Abstract
Transcranial magnetic stimulation (TMS) induces an electric field (E-field) that drives neuronal activation, but the optimal model for predicting cortical responses remains unclear. Traditional TMS motor mapping typically relies on the E-field magnitude or its normal component as a proxy for excitability, overlooking the influence of neuronal morphology and orientation. In this study, we aimed to refine TMS motor mapping by incorporating an average response model that accounts for both E-field magnitude and directional sensitivity. We conducted a regression-based TMS mapping experiment in 14 participants to identify cortical origins of motor-evoked potentials (MEPs) from the first dorsal interosseous (FDI) muscle. Firing thresholds were estimated for excitatory neurons in cortical layers 2/3 and 5, and regression was performed between MEPs and three E-field quantities: the E-field magnitude (magnitude model), the normal component of E-field (cosine model), and an effective E-field that adjusts magnitude by orientation-specific thresholds (neuron model). Models were compared based on regression fit, convergence speed, and functional validation using optimized coil placements tested in 10 additional participants. Results showed that the magnitude and neuron models performed similarly and robustly, whereas the cosine model explained significantly less variance, required more TMS pulses for stable mapping, and produced the weakest MEPs in validation. These findings suggest that while directional sensitivity plays a role, E-field magnitude remains the dominant factor in motor cortex activation.