Thinking Convergent Divergent

gudgeonmaniacalAI and Robotics

Feb 23, 2014 (7 years and 5 months ago)

280 views

What is Thinking?

It is the cognitive rearrangement of
information from the environment & symbols
stored in the LTM.

It is the set of cognitive processes that mediate
stimuli & responses.

The simplest type of thinking is simple
association.
Cognitive (Level of Processing) Approach
Semantic Memory
(Concepts & Schemas)
Verbal Image Memory
(Objects & Locations)
Verbal Graphical Memory
(Letters & Wor
ds)
Sensory-Motor Segments
(Lines & Angles)
Visual Features
(Spots)
Motor Features
(Muscle contractions)
Eye
Hand
Cognition
Sensory
Motor
Input
Output

A symbol is an identity which stands for
an event or item of the world.

A word is a language symbol.
(e.g., ‘dog’
stands for a particular type of
animal)
Symbol
Thinking
(Is private in nature & involves symbols that
have different meanings of each person)
(Aimed at problem solving or creating
something new)
Autistic
Directed
Convergent
Divergent
Images

Mental recollection of a sensory
experience.

It occurs in every sense modality.

Visual imagery dominates the overall
imagery of a person.
Language & Thought

“Thinking is a subvocal
speech”
or a “silent talk”.
-
J.B. Watson
Linguistic Relativity
Linguistic Relativity
(Whorf)
•‘
T
h
e
l
a
n
g
ua
g
e people use, determines & shapes their
experience & thought process.’

Lan
g
ua
g
e can actuall
y
determine the thou
g
hts we
are capable of having.
Phonems

It
is
based
on
universal
sound
units
called
phonemes
(the basic sounds that make up
any language).

By
themselves
phonemes
are
meaningless
&
hence
to
not
play
a
role
in
thinking.
Language & Thought

At
present
researchers
largely
favour
the
primacy of thought over language.

It
is
held
that
language
is
a
relatively
late
invention
&
various
forms
operate
independent
of language.

A
n
interdependence
of
language
&
thought
is
also recognised.
Language & Thought

Researches
(Vernon,
1967;
Furth,
1971)
have
shown
that
deaf
children
with
little
verbal
language
ability score
in
the normal
range
on
standardized
tests
of
cognitive
performance.

Their
cognitive
&
thinking
abilities
develop
relatively normally.
Language & Thought

These
indicate
that
language
plays
little
or
no
role
in
the
thinking
&
cognition
development of the deaf.

Deaf learn visual-gestural
sign language.

They
generate
meaningful
language
by
using phonems.
Concept

Making the same response to a group of
objects having similar characteristics.

Concepts are mental categories for classifying
objects, people, or experience.
Concepts
Simple Concepts
(Defined by a single feature)
E.g., Red, Blue
Complex Concepts
(Defined by several common
properties)
E.g., Red square
Conjunctive Concepts
(Two or more features at
the same time)
Red square: Red as well as
square.
Disjunctive Concepts
(Any one of the several
attributes)
Either red or square or both
This is more difficult to learn.
Relational concepts
(Relationship between features)
A room with more doors than
windows
Concept Attainment
Generalization
Discrimination
It involves two-fold activities
Attribute discovery
Rule discovery
Prototype

The particular example of the concept that readil
y

comes to mind is the prototype of that concept.

A mental model containin
g
the most t
y
pical features
of a concept.
Prototype of bird in one’s mind.
Socio-cultural Origin of Concept

Problem-solving,
as
a
part
of
a
broader
category
of
planning
behaviour,
is
at
the
top
of
all the cognitive activities.
Socio-cultural Origin of Concept

Vygotsky
established
the
relationship
between
language
&
thought.
(…language
is
not
only
a
cognitive
tool
for
communication
but
its
use
has shaped our cultural evolution.)
Socio-cultural Origin of Concept

Luria
(1966)
studied
higher
cognitive
activities
from
a
neuropsychological perspective.

He
systematically
investigated
the
role of language, which provided a
direct
link
between
cognitive
models & socio-cultural aspects
of
the origin of human thinking.
Socio-cultural Origin of Concept

Luria
further
investigated
the
process
of
consciousness.

