extended period of time. While
engaging in such practice, it would also help to recognise these diverse forms of
reasoning when you find them in the research papers and other articles that you happen
to read, identify them, and analyse them, such that paying attention to reasoning
becomes a habit of mind.

11
THE RESEARCH GYM (GYM 10)

SCENARIO
Zeno decides to make some spicy sauce for breakfast. He slices and fries all the
ingredients and puts them in the blender. He turns on the blender switch. The
blender doesn’t turn on. Zeno wonders why.
His first inference is that the machine is not plugged in. He checks, and finds it
properly plugged in.
He remembers that the blender turns itself off when overheated. So his next
inference is that the blender must have turned itself off because of the hot
ingredients. He presses the reset button, expecting the blender to start, but nothing
happens. So this inference has to be abandoned as well.
His third inference is that perhaps the microwave has been turned on when he
turned on the blender, and the overload had blown a fuse. Zeno checks, and finds
the microwave off, and nothing wrong with any of the switches on the switch board.
So this inference is also rejected.
The only remaining possibility is that there is no power supply, and that the
computer and the lights are on because of power from the inverter. Zeno goes back
to his work at the computer.
A few minutes later, he hears the blender whirring. He infers that the electricity is
back, and Athena must be in the kitchen. She must have turned on the blender.
But he finds that Athena is still pottering around in the garden.
The only inference left is that Zeno left the blender switch on by mistake, and when
the electricity came back, the blender started on its own. He finds that this is
indeed the case.

TASK
Provide an analysis of each of the instances of reasoning that Zeno goes through, by
explicitly articulating the premises and conclusion, as well as the intermediate
steps of derivation if any, and decide if it is an example of deduction, induction, or
abduction.

12