Intro to Cognitive Science

1
Intro to Cognitive Science

• The Basic Question
• History
• Methods
• The Current Approach CRUM
• Evaluating Cognitive Theories

2
The Basic Question in Cognitive Science

• How do humans know?
• How is knowledge acquired?
• How is knowledge represented in the mind?
• How is knowledge manipulated by the mind?

3
History

• Our view of how knowledge is acquired has wavered
  back and forth between two broad traditions
  rationalism and empiricism

4
History (2)

• Empiricism
• experience alone give us basic truths regarding
  the world
• knowledge is induced through the senses
• supports a notion of the mind as a tabula rasa
  or wax tablet (Locke)

• Rationalism
• reason alone gives us basic truths regarding the
  world
• knowledge is logically deduced from
  self-evident premises
• supports a notion of mind as possessed of innate
  ideas that allow us to deduce knowledge

5
History (3)

• Rationalists
• Plato (427-347 BC)
• Descartes (1596-1650)
• Chomsky

• Empiricists
• Aristotle (384-322 BC)
• Locke (1632-1704)
• Watson (1878-1958)
• Skinner (1904-1990)
• worksheet

6
History Synthesis of Rationalism Empiricism

• David Hume (1711-1776)
• what is innate is faculties not conscious ideas
• Immanuel Kant (1724-1804)
• empiricism is enabled by faculties that cannot
  themselves be derived from experience
• Jean Piaget (1896-1980)
• humans go through stages of cognitive development
  as they interact with the world
• Steven Pinker (b. 1958)
• every language contains both a system of rules
  and an associationist network

7
Synthesis George Miller

• short term memory has limited capacity it can
  hold only 5-9 chunks of information
• the human mind
• takes in information
• operates on the information to change its form
  and content
• stores the information
• locates the information
• generates responses to the information

8
A Second Question

• How does the mind perform the tasks that Miller
  outlined, i.e.
• How does the mind represent and process
  information?

9
Theories of Mental Representation

• Formal logic Chapter 2
• Rule-based formalisms Chapter 3
• Frames, Schemes, Scripts Chapter 4
• Mental imagery Chapter 5
• Case-based reasoning Chapter 6
• Connectionism Chapter 7

10
Research Methods in Cog Sci

• Issue mental representations and operations are
  not consciously accessible
• Each discipline that contributes to Cog Sci tries
  to get at the mind through its own research
  methods

11
Psychology

• experiments with human subjects to identify and
  explain behavior, and to discriminate between
  theories
• Example

Immediate Recall Delayed Recall
Percent Correct
Early Middle Late Position of Word
in List
12
Computer Science

• Builds models that simulate human performance to
  allow testing of theories and greater precision
  in theory design
• Example
• Rule-based Connectionist
• Noun Phrase -gt Det Noun If Det, then
• Adj. .35
• Noun .65

13
Linguistics

• Builds grammatical models
• Example
• Phonology Syntax Semantics
• sound meaning

Do sound rules and meaning rules communicate
directly? or are they mediated by syntactic rules?
14
Neuroscience

• experiments with the brain during surgery, after
  cerebral accident, with scanning devices
• Example
• PET scans show that many of the brain areas
  activated when people performed a reaction time
  task were the same as those activated during REM
  sleep.
• performance of these people on the reaction time
  task improved with practice and improved even
  more after they got a night's sleep.
  Maquet, P. et al. Nature Neuroscience

15
Cognitive Anthropology

• studies differences in mental operations across
  cultures
• Example
• In a study of Japanese and American perception,
  Japanese subjects looking at a computer simulated
  fishtank saw the background images first, while
  Americans saw the fish first.

16
Philosophy

• asks the basic questions about
• the relation of the mind to the world and
  experience
• the nature of mental representation and
  computation
• the nature of explanation in cognitive science

17
Cog Sci Study Guide on the www

• http//www.chss.montclair.edu/psychology/cogscienc
  e/intromenu.html

18
Central Hypothesis of Cog Sci

• thinking involves
• mental symbols (Representations)
• processes that operate on those symbols
  (Computations)
• The Computational Representational Understanding
  of Mind (CRUM)
• broad enough to encompass the different views of
  representation (logic, rule-based, scripted, etc.)

19
A Mental Representation

• an idea, a symbol that stands for what we know
• Examples
• cat
• dog

linguistic symbols
graphic symbols
20
A (Computational) Process wksht A

• an operation on the representation
• Example
• cat s ? cats
• dog z ? dogs
• horse iz ? horses
• What about mouse?

21
Representation Computation

• draws on the computer analogy
• computers have data structures (representations,
  stored information)
• cats noun singular concrete

- and algorithms (computations) cat s ?
cats (plural)
22
The Analogy of the Mind as a Computer Program

• Program
• data structures
• algorithms
• running programs

• Mind
• mental representations
• computations
• thinking

23
Cognitive Science Procedure

• Propose a theory, a set of representations and
  rules
• Design a model consonant with the theory
• Build a computer program to test the model
• Evaluate the program by comparing it to human
  performance
• (Is it worse than, as good as, or better than
  the human?)

24
Advantages of a Running Program(A Computational
Model)

• allows for a test of the plausibility of the
  cognitive theory (if the model runs, the
  underlying theory is plausible)
• allows comparison with human behavior (a
  plausible model produces the same kinds of errors
  as humans)
• generates detailed predictions about human
  behavior, which can then be tested
• suggests new kinds of mental representations and
  procedures for our cognitive theories

25
Evaluating a Cognitive TheoryDoes it have . . .

• Representational power - can the theory capture a
  wide array of human capabilities?
• rational thought
• sensory input
• emotional affect on thought
• verbal expressiveness
• . . .

26
Evaluating a Cognitive TheoryDoes it have . . .

• Computational power - can the theory efficiently
  capture the various tasks involved in thought,
  including
• Problem solving through
• planning
• decision making
• explanation
• Learning
• Language acquisition, comprehension, and
  production

27
Evaluating a Cognitive TheoryDoes it have . . .

• Psychological plausibility - is the theory
  consonant with experiments that demonstrate how
  humans perform the tasks that the program
  performs?

28
Evaluating a Cognitive TheoryDoes it have . . .

• Neurological plausibility - is the theory
  consonant with experiments that demonstrate how
  the brain functions in the performance of the
  tasks that the program performs?

29
Evaluating a Cognitive TheoryDoes it have . . .

• Practical application in education, design (e.g.,
  human computer interface), and intelligent
  systems
• An intelligent system is a computer program that
  acts as a stand-alone expert or as a tool to
  support human decisions