Transcript PPT Lecture Slides: March 12, 2002

Memory Stores
and
Memory Processes
What is your first memory?
2002/03/12
Psyc202-005, Copyright Jason Harrison
1
Technological metaphors
Plato: memory as an aviary, as a wax tablet
Middle ages: memory as a book, as an empty
cabinet
1944: Our memories are card-indexes consulted,
and then put back in disorder by authorities
whom we do not control (Cyril Connolly)
1950s: memory as a telephone system
1960s: present: memory as a kind of computer
2002/03/12
Psyc202-005, Copyright Jason Harrison
2
Memory
• Change blindness:
– we don’t remember everything about a scene
– but we do remember lots
• Stores: What types of memory do we have?
• Processes: How do these memory types work?
2002/03/12
Psyc202-005, Copyright Jason Harrison
3
Memory Stores
• Sensory register/buffer
– a “copy” of incoming stimulus
– only lasts for a short time
• Short-Term Memory (STM)
– a few important “chunks” or information
– last as long as attention is given to it
• Long-Term Memory (LTM)
– knowledge about the world (events, experiences)
– lasts indefinitely, infinite
2002/03/12
Psyc202-005, Copyright Jason Harrison
4
Memory Processes
• Encoding
– putting information into a store
• Maintenance
– keeping it "alive”
• Retrieval
– finding encoded information
2002/03/12
Psyc202-005, Copyright Jason Harrison
5
Sensory Register: Iconic Memory
Incoming image
Iconic image
Immediately fades away…
Gradually fades away…
2002/03/12
Psyc202-005, Copyright Jason Harrison
6
Sensory Register: Iconic Memory
• How much can it store?
– Total-report technique
2002/03/12
a) show observer brief image (array of letters)
a
q
t
h
k
p
j
b
z
b) ask for a report of all remembered letters
Psyc202-005, Copyright Jason Harrison
7
Total-Report Technique: results
• Observers can reliably report 4-5 items
• Does this mean that only 4-5 items can be
stored?
• No:
– May only mean that 4-5 items can be reported
before items are forgotten.
– Response mechanism (attention?) takes time, iconic
memory decays while reporting.
• How to test?
2002/03/12
Psyc202-005, Copyright Jason Harrison
8
Partial-Report Technique (Sperling)
2002/03/12
a) show observer brief image (array of letters)
a
q
t
h
k
p
j
b
z
b) signal one row via tone (high, medium, low)
c) ask for a report of letters in signaled row
Psyc202-005, Copyright Jason Harrison
9
Partial-Report Technique: Results
• Observers can report 4-5 items from each row
– even when many rows were presented
– (only one row selected by tone)
• Thus:
– capacity of iconic memory is very high
– elements fade rapidly, often before attention can
examine them
– can only report 4-5 (before fading)
2002/03/12
Psyc202-005, Copyright Jason Harrison
10
Summary of Sensory Register
description: a copy (photograph) of input
format: copy of features
capacity: very high
entry of information: non-attentive (automatic)
duration: about 1/4 second
maintenance of information: impossible
2002/03/12
Psyc202-005, Copyright Jason Harrison
11
Short-Term Memory
Working Memory
•
•
•
•
•
Store used for conscious tasks
Semantic not iconic
Lasts about 15 seconds
Constantly being written over
example: pail of water under a water tap:
– new stuff comes in (interference)
– old stuff flows out (decay)
2002/03/12
Psyc202-005, Copyright Jason Harrison
12
Short-Term Memory (Working Memory)
• is short term memory limited in size or
duration?
Test: briefly present subject with three consonants
GCH
When light goes on (after several seconds)
say what the consonants were
2002/03/12
Psyc202-005, Copyright Jason Harrison
13
Consonant Report Task: Results
• Observers can report 3 consonants
2002/03/12
Accuracy
– accurate as long as rehearsal possible
– duration > 30 seconds (a long time!)
