Concentration normalization with a model of gain control in the olfactory bulb

This article presents a biologically inspired model capable of removing concentration effects from the multivariate response of a gas sensor array. The model is based on the first stage of lateral inhibition in the olfactory bulb, which is mediated by periglomerular interneurons. To simulate inputs to the olfactory bulb, signals from a chemosensor array are first processed with a self-organizing model of chemotopic convergence proposed earlier, which leads to odor-specific spatial patterning. Subsequently, a shunting lateral inhibitory network, modeled after the role of periglomerular cells in the olfactory bulb, is used to compress concentration information. The model is validated using experimental data from an array of temperature-modulated metal-oxide chemoresistors.