episodes, time, and the structure of human memory (!) on the agenda: –finish discussion of...
DESCRIPTION
CONTROL and CAPACITY in INFORMATION I/O STM Encoding –Voluntary selection among inputs Partial report tasks –Some contact with semantic LTM in absence of such attention “priming” without awareness (e.g., Marcel, ’70?) –Little apparent “central” interference Free recall recency not affected by card sorting (Murdock, ’65) LTM Encoding –Voluntary selection and selective rehearsal –Elaborative encoding –Great “central” interference Tone decisions impair memory for concurrent events (Carrier & Pashler, 98?)TRANSCRIPT
Episodes, Time, and the Structure of Human Memory (!)
• On the agenda:– Finish discussion of
measurement in memory– Present results of search– Information-processing models
of human memory (Pashler)– Approaches to temporality– Binding elements of events into
episodes• Baddeley’s Episodic Buffer• Craik’s work on binding of
content and context– Settle on presentation
assignments
MEMORY STRUCTURESand PROCESSES
• The information-processing approach and its tools– Elements of Pashler’s model
• Functional level of constructs• Multiple sensory and STM
systems• Dissociative evidence• Tracking the flow of information
– Attentional gating of sensory input– Dual-task / RT methodology
• The Central Processor– Recoding in STM– I/O for episodic memory– Selecting actions and responses
CONTROL and CAPACITY inINFORMATION I/O
• STM Encoding– Voluntary selection among inputs
• Partial report tasks– Some contact with semantic LTM
in absence of such attention• “priming” without awareness
(e.g., Marcel, ’70?)– Little apparent “central”
interference• Free recall recency not affected
by card sorting (Murdock, ’65)• LTM Encoding
– Voluntary selection and selective rehearsal
– Elaborative encoding– Great “central” interference
• Tone decisions impair memory for concurrent events (Carrier & Pashler, 98?)
CONTROL and CAPACITY inINFORMATION I/O (concl)
• LTM Retrieval– Central demands can interfere
• Interference with retrieval from a speeded choice task (Carrier & Pashler, ’95)– Sequential task design– Use of Pschological Refractory
Period (PRP)– RT1 to tone; RT2 cued recall or
recognition– RT2 slowed by short SOA– Slowing is additive with retrieval
factors• May be limited to “strategic”
retrieval (Baddeley)
Carrier & Pashler 1995
• Pashler’s I/O model of Memory– Strong multi-system flavor
• Sensory, STM, LTM• Multiple STM systems
– Capacity and control limits at various stages• Selective attention and
bottlenecks• Capacity limits and the Central
Processor– Contrasts to other multistage
models• Atkinson & Shiffrin 1968• Bjork 1975• Potter 1983• Baddeley 1986
CREATING EPISODES:The Canvas of Time
• The issue of time perception– Relative contribution of attention,
events, and “biological time”– Speculations about bioclocks
• E.g., as marking of synchrony in striatum (Matell & Meck, ’00)
• Time perception (cont’d)– Evidence from patients
• Damasio, Tranel & Jones:– Cued recall of personal and public
events– Placement of events on “time line”– Temporal amnesics:
• Poor event recall, OK placement– Basal forebrain amnesics:
• Fair event recall, poor placement– Recent imaging evidence
• Rao, Mayer & Harrington ’00– Tone-pair standard 1200 ms– 1 sec IPI– Tone-pair time or pitch changed– Early activation of putamen, caudate
(basal ganglia)– Later activation of cerebellum
– Parkinsonism and the timer• Underestimation of intervals
(Malapani, ’98)
• Time perception (concl)– Attentional modulation of time
estimation• Non-temporal central tasks impair
accuracy of both prospective and retrospective interval estimates (S. Brown, ’85)
• General pattern is for underestimation of interval duration
Features, Objects,Events and Episodes: The
Binding Problem(s)• Hierarchical structure of episodes• Feature analysis and integration
– Feature Integration Theory (Triesman)• Attention required (mostly) for
search for conjunctions (Triesman & Gelade ’80)
Distraction can produce “illusory conjunctions” of features
T T T T T T T T T T
T T T F T T T T T T
T T T T F T F T T F
• Multimodal integration and temporal sequencing– The Episodic Buffer (Baddeley
’00)• The “classic” WM model• Problems
– Concurrent articulation, impaired phono-STM, leaves visual digit span of c. 4, not 1
– Impaired phono-STM patient has “sentence span” of 5
– Maintenance of complex images in WM
• The revised model:
• Interfering with the Episodic Buffer– Jeffries, Ralph & Baddeley (JML,
2004)• Dual-task approach• Immediate serial recall of words• Random words or unrelated
sentences• Alone or with concurrent visual 4-
choice RT (x z . /)
Word recall at 50% above span
0102030405060708090
100
Words Sentences Stories
Type of stimulus
Per
cent
Cor
rect
Single
Dual
• Neural basis of integrated WM– Prabhakaran ’00 fMRI study
• Integrated maintenance versus separate maintenance tasks– Letter sequence– Location of x’s– Location of letters
• Match location or letter• Match letter-in-location
• Only in last case, sees right prefrontal cortex activation
• Unintegrated maintenance shows more posterior activity
• Binding events to their context– Craik ’89: Making of episodes
• Importance of the “spatiotemporal context” of objects and actions
• Does binding of event to context require attention?– Present items (words, pictures)– In varied contexts (scenes, voices)– Manipulate attention– Track memory for items and context
• Generally, tight coupling of item and context memory
• Suggestion that attention plays a greater role for “poorly integrated” item-context associations at retrieval
• And leads to better integration at study
Context identification
Item
reco
gniti
on
.1 .3 .5 .7 .9
.1
.3
.5
.
7
. 9