100%
50%
0%
Duration
10 s
20 s
Psyc202-005, Copyright Jason Harrison
30 s
14
Decay of Short-Term Memory
• information quickly decays when rehearsal
stops
Briefly present subject with three consonants
Then have them count backwards by 3s
out loud from given number
KBS
504
When light goes on (after several seconds)
say what the consonants were
2002/03/12
Psyc202-005, Copyright Jason Harrison
15
Decay Task: Result
• Performance declines rapidly with delay
2002/03/12
Accuracy
– essentially zero after 15-20 seconds
100%
50%
0%
Duration
10 s
20 s
Psyc202-005, Copyright Jason Harrison
30 s
16
Interference and Short-Term
Memory
•
•
•
•
Present stimulus to be remembered
Present non-similar stimuli
Present similar “interfering” stimuli
Example:
– RHT, 520, 294, 93, BUMMER
2002/03/12
Psyc202-005, Copyright Jason Harrison
17
Capacity of STM
• Look at how recall depends on number of digits
2002/03/12
3 digits:
273
100%
5 digits:
92612
100%
7 digits:
4678104
90%
9 digits:
954867632
20%
Psyc202-005, Copyright Jason Harrison
18
Capacity of STM
• Number of digits recalled: digit span
• Generically, capacity of STM: memory span
• Miller: memory span is 7(±2) items
– this is true for vision, audition, etc.
– “the magical number 7”
• But.. 7 what?
• What is an “item”
• What are the units of STM?
2002/03/12
Psyc202-005, Copyright Jason Harrison
19
Units of STM: “Chunks”
• Chunk: group of items that have a meaning
For example, this sequence is difficult to remember:
FB
IUB
CIB
MUN
(11 letters)
But this sequence is easy to remember:
FBI
UBC
IBM
UN
(11 letters)
Second sequence has same letters
- arranged as chunks - units have meaning
2002/03/12
Psyc202-005, Copyright Jason Harrison
20
Methods for Searching STM
• Phone up movie theatre and listen to list of
movies currently playing
• Answer the question:
– Are they playing A Beautiful Mind?
• To answer this question you could:
– Parallel search
– Serial, self-terminating search
– Serial, exhaustive search
• Which one is most plausible?
• What are the implications of each one?
2002/03/12
Psyc202-005, Copyright Jason Harrison
21
Searching STM (Sternberg)
• Have subject remember 2 items in STM
– e.g., W X
– ask whether an item was in list (X?)
– measure time taken to say “yes” or “no”
• Increase number of items to 3
– e.g., A X U
• Repeat process for 4 items, 5 items, etc…
• Look at time to say “yes” or “no” versus
number of items in memory
2002/03/12
Psyc202-005, Copyright Jason Harrison
22
Reaction time (ms)
Search time of STM
600
500
400
1
2
3
4
5
Number of items held in STM
Average slope = 152 ms/ 4 items = 38 ms/item
Serial item-by-item scan of short-term memory
-each item needs 38 ms to be checked
-scan is exhaustive - each item in STM checked
2002/03/12
Psyc202-005, Copyright Jason Harrison
23
Applications of STM
• Used for Many perceptual/cognitive tasks
– tracking items across space
– doing addition/subtraction
– temporarily remembering phone numbers
• Specialized STM systems for vision, audition,
spatial, faces, etc.
