Download Memory - Villanova University

Document related concepts

Mind-wandering wikipedia , lookup

Holonomic brain theory wikipedia , lookup

Transcript
Course Overview
Knowledge
Acquisition
(perception)
ch. 3: Vision. How are
objects recognized?
Ch. 6-11:
Ch. 12-14:
Long-term Memory Reasoning
ch.4: Attention.
ch. 5: Working Memory
- Buffer for mental
representations
Use
- to know is
to remember
-
Problem Solving
Memory
Memory Processes
Rehearsal
Attention
Sensory
Memory
Encoding
Retrieval
Short-Term
Memory (STM)
Long-Term
Memory (LTM)
Sensory Memory (aka iconic memory)
• Modality specific (iconic, echoic)
• Very rapid decay (300 ms after a stimulus is removed)
– Partial vs. Whole Report (Sperling 1960)
• An array of letters is presented very briefly.
• Report all letters present.
• READY? (CogLab experiment)
A
D
R
N
M
S
CB
L
Z
F
B
PC
U
GH
H
E D
Y
S
HQ X
IJA N
B
12
10
8
6
4
whole report
Participants can report ~4 letters.
2
# of letters correctly reported
Diagonal:
Perfect
performance
2
4
6
8
10
# of letters in the stimulus
12
Partial Report (Sperling 1960)
• An array of letters is presented very briefly.
• An tone indicates which row of letters to report (‘top’)
• The tone occurs after the letters have disappeared
• READY?
A Z
M P
C G
E
H
J
D
X
N
# of letters correctly reported
(# of letters “available” for report)
12
10
8
6
4
2
Participants see more
than 4 letters,
partial report
advantage
whole report
2
4
6
8
10
# of letters in the stimulus
12
Data -- partial report advantage
How to account for the data?
Propose an internal construct:
- a very brief memory store -- “sensory memory”
- in whole report, information is lost from sensory
memory by the time the first letter or two are
written down
But what exactly is the duration of this memory store?
How could we determine?
Sperling (1960)
# of letters available
10
8
6
4
2
.15
.30
.60
delay of tone cue (seconds)
1.0
To what extent is information in sensory memory processed?
What is the “code”?
- “raw”, visual code?
- categorical code? ( vowel or consonant)
- identity code? (the exact identity of the letters)
Sensory Memory
T V
T C
T C
T V
A
X
S
E
M C
M C
MV
M C
F
K
O
T
B C
B V
B C
B V
N
U
V
I
# of letters correctly reported
(# of letters “available” for report)
12
10
8
6
4
2
Sperling (1960)
partial report
location cue
partial report,
categorical cue
whole report
2
4
6
8
10
# of letters in the stimulus
12
No partial report advantage for category cues!
information in sensory memory is encoded spatially rather than
semantically
“Where” is sensory memory?
- is it just an afterimage on the retina?
- is it a more “central” kind of memory?
McCloskey & Watkins (1978)
- anorthoscopic perception
Memory Processes
Attention
(tone)
Sensory
Memory
•Very rapid decay
•Modality specific
Rehearsal
Short-Term
Memory (STM)
Short Term Memory (STM)
a.k.a. Working Memory (WM)
A cognitive system that allows the maintenance of
information on line or available for
immediate processing.
We should ask:
1. What are the constituent parts of WM?
2. Properties of WM: capacity, code, duration
3. What is WM for? Reading, problem solving,
mental arithmetic, reasoning
4.
Is it really different from long-term memory?
1. What are WM constituent parts?
WM is modality specific (verbal & visual codes are stored separately)
– Visual WM
– Verbal WM
There are subdivisions even within modality
– Visual WM: Spatial ; Object
– Verbal WM: Articulatory ; Buffer
‘Something’ has to coordinate all this info
– Central Executive
WM is modality specific. Prediction: If verbal and
visual WM are independent, verbal and visual STM
loads should not interfere with each other
‘remember these numbers’
3982174
Load verbal WM
‘remember
this shape’
delay
Test Spatial WM
‘compare shapes’
‘recall numbers’
?
Manipulation check
4932687
?
Questions: what is our DV?, what is our IV?
Verbal WM load
High load
Low load
No load
39
3982174
Test Spatial WM
?
?
Manipulation check
Results
Visual WM performance remains unchanged at different levels of
verbal wm load. No interference.
2. Properties of WM
Capacity: How many things can you hold in WM?
