Do structured variabilities in discriminative models match evoked activity and evolve with learning

Determine whether the structured variability arising from input, neural, and synaptic noise in discriminative brain models resembles average evoked activity and whether the degree of this match increases with learning.

Background

Spontaneous cortical activity exhibits structured variability that resembles evoked responses and evolves with experience. A discriminative account could attribute this to noise shaped by learned synaptic structure, but it is uncertain whether this mechanism reproduces the empirically observed similarity to evoked activity and its learning dependence.

Resolving this would clarify whether spontaneous activity in discriminative frameworks can functionally mirror priors or average evoked responses observed empirically.