Causal Measures of Structure and Plasticity in Simulated and Living Neural Networks
Conditional Granger causality was applied to the center block of electrodes (excluding border electrodes) to illustrate the effect of removing mediated influences. For the analysis of each probe the causal effect of each electrode upon each target was conditioned by conditioning out the effects of all other electrodes. The left and right panel shows the results of the conditional analysis for the source and sink measure, respectively. Removing erroneous mediation influences from PGC's causal estimates substantially refined the pattern of causal connections. This is most apparent in the total causality (upper right plots of each panel) where only a few primary electrodes remain, electrodes which may represent the major pathways that were modified by the tetanus.