[Link], DEPT.

OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND

ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
24CSEN3321: Critical Thinking
MODULE 5: OTHER FORMS OF ARGUMENTATION: INDUCTIVE AND
STATISTICAL
Expressiveness of propositional logic. Quantificational logic. Deductions vs.
Induction. Inductive, plausible, Probabilistic and statistical arguments. Inductive
generalisations. Statistical generalisations. Causal and Analogical arguments.
Fallacies in informal arguments.
Course Educational Objectives:
• To expose the students to advanced argumentation and enable them to demonstrate the
capacity to construct compelling arguments and engage with complex texts.
Course Outcomes:
After completion of this module, the student will be able to:
Identify and critique fallacious reasoning, employing critical thinking skills to evaluate the
strength and validity of arguments.

This material was prepared using the textbooks and resources listed below
1. Van Cleave M. J.: Introduction to Logic and Critical Thinking (Version 2.0), Creative
Commons. Open Textbook Library 2016.
[Link]
2. Copi I. M., Carl Cohen C, Rodych V: Introduction to Logic (15th Edition), Routledge Taylor
and Francis 2016
3. Prof.S-Arun-Kumar, “Critical thinking for engineers and scientists,” 2025.
[Online; accessed 01-December-2025, till date]
4. Refresher Course - Presentations prepared by the faculty teaching the Critical Thinking
Course.
* Note: For more examples and explanation, refer to class notes and discussion

Expressiveness of propositional logic

Introduction: Expressiveness
• The ability of a logic system to represent information
• Determines what can be described or reasoned about
• Statements with truth values (True/False)
• Uses logical connectives: AND, OR, NOT, IF-THEN, IF AND ONLY IF

Example
“If the Instagram account is private, then only followers can see the posts.”
P = Account is private
Q = Only followers can see posts
P→Q

1
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
What kind of statements can be modelled?
Propositional logic is good at representing statements that are either true or false as a whole.
What kind of statements cannot be modelled?
Propositional logic cannot talk about objects, properties, or relationships.

What Propositional Logic can Express?

Types Example

Simple Facts I have a book (b)

Compound Statements The system is on, and the file is open

Logical relationships (s^f)

If the server is down, then the website will not load (s -> ¬w)

Truth Functional Reasoning I will drink Tea or Coffee (t V c)

What Propositional Logic cannot Express?

Types Example

Quantification (1) Every student submitted the assignment.

(2) Some students are absent

Needs a quantifier (∀, ∃)

Relations (1) Seetha is taller than Geetha

Tall(Seetha, Geetha)
Needs relations

Objects (1) Every dog has four legs

Needs predicates and quantifiers

Properties (1) Every teacher teaches at least one subject

Not representable as simple true/false atomic propositions.

Exercise
Identify which of the following are Propositional or Predicate

If the sensor detects motion, the alarm sounds

Every employee must update their password

2
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5

Two people are siblings

The network is down AND the server is not responding

• Propositional logic does not allow reasoning about some of the most elementary valid
arguments due to its lack of expressivity.
• Propositional logic does not express the inner logical relationships between
many simple statements
Limitations of Formal Logic
Expressiveness. The fundamental limitation of formal logic is its expressiveness as compared to
natural language.
– nuances of meaning (eg. not bad, )
– contextual dependencies.
– emotions, motives, and intentions
– sub-texts, sarcasm, satire
Limitations of Natural Language
• Does not have a well-defined syntax.
• Does not have a well-defined semantics either.
• Inherently ambiguous.
• Too complicated to analyse, formalise and make rigorous.
• Often too verbose.
• Also has the problem of self-reference.
Self-Reference
Definition: Autological & Heterological
An adjective in a language is autological iff it describes itself or can be applied to itself i.e. the
meaning of the word is also a property of the word. Every adjective in English which is not
autological is said to be heterological.
Example:
• “poly-sylla-bic” is a word consisting of 3 syllables, so it describes itself. Hence it is autological.
• “mono-sylla-bic” consists of more than 1 syllable, so it does not describe itself. Hence it is
heterological.
Examples
Autological Heterological

adjectival adverbial
single multiple
polysyllabic monosyllabic
English French
olde ambiguous

3
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
unambiguous
man-made

• Every natural language has both autological and heterological adjectives.

• Every natural language may be equipped with a (new) adjective for “autological” and a (new)
adjective for “heterological”.
• Every natural language has self-referential assertions whose truth value cannot be established
in any universe of discourse.
Expressiveness of Quantification Logic
Quantification Logic (also called Predicate Logic or First-Order Logic) is far more expressive
than propositional logic.

• Objects
• Properties of Objects
• Relations between Objects
• General Statements (using ∀, ∃)
• Structured information

Objects There is a city that has a metro station.

Objects: cities
Predicate: City(x), HasMetro(x)
∃𝑥(𝐶𝑖𝑡𝑦(𝑥) ∧ 𝐻𝑎𝑠𝑀𝑒𝑡𝑟𝑜(𝑥))

Properties The flower is beautiful

Object: Flower1
Property: Beautiful
Predicate Logic:
𝐵𝑒𝑎𝑢𝑡𝑖𝑓𝑢𝑙(𝐹𝑙𝑜𝑤𝑒𝑟1)

Relations “Bangalore is located in Karnataka.”

Objects :Bangalore , Karnataka
Relation: LocatedIn
Predicate Logic:
𝐿𝑜𝑐𝑎𝑡𝑒𝑑𝐼𝑛(𝐵𝑎𝑛𝑔𝑎𝑙𝑜𝑟𝑒, 𝑘𝑎𝑟𝑛𝑎𝑡𝑎𝑘𝑎)

Quantifiers “Every number greater than 10 is positive”

Object: numbers
Predicate: GreaterThan10(x), Positive(x)
Property : Positive(x)
∀𝑥(𝐺𝑟𝑒𝑎𝑡𝑒𝑟𝑇ℎ𝑎𝑛10(𝑥) → 𝑃𝑜𝑠𝑖𝑡𝑖𝑣𝑒(𝑥))

4
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5

Structures Information Every teacher teaches some subject.

