Encoding sensory information in spiking neural network for the control of autonomous agents

Abstract

The goal of the work presented here was to find a model of a spiking sensory neuron that could cope with small variations of a simulated pheromone concentration and also the whole range of concentrations. We tried many different functions to map the pheromone concentration into the current of the sensory neuron in order to produce a near linear relationship between the concentration and the firing rate of the sensor. After unsuccessful trials using linear currents, we created an equation that would by definition achieve this task and used it as a model to help us find a similar function that is also used in biology. We concluded that by using a biologically plausible sigmoid function in our model to map pheromone concentration to current, we could produce agents able to detect the whole range of pheromone concentration as well as small variations. Now, the sensory neurons used in our model are able to encode the stimulus intensity into appropriate firing rates.