Post on 05-Feb-2016
description
transcript
Psyco 350 Lec #6 – Slide 1
Lecture 6 – Psyco 350, B1Winter, 2011
N. R. Brown
Psyco 350 Lec #6 – Slide 2
Outline
• An Alternative Perspectives on WM– Reading& Operation Span tasks
– Cowan’s Embedded Processes Mode
– WM Capacity as Executive Control
• LTM: Episodic-Semantic Distinction• Memory Basics: Concepts & Associations (nodes &
links)• Factors that influence Storage:
– Rehearsal • maintenance vs elaborate• massed vs spaced
Psyco 350 Lec #6 – Slide 3
Alternatives Perspectives On WM
Motivated by:• Problems w/ Baddeley’s Model• A need to better understand executive
functioning• Predictive power of span task.Three Related Issue1. Reading/Operation Span as a measure of
“capacity”2. WM contents as the active portion of LTM3. WM as executive attention
Psyco 350 Lec #5 – Slide 4
Measuring WM Capacity
• Key Idea: Performance on complex cognitive task reflects a number of different capacities– retrieval efficiency
– processing efficiency
– “attention-free” capacity of relevant slave system
– attentional management (ability to focus on relevant info & inhibit irrelevant info.
– etc.
• WM span tasks developed to measure relation between WM and performance on complex cognitive tasks
Psyco 350 Lec #6 – Slide 5
WM Memory Span
• WM span = # of words recalled
• Demonstrates capacity for holding load while processing.
• Large individual differences in WM span (2-6 items)
• WM span measures predict performance on IQ, achievement tests (e.g., SATs), & g.
• digit/word span uncorrelated with IQ/SAT tests
Psyco 350 Lec #6 – Slide 6
WM Memory Span – Two Interpretations
1. Domain Specific Capacity: Efficient processing of immediate task, leaves additional resources for maintaining load.
– Accounts for dual task performance (e.g. Brooks)
2. Domain General Capacity: General ability to “control attention to maintain information in an active quickly retrievable state.” Engle, 2009, p. 20.
Psyco 350 Lec #6 – Slide 7
An Alternative: WM as Information in an Active State
Psyco 350 Lec #6 – Slide 8
WM as Activation: Cowan’s Embedded Processes Model
• Central Executive: directs and controls voluntary processing.
• Encoding:– Incoming info activities
representation in LTM
Psyco 350 Lec #6 – Slide 9
WM as Activation: Cowan’s Embedded Processes Model
Central Notation:
LTM in one of 3 states:
1. Dormant
2. Activated – fades (decays) unless
reactivated
3. “In focus” (of attention)– limited to 4 items
Psyco 350 Lec #6 – Slide 10
Span from Activation Perspective
Two components:1. read-out from focus
2. activated material, retrieved before decay
Predictions:• factors LTM, span
– concreteness (Walker & Hulme, 1999)– word frequency (Roodenrys & Quinlan, 2000)
• Span > 0 when rehearsal suppressed
Psyco 350 Lec #6 – Slide 11
WM Memory Span – Competing Interpretations
1. Domain Specific Capacity: Efficient processing of immediate task, leaves additional resources for maintaining load. Accounts for Dual-task
2. Domain General Capacity: General ability to “control attention to maintain information in an active quickly retrievable state.” Engle, 2009, p. 20.
Psyco 350 Lec #6 – Slide 12
Span as Executive Control
Main Idea:
• Active, irrelevant info infers w/ performance
• People differ in their ability to inhibited irrelevant info or remain focused on relevant info
Prediction:
Individual differences in reading/operation span predict performance on tasks that require executive control.
Psyco 350 Lec #6 – Slide 13
Three Lines of EvidenceSpan & Proactive Interference
Participants:• High-span • Low-spanMaterials & procedure:• 3 10-word lists (words from same category)• 2 s/word• 16 s Delay• 20 s recall periods
Psyco 350 Lec #6 – Slide 14
Three Lines of EvidenceSpan & Proactive Interference
Results:
• Proactive Interference
• PI worse for Low-span
Psyco 350 Lec #6 – Slide 15
Three Lines of EvidenceSpan & Anti-saccade TasK
Task: Move eyes away from cue to find taget
Measures: Distractibility
Main Finding:
Low Span worse (more distractible) than High-span
Psyco 350 Lec #6 – Slide 16
Three Lines of EvidenceSpan & The Stroop Task
• StroopTask – Name Color of font
Psyco 350 Lec #6 – Slide 17
Three Lines of EvidenceSpan & The Stroop Task
• StroopTask – Name color of font
• General Finding: Incongruent trials (BLUE) slower than congruent trials (GREEN)
• Interpretation: Response slowed because attention required to inhibit color name.