This expresses a structure:
Object 1: Teacher
Object 2: Subject
Relation: Teaches
Quantifiers: ∀, ∃
∀𝑥 (𝑇𝑒𝑎𝑐ℎ𝑒𝑟(𝑥) → ∃𝑦 (𝑆𝑢𝑏𝑗𝑒𝑐𝑡(𝑦) ∧ 𝑇𝑒𝑎𝑐ℎ𝑒𝑠(𝑥, 𝑦)))

Quantification Logic cannot express

Statements About Sets of Sets or First-order logic cannot directly express

Infinite Collections 1) Every subset of a set has a maximum.
2)All sets of numbers are countable.
Because it cannot quantify over sets, only over
individual objects.
Certain Mathematical Concepts “If a set has only a finite number of elements, then it
(like completeness) has a largest element”
• The size of a set (finite vs infinite)
• the concept “maximum element”
But first-order logic cannot talk about finiteness,
because:
• It cannot quantify over sets (only individual
objects).
• It cannot express “finite” or “infinite” as a
property of a set.
• It cannot express “maximum element of a
finite collection.”
• This requires second-order logic, which can
quantify over sets.

Certain Mathematical Concepts (like “If a set has only a finite number of elements,
completeness) then it has a largest element.”
S = a set
≤ = ordering relation on the elements of the
domain
Finite(S) = a second-order property saying S is
finite
∀𝑆 (Finite(𝑆) → ∃𝑚 ∈ 𝑆 ∀𝑥 ∈ 𝑆 (𝑥 ≤ 𝑚))

5
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5

Statements Requiring Finite or Exact First-order logic cannot easily express:

Counting “There are exactly 5 students.”
“There are finitely many primes.”

Statements About Its Own Truth or Quantification logic cannot express:

Structure (Self-reference) “This statement is false.”
“No statement in this system can prove itself.”
This requires meta-logic, not object-level First
order logic

Comparison

Aspect Propositional Logic Quantification Logic

Basic unit Whole statements Objects + relations

Can express relations? No Yes

Can express quantities? No Yes (∀, ∃)

Expressiveness Low High

Applications Circuits, rules, automation AI, databases, semantic reasoning

Key Takeaways
• Propositional Logic expresses simple facts.
• Propositional logic does not allow reasoning about some of the most elementary valid
arguments due to its lack of expressivity.
• Propositional logic does not express the inner logical relationships between many
simple statements
• Propositions have truth values in a universe of discourse.
• Quantification Logic expresses rich, structured, real-world relationships.

Quantification Logic (Predicate Logic)

Terminology
Quantification
A method for describing and symbolizing noncompound statements by reference to their inner
logical structure
Affirmative singular proposition
A proposition in which it is asserted that a particular individual has some specified attribute.
Simple predicate

6
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
In quantification theory, a propositional function having some true and some false substitution
instances, each of which is an affirmative singular proposition.
Predicates are parameterised structures which express properties and relationships.
Universal quantifier

In quantification theory, a symbol, ∀, is used before a propositional function to assert that the
predicate following the symbol is true of everything. Thus, “∀x F(x)” means “Given any x, F is
true of it.
Existential quantifier

In quantification theory, a symbol, ∃, is used before a propositional function to assert that the
function has one or more true substitution instances. Thus “∃” means “there exists an x
such that F is true of it.”
Instantiation
In quantification theory, the process of substituting an individual constant for an individual
variable, thereby converting a propositional function into a proposition.
Generalization
In quantification theory, the process of forming a proposition from a propositional function by
placing a universal quantifier or an existential quantifier before it.

Need for Quantification

Quantification provides a systematic approach for representing noncompound statements and
revealing their underlying logical structure. Traditional analytical methods fall short for
arguments where validity hinges on this internal structure. By applying quantification as a
foundational theory and incorporating additional symbolization, one can elucidate this inner
structure, thereby significantly improving analytical capabilities.
Quantification allows us to view noncompound premises as compound statements, without loss of
meaning. Utilizing that interpretation, we can then apply all fundamental argument forms and
the rule of replacement, as previously demonstrated. Compound statements include drawing
conclusions and establishing validity or invalidity, after which the resultant compound conclusion
can be reverted, through quantification, to its original noncompound form.
Predicates
In the earlier examples studied, all premises and the conclusion are simple statements, rendering
them distinct atomic propositions. The individual atoms lack any mutual connection through truth-
functional connectives, leading to the conclusion that, as propositional arguments, they are
invalid. Despite this, the arguments seem to have some validity.
Components of Objects
Here is another valid argument

7
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
All horses are animals.
Therefore, the head of a horse is the head of an animal.
The relationship between premises and conclusions in logical propositions shows that, although
they are simple statements, they connect through shared properties, such as adjectives (e.g.,
mortal) and collective nouns (e.g., humans, animals). It explains that collective nouns represent
classes of objects, and membership in these classes imparts properties to the objects.
Additionally, it notes that compound objects can also be viewed as collectives, with their
components sharing the property of being members of the larger compound. Finally, it asserts
that adjectives serve to define object properties.
Example:
Simple statements such as
All animals are mortal.
All humans are animals.
Socrates is human.
actually have an inner logical structure in which they share certain common properties, such as
– being animal,
– being human,
– being mortal and
– being a generalisation in the case of the first two statements
– being a particularisation in the case of the last statement.
Properties As (Sub-)Classes
• Every property defines a characteristic class of objects that satisfy the property.
• Every (sub-)class defines a characteristic property defined simply by membership in that (sub-
)class.
• A property may be circumscribed by a class of objects.

