Week 3: Working Memory Model = CHECKED Flashcards
Taxonomy of memory diagram:
declarative memory
It refers to memories which can be consciously recalled/declared such as facts (semantic memory) and events (episodic memory)
short-term memory (STM) is
Keeping a small amount of information in mind and making it accessible for a short time.
Example of STM (working memory/WM)
As an example of working memory, a new phone number is kept in mind until it is dialed and then immediately forgotten.
capacity of short term memory
7 +/- 2
STM and WM often used
interchangeable but defined differently
The mechanism of loss in STM is
Primarily decay
We are consciously aware of STM meaning we can
We can cognitively manipulate the contents of STM in our head and actively rehearse them
Capacity of LTM
High
Mechanism of loss in LTM is
interference
interference theory is theory
Due to structure (2)
the theory that people forget not because memories are lost from storage but because other information gets in the way of what they want to remember
It is due to the structure of brain (e.g., overlapping representations of the memory in the brain
According to the interference theory,
Individuals forget LTM memories caused by memories interfering and disrupting one another (Baddley, 1999)
Are we consciously aware of LTM?
Yes
Decay
When information is immediately forgotten when it is no longer needed/relevant in that moment
Working Memory Model was made by
Baddeley and Hitch (1974)
central executive maintains and manipulates STM in WM model
memory contents
More specifically,
central executive drives
directs (2)
drives the whole WM system
It directs attention and processing to the different subsystems its connected to: visuospatial sketch pad and phonological loop
phonological loop is where information is
WM model
acoustically coded
phonological loop is part of WM
WM
the part of working memory that holds and processes verbal and auditory information
visuospatial sketchpad where info is stored and processed in WM
visually or spatially.
Information in visuospatial sketch pad is visually in WM
coded
In a task, where letters are presented visually, participant’s show errors that indicate information is acoustically coded, for example:
they replace T for G (both sound similar) instead of Q for G (appearance of letters look similar)
Similarly, pps found recalling a wordlist more difficult for similar sounding words and not semantically related words, for example: (after research replace T for G)
recalling ‘rice’ instead of ‘ice’ instead and not recalling ‘frost’
Research of replcating T with G and pps difficulty recalling ‘rice’ instead of ‘ice and repeating nonsense syllables disrupts phonological memory implies that (2)
WM system is not unitary
is a multi-component system with modality specific components , each can be damaged separately.
Each component of WM can be damaged separately:
Phonological WM deficits
For example, damage to Brodman areas 44 and 40 means that (2)
individuals can not hold strings of word in their memory/mind
there will be deficit in the rehearsal process of phonological loop
Each component of WM can be damaged separately:
Visuospatial sketchpad WM deficits
Damage (i.e., lesions) to parieto-occipital causes deficits in visuo-spatial WM for example:
pps with that damage have difficulties in memorising and repeating a sequence of blocks experimenter has touched
Support for the dissociation of both visual (i.e., visuospatial sketch pad) and auditory (i.e., phonologcial loop) WM - (3) in PET scans
This is because there is changes in local cerebral blood flow (PET) in different areas of the brain when participants doing verbal and spatial WM tasks in healthy participants
For Auditory WM tasks: activity in infero-lateral
For Spatial WM tasks: occipital, pariteral, inferior frontal (most RIGHT of the brain)
In Lisman-Idiart Model, neural model of WM, they put forward that
a neural mechanism called ADP helps to carry out WM maintenance
ADP stands for
afterdepolarisation
Purpose of the model
In Lisman-Idiart Model, we do not need to think about
which ion channel is in charge of ADP
Purpose of the model:
In Lisman-Idiart Model, they want to model
we don’t care about (2)
ADP effect on MP
we don’t care about the ion channels
Equation of the model
In previous lecture we introduced: integrate and fire model equation as:
if u = urest then…
if u = urest (inputs are 0) then rate of change is 0 so stay at resting M
Equation of the model
In Lisman-Idiart Model,
u =… (change of MP over Dt)
They have other
(2)
V
they have other contributions of voltage in change of MP/dt
Equation of the model
In Lisman-Idiart Model, Vinh is
The input from inhibitory interneuron which is assumed and not modelled explicitly
Equation of the model
In Lisman-Idiart Model, Vinh is added
every time an AP (i.e. spike) is fired by the pre-synaptic neuron
Equation of the model
In Lisman-Idiart Model, Vosc is
a sin function to cause some membrane potential fluctuation in the background (as observed in real neurons)
Equation of the model
In Lisman-Idiart Model, Vrest is
Resting MP
Purpose of the model and equation
For the Lisman-Idiart neural model, it does not
simplify say after integrating inputs into the equation whether
(2)
calculate the (shape) AP
AP is fired or not
ADP hump in MP diagram
ADP
In depolarisation, (3)
the membrane potential (voltage) increases (e.g.., -70 to 64 mV)
More likely to emit a new spike
Excitatory inputs have this effect
ADP
Hyperpolarisation is when (3)
membrane potential (voltage) decreases and becomes more negative (e.g., -70 to -75 mV)
The neuron is further away from the firing threshold
AHP and inhibitory synapses have this effect
The ADP is a positive ‘hump’ in MP after
each spike
ADP
Lisman and Idiart model ADP as (2)
being similar to a weak synaptic input
Modelled mathematically as an alpha function
ADP
The ADP hump is not due to the (2)
produces after
product of synaptic activity
Produces after each AP and is due to a specific ion channel (we don’t care what channel that is in this model) that operates late.
