Working Memory II Working memory, executive control, and prefrontal cortex

1
Working Memory IIWorking memory, executive
control, and prefrontal cortex

• Cognitive Science, 9.012
• Nuo Li
• 4/27/06

2
What is working memory?
3
What is working memory?

• Some defining characteristics
• goes on in consciousness
• accessible to explicit form of expression
  (declarative memory)
• elaboration of short-term memory
• selective
• involves cognitive processing (e.g. recognition)

4
Atkinson Shiffrin (1971)

• Parallel processing of inputs
• Information are selected for entry into STM
• STM Working Memory Control processes
• Selection
• Rehearsal
• Coding
• Decision making

5
Baddeley Hitch (1974)
WM Executive Control Domain Specific Modules
6
Outline

• Working memory (WM) and short-term memory (STM)
• WM, controlled attention, and fluid intelligence
• Role of pre-frontal cortex (PFC)
• Mapping from functions to structures
• Representation

7
Outline

• Working memory (WM) and short-term memory (STM)
• WM Tasks
• STM Tasks
• Distinctions between WM and STM
• WM, controlled attention, and fluid integellence
• Role of pre-frontal cortex (PFC)
• Mapping from functions to structures
• Representation

8
WM tasks
Reading span (Daneman Capenter, 1980) Subject
read (or listen to) a list of 2 to 6 sentences.
Afterward, the subject recalled the last word of
each sentence.
9
Demo
For many years, my family and friend have been
working on the farm. SPOT
Because the room was stuffy. Bob went outside for
some fresh air. TRIAL
We were 50 miles out at sea before we lost sight
of the land. BAND
ANSWER SPOT, TRIAL, BAND
10
WM tasks
Reading span (Daneman Capenter, 1980) Subject
read (or listen to) a list of 2 to 6 sentences.
Afterward, the subject recalled the last word of
each sentence. Operation span (Turner Engle
1989) Subject solved a string of arithmetic
operations and then read aloud a word that
followed the string. After a series of such
operation-word strings, the subject recalled the
word.
11
Demo
Is (8/4)-11? Bear Is (6 x 2)-2 10? Dad Is (10
x 2)-612? Beans
Answer Bear, Dad, Beans
12
WM tasks (Memory span tasks)
Reading span (Daneman Capenter, 1980) Subject
read (or listen to) a list of 2 to 6 sentences.
Afterward, the subject recalled the last word of
each sentence. Operation span (Turner Engle
1989) Subject solved a string of arithmetic
operations and then read aloud a word that
followed the string. After a series of such
operation-word strings, the subject recalled the
word. Counting span (Case, Kurland, Goldberg
1982) Subject is presented with up to eight
displays. Each display consists of different
number of targets, and two other kinds of
distracters, all randomly placed. Subject is
required to count the targets aloud, and report
the final tally. After the series of displays,
the subject is required to report previous final
tally in order
13
Demo
Answer 5 8 6 3 9 9
14
WM tasks

• These tasks are thought to reflect some
  fundamental aspect of cognition. Score on these
  tasks predict a range of cognitive functions
• Reading listening comprehension
• Following directions
• Vocabulary learning
• Note taking
• Writing
• Reasoning
• Bridge playing
• Computer language learning
• etc

15
STM Tasks
Wechsler Digit Span Task Forward Span The
examiner verbally present digits at a rate of one
per second. Subject is asked to repeat the
digits. The number of digits increases by one
until the participant consecutively fails two
trials of the same digit span length. Backward
Span The backward test requires the participant
to repeat the digits in reverse order.
16
WM STM
17
WM STM
Goes against the classic notion of information
has to be rehearsed in order to be retained.
Proactive interference plays an important role
in short-term retention
18
Effect of Proactive Interference
Others also found similar effects Rosen Engle
(1998) Subjects (with high and low WM score)
learn to associate 12 cue word with 3 list of
other 12 words. Instruction emphasized
accuracy Example list 1 bird-bath list
2 bird-dawn list 3 bird-bath
(re-learn) Measured both timing and accuracy.
Result List 1 same List 2 low WM made more
error (showed intrusion from list 1) List 3 high
WM responded slower than low WM subjects, even
slower than themselves on trial 1. Conclusion
information in list1 is suppressed better for the
high WM subjects, which affected their
performance on later trials.
19
Effect of Proactive Interference
Kane Engle (2000) Subject (with high and low WM
score) had 3 trial in which they saw 10 words to
recall. The subject performed a rehearsal
preventative task for 16s. Result Trial 1
high WM and low WM subject performed the same
(60) Subsequence trials low WM had steeper
decline in recall performance Adding secondary
preventative task (adding interference load) The
two group performed the same. Conclusion When
add more interference load, high WM subjects were
hurt more, suggest that under normal condition,
high WM subjects allocate more attentional
control to combat the interference.
20
WM STM

