Memory

Sensory register: Brief duration, huge capacity, separate stores. 5 stores for senses. Visual iconic --> brief image --> cause of after image. Auditory echoic --> holds sound for up to 4 seconds. Attention --> STM. Rehearsal --> semantic meaning.
STM: Short duration (18-30 seconds), 7+/-2 capacity, acoustic encoding.
LTM: Long duration (forever?), limitless capacity, mainly semantic encoding.

Miller: Chunking increases STM capacity. Simon: size of each chunk important. 

Digit span technique: Measures STM capacity. Increasingly larger (+1) lists of numbers, asked to recall in serial order after each. Lack of consistency across stimuli?

Primacy effect: Rehearse first items. Recency effect: End items remain in STM. 
Murdock serial position effect. PPs learnt list of words varying in length 10-40, free recall, presented for 1-2 seconds, probability of recall depended on position. Mundane realism? Artificial, lab.
Glanzer and Cunitz recency effect. 2 groups of PPs, one recalled words immediately, one after 30 seconds counting back in 3s. Free recall, words at end only recalled if tested immediately, didn't influence primacy effect. Lab, mundane realism?

Atkinson and Shiffrin

Sensory store --Attention--> STM --Elaborative rehearsal--> LTM 

Maintenance rehearsal retains information in STM. 

24 students, measures STM duration. Nonsense trigrams (ecological validity?), count back in 3s aloud (prevent maintenance rehearsal). Incrasing time delays, maximum time = STM duration. 18-30 seconds.

Opportunity sample of 392 American ex-high school students aged 17-74. Independent group design. 

1. Free recall names

2. Photo recognition (identify former classmates)

3. Name recognition

4. Name and photo matching

90% face and name recognition even after 34 years. 48 years - declined to 80% name, 40% face. Free recall less accurate - 60% for 15 years, 30% for 48 years. 

Cues. Only demonstrated VLTM for particular type of information. Mundane realism high. Ecological validity - reasonable (task, but lab experiment). Good control. Limited application.

PPs presented with acoustically similar, acoustically dissimilar, semantically similar and semantically dissimilar words. 75 PPs heard 1 list repeated 4 times, then had to arrange words in order. 

Acoustic confusion in STM (10% recall) - worst = main form of encoding. LTM = semantic.

Makes sense - acoustically rehearse when trying to remember. 

Large sample but any conditions. British volunteers. Standardised. Baddeley and Hitch built on research developing working memory model. Word order not words themselves - internal validity? (Remember words for associations? Ecological validity - order, artificial.

HM: Removed hippocampus for epilepsy, STM normal but couldn't transfer new information into LTM. Anterograde amnesia, supports idea of separate systems.

Clive Wearing: Viral infection. Procedural memories retained (could still play piano), couldn't store new LTM info or episodic/semantic. 

KF: Motorcycle accident. Could remember visual, not verbal (acoustic). 

Case studies: Detailed, rich data but not generalisable/replicable, time consuming and researcher bias. 

Evidence - case studies, experiments. Most accept separate stores. 

STM and LTM different sub-stores? Chunking into meaningful units can only occur if STM knows meaningful unit --> STM draw from LTM? Passive, linear, oversimplified. 

Tulving: Rehearsal less important than suggested. Spent time rehearsing words, no more likely to recall than 'new' words.

Model of STM, not unitary, 'slave systems.' Baddeley and Hitch. 

Central executive - control, decides which system required. Co-ordinates information retrieval from LTM. Decides what working memory pays attention to.

Visuo-spatial sketchpad - layers of information relative to each other, analyse, manipulate stores. Also in LTM. 

Phonological loop - holds spokn words for 1.5-2 seconds. Spoken - direct. Written - converted into articulatory. Capacity - remember more short words (Baddeley's word length effect). Articulatory control process - inner voice circulates info, repeat, remember.

Episodic buffer - needed to explain amnesics remembering book passages despite no LTM. Connects to LTM. 

STM subsystems, experimental evidence. Baddeley's word length effect supports phonological loop. Brain scan studies found different brain regions to activate for different tasks/components. Rehearsal optional, other routes (visual stimuli, episodic buffer). Explains processing as well as storage. Brain damaged patients 'LH' - more difficulty with visuo than spatial tasks, 'KF' - forgot more auditory than visual. 

Little evidence for central executive other than the fact it makes sense. Brain damaged patients - can't accedd STM prior? Too simple? - does not explain full range of day to day phenomena. Central executive poorly understood and unlikely unitary. Understanding of episodic buffer?

Procedural - non-declarative, action.

Semantic - declarative, aware, general knowledge.

Episodic - declarative, aware, events in experience.

