License 1 - License 2
Mentions: We have generated high-resolution data on neural activity over most of one hemisphere of mouse cortex by voltage-sensitive dyes, in both anesthetized and awake animals. In previous work  we have analyzed relations between activity measures at different locations in terms of correlations. Here we fit these data to a predictive model, in which we attempt to predict the next change in activity at every point on cortex from the current pattern of activity over cortex. We fit both linear and non-linear models, whose parameters represent the intrinsic dynamics of local cortical regions and the inputs from distal regions. We find that all regions of mouse cortex appear to have virtually identical patterns of intrinsic dynamics (Figure 1A). We find that even a simple linear fit gives surprisingly sparse patterns of inferred connectivity. Where we have clear anatomical information, these fitted patterns appear to match known anatomy. Furthermore this fit can be used to identify the most prominent functional inputs into anatomically diffusely-connected areas such as the parietal association area (Figure 1B).
No MeSH data available.