Cross-channel and local feedback inhibition shape selectivity-flexibility regimes in a reduced thalamoreticular rate model.

Competitive biological circuits must balance two opposing demands: amplifying small input differences to support selective dominance, while preserving the ability to switch when input priorities change. Here, we study this problem using a reduced inhibitory population-rate model motivated by thalamoreticular competition. The model contains two competing relay populations and two associated inhibitory populations, allowing cross-channel inhibition and local feedback inhibition to be varied independently. Linear analysis shows that competitive amplification is controlled by the effective difference between cross-inhibition and local feedback inhibition, scaled by population gain and recruitment strength. Numerical simulations further show that cross-inhibition amplifies small input asymmetries and promotes selective dominance, but excessive cross-inhibition can stabilize the initially dominant population and impair reversal after stimulus switching. Targeted post-reversal fixed-point analysis identified representative parameter regions in which two locally stable post-reversal equilibria coexisted: a switched equilibrium favoring the newly stronger channel and a persistent equilibrium favoring the initially selected channel. Within the tested parameter ranges, increasing local feedback inhibition reduced post-reversal persistence and shifted the boundary toward successful switching. Noise perturbations show that dominance reliability depends on the magnitude of input fluctuations relative to the imposed input asymmetry, while connection ablations and fixed-total-inhibition controls confirm that selection is a circuit-level property of the closed excitatory-inhibitory loop and depends on the allocation of inhibition between cross-channel and local feedback pathways. Finally, two phenomenological implementations of dopaminergic modulation shift the selectivity-flexibility boundaries in distinct ways, indicating that neuromodulatory effects depend on the functional target of modulation. Together, these results identify a minimal phenomenological inhibitory motif in which the allocation of inhibition between cross-channel and local feedback pathways organizes selection, switching, and persistent dominance within the same dynamical architecture.