Short Term or Working Memory

1
Short Term or Working Memory
2
Experimental Research

• the systematic manipulation and measurement of
  variables and observing their effects on one
  another in controlled settings
• allows for a causal statement to be made about
  the effect of one variable on another
• researchers manipulate factors in order to
  measure the effect on some mental processes
• independent variable the variable that is
  manipulated (factor)
• dependent variable the measured variable
  (outcome)
• Confounding Variables
• a factor that varies along with the independent
  variable, making the results difficult to
  interpret
• a researcher must ensure that the conditions
  being compared are as equivalent as possible

3
The Factorial Design

• completely crossing each level of one independent
  variable with each level of the other independent
  variable
• allows us to look at the joint effects of these
  variables (as well as the effect of each variable
  in isolation)
• types
• between-subjects factorial both variables are
  manipulated between-subjects
• within-subjects factorial both variables are
  manipulated within-subjects
• mixed factorial one of the variables would be
  manipulated between-subjects and the other would
  be manipulated within-subjects
• extremely common in cognition research

4
The Factorial Design, cont.

• A. Analyzing and Presenting Results
• descriptive statistics are used to provide a
  thumbnail sketch of the data
• measures of central tendency (the mean is the
  most commonly used) give an idea of the typical
  score for a given condition
• inferential statistics allow a researcher to
  determine how likely it is that a difference
  between conditions occurred due to chance
• if it is unlikely that the difference occurred by
  chance, the inference is that it must have
  occurred due to the differences in the levels of
  the independent variable (achieved statistical
  significance)

5
The Factorial Design, cont.

• Main Effects
• a statistical comparison of the overall means for
  the levels of each independent variable
• 2. Interactions
• a statistical test to determine if the
  independent variables influence each other
• might be described as a "difference of
  differences"

Note error ?
6
The Factorial Design, cont.

• 2. Interactions , cont.
• example memory test/word frequency experiment
• in recall there is a 14 advantage for
  high-frequency words in recognition, the
  difference is reversed, with a 24 advantage for
  low-frequency words
• recognition is superior to recall overall, but
  the superiority is much more evident in the case
  of low-frequency words
• 54 for low-frequency words, but only 16 for
  high-frequency words

7
Ch 4 Attention, Automaticity, Working Memory

• I. Attention, Theories of
• 1. Attention as a Gateway
• 2. Attention as Capacity
• 3. Multimode Theory of Attention
• II. Automaticity
• A. The Stroop Effect
• B. Automaticity with Practice
• C. Re-evaluating Automaticity
• D. Costs of Automaticity Action Slips
• III. Short-term Memory
• A. Limited Duration
• B. Limited Capacity Chunking and Word Length
• C. Coding in STM
• D. Forgetting in STM Decay Interference
• IV. A Modular Approach to STM Working Memory
• A. The Articulatory Loop and Articulatory
  Suppression
• B. Visuo-Spatial Sketchpad
• C. Central Executive
• D. Evaluation of Working Memory

COGNITION AND CONSCIOUSNESS Failure of
Early-Selection Theory COGNITION AND INDIVIDUAL
DIFFERENCES Sex Differences in Visual
STM COGNITION AND NEUROSCIENCE PETscanning the
Articulatory Loop
8
III. Short-term Memory

• Limited Duration
• Brown-Peterson task
• participants receive a brief presentation of a
  consonant trigram
  (e.g., JDL), immediately followed by presentation
  of a 3-digit number
• participants must count backwards by threes from
  the digit (preventing rehearsal)
• purpose is to prevent the participants from
  thinking about the material
• within 20 seconds, the probability of recalling
  the trigram was only 10
• 2. Limited Capacity (limited memory span)
• the "magical number 7 2"
• assessed through memory span--the longest string
  of digits (or some other simple stimulus) that a
  person can immediately recall
• Chunking in STM
• we can functionally increase the limits by
  recoding incoming information, combining it into
  larger and larger chunks
• ability to chunk is impacted by a number of
  variables
• As STM has relatively brief duration, rate of
  presentation is limiting factor
• affected by knowledge base previous knowledge
  aids in your own personal reorganization of the
  information

9
Chunking and Word Length

• Chase and Ericsson (1985)
• Studied influence of knowledge on STM
• participant spent two years practicing his STM
• read a digit sequence at a rate of one digit per
  second and then tested himself on the sequence
• if recalled perfectly he increased the sequence
  by one digit
• increased memory span to 80 items
• he was an expert long distance runner and grouped
  digits as running times
• The Effects of Word Length
• retaining 7 2 long items (e.g., hippopotamus)
  in STM proves to be much more difficult than
  retaining 7 2 short items (e.g., cat)
• Illustrates the time-based nature of STM (like a
  short tape loop)