Tulving: Separate memory stores for episodic and semantic? Injected radioactive gold into himself, thought semantic (historical fats) or episodic (events from holidays). Showed different brain patterns. Episodic - front, semantic - back. Supports different LTM memory stores. 

Amnesic patients typically have episodic/semantic from before but cannot store new, and have procedural intact. Spiers et al, 147 amnesics. 

Unitary? No, HM learnt to play tennis. 

STM explanations: 

• Displacement (capacity): If information keeps entering, becomes displaced. 
• Decay (duration): Automatic over time. 

LTM explanations:

• Interference (from previous or future). Retroactive: recent interferes with earlier learning. Proactive: earlier learning interferes with present. Postman: PPs in 2 groups, both remember paired words e.g. cat-glass, one had to learn another in which 2nd word different e.g. cat-tree. Struggled. Artificial, low ecological validity, unnatural.
• Retrieval failure: LTM cannot be accessed, available but not accesibly. Cues needed, state or context dependent or external environment e.g. driving test in same car learnt in. e.g. emotional, mood, drunk. Baddeley: deep sea divers, half on beach half on land then split again, remmebered more if recalled where learnt, context. Field experiment, practical uses in education, mundane realism (random info?). Context dependent not necessary to be in same environment but remembering.

Retroactive interference: recent interferes with old. Proactive: old interferes with recent. 

Anterograde amnesia: Can't form new memories. Retrograde amnesia: Loss of memory. 

Post-event discussion, leading questions, anxiety.

45 American students shown film of traffic accident. IV - different verbs. Smashed - 40.8mph, collided - 39.3, bumped - 38.1, hit - 34, contacted - 31.8. Verb altered perceptions. 

1. Response bias - didn't lead to false memory.

2. Memory representation altered - if true, would expect PPs to remember other false information. 

Follow up experiment: 150 students, 'hit' or 'smashed,' 1 week later - broken glass? (none). Smashed more likely - 16.7%. Conclusion: confabulation. 

Two kinds of memory go into an event. 1. Information obtained from perceiving it. 2. Other information supplied after integrated. 

Lacked mundane realism - videos. American students, less confident drivers? Replicable, control, wider implications. Practical application: jury shouldn't convict on EWT alone.

Yuille and Cutshall: Misleading information didn't alter memory of people who had witnessed real robbery.

Slides (artificial?) showing car accident. 

1. Red car stops by 'yield' sign. Half asked consistent, half inconsistent.

2. Same but for 'stop' sign. 

75% consistent picked correct slide, 41% misleading. When delayed, 20% (weak). 

Advertisement. 

1. No cartoon characters.

2. Carboard Bugs Bunny only.

3. Bugs Bunny ad. 

4. Bugs Bunny ad and figure. 

40% group 4 remembered meeting Bugs Bunny. More likely to relate him, used real life memory of event. Practical application in advertising.

Elderly more difficulty recalling source, more prone to misleading information.

Anxiety: Deffenbacher - meta-analysis of 18 studies, high levels impacted accuracy. Christianson and Hubinette - 58 bank robbery witnesses, those threatened more accurate.
Ginet and Verkampt - 'real' electrodes, higher recall. 
Titanic - real life application, 75% said it broke in half.
Yerkes-Dodson law between arousal and performance.

Loftus - greasy pen (49% accurate) or bloody knife (33%).

Peters - Asked to identify nurse (about to give injection)/researcher, researcher easier.

Need to improve effectiveness of police questioning and apply research.

• Reinstate context (cues)
• Change sequence
• Change perspective
• Report everything (insignificant details, but time consuming and lots of irrelevant information). 

Improves recall but not recognition (culprit?). Improves memory - remember but less accuracy? Dependent on skills of interviewer. 

Sensory register, STM, LTM, chunking, digit span technique - STM capacity, receny/primacy effects, serial position effect, Murdock, Glanzer and Cunitz. MSM - Atkinson and Shiffrin, Brown Peterson procedure - STM duration, Bahrick - LTM duration, Baddeley - encoding STM/LTM. HM, Clive Wearing, KF. 

WMM - Baddeley and Hitch, STM model. Central executive, visuo-spatial sketchpad, episodic buffer, LTM, phonological loop - articulatory control process. Subsystems. 

LTM - procedural, semantic, episodic, Tulving - separate semantic/episodic stores. Amnesics.

Forget - STM (decay, displacement), LTM - interference (Postman), retrieval failure (Baddeley). 

EWT - retroactive and proactive interference, anterograde and retrograde amnesia. Loftus and Palmer verb experiments, Loftus et al stop/yield experiment. Loftus and Pickrell advertising. 

Elderly, anxiety, weapon focus. Deffenbacher, Christianson/Hubinette, Yerkes-Dodson. Loftus, Peters.

Cognitive interview - context, perspective, sequence, everything.