Transcript Syllabus P140C (68530) Cognitive Science

Memory II
Reconstructive Memory
Forgetting
Observe this crime scene
What does a penny look like?
Memory Biases
• Memory is better for meaningful significant features than
for details of language or perception
 gist is remembered better than detail
Label distorts memory of objects
Carmichael, Hogan, & Walter (1932)
Reconstructive nature of memory
• Memory is often side-effect of comprehension
– details can be filled in or reconstructed
• Constructive approach to memory:
– Memory = actual events + knowledge, experiences,
expectations
Effect of Expectation on Memory
A simple demonstration experiment
I am going to show you a picture of a graduate student’s
office. Just take a look at it for a while
Now write down all the things you can remember
Potential responses:
Chairs
Desk
Table
Boxes
Bottle of wine
Picnic basket
Books
Skull
Brewer & Treyens (1981): 30% of subjects (falsely)
recalled that books were present
Misinformation Effect
• Memory for event can be influenced by information given
after the event
Elizabeth Loftus
Misinformation Effect
• Subjects view a movie of a car accident
• Different expressions used to describe car contact
• Subjects estimate speed of a car at time of contact
Misinformation Effect
Explaining Misinformation Effect
• Three hypotheses
– Overwriting
• misleading information alters the memory trace
– Source confusion
• Sometimes we misremember the source of a
memory
• Perhaps the memory of the question is confused
with the memory of the visual scene
– Misinformation acceptance
• Ss. believe the information in the postevent is true
Overwriting Hypothesis seems unlikely
• McCloskey and Zaragoza (1985)
• See event: yield sign
• Receive misinformation, “as the car passed the...”
misleading:
“...stop sign?”
nonmisleading:
“...yield sign?”
• Test:
yield sign OR stop sign  35% drop in accuracy for
misleading information
yield sign OR no U-turn  no difference in accuracy
for misleading information
(both groups much higher
than chance)
Relevance to Criminal Justice System
• most obvious case
– crime
 study
– picture of suspect (mugshot)
 misinformation
– Lineup
 test
• Eyewitness may recognize suspect from mugshot, not from
crime scene.
• Conclusions:
– Do not let potential witnesses see suspects.
– Interrogate without asking leading questions
Traditional Lineup
1
2
3
4
5
6
7
8
Sequential Lineup
1
2
3
4
5
6
7
8
Recovery of Lost Memories?
• Several lawsuits have relied on eyewitness testimony of
repressed memories. These memories were “recovered”
by family member or therapist
• Claim: repression follows stress, but repressed material
can be returned to consciousness with the removal of
stress (e.g., Zeller, 1950, 1951; Merrill, 1954)
• Problem:
• Are these repressed memories or false memories
(based on misinformation)?
Recovered memory vs. False Memory
• How do we know whether repressed memories are
accurate? Hard to falsify
• In some cases, traumatic information is misremembered
or simply “made up”
– Loftus has been involved in many cases
– Points out problems of
• hypnosis
• suggestive questioning
• dream interpretations
Can false memories be implanted?
Loftus and Pickrell (1995)
Imagination Inflation
False Memory in the Lab
• Deese, Roediger, McDermott paradigm
• Study the following words
SLUMBER
BLANKET
DROWSY
SNOOZE
DREAM
SNORE
AWAKE
PEACE
TIRED
WAKE
YAWN
DOZE
REST
BED
NAP
• Recall test ....
• Recognition memory test
Use ratings 1) sure new 2) probably new 3) probably old
4) sure old
• TEST:
COFFEE
SNORE
SLEEP
REST
Results
• Critical lure (“sleep”) are words not presented but similar
to studied words. These words are often falsely recalled
(sleep: 61% of Ss.)
• Recognition memory results
proportion of items classified with confidence levels:
confidence rating
4
3
2
1
studied items
not studied
unrelated
critical lure
.75
.11
.09
.05
(e.g. “REST”)
.00
.58
.02
.26
.18
.08
.80
.08
(e.g. “COFFEE”)
(e.g. “SLEEP”)
Accuracy and Confidence
• Eyewitness testimony requires accuracy and confidence
– “eyewitness testimony is likely to be believed by
jurors, especially when it is offered with a high level of
confidence” (Loftus, 1979)
– Should we rely on the confidence level given by a
witness (“I am sure I saw this”)?
– False memory experiment shows sometimes
confidence is high while accuracy is low
Forgetting
Forgetting Functions
• Ebbinghaus (1885/1913): Forgetting over time as
indexed by reduced savings.
Most forgetting
functions show:
Negative
acceleration
Rate of change
gets smaller
and smaller
with delay
Power law of
forgetting
Why do we forget?
Some possibilities:
• Memory has disappeared
 decay theory
• Memory is still there but we can’t retrieve it
 interference theory
e.g. blocking
 inhibitory mechanisms
e.g. retrieval induced forgetting
supression
Example
• You call a friend, but realize you need an older phone
number that you have not used for a while. With effort,
you recall the correct old phone number
Explanation 1: the
old number is
blocked by the
new association
NEW PHONE
NUMBER
FRIEND
OLD PHONE
NUMBER
Example
• You call a friend, but realize you need an older phone
number that you have not used for a while. With effort,
you recall the correct old phone number
FRIEND
NEW PHONE
NUMBER
Explanation 2:
the old memory has been
suppressed
 Retrieval induced
forgetting
OLD PHONE
NUMBER
Evidence for Retrieval Induced Forgetting
• Blocking would predict that using a new cue would
remove blocking effect. Suppression would predict the
memory cannot be accessed with a new cue either
 some evidence for suppression
FRIEND
NEW PHONE
NUMBER
OTHER MEMORY CUES
OLD PHONE
NUMBER
Inhibitory processes in memory?
•
Suppression is an example of an inhibitory
process
•
Two paradigms based on idea of inhibition:
•
Retrieval induced forgetting
•
Think-no-Think paradigm
• Can we voluntarily repress certain thoughts
or memories from our awareness?
Think-no-Think Paradigm
• Subjects studied pairs of
weakly related words
• Recall and say aloud the
response word
“no-think”
• Or avoid thinking of the
response word (“no-think”
condition)
• Recall of “no-think” words
was impaired compared
to “respond” pairs
Anderson and Green (2001)