A Model of Motivation with Chaotic Neuronal Dynamics

摘要

The key problem to creating an autonomous system is: how does the brain choose its reactions, and how are motivation determined by ongoing signals, memory and heredity. In attempts to design a robot brain, several efforts have been made to design a self-contained control system that mimics biological motivation. However, it is impossible to develop an artificial brain using conventional computer algorithms, since a conventional program cannot predict all of the possible perturbations and disturbances in the environment, and hence cannot plan strategies that allow the system to overcome these perturbations and return to an optimal state. An external programmer must constantly update the system about the proper strategies needed to overcome newly-encountered perturbations. In contradistinction, living systems demonstrate excellent goal-directed behavior without the participation of an external programmer, and without full knowledge of the external environment. Biological motivation refers to actions on the part of an organism that lead to the attainment of a specific goal. When the organism attains the goal it is in an optimal state, and no further actions are generated. A deviation from the optimum will result in a change in activity that leads to a return to the optimum. Biologic motivations arise as the result of metabolic disturbances and are related to transient injury of the specific neurons. Treatments which protect neurons satisfy motivations and exert a psychotropic action relative to relief. I have developed a novel hypothesis of how living systems achieve a goal, based on data gathered on the effects of motivation on individual neurons. I claim that if the neuron affects the non-stability of its postsynaptic targets (probably by means of motivationally-relevant substances) in the end it chooses its reaction, although at each instant it acts by chance.