lecture nov 27

Free recall

"Neural model" with sparseness

discussion: mechanistic network vs. abstract graph

 

Individual person parameters

Limit of sparse

Universal formula

Non universal results

Why? Recognition as normalization

 

Possible extensions?

Reaction time - either stochastic, or 

 

 

Working memory

Computational challenge:

prolong beyond natural timescale

Noise

 

Persistent activity

Positive feedback

Eigenvalues and effective timescale

Can slow down and also cause bistability

 

Discrete vs. continuous

Noise

 

Bump attractor dynamics

CRITICAL: why no drift before 3 seconds? Motor?