• It may be different for Visual and Verbal WM
• Verbal WM: digit span task
• Visual WM: change detection task (Luck & Vogel 1997)
Color squares (set size 1-12) for 100 ms
900 ms Blank
Color squares (set size 1-12) for 100 ms
~ 4 objects Remembered
- Again no interference with verbal WM load
Are all objects treated equally in VSTM? No.
2.6
1.6
2.0
2.8
3.7
4.4
Alvarez & Cavanagh (2004)
2. Properties of WM
Code: How is information in WM encoded?
• The code will be different for Visual & Verbal WM
• Verbal WM: we’ll discuss this later in the lecture
• Visual WM:
– A. Do we store objects or features?
– B. Is ‘spatial’ WM different from ‘object’ WM?
2. Properties of WM
Code: Do we store objects or features?
Each object has 2
features (color,
orientation)
Participants either
monitor for a change in
a single feature
Or they monitor for a
change in either
feature (conjunction)
Memory for objects (conjunction condition) is as good as memory
for a single feature (e.g., color), despite the fact that objects require
the storage of twice as many features. This is evidence that visual
STM stores ‘objects’ (Luck & Vogel, 1997)
Four features
color
orientation
size
gap
Monitor for a
change in a single
feature
Or monitor for a
change in any
feature
1. What are WM constituent parts?
WM is modality specific (verbal & visual codes are stored separately)
– Visual WM
– Verbal WM
There are subdivisions even within modality
– Visual WM: Spatial ; Object
– Verbal WM: Articulatory ; Buffer
‘Something’ has to coordinate all this info
– Central Executive
2. Properties of WM
Code: Is spatial WM different from Object WM?
Spatial: is the
encode
initial
display
probe in a same or
different location
than the circle in
the initial display?
delay
Object: is the
probe
probe of a same or
different color as
the one in the
initial display?
Different brain regions active
during storage in working memory
of spatial and object information
2. Properties of WM
Code: Is spatial WM different from Object WM?
Color squares (set size 1-12) for 100 ms
Conjunction visual search (attention)
Color squares (set size 1-12) for 100 ms
Spatial WM: ‘where’
Object WM: ‘what’
VisuoSpatial
Attention
Same location?
Same color?
Results:
- Spatial WM performance interferes with visual search
- Object WM does NOT interfere with visual search
1. What are WM constituent parts?
WM is modality specific
– Visual WM
– Verbal WM
There are subdivisions even within modality
– Visual WM: Spatial ; Object
– Verbal WM: Articulatory ; Buffer
Visual WM
- Capacity: 4-5
- It stores ‘objects’
- & spatial locations
Verbal WM
1. What are the constituent parts of
Verbal Working Memory?
Phonological
short-term store
Verbal information
& phonological decoding
of written language
Sub-vocal
rehearsal
process
Phonological Buffer: Evidence
• Task: Memory Span (CogLab experiment)
– Read a list of items, and repeat them
• Phonological Similarity:
– Confusions occur if words sound alike mad, cat, man, map, cat
– But not for similar meaning: huge, long, tall, big, wide
– nor for similar-looking: bough, cough, dough, through
• So, the code of verbal WM is ‘acoustic’
Subvocal Rehearsal: Evidence
• Articulatory suppression: repeatedly say “the” while
hearing the list of items. “the the the the the the”
– Decreases performance in just 20 secs (duration)
– Reduces phonological similarity effect (convergent evidence for
phonological buffer)
• Word length effect:
– memory span for “sum, wit, harm” better than for
“opportunity, individual, university” because it takes
shorter to articulate
• Neurological overlap with language areas
Neural overlap between verbal WM and language
Speech production areas and language receptive areas are active
when people try to remember phonological information
2. Properties of verbal WM
• Code: Phonological
• Capacity:
How many things can you hold in verbal WM?
7 + 2 items
• But what counts as an item?
– A word? A letter? A sentence? A phoneme?
– An association (a pointer) to a representation in longterm memory (i.e, chunking)
Ready for a test
B F K EJ F I KARA F D
Another trial
F B I J F KC I A F D R
F B I
C I A
F D R
J F K
chunking allows storage of greater amounts of
information…because information is
“packaged” more efficiently
Short-Term Working Memory:
A multi-part system
Central
Executive
-All the WM tasks
discussed so far are
pretty ‘dumb’.
(coordination)
Rehearsal
-Humans are capable
of doing much more
with their WM.
Visual WM
Verbal WM
- Something has to
coordinate all the parts
of WM.
- Capacity: 4-5
- Capacity: 7 + 2
-It stores ‘objects’
- & spatial locations
- It stores ‘sounds’ -- acoustic code
- It has buffer and rehearsal
3. What is Working Memory for?
1.