ADP and Purpose of the model
Lisman and Idiart model:
But did you not say we need HH model to model ion channels? (2)
This is because model does not give us an explanation of ADP
In this model, however, they use ADP use this to explain a higher-level phenomena
Purpose of the model
The aim of the Lisman and Idiart model is to (after saying not use HH model)
demonstrate what ADP can be used for and its effect on MP
Purpose of the model
A full HH model would be completely unnecessary for Lisman and Idiart model as (2)
this complexity does not fulfil the aim of Lisman and Idiart’s model
This is because would not add how ADP functions in terms of networks of neurons
Research working on source of ADP in terms of ion channels will need what model?
HH model
Research working on what ADP is useful for in a neuron, what model will they use?
Integrate and fire model = Lisman-Idiart model
Purpose of the model
Lisman and Idiart model is like a modified
integrate and fire model
In Lisman and Idiart model, threshold and V rest (resting MP) is… (2)
Threshold = -50 mV
Vrest = -60 mV
Diagram of input terms (V rest, VOSC, VADP, Vinh) effect on MP:
Lisman-Idiart model network (4)
Activity of each prefrontal neurons is calculated by the equation change of MP over time
VOSC provides excitatory oscillatory input to the neurons
Vinh has feedback inhibition circuit
If vOSC neuron fires a spike exciting all the prefrontal neurons, eventually transmitted to Vinh neuron which inhbits all neurons including itself
Assume that the firing of each neuron in the Lisman -Idiart model network represents one item in the WM
HOW….?
For example,
In other words…
(2)
These neurons can quickly create a synaptic connection with neuron in phonological loop which encodes and represents the letter ‘G’
In other words, part of phonological loop that represents ‘G’ makes a particular neuron
Assume firing of each neuron in Lisman-Idiart model network represents one item in WM? HOW….? = Synaptic connection with neuron in phonological loop which encodes G so
Both oscillatory inputs and ADP maintains (after example of letter G that create synaptic connection)
spiking
part of the phonological loop that represents ‘G’ makes a particular neuron in the network fire AP and the neuron fires another AP after a while due to
due to both oscillatory inputs and ADP mainting spiking
Assume Firing of each neuron in the Lisman-Idiart model network represents one item in WM
(after osciliatory inputs in combination of neuronal properities maintain spiking)
HOW…
It can only (2)
fire so many spikes in an oscillation cycle
suggesting WM has limited capacity
Assume Firing of each neuron in the Lisman-Idiart model network represents one item in WM HOW…
(after osciliatory inputs in combination of neuronal properities maintain spiking, can only fit so many spikes)
Once activity of neurons (i.e., spiking) in Lisman-Idiart neural network is absent, the content of the STM (item)
cannot be recovered
Proposed mechanism of
Lisman-Idiart Model Network of WM
(4)
Combination of oscillatory input + ADP to maintain letter ‘G’
Neural mechanism for active rehearsal for content of WM
- There is background oscillations in neural network
- A particular neuron in network representing letter’ G’ fires AP due to an external input (e.g., presented with letter G)
- That neuron inhibits itself and all other neurons in network (feedback inhibition circuit)
- Next peak of oscillation comes around, ADP has raised MP high enough for that neuron to fire another AP
With the Lisman-Idiart model network we have the neural mechanisms
for example…
for active rehearsal of content of WM –> repeating firing of neuron represring ‘G’, F,S,W,A,M,R for example
The Lisman-Idiart model provides a neural mechanism for the active rehearsal of WM content using the
each on its own is…
(2)
neural properties (ADP) and network properties (feedback inhibition circuit and oscillatory input among different cells) that make this possible
each on its own is insufficient
We can not add more and more neurons that have the capacity to hold long strings in WM promptly because
The background oscillation has a fixed frequency, which means that… (2)
. the distance in time between two peaks of the oscillatory input (Vosc) is fixed
only fit so many spikes and ADP into one oscillatory cycle
The background oscillation having fixed frequency meaning we can only fit so many spikes and ADP into one osciliatory cycle
Therefore..
WM capacity is limited in this model, as it should be.
An example of WM is limited (7+/- 2 items, Miller et al., 1956)
Trying to memorise a list of words: GPSWAMR then the letter X is added before G
If you try to fit another spike (i.e., AP) that represents the letter ‘X’ then the feedback inhibition circuit will silence the
previously active neuron that spikes for letter ‘R’
Advantages and benefits of Lisman-Idiart model of WM (2)
The model demonstrates how neuronal prosperities (ADP) and network structure (feedback inhibition and oscillatory input) work together to implement function
A criticism is that the authors chosen parameters (e.g., oscillation frequency) to make the number 8 for capacity.