Transcript Working memory

Working memory
Traditional Model of Memory
• Atkinson & Shiffrin (1968) 3 Stage Model
Stimuli
Sensory
registers
Short Term
Memory
(STM)
Information Processing Model
Long Term
Memory
(LTM)
What is short-term memory
(STM) for?
• Is STM for transferring information to longterm memory (LTM)?
• Is it a passive terminal for information
transfer?
Working memory: Conceptual
Background
Working memory
• STM is a temporary storage of information.
• STM is a temporary buffer (control center)
of information.
–
Temporary storage
• Questions:
– From New York to Pittsburgh, it takes about 7hours
and 30 minutes by car. From Pittsburgh to Chicago, it
takes about 8hours and 30 minutes by car. How long
does it take from New York to Chicago, if you want to
drive through Pittsburgh? Easy Ah?
– Tom is taller than Jane. Jane is shorter than Steve.
Lisa is taller than Steve and Tom. And Steve is
shorter than Tom. Is Jane shorter than Tom?
Raven test
Raven test: (Carpenter, Just, &
Shell, 1990)
(2 x 6) -10 =?
3 + 4 +3 x 3 =?
• Do the following calculations while you
keep pronouncing “the, the,…”
(3 x 5) -10 =?
2 + 5 - 3 x 3 =?
Demonstration
Task 1
– Find the answer to the following question as
quickly and accurately as possible.
Lucy came before Jane. Kathy arrived after
Jane. Suzy came before Lucy.
Who came first?
Who came second?
Task 2:
– Find the answer to the following question as
quickly and accurately as possible. While you
are looking for the question, please keep
saying “the-the-the-the….”
Tom arrived after Steve. John came
before Steve. Mike arrived before John.
Who came first?
Who came second?
• Bill went to dinner and met Jane. They decide to
share their French Fries and coke. After the dinner,
they left the cafeteria and spend some time in a
library together. After going over some review
notes for their Sociology class, they went to the
movies together. Later, Steve saw Jane sitting in a
lounge and talking with Bill. Are Bill and Jane
friends?
Temporary storage of
information
• How do we solve these questions?
• In order to answer these questions, you
need temporary storage of information.
– STM working memory
• Working memory  a buffer for
information manipulation
Computer metaphor
• Working memory  Random Access
Memory (RAM)
– 128MB
• Long-term memory
– Hard disk, Zip disk
• After shutting down your computer, you
lose the information stored in RAM.
• But the infor. stored in your hard disk is
OK.
Do we have RAM (Working
memory)?
• Do we have working memory as we have
RAM in our computer?
Operation Span Experiment
(CogLab CD)
Baddeley’s dual task
experiments
• Test the idea that short-term memory is
not just for transferring information to LTM.
• It is for a working buffer (to manipulate
information) for many cognitive activities.
Working memory
• If there is something called “working
memory”, what is it?
• How does it encode and manipulate
information?
• How does it transfer information to LTM?
• What is the capacity? (128MB?)
Hypothesis
We use Short-term memory for Working
memory
Baddeley’s experiments:
• Procedure 1 (control condition):
– The subject was given items to remember
(e.g., a sequence of numbers or letters)
– The subject report the item given in the first
task (recalling).
– The subject carried out a cognitive task (e.g.,
verifying a sentence)
• Procedure 2 (Experimental condition):
– The subject was given items to remember
(e.g., a sequence of numbers or letters)
– The subject carried out a cognitive task (e.g.,
verifying a sentence)
– The subject report the item given in the first
task (recalling).
• (Baddeley & Hitch, 1974)
Design:
Control
condition
Experimental
condition
Memory encoding
Memory encoding
Memory retrieval
(recall)
A sentence
verification task
A sentence
verification task
Memory retrieval
(recall)
Tasks:
• Memory task:
– Briefly present a list of digits (414321090) to
the subject.
– The subject recalls the digits as accurately as
possible.
• Sentence verification task:
– A is preceded by B. AB?  yes/no
– C follows D. DC?  yes/no
Experimental Condition
Memorize
756923
A is preceded by B.
AB
Yes or No?
Recall the number
Memorize
182732
D comes before C.
CD
Yes or No?
Recall the number
Memorize
672392
Memorize
108223
Q follows J.
QJ
Yes or No?
B is before T.
TB
Yes or No?
Recall the number
Recall the number
Memorize
094722
D is ahead of A.
DA
Yes or No?
Recall the number
Memorize
33598
E appears before O
OE
Yes or No?
Recall the number
Control Condition
Memorize
238646
T is preceded by I.
