Short term fellowships (1999-2003).
Final of Joaquín J. Torres-Agudo
Results:
During the stay of Dr. Torres, we have studied the effect of including synaptic mechanisms like synaptic depression in networks of binary neurons. We have discover that including the dynamic of the synaptic weights using a dynamical variable that take in to account synaptic depression, allows for the appearance of a new phase which is characterized by its oscillatory nature. In this phase, the activity of the network reaches one of the stored patterns and then destabilizes jumping to a different pattern. Using techniques from dynamical systems theory, we have analyzed the bifurcation properties of such oscillations and their local stability. In addition, we have developed a mean field approximation for finite number of learned patterns, which agrees qualitatively and quantitatively with the simulations of the network. We are preparing a manuscript with our main conclusions intended for publication in a specialized journal during the coming months.
Collaborative and/or training aspects:
The visit was very useful for dr. Torres for several reasons. The research of the department of Medical Physics and Biophysics in artificial neural networks and computational neuroscience is internationally well known. In general, the visit was useful for dr. Torres to get new knowledge in the modeling of realistic neural systems. In particular, dr. Torres has improved its knowledge in the modeling of synaptic transmission, including concepts like synaptic depression and facilitation. Theoretically, dr. Torres has improved its knowledge in the techniques for the analysis of the dynamics of highly dimensional non-linear flows and maps, for instance, by means of studying of the dynamical bifurcations and the nonlinear phenomena presented in the systems under study. Finally, the results obtained allow for a continuation of a fruitful future collaboration.
Expected future impact of visit:
The visit will be a first step in a fruitful collaboration between the departments of Electromagnetism and Material Physics of the University of Granada and the department of Medical Physics and Biophysics of the University of Nijmegen. The Material physics group at University of Granada will benefit of the strong knowledge in the modeling of biological neural systems from people at the department of Medical Physics and Biophysics. On the other hand, new theoretical techniques and models from non-equilibrium statistical physcis, developed for people belonging to the Material physics group at University of Granada, could be useful for the undertanding and the modeling of some biophysical processes in real neural systems.