Dynamic Brain state for preparatory attention + WM

Question 1

A WM basis for preparatory attention

Answer 1

• Important for maintaining beh relevant info that determines the focus of attention
• WM representations could make up the basis of top-down bias

Question 2

Strong coupling between attention + WM

Answer 2

memory for a specific location helps with detecting a chance stimulus presented at the memory location

Question 3

Non-spatial features in Wm capture attention

Answer 3

• When items in a search array share a feature with the WM item, search times are faster
• Means WM representations are sufficient for biasing visual processing

Question 4

Dynamic population coding - a problem for stable activity states

Answer 4

• Dynamic beh of neural representations in various brain regions has been highlighted in many studies

Question 5

Example study info dynamic population coding

Answer 5

• specific odours trigger activity patterns within anterior lobe projection neurons
• most discriminative info for odor classification is observed along dynamic trajectory through the activity state space
• neurons receiving outputs from these projections respond most vigorously during the most dynamic phase

Question 6

Pattern of activity that drives classification during cue processing

Answer 6

• studies found that the most robust classification during cue processing doesn’t persist into the delay period
• it evolves along a specific trajectory
• information content peaks during the most dynamic phase of the trajectory

Question 7

Remapping + dynamic changes

Answer 7

• from the cue representation –> representations of the anticipated target
• stimulus processing is associated with complex spatiotemporal trajectory state space
• activity seems to travel through a continuous senses of states rather than maintaining a fired memory as anticipation state

Question 8

Hidden state

Answer 8

• stimulus driven activity alters response to subsequent stimulation, this is so that subsequent inputs will trigger a unique response pattern reflecting hidden state, determines response for next input
• New pattern further modulates the hidden state, determining the response to the next input
• hidden states could maintain info in WM

Question 9

silent memory code

Answer 9

• ## more metabolically efficient and general than persistent spiking activity

Question 10

How is WM mediated

Answer 10

• by temporary synaptic weight changes

Question 11

hidden network state

Answer 11

• even random inputs generate a reliable response reflecting the hidden network state
• hidden state isn’t necessarily silent as random spontaneous activity still drives a differential response