Transcript ppt

Activity (2 pts)
1. One main claim from Pinker reading
that you found provocative / insightful /
useful.
2. One main claim from Pinker reading
that you found confusing / problematic.
X
What is Cognitive Psychology?
Computational Theory of Mind
Five Key Features:
1.
2.
3.
4.
5.
Posits a level of analysis wholly separate from the
biological or neurological
Faith that central to any understanding of the human
mind is the computer
Deliberate decision to de-emphasize certain factors
that may be important but complicate things
(emotion, history/culture, role of context)
Faith in interdisciplinary studies (philosophy,
psychology, artificial intelligence, linguistics,
anthropology, neuroscience)
Claim that a key ingredient in contemporary cognitive
psych. is the agenda of issues which have long
exercised epistemologists in the Western
philosophical tradition
Computational Theory
of Mind & Memory


Despite huge variation in human
capacities, we all come equipped with
same info-processing system
Outline of Today’s Lecture



Review fundamental aspects of
human computational system
Contrast humans & computers
(what makes humans unique?)
Connectionist models (briefly)
Initial Reception of Info.



Information, in form of physical energy (light,
sound, pressure etc) is received by receptors to
particular energy forms (e.g. rods & cones)
Receptors send signals in form of electrochemical
impulses to brain
Thus, the first transformation of information:
multiple forms
of energy
(light, sound,
pressure, etc)



one
common
form
Nerve impulse goes to central nervous system,
where it is registered in immediate memory
different immediate memories for different
senses: iconic, echoic
Immediate memory holds a fairly complete
representation of incoming sensory information
for extremely brief period of time
Selective Perception

Selective Attention: Focusing on what is most
important in a stimulus array




Biological tendency (thanks to evolution) = sudden
loud noises, motion, light/dark contrast
Expertise in a domain = ability to attend to the most
important parts of a stimulus (e.g. chess)
The pattern knowledge we acquire over a lifetime
increases the accuracy of selective perception of
items in familiar contexts
Perception is influenced by:
1.
2.
Prior knowledge
Context of the stimulus itself
Perception is as much a “top-down”
process as it is a “bottom-up” process
Working Memory






Working Memory: Awareness (what you are aware
of at any given moment is said to be in your WM)
Information stored here decays within ~ 10
seconds unless rehearsed (e.g. telephone #)
WM holds 7 2 units of information (Miller, 1956),
hence it is the ‘bottleneck’ of info.processing
WM functions as a mental workspace (like a
scratchpad of sorts)
Because of limited capacity, it’s difficult to perform
several (cognitively demanding) tasks as once
One way around the limitations of WM is
“automating” tasks (e.g., decoding text)


Automatic processes: processes performed without
much awareness
Controlled processes: processes that require a lot
attention
Storage &
Long-Term Memory




Information in working memory can be stored in longterm memory (LTM)
Storage: a set of processes by which new information is
integrated in various ways with known information
Function of LTM: store information for later use
LTM: long duration (perhaps even a lifetime)

inability to remember is more a failure of finding a good
retrieval cue than loss of information
Exercise:
 Name people in your high-school graduating class x
 Common results:


recall difficult until good retrieval cues found, such as…
thinking systematically about subgroups of students
o
o
o
“kids in my neighborhood”
“kids in my chemistry class”
“kids who beat me up”
We stopped here
& will continue with
the rest tomorrow.
Retrieval

Retrieval: a set of processes that put information stored in
LTM into a state in which it can be used for current
processing
Different for controlled & automatic processing…

Controlled processing






LTM  WM
Retrieval means becoming aware of the stored info.
activation: Process of re-establishing awareness of it
e.g. when students recall facts for a test
Automatic processing




LTM  effectors
Retrieval by pattern-matching processes
e.g. 26
x 32 x
guide for the sequence of steps you take is automatically
retrieved when you recognize the familiar pattern of
numbers & multiplication sign (contra, ex. Mult.tables)
Response Generation
& Effectors

When we wish to respond to a stimulus, the response has
to be sequenced in some way



e.g., physical responses, language
Response Generation: activities that organize the
sequence of responses and send messages for the
appropriate effectors to execute the sequence
Effectors:





Muscles
Glands
Arms
hands
voice apparatus, etc.
Control Processes



Flow of information in the human system is generally
organized around achieving some purpose
In other words, mental computations are not random
Control Processes: the processes that guide and monitor
information-processing events.






Goal setting
Planning how to achieve goals
Monitoring goal attainment
Revising plans
When we consciously use such processes, we are said to be
using metacognitive processes (‘meta’ = ‘above’)
In contrast to metacognitive processes…



some control processes, either innately or through learning, are
deployed automatically.
Not metacognitive because not ‘above’ the processing event
but rather part of it
In learned, automated skills, the flow of control is embedded in
the skill (e.g., sequence as a control structure)
Computational Theory
of Mind & Memory


Despite huge variation in human
capacities, we all come equipped with
same info-processing system
Outline of Today’s Lecture



Review fundamental aspects of
human computational system
Contrast humans & computers
(what makes humans unique?)
Connectionist models (briefly)
Humans vs. Computers
Similarities
Human
Computer
Input device
senses
keyboard
central processing unit
working memory
CPU
long-term storage medium
LTM
disk
output device
effectors
paper/screen
Human
Computer
size of WM/CPU
5 bits
800,000,000 bits
size of LTM
1,000,000,000 bits*
unlimited
speed of access w/in WM
.025 sec
.000000015 sec
speed of access to LTM
.200 sec
.020 sec
Differences
Humans vs. Computers
Human
Computer
preferred storage mode
analog;
time-oriented
digital;
list oriented
retention of information
graded
all-or-none
efficiency (bits/sec)
low
high
capacity
dependent
on experience
independent
of experience
very dependent
dependent
independent
independent
retrieval:
* relative to context
* relative to previous retrievals
interconnectedness
general purpose;
purpose
of information inopen set of
functions
human memory
special or
general
purpose;
closed set of
functions
Human vs. Computers


Human dependency on context allows for learning &
adaptation to take place
Our tendency for partial memory allows us to think
of novel solutions to problems
Thus, the human information processing system has
evolved to be adaptive
Computational Theory
of Mind & Memory


Despite huge variation in human
capacities, we all come equipped with
same info-processing system
Outline of Today’s Lecture



Review fundamental aspects of
human computational system
Contrast humans & computers
(what makes humans unique?)
Connectionist models (briefly)
Connectionism

How is the mind instantiated in the brain?



We have detailed neural models of low level cognition
(e.g. how we perceive a line)
We have NO specified neural models of more advanced
cognitive processes (e.g., how we remember facts
about people or judge their character)
Traditionally, the more advanced aspects of human
cognition have only been discussed in terms of
abstract information processing models
EXCEPT…
 Recent efforts to develop models of higher-level
processes that are better grounded in our
understanding of neural processing



Start with general knowledge of how neurons work
Ask the Q: how could higher-level function be achieved
by connecting together basic elements like neurons
Connectionism
Connectionism


Connectionism: Theory and research concerned
with ways of connecting neural elements together
to account for higher-level cognition
Example: McClelland & Rummelhart’s PDP model


PDP = parallel distributed processing
models how neural elements are interconnected &
simultaneously interact with each other 
X

How many can you
remember?
39583129553468401548
Activity (3 pts)


Read 'Thanks for the Memories'
What import might this have for a
teacher in a classroom? (One paragraph)
X