Dynamic causal modeling of layered magnetoencephalographic event-related responses

Abstract

The layered architecture of cortex is thought to play a fundamental role in shaping cortical computations. However, direct electrophysiological measurements of layered activity are not possible non-invasively in humans. Recent advances have shown that a distinction of two layers can be achieved using magnetoencephalography in combination with head casts and advanced spatial modeling. In this technical note, we present a dynamic causal model of a single cortical microcircuit that models event related potentials. The model captures the average dynamics of a detailed two layered circuit. It combines a temporal model of neural dynamics with a spatial model of a layer specific lead field to facilitate layer separation. In simulations we show that the spatial arrangement of the two layers can be successfully recovered using Bayesian inference. The layered model can also be distinguished from a single dipole model. We conclude that precision magnetoencephalography in combination with detailed dynamical system modeling can be used to study non-invasively the fast dynamics of layered computations.