Stimulus discrimination in networks of spiking neurons

Discriminating amongst stimuli in the environment is a fundamental aspect of brain function. Research has shown that chaotic neural networks are exquisitely sensitive to small perturbations making them unreliable and unpredictable. Here, we examine how neuronal oscillations (i.e., temporal waves of activity) may be tuned to enhance the discrimination performance of chaotic neural networks. Using a computational model of randomly connected leaky integrate-and-fire (LIF) neurons, we examine the possibility that oscillations enhance the reliability of spike times across several repetitions of the same stimuli. Compared to networks with no oscillations, networks injected with oscillations yielded markedly superior stimulus discrimination. Furthermore, the discrimination performance of the model was sensitive to the frequency and amplitude of oscillations, as well as the phase at which stimuli were presented. In sum, our work suggests that oscillations augment spike timing reliability, thus leading to enhanced stimulus discrimination performance. Overall, results highlight the importance of background rhythmic activity on information processing in neuronal circuits.