• Baddeley: general purpose “working memory”
system
2002/03/12
Psyc202-005, Copyright Jason Harrison
24
Spatial memory: Monkey vs. Human
2002/03/12
Psyc202-005, Copyright Jason Harrison
25
Baddeley’s Model
2002/03/12
Central
Executive
Phonological
Loop
PL & VS are
"slave"
systems
Visuospatial
Sketchpad
Psyc202-005, Copyright Jason Harrison
26
Baddeley’s Model of WM (STM)
1. Auditory STM - “phonological loop”
• allows maintenance/manipulation of speech-based
information (stink vs. sink vs. smell)
• maintenance allows learning of new words
2. Visual STM - “Visuospatial sketchpad”
• allows maintenance/manipulation of visual and
spatial information (shape, colour, position)
• maintenance allows learning of new objects
3. Control System - “central executive”
• selects which STM system to use
• selects strategies for information manipulation
2002/03/12
Psyc202-005, Copyright Jason Harrison
27
Summary of Short Term Memory
description: data for mental operations
format: “chunks”
capacity: about 7 “chunks”
entry of information: requires attention
duration: > 30 seconds (if not disturbed)
maintenance of information: continued attention
and rehearsal required
2002/03/12
Psyc202-005, Copyright Jason Harrison
28
Long Term Memory (LTM)
• Separate system from STM
• Evidence: damage to medial temporal complex
– HM (see Ramachandran)
hippocampus removed from both sides of brain
– Patients with Korsakov’s syndrome
due to chronic alcoholism
• For these patients, long-term learning is
impossible, but:
– normal STM memory span
– normal STM scanning speed
2002/03/12
Psyc202-005, Copyright Jason Harrison
29
Capacity of LTM
• Essentially unlmited -- can always add more
• But, this does not mean that everything is
available later…
• Recall different from Recognition
2002/03/12
Psyc202-005, Copyright Jason Harrison
30
Recall versus Recognition
• Recall
– What did your learn in class today?
– need to generate facts, ideas
• Recognition
– Did you learn about LTM in class today?
– need to verify given facts, ideas
• Which is more difficult?
2002/03/12
Psyc202-005, Copyright Jason Harrison
31
Conway, Cohen & Stanhope (1991)
• Test what was learned in university classes
– recall of names and concepts
– interval between 3 to 12 years
• How much of this kind of knowledge is remembered and for
how long?
• Are some types of knowledge better remembered than others ?
• Does knowledge undergo structural changes over time?
• Do older people forget faster than young people?
• Is very long term retention superior when students originally
obtained high grades?
• Do they remember it better if they found the material very
interesting when they studied it?
2002/03/12
Psyc202-005, Copyright Jason Harrison
32
Results: Cued recall (fill in the blank)
Concepts
Names
% correct recall
80
60
40
20
0
3
25
50
100
125
-memory for names decays sooner than for concepts
-both average to about 25% retention over long term
2002/03/12
Psyc202-005, Copyright Jason Harrison
33
Serial Position Effect (Rundus)
• Does recall accuracy depend on position in list?
• List of 20 nouns, one every 5 seconds
– cat, ball, house, truck, pencil, vase, hose, book,
lake…
• Test recall as function of position
• Two effects: primacy and recency
2002/03/12
Psyc202-005, Copyright Jason Harrison
34
% correct recall
Results: Serial Position Effect
80
60
40
20
0
1
5
10
Position
15
20
Primacy effect - better recall for words at beginning
Recency effect - better recall for words at end
2002/03/12
Psyc202-005, Copyright Jason Harrison
35
Primacy and Recency Effects
• Primacy effect is due to…
– greater rehearsal of items
– more rehearsal, more chance to get into LTM
– if rehearsal prevented, primacy effect disappears
• Recency effect is due to…
– items still in STM
– if use of STM prevented (introduce another task
before testing) recency effect disappears
2002/03/12
Psyc202-005, Copyright Jason Harrison
36
Types of LTM
• Most studies test for retention of words and
concepts: semantic memory
– general knowledge
– not connected to any particular time or place
– e.g.,
• meanings of words,
• random facts about world
• rules of multiplication
• What other type of memory is there?
2002/03/12
Psyc202-005, Copyright Jason Harrison
37
Episodic Memory
• knowledge connected to a particular time or
place
• not general
– particular time or place or people
• e.g.,
– dinner I ate yesterday
– house I lived in when I was 10
– what I did Sunday afternoon
• often autobiographical events
2002/03/12
Psyc202-005, Copyright Jason Harrison
38
Semantic versus Episodic Memory
• Semantic:
– who is the president of UBC?
• Episodic:
– when did you first hear of Martha Piper?