10
Coding in STM

• auditory coding is the dominant mode of
  processing
• STM like inward ear
• supported by the phonological similarity effect
• lists of similar-sounding items are more
  difficult to keep track of in STM than are lists
  of different-sounding items
• even occurs when material is presented visually,
  strongly implying that visually-presented
  information is quickly converted into an auditory
  form
• visual coding also exists, especially so in
  children. inner eye
• Phonological similarity effect also influences
  visual coding!
• Brandimonte and Gerbino (1993)
• can mental images (i.e., visual codings) of these
  reversible figures can be ambiguous, and
  interpreted in two different ways
• employed reversible figures (pictures with two
  possible interpretations that can alternate as
  you're viewing it)

11
Brandimonte and Gerbino (1993) , cont.

• training phase
• trained participants with a couple of reversible
  figures in order to familiarize them with the
  reversals that could occur
• after the training phase, they were presented
  with a new reversible figure
• the presentation duration was brief (2 sec) so
  that participants could not experience the
  reversal while viewing it but was sufficient for
  auditory coding of the stimulus
• during this brief presentation, half of the
  participants remained silent the other half was
  required to say "la, la, la..." (designed to
  prevent auditory rehearsal)
• after the image was taken away, participants were
  asked what they had seen
• imagery phase
• participants were told to hold the figure in
  mind, and attempt to reverse it to reveal the
  second possible interpretation
• would image reversals depend on whether or not
  participants were able to use an auditory code
  during the initial encoding
• tested children and adults
• younger children tend to rely on a visual code in
  STM tasks,
• hypothesized that young children's ability to
  imagine the reversals would not differ in the
  auditory and visual coding conditions--young
  children don't rely on auditory coding so it
  should not matter whether they were allowed to
  rehearse or not

12
Brandimonte and Gerbino (1993) , cont.

• results
• reversals of the mental image were much more
  likely to occur when auditory rehearsal was
  prevented
• the effect did depend on age younger children
  experienced just as many image reversals when
  auditory rehearsal was prevented as when it wasn't

13
Sex Differences in Visual STM

• four components of visuospatial processing in STM
  
• image generation--bringing an image from LTM into
  STM
• participants committed 10 different block letters
  to LTM through extensive encoding and drawing
  practice they were then presented with a lower
  case/script version of a letter, and finally
  an"X" was presented in a certain spot on the
  screen
• task was to imagine the block-letter version of
  the presented lower-case letter, and decide
  whether the block version would cover the "X"
• image maintenance--holding the image in STM once
  it has been formed
• participants encoded a presented pattern, held
  the image in STM when it was removed
• then, an "X" appeared on the screen and they were
  to indicate whether the image they were
  maintaining would have covered the presented "X"
• image scanning--searching the image being held in
  STM for some feature
• participants memorized a pattern of small squares
  that formed a large square some of these squares
  were filled, and some were not.
• the pattern was then removed and an arrow
  appeared, pointing to one of the small squares
• task was to indicate whether the arrow pointed to
  a filled or unfilled square
• image transformation--actively manipulating an
  image being held in STM
• participants had to decide whether 2 figures,
  presented side-by-side, were identical
• the figures differed from each other only by
  varying degrees of rotation the decision
  required mental rotation of the right-hand figure
  to see if it matched the left-hand figure

14
Loring-Maier and Halpern (1999)
? Note error

• results
• males were faster than females in each of the
  four components of visual/spatial processing
• does not represent a speed/accuracy tradeoff
• error rate did not differ between males and
  females for any of the four tasks
• implication
• standardized tests that involve speeded tests of
  visual processing will lead to lower scores for
  women than men
• May lie behind differential GRE scores of men and
  women

15
Forgetting in STM

• Decay
• the loss of information from STM due to the
  passage of time
• original support came from early studies using
  the Brown-Peterson task
• forgetting occurred with minimal interference
• counting was assumed to involve stimuli (numbers)
  that were too dissimilar to letters to cause
  interference
• explanation seems to be inadequate it is not
  time but something that occurs in that time
  interval that is responsible for forgetting (cf.
  rust)
• Interference
• information is lost from STM because information
  currently being processed is negatively
  influenced by the presentation of other
  information
• Types
• proactive interference earlier information
  interferes with the ability to retain information
  that comes later
• retroactive interference later information
  interferes with the ability to retain information
  that occurred earlier

