Block 1

UNIT 1 INTRODUCTION TO
INTELLIGENCE AND ARTIFICIAL

INTELLIGENCE
Structure
1.0
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
Introduction
Objectives
Some Simple Definition of A.I.
Definition by Eliane Rich
Definition by Buchanin and Shortliffe
Another Definition by Elaine Rich
Definition by Barr and Feigenbaum
Definition by Shalkoff
Summary
Further Readings/References
Introduction to Intelligence
and Artificial Intelligence
Page Nos.
5
6
6
6
8
12
13
18
19
20
1.0 INTRODUCTION
In this unit, we discuss intelligence, both machine and human. However, as our
subject matter in the course is machine intelligence, or artificial intelligence, our
discussion of the subject matter is mainly from the point of view of machine
intelligence. Machine intelligence is popularly known as Artificial Intelligence and is
generally referred to by its abbreviation viz. AI. We also shall use the name AI for
the discipline throughout. The style of discussion in this unit is to start with a
definition of AI by some pioneer in the field, and then elaborate the ideas involved in
the definition. Further, while elaborating the ideas involved in the definition, we
introduce a number of relevant new ideas, concepts and definitions to be used later. In
this process, we have introduced and/or explained the following:
i)
ii)
iii)
iv)
v)
vi)
vii)
viii)
Artificial Intelligence & Human Intelligence

When a problem necessarily requires parallel processing for its solution
Symbol vs. number issue
Numeric vs. symbolic processing
Algorithm vs. non-algorithmic method and limitation of algorithmic approach
Limitations of computational abilities of logical devices
Heuristics an important A.I technique
Time/space complexities of programs and problems, exponential time vs.
polynomial time, hard problems
ix)
Role of search and knowledge in solving hard problems; search as an important
AI technique
x)
Enumeration of issues about knowledge
xi)
Information: one of the four fundamental properties of nature
xii) Organisation; relations between information and organisation and between
information and intelligence
xiii) A principle of intelligence
AI as a science and as an engineering discipline
xiv) Controversial issue about the possibility of machine intelligence at least
equating or surpassing human intelligence.
xv) Brief history of AI the name and as a subject
Introduction to A.I
1.1 OBJECTIVES
After going through this unit, you should be able to:
discuss the concepts of intelligence and artificial intelligence as visualised by
a number of leading experts in the field;
enumerate the fields in which human beings are still better than computers;
tell the difference between the concepts of:
(i) Symbol and number
(ii) Algorithmic and non-algorithmic methods
(iii) Information and knowledge
(iv) Polynomial time and exponential time complexities
tell the relation of information to organisation and to intelligence.
1.2 SOME SIMPLE DEFINITIONS OF A.I.

Before looking at what A.I. is in the experts opinions that involve technical terms
needing some explanation, we state below three simple definitions from completely
non-specialists point of view:
1. A.I. is the study of making computers smart.
2. A.I. is the study of making computer models of human intelligence; and finally
3. A.I. is the study concerned with building machines that simulate human
behaviour.
The first one of the above definitions is based on behaviour-oriented approach to
A.I. According to this approach, AI is concerned with programming computers to
behave intelligently. The next definition is more from a psychologists point of view,
where the purpose is to use computer as a tool to understand better the mechanisms of
the human mind, and the final definition, which we may call robotic approach to
A.I., includes under the domain of A.I., not only writing of computer programs but
building also the whole of an intelligent system or machine including its mechanical,
electronic, optical components and other components.
In order to have still better and concrete opinion about what is AI and its subjectmatter, we consider definitions suggested by leading writers and pioneer contributors
to the development of A.I. We supplement these definitions with comments to
facilitate the understanding of the underlying ideas and of the technical terms involved
in the definitions.
1.3 DEFINITION BY ELIANE RICH

Definition 1: The first definition we consider is by Elaine Rich, the author of the
book entitled Artificial Intelligence[1]. It states: Artificial Intelligence is the
study of how to make computers do things, at which, at the moment, people are
better.
Comment 1, Definition 1: Implicit in the Richs definition is the idea that there are
mental tasks that computers can do better than human beings and vice-versa, there are
tasks which at the moment human beings can do better than computers. It is wellknown that computers are better than human beings in the matter of
numerical computation,
information storage, and

repetitive tasks.
On the other hand, at the moment, human beings are much better than machine in
the matter of
understanding including the capability of explaining,
predicting the behaviour and structure of a system,
in the matter of common-sense reasoning,
in drawing conclusions when available information is either incomplete,
inconsistent or even both, and
also, in visual understanding and speech understanding, which require
simultaneous availability (availability in parallel) of large amount of information.
In essence, it is found that computers are better than human beings in tasks
requiring sequential but fast computations, where human beings are better than
computers in tasks, requiring essentially parallel processing. In order to clarify
what it is for a problem to essentially require parallel processing for its solution, we
consider the following problem:
C
Figure 1.1
We are given a paper with some letter, say, C written on it and a card-board with a
pin-hole in it. The card board is placed on the paper in such a manner that the letter is
fully covered by the card board as shown in Figure 1.1. We are allowed to look at the
paper only through the pin-hole in the card-board. The problem is to tell correctly the
letter written on the paper by just looking through the pin-hole. As the information
about the black and white pixels is not available simultaneously, it is not possible to
figure out the letter written on the paper. The figuring out the letter on the paper
requires, simultaneous availability of the whole of the grey-level information of all the
points constituting the letter and its surrounding on the paper. The gray-level
information of the surrounding of the letter provides the context in which to interpret
the letter.
We consider another example that shows the significance of contextual information or
knowledge and its simultaneous availability for visual understanding. From the
following picture, we can conclude that one of the curved lines represents a river and
other curved lines represent sides of the hills only on the basis of the simultaneous
availability of information of the pixels.
Introduction to A.I
Contextual information plays a very important role not only in the visual
understanding but also in the language and speech understanding. In case of speech
understanding, consider the following example, in which the word with has a
number of meanings (or connotations) each being determined by the context.
Mohan saw the boy in the park with a telescope.

Mohan saw the boy in the park with a dog.
Mohan saw the boy in the park with a statue.
Further, the phrase for a long time may stand for a few hours to millions of years,
but again determined by the context, as explained below.
For a long time..
He waited in the doctors room for a long time.

It has not rained for a long time.
Dinosaurs ruled the earth for a long time.
Comment 2, Definition 1: In addition to the advantage that human beings have in

the matter of parallel processing as explained above, Boden [12] says: humans have
two psychological strengths which are yet to be approached by computer systems:
a teeming richness of conceptual sources and the ability to evaluate new ideas in
many different ways. The first of these is difficult enough for AI to emulate, the
second is even more problematic.
Comment 3, Definition 1: The definition is rather weak in the sense that it fails to
include some areas of potentially large importance viz, problems that can be solved at
present neither by human beings nor by computers. Also, it may be noted that, by
and by, if computer systems become so powerful that there is no problem left, which
human beings can solve better than computers, then nothing is left of AI according to
this definition.
1.4 DEFINITION BY BUCHANIN AND

SHORTLIFFE
Next, we consider a definition obtained by rephrasing and combining the two
definitions, viz., the first by Bruce G. Buchanin as given in Encycolopedia
Britanica and the second by BUCHANIN & SHORTLIFFE as given in Rule-Based
Expert Systems [2]. It states:
Definition 2 AI is the branch of computer science that deals with symbolic rather
than numeric processing and non-algorithmic methods including the rules of
thumb or heuristics instead of algorithms as techniques for solving problems.
Comments/Explanations 1, Definition 2: Symbolic processing vs numberic

processing: We generally think and use 128 as a number which has a definite relation
with the number say 105 (that of greater than), also with 64 (that of being double of)
and again with 2 (that of being a multiple of). Also, 128 can be multiplied, through
built-in mechanisms, with any number say 3 to get 384. However, if the numbers
mentioned above including 128 denote the route numbers of buses or house numbers
a residential colony then none of the relations or operations mentioned above, may
hold. Rather, in this context, these relations of 128 w.r.t. 105, etc. and the operations
like multiplication even do not make any sense. We cannot tell what is meant by
saying House Number 128 is greater than House Number 105 in a normally
acceptable way.
On the other hand, even a non-digital character sequence say ABC may represent a
number, for example, in hexadecimal number system. Also, words of English (or any
other) language when considered lexicographically ordered, acquire some numeric
attributes.
The conclusion we draw from the above discussion, is that a word as a sequence of
characters (including digits) may denote a number or a symbol (henceforth, a symbol
stands for non-numeric symbol) depending upon the context in which it is used.
And the context is determined by the nature of the problem under consideration. If
the problem can be solved using only numerical aspects of the objects in the domain
and environment of the problem, then we have the advantage of having built-in
relations (like less than, equal to etc.) and the built-in operations (like +, -, * etc.) that
can be readily used without having to define these relations and operations explicitly.
But, unfortunately, most of the problems, we encounter for our day to day survival or
even for our intellectual pursuits, involve not only quantitative, but qualitative aspects
also of the objects of the problem domain. In order to solve these problems, we use
common sense reasoning, exploit our capability for visual and linguistic
understanding, try to get meaning out of incomplete and even inconsistent information
that is available, in addition to a number of other known and unknown mechanism.
Qualitative aspects, their ideal representations, defining relations and operations
involving these aspects, are generally different for different types of problems.
Hence, it is impossible to capture in general relevant relations and operations for all
types of problems, and then defining these as built-in operations of the machine,
because there are potentially infinite types of problems that we encounter and try to
solve.
This discussion explains the basic difference between numeric processing and
(non-numeric) symbolic processing. Summarizing, numeric processing involves
only a small number of well-defined relations and operations having universally
accepted meanings, and hence, these relations and operations can be incorporated as a
part of a computer system. On the other hand, in symbolic processing the relations
and operations required to solve a problem depend upon the problem under
consideration, and hence, have to be defined explicitly along-with or as a part of
programs constituting the solutions of the problems.
The weakness of numeric processing, however, is that it can be used in solving a
small fraction of the set of problems we want to or need to solve. The numeric
processing can be used in solving only those problems, the solutions of which involve
only numeric aspects of the objects involved in the domain and environment of the
problem under consideration. For the solution of other solvable problems, we need to
Introduction to A.I
use symbolic processing. It is not out of place to mention that not all problems
which even can be stated precisely or formally, are amenable to computer
solutions using even symbolic processing. More discussion in this respect follows
next, under Comments/Explanations 2 for Definition 2.
Comments/Explanation 2, Definition 2 : Algorithmic method vs non-algorithmic
method, heuristics : We recall that an Algorithm is a step-by-step procedure with
well-defined starting and ending points, which is guaranteed to reach a solution to a
specific problem. A solution to a problem which can be expressed as an algorithm is
called an algorithmic solution. An algorithmic solution may involve only numeric
processing or may involve symbolic processing with/without numeric processing. For
the purpose of further discussion, symbolic processing includes/subsumes numeric
processing. Algorithmic approach even when using symbolic processing has
limitations. During 1930s, a number of logicians and mathematicians including
Gdel, Church, Post, Turing and Kleene suggested a number of mathematical
models of a computer, and through these models tried to explain the nature of
computation, established a number of useful results about computation and also
found the limits of computational power.
They proved that even through a problem may be expressed precisely or formally (i.e.,
in terms of mathematical entities like sets, relations functions etc.), yet it need not
yield to an algorithmic solution. A problem which has at least one algorithmic
solution is called a solvable problem. They further proved that out of even solvable
problems, only a small fraction can be solved if only feasible amount of resources
like, time and space are used. Informally, feasible amount of resources means that
the requirement for resources does not increase too rapidly with the increase in size of
the problem. The notion of the size of a problem will be defined formally later on
(under comment 1 on Definition 3). However, an intuitive idea about the concept of
the size of a problem and its role in estimating the resource requirement for solving
the problem can be had through the simple problem of calculating income tax for each
of the tax-payers. The requirement of resources like, time and computing equipment
for 1000 tax-payers would be much less, as compared to the requirement of resources
for computing income-tax for one million tax payers. In this problem, n, the number of
tax-payers for whom the income-tax is to be calculated, may be taken as size of the
problem.
This limitation and other difficulties with algorithmic solutions has given impetus to
efforts for finding non-algorithmic solutions of problems. Neural Network
approach to solving many difficult problems, is a well-known alternative to
algorithmic methods of solving problems. In AI, there are mainly two approaches to
solve problems, which generally difficult to solve with algorithmic methods. One
approach is Neural approach, mentioned just above. The other approach is called
symbolic approach. The symbolic approach cannot be said to be non-algorithmic. The
main difference between symbolic approach of AI and algorithmic approach is that
symbolic approach of AI emphasizes exploitation of the knowledge of the domain and
the environment of the problem under consideration. Some of this knowledge is in the
form rules of thumb, generally, called heuristics in AI.
In order to realise the limitations of algorithmic approach to solving problems, we
need not refer to highly theoretical work by the earlier mentioned
logicians/mathematicians. The limitation of the approach may be appreciated through
the following simple example.
Consider the problem of crossing from one side over to the other side of a busy road
on which a number of vehicles are moving at different velocities. A step-by-step (i.e.,
algorithmic) method of solving this problem may consist of:
10
(i)
(ii)
(iii)
(iv)
Knowing (exactly) the distances of various vehicles from the path to be

