Immediate Memory
______
1) Provide a historical perspective on the study of the conscious portion of memory.
2) Present the key components of the ‘modal’ model of memory (Atkinson & Shiffrin) along with the empirical data that support/contradict the model.
3) Discuss several reasons why researchers have become dissatisfied with the ‘modal’ model.
4) Describe the architecture of the WM model and the empirical data that support/contradict the model.
- Central executive
- Phonological loop
- Visuo-spatial sketchpad
5) Review two recent attempts to update the WM model:
- Activation models
- Feature models
Historical Perspectives on Immediate Memory
______
Different theories … different names
- Working Memory
- working memory
- STM
- primary memory
Theoretically neutral terms
- Immediate memory
- Contents of consciousness
______
Out with the old, in with the new
- Unitary vs. multi-faceted construct
- Associationism vs. strategic behavior
- Behaviors vs. peering inside the box
Limited Capacity of Immediate Memory
______
Demo:
______
Q: how much information can it hold?
A1: 7 +/- 2 items
A2:
Q: Why do capacity constraints matter?
A:
Q: But what would be sufficient proof?
- Different properties
- Respond differently to experimental variables
- Make separate contributions to memory
Let’s give it a name
______
Conscious / UnconsciousWaugh & Norman / Primary / Secondary
Atkinson & Shiffrin / STM / LTM
We are going to use the A&S terminology because:
- STM and LTM captures an important distinction between the two proposed memory systems
- A&S introduced terminology to describe the subjects mental activity
- rehearsal
- search
- transfer
Distinguishing STM & LTM: Different Properties I
______
STM / LTMCapacity / 7 2 / functionally infinite
Duration / relatively brief / relatively long
Serial Position Effects / Recency / Primacy
Supportive Data for Serial Position Effects:
- Rundus & Atkinson (1970) – # of rehearsals correlated with primacy, not recency.
Distinguishing STM & LTM: Different Properties II
______
More on Serial Positions Effects:
Glanzer & Cunitz (1966) – distractor tasks eliminate recency; no effect on primacy.
Distinguishing STM & LTM: Different Properties III
______
Maintenance of information
STMrehearsal
EX: new cutie’s phone #
LTMrehearsal is not necessary
EX: current email address
old email addresses
Effect of Decay
STMRapid
LTMSlow, but measurable
Effect of interference
STMhuge
EX: new cuties’ phone #
LTMrelatively minor
EX: current email address
But: old phone #s / email addresses
Distinguishing STM & LTM: Experimental Manipulations and Separate Contributions
______
Experimental Manipulations
What does it take to eliminate primacy and recency?
- How can you get rid of primacy?
A: Presentation rate
- How can you get rid of recency?
A: Delay recall (Craik, 1970)
Confusion errors
STM (short RI, lists)phonological confusions
LTM (long RI, lists) semantic confusions
Separability of function
Two letters: HM
- Normal LTM
- Highly impaired STM (across delays)
- Specific implications
- Transfer
- Neural correlate
Critics of the two-store model
______
Primacy / Recency effects
- Primacy without rehearsal
- Recency with RIs of several weeks
- Continual distractor paradigm
- Bjork & Whitten (1974)
Is rehearsal necessarily correlated with memory?
Confusion errors are not so cut and dried
- STM - some semantic confusions
- LTM - some phonological confusions
True separability is logically impossible
- Semanticity affects STM
- LTM retrieval must involve STM
The next step: Working Memory
Baddeley & Hitch (1974)
______
Q: What made modal model revolutionary?
A: introduction of terminology to describe covert mental activities like rehearsal, storage, transfer,
Working Memory (Baddeley & Hitch, 1974)
Emphasized processes over contents
Why jump on the WM bandwagon?
- Digit spanvs. working memory span
- Builds on the advantages of the modal model
- More in step with cognitive revolution
- Can I remember new cutie’s phone # vs.
- Is it safe to cross Rte. 9?
- Reactive parenting
- Mental wandering
Architecture of the Working Memory model
______
Theoretical approach
Immediate memory= workbench of the mind
- What do you need to build a house?
- General Contractor
- Expertise
- Workers
- Tools
- Raw Materials
Working Memory
- Central Executive
- Strategies, experience
- Phonological Loop
- Phonological store
- Articulatory control process
- Visuo-spatial sketch pad
- Mental capacity / energy (limited)
Working memory: Overview
______
The CE co-ordinates all mental activity.
