Neuronal Working Memory Mechanisms Flashcards
1
Q
SUMMARY I
A
- neuronal mechanisms of WM
- “standard model” of role of prefrontal cortex (PFC) in WM:
1) info stored in PFC
2) PFC organised according to type of info stored
3) WM for objects VS locations - evidence against standard model:
1) PFC organised according to type of processing NOT info type (manipulation VS maintenance)
2) no info about stored items in PFC
3) role of PFC in WM = attentional
2
Q
FUSTER (1974): PROCEDURE
A
- monkey neurophysiology studies suggest role for prefrontal cortex (PFC) in WM
- monkeys see piece of food on tray; shutter comes down -> tray closes
- when shutter opens, monkey has to remember where food = located
3
Q
FUSTER (1974): RESULTS
A
- single neuron recording from PFC showed elevated neuronal firing during delay period (ie. when shutter is down)
- interpreted as showing that neurons in PFC hold representation of to-be-remembered stimulus (ie. food location)
4
Q
GOLDMAN-RAKIC (1987)
A
- proposed standard model of WM
- suggests that this PFC activity reflects neuronal instantiation of Baddeley’s WM storage buffers containing template (ie. temporary representation) of info maintained during delay
- became dominant view of PFC role in WM for some time
5
Q
FUNAHASHI ET AL. (1989): PROCEDURE
A
- evidence from monkey neurophysiology
- oculomotor delayed response task
- monkeys saw cue on left/right of fixation; had to maintain eye gaze at centre for 3s then make eye movement (saccade) in cue direction
- had to hold cue direction in memory for 3s
6
Q
FUNAHASHI ET AL. (1989): RESULTS
A
- found that single neurons in PFC showed direction-specific firing during delay period of task (between cue/response)
- this particular neuron fired strongly to locations in upper left quadrant
- interpreted as showing direct neurophysiological correlate of WM template aka. temporary representation of spatial location indicated by cue
7
Q
PETRIDES & MILNER (1982): PROCEDURE
A
- further evidence for standard model from human neuropsychology studies
- administered self-ordered task to patients w/frontal lobe lesions
- patients had to touch 1 pic; would then be given next sheet where they’d have to touch a dif pic etc. until they go through 12 sheets & touch all 12 pics
- aka. task requires them to remember from 1 sheet to next which images they’ve touched
8
Q
PETRIDES & MILNER (1982): RESULTS
A
- patients w/frontal lobe lesions were disproportionately impaired on self-ordered task; indicates that patients had deficit w/WM
- interpreted this as evidence that PFC holds representation of to-be-remembered info over short periods of time of WM task
9
Q
WHAT VS WHERE IN WM
A
- 1 of key tenants in standard model = 2 visuospatial WM types:
1) objects system
2) spatial locations - appealing idea as extended what we already know about visual object recognition into PFC
- aka. idea that 2 visual streams (1 for locating, other for identifying) extends into PFC & enables remembering info over short time periods
- evidence came predominantly from monkey neurophysiology studies
10
Q
WILSON ET AL. (1993): PROCEDURE
A
- majority evidence for ventral/dorsal dissociation in PFC for objects VS spatial WM came from monkey neurophysiology
- monkeys trained to perform oculomotor delayed response task
- saw cue instructing them to make eye movement in particular direction then had to remember info before making response
- key change = pattern cues (pattern appeared in screen centre instructing eye movement) in addition to standard spatial cues (cue in actual location)
- so response in both trial types = same BUT info type to be remembered differed
11
Q
WILSON ET AL. (1993): RESULTS
A
- neuron in inferior ventral PFC showed higher activation to patterns BUT lower activation to spatial cues
- neuron in upper dorsal PFC shows higher activation to spatial cues BUT lower activation to patterns
- each neuron preferred particular pattern/direction
12
Q
COURTNEY ET AL. (1996): PROCEDURE
A
- supporting evidence from fMRI for ventral/dorsal what VS where distinction in WM
- pps required to remember identity/location of 3 faces & to say if test stimulus matched any identities/locations
- activation for object WM task seen in ventral PFC
- activation for spatial WM task seen in dorsal PFC
13
Q
RAO ET AL. (1997): PROCEDURE
A
- evidence AGAINST standard model aka. PFC doesn’t store representations of stored stimuli
- task required monkey to remember object & 1st make eye movement to correct object; then make eye movement to correct location
14
Q
RAO ET AL. (1997): RESULTS
A
- single neurons in PFC can encode both location/object as well as its identity
- suggests flexible adaptation of responses in PFC
- aka. neurons can adapt to represent whatever info is task relevant
15
Q
NEUROPSYCHOLOGICAL EVIDENCE FOR PFC ROLE IN WM
A
- self-ordered pic task requires processes including:
1) storage of previously touched item
2) suppression of previously touched item
3) selective attention to novel item
4) planning/strategy use
5) sustained attention
16
Q
WM x PFC: POSSIBLE CONFOUNDS
A
- activation in dorsal PFC = spatial WM
- activation in dorsal PFC = object WM
- possible confound = processing type aka. pps may use strategies (ie. chunking) to help performance for spatial task
- requires manipulation of stored info in WM
17
Q