followed to cross over.
Knowing the velocities and accelerations of the various vehicles moving on the
road within a distance of, say, one kilometer.
Using Newtons Laws of motion and their derivatives like s = ut +
1 2
at , and
2
calculating the times that would be taken by each of the various vehicles to
reach the path intended to be followed to cross over.
Adjusting dynamically our speeds on the path so that no collision takes place
with any of the vehicle moving on the road.
The above is a systematic step-by-step method, i.e., an algorithm, of crossing the road
that may ensure no collision with any vehicle. But, how many of us can follow it?
Hardly anybody! First of all, it is practically impossible to measure distances,
velocities and accelerations of various vehicles on the road, even within a radius of
one kilometer. Secondly, even if we assume theoretically that it is possible to measure
distances, velocities and accelerations of various vehicles and to calculate safe timings
to cross the road, we would not like or care to follow the above-mentioned algorithm,
because our past experience, our sense of survival and other built-in mechanisms have
allowed us, in the past, to cross over safely without following any systematic method.
All of us just guess the distances of the vehicles, safe enough to cross over, and then
actually cross over at an appropriate time. Not even one in 1000, on an average gets
hurt when crossing a road using only guesses, in a crowded city like, Delhi, where
movement of vehicles is one of the most chaotic and unruly in the whole world.
However, this is not to deny that once in a while, the guess is incorrect and someone
or other gets hurt or even is killed almost every day.
Each one of us every day, comes across hundreds of problems similar to the one of
crossing of a road. And, for each such problem one uses a good guess and one
generally is able to solve the problem satisfactorily each time, though the solutions
may not be the best possible ones. And, or once in a while, we even fail to get any
solution using the guess. However, if we insist on only following a systematic stepby-stop method that guarantees best possible solution for solving each problem, then
we would hardly be able to make any progress in our day to day business of even
mere survival.
The essence of the above discussion is that while attempting solutions of many of the
problems, it is not only desirable but almost essential that for each of such problems
we follow some good guess instead of following a step-by-step systematic method
that guarantees the best solution. In A.I, these guesses are called heuristics. In later
chapters, we discuss heuristics in detail. However, for the time being, we state that
heuristics are good guesses, possibly based on past experience, judgement, intuition
or hunches, which lead us most of the time to reasonably good solutions, though these
guesses do not guarantee the best solutions or even any solution for every instance of
the problem under consideration.
The advantage of using heuristics is that we do not have to rethink completely
everytime we are faced with a problem of the type of which another problem has
already been solved satisfactorily. If we have a handy rule of thumb that may apply to
the current problem, it may suggest to us how to proceed.
11
Introduction to A.I
1.5 ANOTHER DEFINITION BY ELAINE RICH

The next definition, again by Elaine Rich [1] is more technical and involves some
concepts from Theory of Computation. It sates:
Definition 3: Artificial Intelligence is the study of techniques for solving
exponentially hard problems in polynomial time exploiting knowledge about the
problem domain.
Comments/Explanations 1, Definition 3: For deeper understanding of the concepts
like hard, solvable and unsolvable problems, any one of the books by Brady [3], by
Lewis and Papadimiriou [4] or by Hopcroft and Ullman [5] may be consulted.
However, for our purpose of appreciating Definition 3 of A.I., we briefly discuss only
the required essentials from Theory of Computation (TOC). In the comments on
Definition 2, we have already talked about the mathematical models of computation
and also about the limitations of algorithmic solutions.
As computer study is partly engineering in nature, in the sense that we design and
implement or produce computer solutions for different types of problems and hence
these products, i.e., solutions, need to be evaluated vis-a-vis problem specifications
and other measures like, efficiency in respect of time and space requirements of the
solutions. In order to measure the efficiency of a suggested computer solution of a
problem, the earlier mentioned logicians/mathematicians suggested the concepts of
time complexity and space complexity for the solutions and even for the problems.
The basic idea behind these complexity measures is that all the operations that a
computer (present or future generations) can execute, may be thought of as composed
of a small number of basic operations. These basic operations can be easily compared
for their relative requirements for time and space. For the basic operation say O1,
which is expected to take minimum time (or space) among all the basic operations, the
time (or space) complexity is assigned the number one. For any other basic operation,
complexity is a positive number depending upon the expected relative requirement for
time (or space) for the operation as compared to that for the operation O1. For other
computer operations, time/space complexity may be computed from those for the
basic operations. Also from these complexities, we can compute the complexities of
the programs using the size of the input data as an additional parameter. For example,
to multiply two n x n matrices we require n3 multiplications and (n3 n2) additions.
12
Thus, complexity of the straight-forward method of multiplication of two n x n

matrices is n3 . + (n3 n2) , where and are complexities of, respectively, the
operations of addition and that of multiplication of two numbers. The time/space
complexity of a problem may be defined as the time/space complexity of the program
which has the least complexity among all the known programs that solve the problem.
Further, a problem is said to be polynomial time problem, if the time complexity of
the problem is some polynomial a0 nk + a1 nk-1 + + ai nk-i + + ak, where n is the
size of the data. Similarly, exponentially hard problem is one for which time
complexity is of the form an, with a > 1. For large n, the value of an exponential
function increases at a much faster rate than the increase in the value of any given
polymial functions in n. For a given polynomial function f(n) and an exponential
function g(n), it is always possible to find a positive integer k such that g(n) > f(n) for
all intergers n k. Thus, the problems requiring exponential time are considered
harder than the problems requiring polynomial time. Polynomial time is
considered as reasonable amount of time, and on the other hand, exponential time
is considered as impractical or infeasible amount of time from computational
point of view. This is why, the problems requiring exponential time are considered as
hard problems. Also, using the fact that the complexity of a problem is the least of
the complexities of its known algorithms, we can not solve an exponential time
problem in polynomial time.
Comment 2, Definition 3: Role of knowledge in solving hard problems:
In view of the previous comments, no polynomical time algorithmic solution can exist
for any (exponentially) hard problem. However, there are mechanisms/techniques
which when used in a solution of a hard problem, though divest the solution of its
step-by-step or algorithmic characteristic, yet may make it a polynomial time solution.
Use of appropriate knowledge of the problem domain has been found useful in
techniques that when used, solve hard problems in polynomial time. Definition 3
declares the scope of (or the subject-matter) AI as the study of techniques that
exploit appropriate knowledge to solve hard problems in polynomial time. The
role of appropriate knowledge in reducing time complexity of a solution cannot be
overemphasized. The following simple example supports this claim abundantly:
Ms X is to meet Ms Y at her residence. Initially, let us assume that Ms X knows only
that Ms Y lives in Delhi and knows nothing else about Ms Ys residence. A step-bystep or algorithmic solution to the problem may be to search the residential places, one
by one, in some order, in Delhi and to stop when Ms. Ys place is located. The
complexity of the algorithm, on the average, is undoubtedly very large. However, if
X further knows that Y lives in some particular colony say Hauz Khas in Delhi, then
search is substantially reduced by searching residential places only within Hauz Khas.
Further, if Ms X also knows the house number in Hauz Khas, then there is hardly any
search required and X can directly reach Ys residence. Next, consider just opposite
situation so far as availability of knowledge is concerned. Let us X even do not know
that Y lives in Delhi. We can easily guess the plight of X when she, if follows a stepby-step method, is required to search, possibly all over the world, for the residence of
Y.
The importance of (relevant) knowledge in solving difficult problems was recognised
by the pioneers in the very early stages in the development of A.I. As we shall find
subsequently, major portion of A.I. is constituted of discussion of various issues
about knowledge: methods for acquisition of knowledge, for representation of
knowledge, for organisation of knowledge, for manipulation of knowledge, for
maintenance of knowledge and for restricting search of the problem domain by
exploiting the knowledge of the domain.
1.6 DEFINITION BY BARR AND FEIGENBAUM

Next, we come to another definition of A.I. which involves human intelligence a
phenomenon only partially understood yet. Rather, computers and some A.I.
techniques are being used in helping the psychologists in establishing their theories
about intelligence and other mental processes. But this definition provides another
angle to look at A.I. as the study of attempts at incorporating intelligence, whatever
we understand of it yet, in machine. This definition, in a way, would also justify the
inclusion of the word intelligence in the name Artificial Intelligence for the
subject-matter of our study. The definition, by Barr and Feigenbaum in The
Handbook of Artificial Intelligence [6], is as given below.
Definition 4: Artificial Intelligence is the part of computer science concerned
with designing intelligent computer systems, i.e., systems that exhibit the
characteristics we associate with intelligence in human behaviour.
Discussion/Comments 2. Definition 4: What is intelligence or intelligent behaviour
in humans? In order to have good grasp on the intent of this definition of A.I., we
attempt to enumerate some known characteristics of Intelligence. There must be some
basic mechanisms behind intelligent behaviour and some important
attributes/characterises of intelligence which have defined human recognition or
13
Introduction to A.I
understanding, because of which we are not able to describe the phenomenon of

intelligence in its totality. Capturing the total essence of the phenomenon of
intelligence in humans through a definition is almost impossible, as is noted by one of
the leaders of A.I viz. Patrick Winston [7] of Massachusetts Institute of Technology
(MIT), when he states defining intelligence usually takes a semester-long
struggle, and even after that I am not sure we ever get a definition really nailed
down. However, there are some characteristics of intelligence which are readily
acceptable, some others acceptable after some thinking and still others that may be
controversial. We enumerate the characteristics as considered by some A.I. writers
and contributors and others. Enumeration of these characteristics here is essential
because as A.I. technologists, we would study various techniques that help us in
incorporating these characteristics, through computer programs, into machines, which
we attempt to make intelligent according to Definition 4 of Artificial Intelligence. We
give below the attributes verbatim from the respective sources.
Douglous R. Holstadter in his book: Gdel Escher, Bach: An Eternal Golden
Braid [8], which won him Pulitzer Prize and was a best-seller mentions on Page
26 of the book, the following as essential abilities for intelligence:
to respond to situations very flexibly;

to take advantage of fortuitous circumstances;
to make sense out of ambiguous or contradictory messages; to recognize the
relative importance of different elements of situation;
to find similarities between situations despite differences which may separate
them;
to draw distinctions between situations despite similarities which may link
them;
to synthesize new concepts by taking old concepts and putting them together in
new ways;
to come up with ideas which are novel.
Fisher and Firschein in their book Intelligence: The Eye, the Brain and the
Computer [9] on Page 4 state that they expect an intelligent agent to be able to:
Have mental attitudes (beliefs, desires and intentions)