- How the problem will be solved: sub-steps needed to arrive at a solution
- Schedules tasks
- Allocates limited resources
- Chooses which subsystems to employ: what each subsystem will do
- How to efficiently co-ordinate different tasks to maximize available resources
Two lackies:
Phonological Loop
Auditory / linguistic information
VSSP
Visual information
Working memory: Behavioral support for the
existence of the Central Executive
______
Concurrent task paradigm –
Primary task: mental arithmetic
Function of CE?
Secondary tasks:
- random sequence generation
- abstract line drawings
- simple motor sequence.
Typical results:
Which has the biggest dual task cost?
Problems:
circularity
Relation to personal experience:
cognitive failures
Working memory: Neuropsych support for the
existence of the Central Executive
______
Dysexecutive syndrome – Frontal lobe damage
Case study
- EVR –lost decision making ability
Behavioral tasks
- Wisconsin Card Sort
- Verbal fluency tasks
- Copying simple motor behaviors
- Stroop task
Two classes of errors
- Perseveration
- inferior and orbital frontal cortex
- Distraction
- dorsolateral frontal cortex
Neuroimaging data
Frontal lobe activation consistently tied to adoption of strategic behavior
- Detecting novelty
- Difficult conditions of the Stroop
Working Memory: Training the Central Executive
______
Space Fortress improves with practice.
Big Question: How do we know this relates to CE?
A: Because Space Fortress training improve pilot performance
Bigger Question: What does Space Fortress have to do with Piloting?
A: Nothing, except…the ability to pay attention to multiple streams of information simultaneously and to co-ordinate a variety of actions simultaneously.
Biggest Question: Why is this such a big question?
A: Because ultimately, your evaluation of the CE is tied to the extent to which you see a connection between piloting and Space Fortress and medical decision making, and Stroop, and crossing the street.
Working Memory: Phonological Loop
______
Phonological store
Function: temporarily holds auditory input
Articulatory control process
Function 1: converts visual information into articulatory code
Function 2: refreshes memory traces in Phonological Store via subvocal articulation (i.e., talking to yourself)
Capacity of Phonological loop
Decay rateFixed
Rehearsal ratefunction of both stimulus and personal factors
Demo: Articulation rate
Phonological loop: what is it good for?
______
Phonological similarity effect – immediate memory is worse for similar sounding items than for items that sound different.
Why?
Word-length effect – longer words are more difficult to remember. Demo
Why?
Articulatory suppression – repeating a nonsense syllable ‘bap’ impairs memory.
Why?
Secondary prediction: larger effect on visual presentation than auditory presentation. Why?
Irrelevant speech effect - Recall visually presented consonants either in silence or while listening to irrelevant speech
Prediction:
Explanation:
Effect of articulatory suppression:
Effect of irrelevant music:
Note: data are mixed
Martin, Wogalter, & Forlano (1988)
______
What was the goal of the research?
They wanted to examine whether unattended speech/music would affect reading comprehension.
Why were they interested in that?
- Does background noise mask inner speech?
- Is phonological representation necessary?
- Previous data were mixed
- Auditory vs. visual presentation
- Meaningfulness of stimuli
- verbatim recall or comprehension
Experiment 1: Speech vs. Music?
Method:
- continuous spoken speech
- random speech
- instrumental music
- random tones
- silence.
Results:
- Continuous and random speech worse than control
Martin, Wogalter, & Forlano (1988) II
______
Experiment 2 – Music vs. language
Method:
- Sung lyrics vs. spoken lyrics vs. no lyrics
- with or without musical accompaniment.
Results:
- Sung lyrics not different from spoken lyrics but both were worse than no lyrics.
Experiment 3 – speech more distracting in general?
Method:
- Speech vs. musical background
Results:
- Music more disruptive than speech
Experiment 4 – Semanticity?
Results:English < Russian = White noise
Experiment 5– Phonology?
- English non-words had no effect
______
What is the overall interpretation?
- Phonological interference does not affect comprehension
- Relation to WM model? Function of PL?
Balch and Lewis (1996)
______
Theoretical question: How does music produce CDM?
Place, mood, smells
Empirical question: How will manipulating various aspects of the music affect CDM?