He
linked
consciousness
to
the
activity
of
the
frontal
lobes,
which
is
deeply
rooted
in
the socio-cultural history of the individual.
Socio-cultural Origin of Concept

Piaget says that thinking is
reflected
in
language
and
by
examining
a
person’s
language,
we
can
know
about his/her thinking.
Socio-cultural Origin of Concept

Chomsky
(1957)
became
the
psycholinguist
to
study
the
process of language in depth.

Theory
of
Universal
Grammar
Basic
language
structure
resides
internally
and
basic
patterns
are
innate
to
human
beings.

Chomsky
has
gone
as
far
as
to
implement
his
innatist
theory
on
other
disciplines
such
as
aesthetics,
concept
formulation,
science,
and
morality.

This
is
believed
so
by
Chomsky
because
if
there
were not a basic universal
"organ",
as
it is
called,
then
the
systems
of
understanding
which
humans
use
would
be
much
less
complex
and
vary
far
too
much
between
individuals.
Problem Solving

It is the process of moving from the given state to
the goal state of a problem.
Stages
Understanding the problem
Devising a plan
Carrying out the plan
Evaluation
Simple Decision Making

Decisions about physical differences are based on jnd
(Just
Noticeable Difference).

Decisions about symbolic differences are based upon
Semantic Congruity Effect.
Semantic Congruity Effect refers to the match (contiguity)
between stimuli, on one hand, and the dimension being
judged on the other.
It states that the decision will be faster when the dimension
being judged is congruent with the stimuli being evaluated.
Example, pick the smallest number from a set of numbers.

There
is
a
separate
psychological
dimension
that
influences
our
perception.
The
greater
the
d
ifference
between
two
stimuli,
the
easier
it
is
to
judge
them
as
different.
The
more dissimilar
two concepts
are, the
easier
it
i
s
to
judge
them
as
different.
When
asked
to
compare,
the
congruity
between
the
concepts
and
the
dimension
being
judged
also
affects
performance,
the semantic congruity effect.
•The
symbolic
distance
and
semantic
congruity
effect
works
for
numbers,
visual
imagery,
semantic
judgments
and
judgment based on geographical location.
Characteristics of Problem Solving
Goal directedness
Sequence of steps
Cognitive operations
Subgoal
decomposition
Problem Solving Strategies
Backward Search
Mean-End Analysis
Algorithms
Heuristics
Problem Solving Strategies:
Algorithms

Procedures
that
guarantee
solution
to
a
given problem.
(E.g., scientific formulas).

It leads to mechanical solution.
Problem Solving Strategies: Heuristics

Short-cut
rules
of
thumb
that
often
lead
to
quick
solutions.

They
are
based
on
past
experiences
with
these
problems & involve reasoning.
(E.g., Anagram problem or scrambled words)

It
requires
rearranging
the
letters
in
all
possible
combinations.
5 sets would lead to 120 combinations.
We
consider
only
those
letter
combinations
that
occur
frequently
&
rule
out
combinations
that
are
unlikely
in
ordinary English words.
Problem Solving Strategies:
Mean-End Analysis

The
problem
is
divided
into
a
number
of
sub-
problems
each
of
which
is
a
little
closer
to
the
devised goal.

This
reduces
the
differences
between
the
current state of the problem & its solution.

(E.g.,
Chess,
Working
through
mathematical
proofs, Computer programming, etc.)
Problem Solving Strategies:
Backward Search

Useful
when
the
goal
has
more
information than the given problem.
Factors affecting Problem Solving

Mental
set:
The
way
of
perceiving
certain
structure
or
situation.

R
igidity:
Use
a
response
that
worked
in
other
situations.

Functional
Fixedness:
Functions
of
a
particular
object
is
fixed
by
its
use
in
a
particular
way
just
before
the
problem is presented.
The
tendency
to
use
objects
or
concepts
in
ordinary
ways
during
problem
solving,
rather
than
use
them
in
creative or insightful ways.
Factors affecting Problem Solving
contd…
•Well defined & ill-defined problems
•Anxiety
•Incubation
•Brainstorming
Insight & Problem Solving by Analogy

Insight
&
problem
solving
by
analogy
are
on
the
positive sides of problem solving.

Insight
is
usually
thought
as
a
deep,
useful
understanding of the nature of something,
especially
a
difficult problem.

A
nalogy
is
a
relationship
between
two
similar
situations, problems, or concepts.

