Perception, Coordination And Evolution -Intelligence Lies In Structure And Movement

Biological-inspired computational model is one of the most important driving forces for the development of artificial intelligence. For a long time, a large number of schol-ars have proposed various kinds of biological-inspired models based on different bio-logical phenomena. They have made many pioneering contributions in the field of arti-ficial intelligence. We attempt to follow the footsteps of these scholars, and investigate the physiological experiments and discoveries in recent years. Based on the investiga-tions, it is found that the structure of the nervous system and the movement patterns of the nerve impulses on the structure have a vital influence on the intelligent behaviour of the organism. It can be said that the intelligence lies in the structure of the nervous system and the movement patterns of the nerve impulses. We attempt to simulate the structure of the brain, the movement patterns of nerve impulses on the structure, and the expansibility of the structure. And two biological-inspired models are proposed, which are the perception coordination network and the perception evolution network. The main work are as follows.(1) To simulate the perception coordination between different senses in the brain, a perception coordination network is proposed. It is inspired by the brain’s hierarchi-cal structure and the coordination between different functional modules in the struc-ture. Thus, the perception coordination network is a hierarchical structure, which is functionally divided into the primary sensory areas, the primary sensory association areas, and the higher order association areas. The primary sensory areas process many scattered feelings, e.g., the feeling of color, shape, and syllables. The primary sen-sory association areas combine the scattered feelings in the primary sensory areas to represent unity object, e.g., to form visual concepts, auditory concepts, and gustato-ry concepts, etc. The higher order association areas connect several primary sensory association areas, e.g., the association of vision and audition, vision and gustation, etc. This area executives a function just as a kind of synaesthesia. The perception coordination network gives three types of artificial neurons according to their func-tions, which includes the feature cells, the primary concept cells, and the associated cells. Different types of cells have different behaviours. Through the establishment of synapses between neurons, perception coordination network forms new memories quickly. Unconscious impulse and introspection mechanism are designed to guarantee the consistency of the learned network structure and the external data structure. This network can be applied to the field of concept acquisition, information fusion, online learning system, and robot system, etc.(2) Experiment in [87] shows that the cDNA knock-in mice show enhanced long-wavelength sensitivity and acquire a new capacity for chromatic discrimination. This experiment implies that the structure of the perception system or the nervous system of the organism can be expanded artificially, e.g., using the genetic engineering tech-nology in [87]. Inspired by the experiment, a question is introduced, that is, can we exploit an agent that is able to expand its cognitive dimension online freely to break the perception limitation? In order to solve this problem, a kind of perception evolution network is proposed, which includes Type I. emergence of novel receptor in existing perceptual channel and Type II. emergence of novel perceptual channel. When novel receptors emerge in existing perceptual channel, the simulates received from the novel sensory receptors are transmitted to the primary sensory areas along with simulates received from the existing sensory receptors. And the feature, cell associates the in-herent features with the novel features to form a "deeper" feature concept. That is, by establishing the synapse between the feature cells and the novel receptors, the response dimension of the feature cells is expanded online. When novel perceptual channel emerge, simulates received from the novel perceptual channel are transmitted to the primary sensory areas along with that from the existing perceptual channels. These simulates are processed into many scattered feelings. Then the impulses from the fir-ing feature cells are transmitted ascendingly to the primary sensory association areas to fire their target concept cells. At last, the firing concept cells transmit their impuls-es to the higher order association areas, the novel sense and the existing sense will intercommunicate with each other through the associated cells to form a coordinated unity. This work also gives a partial answer, from the computational viewpoint, to the question proposed in [87], that is, "... create an additional class of sensory neurons or to facilitate the emergence of neural circuitry for comparing new and existing sensory responses." The perception evolution network can be applied to the field of information fusion, online learning system, and robot system, etc.(3) Many neurophysiological experiments and medical cases imply that cerebral cortex and sense organ have a high expansibility [142,146,187]. Based on these two models, these neurophysiological experiments and medical cases, a theory of cortex-receptor artificial extension is introduced. Then a concept of artificial evolution is introduced. The artificial evolution aims to realize the biological evolution at the living level by sense-brain reforming and expanding. That is the living body itself gains evolution. Though the concept and theory are in the imagination stage currently, I think that they are worthy of deep thinking and research.The perception coordination network and perception evolution network aim to model the perception, coordination, and evolution of the organisms. The validity of the two models is verified by some experiments. However, to put these two models into practical application, there is still a long way to go.The addendum part discusses some problems which mainly focus on building a bridge between the micro physiological structure, micro physiological coordinations and human’s macro behaviour. Currently, the discussion of these problems in the dis-sertation is very rough. However, they all are worthy of deep thinking. It is hoped that these problems can be theorized and engineered in the future.