Upload
gary-carpenter
View
222
Download
0
Tags:
Embed Size (px)
Citation preview
Memory: A strange brew• Doctor: Good morning, I’m doctor X, I’m here to test your memory• H.M. Hi, nice to meet you. Anybody ever tell you that you look just like Frank Sinatra?• Doctor: I get that a lot. I’m going to start by showing you some pictures and I want to you to name them as fast as you can and then
try and remember them because I will ask you later about them. OK?• H.M. OK.• H.M. Slowly - Rhinoceros, Camel, Anchor,, lamp• Doctor: good -- can you repeat those back to me:• H.M. Sure, Rhinoceros, Camel, anchor, and lamp• Doctor: that’s very good (pager rings), I’m sorry, I’ll be right back• H.M. No problem.• Doctor leaves room for 3 minutes and comes back.• Doctor: I’m going to ask you now if you remember any of the words I asked you to remember before.• H.M. What words? Anybody ever tell you that you look just like Frank Sinatra?• Doctor: I get that a lot. Do you remember any of the words I told you before?• H.M. What are you talking about? I’ve never met you before in my life,• Doctor: OK, I’m doctor X. I’m going to start by showing you some pictures and I want to you to name them as fast as you can OK?• H.M. No problem. - quickly. (F)Rhinoceros, (S)car, (F) camel, (S) horse, (F) anchor.• Doctor: do you remember naming these pictures for me before?• H.M. Huh? I told you, I just met you,, I’ve never seen these pictures before.• Doctor: Why do you think you named some of the pictures (like the rhinoceros) more quickly?• H.M. They’re were easier to recognize • Doctor: Does it make sense that a rhinoceros would be easier to recognize than a car given that you see cars a lot more often than
rhinoceroses?• H.M. No, not really -- something weird is going on.
Memory Systems: Explicit and Implicit
• Memory is divided into two systems
• Neurally separate
• Behaviorally dissociable to some degree
Explicit (related to “declarative” memory)
• “Conscious recollection”
• Episodes (e.g. The trip I took to mars last year)
• Facts (e.g. Paris is the capital of France)
• Behavioral measure usually used - Recall
Explicit Memory: Neural Architecture
• Related to “Hippocampal formation (HF)” in medial temporal lobe can cause of explicit memory deficits
• Korsakoff’s syndrome damages mammillary bodies with alcohol abuse
• Amnesics have deficits in explicit memory
Amnesia• Retrograde: Can’t remember things in past
– Go on trip to mars for 1 year (return trip was 1 month)
– Hit head on capsule when landing in ocean– Temporally graded (events just before
accident are forgotten)• (e.g.Forget return trip but remember being on
mars)
Why is Amnesia temporally graded?
• HF holds info for a while and “teaches” information to rest of cortex
• This “teaching” is part of consolidation
• After new info has been “taught”, then it is stored in rest of cortex– Time on mars has already been “taught” (OK!)
– Return trip not completely “taught” yet (GONE!)
Amnesia
• Anterograde: New information cannot be learned– H.M cannot learn new information (e.g.
who doctor “X” is)
Implicit Memory: Everything that’s not Explicit
• Repetition priming (e.g. object recognition priming)
• Savings upon relearning (relearning the psych 120 material 10 years from now)
• Motor and skill learning (e.g. riding a bike)
Explicit vs. Implicit Memory Experiments
• Study phase - subjects look at words that have common stems (e.g. market)
• Test phase - cued recall:– Given “mar”, say which word on previous list
• Test phase - completion:– Given “mar”, say any word on which comes
to mind
Impaired Implicit Memory with Intact Explicit
• Right occipital lobe removed at age 14
• Normal functioning (owner of computer software company)
• Recognition Task: OK
• Priming (perceptual identification): Impaired
• Priming impairment specific to vision: Normal auditory priming
Explicit vs. Implicit in Normals
• Study phase - subjects look at words in 3 condition– Generate (hot - ????)– Context (hot - cold)– No context (xxx - cold)
• Test phase - recognition - say if on old list
• Test phase - perceptual identification (detect word very briefly presented)
Explicit vs. Implicit in Normals: Results
Generate Context No Context0.0
0.2
0.4
0.6
0.8
Recognition (Explicit)Priming (Implicit)
Pe
rfo
rm
an
ce
hot - ???? hot - cold xxx - cold
Eyewitness Testimony: Leading the Witness
• Loftus, Burns and Miller 1978
• Subjects shown series of slides depicting accident
• One group fed misleading questions like
• “Did any car pass the red sports car when it was stopped at the stop sign?” (if yield sign was actually present)
Results:
• No misleading info: 85% correct
• Misleading info: 38% correct
• Called “the misinformation effect”
Flashbulb Memories
• Supposedly very good memory for huge events (e.g. JFK assassination)
• But are they correct?