Learn (ability to acquire new knowledge)
Solve problems, including the ability to break complex problems into simpler
parts.
Understand, including the ability to make sense out of ambiguous or
contradictory information.
Plan and predict the consequence of contemplated actions, including the ability
to compare and evaluate alternatives.
Know the limits of its (own) knowledge and abilities.
Draw distinctions between situations despite similarities.
Be original, synthesize new concepts and ideas, and acquire and employ
analogies.
Generalize (find a common underlying pattern in superficially distinct
situations)
Perceive and model the external world
Understand and use language and related symbolic tools.
They further state that there are a number of human attributes that are related
to the concept of intelligence, but are normally considered distinct from it:
14
Awareness (consciousness)
Aesthetic appreciation (art, music)
Emotion (anger, sorrow, pain, pleasure, love, hate)
Sensory acuteness
Muscular coordination (motor skills)
Next, we discuss intelligence from more fundamental level. The ideas explained
below are based on the Information Transfer Model of scientific phenomena due to
Norbert Wiener (1894-1964). Norbert Wiener, an intellectual prodigy and author of
the famous book entitled Cybernetics [14], suggested the Transfer of Information
model to be a better model than the prevailing model based on Transfer of Energy for
explanation of a number of scientific phenomena. Through the Wieners theory, a new
discipline was born, also, called Cybernetics
However, our discussion is mainly based on ideas explained in the book Beyond
Information by Tom Stonier [10]: According to the ideas explained in Stonier,
there are four fundamental properties of the universe viz. energy, matter,
information and evolution (or change). The cardinality of information in the
universal scheme of things can be judged from the following argument: All the
entities from down to nucleons to the whole of the universe, each is known to us as an
organised system of simpler objects, e.g., fundamental particles organise into
nucleolus, nucleolus organise to form atomic nuclei, which alongwith electrons and
protons organise into atoms and so on. Molecules, polymers, membranes, organs,
living beings, societies, planets, planetary systems, galaxies and finally the whole
universe, each is known as an organised system of some simpler objects. An
organisation builds upon pre-existing organisations. Thus an organised system is
recursively obtained (or defined) as an interdependent assembly of elements
and/or organised systems. And it is information what is exchanged between
components of an organised system to effect their interdependence and to maintain
the integrity of the system as long as the system survives against the fourth
fundamental property of the universe, i.e., evolution or change. Gravitational pull,
now an established entity, is just an information processing activity. Thus
information is no more or no less an abstract concept than energy or matter.
What mass is to matter and the heat is to energy, so is organisation to
information. Each of the former is a visible and measurable form of the
corresponding latter. More the mass, more the matter in a system; more the heat,
more the capacity to do work, i.e., energy in the system; similarly higher the degree
(or more the complexity) of the organization (in terms of underlying organizations of
the components and their components and so on, and in terms of the number and
levels of interactions and relations between components at a particular level) higher is
the information content of the system.
The relation between information and organisation and the characteristic
difference between the two is exactly what is the relation and characteristic
difference between a number and a numeral. A number is an abstract concept,
wheras a numeral is its physical manifestation or representation. A number may have
many representations and even may use many mediums for representations or
manifestations. In the form of, writing on the paper, as patterns of ink dots on a
piece of paper, the same number may be represented as 7 in decimal, 111 in binary,
and even 4 + 2 + 1 again in decimal. In computers memory, the same number is
represented with the help of electronic components, a different medium, and not as
shapes composed of ink-dots. In human brain the same number is represented,
possibly, as some neural net.
Summarising, a number is a concept which needs a medium for its manifestation
or physical representation for the purpose of conveying, or transformation. This
representation is called a numeral. But it should be clear that when we say that I
need two books, the word two is not just the sequence of three letters viz t, w
and o i.e., the representation, which is intended to be conveyed but just it is the
abstract number which is intended to be conveyed. Because of the tangibility or
15
Introduction to A.I
perceptual visibility of the representation, we always use the representation for

various purposes like applying some operations or for conveying, but it is not the
representations, but the instances of the idea or concept (of number) which are
intended to be transformed or conveyed.
Similarly, information is a concept and an organisation is its representation, i.e.,
physical manifestation. For the purpose of applying operations (like refining
information, adding information etc) or for conveying information we use
organisation (as patterns of ink dots on paper or as neural net in brain etc). Then, we
manipulate the organisation or representation for applying operations on information
(operations again are abstract, whereas manipulations are their physical realizations).
Also, we communicate the organization for conveying information (communication is
physical realisation of conveying). As in the case of number, informations
representation may be through various organisations on various type of media such as
patterns of ink dots on paper, neural nets in brain, or on flip-flops in electronic
memory. For example, the information content of the organisation in the form of
pattern of inkdots in the sentence Heat is a form of energy is stored in the brain as an
organization in the form of a Neural Network etc.
Remark 1: We have already mentioned that an atom is an organsied system and so
are organs in the human body and so are the galaxies in the universe. Also every
organized system contains information. Hence, as we say God gave the numbers so
we can say God created information and information is not just a product of
human mental activities.
Remark 2: Information organises not only matter and energy but itself as well.
Evolution leads to discontinuities, i.e., to something which is qualitatively different
from the earlier existing entities. And intelligence is the phenomenon which has
evolved out of information but which is qualitatively different from information.
Remark 3: Intelligence, being an outcome of evolution from information over a
period spanning back almost upto the Big Bang, must be a spectrum of phenomena
and can not be an all or nothing affair. Further, intelligence can not be a singledimensional phenomenon. The veracity of this claim can be judged by analogy with
evolution of matter and energy into myriad forms differing from each other in almost
innumerable ways.
However, in order to draw a sort of fuzzy boundary between intelligent organisations
or systems and the other systems, let us consider the case of living matter. Matter
evolved from subatomic particles to atoms, molecules and so in potentially infinite
number of different material objects. Out of these materials, there is a large number
of types of objects (for example, human beings) for which we can say with surety that
these are living matter and again for a large number of types of material objects (for
example soil), we can say with surety that these are not living matters. Of course,
there may still be large number of objects which may not be distinctly characterised as
either. How we decide living or non-living is based on a finite collection of attributes
of matter and degree of each such attribute.
In the similar manner, we consider a finite set of attributes and degrees for each
attribute for organizations, i.e., information processing systems, which allow us to
categorise systems as intelligent or otherwise in such a way that the systems which are
generally considered as intelligent are categorised as intelligent and further whatever
systems are generally considered as non-intelligent are categorized as non-intelligent.
As evolution has taken over billions of years, hence divergence among information
processing systems intelligence-wise must be potentially infinite. Thus any
categorization based on only finite number of attributes would always be incomplete
16
and leave large number of cases uncategorisable. To begin with, we start with a
working definition of intelligence and then later expand on it:
Intelligence is a property of advanced information processing systems, which not

only engage in information processing, but are able to analyse their dynamically
changing environment and to respond to it in such a way that:
i)
ii)
iii)
survivability of the system is enhanced

its reproducibility is enhanced (reproducibility is sort of self propagation
through another system)
if the system is goal-oriented, then achievability of goal is enhanced.
In stead of attempting to categorize most of the information processing systems as

either intelligent or non-intelligent, if we are interested in their relative merit as
intelligent systems, then the following principle of intelligence may be useful. The
principle quoted in Stonier [10] states: The intelligence exhibited by a system
may, at least in theory, be measured as a ratio, or quotient, of the ability of a
system to control its environment, versus the tendency of the system to be
controlled by the environment.
The above principle fits best, at least, in the limiting cases: At one extreme is a cube
of sugar dissolving in a cup of tea. Although highly organised, the cube is totally
controlled by environmental elements and hence, according to the above principle, it
has zero intelligence. This is exactly what we also feel. On the other extreme is
technologically advanced human society which can divert the waters of rivers to
irrigate plains to provide an assured supply of food to its population. Thus
intelligence measure of a technologically advanced society as a whole is, according to
the above principle, quite high. This conclusion of the above principle is in
consonance with what we also feel.
Below we include some more attributes and/or definitions of intelligence by leading
computer scientists and A.I. researchers. The purpose is to be aware of as many facts
as possible of yet-to-be completely understood phenomena of intelligence. Only then,
it may be possible to design and develop really intelligent programs to solve those
hard problems which are so far not amenable to computer solutions. Also, in this
process, we are providing a list of attributes, against which an A.I. engineer can test
their products for the quality of their product as intelligent one.
Hofstadter [8] on Page 37 says: It is an inherent property of intelligence that it can
jump out of the task which it is performing and survey what it has done. Selfevaluation and self-criticism are part of intelligent behaviour.
Fishler and Firschein [9] on Page 4 state: Intelligence involves learning capability
and goal-oriented behaviour. Additional attributes of intelligence include reasoning,
common-sense, planning, perception, creativity, memory retention & recall.
Shanks [11] on Page 49 observes: The simplest and perhaps safest definition of
intelligence is the ability to react to something new in a non-programmed way. The
ability to be surprised or to think for oneself is really what we mean by intelligence.
In order to explain the concept of A.I. through Definition 4, we discussed the
concept of intelligence itself as a phenomenon. Next, we quote another definition of
A.I. again based on the concept of intelligence and given but from engineering point
of view by another pioneer in the field, viz Shalkoff, a Professor of Electrical
Engineering.
17
Introduction to A.I
1.7 DEFINITION BY SHALKOFF

