Influence of model type and network resolution on whole-brain simulation results and potential modeling artifacts

Determine how the choice of whole-brain model type and the spatial resolution of the structural connectome influence simulation outcomes, and ascertain the extent to which conclusions drawn from these simulations are limited by modeling artifacts stemming from these choices.

Background

Whole-brain modeling is widely used to explore mechanisms underlying macroscopic brain activity patterns. However, models vary substantially in biophysical detail and in the spatial resolution of the connectome, and these choices may affect the dynamics produced in simulations. The degree to which such differences alter results—and therefore the reliability of inferences drawn from them—has not been conclusively established.

This paper compares a biophysically realistic adaptive linear-nonlinear (aLN) model and a phenomenological Wilson–Cowan model across Schaefer parcellations with 100, 200, and 500 nodes, aiming to assess robustness of emergent spatiotemporal dynamics. The authors explicitly note uncertainty about the effects of model type and resolution on simulation results, motivating the need to determine whether observed phenomena reflect true mechanisms or modeling artifacts.