Psyco 350 Lec #6 – Slide 18
Three Lines of EvidenceSpan & The Stroop Task
Participants:Low Span Vs High SP
Materials:Congruence: 0%, 50%, 75%
Difficulty increases w/ congruence
Psyco 350 Lec #6 – Slide 19
Three Lines of EvidenceSpan & The Stroop Task
Results:
• Difficulty increases w/ congruence
• Low-span worse than High-span
• Effect increase w/ congruence
Interpretation:
Inhibitory capacity particularly important when task is difficult.
Psyco 350 Lec #6 – Slide 20
WM Capacity as Executive Attention Engle (2009)
Main Claim:
WM capacity (as measure by reading & operation span) reflect individual differences in ability inhibit/exclude competing (partially active) sources of information.
Psyco 350 Lec #6 – Slide 21
Summary: Alternative View
Contents of WM
Active portion of LTM
Differences in Capacity
Reflects individual differences in ability to remain focused and inhibit irrelevant info.
Reflect in LTM and task perceptual tasks
Psyco 350 Lec #6 – Slide 22
Conclusions
• No pure measure of STM– Contents: focal info + activated LTM
• Covert Rehearsal: – one way of keeping info active
• Functional importance: – WM provides ability to access info and maintain
in active state required for thought, language, problem-solving, etc
Psyco 350 Lec #6 – Slide 23
Procedural • Knowing how
Declarative• Knowing that
Memory
Systems
DeclarativeProcedural
Episodic Semantic
Psyco 350 Lec #6 – Slide 24
Episodic-Semantic Distinction: Tulving (1972)
Episodic Memory:
• autobiographical
• temporally dated
• interference from similar episodes
• retrieval also serves as input
(i.e., episodic memory is continually being updated)
Psyco 350 Lec #6 – Slide 25
Episodic-Semantic Distinction Tulving (1972)
Semantic Memory:
• language (lexical memory)
• world knowledge
• not temporally dated
• very well-organized (organization protects memory from interference)
• not changed or modified by retrieval of information (i.e., not continually updated)
Psyco 350 Lec #6 – Slide 26
(Episodic) Memory Basics
• Content episodic memory– representation of specific events
• mini-events – “the work BOOK on List 2”• real events – “Diving in Cozumel”
• An event representation (ER) = a particular instantiation of a configuration of intersecting concepts
Psyco 350 Lec #6 – Slide 27
Memory Trace: “the word BOOK on List 2”
Event34272
BOOK List 2
WM
SM
EM
Psyco 350 Lec #6 – Slide 28
Memory Trace: “Diving in Cozumel”
Event20342
Cozumel Diving
WM
SM
EP
Psyco 350 Lec #6 – Slide 29
(Episodic) Memory Basics
• Links between ER & concepts differ in strength (as do the links between concepts).
• Strength depends of:– frequency
– contiguity
– recency
Psyco 350 Lec #6 – Slide 30
(Episodic) Memory Basics
• Cues in WM access conceptual information in LTM.
• Origin of cues:– environment– self-generated– provided by experimenter
• Concepts “spread activation” to linked nodes (other concepts & ERs)
• ERn activation > threshold, “ERn retrieved”
Psyco 350 Lec #6 – Slide 31
Cue word: “Cozumel”
Event20342
Cozumel Diving
WM
SM
EP
“Cozumel”
Psyco 350 Lec #6 – Slide 32
(Episodic) Memory Basic: Implications
• The stronger the link between a cued concept and an ER, the greater probability that the ER will be recalled.
[non-elaborative rehearsal]
Psyco 350 Lec #6 – Slide 33
Cue word: Cozumel w/ strong link
Event20342
Cozumel Diving
WM
SM
EP
“Cozumel”
Psyco 350 Lec #6 – Slide 34
(Episodic) Memory Basic: Implications
• The more ER-to-concept links there are, the greater the probability that a given cue will serve as an effective retrieval cue. [elaboration, depth or processing]
Psyco 350 Lec #6 – Slide 35
Cue words: “Cozumel” “diving”
Event20342
Cozumel Diving
WM
SM
EP
“hurricane”
hurricane
Psyco 350 Lec #6 – Slide 36
(Episodic) Memory Basic: Implications
• Context (internal & external) is encoded as part of the ER, and thus contextual features can serve as retrieval cues.