Adjectives: Properties of Objects

• Adjectives are used to “describe” nouns, e.g.
God is omniscient.
God is omnipotent.
• Adjectives actually define a property (e.g. omniscient and omnipotent) satisfied by the nouns
(e.g. God) they describe.
Binary Relationships
• Verbs are often used to define relationships between two “objects”.
Rama married Sita
• Adjectives may also be used to define relationships between “objects”.
Vizag is to the east of Mumbai
married and east of are both binary relationships (i.e between “objects” in
the domain of discourse)
Ternary and Quaternary Relationships
• Prepositions may be used to define relationships.
The Danube flows between Buda and Pest.
8
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
This is a 3-way relationship between the Danube and the towns of Buda and Pest.
• Here is a 4-way relationship between cities.
Chennai is further away from Kolkata than Delhi is from Mumbai
Generalising Relationships
Consider the following simple sentences.
Rama married Sita.
Karna married Draupadi.
Sugreeva married Tara
. . .married . . .
• Clearly, the binary relation married can be used as a general relation between the various
“objects” in the universe of discourse.
• Each of the above statements provides an instance of this relation.
• Some of the instances may be true and some may be false.
Relationships Parameterised
• We may parameterise the relationship by the use of variables, such as x, y etc.
• The variables are placeholders and may be replaced by the actual objects in the universe
of discourse to obtain instances.
• Each instance of a predicate is a proposition.
• x married y is a parametrised relationship and is called a predicate.
• Similarly, other relationships (ternary, quaternary) may also be parameterised to obtain
ternary or quaternary predicates.
Predicate Symbols
Generalising the other relationships, we have the following predicate symbols as examples.
• x flows between y and z.
• w is further away from x than y is from z.
Unary Predicates
• Every property of objects may be parameterised to unary predicates, e.g.
x is omniscient.
y is omnipotent.
• Since classes and subclasses may also be characterised as properties, they may also be
parameterised as unary predicates.
– x is/are human
– x is/are animal
– x is/are mortal
Predicates: Prefix Notation
A uniform prefix notation for predicates (parameterised relations).
• human (x)
• mortal (x)
• married (x, y)
• between (x, y, z)
• further (w, x, y, z)

9
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
Key points
• Predicates are parameterised structures which express properties and relationships.
• Predicates do not themselves have truth values. They have formal parameters (place-holders)
which may be instantiated to yield propositions which have truth values.
• The formal parameters of a predicate are called variables, and they may be instantiated by
substituting objects from the universe of discourse for variables.
• Predicates whose parameters have been instantiated are propositions. These propositions
have truth values.

Universal Generalisations
The word “All” as in

adjective. All humans are mortal.

adverb. All prime numbers are positive integers.
pronoun. All teenagers are perpetually hungry.

is used as a universal generaliser which defines a possibly empty class of objects, all of whose
members share a certain property.
Formal Translations of Universals
Natural Language Equivalents
• ∀x[human(x) → mortal(x)]
• ∀y[prime(y) → positive integer(y)]
• ∀z[teenager(z) → perpetually hungry(z)]
• ∀x[∀y[(man(x) ∧ man(y)) → equal(x, y)]]
Existential Generalisations
In contrast to “All” the word “Some” as in

adjective. Some humans are mortal.

adverb. Some prime numbers are positive integers.
pronoun. Some teenagers are perpetually hungry.

is used as an existential generaliser which defines a nonempty class of objects, all of whose
members share a certain property.
Formal Translations of Existentials
Alternative Natural Language Equivalents
• ∃x[human(x) ∧ mortal(x)]
• ∃y[prime(y) ∧ positive integer(y)]
• ∃z[teenager(z) ∧ perpetually hungry(z)]
Predicate Logic
Free and Bound Variables
A positive integer x is a prime iff x ≠ 1 and for each y and z such that x = yz either y = 1 or
z = 1.
• The universe of discourse is the set of the positive integers.
10
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
• prime(x) is a unary predicate denoting the property of being a prime.
• Symbolically, we have the definition of prime(x) as
(x > 1) ∧ ∀y[∀z[x = yz → (y = 1 ∨ z = 1)]]
• Notice that all occurrences of y and z are bound by the quantifiers ∀y and ∀z, respectively,
whereas the variable x is free.
Naming Predicates: Definiens and Definiendum
• We may have large predicates (the definiens) defining some intuitive property (named by
the definiendum) of elements in the universe of discourse, as in the definition of prime.
• We write definitions in the form

• In the following definition

prime(x) is the definiendum and its definition is the definiens.

• The free variables of the definiens are automatically the parameters of the definiendum.
Syntax of Predicate Logic
We extend the object language of Propositional Logic to Predicate logic by
• allowing predicates i.e. propositions parameterised by variables
• allowing the two generalisations.
This gives us the following formal language of Predicate logic.
Definition
ϕ, ψ ::= p(x) | (¬ϕ) | (ϕ ∧ ψ) | (ϕ ∨ ψ)
| (ϕ → ψ) | (ϕ ↔ ψ) | ∀x[ϕ] | ∃x[ϕ]
where p(x) denotes a predicate parameterised by a vector x = x1, . . . , xn of variables
Propositions in Predicate Logic
Definition
A predicate with no free variables is called a proposition.
Notice that
• Propositions previously defined also had no free variables.
• A predicate with no free variables bears a truth value in the universe of discourse and is
hence a proposition.
• Instead of atomic propositional symbols, we now have a collection of predicate symbols,
each with a fixed arity.
Closures and Closed Formulae
Definition
If ϕ(x) is a predicate with free variables x = x1, . . . , xn for some n > 0, then its closures are
defined as follows.
Universal closure. ∀x[ϕ(x)].
Existential closure ∃x[ϕ(x)].
• In general, when the free variables are unknown, we write ∀[ϕ] and ∃[ϕ] to denote the

11
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
respective closures.
• The closures of predicates are propositions.
Semantics of Quantified Propositions
Definition
Given a universe of discourse U, a proposition of the form
• ∀x[ϕ(x)] is true if for every a ∈ U, ϕ(a) is true, and false if there is at least one b ∈ U such
that ϕ(b) is false.
• ∃x[ϕ(x)] is true if there is at least one b ∈ U such that ϕ(b) is true and false if for every a ∈
U, ϕ(a) is false.
De Morgan laws for Quantification
The identities previously defined for propositions may be extended to quantified propositions.
Theorem
¬∀x[ϕ(x)] ⇔ ∃x[¬ϕ(x)]
¬∃x[ϕ(x)] ⇔ ∀x[¬ϕ(x)]
Corollary
∀x[ϕ(x)] ⇔ ¬∃x[¬ϕ(x)]
∃x[ϕ(x)] ⇔ ¬∀x[¬ϕ(x)]
Tautological Forms of Predicates
Even though a predicate with free variables cannot have a truth value, all predicates that
have tautological forms, such as the equivalences expressed as biconditionals, continue to
remain tautologies even in predicate logic.
Example
Consider the unary predicate p(x) ∨ ¬p(x) which is a trivial tautological form. It is clear that
every instance p(a) ∨ ¬p(a) is a tautology. Hence both p(x) ∨ ¬p(x) is always true and
existential closures viz. ∀x[p(x) ∨ ¬p(x)] and ∃x[p(x) ∨ ¬p(x)] are also always true.
Tautological Forms
Definition
A tautological form is a syntactic pattern made up of propositional variables such that any
uniform replacement of the propositional variables by propositions yields a tautology.
Example
(X ∧ Y ) → X is a tautological form. It is easy to see that for any uniform substitution of
propositions for the propositional variables such as ϕ for X and ψ for Y we get the formula (ϕ
∧ ψ) → ϕ which is a tautology.
Inference Rules for Quantification
We have just four rules for quantification:
UI. Universal Instantiation
UG. Universal Generalisation
EI. Existential Instantiation
EG. Existential Generalisation

12
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
Inference Rule: Universal Instantiation
We are still interested in truth-preserving reasoning. Hence if a universally quantified formula
is (assumed) true then every instance of it must also be true.

