Organising sensory and motion information for use
All information received by the five senses, vision, hearing, smell, taste and touch, are processed and interpreted in the brain to create an internal representation of the outside world. On the other hand, internal senses inform on the position of different parts of the body, as well as how the body is to move. To further current understanding of how all this input is organised and integrated, a computational model that summarises existing results of neurophysiological studies was proposed by the AMOUSE project. Based on simple and well established mechanisms for the integration of information across different sensory modalities, it was used to investigate to what extent motion perception contributes to planning movements. The autonomous robot developed within the AMOUSE project and a simulated approximation of this robot was used in the experiments conducted in the laboratories of the University of Osnabrück in Germany. The neural model was used to control the robot as it learned to move about within the confines of a maze. Direct evidence of changes in the planning of their movements was provided by investigating the robot's trajectories. The robot's uncertainty to drive straight, turn to the left or to the right at crossings was gradually reduced as information from the light and touch sensors on the robot were mapped. Nevertheless, interpretation of motion information was seen as an integral component in the process of making the optimum decision under rational constraints. The AMOUSE project partners suggested that sensory neurons can encode all the essential information, but they neither contribute to the process of making a decision nor carry its outcome. On the other hand, the part of the peripheral nervous system that is involved in movement control could regulate whether high-level decision making is required by assessing the complexity of task demands. The emerging reconceptualisation of information processing principles has already been exploited for the construction of flexible and task-adaptive systems for the AMOUSE robot.