Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
2/18/2015
1
Chapter 6: Retrieving Memories From Long-Term
Storage
1
2Long Term Memory
IntroductionIndDiff & App
Retrieval
RetentionStorageSem v Epi
Exp v Imp
Dec v Proc
S�R�R
Rote
Org
SemanticNovel
Elab
ImageryNarrative
Consol
Interf
Pro
Retroactive
Deep Amn
Implicit
Durab
Recon
EncSp
Ghosts Lead
FM
Age
Edu
…
Long-Term Memory
• Subdivisions– LTM v STM – Semantic v Episodic
(+1)• General Knowledge v
Events at Time & Place
– Explicit v Implicit• Consciously
recollected v Not conscious, shown in other ways ()
– Declarative v Procedural
• Facts v How
3
4
4
• LTM important for many areas of life– Individual Differences: Aging
(�), Gender, Culture, …– Medical Instructions, Education
& Training, Clinical Psychology, Justice and Law (below), …
– Re-examine later with recommendations based on well-founded cognitive theory
5 LTM - Storage• Repetition: Rote Rehearsal, Shallow processing
– Not effective way to store information in LTM– Especially poor if repetitions massed all together: %
delayed recall after # repetitions in STM (below)– Better if spaced (+1)
6
MaintenanceRehearsal
2/18/2015
2
• Recall increases with spaced repetition (Madigan, 1969)
• Spacing may explain relation between Primacy Effect and # Rehearsals (�)
7 LTM - Storage• Deep Processing (v Shallow, Repetition)
– When information processed appropriately, capacity virtually unlimited
– Estimate 1 billion bits of information for middle-aged adult
– Effective coding based on Meaning• “Big” and “large” confused in LTM v “big” and “pig” in STM
• Evidence– Effect of Rate of Presentation & Levels of Processing
(+1)• Types of Deep Processing
– Elaboration: meaningful processing of individual elements
– Organization: link elements together
8
• Recall increases with slower presentations: allows for deeper processing (�)
• Incidental + Shallow less than Intentional OR Incidental + Deep ( Hyde & Jenkins, 1969)
9
Elaboration• Levels of Processing
(Craik & Lockhart): better memory if processed to “deeper" level (�)– Structural / Physical /
Orthographic• Is word in capital letters?
TABLE table– Phonemic / Acoustic
• Does word rhyme with weight?
crate MARKET– Semantic / Category
• Is word type of fish?SHARK heaven
– Sentence• Would word fit sentence?
FRIEND cloud“He met a ______ in the street?
10
ElaborationSelf-Reference
Encoding– Personal relevance
of material (adjectives)
– Rogers et al. (1977): “Describes you?” produced best recall
– Conditions• Physical (P)
• Acoustic (A)
• Semantic (S)
• Self-Reference (SR)
– Results (�)
11 Generation Effects• Lists of words or fragments
– Read ANIMAL
– Generate A_IM_L
– Better recall in generate condition
• Pressley et al. (1987 )
– Sentence memory: The toothless man
wrote a check. ... who wrote a check?
– 3 conditions
• Base only
• Base + supplied elaboration (... to pay for
the dentures.)
• Base + generate elaboration (why did that
particular man do that?)