WM x PFC: CONCLUSIONS
A
- evidence doesn’t seem to support idea of dorsal/ventral distinction for storing locations/objects in WM
- maybe PFC = organised according to processing type > stored info type
18
Q
D’ESPOSITO ET AL. (1999): PROCEDURE
A
- fMRI evidence to support idea that PFC = organised according to processing type carried out > stimulus type maintained
- asked pps to perform WM task; had to either:
1) maintain info in WM by simply holding letter string in memory then judging if probe letter = part of memory set
2) (manipulation condition) rearrange letters into alphabetical order & judge if probe letter = in specific location in letter sting post rearranging
19
Q
D’ESPOSITO ET AL. (1999): RESULTS
A
- found activation in both dorsal/ventral PFC during delay period of WM task
- BUT dorsal PFC activation = greater during delay period of manipulation trials > during delay period of maintenance trials
- suggests that PFC = organised according to processing type (dorsal = manipulation; ventral = maintenance) > stimulus maintained type (spatial VS nonspatial)
20
Q
HUMAN x MONKEY STUDY CONVERGENCE
A
- both suggest that standard model = incorrect
- aka. PFC isn’t organised according to stimulus type held in WM
- PFC may be organised according to processing type (dorsal = manipulation; ventral = maintenance)
- PFC may not even hold representations of stimuli held in WM (where is it stored? what’s its real role?)
21
Q
MULTIVOXEL PATTERN ANALYSIS (MVPA): FMRI DATA
A
- used to try and answer where in the brain WM info is stored
- traditional fMRI = smooth data so groups of individual vowels are treated as clusters; useful way to reveal where in the brain a particular process is happening
- BUT risks missing important info that might be contained in individual vowel responses
22
Q
MULTIVOXEL PATTERN ANALYSIS (MVPA)
A
- takes advantage of fine-grained patterns of activation in brain
- uses machine learning techniques to teach algorithm about neural activation pattern associated w/particular stimuli
- so algorithm = able to “decode” what pp is looking at simply by viewing brain activity pattern
23
Q
LINDEN ET AL. (2012): PROCEDURE
A
- pps performed task requiring them to hold several objects in WM
- each trial required pps to decide if single object = part of memory test
- 4 object categories: faces/bodies/flowers/scenes
- trained pattern classifier to learn activation patterns for each category
- tested ability of pattern classifier to predict which category pps were holding in WM in each trial
24
Q
LINDEN ET AL. (2012): RESULTS
A
- regions holding category related info = exclusively in posterior brain region aka. visual processing regions (ie. FFA/other ventral visual stream regions activated when object categories present to pps)
- implication = same regions enabling processing object when seen w/eyes also involved in storing temporary representations of said objects in WM
- PFC doesn’t hold temporary representations of WM info; so what is its role?
25
RIGGALL & POSTLE (2012): PROCEDURE
- also looked into where dif info types are represented in brain during WM
- scanned pps performing WM task; had to memorise moving dot array; cued during delay period to remember direction OR speed of dots
- then shown probe (match/mismatch)
- when cue appeared pps had to follow rule (attend to speed OR direction) while holding specific stimulus in WM (particular speed/direction)
26
RIGGALL & POSTLE (2012): RESULTS
- could decode which direction/how fast dots were moving BUT only from visual cortex/temporal cortex; PFC provided no info about this
- task instructions could be decoded from PFC/parietal regions
- consistent w/Linden (2012); stimulus specific info = stored in posterior sensory specific cortex
- BUT info about task rules = stored in frontal/parietal cortex
27
HIGO ET AL. (2011): PROCEDURE
- provided direct evidence for idea that PFC performs attentional role in WM
- pps held 2 objects in WM; subsequently cued either to:
1) maintain both (non-selective attention condition)
2) maintain 1 (selective attention condition)
- finally asked to decide if any objects in array matched objects they were holding
28
HIGO ET AL. (2011): RESULTS
- activation in inferior PFC = greater for selective attention > non-selective
- implication = PFC performs attentional function by prioritising processing of item in WM that is most relevant to current task
- activation in PFC directly modulated activation in posterior regions; suggests that PFC biases activation in sensory-specific regions during WM in same way as during attention to external stimuli
- aka. (at least in this task) WM involved process of directing attention to specific items held in WM
29
DISTRIBUTED NEURAL ARCHITECTURE OF WM
- WM model where:
1) lower level visual regions maintain temporary representations of items held in WM (templates)
2) PFC/parietal regions hold representation of task rules for manipulating info (central executive)
30
SUMMARY II
- PFC neurons show sustained activity during delay period of WM tasks
- BUT activity doesn't reflect WM item storage
- representations of stored items = maintained in sensory-specific cortex
- PFC seems to play role in processing stimuli held in WM; role may involve enhancing attention to internal representations of task relevant stimuli in WM & manipulating this info