Keeping information available: mentally reciting a telephone
before writing it down. Pretty dumb task
2.
3.
Reading,
problem solving: mentally rotating the image in the instructions
when building IKEA furniture
4.
5.
mental arithmetic: Calculate how much to tip the waiter?
Reasoning
The Central Executive
• Supervise attention
• Planning/Coordination
• Monitoring
Frontal lobe syndrome
• Distractibility, difficulty concentrating
• Problems with organization, planning
• Perseveration: fail to stop inappropriate
behavior
Memory Processes
Rehearsal
Attention
Sensory
Memory
•Very rapid decay
•Modality specific
Encoding
Retrieval
Short-Term
Memory (STM)
Long-Term
Memory (LTM)
• Limited Capacity (7+2) •Unlimited capacity
•Consciously available
•Hard to get stuff in it
• Flexible material
•Organized
•Decays if not rehearsed
semantically
++++++++ Different domains: Visual, Verbal, etc.+++++++++
4. Is Working Memory really a process
different than Long-term Memory?
1.
Reading, problem solving, mental arithmetic, reasoning
2.
Is it really different from long-term memory?
Free Recall Task
- Listen a list of words (10-40),
- at the end write all the words you remember
- you can list them in any order.
Serial Position Function
Probability
of reporting
the item
?
12
………
Position in Original List
1. Monster
2. Camera
3. Tricycle
4. Melon
5. Window
6. Guest
7. Quiet
8. Cherish
9. Waiting
10. Villanova
11. Computer
12. Child
13. Chicken
14. Ghost
15. Slave
30
List Length
20
30
10
20
Position in Original List
30
40
Prob.
Of
Rept.
1
40
Serial position effects are consistent over different list sizes...
distinctiveness
Villanova
Primacy
Recency
Privileged rehearsal
better LTM encoding
STM contribution
(Glanzer & Kunitz, 1966)
distinctiveness
Lincoln
Washington
Adams
Jefferson
Primacy
W
Clinton
Bush senior
Reagan
Johnson
Recency
STM
LTM
Primacy and Recency Effects
STM
LTM
early sensory
processing
•Unlimited capacity
• Consciously available •Hard to get stuff into it.
•Organized semantically
• Flexible material
•Fixed # of slots
(7+2 chunks)
•Decays if not rehearsed
STM
(Murray Glanzer)
(Murray Glanzer)
LTM
Independence of LTM and STM:
Neurological evidence
Patient H.M.
- Normal working memory: normal digit span
- Impaired Long-term memory (anterograde amnesia): unable to
learn most new information.
Patient K.F.
- Impaired working memory: Digit span of 1 item
- Normal Long-term memory (learn word lists when lists presented
repeatedly, and do fine on long-term recognition).
(Alan Baddeley)
Prob.
Of
Rept.
Normals
STM Patients
Position
Anterograde Amnesia
might be explained as a
blockage of the flow of
information from STM
to LTM
Sensory
LTM
STM
BUT…short term memory deficits in the absence
of LTM deficits spell trouble for this gateway
model of LTM acquisition...
Sensory
LTM
STM
Entry into STM is not necessary for entry into LTM
• Double dissociations guard against resource
artifacts (differences in task performance that
stem from differences in task difficulty)
• For example,
– I can juggle 3 balls, but
– I cannot juggle 5 balls,
• Should we conclude that juggling 3 balls is a
process independent from juggling 5? Or that
juggling 5 balls is a more difficult task?
– We’ll argue for independence only if we find
someone who is unable to juggle 3 balls but can
juggle 5 (double dissociation). Quite unlikely :-)
Double dissociations guard against resource
artifacts (differences in task performance that
stem from differences in task difficulty)
For example, Patient H. M. has:
- impaired LTM but,
- normal STM
Should we conclude that LTM is a process independent
from STM? Or that LTM is a more difficult task?
We’ll argue for independence only if we find
someone who is unable to hold things in STM but
can retain them in LTM (patient K.H.).
Four memory systems for visual information
1. Visible persistence (afterimage)
Complete sensory (low-level) representation that lasts for 100 ms or so after a
stimulus appears. Information preserved in precise retinotopic coordinates.
2. Informational persistence (iconic memory)
Decays within ~300 ms after a stimulus is removed.
3. Visual Short-term memory
Limited-capacity visual representation lasting up to many seconds and abstracted
away from precise retinotopic organization. Can store ~ 3-4 objects (Luck &
Vogel, 1997)
4. Visual Long-term memory
Highly robust retention of visual representations (at least up to a year).