Definition 5: Shalkoff [13] says: Perhaps broadest definition is that AI is a field
of study that seeks to explain and emulate intelligent behaviour in terms of
computational processes.
Comments 1, Definition 5: According to the above definition, AI is partly scientific
in nature because it seeks to explain the phenomena of intelligence, and partly
engineering because it seeks to emulate intelligent behaviour through computational
processes, i.e., by generating representations (of knowledge) and development of
programs that automatically (autonomously) solve problems, so far solved by only
intelligent humans beings.
In view of the fact that A.I. is partly an engineering discipline according to the above
definition, let us recall what is meant by the concept engineering.
Engineering may be thought of as the application of science and mathematics by
which properties of matter and sources of energy in nature are made useful (meeting
some requirements and according to some specifications) to man in structures,
machines, products, systems and processes etc.
Again, in the light of the definition of Engineering given above, a part of the
definition by Shalkoff may be paraphrased as through application of A.I., products
are obtained that exhibit intelligent behaviour. This paraphrased part of the
definition by Shalkoff raises another issue: How to judge/evaluate whether a product
obtained through an application of A.I., is actually intelligent.
The issue of testing an A.I. product as intelligent product was considered by the
pioneers themselves including Alan Turing, the most well known name among the
pioneers. In honour of Turing, the most prestigious award for contributions to the field
of computer science, has been instituted and is given annually.
Turing suggested a test, which is well known as Turing Test, for testing whether a
product has intelligence. An outline of the Turing test is given below.
For the purpose of the test, there are three rooms. In one of the rooms is a computer
system claimed to have imbedded intelligence. In the other two rooms, two persons
are sitting, one in each room. The role of one of the persons, let us call A, is to put
questions to the computer and to the other person to be called B, without knowing to
whom a particular question is being directed, and, of course, with the specific purpose
of identifying the computer. On the other hand, the computer would answer in such a
way that its identity is not revealed to A.
The communication among the three is only through computer terminals so that
identity of the computer or the person B can be known only on the basis of quality of
responses as intelligent or otherwise, and not just on the basis of other human or
machine characteristics. If A is not able to know the identity of the computer, then
computer is intelligent. More appropriately, if the computer is able to conceal its
identity from A, then the computer is intelligent.
We may note here that, in order to be called intelligent, the computer should be clever
enough not to give answer too quickly, at least not within a fraction of a second, even
if it can, say, to a question involving finding of the product of two numbers each of
more than 20 digits.
18
Objections to Turing Test: There have been a number of objections to the Turing
test as a test of intelligence of a machine. One of the most well known objections is
called Chinese Room Test proposed by John Searle. The essence of the Chinese
Room Test, that we are going to explain below, is that convincing successfully by a
system, say A , of possessing qualities of another system, say B, does not imply that
the system A actually possesses the qualities of B. For example, the capability of
convincing others by a male human of being a woman, does not give the male the
quality of bearing a child like a woman.
The scenario for the Chinese Room Test consists of a single room with two windows.
In the room a scholar on Shakespeare, knowing English, but not knowing Chinese, is
sitting with a sort of encyclopedia on Shakespeare. The encyclopedia is printed in
such a way that for each pair of facing pages, one page is written in Chinese
characters and the other page is translation in English of the contents of the facing
page in Chinese. Through one of the windows questions on Shakespeares literature in
Chinese characters are sent to the person sitting inside. The person looks through the
encyclopedia and on finding in the encyclopedia the exact copy of the sequence of
characters sent in, reads its translation in English, thinks of its answer and writes the
answer in English for his/her own understanding, finds the corresponding sequence of
Chinese characters in the encyclopedia, and sends the sequence of Chinese characters
through the other window. Now, Searle says that, though the scholar successfully
behaves as if s/he knows Chinese, but, as per assumption it is not so. Just from the fact
that a system is able to simulate a quality, it can not be inferred that the system
possesses the quality.
1.8 SUMMARY
This is an introductory unit to the course. The unit gives a birds eye view of the
whole of the course of Artificial Intelligence. The approach, in the unit, is to start with
a definition by some pioneer in A.I. In the process of discussion of the definition, a
number of relevant new concepts are gradually built up and discussed.
In Section 0.3, we discuss definition of A.I., as given by Eliane Rich . It states:
Artificial Intelligence is the study of how to make computers do things, at which,
at the moment, people are better.
.
In this context, it was discussed that human beings are still better than computers in
the problem areas, which require parallel processing and simultaneous availability of
information.
According to the next definition of A.I., as given by Buchamin & Shortliffe:
AI is the branch of computer Science that deals with symbolic rather than
numeric processing and non-algorithmic methods including the rules of thumb or
heuristics in stead of algorithms as techniques for solving problems.
In Section 0.4, we discuss the differences (i) between number and symbol, (ii)
between algorithmic and non-algorithmic methods of solving problems.
In the Section 0.5, another definition by Eliane Rich, as given below, is discussed:
Artificial Intelligence is the study of techniques for solving exponentially hard
problems in polynomial time exploiting knowledge about the problem domain.
In context of this definition, we discuss the difference between exponentially hard
problems versus polynomial time problem.
In section 0.6, we discuss the following definition of A.I. by Barr & Feigenbaum:
Artificial Intelligence is the part of computer science concerned with designing
19
Introduction to A.I
intelligent computer systems, i.e., systems that exhibit the characteristics we

associate with intelligence in human behaviour.
In context of this definition, we discuss various characteristics of human intelligence.
In the process, we discuss, relation between information and organisation and relation
between information and intelligence.
Finally, in Section 0.7, we discuss a definition of A.I by Shalkoff, an engineer.
According to this definition, A.I. is partly an engineering and partly a scientific
discipline. As an engineering discipline A.I. is the study of designing and developing
intelligent machines. In context of testing whether a machine is intelligent, we discuss
Turing test and its criticism.
1.9 FURTHER READINGS/REFERENCES

1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
20
Rich E. & Knight K. (1991). Artificial Intelligence.

Tata McGraw-Hill Publishing Company Limited
Buchanan B.G. & Shortlife E.H. eds. (1984), Rule-Based Expert Systems.
Addison-Wesley
Brady J.M. (1977). The Theory of Computer Science. Chapman and Hall.
Lewis H.R. & Papdimitriou C.H. (1981). Elements of Theory of Computation
Prentice-Hall International Editions.
Hoperoft J.E. & Ullman J.D. (1987). Introduction to Automata Theory,
Languages and Computation. Narosa Publishing House
Barr A. & Feigenbaum E.A. (1981-82): The Handbook of Artificial
Intelligence Vol.1, Kaufman
Winston P.H. & Prendergast K.A (1984): Perspective in the AI Business,
eds. MIT Press
Hofstadter D.R. (1979): Gdel, Escher, Bach: An Eternal Golden Braid
Penguin Books.
Fischler M.A. & Firschein O. (1987). Intelligence: The Eye, the Brain, and the
Computer: Addison-Wesley Publishing Company
Stonier T. (1992). Beyond Information. Springer-Verleg.
Schank R.C. with Childers P.G. (1984). The Cognitive Computer on Language
Learning and Artificial Intelligence. Addison-Wesley Publishing Company
Boden M. (1998). The Computer can act as a Brain stormer, The Times of
India, New Delhi dated March 06, 1998.
Schalkoff R.J. (1990). Artificial Intelligence: An Engineering Approach
McGraw-Hill International.
Wiener N. (1948). Cybernatics. Wiley, New York.
Boden M. (1990). The Philosophy of Artificial Intelligence. Oxford University
Press.
The Propositional
Logic
UNIT 2 THE PROPOSITIONAL LOGIC

Structure
2.0
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
2.10
2.11
2.12
2.13
2.14
Introduction
Objectives
Logical Study of Valid and Sound Arguments
Non-Logical Operators
Syntax of Propositional Logic
Semantics/Meaning in Propositional Logic
Interpretations of Formulas
Validity and Inconsistency of Propositions
Equivalent forms in the Prepositional Logic (PL)
Normal Forms
Logical Deduction
Applications
Summary
Solutions/Answers
Further/Readings
Page Nos.
21
23
23
25
26
27
29
30
32
33
35
37
38
38
43
2.0 INTRODUCTION
Symbolic logic may be thought of as a formal language for representing facts about
objects and relationships between objects of a problem domain alongwith a precise
inferencing mechanism for reasoning and deduction. An inferencing mechanism
derives the knowledge, which is not explicitly/directly available in the knowledge
base, but can be logically inferred from what is given in the knowledge base.
The reason why the subject-matter of the study is called Symbolic Logic is that
symbols are used to denote facts about objects of the domain and relationships
between these objects. Then the symbolic representations and not the original facts
and relationships are manipulated in order to make conclusions or to solve problems.
Also, we mentioned that a Symbolic Logic, apart from having other characteristics, is
a formal language. As a formal language, there must be clearly stated unambiguous
rules for defining various constituents or constructs, viz. alphabet set, words, phrases,
sentences etc. of the language and also for associating meaning to each of these
constituents.
The study of Symbolic Logic is significant, specially, for academic pursuits, in view
of the fact that it is not only descriptive (i.e., it tells how the human beings reason)
but it is also normative (i.e., it tells how the human beings should reason).
In this unit, we shall first study the simplest form of symbolic logic, viz, the
Propositional Logic (PL). In the next unit, we consider a more general form of logic
called the First-Order Predicate Logic (FOPL). Subsequently, we shall consider other
symbolic systems including Fuzzy systems and some Non-monotonic systems.
In the propositional logic, we are interested in declarative sentences, i.e., sentences
that can be either true or false, but not both. Any such declarative sentence is called a
proposition or a statement. For example
(i)
(ii)
The proposition: The sun rises in the west, is False,

The proposition: Sugar is sweet, is True, and
21
Introduction to A.I
(iii)
The truth of the proposition: Ram has a Ph. D degree. depends upon whether
Ram is actually a Ph. D or not.
Though, at present, it may not be known whether the statement is True or False,
yet it is sure that the sentence is either True or False and not both True and False
simultaneously.
For a given declarative sentence, its being True or False is called its Truth-value.
Thus, truth-value of (i) above is False and that of (ii) is True.
On the other hand, none of the following sentences can be assigned a truth-value, and
hence none of these, is a statement or a proposition:
(i)
Who was the first Prime Minister of India? (Interrogative sentence)
(ii) Please, give me that book. (Imperative sentence)
(iii) Ram must exercise regularly. (Imperative, rather Deontic)
(iv) Hurrah! We have won the trophy. (Exclamatory sentence)
In propositional logic, as mentioned earlier also, symbols are used to denote
propositions. For instance, we may denote the propositions discussed above as
follows:
P : The sun rises in the west,
Q : Sugar is sweet,
R : Ram has a Ph.D. degree.
The symbols, such as P, Q, and R, that are used to denote propositions, are called
atomic formulas, or atoms. As discussed earlier, in this case, the truth-value of P is
False, the truth-value of Q is True and the truth-value of R, though not known yet, is
exactly one of True or False, depending on whether Ram is actually a Ph. D or
not.
At this stage, it may be noted that once symbols are used in place of given statements
in, say, English, then the propositional system, and, in general, a symbolic system is
aware only of symbolic representations, and the associated truth values. The system
operate only on these representations. And, except for possible final translation, is not
aware of the original statements, generally given in some natural language, say,
English.
We can build, from atoms, more complex propositions, sometimes called compound
propositions, by using logical connectives.
Examples of such propositions are:
(i) Sun rises in the east and the sky is clear, and
(ii) If it is hot then it shall rain.
The logical connectives in the above two propositions are and and ifthen. In the
propositional logic, five logical operators or connectives, viz., ~ (not), (and),
(or), (if then), and (if and only if), are used. These five logical connectives can
be used to build compound propositions from given atomic formulas. More generally,
they can be used to construct more complicated compound propositions from
compound propositions by applying the connectives repeatedly. For example, if each
of the letters P, Q, C is used as a symbol for the corresponding statement, as follows:
P: The wind speed is high.
Q: Temperature is low.
C: One feels comfortable.
22
then the sentence:

If the wind speed is high and the temperature is low, then one does not feel
comfortable
may be represented by the formula (( P Q ) (~ C)). Thus, a compound
proposition can express a complex idea. In the propositional logic, an expression that
represents a proposition, such as P, or a compound proposition, such as ((P Q) (~
C)), is called a well-formed formula.
The Propositional
Logic
2.1 OBJECTIVES
After going through this unit, you should be able to:
tell about what is Logic, Symbolic Logic, and Propositional Logic (PL); further,
about why we study each of these; and about some detailed subject matter of
each of these;
tell the difference between a Proposition/Statement, which forms the basis of PL,
and a sentence in a natural language;
explain the difference between a logical operator and a non-logical operator; any
symbolic logic uses only logical operators;
explain the concept of arguments in a logical system and further should be able to
explain mutual differences between a (i) valid argument (ii) sound argument
(iii) invalid argument, and (iv) unsound argument;
differentiate between an expression that is a well-formed formula wff of PL and

an expression which is not a wff.;
find the truth-value, or meaning, of a wff of PL and should be able to explain

how the truth value of a wff is obtained from the truth values of atomic wffs.
explain the difference between various types of wffs, viz, valid wff; consistent
wff, invalid wff and inconsistent wff;
explain about the various tools, like truth table, logical deduction and reduction
to normal forms that are used to establish validity/invalidity of arguments, and
further should be able to use these tools for the purpose, and
use the tools and techniques of PL in solving problems that can be solved within
a PL system.
2.2 LOGICAL STUDY OF VALID AND SOUND