Test of Flashbulb memory: Challenger
• Students recorded info at time of disaster: Tested 2 years later
• Many had vivid memories• Scored 0 to 7, (2 - where, 1 -who they were
with, 1- time of day etc.)• 3/ 44 scored 7, mean =2.9, 50% <2.• Almost no correlation of accuracy and
confidence
Learning: Acquiring New Knowledge
• Associations: (e.g.) The smell of coffee means that you will get some Folger’s
• Structures: (e.g. how U.S. government is put together)
Learning Associations: Classical Conditioning
• Every time I play mozart, my dog salivates
• Unconditioned Stimulus (US) (meat)
• Unconditioned Response (UR)(salivates)
• Conditioned stimuli (CS) (Mozart)
• Conditioned response (CR) (salivating to mozart)
Classical Conditioning: Timing
• CS comes on a little before US
• Adaptive: CR prepares the organism for the US
• Some pairings take longer
Classical conditioning one one trial
• Danger pairings can happen very fast– E.g. conditioning dog to fear the mailperson
(shock him once after showing him mailperson)– Conditioned taster aversion
• US can follow a long time after• Give rats flavored water• Make them sick with a drug• They will avoid flavored water in future• Associate flavor with getting ill
Constraints of Conditioning
• Not all things get associated equally as well– Flavor with sickness - good association
– Loud noise and sickness - not as well
– Hard-wired pathways that affect ability for things to associate
Stages of Classical Conditioning• Acquisition (training): pair CS and US
(e.g. playing mozart and giving dog food)
• Extinction (testing): don’t pair CS and US, but still get some response for a while
• Spontaneous Recovery: wait a while after extinction trials and try CS again and get UR, but then goes away again
Stimulus Specificity and Stimulus Generalization
• What gets paired with US?
• CS is middle “C’ and is paired with meat
• Dog learns that middle “C” means meat and salivation
• What if tested with “D above middle “C”?
• If he salivates he is showing stimulus generalization
Instrumental Conditioning• How do you train your dog to bark jingle
bells?• Get her to bark and then slowly shape her
behavior with reinforcement• “Law of Effect”• Can reinforce with positive (good dog) or
negative (bad dog)• Learnability of responses varies
Non-associative Learning• Habituation - less response with experience
– Construction noise outside my window every day• Day 1 - I want a bb gun, Day 20 - I sleep blissfully
– Major way of studying cognitive processes in babies
• Sensitization - more response with experience– Often associated with strongly aversive stimuli
Example: Study and test on land and under water
Recalled on Land Recalled underwater0
10
20
30
40
50
60
Test condition
Pe
rce
nt
Co
rre
ct
Studied Underwater
Studied on Land
Context and Mood
• Subjects learned words either inside (I) or outside (O)
• Recall was better when study and test matched
• But (I) was dreary office and (O) was nice sunny place outside
• Maybe outside caused a good mood and inside caused a bad mood
Context and Mood
• Same experiment, but mood manipulated
• I+ (Inside with good mood)
• I- (Inside with bad mood)
• 0+ (Outside with good mood)
• 0- (Outside with bad mood)
• Results: mood mattered, Inside/Outside didn’t
Assessing Memory: Recall and Recognition
• High-Frequency words recalled better than Low-frequency
• Low-Frequency words recognized better than High-frequency
Context-dependent learning
• A double edged sword– If you study only in one context and reproduce
all conditions at test, then similar context is good
– But there are contextual elements you can’t control (state of your brain fluctuates)
– You can’t always reproduce study conditions (can’t study in 2278 all the time)
Context-Independent Learning
• You want to take your knowledge on the road
• You want to retrieve this information anywhere, anytime
• Vary study contexts so that memory traces are encoded with variability
Memory: Finding the Way
• Even information is stored somewhere in your head, you still have to access that information– Many “failures” of memory are really failures
to access– Things you thought you forgot can often be
relearned easily (savings upon relearning)