16
Decay or Interference?

• little forgetting on the first trial of a set of
  Brown-Peterson trials regardless of delay
• delay has no effect until there is an opportunity
  for interference to accumulate
• Theoretical explanations for detrimental effects
  of interference
• encoding views
• displacement view new item entering STM "bumps
  out" (displaces) previously stored item
• overwriting view new item entering STM
  overwrites previously stored item
• find more interference in a Brown-Peterson task
  if newer information was presented in the same
  modality as previously stored information (i.e.,
  both visual or both auditory) than if the two
  were presented in different modalities (i.e., one
  visual and one auditory)
• indicates that the overwriting view may be a
  better description of the effects of
  interference-if the interference effect was due
  to items being "bumped out" of STM, the new
  information would bump the old stuff out,
  regardless of modality
• retrieval views
• according to the notion of blurring and
  deblurring, items in STM can blur into one
  another and become difficult to tell apart (or
  deblur) at retrieval.

Keppel Underwood (1962)
17
Wickens, Dalezman, and Eggemeier (1976)

• employed the Brown-Peterson task, but used
  categorized lists instead of letters
• each trial involved reading 3 items from a
  category and then counting backward at which time
  the items were to be recalled
• presented 3 of these Brown-Peterson trials in a
  row, each time using three more items from the
  same category
• recall fell by 60 from the first to third trials
• on the fourth trial, participants were presented
  with three items but this time from a different
  category--fruits
• found release from proactive interference--recall
  bounced back up
• the degree of release is greater, the greater the
  difference between the original lists and the
  trial four list
• reveals that coding in STM is at least partially
  semantic

18
IV. Modular Approach to STM Working Memory

• A. The Articulatory Loop two sub-components
• phonological store
• allows one to hold information temporarily
• word-length effect derives from limited duration
  of phonological store
• subvocal rehearsal mechanism
• an articulatory process used to rehearse
  information
• responsible for the phonological similarity
  effect
• relies on the articulatory mechanism of the
  person
• auditory coding is prominent in WM not because
  the basic mechanisms of WM rely on the auditory
  modality but because WM relies on articulatory
  processes, which just happen to be auditory in
  hearing adults
• phonological similarity effect the critical
  factor is not that the words sound similar, but
  that similar motor movements are required to
  articulate the words
• analogous finding is found for users of ASL -- a
  sign-based similarity effect (similar-looking
  signs are more difficult to keep track of than
  different-looking signs because similarly looking
  signs share similar motor movements)
• word-length effect for users of ASL -- long signs
  are just as difficult to keep straight for deaf
  signers, as long words are for hearing speakers

19
Working Memory Model
20
Effects of Articulatory Suppression

• articulatory suppression task ("la, la, la..."
  task from rabbit/duck study)
• designed to prevent the a person from using
  articulation mechanisms to recycle information by
  tying up the "inner voice" (subvocal rehearsal
  mechanism) with other matters
• when used during the encoding of
  visually-presented items, there is no negative
  effect of phonological similarity or word length
• the item can not be coded in terms of the
  articulatory mode of your communication
  system--sound--so the way the words sound or
  their length doesn't influence performance
• articulatory suppression in users of ASL and its
  effect on word length
• participants were required to touch their middle
  fingers to the respective thumbs (the ASL "8"
  handshape), while at the same time having their
  hands circle one another, with contact at the end
  of each circle (this is the sign for "world")
• just as saying "la, la, la..." prevents someone
  from using the "inner voice" to vocally rehearse,
  making hand motions prevented one from using
  what might be termed the "inner hands" to
  manually rehearse
• articulatory suppression also eliminated the
  sign-based similarity effect

21
Effects of Articulatory Suppression
22
PETscanning the Articulatory Loop

• Awh, Jonides, Smith, Schumacher, Koeppe, and Katz
  (1996)
• attempted to isolate the brain areas involved in
  the two components of the articulatory loop
• procedure
• participants tested in one of three conditions
  designed to involve or not involve the two
  components of the articulatory loop
• two-back condition
• participants saw a string of letters presented
  one at a time on a computer screen
• after the presentation of each letter, their task
  was to determine whether that letter matched the
  one presented two items back by clicking a mouse
  button
• involves both components of the articulatory loop
• participants must constantly manage a WM load
  (which uses the phonological store), and silently
  repeat each letter as it is presented (which uses
  the subvocal rehearsal mechanism)