• Other evidence for Generation
– Generate Image for words
– Pictures > Concrete Words > Abstract
Words
12
Elaboration
2/18/2015
3
Organization
• Organization involves linking
multiple to-be-remembered
items together
– Vs elaboration: processing
individual elements
• Several ways to organize
items
– Semantic: use pre-existing
relations (top �)
– Novel Associations
• Narratives
• Interactive Images (�)
13 Semantic
Organization• Recall increases with
associative relatedness
of words in list (Deese,
1959 �) r = .88
14
• Category Clustering
– Words presented in
categories (-1)
recalled much better
than random order
(Bower, 1969 )
• More on Category Clustering
– Better recall for lists of categorized words
• Lists of words from categories (e.g., dog cow horse shirt jacket
pants ...) vs. unrelated words, presented in random order
– Words from same category recalled together
– Organization and recall tend to increase across trials
– High clusterers recall more than Low Clusterers
• Thompson, Hamlin, & Roenker (1972)
– Organizing leads to better recall under Incidental
learning (Mandler, 1967): Orienting instructions
Organize
No Yes
Recall No 23.5 32.8
Yes 31.4 32.9
15 SemanticOrganization
• Scripts (or Schemas)– Sequential events that make up
episodes in our lives
– Scripts about routine activities: e.g., restaurant, visit to doctor
– Script generation (Below 1,2,3 = strong, medium, weak)
• Visit-to-Doctor Script
check in (1) sit down (1)wait (3) look at people (3)read magazine (1) name called (2)follow nurse (3) enter exam room (2)undress (3) sit on table (2)talk to nurse (3) nurse tests (1)wait (3) doctor enters (3)talk to doctor (3) doctor questions (3)doctor examines (1) dress (3)make appointment (3) leave (1)
• Bower, Black & Turner (1979)– Memory for script actions– Good, for stated script actions
relative to nonscript actions– False Alarms high: high ratings
for unstated script actions (3.91)– Intrusions increased from 1 to 3
versions (e.g., doctor, dentist, chiropractor)
RecallStated Unstated Nonscript
1 3.03 .80 .392 2.27 1.26 .353 2.56 1.16 .36
Recognition (1 = new to 7 = old) 1 5.46 3.91 1.712 5.40 4.62 1.763 5.59 4.81 1.86
16
Novel
Organization– No pre-existing
relationship between items: require novel organization (way to link or relate items)
• Narratives & Sentences– Make meaningful text– Bower & Clark (1969)
• Serial Recall of 12 lists of 20 words (e.g., bottle, curtains, airplane, gun, trowel, book, broom, telephone, ferry, desk, doorknob, cup, glass, lion, tiger, boat, bed, desk, fish, lobster)
• Half participants told to create Story (93%), half to Rehearse (13%) ()
17 • Imagery and Organization
– Interactive Imagery: Paired-Associate Learning (Bower)
Rehearsal 33% 30%
Separate Imagery 46% 27%
Interactive Imagery 80% 53%
– Imagery and Sentence Effects• Bower & Winsenz: Imagery Sentence Rehearsal
Recall � 87% 77% 37%
– Imagery works better with Concrete materials
– Imagery Mnemonics• Pegword: 1-Bun, 2-Shoe, 3-Tree, 4-Door, 5-Hive, 6-Sticks, 7-
Heaven, 8-Gate, 9-Wine
• Method of Loci (+1)
• Keyword (+1)
– Other Mnemonics
• Great Lakes: HOMES, She Made Harry Eat Onions, …
18
2/18/2015
4
19
Carta
Foreign
word
Cart
Similar
Sounding
Native Word
Letter
Native
Translation
Interactive
Image
Method of Loci: visualize to-be-remembered items at series of familiar locations
KeywordMnemonic
20Imagery & Narratives
Image in this case provides a meaningful interpretation of otherwise difficult to understand text: provides Context
Why Organization is Effective
• Semantic Strategies
– Make use of knowledge in Semantic Memory
• Imagery
– Dual-Coding Theory
• Two (Verbal + Imaginal) codes better than one
• Interactive Imagery integrates elements, so that given one element, other “automatically” comes to mind (Retrieval)
– Relational-Organizational Hypothesis
• Interactive imagery allows for more connections between elements
• Narratives and Sentences
– Create meaningful structures, may evoke Imagery
21
Loss of
Storage –
Amnesia
• Certain brain insults result in Anterograde Amnesia– Fail to store new items in LTM
– Amnesia for future events after insult
• HM is classic case– Hippocampus removed to control
epilepsy
– Retained most information from distant past, some “recent” past loss due to Retrograde Amnesia
– Able to perform STM tasks
– Affects Explicit (deliberate) storage in LTM, not Implicit (, slide 3, +1)
• Various ways to assess memory without explicit retrieval: Fragment Completion (words, pictures), Mirror Tracing, Mirror Reading, Towers of Hanoi, …
– Implicit & Explicit processes differ (+1)
22
Implicit v
Explicit LTM
• Levels of Processing
effect for Explicit, not
Implicit memory (�)
23
• HM & Mirror Tracing
– Better at task (), but did not remember performing task before
LTM – Retention (Forgetting)– Various factors contribute to forgetting from LTM
• Consolidation
– Information failed to register: requires time “consolidate”
– Retrograde Amnesia (v Anterograde Amnesia)
• Forget information PRIOR to disrupting event (e.g., head injury)
• Electroconvulsive Shock Therapy (ECT, ECS) (+1)
– Consolidation and Sleep (+1)
– One purpose of dreams may be to consolidate memories?
