2/27/12
Organiza on of knowledge
• Marie’s phone rings. She answers. A woman’s voice shrieks, “They’re making a movie about Schemas and Scripts them!” • Marie is confused. • Helps if you know that – Caller was vet recep onist – Marie has two cats named Shaggy and Velma • You have to give people enough context.
Organiza on of knowledge Organiza on of knowledge
• Test example • What we know • Schema (pl. “schemata” or “schemas”) – John possessed a copy of – John picked up a test – Knowledge about complex situa ons from the TA. the test. – – He worked for an hour • What we “know” Helps you understand the current situa on and twenty minutes. – John took an exam. • You’re not trying to remember a list of events to report – He le feeling extremely – He worked on the exam. later to a scien st (as in a serial recall experiment)— worried. – He was worried about • You want to get what you need out of the situa on his exam performance. – Top-down knowledge (fill-in)
The last me you dined out… Schemas
• Can you remember… • Test example • Assump ons are filled in from Less – What was waitperson’s name? – John picked up a schemas schema- test from the TA. relevant – Meaning = input + ac vated knowledge – How they took your order, word for word? – He worked for an • (Hmm, what if you ac vate the wrong hour and twenty knowledge?) – Was the service good? minutes. – You’re ac vely construc ng meaning Schema- – He le feeling relevant – What you ate? – To understand is to come up with an integrated extremely representa on. – Whether the food contained metal shreds? worried.
1 2/27/12
Schemas Schemas
• Ac va ng different knowledge results in a • What they are different understanding – Your knowledge about how the world works – Anderson et al. (1977): prisoner story (based on your experiences) • Could also possibly be construed as wrestling – General--about type of situa on (not token/ • Test on people who are/aren’t ac va ng wrestling episode) knowledge a lot – Structured rela onships, not just set of facts – Are (PE majors): 64% “wrestling” responses – Not (music majors): 28% “wrestling” responses – Used to understand the world
Example: CLOTHING schema Schemas
Slots Values • Can have embedded schemas • Torso covering: • T-shirt – Going-to-dinner schema Purchase-stuff schema • Leg covering: • Jeans Nothing – Ice-cream-truck schema • Head: • Nothing Jeans • Benefits of schemas – Infer things that aren’t directly observed • Feet: • Sandals , sneakers, pumps – Predict upcoming stuff • Jan was at a party talking to a very a rac ve individual. • Slots are specific and • She then no ced a ring on that person’s le hand. contain defaults • What is Jan going to do?
Schemas Markman & Gentner
• Influences on memory Show 2 pictures Show 2 pictures – Place schemas Man, who Man, who Woman with has dropped has dropped Girl looks at • Dorm room; grad student office camera cigare e, cigare e, xmas tree, photographs • Are there books in your TA’s office? paints paints holding dripping boy scout – Brewer & Treyens (1981): 30% say “yes” when no books were woman woman candle actually present Rate similarity Rate similarity Picture-taking schema ac vated Arson schema ac vated – Markman & Gentner (1997) • Ac vated schemas by juxtaposing similar pictures… Now recall picture A given a cue
Cue: woman cigare e Cue: woman cigare e Mem: GOOD BAD Mem: BAD GOOD
2 2/27/12
Schemas Schemas
• Influences on memory • Influences on memory – Place schemas (Brewer & Treyens, 1981) – Place schemas (Brewer & Treyens, 1981) – Markman & Gentner (1997) – Markman & Gentner (1997) • Ac vated schemas led to different encoding of • Ac vated schemas led to different encoding of depicted events depicted events • Poor encoding of schema-irrelevant events – Boutla et al. (2004)
Schemas Scripts
• Stereotypes • Specific type of schema – Scien st • Used for stereotyped event sequences • Test tubes and symbols – Going to dinner, ge ng ice cream, taking exam • Caucasian (?) • Male • Contains: – Boutla et al. – Set of ordered ac ons • Of those using pronoun in discussing this on a problem – Causal links between events set the first me this class was taught, all used he/him • E.g. p depends on good service • Boutla is a woman!