U
nderstanding
an
analogy
means
putting
the
two
situations
into
some
kind
of
alignment
or
relationship
so that the similarities and differences can be seen.
Multiconstraint
Theory

This is an overall theor
y
of analo
g
ical reasonin
g
&
problem solving proposed by Holyoak
& Thagard
(1997).

People are constrained b
y
three factors when the
y

try to use or develop analogies:
1.
Problem similarity
2.
Problem structure
3.
Purpose of the analogy
Problem similarity

There must be a reasonable de
g
ree of similarit
y

between the already-understood situation, the source
domain, and the tar
g
et domain (the current problem
being solved).
Problem Structure

People must establish a parallel structure between
the source & tar
g
et problems, so the
y
can map
elements from the source to comparable
elements in
the target.
Anderson’s (1983) ACT Model
{Adaptive Control of Thought}
Storage
Match
Retrieval
Execution
Encoding
Application
Production Memory
Declarative Memory
Working Memory
Performance
Outside World
Decision Making

It
is
a
kind
of
problem
solving
in
which
several
alternatives
are
presented
to
choose from.

While
making
complex
real
life
decisions
the
precise
likelihood
of
various
outcomes
are not known.

We
tend to
make our
own estimates
of the
probabilities, i.e., Subjective Probabilities.
Decision Making
Mathematical Model of Decision Process
(Subjectively Expected Utility)

Given a choice among alternatives, utility &
subjective probability are taken into account &
multiplied.

The alternative with the highest product is
chosen.
Heuristics in Decision Making
Heuristics of Representativeness

Whether
the
current
situation
is
a
representative
of
something
already
experienced.

In
the
representativeness
heuristics,
we
judge
the
likelihood
of
outcomes
based
on
how
representative
they
seem,
for
example
if
a
sequence
of
heads
&
tails
looks
like
a
random
sequence,
or
if
an
individual
matches
a
stereotype.
Heuristics in Decision Making
Availability

Some events are easier to imagine/ remember.

Easily
imagined/
remembered
events
are
l
ikely
to be frequently available.

Thus, the
ease with which
we
remember certain
things
help
us
in
making
subjective-probability
estimates.
Availability contd….

In
the
availability
heuristic,
people
judge
the
likelihood
of
events
by
how
easily
can
remember examples or instances.

These
judgments
can
therefore
be
biased
by
any
factor,
such
as
f
requency,
familiarity,
or
salience
and
vividness,
that
affects
the
information stored in memory.

In
combination
with
the
confirmation
bias,
this
kind of heuristic
may
explain how people
hold
&
strengthen prejudices.
Heuristics in Decision Making
Adjustment

W
e start with a certain sub
j
ective probabilit
y
& raise
or lower it depending on the circumstances.

Anchorin
g
(a bias: In estimatin
g
sub
j
ective
p
robabilit
y
the initial level which provides an anchor
that bias the estimate) may crop-up.
Heuristics in Decision Making

In
simulation
heuristic,
people
forecast
how
some outcome could have been different.

These
factors
are
influenced
by
how
easily
the
alternative outcomes can be imagined.

These
f
orecasts
are
based
on
informal
mental
models,
as
are
situations
like
predicting
the
path of a moving object.

Incompleteness
or
misconceptions
in
the
mental model lead to errors in reasoning.
Improving Problem Solving
1.
Increase domain knowledge
2.
Automate some components of the solution
3.
Follow a systematic plan
4.
Draw inferences
5.
Develop subgoals
6.
Work backward
7.
Search for contradictions
8.
Search for relations among problems
9.
Find a different problem representation
10.
If all else fails, try practice
Creative Thinking

It is characterised
by originality, novelty &
appropriateness.

The sudden appearance of new ideas is
called insight.
Stages of Creative Thinking
Preparation
Incubation
Illumination
Evaluation
Revision
Creative Thinking
Thinking
Convergent
Divergent
The thinker gathers
information relevant
to the problem & then
proceeds by using
problem solving
rules.
Variety of thoughts
about
a problem are
involved.
It includes autistic
thinking.
Personality Characteristics of Creative Thinkers

Prefer
complex
&
some
degree
of
apparent
imbalance
in phenomena.

Psychodynamically
complex
&
greater
personal
scope.

Independent in their judgments.

Self-assertive & dominant.

R
eject
suppression
as
a
mechanism
for
control
of
impulse.

