We present in this work a theoretical and conceptual study and some reflections on a fundamental aspect concerning with the structure and brain function: the Cellulr Communication. The main interests of our study are the signal transmission mechanisms and the neuronal mechanisms responsible to learning. We propose the consideration of a new kind of communication mechanisms, different to the synaptic transmission, "Diffusion or Volume Transmission". This new alternative is based on a diffusing messenger as nitric oxide (NO). Our study aims towards the design of a conceptual framework, which covers implications of NO in the artificial neural networks (ANNs), both in neural architecture and learnign processing. This conceptual frame might be able to provide possible biological support for many aspects of ANNs and to generate new concepts to improve the structure and operation of them. Some of these new concepts are The Fast Diffusion Neural Propagation (FDNP), the Diffuse Neighbourhood (DNB), the Diffusive Hybrid Neuromodulation (DHN), the Virtual Weights. Finally we will propose a new mathematical formulation for the Hebb learnign law, taking into account the NO effect. Along the same lines, we will reflect on the possibility of a new formal framework for learning processes in ANNs, which consist of slow and fast learning concerning with co-operation between the classical neurotransmission and FDNP.
展开▼