Scripts Scripts
• Evidence for scripts (Bower et al. 1979) • Evidence for scripts (Bower et al. 1979) – Study 1 – Study 2 – Presented 6-ac on passages – Present 10 lists of ac ons – Later, gave tles & asked to recall exactly – Some lists in order, others out of order • Correctly recalled: 3 out of 6 – Asked people to recall ac ons • Filled-in: 1 extra fact that didn’t take place • In order lists: 50% correct order at recall – Recall was based on familiar series of events • Out of order lists: only 18% correct order – Recall was structured around familiar order
3 2/27/12
Scripts Schemas
• Problems • General problems – What about things that don’t have a par cularly – Slots: stereotyped order of occurrence? • func on dineOut(diner, food_type, transporta on) • Going to the bathroom at a restaurant – Dependence between slots • Going-to-dinner schema – If diner is person A, Thai food – If diner is person B, nothing with meat • Be er captured by PDP-type models
Is memory accurate?
When memory goes bad
Reconstruc ve memory False memory: your data N=219 • In recalling an event, some mes other stuff is False memories! 0.9 recalled with it that’s not part of it 0.8 – Esp. for complex events, may put mul ple pieces 0.7 together--reconstruct 0.6 *** 0.5 – Errors when you probe with cues from part of a 0.4 recollec on to retrieve the rest 0.3
Average FM per trial 0.2 0.1 0 Unrelated distractor Special distractor Error types 194 out of 219 of you made more “special” than unrelated errors.
4 2/27/12
Reconstruc ve memory Reconstruc ve memory
• Example: • Error-prone memories (episodes) – Actual event: – Poorly encoded ones • Not processing lecture much • dinner (Cuban) & movie (Wordplay) w/Julia – – Probe: movies seen with German friends Ones similar to other memories (≈ encoding cues) • » Wordplay Winter cog sci lectures • German postdoc friends in Philly » An Inconvenient Truth – Pull up wrong movie – Not recent » Remember: Cuban & Inconvenient Truth w/Julia • General picture: if you can retrieve only bits and pieces, you fill in to get a whole memory
Reconstruc ve memory Reconstruc ve memory
• Effects of retrieval cues – Anderson & Pichert (1978) • Stereotypes (e.g. • Par cipants read burglar/home buyer story scien sts, band nerds) • Asked to recall details from one perspec ve – 64% perspec ve-relevant facts recalled – Guide retrieval of events – 46% other-perspec ve facts recalled • How badly injured was this – (I.e., perspec ve ma ers) person? • Then asked to recall from the other perspec ve – Probably encoding effects – Another 10% of facts suddenly came to mind! too – Retrieval alone can “jog” memory • Expecta on that band – Tversky & Marsh (2000) nerds are a li le clumsy • Recalling from a perspec ve can alter memory itself
www.xkcd.com
Reconstruc ve memory Memory issues in real life
Big point: remembering isn’t just about pulling • Eyewitness tes mony an experience out of a li le pigeonhole in – Misinforma on effect your mind. You filter it through the rest of your world knowledge. • Flashbulb memories • False memory (Which usually works, but can some mes get you in trouble.)
5 2/27/12
Eyewitness tes mony Eyewitness tes mony
• Assump ons • The misinforma on effect (Lo us, Burns, & Miller, 1978) – It’s accurate – Slide show of car accident – Certainty and accuracy are correlated – Half saw YIELD sign, half saw STOP sign • Data – Ques onnaire • Misleading ques on (“stopped at stop/yield sign?”) – 75000 suspects ID’ed per year • No misinforma on (“stopped at intersec on?”) – Some mes right, but not always – Pick YIELD slide or STOP slide • No misinforma on: 85% correct • One problem: Misleading ques ons • Misleading-Q group: 38% correct :-( – Memory has been overwri en/revised
Eyewitness tes mony Misinforma on effect
• The misinforma on effect (Lo us, Burns, & Miller, • Source confusion explana on 1978) – Like trace interference – How does this happen? • Overwri ng (“destruc ve upda ng”) – Original memory is there, but not clear where it • Source confusion came from • Misinforma on acceptance – Lindsay & Johnson (1989): • If given a misleading sugges on, it is also recalled and it may be incorrectly remembered as the thing you saw
Misinforma on effect Misinforma on effect
• Misinforma on acceptance • Lo us: overwri ng – You’re totally aware that you didn’t know • McCloskey & Zaragoza (1985): misinforma on – But you assume that the misleading informa on acceptance was correct • Why would a lawyer say something inaccurate?