∀x[X(x)]
∴ X(a) ∴ X(y) (UI)
Notice that the place-holder x may be replaced by any value a in the universe of discourse or
even by another variable y or even the same variable x.
Example 1: The Mortality of Socrates

All humans are mortal. ∀x[human(x) → mortal(x)]

Socrates is human. human(Socrates)
Therefore Socrates is mortal. ∴ mortal(Socrates)

1. ∀x[human(x) → mortal(x)] (Premise 1)

2. human(Socrates) (Premise 2)
3. human(Socrates) → mortal(Socrates) (1, UI)
4. mortal(Socrates) (3, 2, MP)
Inference Rule: Existential Generalisation
Existential quantification of a formula can be inferred from any substitution instance that is
deemed to be true.
X(a)
∴ ∃x[X(x)] (EG)
Since the universe of discourse is nonempty any particular instance may be generalised
existentially
Inference Rule: Existential Instantiation
If an existentially quantified formula is deemed to be true then there must be some instance of
it which is true. In general there is an element a (a constant) in the Universe of discourse for
which the body of the predicate is true.
Hence the converse of rule EG would automatically hold.

∃x[X(x)]
∴ X(a) (EI)
In general the constant symbol a chosen as instance must be “fresh” i.e. it should not have been
chosen earlier in the argument as an instance of some other existentially quantified formula.
The constant chosen is particular and not arbitrary.
Inference Rule: Universal Generalisation
X(x)
∴ ∀x[X(x)] (UG)
This rule at first sight appears to be wrong! However, in extenuation of it, the occurrence of a
free variable x makes the choice of instance arbitrary rather than particular.
13
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
Example 2:

1. All animals are mortal. ∀x[animal(x) → mortal(x)]

2. All humans are animals. ∀x[human(x) → animal(x)]
Therefore all humans are mortal. ∴ ∀x[human(x) → mortal(x)]
3. animal(x) → mortal(x) (1, UI)
4. human(x) → animal(x) (2, UI)
5. human(x) → mortal(x) (4,3, HS)
6. ∀x[human(x) → mortal(x)] (5, UG)
Reasoning With Quantification
Example 3:

1. No mortal is perfect. ¬∃x[mortal(x) ∧ perfect(x)]

2. All humans are mortal. ∀x[human(x) → mortal(x)]
Therefore no human is perfect. ¬∃x[human(x) ∧ perfect(x)]

3. human(z) → mortal(z) (2, UI)

4. ∀x[mortal(x) → ¬perfect(x)] (1, not-forall)
5. mortal(z) → ¬perfect(z) (4, UI)
6. human(z) → ¬perfect(z) (3, 5 HS)
7. ∀x[human(x) → ¬perfect(x)] (5, UG)
Example 4:
Consider the statement Not everything is edible.
Then
1. ¬∀x[edible(x)] (Premise 1)
2. edible(y) (1, UI)
3. ∀x[edible(x)] (2, UG)
and
1. ¬∀x[edible(x)] (Premise 1)
2. ¬edible(y) (1, UI)
3. ∀x[¬edible(x)] (2, UG)
are both wrong!
Inference: General Principle
The application of UI in both proofs of example 4 is wrong. An instantiation rule may be
applied to an entire formula, but never to a quantified proper sub-formula of a formula.
Otherwise, truth may not be preserved.
However,
Any subformula of a formula may be replaced by a logically equivalent formula, since truth
will still be preserved.

14
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
Example 5:

1. All dogs are carnivorous. ∀x[dog(x) → carnivorous(x)]

2. Some animals are dogs. ∃x[animal(x) ∧ dog(x)]
Therefore some animals are carnivorous. ∃x[animal(x) ∧ carnivorous(x)].
3. animal(a) ∧ dog(a) (2, EI)
4. dog(a) → carnivorous(a) (1, UI)
5. dog(a) (3, Simp)
6. animal(a) (3, Simp)
7. carnivorous(a) (4,5, MP)
8. animal(a) ∧ carnivorous(a) (6,7, Conj)
9. ∃x[animal(x) ∧ carnivorous(x)] (8, EG)
Example 6:

1. Some dogs are animals. ∃x[dog(x) ∧ animal(x)]

2. Some cats are animals. ∃x[cat(x) ∧ animal(x)]
Therefore some cats are dogs. ∃x[cat(x) ∧ dog(x)]
3. dog(a) ∧ animal(a) (1, EI)
4. cat(a) ∧ animal(a) (1, EI)
5. dog(a) (3, Simp)
6. cat(a) (4, Simp)
7. cat(a) ∧ dog(a) (5,6, Conj)
8. ∃x[cat(x) ∧ dog(x)] (7, EG)
In this example, while the constant a was used to instantiate premise 1, the same constant was
used to instantiate premise 2 as well. This has led to the wrong conclusion derived in step 8.
Clearly, if a fresh constant, say b, were used to instantiate premise 2, such a conclusion would
have been impossible.
In the above case, the invalidity of the argument was obvious because in the universe of
discourse as we know it, cats are not dogs, and hence the conclusion is false even though the
premises are true.
A Re-look at Logical Validity
• Starting with an informal notion of validity, there were questions about restricting them to
only sound arguments.
• We had formalised it in propositional logic by requiring precise and objective criteria which
included the burden of proof.
• Validity or invalidity could be evaluated by evaluating the correctness of the proof.
• Finally, we defined logical validity formally in terms of truth functionality leading to logical
consequence.
Validity of Closed Formulae
• For an argument to be logically valid, it does not suffice that the premises and the conclusion
are true in the universe of discourse.
• It is necessary that the argument expressed as a proposition be a tautology.

