MODELING GAIN CONTROL IN SENSORY NETWORKS: THE OLFACTORY CASE

MARACHLIAN EMILIANO; LOCATELLI FERNANDO; HUERTA R.

Huerta Grande Cordoba

Congreso; Sociedad Argentina de Neurociencias 2012; 2012

Sociedad Argentina de Neurociencias

The natural stimuli are presented in a large range of intensities. It is important for animals to recognize the stimulus identity independently of it´s intensity; especially increasing sensitivity when the stimulus is weak and avoiding saturation when it is too intense. This property is named gain control. It allows us to sense the presence of a flavor regardless of its intensity, or an odorant irrespective of the concentration. In particular for bees it is important to distinguish an odor regardless of it´s intensity because during the foraging the odor concentration fluctuates while approaching the targets.The computational models of the Antenal Lobe (AL) are normally based on random connectivity among neurons with specific defined probabilities for the connections between different types of neurons. In order to show gain control properties the probabilities have to meet functional relationships that do not seem to have biological meaning. In the present work we show results using a realistic model that considers our current knowledge of the AL structure. The model considers the glomerular structure and the connections between Projection Neurons and Local Neurons inside them. The neurons activity is calculated using the Hodgkin-Huxley model. Our simulations show that the current model accomplishes gain control properties without requiring specific functional restriction in the connection probabilities between neurons. The new model is more robust for gain control property than previous ones.