6 2/27/12
Misinforma on effect Misinforma on effect
McCloskey & Zaragoza (1985) • But Lindsay (1990): source confusions • Set-up – Subjects in Lo us, M&Z weren’t aware • To test non-awareness of where info came from: – See the from under hammer – Event happens – Narra ve w/ or w/o misleading screwdriver – Misleading narra ve happens • Test – 48 hours go by… – Lo us version: hammer vs. screwdriver – Just before test: “narra ve was made up” • If they know the source, should dismiss misleading info – Modified version: hammer vs. wrench • But they didn’t--o en recalled narra ve informa on!! • Result: screwdriver misleads, wrench not • Source confusions do occur • M&Z: misinforma on acceptance – Retrieval cues? (Screwdriver > wrench) – Does not support overwri ng
Screwdriver must not overwrite because hammer “s ll there”
Misinforma on effect Memory issues in real life
• Effect is widely accepted • Eyewitness tes mony • Underlying explana on, less so – Misinforma on effect – Some s ll hold to overwri ng • Flashbulb memories – More likely: source confusion • False memory • Real and fake info compete for recogni on • Plausibility important too – “The car stopped at the stop sign, then a flying saucer hit the pedestrian. The car then fled the scene.”
Flashbulb memories Flashbulb memories
• Memories that “stand out” from others • Indelibility: Brown & Kulik (1977) – College admission – Asked N=80 about JFK • What were you doing when you found out? – First date – 79/80 remembered – Historical events (easier to inves gate) • 13-year delay (1963-1977) • JFK – Argued for dis nct biological mechanism for storing • Challenger surprising memories • 9/11 • Would have been crucial to survival • Li le forge ng • Really as “indelible” as they seem? • Highly detailed, including circumstances surrounding
7 2/27/12
Flashbulb memories
• Neisser & Harsch (1992): not so fast. – Morning a er 1986 Challenger explosion When I first heard about the explosion I was si ng on my freshman dorm room with my roommate and we were • Ques onnaire to ugrads at Emory watching TV. It came on a news flash and we were both totally – What happened? shocked. I was really upset and I went upstairs to talk to a – What were you doing? friend of mine and then I called my parents. Neisser & Harsch, 1992, p. 9 / Textbook p. 234 – Who told you? (and so on) – A er 2.5 years, recontacted to be in study • Only 25% recalled taking ques onnaire
Flashbulb memories
• Neisser & Harsch (1992): not so fast.
