This talk will describe two models for brain dynamics and their relation to data
The first is a model for slow delta-band activity over the surface of mouse cortex. We develop a linear model for both intrinsic regional dynamics and communication between cortical regions, and use this model to estimate some of the effective connectivity between different cortical regions. Our estimates from dynamical data correspond well to known anatomical connections. Short-term predictions from this model are correlated 65 percent with observed data. The model is being extended to non-linear dynamics. The second model reproduces the gamma rhythm found in active regions of cortex. There have been several models that generate plausible rhythms in the gamma range, 30-50 Hz; but the parameters of these models are not realistic. Furthermore recent genomic data from healthy human subjects indicates very high variability in key parameters of these models, and current models are not robust to this variability. We propose a more realistic model drawing on data; the distinctive feature is high diversity among connection strengths. This model gives much more realistic gamma rhythms, on all measures, and is also more robust to inter-individual variation.