When Firing Rate Falls Short: Spike Synchrony to Efficiently Disentangle Stimulus Saliency and Familiarity

Abstract

Whether neural computation relies on firing rate or spike timing has been debated for decades, with no definitive resolution. Here, using a recurrent spiking network simulation and realistically varying stimulus saliency, we demonstrate that the two mechanisms are complementary: rate reliably detects stimulus class, spike synchrony reliably detects stimulus familiarity. This division of labor is necessary: rate coding fails to distinguish novel high-saliency from familiar low-saliency stimuli, while synchrony succeeds. We validate this complementary coding across a biologically realistic V1 model and an abstract associative memory network, demonstrating robustness across connectivity regimes and memory loads. Crucially, when inputs are already temporally coordinated through familiarity in a previous network layer, spike synchrony alone can encode both stimulus identity and familiarity. Our findings reconcile the rate-timing debate: familiar stimuli exploit efficient synchrony-based codes reinforced by recurrent connectivity across the cortical hierarchy, while novel stimuli depend on rate codes.