When I first heard about the explosion I was si ng on my – Morning a er 1986 Challenger explosion freshman dorm room with my roommate and we were • Ques onnaire to ugrads at Emory watching TV. It came on a news flash and we were both totally – What happened? shocked. I was really upset and I went upstairs to talk to a – What were you doing? friend of mine and then I called my parents. – Who told you? (and so on) Neisser & Harsch, 1992, p. 9 / Textbook p. 234 – A er 2.5 years, recontacted to be in study • Only 25% recalled taking ques onnaire • Only 3/44 had perfect recall (assuming q’aire true) • Li le rela on between confidence & accuracy
Actual: found out in class, felt sad, watched TV for details
Flashbulb memories Flashbulb memories
• Objec ons to N&H • Maybe no special mechanism – Conway et al (1994) – Important, so likely to get rehearsed a lot • Challenger explosion not consequen al for people • Tested memory for Thatcher’s 1990 resigna on – Very unusual--less interference – UK, US, Denmark – Strong emo onal tone may affect memorability – Tested at 2 weeks & 11 months » UK: 86% highly accurate » US, Denmark: 29% • Good challenge • But maybe less well encoded to begin with
8 2/27/12
Memory issues in real life Recovered memories
• Eyewitness tes mony • Terrible event forgo en for many years – Misinforma on effect • Some mes necessitates criminal prosecu on • Flashbulb memories – Statute of limita ons excep on--criminal doesn’t • False memory benefit from causing witness trauma • Some mes the accused protests
Recovered memories Recovered memories
• Agreed by all par es • Why aren’t these extra memorable? – Child abuse (or other criminal acts) are frequent, – Repression. and should be punished • Is repression real? – But also, innocent people shouldn’t be punished – Maybe. – Some mes no way to verify/falsify • Proponents: once retrieved, very accurate and vivid – Again, hard to verify – Remember that confidence ≠ reality
Recovered memories Recovered memories
• Opponents • False memories induced – No evidence that memories real – Ge ng lost in a mall (but not enema) – Therapists (trying to be helpful) may ask – Details of childhood crib misleading ques ons or encourage erroneous reconstruc on of events • Alien abduc on vic ms: Clancy et al. 02 – We know recall can be inaccurate – More sugges ble • Response: Sure, but lab experiences can’t achieve – More prone to false memory effect (see Ch. 6) ecological validity • – Emo onal trauma very strong Familiarity = fame – Extended abuse ≠ stop sign!
9 2/27/12
Memory issues in real life
• Eyewitness tes mony – Misinforma on effect Memory & Metamemory • Flashbulb memories • Recovered memory • False confession
A bit more on recovered memories Final issue: false confession
• Sugges bility • Causes: – Emo onal stress – Porter & colleagues: – Social pressure • Lost in mall (15% implanted) – Sugges on • Enema (0% implanted) • Distrust memory enough • Serious childhood animal a ack: about 30% – If you think it’s possible to repress horrific memory, you – Not a minor incident might believe you’ve done it – A large propor on of people are sugges ble • “Interroga ve sugges bility” (Gudjonsson) – Tested people who made confession, later retracted – They were more prone to sugges ve ques ons
Final issue: false confession Memory issues in real life
• Experimental false confession (Kassin) • Eyewitness tes mony – “Type spoken le ers--don’t hit ALT!” – • Fast or slow Misinforma on effect – Experimenter: false accusa on • Flashbulb memories – Subject: no, I didn’t 1 • Recovered memory – Confederate: “I saw you do it!” ( /2 subj’s) – Overall, 70% signed a confession Low certainty of own • memory False confession – If typing fast and confederate, all • And 35% had a detailed recollec on about it!! Misleading ques on • But again, problem of scale – ALT key isn’t exactly a dagger
10 2/27/12
Memory issues in real life Hypnosis and memory
• Common threads: • Used in therapy, on eyewitnesses – Great confidence • Scien fic findings – Possible inaccuracy – Does not improve list recall • Driest lab test possible • Hypno zed subjects recall more vs. controls – Stuff on the list – Stuff NOT on the list • Overall, no improvement (+signal but +noise too) • Some mes not as well as control subjects who are encouraged strongly to try their best
Hypnosis and memory Hypnosis and memory
• Used in therapy, on eyewitnesses • Used in therapy, on eyewitnesses • Scien fic findings • Scien fic findings – Does not improve list recall – Does not improve list recall – Does increase confidence in memory – Does increase confidence in memory • • Some states exclude tes mony
Alterna ves to hypnosis “Special” memories?
• Are there be er ways? • Eyewitness tes mony – Cogni ve interview (Fisher & Geiselman) – Misinforma on effect • No misinforma on provided • “Report everything” (not just specific ques ons) • Flashbulb memories • Ask witness to reinstate context • Recovered memory • Ask witness to take different perspec ve • False confession – Results: be er recall, with a li le bit of erroneously-recalled info Upshot: there’s no such thing as a free lunch. – Used to train English & Welsh police These memories are prey to everything that “normal” memories are suscep ble to, despite the subjec ve feeling of certainty that o en surrounds them.