• STM
• Storage of information.
• Limitation How many item can be stored.
• WM
• Different from passive storage, WM is active
  maintenance of goal related information relevent
  to a task in the face of proactive distractors.
  There is addition of mental work (cognitive
  processing) and its combination w/ memory. (
  more than just STM), involves recognition
  comprehension.
• Limitation memory component control component

21
Outline

• Working memory (WM) and short-term memory (STM)
• WM, controlled attention, and fluid intelligence
• Role of pre-frontal cortex (PFC)
• Mapping from functions to structures
• Representation

22
WM is Capacity Limited
Memory capacity vs. Cognitive control capacity
Greater working memory capacity does not mean
that more items can be maintained as active, but
this is a result of greater ability to control
attention,ability to use attention to avoid
distraction. -Engle RW
23
Model of WM, Engle
WM STM controlled attention

• Three components
• a store in the form of long-term memory traces
  active above threshold
• processes for achieving and maintaining that
  activation
• Controlled attention (capacity limited,
  individual performance difference arise)

24
Model of WM, Engle
Central Executive

• (working memory capacity, controlled attentions)
• Achieve activation through controlled retrieval
• Maintain activation (to the extent that
  maintenance activities are attention demanding).
• Blocking interference through inhibition of
  distraction

Magnitude of this link is determined by the
extent to which the procedures for achieving the
maintaining activation are routinized or
attention demanding. Thus, it is assumed that, in
intelligent, well-educated adults, coding and
rehearsal in a digit span task would be less
attention demanding than in a 4-year old children.
Strategies, procedures for maintaining activation
STM

• Activated portion of long term memory

1. Could be phonological, visual spatial, motor,
   auditory, etc
2. More or less attention demanding depending on the
   task and subject

LTM
25
Empirical Support

• Two questions
• Is separate construct of STM and WM necessary?
• Once the common variance to WM and STM is
  removed, do the WM residual variance (which
  should reflect controlled attention) correlate
  with the residual for general fluid intelligence?

26
Empirical Support
133 subjects performed 8 different tasks
Engle et al (1999)
27
WM and Fluid Intelligence

• Controlled attention is closely related to
  general fluid intelligence
• Performance task on WM task predicts performance
  on other cognitive tasks
• e.g. performance on reading span task correlates
  well with reading comprehension

28
WM and Fluid Intelligence
Antisaccade Task
Identification of target by pressing key
A
Pro-saccade
cue
target
Identification of target by pressing key
A
Anti-saccade
cue
target
Require suppression of the natural tendency to
saccade to the flashed cue. WM span assessed on
operation-span task.
Kane et al (1999)
29
WM and Fluid Intelligence
Result
Reflexive Eye Movement
Reaction Time
Eye movement not monitored in this case
Kane et al (1999)
30
WM and Fluid Intelligence
Stroop task
Result
31
WM and Fluid Intelligence
Dichotic listening task
Subjects repeat aloud words presented to one ear
while ignoring information presented to the other
ear. At some point, subjects first name is
spoken to the ignored ear. Report whether they
heard their name during the trial.
Result
20 high WM span subject reported 65 low WM span
subject reported
Conclusion
High WM span people suppressed distractor
information better
32
Current WM Models
?
Engle
Baddeley Hitch
Central Executive
Strategies, procedures for maintaining activation
STM
LTM
33
Outline

• Working memory (WM) and short-term memory (STM)
• WM, controlled attention, and fluid intelligence
• Role of pre-frontal cortex (PFC)
• Representation
• Mapping from functions to structures