• Retrieval Failure (in Retrieval section)
• Interference
– Retroactive & Proactive Interference
24
2/18/2015
5
Electroconvulsive
shock disrupts
consolidation of
learning (
Chorover & Schiller)
Sleep onset
disrupts
consolidation (�)
25 Interference
• Interference can be from Prior Learning or from
Subsequent Learning (T6.1 below)
– Notation (A-B, A-C) from Paired Associate Learning
• Retroactive Interference (+1)
26
Retroactive Interference: events subsequent to learning affect memory (Jenkins & Dallenbach, 1924)
• More learning of List 2 leads to poorer recall of List 1 (�)
• Stronger learning of List 1, more resistance to RI (Overlearning!)
• More similar material, greater interference
27
Barnes &Underwood1959
• Proactive Interference (PI)– Interference from previously
learned material
– See T6.1 (-2): usually delay (retention interval) between learning A-C and testing A-C
– Underwood (1957)• Noted considerable variability in rate
of forgetting across different studies
• Classified 14 studies by number of prior lists (�)
– Relevant to Ebbinghaus? (+1)
– PI and STM• Interference from prior items on
Brown-Peterson task
• Similarity: Release from PI when materials change (e.g., letters to numbers) (+1)
• Perhaps contribution of LTM on some STM tasks
28
• Ebbinghaus: used self in many studies of memory
• Rapid forgetting () due to massive amounts of PI?
29
• Brown-Peterson STM task• Trials 1 to 3 all Fruits
• PI develops• Trial 4 switch categories for
some groups• Release from PI varies
with similarity to Fruits
Durability of LTM
• Despite sources of
interference, LTM can
be extremely durable
– Bahrick’s research on
memory for university-
learned material over
very long time periods
• Student names, campus
landmarks, Spanich (�)
• Recall v Recognition
– Linton’s diary study
(+1)
30
2/18/2015
6
• Kept “diary” on index
cards of day-to-day
events
– Randomly selected cards
at later time to test
memory for events
– Rather little forgetting of
about 6% a year (�)
– Forgetting steady (rather
than curvilinear)
– Forgetting influenced by
repeated testing (�)
– Meaningful, Personally
Relevant information!
31 Making Memories Durable• Previously discussed methods:
Elaboration, Organization, Spacing BUT students thought massed better! (�), …
• Testing Effect (+1)
– Taking tests on material improves learning.