IT
Yes or No?
Recall the number
Memorize
521890
D comes before U.
UD
Yes or No?
Recall the number
Memorize
589886
Memorize
332189
Q follows N.
NQ
Yes or No?
B is before Y.
YB
Yes or No?
Recall the number
Recall the number
Memorize
122649
D is ahead of O.
DO
Yes or No?
Recall the number
Memorize
443284
E appears after C
CE
Yes or No?
Recall the number
What is the difference between
the two conditions?
• The memory task is divided by the
verification task in the experimental
condition.
• The memory task is not divided by the
verification task in the control condition.
Why is this difference
important?
• Test:
– Short-term memory is used as working
memory.
What is the implication of
these, given the experimental
design?
• In the experimental condition,
– you need to hold memory items in your
short-term memory while answering the
verification questions.
• In the control condition,
– you don’t need to store memory items
while answering the verification questions.
Dependent Measure:
• Accuracy/response time for the verification
task
– What results would you expect if Baddeley’s
hypothesis is correct?
• Hint (computer metaphor).
– If you open lots of programs in your computer,
what happens?
So what?
• RAM is filled with lots of information.
– E.g., Excel, Words, Photoshop, Netscape,
Eudora, …….
• Each operation gets very cumbersome
(slow).
Baddeley’s hypothesis:
• If short-term memory is something to do
with temporary manipulation of information
(solving math problems, reasoning, etc.),
– Then performance for the verification task
should decline in the experimental condition,
as compared to the controlled condition.
Results:
• Control condition vs. Experimental condition
– Subjects in the Experimental condition were less
accurate than subjects in the Control condition.
– They also required longer time to make decision.
• Mean reasoning time:
– Control  2.73 sec
– Experimental  4.75 sec
Similar results were found in other tasks:
– Sentence comprehension, grammar judgment, and
problem solving tasks.
Interpretations:
• Why longer response time required in the
experimental condition?
– The experimental condition and the control
condition differed only in one aspect.
• The verification task was given while carrying out
the memory task.
• The short-term memory task and the
sentence verification task required the
same temporary storage (buffer) space.
Working memory
• Short-term memory is “working memory.”
• It is a temporary storage for manipulating
information.
• Working memory is required to perform
many cognitive tasks (reasoning, problem
solving, language comprehension, etc.).
Encoding format of working
memory
• What is the encoding format of working
memory?
– == the encoding format of short-term memory.
– Auditory (verbal ) encoding
• Conrand’s confusion matrix.
– auditory rehearsing sends information stored
in working memory to long-term storage.
Demonstration
Task 1
– Find the answer to the following question as
quickly and accurately as possible.
Lucy came before Jane. Kathy arrived after
Jane. Suzy came before Lucy.
Who came first?
Who came second?
Task 2:
– Find the answer to the following question as
quickly and accurately as possible. While you
are looking for the question, please keep
saying “the-the-the-the….”
Tom arrived after Steve. John came
before Steve. Mike arrived before John.
Who came first?
Who came second?
Verbalization hampers the task
performance
• Why?
• Because the capacity of working memory
is limited.
• The verbalization (the-the-the-…) left little
room for the sentence verification task.
Working memory (Baddeley,
1986, 1992)
• Working memory consists of 3 components
– Central executive, phonological loop, visuospatial
sketchpad
– Phonological loop:
• Hold acoustic information (verbal rehearsal)
• (sound card?)
– Visuospatial sketchpad
• Hold visual and spatial information
• E.g., mental rotation
– Central executive
• Allocate attention, coordinate subsytems
Working memory
• Working memory is related to many
cognitive activities.
– reading comprehension, solving problems,
acquiring new vocabulary...
Working memory and learning
new words in children
• Gathercole and Baddeley (1989)
• Subjects:
– Children (4-5yrs old)
• Test1:
– measure their cognitive abilities
• general intelligence, reading ability, non-verbal
cognitive skills…
• Test 2 (non-word repetition task):
– the experimenter read aloud non-word
syllables (e.g., “mashpole,” “woop”
“kintent.” ) to children. Then, children
were asked to repeat the syllables.
– Children’s ability to repeat the syllables
was scored.
• Test 3
– 1 yr later, the experimenter measured
children’s vocabulary score (majoring roughly
the number of new words these children
acquired in a year).
• Results:
– children’s verbalization ability (as
measured by their ability to repeat nonword syllables) correlated with their
vocabulary acquisition score.
– --->
• children’s verbalization ability is related to their
ability to acquire new words.