34
PFC Anatomy
Dorsolateral Input mainly from medial,
dorsolateral cortical areas. (somatosensory,
visuospatial information)
Lateral Inputs mainly from ventrolateral and
ventromedial cortical areas (auditory, visual
pattern information)
Segregation of functions? More on that later
35
PFC Deficits in Human
Wisconsin card sorting
Results No difficulty learning the rule
initially. (memory component) But once learning a
rule, patient could not switch.
36
PFC Deficits in Human
Other cognitive tasks
Stroop task London tower
All these cognitive task involves a memory
component Q deficit in memory component vs.
executive control?
37
PFC Deficits in Human
Owen et al (1996)
spatial
Task Find the hidden squares Result Normal
controls developed successful self-ordered
search strategy Temporal lob lesion patients
only failed on most difficult task PFC lesion
failed on most easy task
verbal
visual
Conclusion prefrontal contribution to WM is the
mediation of problem-solving strategies and not
in memory per se.
38
PFC Deficits in Human
Other memory tasks
Shimamura 1995
Inability to suppress irrelevent information,
sensitive to proactive interference
39
PFC Deficits in Human
Recall temporal-sequence
Milner Petrides, 1984
Shimamura 1995
Deficits in recalling temporal ordering, but no
deficit in recognition.
40
PFC Anatomy in Other Species
41
PFC Deficit in Nonhuman Primates
Working memory task
Associative memory task
PFC lesioned primates Deficit in working memory
task, but not in discrimination task
PFC lesioned primates also show deficit in
spatial-delayed alternation task
42
PFC Deficit in Nonhuman Primates
Lateral PFC Lesion Slower to learn new criteria
when the diagnostic dimension is
changed. Orbital PFC Lesion Learn new criteria
normally, slower to relearn.
Dias, Robbins Roberts, 1996
43
PFC Deficit in Nonhuman Primates
Funahashi , Bruce, Goldman-Rakic (1993)
Delayed saccade task
Fixation Point
Target
Saccade!
After dorsolateral PFC lesion
Error on contralateral visual field Performance
decay with time
44
PFC Neural Response
Anti-saccade task
Funahashi, Chafee, Goldman-Rakic (1993)
45
PFC Deficits
Dorsolateral - Spatial delayed response - Spatial
delayed alternation task - No deficit in
discrimination Lateral - Object alternation -
Delayed non-match to sample Orbital - Deficits
in olfactory, taste, visual and auditory
discrimination - Discrimination reversal
learning.
46
Outline

• Working memory (WM) and short-term memory (STM)
• WM, controlled attention, and fluid intelligence
• Role of pre-frontal cortex (PFC)
• Mapping from functions to structures
• Representation

47
Verbal storage
storage
storageprocessing
48
Verbal storage
Activation in left posterior parietal cortex
(Brodmanns area) 3 frontal sites (Brocas
area) Left supplementary motor and premotor area
storage
0- and 1-back Item recognition
Storageprocess
2- and 3-back Free recall
Smith Jonides, 1999 (review)
49
Visuospatial
visual
spatial
50
Visuospatial
Spaital (blue) Object (red)
Smith Jonides, 1999 (review)
51
Some insights from Imaging studies

• Imaging studies show that working memory is
  mediated by frontal cortex and a collection of
  posterior regions differing in the types of
  information maintained.
• Posterior cortical regions seem to specialize in
  the type of information held in working memory
• Frontal area may have a special role in
  integrating different type of information

52
Outline

• Working memory (WM) and short-term memory (STM)
• WM, controlled attention, and fluid intelligence
• Role of pre-frontal cortex (PFC)
• Mapping from functions to structures
• Representation

53
WM Representation
Domain general vs. Domain specific?
Engle
Baddeley Hitch
Central Executive
Strategies, procedures for maintaining activation
STM
LTM
54
Domain General?
Rao, Rainer, Miller (1997)
Conclusion More than half of the neuron (64/123)
contain both what and where information
55
But wait
Cue
Delay
Response
Wilson, Scalaidhe, Goldman-Rakic 1993
Lateral PFC 24/31 selective for pattern 6/31
selective for both 1/31 selective for spatial
information
Conclusion segregation of what (lateral) and
where (dorsolateral)
56
Thank you!
57
Backup slides
58
PFC Neural Response
Primate performing a delayed response task
PFC Neurons
Milner Petrides (1984)
59
PFC Neural Response
Cue-delay-saccade
Funahashi , Bruce, Goldman-Rakic (1989)
Q memory of cue vs. motor preparation?
60
PFC Neural Response
Quintana Fuster (1992)
Dissociation of memory of cue from motor
preparatory cue
61
PFC Neural Response
Delayed match-to-sample task
Rainer Asaad Miller (1998)
62
Spatial
Jonides, Smith, Koeppe, Awh, Minoshima, Mintun.
(1993)
Areas for spatial memory task
63
Domain General?
Cross-modal processing
Suchan, Linnewerth, Koster, Daum, Schmid (2006)