11 2/27/12
Really, really good memory
• What about those people who can memorize Extraordinary memory phone books? – Yes, they exist. and metamemory – No, they’re not fundamentally different from anyone else.
Really, really good memory Really, really good memory
• Individual differences • Strategies we’ve already discussed – Aren’t we all working with the same equipment? – Pay a en on at encoding! • If you don’t think about someone’s name, you won’t remember it – Factors later • Mo va on to learn – Rehearse in mul ple ways • More interested in material • Relate to knowledge (looks like a Roxanne I went to grade school • Just plain be er memory with) – Evidence: • Elaborate (imagine her at La Jolla Cove where there were rocks and sand) • Strategies – Set up a good retrieval plan • Prior knowledge • Imagine (or prac ce) giving your report in auditorium • Retrieval cues will be er match encoding cues
Really, really good memory Memory experts
• Exis ng schemas/domain knowledge • 7 ± 2 is normal STM limit – Spilich et al. (1979): baseball • Chunking helps (even if not meaningful) – Memory for melodies • Chase & Ericsson (1981)
• In an unfamiliar domain – Things go “in one ear and out the other” – Expert performance seems ‘magical’ to you
12 2/27/12
Memory experts Memory experts
• Chase & Ericsson (1981) • – Trained people to chunk Chase & Ericsson (1981): HOW? – Subject 1: runner, chunked into running mes for Incoming: 2141034084750 different races; got up to 80 digits 2141034084750 Huff, puff – Crazily-talented person? – Subject 2: trained w/same strategy, got up to 40 digits with ≈ same rate of improvement marathon 10 miles LTM LTM 100-yd mile dash
No strategy Strategy
Memory experts Memory experts
• Chase & Ericsson (1981) • Mnemonists – Great, but so what? – Use mnemonics – – Wouldn’t generalize to (e.g.) le er sequences or Examples: • Bizarre images grocery lists – Problem set 3 grows legs, leaps into your bag before Thursday – Your chunks have to match what you’re morning class • Method of loci (for ordered things) memorizing – Mentally navigate a familiar path – Put the to-be-remembered things at points along path • PPMDAS, EGBDF (order of opera ons, lines on staff) – Point: you aren’t going to forget sentence order – Banks off of knowledge of grammar
Memory experts But wait… 1 2 3 4 2 3 4 5 • Luria (1968): “S” 4 5 6 7 … • Are there real memory experts? – 70 words in memory span task • And backwards • And the next/previous word, given any word • Part 1 – Synaesthete (A is blue, etc.)--rich encoding • Part 2 – Had to also use method of loci and others – Quit journalism to become professional mnemonist – Some mes didn’t pick up on simple consistencies
Thanks to a student from last year for bringing this to the class’ a en on.
13 2/27/12
Thought ques ons Metamemory
• Are these people qualita vely different from • Knowing what you know (and how well you people with “regular” memory? know) • Think about a sports-obsessed or music-obsessed friend. Would astounding content knowledge in – Do I know James Bond’s phone number? their domain of interest surprise you? – Is it reasonable to forget my keys? • Does somebody’s life provide an unusually – Do I need to read over my notes again? coherent reference frame? • What if people mentally rehearse the events in • Are we good at this? their lives constantly—could that constant – E.g. false confidence in memory rehearsal lead to be er encoding? – Usually we are good – How would you test this?
Metamemory skill Metamemory skill
• Nelson et al. (1994): good metamemory • Infer well-encodedness by property of – Par cipants learned new (Swahili) words memory – Experimental group es mated how well each – Level of detail word was learned – Speed of recollec on – Experimenters used these es mates to increase/ • Big area of interest decrease study of poorly/well-learned words – Experimental group > control group
14