More
origence
(Welsh,
1975,
has
given
74
characteristics of origence
personality dimension)
Reasoning

When logical principles are applied to thinking.
Categorical Syllogism
Formal / Aristotelian Syllogism
Major Premise: All businessmen
are cunning.
Minor Premise: All businessmen
are men.
Conclusion: All men are cunning
Conditional Syllogism
E.g., If you do well in exams
you will be promoted to the
next class.
Inductive Reasoning
Deductive Reasoning
Artificial Intelligence & Cognition

This is the branch of computer science that emulate
human cognitive functions.

Present da
y
Co
g
nitive
Science is a shared platform
of AI, Cognitive Psychology and Neurosciences.

It is assumed that human thinkin
g
can best be
emulated b
y
modelin
g
a machine after basic
neurological structures.

A Japanese scientist, Aizawa, is making a brain-like
computer.

He has taken real nerve cells intermingled with electronic
devices.

He has succeeded combining neurons with the semi
conducting compound indium tin oxide.

Under weak electrical stimulation cells do respond with
controlled growth.
Artificial Intelligence & Cognition:
Superbiology

Success in this area would open a new
dimension of interface between the nervous
system and prostheses.

A whole new area of developing and
implanting artificial organs controlled by the
nervous system would emerge.
Artificial Intelligence & Cognition:
Superbiology

Human perception is dependent upon external
si
g
nals such as li
g
ht, sound, molecular
composition,
pressure, etc.

These signals are detected/ sensed and transduced.

A large chunk of signals are sent to the brain.

The brain receives 4.3 x 106
b
its of information per
second from the visual system alone.
Artificial Intelligence & Cognition:
Perception
•A
n
a
l
y
sis of lines and pattern reco
g
nition is a crucial
task in visual perception.

Because of its strate
g
ic importance, AI and
p
erceptual processes are an important area of
research.
•A
c
o
m
p
l
e
x

g
eometric form is composed of simpler
forms. A computer can be made to anal
y
ze local
features of the object.
Artificial Intelligence & Cognition:
Perception

The system depends on a program that uses a number of
small templates that systematically passes over each object
in search of a match.

The pattern recognition systems uses hardware such as a
raster or a matrix of photoelectric cells.

Though transduced
binary codes (0: off/ light; 1: on/ dark)
and optical scanning are being used, a replicable human
information analysis used in identifying letters, words and
patterns still elude researchers.
Artificial Intelligence & Cognition:
Perception

AI alphanumeric recognition systems
were based
template concept. A pattern of letters & numbers
were stored in a computer which reads b
y
matchin
g

a pattern.

DYSTAL (DYnamically
STable
Associative
Learning)
acquires letters and letter sequences and
reco
g
nize them even when a part of the pattern is
present.
Artificial Intelligence & Perception:
Examples

Alkon’s
artificial network reco
g
nizes a pattern usin
g

the rules followed by the biological system.

When trained to reco
g
nize a pattern the receivin
g

sites participatin
g
in the reco
g
nition process receives
more “weight”
than the ones not participatin
g
. In
other words, their excitability increases.
Artificial Intelligence & Perception:
Examples

ELIZA
is one of the first program written by Joseph
Weizenbaum
(1966). This was programmed to respond to
certain key words that only transformed the original
sentence.

PARRY
was a program developed by Colby et al. (1972)
that simulated a paranoid patient.

NETtalk
was developed by Sejnowski
which reads
letters
and pronounces them aloud. The neural net simulation
model hundreds of units and thousands of their connection.
It reads by considering letters one by one and by scanning
three letters on either side for contextual information.
Artificial Intelligence & Language:
Examples
Artificial Intelligence & Language

Continuous speech reco
g
nition (CSR) s
y
stems are
programs that recognize and record natural speech.

NETtalk
and CSR pro
g
ram react reasonabl
y
and are
considered to have incorporated some form of
human understanding.

RKCP (Ray Kurzweil’s
C
y
bernetic Poet) is
a
program developed by Kurzweil
to creat
poems.

CAD (Computer-assisted desi
g
n) are bein
g
used in
architectural and industrial design.

EMI (Experiments in Musical Intelli
g
ence) is used
to record computer-generated music.

Computational devices are also bein
g
thou
g
ht of in
art and painting.

Deep sea exploration and sur
g
er
y
is alread
y
usin
g

AI.
Artificial Intelligence & Cognition