15
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
• If all the premises and the conclusion are closed formulae, then the same notion of validity
that holds for propositions should also hold for closed formulae.
Validity in Predicate Logic
• Our notion of the validity of an argument still remains the same as before.
• Our notion of the invalidity of an argument remains the same, too. But it requires us to give a
more complicated counter-example involving interpretations.
Proving Invalidity in Predicate Logic
Example 1:

1. All whales are heavy. ∀x[whale(x) → heavy(x)]

2. All elephants are heavy. ∀x[elephant(x) → heavy(x)]
Therefore all whales are elephants. ∴ ∀x[whale(x) → elephant(x)]
If the Universe of discourse consists of only one animal a, then the given argument is logically
equivalent to the propositional argument
3. whale(a) → heavy(a) (1, UI)
4. elephant(a) → heavy(a) (1, UI)
5. ∴ whale(a) → elephant(a) (41)
This argument is invalid if we assign the truth value 1 to both whale(a) and heavy(a) and the
truth value 0 to elephant(a).

1. All whales are heavy. ∀x[whale(x) → heavy(x)]

2. Some elephants are heavy. ∃x[elephant(x) ∧ heavy(x)]
Therefore all whales are elephants. ∴ ∀x[whale(x) → elephant(x)]

Again for a model containing a single element a, we get the following equivalent
propositional argument.

1. whale(a) → heavy(a) (1, UI)

2. elephant(a) ∧ heavy(a) (1, UI)
3. ∴ whale(a) → elephant(a)
For this argument to be invalid it is necessary that whale(a) is assigned the truth value 1 and
elephant(a) is assigned the truth value 0. However, then the premise elephant(a) ∧ heavy(a)
becomes false. Hence, the propositional version of this argument is actually valid.
Example 2:

1. All whales are heavy. ∀x[whale(x) → heavy(x)]

2. Some elephants are heavy. ∃x[elephant(x) ∧ heavy(x)]
Therefore all whales are elephants. ∴ ∀x[whale(x) → elephant(x)]
Again for a model containing a single element a we get the following equivalent propositional
argument.
1. whale(a) → heavy(a) (1, UI)
2. elephant(a) ∧ heavy(a) (1, UI)

16
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
3. ∴ whale(a) → elephant(a)
For this argument to be invalid it is necessary that whale(a) is assigned the truth value 1 and
elephant(a) is assigned the truth value 0. However, then the premise elephant(a) ∧ heavy(a)
becomes false. Hence, the propositional version of this argument is actually valid.

Deductions vs. Induction

Deduction
• Deduction states that if the premises are true, the conclusion must be true.
• Deduction guarantees the truth of the conclusion.
• Deductive reasoning is truth-preserving
• Deduction moves from general → specific.
Example:
1. If a circuit receives no power, it will not operate.
2. This circuit receives no power.
3. ∴ This circuit will not operate.
• The conclusion cannot be false if the premises are true.

• Deduction is used when engineers:

• Prove theorems
• Validate algorithms
• Derive constraints
• Verify safety conditions
• Use logic circuits
• Apply mathematical models
• Deduction is evaluated using:
• Validity
• Soundness
• Truth tables
• Rules of inference
• Proof techniques

Induction
• An inductive argument is one whose conclusion follows from the premises, with some degree of
probability rather than with certainty.
• There is no certainty about the conclusion in a valid inductive argument.
• At best, the premises act as evidence to support the conclusion
Example:
Consider the following argument.
17
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
Everyone in Gen Z suffers from attention deficiency.
All students in your class belong to Gen Z.
Therefore all students in your class suffer from attention deficiency.
is a valid argument (one can deductively prove it).

We could replace it by the following

Most of Gen Z suffers from attention deficiency.
All students in your class belong to Gen Z.
Therefore most students in your class suffer from attention deficiency
• The argument is inductively valid because it is implicitly assumed that your class consists of
students from Gen Z chosen independently of considerations such as attention deficiency
• Induction states that if the premises are true, the conclusion is likely to be true.
• The conclusion is probable, not guaranteed.
• Inductive reasoning is NOT truth-preserving
• Even with true premises, the conclusion may still be false.
• Induction moves from specific → general
Example
• Every battery tested in this batch so far has failed early.
• ∴ The next battery in the batch will also likely fail early.
• The above conclusion is a probabilistic claim, not a certainty.
• Induction is used when engineers
• Predict system performance
• Infer reliability from samples
• Estimate population parameters
• Do machine learning
• Perform risk assessment
• Build statistical models
• Types of inductive reasoning:
• Inductive generalization
• Causal inference
• Statistical reasoning
• Analogical reasoning
• Probabilistic inference
Quick check:
Example 1: Identify the type of the argument? Why?
1. If pressure exceeds 300 psi, the system shuts down.
2. Pressure exceeded 300 psi.
3. System must shut down
Example 2: Identify the type of the argument. Why?
1. The last 50 sensors calibrated to 0.5 mm tolerance worked perfectly.
2. Future sensors calibrated similarly will likely work perfectly.

18
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
Inductive Arguments
• Induction is reasoning where the conclusion goes beyond the information in the
premises.
• It aims to generalize from specific observations to a broader rule.
• Structure:
• Premises describe observed cases.
• Conclusion describes unobserved cases (generalization).
• Inductive arguments give:
• Strong or weak support, not guaranteed truth.
• General Form:
• Most observed Xs have property Y. Therefore all (or most) Xs have
property Y.
Example 1:
• You test 20 steel rods from a batch:
• 18 withstand 1500 MPa
• 2 withstand slightly less
• Conclusion (inductive):
• The whole batch is likely to withstand ~1500 MPa.
• But one faulty rod could still exist → not guaranteed.
• Example 2:
• You observe that every acidic solution tested turns blue litmus red.
• Conclusion: All acidic solutions will turn blue litmus red.
• Using such repeated observations, the scientific laws are discovered

Plausible Arguments
• Plausible arguments are arguments that sound reasonable, believable, or intuitively
acceptable, even if we don’t have statistical data or conclusive evidence.
• They rely on Common sense, Prior experience, Domain knowledge,
Expert judgment or Reasonable assumptions
• Key Idea in Plausibility depends on rational believability but not
certainty.
• Plausible Arguments are stronger than “guessing” but weaker than
empirical induction.
• Example 1:
• Given that the network latency increases whenever server load spikes,
• Today’s slow system performance is due to increased server load
• In the above conclusion: we don’t have data yet, but based on previous
experience, the reasoning is plausible
• Example 2:
• If a new bacterial strain survives in extreme heat,
• It is believable that it contains special heat-resistant proteins.
• The above belief is reasonable but not proven until tested.
• When used?