• Group 1 studies material, no quiz
• Group 2 studies material, takes quiz
• Group 2 performs better 2 days later, or even 2 weeks later
– Quizzing better study technique than rereading or highlighting
• Overlearning (+1)
32
Testing Effect• Students given repeated
testing (T) remembered more
than repeated study (S),
especially after one week
(�)
33
• Repeated testing and study (overlearning) may explain why material learned in 6 mth course is remembered better than in 8 wk course ()
LTM -
Retrieval
• Material stored in LTM must be retrieved (like library book)
• Strategies for Storing & Retaining information in LTM help retrieval– Retrieval Cues: category clustering,
interactive imagery, story, …
– Retrieval Practice: spacing, testing, …
• Failure to retrieve contributes to forgetting– Recall v Recognition
• Retrieval is reconstructive process that can be facilitated or hindered by experimental manipulations
34
Retrieval• Reconstructive memory
– Bartlett: War of the Ghosts and similar exotic stories
• From different cultures to include “unusual” (from Western perspective) features (e.g., ghosts)
• Retrieved stories shorter: 180 words vs 330 in original
• Memory distortions made more “sensible” schema for story
– Levelling: omit illogical elements
– Sharpening: add, rationalizations
– Transform words, reorder events
– Retelling leads to distortions in
memory, also images (�)
35 Retrieval• Encoding Specificity
– At recall, helps to have same
context as when material
originally learned (encoded)
– Divers who studied
underwater recalled more
under water; divers who
studied on land recalled more
on land (�)
– Similarly, mood (Bower, �),
drugs, …
– Take tests in same room where
learned material if possible!
36
Learn Sad
Learn Happy
2/18/2015
7
Retrieval• Memory subject to
distortions ( F8.13)
• Leading questions & eyewitness memory
– “Did a car pass the red Datsun at stop sign?”
• Sign actually yield sign
• Wrongly recognized stop sign 59% of time
– Effect of verb on estimated speed ()
– No better $, hypnosis, second guess, …
37 Retrieval
• False Memories
– Controversial: vs Recovered Memories
– Loftus induced false memory in subjects
• “You went on shopping trip with your mom and cousin. You
wandered away in store and got lost. A security guard found
you and you were reunited with mom about an hour later.”
• This never happened. But after repeated questioning, 29% of
participants “recalled” details of the false event!
– Other experimental tasks (+1)
• Many people, including professionals, believe
– “Everything we learned is stored permanently in brain.”
• Not consistent with evidence
38
Experimental False Memories
• Deese/Roediger-
McDermott
paradigm
• Study list of words
related to sleep:
Dark, Dream, Pillow,
Nap, Night, Quiet, …
• 80% of participants
falsely recognize
“Sleep” as on list,
but not (recall �)
3940• “Flashbulb Memories” appear
distinct from typical memories: vividness, confidence, …– Remember where you were at time of
traumatic event• Kennedy shot, John Lennon, Shuttle
disaster, 9/11, …
– Many issues• How does one verify the “memory”?• Confidence can be poor guide to accuracy• Unknown questioning and retelling
contaminate memory trace
– Neisser & Harsch (1992)• “Flashbulb memories no more
accurate than ‘ordinary’ memories but confidence higher” (+1)
– Some evidence for Flashbulb Memories
• Conway: contacted 678 adults few days after 9/11 and obtained diverse information: where were you, how did you hear about it, …
• Contacted people year later (+1)
Flashbulb
Memories
40
41
41
% (Conway)
Individual Differences & Applications• Age
– Elderly (slide 5) (�)– Childhood Development
• How do LTM strategies develop?• Metamemory (Flavell): Young children
may know how to use strategy (e.g., organization), but only use it when told
• Younger children cannot perform strategy even when told, and older children do not need to be told to
MetamemoryCan Use Do UseStrategy Strategy
Very Young No NoMiddle Yes NoOlder Yes Yes
– e.g., Rehearsal (Lip reader); clustering (+1)
Age % Rehearsal5 107 60
10 85
42
2/18/2015
8
