Dynamic Neural Interactions Revealed by the State-Space Ising Model

Stimulus information and cognitive states of an animal are represented by correlated population activity neurons. The maximum entropy method provides a principled way to describe the using much less parameters than number possible patterns. This successfully explained stationary spiking neural populations such as in vitro retinal ganglion cells. Modeling cortical circuitries vivo, however, has been challenging because both spike rates interactions among neurons can change according sensory stimulation, behavior, or internal state brain. To capture non-stationary neurons, we augmented model (Ising model) state-space modeling framework, which call Ising model. We will demonstrate that applications reveal dynamic interactions, how they contribute sparseness fluctuation well stimulus coding.