• Encoding with variability sets up accessible memory traces
Encoding Variability: Spacing
• Massed practice: Large blocks of practice– Performance during training is often better– Long-term performance is crummy
• Spaced (distributed) practice: Smaller blocks of practice separated by time– Performance during training is sometimes not
as good as massed– Long-term learning is much better than massed
Spacing Effects Example
• Baddeley (1990) taught subjects typing
• 4 training schedules– (1) 1 hour sessions/day
– (2) 1 hour sessions/day
– (1) 2 hour sessions/day
– (2) 2 hour sessions/day
Spacing Effects: Results
8575655545353550
60
70
80
90
Hours of practice
Co
rre
ct K
ey
stro
ke
s
(1) 1-hour session(2) 1-hour sessions(1) 2-hour session
(2) - 2 hour sessions
Spacing Effects
• Applies to almost all kinds of learning– Skill learning (motor)
– Fact learning (explicit learning)
Spacing Effects: Short Time Frames
5040302010020
30
40
50
60
Events Between the 2 presentations
Pe
rc
en
t C
orre
ct
1.3 Sec
2.3 Sec
4.3 Sec
Spacing Effects: Long Time Frames
654321030
40
50
60
70
80
Years
Pe
rc
en
t C
orre
ct
56 day
28 day
14 day
Encoding Variability: Context
• Studying in different contexts will help you generalize the information– Example: Study in more than one place if you
cannot take the test in the place where you study
• One exception: If you have a hard time getting focused except in one place
• Better to learn to have a few places where you can study
Encoding Variability: Motor Skills
Kids had to throw a bean bag at a post 4 feet away
– Group A: Practice at 4 feet only– Group B: Practice at 3 and 5 feet
– Wait a while then test
– Group B did better at 4 foot toss!
Encoding Variability: Motor Skills
Practical advice for Shaq’s free throw problem
–Don’t shoot 40 free throws in a row
–Do shoot 14 ft, 15 ft, 16 ft, 15 ft, hook shot, 15 ft, etc…
Mental Imagery Experiment– Custard-Lumber– Jail-clown– Envelope-slipper– Sheepskin-candle– Freckles-apple– Hammer-star– Ivy-mother– Lizard-paper– Scissors-bear– Candy-mountain– Book-paint– Tree-ocean
Using Imagery to Enhance Memory
• Interaction between elements is especially useful – Example: Dollar Bill and Elephant
• Imagine elephant paying with a dollar bill at a checkout line
• Bizarreness is not especially important
The Method of Loci
• Associate each item on a list with a place
• Take a “Mental Walk” and associate each place with an item
Loci: Does it Really Work?
1 Week 5 Weeks0
10
20
30
40
50
60
70
80
90
100
Retention Interval
Pe
rc
en
t C
orre
ct
LOCI
Control
Enhancing Memory: Depth of Processing
• Processing at deeper levels lead to better memory– Example (meaning>rhyming>case judgments)
• Try to connect new material with old material in as as many ways as possible
Enhancing Memory: Interaction
• Interact with material as much as possible
• Read (less than 50%)
• Organize, Summarize and Test (>50%)
Enhancing Memory: Testing
• Testing is powerful learning event
• Also reveals what you really know and what you really don’t know
Enhancing Memory: Expanding Retrieval Practice
• Bob Bjork invented this to facilitate learning• Practice retrieving the information• Slowly expand the delay interval between study
and test– E.g. (Job interview - remember names)
• Retrieve all names immediately after you leave
• Retrieve again in 5 minutes
• Retrieve again in 20 minutes
• Retrieve again 60 minutes
Learning vs. Performance
• Not the same thing
• Sometimes performance and learning trade off
• Massed vs spaced practice
• Massed - better for performance
• Spaces - better for learning
False Intuitions
• During learning:– Hey - I really know this stuff (massed practice)– I’m not so sure about this stuff (spaced
practice)
• Later when it counts:– Hey - I actually remember that stuff (spaced
practice)– Huh? (massed practice)
Testing: A Great Way to Assess Learning
• Test yourself– Assessed what you really know
– Testing is itself a powerful learning event
•Memory Enhancement– “Seven Tigers Invade Indiana Mall”
– Space out study periods
– Test yourself often (expanding retrieval if possible)
– Interact with material, process as deeply as possible by connecting to existing knowledge
– Imagery (method of loci)
– Match study and test conditions when possible