Information encoding based on spike timing has attracted increasing attention due to the growing evidence for the relation between synchronization in neural networks and higher brain functions, such as memory, attention and cognition. And it has been shown that first-spike latency (arrival time of the first spike associated with information) carries a considerable amount of information, possibly more than other spikes.
The researchers analyzed the presence of noise in the nervous system, detected by changes in first-spike latency (the time it takes for brain cells to first respond to an external stimulus) and jitter (variation in spike timing).
The noise is generated by the synaptic bombardment of each neuron by a large number of incoming excitatory and inhibitory spike inputs and because chemical-based signalling does not always work.
Previous attempts at noise modeling used a generic bell-shaped signal, referred to as a Gaussian approximation. The new noise model, published in European Physical Journal B, is closer to biological reality, the engineers suggest.
They showed there is a relation between the noise and delays in spike signal transmission, and identified two factors that could be tuned, thus influencing the noise: the incoming excitatory and inhibitory input signaling regime and the coupling strength between inhibitory and excitatory synapses. Modulating these factors could help neurons encode information more accurately, they found.