23
PETscanning the Articulatory Loop
24
PETscanning the Articulatory Loop cont.

• search-control condition
• participants saw the same stimuli, and made the
  same physical response (clicking a mouse), but
  the task was far simpler
• task was to judge whether each presented letter
  matched the first one they had seen in the
  sequence
• severely minimizes, if not eliminates, the need
  for either component of the articulatory loop
  very little must be held in the phonological
  store (the same letter) and there is really no
  need to continually rehearse the one letter so
  the subvocal rehearsal mechanism is not needed
• so the only brain activity in this task is
  clicking the mouse and watching digits
• looking at the difference in brain activation
  between the two-back and search-control
  conditions allowed the researchers to find the
  brain areas involved when the articulatory loop
  is active
• brain areas for the articulatory loop were the
  speech areas in the frontal lobe (Brocas area,
  which is involved in speech planning and
  execution), and the posterior regions of the
  parietal cortex (which are involved the storage
  of verbal information)

25
PETscanning the Articulatory Loop cont.

• rehearsal-control condition
• participants saw the same stimuli, and made the
  same physical response (clicking a mouse), but
  they had to repeat each letter they saw silently
  until the next letter appeared
• designed to isolate the brain areas active during
  the use of the subvocal rehearsal mechanism task
  does not require the use of the phonological
  store because there is essentially nothing to
  store (one letter)
• subtracting this area of activation from the
  total area of activation involved in the
  articulatory loop leaves only a view of the brain
  areas involved in the phonological store
• revealed that activation in the parietal regions
  seems to be the basis for the phonological store,
  while activation in the frontal lobe is
  associated with the subvocal rehearsal mechanism
• the authors suggest that the subvocal rehearsal
  mechanism depends on the articulatory mechanisms
  similar to the ones used in overt speech (i.e.,
  Broca's area)-further supporting the idea that
  the subvocal rehearsal mechanism is based on
  articulation rather than sound

26
Modular Approach Working Memory

• B. Visuo-Spatial Sketchpad
• responsible for the storage and manipulation of
  visual and spatial information, and seems to
  operate (in large part) independently of the
  other subsystem (the articulatory loop)
• This was shown by Brooks (1967)
• manipulated the task mode (visual or verbal) and
  the response mode (visual or verbal)gave two
  concurrent WM memory tasks (both visual, both
  verbal, or one of each)
• participants had a great deal of difficulty
  combining tasks if they were both visual (i.e.,
  both engaging the visuo-spatial sketchpad) or
  both verbal (i.e., both engaging the articulatory
  loop), they were quite doable if the task
  combination was one of each
• C. Central Executive
• might be conceived of as the "gatekeeper" or
  "capacity allocator" for the attentional system
• when a particular task demands extensive
  involvement of either the articulatory loop or
  visuo-spatial subsystem of working memory, the
  central executive deploys the necessary resources
• thought to be responsible for the higher-level
  thought processes involved in reasoning and
  language comprehension

27
Evaluation of Working Memory

• does a better job of describing STM than that
  proposed in the original modal model
• original model simply viewed immediate memory as
  a static holding place for information the
  working memory model represents immediate memory
  as a set of processes involved in the dynamic
  manipulation as well as static storage of
  information
• working memory span
• provides a more dynamic and valid
  characterization of immediate memory capacity
  than simple short-term memory span
• example--operation span
• participants receive trials like the following
• Is (8/4) - 1 1? BEAR
• Is (6 x 2) - 2 10? DAD
• Is (10 x 2) - 6 12? BEANS
• participants are to read the problems aloud and
  solve them
• after each of the three problems in a trial has
  been presented, they are to recall the words that
  accompanied each problem
• WM span performance is a good predictor of a
  broad range of more complex abilities (spoken and
  written language comprehension, writing,
  note-taking, etc.)

28
Evaluation of Working Memory

• WM span performance is a good predictor of a
  broad range of more complex abilities (spoken and
  written language comprehension, writing,
  note-taking, etc.)
• Why?
• tasks involved in WM span are a microcosm of what
  is required during complex cognitive
  processing--must process multiple streams of
  information, keeping some of it active and easily
  retrievable even in the face of interference from
  other material that may be more relevant at the
  time
• predictive power of working memory span
  underscores the important role that attention and
  information manipulation play in more complex
  cognitive processing
• working memory capacity is about attention in the
  service of memory
• greater WM capacity means that more items can be
  maintained in the focus of attention, but it also
  means that information can be effectively blocked
  from the focus of attention