• Amnesiacs generally more affected at level of
episodic memory
• Episodic memory forms basis for mental
imagery
– generation of internal images (e.g., mom’s face)
2002/03/12
Psyc202-005, Copyright Jason Harrison
39
Mental Imagery
• Internal genration of “images”
• Provides answers to questions about particular
things:
– Where is the clock tower in relation to the Main
Library?
• Kosslyn
– experiments on brain activity
– imagery is vision “run backwards”
2002/03/12
Psyc202-005, Copyright Jason Harrison
40
Vision: forward process
• input activates visual cortex (V1)
• this activates higher centres
2002/03/12
Psyc202-005, Copyright Jason Harrison
41
Imagery: backwards process
• higher centres activate V1
• activation of V1 gives “visual impression”
• 5% of population cannot do this
2002/03/12
Psyc202-005, Copyright Jason Harrison
42
Application of mental imagery
Roman Room (“Method of Loci”)
•
1.
2.
3.
used by Roman orators to memorize long speeches
Pick a place you know well (e.g., your bedroom)
Find a path to travel around the room
Find items that suggest topics to be remembered
• e.g, hospital, doctor, cash
4. Place small-scale versions of these along path
• according to order of speech
5. To recall, imagine walking around room
• path helps you remember items
• items help you remember topics
2002/03/12
Psyc202-005, Copyright Jason Harrison
43
Improving LTM
• “Use it or lose it”
– LTM doesn’t become stronger with use
– But techniques you use do!
• To improve memory retention
– More rehearsal (e.g., writing, review)
• increase probability of transfer to LTM
– More connections to existing knowledge
• either semantic or episodic memories
– Sleep (8+ hours every night)
2002/03/12
• memory consolidation occurs during REM
Psyc202-005, Copyright Jason Harrison
44
Any other types of memory?
• Explicit memory
– semantic
– episodic
• Implicit memory
– procedural
– priming
2002/03/12
Psyc202-005, Copyright Jason Harrison
45
Implicit Memory
• Memory formed without awareness
– you didn’t try to remember it but you did anyway
– you didn’t try to recall it but you did anyway
• Two main types
– procedural memory: how to do things
– priming: bias towards recall based on cue
2002/03/12
Psyc202-005, Copyright Jason Harrison
46
Procedural Memory: Notes
• memory for actions, operations, skills, etc.
– e.g, how to tie shoelaces, ride a bike, start your car
• “automatic” but difficult to know how you do
something:
– can only start at particular points
– memory not easily accessed consciously
• May involve “zombie”
– procedural amnesia when striatum damaged
(part of midbrain near neocortex)
– HM has procedural memory
(hippocampus removed; striatum intact)
2002/03/12
Psyc202-005, Copyright Jason Harrison
47
2002/03/12
From: www.mareshbrainsatwork.com/ B2B/SB10.html
Psyc202-005, Copyright Jason Harrison
48
2002/03/12
Psyc202-005, Copyright Jason Harrison
49
Priming
• Facilitation of a ability to recognize words,
images based on prior exposure
• May not consciously remember prior exposure
• May not consciously see “prime”
• Increase of 5-10% over non-primed stimuli
• e.g.,
– prime with word -- show incomplete word -- ask for
completions
– prime with named images -- show many more
images -- ask for their names
2002/03/12
Psyc202-005, Copyright Jason Harrison
50
Priming with words
2002/03/12
clown
Psyc202-005, Copyright Jason Harrison
51
Priming with words
• Complete the word:
2002/03/12
c l o_ _
Psyc202-005, Copyright Jason Harrison
52
Priming with words
•Look at frequency of chosen words
•
•
•
•
•
•
•
cloak
clock
clods
clogs
clone
clonk
close
•
•
•
•
•
•
•
cloth
clots
cloud
clout
clove
clown
cloys
•primed word chosen more often than chance
•requires working neocortex
2002/03/12
Psyc202-005, Copyright Jason Harrison
53
Summary of LTM systems:
LTM
Explicit
Semantic
Implicit
Episodic
Medial temporal complex
2002/03/12
Procedural
Priming
Striatum
Neocortex
Psyc202-005, Copyright Jason Harrison
54