ARGUMENTS
Logic is the analysis and appraisal of arguments.
An argument is a set of statements consisting of a finite number of premises, i.e.,
assumed statements and a conclusion.
Valid Argument: A valid argument is one in which it would be contradictory for the
premises to be true but the conclusion false.
In logical studies we are interested in valid arguments.
Example of Valid Argument
(i) If you overslept, you will be late
(ii) You are not late.
you did not oversleep.
Example of Invalid argument
23
Introduction to A.I
(i) If you overslept, you will be late

(ii) You did not oversleep
you are not late
(This argument is invalid, because despite not having overslept, one may be late
because of some other engagements or lazyness.)
Another Invalid Argument
(i) If we are close to the top of Mt. Everest then we have magnificent view.
(ii) We are having a magnificent view.
Therefore,
(iii) We are the near the top of Mt. Everest.
(This argument is invalid, because, we may have a magnificent view even if we are not
close to the top of Mt. Everest. The two given statements do not falsify this claim)
How to establish logical validity/invalidity of an argument
We have already discussed invalidity of some arguments, but invalidity above was
based on our intuition. However, intuition may also lead us to incorrect conclusion.
To be sure about the validity of our argument, we need some formal method. In
Section 1.5, we discuss how a Truth table (a formal tool) can be used to establish the
validity/invalidity of an argument.
Sound Argument
We may note that, in the case of a valid argument, it is not required that the
premises/axioms or assumed statements must be True. The assumptions may not be
True, and still the argument may be valid. For example, the following argument is
valid, but its premises and conclusion both are false:
Premise 1: If moon is made of green cheese
Then 2 + 2 = 5
Premise 2: Moon is made of green cheese
(False premise)
From Premise 1 and Premise 2, by applying Modus Ponens, we conclude through
valid argument that 2 + 2 = 5 (which is False).
However, in order to solve problems of everyday life, we need generally to restrict to
only true premises and valid arguments. Then such an argument is called sound
argument.
Sound Argument: is an argument that is valid and has true premises.
(i) If you are reading this, then
you are not illiterate
(ii) You are reading this (true premise)
You are not illiterate (sound conclusion)
Example of valid but not sound argument with correct conclusion.
(i)
(ii)
24
If moon is made of green cheese then 2 + 2 = 4

Moon is made of green cheese (False premise)
To conclude 2 + 2 = 4 (correct) makes the argument a Valid Argument
The Propositional
Logic
Example of Invalid Argument

I (i) If you overslept, you are late.
(ii) you are late.

Therefore, you overslept.
II (i) If you are in Delhi, you are in India.
You are in India.
Therefore, you are in Delhi (invalid argument, though conclusion may be True)
2.3 NON-LOGICAL OPERATORS

One of reason why special symbols:
~
are used in symbolic logic in stead of the corresponding natural languages words:
and, or, not, if. Then,
if and only if, is that the words may have different
meaning in different contexts. For example, the use of the word and in one sentences
has different connotation or meaning from the use in others in the following:
(i) Ram and Mohan are good hockey players.
(the statement can be equivalently broken into two statements:
(i) Ram is a good hockey player (ii) Mohan is a good hockey player)
(ii) Ram and Mohan are good friends.
(though the word and joins two words Ram & Mohan, but can not be equivalently
broken into two statements viz. (i) Ram is a friend (ii) Mohan is a friend)
(iii) Mohan drove a car to reach home, met an accident and got slightly injured.
(Here, the use of the word and is not in a logical sense, but, it is in temporal sense of
and then because statement (iii) has different sense from the statement given in (iv)
below)
(iv) Mohan met an accident, got slightly injured and drove a car to reach home.
Thus from the above statements, it can be seen that the natural language word and
may have many senses, both logical and non-logical. Similarly, the words since,
hence and because are frequently used in arguments to establish some facts. But as
shown from the following two arguments, their use in logical arguments is risky in
the sense that some of the arguments involving any of these words may lead to
incorrect conclusions:
Argument (1): Using the word because, we get correct conclusion from
True statements.
Let
P: Dr. Man Mohan Singh was Prime Minister of
India in the year 2006
(True statement)
Q: Congress party and its allies commanded majority in Indian Parliament in the year
2006
(True statement)
Then the following statement:
P because Q
(True statement/conclusion)
25
Introduction to A.I
Thus, by using the connective because we get a correct/True conclusion from two
True statements viz. P and Q.
Argument ( 2)
In the following using the word, because, we get incorrect/false conclusion from
True statements
Let
P: Dr. Man Mohan Singh was Prime Minister of
India in the year 2006
(True statement)
R: Chirapoonji, a town in north-east India, received maximum average rainfall in the
world during 1901-2000.
(True statement)
However to say
P because R, i.e., to say
Dr. Man Mohan Singe was Prime Minster of India in 2006, because Chirapoonji, a
town in north-east India, received maximum average rainfall in the world during
1901-2000.
is at least incorrect, if not ludicrous.
Thus from two True statements, P and R and by using connective because, in this
case, the conclusion is incorrect.
Thus, by using connective because, in one argument we get a correct conclusion from
two True statements and, on the other hand, we get an incorrect conclusion from True
statements.
2.4 SYNTAX OF PROPOSITIONAL LOGIC

A Well-formed formula, or wff or formula in short, in the propositional logic is
defined recursively as follows:
1.
2.
3.
4.
An atom is a wff.
If A is a wff, then (~A) is a wff.
If A and B are wffs, then each of (A B), (A B), (A B), and (A B) is a
wff.
Any wff is obtained only by applying the above rules.
From the above recursive definition of a wff it is not difficult to see that expression:
(( P ( Q ( ~ R))) S) is a wff; because , to begin with, each of P, Q , ( ~ R) and
S, by definitions is a wff. Then, by recursive application, the expression: (Q ( ~ R))
is a wff. Again, by another recursive application, the expression: (P (Q ( ~ R)))
is a wff. And, finally the expression given initially is a wff.
Further, it is easy to see that according to the recursive definition of a wff, each of the
expressions: (P (Q )) and (P ( Q R )) is not a wff.
26
Some pairs of parentheses may be dropped, for simplification. For example,

A B and A B respectively may be used in stead of the given wffs ( A B) and (A
B), respectively. We can omit the use of parentheses by assigning priorities in
increasing order to the connectives as follows:
, , , , ~.
Thus, has least priority and ~ has highest priority. Further, if in an expression,
there are no parentheses and two connectives between three atomic formulas are used,
then the operator with higher priority will be applied first and the other operator will
be applied later.
The Propositional
Logic
For example: Let us be given the wff P Q ~ R without parenthesis. Then among
the operators appearing in wff, the operator ~ has highest priority. Therefore, ~ R is
replaced by (~R). The equivalent expression becomes P Q (~ R). Next, out of the
two operators viz and , the operators has higher priority. Therefore, by
applying parentheses appropriately, the new expression becomes P (Q (~ R)).
Finally, only one operator is left. Hence the fully parenthesized expression becomes (P
(Q (~ R)))
2.5 SEMANTICS/MEANING IN PROPOSITIONAL

LOGIC
Next, we define the rules of finding the truth value or meaning of a wff, when truth
values of the atoms appearing in the wff are known or given.
1. The wff ~ A is True when A is False, and ~ A is False when A is true. The wff
~ A is called the negation of A.
2. The wff (A B) is True if A and B are both True; otherwise, the wff A B is
False. The wff (A B) is called the conjunction of A and B.
3. The wff (A B) is true if at least one of A and B is True; otherwise, (A B) is
False. (A B) is called the disjunction of A and B.
4. The wff (A B) is False if A is True and B is False; otherwise, (A B) is True.
The wff (A B) is read as If A, then B, or A implies B. The symbol is
called implication.
5. The wff (A B) is True whenever A and B have the same truth values;
otherwise (A B) is False. The wff (A B) is read as A if and only if B.
Table 1.5
A
(i) T
(ii) T
(iii) F
B
T
F
T
~A
F
F
T
(A B)
T
F
F
(A B)
T
T
T
(iv) F
(A B)
T
F
T
(A B)
T
F
F
The above relations can be summarized by Table 1.5 given below.

The table may be read as follows:
Let the symbol T stand for True and the symbol F stand for False. Then, Row (i) is
interpreted as: if we assign T (i.e. True) to A and T to B then the truth values of (~ A),
(A B), (A B), (A B) and (A B) are respectively F, T, T, T, T.
Further row (iii), for example, is interpreted as:, if we assign truth-value F (False) to
A and T (True) to B then truth values of (~ A), (A B), (A B), ( A B) and ( A
B) are respectively T, F, T, T, F.
This table, shall be used to evaluate the truth values of a wff in terms of the truth
values of the atoms occurring in the formula.
Now, we discuss the issue, raised in Section 1.2, of how to check validity/invalidity of
an argument through formal means.
27
Introduction to A.I
Validity through Truth-Table.

(i) If I overslept, then I am late, i.e., symbolically
SL
(ii) I am not late, i.e., symbolically
~ L
To conclude
(iii) I did not oversleep, i.e., symbolically
~S
To establish the validity/Invalidity of the argument, consider the Truth-Table
S
F
F
T
T
L
F
T
F
T
SL
T
T
F
T
~L
T
F
T
F
~S
T
T
F
F
There is only one row, viz., first row, in which both the premises viz. S L and ~ L
are True. But in this case the conclusion represented by ~ S is also True. Hence, the
conclusion is valid.
Invalidity through Truth-Table
(i) If I overslept, then I am late
SL
(ii) I did not oversleep, i.e.,
~S
To conclude
(iii) I would not be late, i.e.,
~ L (invalid conclusion)
S
F
F
T
T
F
T
F
T
(S L) ~ S
T
T
F
T
~L
T
T
F
F
T
F
T
F
The invalidity of the argument is established, because, for validity last column must
contain True in those rows for which all axioms/premises are True. But in the second
row both S L and ~ S are True but ~ L is False
Ex. 1 Express the following statements in Propositional Logic.
a) If he campaigns hard, he will be elected.
b) If the humidity is high, it will rain either today or tomorrow.
c) Cancer will not be cured unless its cause is determined and a new drug for
cancer is found.
d) It requires courage and skills to climb a mountain.
Ex. 2: Let
P : He needs a doctor,
R : He has an accident,
U : He is injured.
28
Q : He needs a lawyer,
S : He is sick,
The Propositional
Logic
State the following formulas in English.

