A Tale of Two Learning Rules

Learning from past experiences is a dening feature of the brain. The biology of learning manifests as experience induced changes in the synaptic coupling between neurons. While there has been progress in the mathematical modeling of learning in single synapse models, there is much less understanding of how
stimulus experience shapes the recurrent coupling within networks of stochastically driven neurons. We review two mathematical frameworks where clustered network wiring is learned; one based strictly upon the time integrated activity and the other upon the ne timescale structure of network dynamics. The stability of learned structures in the face of ongoing stochastic activity will be contrasted between the two frameworks.