Mechanistic Role of Layer 5-to-2/3 Recurrent Feedback in TMS-Evoked Motor Potentials.

Abstract

Objective We aimed to develop and validate a personalized neural field theory (NFT) model of the motor cortex incorporating recurrent connections from layer 5 to layers 2/3. Methods We extended the established feedforward model of Wilson et al. (2021) by adding recurrent feedback connections. Synaptic coupling coefficients were derived using a bottom-up approach from microanatomical data and calibrated via Particle Swarm Optimization (PSO) against experimental input-output (IO) curves. The model was then personalized for individuals by optimizing three recurrent weighted gains ([Formula: see text]). We simulated single-pulse IO curves, paired-pulse protocols (SICI, ICF, LICI), and responses to intermittent and continuous theta-burst stimulation (iTBS/cTBS). Results The model revealed a sigmoidal, non-linear relationship between TMS intensity and neuronal input, in contrast to previous linear assumptions. Recurrent weighted inhibitory gain ([Formula: see text]) was critical for generating realistic short-interval intracortical inhibition (SICI). Also, this parameter was the dominant parameter for personalization, successfully fitting 13/18 iTBS and 15/18 cTBS individual response profiles. The model demonstrated that SICI/ICF and LICI emerge from protocol-specific tuning of the same recurrent microcircuit, with LICI requiring a greater suppression of recurrent excitation consistent with GABAB mediation. Our model showed superior generalizability, achieving robust fits on an independent cTBS dataset not used for initial calibration. Conclusion We present a unified, mechanistically interpretable NFT model identifying inhibitory feedback as a primary regulator of TMS dynamics and inter-individual variability. Significance This work advances the Wilson (2021) model by providing a biologically grounded framework that links clinical TMS metrics to specific cortical circuits, offering a powerful tool for optimizing personalized neuromodulation therapies.