a)
c)
(S P) (R Q)
(P Q) R
b) P (S U)
d) (P Q) (S U)
2.6 INTERPRETATIONS OF FORMULAS

In order to find the truth value of a given formula G, the truth values for the atoms of
the formula are either given or assumed. The set of initially given/assumed values of
all the atomic formulas occurring in a formula say G, is called an interpretation of
the formula G. Suppose that A and B are two atoms and that the truth values of A and
B are T and F respectively. Then, according to third row of Table 1.5, when A is F
and B is T we find that the truth values of (~A), (A B),
(A B), (A B), and (A B) are T, F, T, T and F, respectively. By developing a
Truth-table of a(ny) formula, its truth value can be evaluated in terms of its
interpretation, i.e., in terms of the truth values associated with the constituent atoms.
Example
Consider the formula
G : ((A B) (R (~ S))).
(Please note that the string, in this case G, before the symbol :, is the name of the
formula which is the name of the string of symbols after :. Thus, G is the name of the
formula ((A B) (R (~ S))).
The atoms in this formula are A, B, R and S. Suppose the truth values of A, B, R, and
S are given as T, F, T and T, respectively. Then (in the following and elsewhere also,
if there is no possibility of confusion, we use T for True and F for False.)
(A B) is F since B is F;
(~S) is F since S is T;
(R (~ S)) is F since R is T and (~S) is F; and hence,
(A B) (R (~S)) is T since (A B) is F (and (R (~S)) is F, which
does not matter).
Note: In view of the fact that when ( A B) is F, the truth-value of
(A B) Any Formula
must be T and, hence, we need not compute the value of (R (~ S)).
Therefore, the formula G is T if A, B, R, and S are assigned truth values T, F, T and T,
respectively.
The assignment of the truth values T, F, T, T to A, B, R, S, respectively, is called an
interpretation of the formula G. Since, each one of A, B, R, and S can be assigned
one of the two values, viz., either T or F, there are 24 = 16 possible interpretations of
the formula G. In Table 1.6, we give the truth values of the formula G under all these
16 interpretations.
The above procedure may be repeated to find truth value of any formula from any
interpretation, i.e., from any assignment to the atomic formulas occurring in the given
formula.
29
Introduction to A.I
Table 1.6 Truth Table of (A B (R ( ~ S)
~S
(A B)
(R (~S))
(A B) (R
( ~ S)
T
T
T
T
T
T
T
T
F
F
F
F
F
F
F
F
T
T
T
T
F
F
F
F
T
T
T
T
F
F
F
F
T
T
F
F
T
T
F
F
T
T
F
F
T
T
F
F
T
F
T
F
T
F
T
F
T
F
T
F
T
F
T
F
F
T
F
T
F
T
F
T
F
T
F
T
F
T
F
T
T
T
T
T
F
F
F
F
F
F
F
F
F
F
F
F
F
T
T
F
F
T
T
F
F
T
T
F
F
T
T
F
F
T
T
F
T
T
T
T
T
T
T
T
T
T
T
T
A table, such as given above, that displays the truth values of a formula G for all
possible assignments of truth values to atoms occurring in G is called a Truth table
of G.
NOTATION: If A1,.An are all the atoms in a formula, it may be more convenient to
represent an interpretation by a set (m1,.mn), where mi is either Ai or ~Ai. mi is
written as Ai if T is assigned to Ai. But mi is written as ~ Ai if F is assigned to Ai.
For example, the set {A, ~B, ~R,S} represents an interpretation of a formula in which
A, B, R, and S are the only atoms and which are, respectively, assigned T, F, F, and T.
We will use the notation throughout.
Ex. 3: Construct a truth table for the formula.
P: (~ A B) (~ (A ~ B))
2.7 VALIDITY AND INCONSISTENCY OF

PROPOSITIONS
It may noted that in Section 1.2, we discussed the concept of valid Argument. Here,
we study formulas or propositions. Next, we shall consider wff that are true under
all possible interpretations and wff that are false under all possible interpretations.
Example
Let us consider the wff
G : (((A B) A) B).
The formula G has 22 = 4 possible interpretations in view of the fact it has two atoms
viz A and B. It can be easily seen from the following table that the wff G is True
under all its interpretations. Such as a wff which is True under all interpretation is
called a valid formula (or a tautology).
30
Truth Table of (((A B) A) B)
A
T
T
F
F
B
T
F
T
F
(A B)
T
F
T
T
(A B) A
T
F
F
F
((A B) A) B
T
T
T
T
The Propositional
Logic
Consider another formula

G : ((A B) (A ~ B))
The truth table of the formula G given below shows that G is False under all its
interpretations. Such a formula which is False under all interpretations is called an
inconsistent formula (or a contradiction).
Truth Table of (A B) (A ~ B)
A
T
T
F
F
B
T
F
T
F
~B
F
T
F
T
(A B)
T
F
T
T
(A ~ B)
F
T
F
F
((A B) (A ~ B)
F
F
F
F
Next, we formally define the concepts discussed above.

Definition: A formula is said to be valid if and only if it is true under all its
interpretations. A formula is said to be invalid if and only if it is not true under at
least one interpretation. A valid formula is also called a Tautology. A formula is
invalid if there is at least one interpretation for which the formula has a truth value
False.
Definition: A formula is said to be inconsistent (or unsatisfiable) if and only if it is
False under all its interpretations. A formula is said to be consistent or satisfiable if
and only if it is not inconsistent. In other words, a formula is consistent if there is at
least one interpretation for which the formula has a truth value true.
From the definitions given above, it is easily seen that
(i)
(ii)
(iii)
(iv)
(v)
A formula is valid if and only if its negation is inconsistent.

A formula is invalid if and only if there is at least one interpretation under
which the formula is false.
A formula is consistent if and only if there is at least one interpretation under
which the formula is true.
If a formula is valid, then it is consistent, but not vice versa. (example given
below)
If a formula is inconsistent, then it is invalid, but not vice versa. (example given
below)
Definition: If a formula P is True under an interpretation I, then we say that I

satisfied P, or P is satisfied by I. If a formula P is False under an interpretation I,
then we say that I falsifies P or P is falsified by I.
As for an example, the formula (A (~B)) is satisfied by the interpretation {A, ~ B}
i.e, by taking A as T and B as F, but is falsified by the interpretation {A, B} i.e., when
A is taken as T and B is taken as T. An interpretation I that satisfies a formula P, is
called a model of the formula F.
31
Introduction to A.I
Examples:
(i) A Valid Formula:
(a) Even True is a wff which is always True and, hence, True is a valid formula.
(b) G1: A (~A) is True for all its interpretations. As G1 has only one atom viz. A,
terefore, it has only two interpretations. Let one interpretation of G1 be : A is
True. But then G1 assumes the value (True (~ True)) = True. The other
interpretation of G1 is : A is False. Then G1 assumes the value (False ~ False) =
True.
(ii) Consistent (True for at least one interpretation) but not valid Formula (i.e. is
invalid, i.e., False for at least one interpretation):
(a) The simplest example of such a formula is the formula G2: A. Then, for the
assignment A as True, G2 is True. Therefore G2 is consistent. On the other
hand, the interpretation of G2 with A as False, makes G2 false. Therefore, G2:
A is not valid.
(b) Both G3 : A B and G4 : A B are consistent but not valid. Both G3 and G4
are True under the assignment A as True and B as True. On the other hand,
both are False under the interpretation A as False and B as False.
(iii) Invalid (False for at least one interpretation) but not inconsistent (not False
for all interpretations): Any one of the examples in (ii) above
(iv) Inconsistent formula (i.e., which is false for all interpretations)
(a) Even False is a wff; which is always False, and hence is inconsistent.
(b) G5 : A (~A) is False, for all interpretations of G5. Actually, there are only
two interpretations of G5. One is : A is True. The other is : A is False. In both
cases G5 is False.
It will be shown later that the proof of the validity or inconsistency of a formula is a
very important problem. In the propositional logic, since the number of interpretations
of a formula is finite, one can always decide whether or not a formula in the
propositional logic is valid (inconsistent) by exhaustively examining all of its possible
interpretations.
Ex. 4: For each of the following formulas, determine whether it is valid, inconsistent,
consistent or some combination of these.
(i) E: ~ (~A) B
(ii) G: (A B) (~ B ~ A)
(iii) H: (A ~ A) (A B ) ( ~ A)
(iv) J: (A B) (~ A) ( B ~ B)
2.8 EQUIVALENT FORMS IN THE

PROPOSITIONAL LOGIC (PL)
Definition: Logically Equivalent Formulas: Two formulas G1 and G2 are said to be
(logically) equivalent if for each interpretation i.e., truth assignment to all the atoms
that occur in either G1 or G2; the truth values of G1 and G2 are identical. In other
words, for each interpretation, G1 is True if and only if G2 is True. And, for each
interpretation, G1 is False if and only if G2 is False.
32
As will be clear later, it is often necessary to transform a formula from one form to
another, especially to a normal form. This is accomplished by replacing a formula in
the given formula by a formula equivalent to it and repeating this process until the
desired form is obtained.
The Propositional
Logic
Example
We can verify that the formula E: ~ (A B) is equivalent the formula G: to A ~ B
by examining the following truth table. The corresponding values in the last two
columns are identical.
Table Joint Truth table of ~ (A B) and (A ~ B)
A
T
T
F
F
B
T
F
T
F
~B
F
T
F
T
(A B)
T
F
T
T
~(A B)
F
T
F
F
A~B
F
T
F
F
Solutions of problems using symbolic logic can be simplified, if we can simplify

involved formulas by some equivalent simpler formulas given in table below. These
equivalences can be verified by using truth tables.
Table of Equivalences of PL
(1.1)
(1.2)
(1.3)(a)
(1.4)(a)
(1.5)(a)
(1.6)(a)
(1.7)(a)
(1.8)(a)
(1.9)
(1.10)(a)
E G = (E G) (G E)
EG=~EG
E G = G E;
(E G) H = E (G H);
E (G H) = (E G) (E H);
E False = E;
E True = True
E ~ E = True;
~ (~ E) = E
~ (E G) = ~ E ~ G;
(b) E G = G E
(b) (E G) H = E (G H)
(b) E (G H) = (E G) (E H)
(b) E True = E
(b) E False = False
(b) E E = E
(b) ~ (E G) = ~ E ~ G
In the table given above, True denotes the fact that the wff is True under all
interpretations and False denotes the wff that is False under all interpretations.
Laws (1.3a), (1.3b) are often, called commutative laws; (1.4a), (1.4b) associative
laws; (1.5a), (1.5b), distributive laws: and (1.10a), (1.10b), De Morgans laws.
2.9 NORMAL FORMS

Some Definitions: A clause is a disjunction of literals. For example, (E ~ F ~ G)
is a clause. But (E ~ F ~ G) is not a clause. A literal is either an atom, say A, or
its negation, say ~ A.
Definition: A formula E is said to be in a Conjunctive Normal Form (CNF) if and
only if E has the form E : E1 . En, n 1, where each of E1,., En is a
disjunction of literals.
Definition: A formula E is said to be in Disjunctive Normal Form (DNF) if and only
if E has the form E: E1 E2 .En, where each Ei is a conjunction of literals.
Examples: Let A, B and C be atoms. Then F: ( ~ A B ) (A ~ B ~ C) is a
formula in a disjunctive normal form.
33
Introduction to A.I
Example: Again G: (~ A B) (A ~ B ~ C) is a formula in Conjunctive Normal