19
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
• Troubleshooting
• Early hypothesis formation
• Engineering intuition
• Preliminary design justification
• Real-time decision-making

Probabilistic Arguments
• Probabilistic arguments use explicit numerical probability to support a conclusion.
• They quantify uncertainty.
• Key Idea: The premises assign probability values to outcomes.
• Conclusions are expressed using:
• Likelihood
• Probability
• Confidence levels
• Risk percentages
• Probabilistic arguments are stronger than plausible arguments because they
rely on quantifiable evidence.
• Example 1:
• Premise: “In our last 1000 flight hours, the failure probability of the
hydraulic pump is 0.003.”
• Conclusion (probabilistic): The likelihood of pump failure in the next
flight is very low (0.3%).
• This is not certainty but a quantified measure of belief.
• Example 2:
• Given the drug's success rate in clinical trials is 92%,
• Conclusion: There is a 92% probability it will be effective in the next
patient.
• Applications in the domain:
• Risk assessment
• Reliability engineering
• AI/machine learning
• Statistical physics
• Probability models for safety systems.

Statistical Arguments
• Statistical arguments reason from samples → populations using statistical methods.
• All statistical arguments are inductive, but not all inductive arguments are statistical.
• Key Idea is to Sample data to infer something about a larger population.
• It involves:
1. Sample size
2. Sampling method
3. Statistical significance
4. Confidence intervals

20
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
5. Distributions
• Statistical arguments provide probability-based generalizations relying on data.
• Example 1
1. A manufacturer tests 50 batteries from a shipment of 10,000:
• Mean lifespan = 200 cycles
• Standard deviation = 12 cycles
• Conclusion (statistical): The average lifespan of all 10,000
batteries is likely around 200 ± 12 cycles.
• Example 2
1. From a random sample of 1,500 people: 23% show a genetic marker.
2. Conclusion: About 23% of the entire population carries the marker.
3. This conclusion has statistical confidence (e.g., 95%) and margin of
error.
• Application domains
1. Quality control
2. Experimental science
3. Machine learning / AI models
4. Engineering product testing
5. Predictive modeling

Summary of the Types of Arguments

Type of Key Feature Strength Example
Argumentation
1. Inductive From specific to Medium–High Testing rods →
general (depending on whole batch
2. Plausible Based on Medium
data) “Likely server load
reasonability, common issue”
3. Probabilistic Uses explicit
experience High (quantified Pump failure
probability values uncertainty) probability =
4. Statistical Based on sample data High Battery
0.003 testing
sample →
How do they differ? population
1. Inductive = observational → generalization
2. Plausible = reasonable → intuitive
3. Probabilistic = quantified likelihood
4. Statistical = sample data → population inference

21
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
Inductive generalisations

Inductive reasoning
• Moves from specific observations -> broad conclusion
• Conclusion is probable, not certain
• Foundation of scientific reasoning
• Example:
• Observation: “Three students who study daily score high.”
• Conclusion: “Regular study leads to high performance.”

What Is Inductive Generalisation?

• A conclusion about a whole group based on a sample
• Also called simple enumeration
• Example:
• “Many modern phones overheat, so modern phones are unreliable.”
• Perceptual similarity can mislead.
A rabbit and a squirrel look alike superficially (both small, furry, four-legged), so
someone might generalise properties from one to the other based on appearance
alone.
• Category (conceptual) similarity is stronger.

22
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
• The grey animal on the right is another lagomorph (rabbit/hare family). Generalising
from one rabbit to another is more reliable because they share deeper biological
properties.
• Important fact to mention: Rabbits are lagomorphs, while squirrels are rodents despite
visual similarity, they belong to different taxonomic groups. This is a clear example of
why surface appearance (perceptual match) is a weaker basis for inductive
generalisation than relevant categorical similarity.

EXAMPLES
EXAMPLE 1:
I observed that students who attended revision classes consistently scored higher in the last
three internal tests.
Therefore, it is reasonable to conclude that attending revision classes generally improves
academic performance.
EXAMPLE 2
The last five times I visited the café, the service was quick and consistent.
So, it seems likely that the café generally maintains fast service.
EXAMPLE 3:
I noticed that several students who practiced past-year papers performed better in the final
exam.
So, practicing previous papers seems to improve exam performance.
EXAMPLE 4:
Every plant I watered daily grew faster than the ones watered irregularly.
Thus, regular watering seems to promote healthier plant growth.

Weak Generalisation

A weak generalization is an inductive conclusion drawn from too few examples,

unrepresentative cases, or limited evidence, making the conclusion unreliable or unjustified.
It often results from insufficient sample size, biased samples, or hasty observations, leading to
poorly supported general claims.

23
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
EXAMPLE1:
1. From One Student to All Students
Observation: One student submits an assignment late.
Weak generalisation:
“Students these days never submit work on time.”
Why weak?
Only one case → applied to all students.
EXAMPLE 2:
From One Personal Experience to Everyone
Observation: A person had a bad experience with one online order.
Weak generalisation:
“Online shopping is always unreliable.”
Why weak?
One incident → generalised to entire system.
Why These Are Weak Inductive Generalizations?
• Small evidence -> big conclusion
• No counter-examples considered
• No alternative explanations
• No supporting data
• Sample size too small

What Is a Strong Inductive Generalisation?

A strong generalisation is an inductive argument where the conclusion is drawn from a large,
diverse, and representative sample, making the general statement reliable and well-supported.
A strong generalisation does not guarantee truth, but it makes the conclusion highly reasonable.

