We study the effect of finite-size synaptic noise on the response to time-varying stimuli of a detailed neurodynamical model of perceptual-decision making. Specifically, the reaction to a periodic stimulus is analyzed, while controlling the noise level through the number of neurons in the perception module. We find an enhanced response to the dynamical stimulus for intermediate noise, i.e. for an optimal system size. We thus propose a possible functional role of system-size resonance in a complex cognitive task.