| THE EMERGING PROPERTIES OF NEURONAL NETWORKS |
Egidio D'Angelo
Istituto di Fisiologia Generale
Via Forlanini 6, I-27100 Pavia, Italy
e-mail: dangelo@ipv36.unipv.it
The neurones are the computational units of the nervous system of living organisms. Neurones are interconnected by synapses, building-up neuronal networks. Even the simplest neurones are capable of complex computation exploiting the non-linear properties of membrane currents, and their uneven distribution over the soma and dendrites. In this way, neurones transform the synaptic input into a related discharge of action potentials. The "neurone doctrine" considers these concepts as the basis of nervous system computation, and ultimately of animal behaviours and higher human functions. However, it is becoming evident that complex functions cannot emerge from single neurones, but from network properties instead. The neuronal network theory assumes that the network performs input-output vector transformations depending on synaptic weights between neurones. A conceptual advance in the field has come with the discovery that neuronal memory can be stored in the form of permanent changes in synaptic efficacy, called long-term potentiation and depression. Though short-term plasticity may be equally important for procedural memory, long-term plasticity matches the Hebbian rule for associative memory. Thus, long-term synaptic plasticity would ensure that stable states are achieved in a given network by training. This conceptual framework is promoting investigations both toward a deeper understanding of single neurone and synaptic functions, and toward the development of neuronal networks incorporating realistic representations of neurones and synapses. This two-prong approach is inherently bottom-up. There are several examples of how this approach applies to neurobiological questions, showing how network properties emerge from intrinsic network architecture and neuronal and synaptic properties. We are currently investigating synaptic plasticity and neuronal excitability, and are developing a compartmental model of granule cell neurones of the cerebellum. Our perspective is to implement synaptic and neuronal properties into a large-scale network of the cerebellum, a brain structure involved in co-ordinating sensori-motor and cognitive processing.
e-mail: hmp@ipvvis.unipv.it