EXAMPLE 1:
Evidence:
A company collects productivity data from 25 teams over one year, showing teams with clear
weekly goals perform better.
Strong generalisation:
“Teams with structured weekly goals are generally more productive.”
Why strong?
Long duration + many teams + measurable data.
EXAMPLE 2
Evidence:
Across 20 schools, teachers observe that students engaged in group discussions participate
more actively and retain content better.
Strong generalisation:
“Group discussions usually improve student engagement and retention.”
Why strong?
Multiple schools + repeated observations + variety of contexts.
Features of Strong Generalisation
• Large sample size
• Representative sample (age, gender, region,

24
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
department, etc.)
• Consistent trend
• Reliable data collection method
• Avoids biased or narrow examples

Weak Generalisation vs Strong Generalisation

Feature Weak Generalisation Strong Generalisation

Sample Size Very small Large enough to support conclusion

Sample Type Unrepresentative / biased Representative and diverse

Evidence Quality Limited, anecdotal, subjective Systematic, reliable, objective

Conclusion Strength Exaggerated and unjustified Reasonable and well supported

Logic Issue Hasty generalisation fallacy Sound inductive reasoning

Predictability Low High

Risk of Error High Lower

How to Convert Weak → Strong Generalisation

Weak Generalisation
“My friend lost weight by skipping breakfast; so skipping breakfast helps everyone lose
weight.”

Strong Generalisation
“A study involving 5,000 participants over 6 months found no consistent evidence that skipping
breakfast leads to weight loss for most people.”
Why strong?
• Scientific study
• Many participants
• Long duration

25
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
Statistical Generalisations

1. What is a Statistical Generalisation?

A statistical generalisation is a type of reasoning where we draw a conclusion about an entire
population based on information collected from a sample of that population.
It answers questions like:
• "What is true of the sample is probably true of the whole group."
It is a probabilistic argument, not a deductively certain one.

Structure of a Statistical Generalisation

Sample → Inference → Population
• We study a sample (a smaller group).
• We observe a pattern in that sample.
• We generalise the pattern to the whole population.

General Form
1. X% of observed sample S has property P.
2. Therefore, probably X% (or many) of population G has property P.

Conditions for a Good Statistical Generalisation

1. The sample is large enough
Small sample = weak conclusion.
2. The sample is representative
Not biased (age, region, gender, interest, etc.).
3. The sampling method is appropriate
Random sampling → best.
4. The conclusion does not overstate the evidence
Use words like probably, likely, most, in general.

Weak Statistical Generalisation

A weak generalisation happens when:
• sample is too small
• sample is biased
• conclusion is too broad

Example of Weak Generalisation

Sample: “I asked 3 students in my class. All three said the exam was difficult.”
Conclusion: “Therefore, all students in the college found the exam difficult.”
Why weak?
• Sample size: only 3
• Sample not representative
• Conclusion is too strong

Strong Statistical Generalisation

A strong generalisation happens when:
26
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
• the sample is large
• randomly selected
• representative
• conclusion is proportional to evidence

Example of Strong Generalisation

Sample: “A survey of 800 randomly selected students shows that 72% prefer online library
services.”
Conclusion: “Therefore, around 70% of all college students probably prefer online library
services.”
Why strong?
• Large sample
• Random selection
• Population proportion is not exaggerated

Converting Weak to Strong Statistical Generalisation

Weak Version
“I spoke to 5 teenagers, and they all like gaming. Therefore, all teenagers love gaming.”
Strong Version
“I surveyed 500 randomly selected teenagers across different schools, and 68% reported that
they enjoy gaming. Therefore, a large proportion of teenagers probably like gaming.”

Inductive vs Statistical Generalisation

Feature Inductive Generalisation Statistical Generalisation

Basis Few observations Large sample with numerical data

Type of Qualitative Quantitative
reasoning
Strength Often weaker Statistically stronger

Evidence “I have seen 3–4 cases” “Survey of 1000+ people”

Conclusion Broad guess based on Probabilistic conclusion

type patterns
Example “All crows I saw are “92% of sampled crows are black → most
black” crows are black”
EXAMPLES
Example
Observation:
• One rabbit I saw was fast.
• Another rabbit was fast.
• A third rabbit was also fast.

27
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
Conclusion:
“Rabbits are fast animals.”
(Generalising from a few examples)
Example
Survey:
Out of 2,000 college students surveyed, 78% said they prefer online resources for study.
Conclusion:
“Most college students prefer online resources for study.”
(Generalising from statistical evidence)
Example 1 :The “One Cup Coffee” Pattern
Observation:
• Yesterday, my friend drank one cup of coffee and acted like a philosopher.
• Today, she drank one cup and gave a motivational speech.
• Last week, she drank one cup and decided to start a startup.
Conclusion (Inductive):
“Anyone who drinks one cup of coffee suddenly believes they can change the world.”
Example 2: Snack Consumption Pattern
Survey of 2,500 office workers:
• 74% said they come to meetings mainly for the snacks.
• Only 26% said the meeting content was important.
Conclusion:
“Most employees attend meetings for biscuits, not business.”
Example 3:The Overconfident Cat
Observation:
• My cat knocked one glass from the table and walked away like a boss.
• Next day, it pushed a book and stared at me like it won a fight.
• Yesterday, it kicked the TV remote and proudly sat on the couch.
Conclusion:
“Cats think they own the house if they successfully knock down two or more objects.”
Example 4:Data:
In a study of 3,800 students,
• 62% admitted they turned off their camera just to eat snacks.
• 21% turned it off to take a nap.
• 17% actually had technical problems.
Conclusion:
“Most students turn cameras off in online classes for everything except technical issues.”

Statistical Generalisations
Statistical generalisations are those that are supported by empirical or experimental
observations.
Statistical generalisations are of two kinds:
• Universal statistical generalisations
• Partial statistical generalisations

28
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
While a conclusion drawn from a universal generalisations could be deductive any conclusion
drawn from a partial generalisation is mostly inductive.

Universal Statistical Generalisations

100% of Gen Z suffers from attention deficiency.
is an example of a universal statistical generalisation. It is equivalent to the universal
statement

Everyone in Gen Z suffers from attention deficiency.

Any conclusion that is derived from purely universal statistical generalisations
is deductive in nature and therefore certain.
Example:
Everyone in Gen Z suffers from attention deficiency.
Every student in your class belongs to Gen Z.
Therefore every student in your class suffers from attention deficiency.

