AIMS
OF TODAY’S SESSION
·
to
go over the main points of the first lecture;
·
to
provide an opportunity to ask any questions about the course;
·
to
discuss some basic issues concerning
·
AI,
·
cognitive
science & cognitive modelling
·
computer-based
approaches to cognitive modelling.
Reading for today:
(1)
Foreword
to M Sharples et al, Computers and Thought (1989): 'A Personal View of Artificial Intelligence', by Aaron
Sloman
(2)
Ch.
3, 'Stored Knowledge', by Steve Torrance, in the same volume: (Sections 3.1 – 3.6)
Cognitive Science: Some key questions (from Monday’s lecture):
How
do we
·
use
language?
·
solve
problems?
·
operate
effectively in a complex, dynamic world?
·
learn
to carry out these functions?
Cognitive
Modelling
·
Trying
to understand cognitive processes
by building computer models.
·
Compare
to computer modelling in general
·
Example: The movie Twister.
·
Building
complex computer models to understand more about tornadoes.
·
Practical
objective: to save lives. (Increase
warning time from a few seconds to a couple of minutes.)
·
See
A Sloman, foreword to Computers and Thought p. xx:
·
Trinity
of science:
·
empirical
·
theoretical
·
practical
Cognitive Science:
An
interdisciplinary
study of mental processes:
which disciplines
involved ? ? ? ?
Kinds
of problems addressed:
·
Empirical
·
Theoretical
·
Practical
what kinds of problems
? ? ? ?
Course Overview:
·
Aims
of this course
o AI/computer modelling
perspective within cognitive science;
o basic programming concepts
(not practical programming skills)
o groundwork for later Major
and School courses
·
Reading:
o Green et al. Cognitive
Science: An Introduction.
·
Computer
Classes
o Week 2 onwards
o working in pairs (if
possible)
·
Seminars
o Week 1 onwards
·
Assignment
o Log of computer-based
experiments
·
Assessment
§
Essay
plus
Unseen
exam
Computer models of cognition.
Two
initial examples:
(a)
BUGGY:
simulating children’s subtraction behaviour;
(b)
blocksworld:
simulating simple conversations about a basic world of objects.
Buggy: Child A – where is she going wrong?
2 3
4
-----
2 1
9 2
9
2 3
5
--------
7 1
4
See Handout for today.
·
Program
‘knows’ about ‘microworld’ of boxes of different sizes
and colours.
·
Can
interpret simple English sentences, like ………
·
Put
the big green box on the small green one.
·
Place
a blue box on a red one.
·
Is
there a large box on the table?
·
Can
form plans to execute commands
· Based on Terry Winograd’s SHRDLU.
Sloman’s model (design) for an intelligent
system. (pp. xxv – xxvii)
Perceptual
mechanisms – finding out about
the world.
Database of information about the world.
Analysis
and interpretation procedures –
making sense of the perceptual input.
Reasoning
procedures – inferring new
knowledge from stored info and new input
Database
of goals – what the system
‘wants’ (or is directed) to do.
Planning
procedures – given goals and
information, how to produce a sequence of actions to achieve goal.
Executive
mechanisms – getting the world
to change by acting according to the plan.
Some questions for
discussion:
HOW FAR DO SHRDLU (AND
MSBLOCKS) CONFORM TO THIS MODEL?
WHAT ARE THE LIMITATIONS
OF THIS MODEL?