• Development of clustering– Niemark et al (1971)
• 24 pictures from 4 categories presented in random order
• Bear, Camel, …; Chair, Lamp, …; Jacket, Mitten, …;Boat, Car,…
Grade 1 3 4 5 6 College
Recall (Trial 3) 12.7 18.5 20.3 22.3 21.8 23.9
Clustering .34 .46 .56 .65 .76 .95
• Culture and LTM– Cole et al. (1971)
• Nonliterate Africans did not do well when tested with lists of words
– Effect of schooling• Educated Africans recalled lists as well as Americans, and Uneducated
Africans recalled less (Scribner, 1974)
• Primacy effect stronger in schooled Moroccan children (Wagner, 1980)
– Type of Material• Memory for Stories : Ghanian college students (oral tradition) better than
Americans at remembering stories (Ross & Millson, 1970)
• Franklin: familiarity with categories and ethnicity
• Australian Aborigines: better spatial memory than Euro-Australians
43 • Education
– Spacing repetitions (�)
– Instruction
• Structuring & Organizing
lessons: Ausubel
– Texts
• Advanced Organizers
• Mayer: diagrams and
scientific text (�)
– Mnemonics
• Keyword Mnemonic (+1)
44
• Pressley et al (JEd Psy, 1982)
– University students; unfamiliar English words (30)
Group %Correct
Keyword Imagery 48.90
Key Sentence 55.00
No strategy 28.9
Controls/Copy/Syn/Imagery 24.25 24.1 20.9 23.1
• Generalization & Age– Trained on city-product: Lock Haven - newspaper
– Transfer to latin (mannus-pony), with or without reminder
Recall of Latin Translations
10-13 yrs 16-19 yrs
Control 5.6 11.7
No Reminder 5.5 16.7
Reminder 9.9 18.7
45 Applications• Medicine
– Poor memory for medical instructions
– Patients' Recall of Information presented by GPs before and after doctors started to use procedures designed to increase recall: Concrete/Specific, Categorize, Repetition
4 studies Mean
Before 52, 56, 57, 59% 56%
After 61, 70, 73, 80% 71%
• Clinical and Forensic Psychology
– Depression and Memory for autobiographical events
– Eyewitness identification and ethnicity
– Repressed and False Memories• Identify people at risk: e.g., children, suggestible, …
• Identify procedures conducive to FM: e.g., imagining, drawing, …
– Jurors (+1)
46
“Jurors” and Misconceptions about Memory• Surveyed of 1,000 potential jurors in Washington, DC area about memory
and eyewitness testimony. Many misconceptions– “Act of remembering traumatic event like a video in that one can recall details as
if imprinted or burned into one’s brain.” 52% true or did not know
– Whether “weapon” makes “eyewitness memory about details of crime more reliable, less reliable or no effect.” 37% thought weapon made witness’s memory more reliable, 33% thought no effect or not sure
– Whether fact that “crime is violent” makes “eyewitness memory about details of crime more reliable, less reliable or no effect.” 39% thought violence makes memory more reliable, 33% thought no effect or not sure
– Compare reliability of witness “absolutely certain” of identification with witness who was not. 31% certain witness “much more reliable”
– 40% agreed “eyewitness confidence in identification is excellent indicator of reliability”
– Compare reliability of identification by eyewitness “of same race as person being identified” with eyewitness “of different race.” 48% cross-race and same-race equally reliable. Only 36% cross-race less reliable.
– Compare reliability of “lineup of potential suspects standing next to one another” with “one at a time.” 76% thought simultaneous more reliable or equal to sequential lineup, or not sure. Similar results (61%) for photo lineups.
– 52% do not understand importance of conducting “photo array where police officer running array is unaware who suspect is.” 30% incorrectly believed array where identity known is more reliable than “double blind,” and another 22% believed that two procedures equally reliable or unsure.
47 • Attitude Change
– Relevant to many applied domains: clinical, health, education, …
– Memory for statements in arguments
Smoking Statement Type
Anti- Pro-
Smokers 1.26 1.53
Ex-smokers 1.83 1.08
Non-smokers 2.56 1.50
Seatbelt Use Anti- Pro-
Never wear 2.07 1.60
Sometimes wear 1.78 1.72
Always wear 1.61 2.29
48