Partial Statistical Generalisations

90% of Gen Z suffers from attention deficiency.
is an example of a partial statistical generalisation.
Hence the argument
90% of Gen Z suffers from attention deficiency.
Every student in your class belongs to Gen Z.
Students who have been admitted have been randomly chosen from Gen Z.
Therefore 90% of students in your class suffer from attention deficiency. is a “good” inductive
argument.

The Law of Large Numbers

Good partial statistical generalisations follow the law of large numbers i.e. Given a sample of
independent and identically distributed values, the sample mean converges to the true mean.
Example
Given a six-sided die, marked with numbers 1 to 6 each with equal probability the expected
value of the throw is 1+2+3+4+5+6 6 = 3.5. By the law of large numbers a large number
of throws of the six-sided die would yield a mean value approaching 3.5.

Casual and Analogical arguments

A causal argument tries to show that one event causes another

What caused this?
Why did this happen?
What will happen if X occurs?
1. Cause occurs
2. Therefore, effect happens
OR
1. Effect is observed
2. Therefore, we infer the cause
Keywords
29
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
• Because
• Leads to
• As a result
• Causes
• Therefore
Example
• Premise: Smoking damages the lungs.
Conclusion: Therefore, smoking causes health problems.
Example
• (Reverse Cause)
• Premise: The plants look dry.
Conclusion: They must not have been watered.

Analogical Arguments
An analogical argument compares two similar things and reasons that because they are similar
in some respects, they will also be similar in another respect.
A and B are similar in features F1, F2, F3
A has property G
Therefore, B probably also has property
Keywords
• Similar to
• Just like
• In the same way
• Likewise
Example
• Premise: A smartphone and a laptop are both digital devices used for communication
and browsing.
Conclusion: So, if a laptop can get a virus, a smartphone can also get one.
Example
• Premise: A seed growing into a plant is like a child growing into an adult.
Conclusion: Both need proper care and nourishment.

Fallacies in informal arguments

Errors in reasoning that make an argument weak or invalid.
In informal arguments, fallacies usually occur because of language misuse, false assumptions,
or irrational reasoning.

Informal Fallacies
Informal fallacies are those which cannot be identified without understanding the concepts
involved in the argument.
Composition
Composition fallacy is the fallacy of concluding a property of the whole from a property of its
parts.
Compositions Examples
We have already seen “is-a” relations.
30
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
Every human is a mammal.
Every mammal is an animal.
Therefore every human is an animal.
Every part of this car is metallic.
Therefore the whole car is metallic.
Every part of me is made up of atoms.
Therefore I am made up of atoms.
are all valid arguments. However already seen “is-a” relations

The Fallacy of Composition

Examples of invalid arguments
Example 1:
Every part of me is made up of atoms.
Therefore I am made up of atoms.
Atoms have no life.
Therefore every part of me has no life.
Therefore I have no life.
Example 2:
Every brick in the building is a cuboid.
Therefore the whole building is a cuboid.

The Division Fallacy

Conversely,
The division fallacy is the fallacy of concluding a property of the constituents from a property
of the whole.
Example 1 : valid
The whole team died in a plane crash.
Therefore every member of the team died in the plane crash.
Example 2 : Invalid
Gen Z suffers from attention deficiency.
Therefore every member of Gen Z suffers from attention deficiency.

False Dichotomy
A false dichotomy is a disjunction that does not exhaust all possibilities.
• The problem is not with the validity of the argument. Such arguments
are usually deductively proved using the rules of disjunctive syllogism (DS), constructive
dilemma (CD) or destructive dilemma (DD).
• The problem is with an insufficient number of choices in the premise. Hence the argument is
valid but not sound

Fallacy of Equivocation
The fallacy of equivocation uses ambiguity or double meanings deliberately to conclude
perversely from the premises.
General Hush said the terrorists must be stopped before rising from his desk.
Any person who sincerely believes that people are waiting to arise from inside his desk is mentally

31
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
ill.
Therefore, General Hush is mentally ill.
Humourists and comedians often use this persuasively

Fallacy of Accent
Interpretations of the same statement may be modified by placing the emphasis on different
words of the same statement.
The fallacy of accent is committed when one interpretation is used to gain acceptance of a
premise and a different interpretation is used to draw a conclusion.

Example:
Children are a headache.
Aspirin makes headaches go away.
Therefore Aspirin makes children go away.

Ad Hominem (Fallacies in informal Arguments)

Is a fallacy where someone rejects or attacks an argument by attacking the person, instead of
addressing the actual reasoning or evidence.
The focus shifts from the argument → to the person making it.
A person's character, behavior, or background does not determine whether their argument is true
or false.
Example
Person A: “Regular exercise improves health.”
Person B: “What do you know? You’re lazy and unfit.”
Person B attacks the person, not the claim about exercise.

Example
“Don’t listen to her argument on climate change; she failed science in school.”
The argument should be judged on facts, not the student's grades.

Straw Man fallacy

Happens when someone misrepresents or exaggerates another person’s argument so it
becomes easier to attack.
Example
Person A: “We should have more healthy food options in the school canteen.”
Person B: “So you want to ban all tasty food and force everyone to eat salad?”
Person B twists the argument into something extreme.
Example
Person A: “We should use public transport to reduce pollution.”
Person B: “So you’re saying no one should ever own a car again?”
Misrepresentation of the original suggestion.

Appeal to Emotion
An Appeal to Emotion is a fallacy in which someone tries to win an argument by using emotions
(fear, pity, anger, sympathy, guilt) instead of logical reasons or evidence.
Example
32
[Link], DEPT. OF CSE, GITAM SCHOOL OF COMPUTER SCIENCE AND
ENGINEERING, VISAKHAPATNAM
CRITICAL THINKING – MODULE-5
“Please give me full marks on my assignment. I worked very hard and stayed up all night.”
The student uses sympathy, not academic reasons.
“Look at these sad pictures of abandoned animals. You must adopt one.”
Uses emotion, not argument about capability or care.

False Cause
This fallacy happens when someone wrongly assumes that:
If event A happens before event B, then A must have caused B.
In reality, the two events may be unrelated.
Example
“The street lights turned on, and then the traffic reduced.
So street lights reduce traffic.”
Traffic reducing has other causes; lights didn’t cause it.
Example
“My Instagram account got more likes after I changed my phone.
So the new phone is giving me more likes.”
Likes depend on content and audience, not the device.

33