Decoupling synaptic weight and connection sparsity reveals asymmetric control of network dynamics

Circuit dynamics arise from the interaction between the network’s connectivity structure and intrinsic neuronal nonlinearities, yet the roles of key structural parameters — synaptic weight (W) and connection probability (P) — are usually examined in simplified network models. Using a biologically grounded, multilayer spiking model of thalamocortical microcircuitry, incorporating conductance-based neurons and rodent somatosensory cortex connectivity, we systematically scaled W and P and identified four organising principles of population dynamics. First, stronger synapses monotonically amplified spiking across all populations. Second, increasing connection density produced a weight-dependent bidirectional outcome: adding weak synapses preserved baseline activity, whereas adding strong ones suppressed firing. Third, concurrent increases in W and P yielded sublinear effects, where population activity increased less than expected from the sum of their individual impacts. Fourth, two functional neuronal classes emerged — scaling-invariant neurons that reliably transmitted thalamic input across connectivity regimes, and variant neurons that spiked selectively under specific connectivity scales. These classes differed in their excitation–inhibition balance, shaped by the strength of recurrent inhibition. Together, our findings show that synaptic weight and connection probability work in concert to define cortical operating regimes and generate the functional diversity in neuronal responses that supports flexible computation.