Working Memory (8b) Flashcards
(15 cards)
Outline the research into the PFC by Fuster (1974)
Monkey neurophysiology studies suggested a role for the prefrontal cortex (PFC) in working memory
- monkeys see piece of food in a tray, shutter comes down, try is closed
- when the shutter opens the monkey has to remember where the food was located
- single neuron recordings from PFC showed ELEVATED FIRING during the delay period (when shutter down)
Interpreted as: showing that neurons in the PFC hold a representation of the to-be-remembered stimulus (eg. location of the food)
Outline the Goldman-Rakic (1987) Standard Model of Working Memory
It was proposed that this PFC activity reflects the neuronal instantiation of Baddeley’s WM storage buffers:
Sustained delay activity in PFC = Baddeley’s WM storage buffers
–> Sustained activation in PFC during the delay period of a WM task reflects a neuronal WM ‘template’ (a temporary representation of to-be-remembered info)
Outline the direct WM correlate research from monkey neurophysiology (Funahashi et al. 1989)
Used a OCULOMOTOR DELAYED RESPONSE TASK
–> further evidence for the model came from monkey single neuron recording
- in this task monkeys sawa cue on the left or right of fixation and had to maintain their eye gaze at the centre for 3 seconds and then make an eye movement in the direction of the cue
FOUND: that single neurons in the PFC showed direction-specific firing during the delay period of this task (between the cue and response - holding in memory)
- direct neurophysiological correlate of a WM template (temporary representation of the spatial location)
Wilson et al. 1993 - outline the evidence from monkey neurophysiology for what/where dissociation in working memory
The majority of evidence for a ventral/dorsal dissociation in the PFC for object v spatial working memory came from MONKEY STUDIES
–> monkeys trained to perform delayed response task
–> monkeys saw a cue that instructed them to make an eye movement in a particular direction - then had to remember that info before making a response
- there were PATTERN delayed response tasks and SPATIAL delayed response tasks different info to remember
FOUND:
- neurons in VENTRAL PFC showed higher activation to patterns compared to spatial cues
- neurons in DORSAL PFC showed higher activation to spatial cues compared to patterns
What v Where in working memory - recap our two streams
Reminder: we have the dorsal stream and ventral stream
Dorsal PFC: SPATIAL working memory (remembering WHERE an item was)
Ventral PFC: OBJECT working memory (remembering what an item was)
–> model suggests that there are two types of visuospatial working memory - one system for objects and another for spatial locations
Supporting evidence from PET - briefly summarise Courtney et al. 1996
Supporting human fMRI
–> object working memory and spatial working memory tasks
FOUND SIMILAR RESULTS:
- ventral (what) - higher activation for object
- dorsal (where) - higher activation for spatial
Testing the idea that PFC stores temporary representations - outline the WVPA work (Linden et al. 2012)
Multivoxel Pattern Analysis - MVPA
–> takes advantage of fine-grained patterns of activation in the brain
–> uses machine learning technique to teach an algorithm about the pattern of neural activation associated with a stimulus
–> Algorithm can then decode what the stimulus is, simply by looking at brain activity pattern
Found:
- regions holding category related info are exclusively in the posterior part of the brain (involved in sensory processing)
- PFC does not seem to hold temporary representations of info in WM
Implication:
- the same regions that enable processing are also involved in storing temporary representations of those objects in WM
Outline the evidence AGAINST ventral-dorsal what-where distinction in PFC
Has used monkey tasks to remember an object and make relevant eye movements
- over half of neurons showed both object and location selectivity
The fact that the neurons could encode both object and location - suggests flexible adaptation of responses in PFC - neuron scan adapt to represent whatever information is task relevant
Duncan - what is the adaptive coding model of PFC neurons
PFC neurons show flexibility in their responses ‘tuning’
–> unlike neurons in other brain regions, response properties of PFC neurons are not fixed, but adapt to encode currently task-relevant information
“Chunking” refers to the process by which the brain, particularly the prefrontal cortex (PFC), groups individual pieces of information or actions into a single, cohesive unit or “chunk.”
–> This allows for more efficient processing, learning, and task execution.
What is the role of OTHER PROCESSES (PFC in working memory)
What other processes to tasks like these require
- storage of previously touched items
- suppression of previously touched items
- selective attention to a novel item
- planning/strategy use
- sustained attention
Studies converge to suggest the PFC is NOT organised according to the type of stimulus held in WM (what v where)
–> Perhaps instead the role of the PFC is to follow complex task rules
Can we decode information about items held in WM from patterns of activation in prefrontal/parietal cortex
Rigall and Postle: found that ppts can decode which direction and speed dots are moving - but only from VISUAL and TEMPORAL cortical areas
- PFC provided no info about this
BUT: task instructions could be decoded from PFC (task rule)
An alternative view to the standard model - what is the idea that WM as an emergent property (Postle)
Working memory functions are produced when attention is directed to systems that evolved to accomplish representation related functions…
… therefore Working memory may simply be a property that emerges from a nervous system that is capable of representing many different kinds of information - and that is endowed with flexibly deployable attention
internal attention hypothesis
Outline the evidence by Higo that PFC performs an attentional role in WM
Subjects held two objects in WM and were cued either to maintain both objects (non-selective) or a single object (selective attention) in WM
–> asked to then decide whether any of the objects in an array matched what they were holding in WM
FOUND:
- activation in PFC was greater for the selective condition (hold one item) than the two item
–> Activation in PFC modulated activation in different occipitotemporal regions depending on which stimulus the subject held
–> Activation in PFC modulated activation in posterior regions too
WM seems to involve a process of directing attention to specific items held in working memory
Describe the distributed neuronal architecture of WM
Lower level visual regions: maintain the temporary representations of items held in WM (the templates)
The PFC/Parietal regions: hold a representation of the task rules for manipulating the info (the central executive)
Outline the key points of the lecture - role of PFC neurons and what it reflects
PFC neurons show sustained activity during the delay period of a WM task
HOWEVER - this does not reflect STORAGE of items in WM
Representations of items stored in WM are maintained in sensory-specific cortex
–> PFC seems to play a role in the processing of stimuli held in WM
PFC seems to play a role in the PROCESSING of stimuli held in WM
–> involves enhancing attention to internal representations of task relevant stimuli in WM
–> and manipulating such information according to a set of stored task rules