Form, because it is a conjunction of the two disjunctions of literals viz of (~ A B)
and (A ~ B ~ C)
Example: Each of the following is neither in CNF nor in DNF
(i)
(ii)
(~ A B) (A ~ B C)
( A B) ( ~ B ~ A)
Using table of equivalent formulas given above, any valid Propositional Logic
formula can be transformed into CNF as well as DNF.
The steps for conversion to DNF are as follows
Step 1: Use the equivalences to remove the logical operators and :
(i) E G = (E g) (G E)
(ii) E G = ~ E G
Step 2 Remove ~s, if occur consecutively more than once, using
(iii) ~ (~E) = E
(iv) Use De Morgans laws to take ~ nearest to atoms
(v) ~(E G) = ~ E ~ G
(vi) ~(E G) = ~ E ~ G
Step 3 Use the distributive laws repeatedly
(vii) E (G H) = (E G) (E H)
(viii) E (G H) = (E G) (E H)
Example
Obtain a disjunctive normal form for the formula ~ (A (~ B C)).
Consider A (~B C) = ~ A (~B C)
Hence, ~ (A (~ B C)) = ~ (~ A (~B C))
= ~ (~ A) (~ (~ B C))
= A (B (~ C))
= (A B) (A (~ C))
(Using (E F) = (~ E F))
(Using ~ ( E F) =
~ E ~ F)
(Using ~ (~ E) = E and
~ (E F ) = ~ E ~ F
(Using E (F G) =
(E F) (E G))
However, if we are to obtain CNF of ~ A ( (~ B C)), in the last but one step, we
obtain
~ (A (~ B C)) = A (B ~ C), which is in CNF, because, each of A and
( B ~ C) is a disjunct.
Example: Obtain conjunctive Normal Form (CNF) for the formula: D (A
(B C))
34
Consider
D (A (B C))
(using E F = ~ E F for the inner implication)
= D (~ A (B C)) (using E F = ~ E F for the outer implication)
= ~ D (~ A (B C))
= ( ~ D ~ A) (B C) (using Associative law for disjunction)
= (( ~ D ~ A B) (~ D ~ A C)
The Propositional
Logic
The last line denotes the conjunctive Normal Form of D (A (B C))

(using distributivity of over )
Note: If we stop at the last but one stop, then we obtain (~ D ~ A) (B C) = ~ D
~ A (B C) is a Disjunctive Normal Form for the given formula: D (A (B
C))
Ex. 5: Transform the following into disjunctive normal forms.
(i) ~ (A ~ B) (S T)
(ii) (A B) R
Ex. 6: Transform the following into conjunctive normal forms.

(i) (A B) R
(ii) (~A B) (A ~ B)
Ex. 7: Verify each of the following pairs of equivalent formulas by transforming
formulas on both sides of the sign = into the same normal form.
(i) (A B) (A B) = (~ A B) (B A)
(ii) A B (~A ~ B) = ~ A ~ B (A B)
2.10 LOGICAL DEDUCTION

Definition: A formula G is said to be a logical consequence of given formulas E1,,
En (or G is logical derivation of E1,.E2) if and only for any interpretation I in
which E1 E2 En is true, for the interpretation I, G is also true. The proposition
E1, E2,En are called axioms/premises of G.
Next, we state without proof two very useful theorems for establishing logical
derivations:
Theorem 1: Given formulas E1,, En and a formula G, G is a logical derivation of
E1,.,En if and only if the formula ((E1 .En) G) is valid, i.e., True for all
interpretations of the formula.
Theorem 2: Given formulas E1,,En and a formula G, G is a logical consequence or
derivation of E1,En if only if the formula (E1 .. En ~ G) is inconsistent, i.e.,
False for all interpretations of the formula.
The above two theorems are very useful. They show that proving a particular
formula as a logical consequence of a finite set of formulas is equivalent to
proving that a certain single but related formula is valid or inconsistent.
Note: Significance of the above two theorems lies in the fact that logical consequence
relates two formulas, where as validity/inconsistency is only about one formula. Also,
there are a number of well-known methods, including truth-table method, for
35
Introduction to A.I
establishing inconsistency/validity of a formula. Thus, formula G logically follows

from a given set of formulas, we check validity of single formula. And, for checking
validity of a single formula, we already have some methods including Truth-table
method.
Definition: If the formula G is a logical consequence of the formula E1,.En, then the
single formula ((E1 . En) G) is called a theorem, and G is also called the
conclusion of the theorem.
There are at least three alternative methods of establishing formula G as a logical
consequence of given formulas E1, E2, .En.
According to one of these methods, through truth table or otherwise, it should be
established that for any interpretation for which each of E1, .En, is true then for that
interpretation G must be true.
According to second method, using Theorem 1, we should show that the formula:
(E1 E2 .. En) G
is valid, i.e., True for each of its interpretations. Again validity can be shown either
through a truth table or otherwise.
The last of the three methods uses Theorem 2. According to this method, in order to
show, G as a logical consequence of E1, E2,En, it should be established that the
formula (E1 E2 .. En ~ G) is inconsistent, i.e., is False under all its
interpretations. Next, we apply these methods through an example.
Example: We are given the formulas
E1 : (A B), E2 : ~B , G : ~ A
We are required to show that G is a logical consequence of E1 and E2.
Method 1: From the following Table, it is clear that whenever E1: A B and
E2: ~ B both are simultaneously True, (which is true only in the last row of the table)
then G: ~ A is also True. Hence, the proof.
A
T
T
F
F
B
T
F
T
F
AB
T
F
T
T
~B
F
T
F
T
~A
F
F
T
T
Method 2: We prove the result by showing the validity of E1 E2 G, i.e., of ((A

B) ~ B) ~ A by transforming it into a conjunctive normal form.
(A B) ~ B) ~ A = ~ (( A B) ~ B) ~ A (using E F = (~ E F))
= ~ (( ~ A B) ~ B) ~ A
= ~ ((~ A ~ B) (B ~ B)) ~ A
= ~ ((~ A ~ B) False) ~ A
= ~(( ~ A ~ B)) ~ A (using De Morgans Laws)
= (A B) ~ A =
= (B A) ~ A
= B (A ~ A)
= B True
36
The Propositional
Logic
= True (always)
Thus, ((A B) B) ~ A is valid.
Ex. 8: Using Truth Table show that G is a logical consequence of E1 and E2
where E1 : (A B), E2 : ~B , G : ~ A, by establishing validity of the formula (E1 E2
G).
Ex. 9: Use (i) the truth table technique (ii) reduction to DNF/CNF to show that
(A B) ~ B A is inconsistent which, in turn proves that ~ A is a logical
consequence of (A B) and ~ B.
2.11 APPLICATIONS
Next, we discuss some of the applications of Propositional Logic.
Example
Suppose the stock prices go down if the interest rate goes up. Suppose also that most
people are unhappy when stock prices go down. Assume that the interest rate goes up.
Show that we can conclude that most people are unhappy.
To show the above conclusion, let us denote the statements are as follows:
A : Interest rate goes up,
S : Stock prices go down
U : Most people are unhappy
The problem has the following four statements:
1)
2)
3)
4)
If the interest rate goes up, stock prices go down.

If stock prices go down, most people are unhappy.
The interest rate goes up.
Most people are unhappy. (to conclude)
The above-mentioned statements are symbolised as,

(1) A S
(2) S U
(3) A
(4) U. (to conclude)
In order to establish the conclusion, we should show that (4) is logical consequence
of (1) , (2) and (3). For this purpose, we show that (4) is true whenever (1) (2)
(3) is true.
We transform ((A S) (S U) A) (representing (1) (2) (3)) into a normal
form:
((A S) (S U) A) = ((~A S) (~S U) A)
= (A (~A S) (~ S U))
(by using E F =
~ E F)
(by using E F =
F E, (to bring the
last clause A in the
beginning)
37
Introduction to A.I
= (((A ~A) (A S)) (~ S U)) (by using associative

laws and then using
distributivity of
A over the next
disjunct (~ A S))
= ((False (A S)) (~ S U))
(using False
E = E)
= (A S ) ( ~ S U)
= (A S ~ S) (A S U)
= (A False) (A S U)
(using A False =
False)
= False (A S U)
=ASU
Therefore, if ((A S) (S U ) A) is true, then (A S U) is true. Since
(A S U) is true then each of A, S, and U is true, we conclude that U is true. Hence,
U is a logical consequence of 1), 2) and 3) given above.
Ex. 10:Given that if the Parliament refuses to enact new laws, then the strike will not
be over unless it lasts more than one year and the president of the firm resigns, will
the strike not be over if the Parliament refuses to act and the strike just starts?
2.12 SUMMARY
In this unit, to begin with, we discuss what is Symbolic Logic and why it is it is
important to study it. The subject matter of symbolic logic consists of arguments,
where an argument consists of a number of statements one of which is called
the conclusion and is supposed to be logically drawn from the others. Each one of the
other is called a premise, To be more specific, the subject of Symbolic Logic is the
study of how to develop tools and technique to draw correct conclusions from a given
set of premisses or to verify whether a conclusion is correct or not. A conclusion is
correct in the sense: Whenever all the premisses are True then conclusion is
necessarily True. An argument with correct conclusion is called a valid argument.
Next, a sound argument is defined as a valid argument in which premises also have to
be True.
(in some world).
In this unit, we study only a specific branch of symbolic logic, viz. Propositional
Logic (PL).
Next, we discuss how a statement, also called a well-formed formula (wff) and also a
Proposition, which is the basic unit of an argument in PL, is appropriately denoted
and how it is interpreted, i.e., how a wff is given meaning. The meaning of a wff in
PL is only in terms of True or False. The wffs are classified as valid, invalid,
consistent and inconsistent.
Then tools and techniques in the form of Truth-table, logical deduction, normal forms
etc are discussed to test these properties of wffs and also to test validity of arguments.
Finally a number of applications of these concepts, tools and techniques of PL are
used to solve problems that involve logical reasoning of PL systems.
2.13 SOLUTIONS/ANSWERS
38
Ex. 1
(a) Let H: He campaigns hard ; E: He will be elected
Then the statement becomes the formula:
HE
(b) Let H: The Humidity is high, RTY: It will rain today
RTW: It will rain tomorrow.
Then
H RTY RTW
(c) Let C: Cancer will be cured
D: Cancers cause will be determined
F: A new drug for cancer will be found
(~ C) (D F). This formula may also be written as:
CDF
(d) Let C: One has courage
S: One has skill
M: One climbs mountain
MCS
Ex 2: (a) If he is sick then he needs a doctor, but, if he has an accident then he needs a
lawyer
(b) If One requires a doctor then one must be either sick or injured.
(c) If he needs both a doctor and a lawyer then he has an accident.
(d) One requires a doctor and also a lawyer if and only if one is sick and also
injured.
The Propositional
Logic
Ex. 3:
(i) Truth table of the formula: P: (~ A B) ( ~ (A ~ B)) is as given below.
A
T
T
F
F
B
T
F
T
F
Ex. 4:
(i)
(ii)
(iii)
(iv)
~A
F
F
T
T
~B
F
T
F
T
~A B
T
F
T
T
A~B
F
T
F
F
~ (A ~B)
T
F
T
T
P
T
F
T
T
Consistent but not valid, because, for For B as T and A as F, the formula
is T. But, for A as T and B as F the formula is F.
It can be easily that ~ B ~ A has same truth-value as (A B) for any
interpretation. Therefore, in stead of the given formula, we may consider
the formula
(A B) (A B)
which can be further written as P P, writing (A B) as P. Even P
can be written as P P P (A B), The last formula is F when F and
A is T. The formula is T when A is F and B is T. Hence, the formula is
neither valid nor inconsistent.
Therefore, the formula is consistent but not valid
For all truth assignments to A and B, L. H.S. of the formula is always T
and R. H.S. is always F. Hence the formula is inconsistent, i.e., always F
The L. H. S. of the given formula is F under all interpretations. Hence, the
formula is T under all interpretation. Therefore, the formula is valid.
Ex. 5: (i) Removing , we get

