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

The function of temporal coding in the brain remains controversial, with debate centering on the following question: does spike timing information, such as their temporal synchrony, play a meaningful role in neural computation which cannot be attributed to firing rate? We propose the solution to this dilemma: spike synchrony provides crucial information about stimulus familiarity under conditions when the firing rate alone is insufficient − namely, when the input stimulus is varied in saliency. Using simulations of recurrent spiking networks, we show that synchrony is particularly effective in distinguishing familiar stimuli of low saliency from novel stimuli of high saliency − an important distinction for both biological perception and artificial agents navigating dynamic environments. We evaluate familiarity detection across two frameworks: a biologically inspired model of the primary visual cortex (V1) and an abstract associative memory model. We show that in both cases, synchrony is more sensitive to recurrent connectivity, that encodes prior experiences, compared to input firing rate. This highlights the relevance of synchrony for familiarity encoding in a scenario of realistic input variability.