The discriminatory power of the olfactory system is enormous both qualitatively and quantitatively. The initial step of discrimination is the joining of the odorous ligand and the corresponding receptor protein. As a result, ion channels open in teh membrane of the neuron raising the membrane potetntial above the threshold value. This may sometimes be triggered by a single molecule. Thermal fluctuation can also cause 'spontaneous' firing. A minimial ratio of the population must fire simultaneously to give a signal..

The noise output of first order neurons are transformed into signals by two mechanisms. Spontaneous activity is consiodered to be a mere hindrance. It turned ot that this is not necessariliy the case: spontaneous activity can also increse the sensitivity of neurons. It may occur that the mean value of the receptor potential is below the threshold value, outliers still are able to cause firing. This is the case of stochastic resonance.

Thus, olfaction is an area where the geenral methods of stochastic processes and neural networks can be applioed. This is the aim of the present work, in which experts in biophysics, physical chemistry, neural networks and stochastic processes join to approach the problems.

Modelling olfaction shows similar traits than that of formal reaction kinetics. The evolution of asystem can be described by a master equation or (using a stochastic model) by Langevin- or Fokker-Planck equation. We are going to concentrate to to impoirtant phenomena: stochastic resonance and self-organizing criticality. We shall study the effect of random fluctuations onto the transfer properties of axons, as it has turned out from formal models: such fluctuations maty evoke stochastic resonance with positive consequences. We also think that spontaneous tending to the critical state will be illustrated on the olfactory system. The complexity of the system will also require computer simulations.

László Zalányi and Fülöp Bazsó were working on this project in cooperation with

* János Tóth, Department of Computer Science, Agricultural University
* Rospars, Jena-Pierre, Biometrie, INRA, Versailles
* Moreau, Michel, Laboratoire de Physique Theorique des Liquides, Univ. P. and M. Curie
* within the framework of the BALATON project