• Increasing the similarity between encoding contexts and retrieval contexts increases the probability of retrieval. [context effects, TAP]
Psyco 350 Lec #6 – Slide 37
Contextual Cue
Event20342
Cozumel Diving
WM
SM
EP
taste of t. scrimp
tequila scrimp
Psyco 350 Lec #6 – Slide 38
(Episodic) Memory Basic: Implications
• Probability of recall decreases, as # of ERs linked to a cued concept increases. [interference]
Psyco 350 Lec #6 – Slide 39
Cue word: “diving” interference
Event20342
Cozumel Diving
WM
SM
EP
“diving”
Grand Banks
Event5632
Psyco 350 Lec #6 – Slide 40
Memory Processes
• Encoding:– process of storing information in memory
• Storage:– the retention (& loss) of information over time
• Retrieval:– recovery of previously stored information
Psyco 350 Lec #6 – Slide 41
Encoding Processes: Rehearsal
• Rehearsal: a set of techniques/strategies for encoding information into long-term memory
• Two kinds of rehearsal:
– Maintenance: keeps information "alive" in WM; rote recycling; little effect on LTM
– Elaboration: "promotes" information to LTM; think about and connect
Psyco 350 Lec #6 – Slide 42
Maintenance Rehearsal:Craik and Watkins (1973)
Task: • Monitor auditory list for words beginning w/ target
letter (e.g., G)• required to recall “last” target word at end of list (list
could end at any time).
• Session final recall – recall as many words as possible.
Manipulation:# of words between appearance of target word (assumed equal to amount of rehearsal).
Psyco 350 Lec #6 – Slide 43
Craik & Watkins (1973)
Results:• recall unaffected by #
of rehearsals.Interpretation:
rote (maintenance) rehearsal fails to transfer info to LTM.
Question:• What about Rundus
(1971)? • Maintenance +
(elaboration, relational coding, etc)
Psyco 350 Lec #6 – Slide 44
Rundus (1971) again
• Analysis:– # rehearsals for each
word (position)
– % recall for each word (position)
• Results:– “For a given amount of
rehearsal, items from the initial serial positions are no better recalled than items from the middle of the list” – Rundus, 1971, p. 66
Psyco 350 Lec #6 – Slide 45
Elaborative Encoding: Bradshaw & Anderson(1982)
Tasks:
Recall “target” fact
Design:
Encoding Context X Delaytarget only immediate (no delay)
target + 2 irrelevant facts 1 week
target + 2 relevant facts
Psyco 350 Lec #6 – Slide 46
Elaborative Encoding: Bradshaw & Anderson(1982)
Target only:
Newton became emotionally unstable and insecure as child.
Target+2 irrelevant facts:
Locke was unhappy as a student at Westminster.plus
Locke felt fruits were unwholesome for children.
Locke had a long history of back problems.
Target+2 relevant facts
Mozart made a long journey form Munich to Paris.plus
Mozart wanted to leave Munich to avoid a romantic entanglement.
Mozart was intrigued by musical developments in Paris.
Psyco 350 Lec #6 – Slide 47
Bradshaw & Anderson(1982): Results
• Encoding relevant facts improved recall.• Encoding irrelevant facts hampered recall.• Effect was magnified by delay.Interpretation:• relevant elaboration increases # of retrieval path• relevant facts fosters generation of appropriate cues• irrelevant facts cause interference
Psyco 350 Lec #6 – Slide 48
Spacing Effects
• Issue:– Rehearsal improves memory.
– Does the temporal distribution of rehearsals matter?
• Research Strategy– holding # of presentations constant, manipulate
the lag (delay) between presentations.
Psyco 350 Lec #6 – Slide 49
Madigan (1969): Spacing Effect• Method:
– words studied twice at 6 different lags.
• Results:– recall w/ lag
• Other findings:– spacing affects
recognition (Glenberg, 1979)
– obtained w/ textbook materials
Psyco 350 Lec #6 – Slide 50
Accounting for the Spacing Effect• Deficient Processing
– habituate to recently presented material (less “interesting”)
– short lags overestimation of learning --> decreases (or redistributed) rehearsals.
• Encoding Variability:– Recall depends in part on study context matching test
context. – Context changes w/ time– the greater the lag, the more different the encoding
contexts, and therefore the more likely that one of them will overlap with the test context.