A GENERAL-METHOD TO CHARACTERIZE HOMOGENEOUS NEURONAL POPULATIONS

Nonlinear methods have been applied to the study of brain systems. However, experimental results have been disappointing in that higher-order terms are often found to have no effect, or increase the error in prediction for kernel-derived models. While simple neural systems may be second-order, it is possible that the efficacy of higher-order terms for complex systems may be degraded by computational error. Input parameters, output measures and computations which incorporate known parameters of neural systems are described. Results should provide system definition that is more accurate and amenable to modeling and simulation efforts.