Experiment 1:
Key manipulations:Melody and Tempo
Results:
- Changing melody did not reduce memory
- Changing tempo did
Experiment 2:
Key manipulations:Tempo and Timbre
Results:
- Changing timbre did not reduce memory
- Changing tempo did
Experiment 3:
Key manipulation: Timbre and Tempo
Results:
- Timbre change did not affect emotionality
- Tempo change did
Balch and Lewis (1996)
______
Experiment 4:
Key manipulation:
Tempo at encoding
Induced mood at test
Results:
Match facilitated memory even though
music was not presented at test
Implications:
Music influences memory by manipulating mood.
Mood explains everything.
______
Questions:
- What are the implications of mood-dependent memory for how memory functions?
- So, you and your roommate are studying for your Memory exam. S/he says, do you mind if I turn on the stereo? What do you say?
- What do you think about IMing, texting, and so forth while studying?
- How should our legal institutions resolve the issue of driving while talking on the cell phone?
Fürst and Hitch (2000)
______
Theoretical Question: Do the phonological loop and central executive make independent contributions to strategic memory tasks?
Empirical Question: Will secondary tasks designed to stress the CE and PL produce differential performance deficits on a primary mental arithmetic task?
What are the three component processes of mental math?
Experiment 1:
Q: What does the PL do?
A: temporarily stores articulatory information
Q: How would you interfere with the PL?
A: articulatory suppression
Method:
- Articulatory suppression or silence
- Math problem remained in view or not
Results:
- AS interfered with performance, but only if the problem was removed from view.
Interpretation:
- PL holds temporary results of calculations
Fürst and Hitch (2000): What about the CE?
______
Experiment 2:
Q: How does the CE contribute to mental math?
A: Strategic behavior; specifically: carrying
Q: How did they tax the CE?
A: Trails task
Method:
Completed trails task with visible math problems
Results:
- Dual task decreased math performance
- Vast majority of errors were carries
- Trails task varied inversely with # of carries
Experiment 3:
Q: Did # of carries influence trails performance because carrying takes time?
A: Length of trails task unrelated to # of errors
Dual task influenced errors on trail task
Fürst and Hitch (2000): What does it all mean?
______
Implications:
- PL: retains needed problem information
- CE: strategic behavior (carrying)
- Independent functions constitutes convincing evidence for WM architecture
Questions:
- How can we explain discrepancies with previous research regarding the role of WM in carrying during mental arithmetic?
- Forgetting a carry common; mistakenly including a carry was uncommon. What does that imply?
- Are the functions of the PL and CE completely independent?
Critiques of Working Memory
______
Consistency
- Word length effects
- Irrelevant speech effects
- Is articulatory rate related to speaking rate?
- Neuropsych data:
If reduced spans caused by faulty ACP, then should be immune to articulatory suppression. Some are, some are not.
Model is overly descriptive
EX: Why are phonologically similar items more confusable than phonologically distinct items?
Qualitative rather than quantitative
- Makes it difficult to falsify
______
Overall evaluation
- Extremely valuable
- Needs further development / specification
What else is out there? Feature models
______
Feature model (Nairne, 1990)
- Similar to the way a computer represents information (binary code)
How might you represent a person?
Important features of Feature Models
______
One data point is meaningless
- It's the overall pattern
- Thus, you don't need many features to distinguish between objects ( 20).
320 = 3.5 billion
Two kinds of features
- Modality-dependent
- Modality-independent.
- Note: feature models do not posit different stores for different kinds of information.
How do feature models work?
______
Distinction between primary and secondary memory
Primary memoryconsciousness.
Secondary memorystorehouse of knowledge
Primary functions
Primary Memory:
- assemble and maintain cues that will aid subsequent memory searches.
Secondary Memory:
- permanent storage of knowledge
How does memory work?
Encoding:
1st items features are activated
2nd items features are activated
- overlap and interfere with 1st item
3rd items features are activated
- overlap / interfere with items 1 and 2
Test:
Match degraded cues with items in secondary memory
How do Feature Models account for different effects?
______
Recency
Features of final object are not overwritten
Suffix Effect
Features of final item ARE overwritten
Phonological similarity
Similar items have more overlapping features. Therefore, will be harder to distinguish from one another at test
Articulatory suppression
Features of the repeated item are incorporated into the cue used to retrieve each item. Therefore, items are less discriminable from one another.
Word-length effect
The more features, there are, the more opportunities there are to make mistakes.
EX: 5-piece jigsaw puzzle vs.
100-piece jigsaw puzzle
Evaluation of Feature Models
______
Negatives:
weak on SP effects
Positives:
Much more explicit model than WM
Question to ponder:
How could you use feature models to simulate the advantage of spaced over massed practice?