~ ( A ~ B) (~ S T)
Taking ~ inside we get
(~ A B) (~ S T) (using De Morgans Law)
Using distributivity of over we get
(~ A B ~ S) (~ A B T)
which is the required form
39
Introduction to A.I
(ii) Removing outer we get

~ (A B ) R
Removing the other we get
~ (~ A B) R
Taking ~ inside, we get
(A ~ B) R,
which is the required form
Ex. 6:
(i) Using distributive law in the last formula of 5 (ii) above, we get
(A R) (~ B R)
which is the required CNF
(ii) Using Left distributivity of over we get
(( ~ A B) A ) ( ~ A B) ~ B)
Using Right distributivity inside each pair of parenthese of over we get
(( ~ A A) (B A) (( ~ A ~ B) (B ~ B))
Using ~ A A = T = B ~ B, we get
( T ( B A)) (( ~ A ~ B) T)
which is equivalent to
( B A ) (( ~ A ~ B) = (A B) (~ A ~ B)
is the required CNF.
Ex. 7: (i) Consider L.H.S
Removing inner on L. H.S., we get
(~ A B) (A B)
removing the other
= ~ ( ~ A B) (A B)
Using De Morgans Laws, we get
= (~ (~A) (~ B)) (A B)
= (A ~ B) (A B)
which is in DNF
For R.H.S, removing the two implications, we get
(~ (~ A ) B) (~ B A)
= (A B) (~ B A)
(which is in CNF, but we require DNF)
Using Left distributivity of over , we get
= (( A B) ( ~ B)) (( A B) A)
Using Right distributivity of over , we get
= (( A ~ B) (B ~ B)) ( ( A A) (B A))
Using B ~ B = F
AA=A
And P F = P we get
= (A ~ B) ( A (B A))
= (A ~ B) (A) = (A ~ B) (A T)
= ( A ~ B) (A (B ~ B))
= (A ~ B) (A ~ B) (A B)
Using P P = P, we get
= (A ~ B) (A B)
40
(i)
(ii)
(ii) R.H.S Applying associative laws, we get

( ~ A ~ B) (A B)
Using left distributivity of over we get
= (( ~ A ~ B) A) ((~ A ~ B) B)
Again using associativity of and using ~ A A = F = ~ B B we get
The Propositional
Logic
R.H. S. = F
Consider L.H.S, applying associativity of , we get
= (( A B) (~ A ~ B )),
using left distributivity and commutativity of we get
= (( A B) ~ A) (( A B) ~ B)
Using associativity of and using A ~ A = F = B ~ B
= (B F) ( A F)
Using A F = F = B F
=F
Ex. 8: The following table shows that ((A B) ~ B) ~ A is true in every
interpretation. Therefore ((A B) ~ B) ~ A is valid and according to the First
theorem, ~ A is a logical consequence of (A B) and ~ B.
Truth Table of ((A B) ~ B) ~ A
A
T
T
F
F
B
T
F
T
F
AB
T
F
T
T
(AB) ~ B
F
F
F
T
~B
F
T
F
T
~A
F
F
T
T
(AB) ~ B) ~ A
T
T
T
T
Ex. 9: (i) From the following table, ((A B) ~ B A) being False for all
interpretations, is inconsistent.
Truth Table of (A B) ~ B A
A
T
T
F
F
B
T
F
T
F
AB
T
F
T
T
~B
F
T
F
T
(A B) ~ B A
F
F
F
F
(ii) Prove the inconsistency of E1 E2 ~ G, i.e., of (A B) ~ B A by

transforming, into a disjunctive normal form:
(A B) ~ B A = (~ A B) ( ~ B A )
= (~ A ~ B A) (B ~ B A) (Distributive Law)
= (~ A A ~ B) (F A)
= False False = False
Thus (A B) ~ B A is inconsistent.
Ex. 10:
Let us symbolize the statements in the problem state of above as follows:
A:
The Parliament refuses to act.
B:
The strike is over.
R:
The president of the firm resigns.
S:
The strike lasts more than one year.
Then the facts and the question to be answered in the problem can be symbolized as:
E1: (A (~ B (R S))) represents the statement If the congress refuses to enact
new laws, then the strike will not be over unless it lasts more than one year and the
president of the firm resigns.
E2 : A, represents the statement The congress refuses to act, by and
41
Introduction to A.I
E3: ~ S represent the statement The strike just starts.

E4: ~ B (to be concluded)
Ex. 10: We solve the problem by showing that the formula P: ((A (~ B (R S)))
A ~ S) ~ B is valid by two methods: (i) by reducing to CNF/DNF
(ii) by constructing truth-table of the formula.
Methods (i) Removing the two occurrences of , we get
P = ~ (( ~ A (~ B (R S))) A ~ S) ~ B
Using De Morgans Laws, we get
= (~ ((~ A) (~ B (R S))) ~ A ~ ~ S) ~ B
= (A ( ~ ~ B ~ ( R S))) ~ A S) ~ B
= (A (B ~ ( R S))) ~ A S) ~ B
P = (A (B (~ R ~ S))) ~ A S ~ B .. (i)
Consider the case R is assigned value F
Then the formula P becomes
(A (B (~ F ~ S))) (~ A ~ B S)
= ((A B) T) (~ (A B) S)
= (A B) (~ (A B) S)
By denoting A B by H we get P = H (~ H S) = T whether (A B) is T or F
Consider the case when R is assigned T
Then the formula P given by (i) becomes
(A (B (~ T ~ S) (~ A ~ B S) (using De Morgan Laws)
= ((A B) ~ S) (~ (A B) S)
= (( A B) ~ S) (~ (A B ~ S))
Denoting (A B ~ S) by K we get
P=K~K=T
Hence P is valid. Hence, the proof.
Method (ii)
The solution of the problem lies in showing that ~ B logical follows from E1, E2, and
E3. This is equivalent to showing that P: ((A (~B (R S ))) A ~ S) ~ B is
a valid formula. The truth values of the above formula under all the interpretations are
shown in given table
A
T
T
T
T
T
T
T
T
F
F
F
F
F
F
F
F
42
B
T
T
T
T
F
F
F
F
T
T
T
T
F
F
F
F
R
T
T
F
F
T
T
F
F
T
T
F
F
T
T
F
F
S
T
F
T
F
T
F
T
F
T
F
T
F
T
F
T
F
~B
F
F
F
F
T
T
T
T
F
F
F
F
T
T
T
T
~ B (R S)
T
F
F
F
T
T
T
T
T
F
F
F
T
T
T
T
E1
E2
E3
~B
T
T
T
T
T
T
T
T
F
F
F
F
F
F
F
F
T
T
T
T
F
F
F
F
T
T
T
T
F
F
F
F
T
T
F
F
T
T
F
F
T
T
F
F
T
T
F
F
T
F
T
F
T
F
T
F
T
F
T
F
T
F
T
F
T
F
F
F
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
F
F
F
F
F
F
F
F
F
T
F
T
F
T
F
T
F
T
F
T
F
T
F
T
F
F
F
F
T
T
T
T
F
F
F
F
T
T
T
T
~ B (R
S)
T
F
F
F
T
T
T
T
T
F
F
F
T
T
T
T
E1
T
F
F
F
T
T
T
T
T
T
T
T
T
T
T
T
(E1 E2 E3)
~B
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
T
The Propositional
Logic
Under all interpretations formula is True. Hence, the formula P a valid formula. ~ B is
a logical consequence of E1, E2 and E3. Hence, the The strike will not be over is a
valid conclusion.
2.14 FURTHER READINGS

(In the order from elementary to advanced)
1.
McKay, Thomas J., Modern Formal Logic (Macmillan Publishing Company,

1989).
2. Gensler, Harry J. Symbolic Logic: Classical and Advanced Systems (Prentice
Hall, 1990).
3. Klenk, Virginia Understanding Symbolic Logic (Prentice Hall 1983)
4. Copi Irving M. & Cohen Carl, Introduction to Logic, IX edition, (Prentice Hall of
India, 2001).
5. Carroll, Lewis, Symbolic Logic & Game of Logic (Dover Publication, 1955).
6. Wells, D.G., Recreations in Logic (Dover Publications, 1979).
7. Suppes Patrick, Introduction to Logic (Affiliated East-West Press, 1957).
8. Getmanova, Alexandra, Logic (Progressive Publishers, Moscow, 1989).
9. Crossely, J.N. et al What is Mathematical Logic? (Dover Publications, 1972).
10. Mendelson, Elliott: Introduction to Mathematical Logic (Second Edition) (D.Van
Nostrand Company, 1979).
43

Block 1

Uploaded by

Copyright:

Available Formats

Block 1

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Block 1

Uploaded by

Copyright:

Available Formats

UNIT 1 INTRODUCTION TO

INTELLIGENCE AND ARTIFICIAL

Artificial Intelligence & Human Intelligence

1.2 SOME SIMPLE DEFINITIONS OF A.I.

1.3 DEFINITION BY ELIANE RICH

information storage, and

Mohan saw the boy in the park with a telescope.

For a long time..

He waited in the doctors room for a long time.

Comment 2, Definition 1: In addition to the advantage that human beings have in

1.4 DEFINITION BY BUCHANIN AND

Comments/Explanations 1, Definition 2: Symbolic processing vs numberic

Knowing (exactly) the distances of various vehicles from the path to be

1.5 ANOTHER DEFINITION BY ELAINE RICH

Thus, complexity of the straight-forward method of multiplication of two n x n

Comment 2, Definition 3: Role of knowledge in solving hard problems:

1.6 DEFINITION BY BARR AND FEIGENBAUM

understanding, because of which we are not able to describe the phenomenon of

to respond to situations very flexibly;

Have mental attitudes (beliefs, desires and intentions)

perceptual visibility of the representation, we always use the representation for

Intelligence is a property of advanced information processing systems, which not

survivability of the system is enhanced

In stead of attempting to categorize most of the information processing systems as

1.7 DEFINITION BY SHALKOFF

intelligent computer systems, i.e., systems that exhibit the characteristics we

1.9 FURTHER READINGS/REFERENCES

Rich E. & Knight K. (1991). Artificial Intelligence.

UNIT 2 THE PROPOSITIONAL LOGIC

The proposition: The sun rises in the west, is False,

then the sentence:

differentiate between an expression that is a well-formed formula wff of PL and

find the truth-value, or meaning, of a wff of PL and should be able to explain

2.2 LOGICAL STUDY OF VALID AND SOUND

(i) If you overslept, you will be late

If moon is made of green cheese then 2 + 2 = 4

Example of Invalid Argument

(ii) you are late.

2.3 NON-LOGICAL OPERATORS

2.4 SYNTAX OF PROPOSITIONAL LOGIC

Some pairs of parentheses may be dropped, for simplification. For example,

2.5 SEMANTICS/MEANING IN PROPOSITIONAL

The above relations can be summarized by Table 1.5 given below.

Validity through Truth-Table.

State the following formulas in English.

2.6 INTERPRETATIONS OF FORMULAS

2.7 VALIDITY AND INCONSISTENCY OF

Truth Table of (((A B) A) B)

Consider another formula

Next, we formally define the concepts discussed above.

A formula is valid if and only if its negation is inconsistent.

Definition: If a formula P is True under an interpretation I, then we say that I

2.8 EQUIVALENT FORMS IN THE

Solutions of problems using symbolic logic can be simplified, if we can simplify

2.9 NORMAL FORMS

Example: Again G: (~ A B) (A ~ B ~ C) is a formula in Conjunctive Normal

The last line denotes the conjunctive Normal Form of D (A (B C))

Ex. 6: Transform the following into conjunctive normal forms.

2.10 LOGICAL DEDUCTION