Yasuo Kinouchi

A LOGICAL MODEL OF CONSCIOUSNESS ON AN AUTONOMOUSLY ADAPTIVE SYSTEM

Consciousness is a tremendously complex phenomenon. We examined the configurations and functions of an autonomously adaptive system that can adapt to an environment without a teacher to understand this complex phenomenon in the easiest way possible, and proposed a modeling method of consciousness on the system. In modeling of consciousness, it is important to note the difference between phenomenal consciousness and functional consciousness. To clarify the difference, a model with two layers, a physical layer and a logical layer, is proposed. The functions of primitive consciousness on the autonomously adaptive system were clarified on the model. The physical layer is composed of an artificial neural node. All signals are processed in detail by the neural nodes. Contrarily, minimum information, necessary for the system to adapt itself, selected from the physical layer composes the logical layer. The operations in the logical layer are represented by interactions between only the selected information. Our daily conscious phenomenon is expressed on the logical layer.

An Approach for the Binding Problem Based on Brain-oriented Autonomous Adaptation System with Object Handling Functions

Abstract An approach for the binding problem is proposed, based on an autonomous adaptive system designed using artificial neural networks with object handling functions. Object handling functionality, such as object files, has been reported to have a relationship with perception, and working memory. However, in order for a brain-oriented system to decide actions based on object handling, the system must clarify the “binding problem”, or the problem of processing different attributes such as shape, color and location in parallel, then binding these multiple attributes as a single object. The proposed system decides semi-optimum actions by combining nonlinear programming and reinforced learning. By the introduction of artificial neural networks based on dendritic structures of pyramidal neurons in the cerebral cortex, together with a mechanism for dynamically linking nodes to objects, it is shown that deciding actions and learning as a whole system, based on binding object attributes and location, is possible. The proposed features are verified through computer simulation results.

A Model of Consciousness and Attention Aimed at Speedy Autonomous Adaptation

Abstract An advanced model is proposed that can explain the function in short term adaptation. Associative temporal memory and the function of top-down attention are adopted in the model, in which consciousness and attention are defined as clearly different functions; consciousness is mainly defined as a process of learning as a whole system, and attention is defined as functions that quickly select resources with priority based on the information for learning. These functions work by complementing each other via associative temporal memory. As a result, consciousness and attention are explained as fundamental functions for an autonomous adaptive system to be efficiently and speedily adapted to an environment. Even though the model proposed here has a configuration similar to Global Workspace Theory (GWT) including Global Neural Workspace (GNW), it differs greatly from our model in its basic purposes and functions.

An adaptive associative memory system based on autonomous reaction between image memories

To construct a “thinking-like” processing system, a new architecture of an adaptive associative memory system is proposed. This memory system treats “images” as basic units of information, and adapts to the environment of the external world by means of autonomous reactions between the images. The images do not have to be clear, distinct symbols or patterns; they can be ambiguous, indistinct symbols or patterns as well. This memory system is a kind of neural network made up of nodes and links called a localist spreading activation network. Each node holds one image in a localist manner. Images in high-activity nodes interact autonomously and generate new images and links. By this reaction between images, various forms of images are generated automatically under constraints of links with adjacent nodes. In this system, three simple image reaction operations are defined. Each operation generates a new image by combining pseudofigures or features and links of two images.

A Basic Architecture of an Autonomous Adaptive System With Conscious-Like Function for a Humanoid Robot

In developing a humanoid robot, there are two major objectives. One is developing a physical robot having body, hands and feet resembling those of human beings and being able to similarly control them. The other is to develop a control system that works similarly to our brain, to feel, think, act, and learn like ours. In this paper, an architecture of a control system with a brain-oriented logical structure for the second objective is proposed. The proposed system autonomously adapts to the environment, and implements a clearly defined “consciousness” function, through which both habitual behavior as well as goal-directed behavior are realized. Consciousness is regarded as a function for effective adaptation at the system-level, based on matching and organizing the individual results of the underlying parallel-processing units. This consciousness is assumed to correspond to how our mind is “aware” when making our moment to moment decisions in our daily life. The binding problem and the basic causes of delay in Libets experiment are also explained by capturing awareness in this manner. The goal is set as an image in the system, and efficient actions towards achieving this goal are selected in the goal-directed behavior process. The system is designed as an artificial neural network, and aims at achieving consistent and efficient system behavior, through the interaction of highly independent neural nodes. The proposed architecture is based on a two-level design. The first level, which we call the “basic-system” is an artificial neural network system which realizes consciousness, habitual behavior and explains the binding problem. The second level, which we call the “extended-system” is an artificial neural network system which realizes goal-directed behavior.

A Model of Primitive Consciousness Based on System-Level Learning Activity in Autonomous Adaptation

Although many models of consciousness have been proposed from various viewpoints, these models have not been based on learning activities in a whole system with capability of autonomous adaptation. Through investigating a learning process as the whole system, consciousness is basically modeled as system level learning activity to modify both own configuration and states in autonomous adaptation. The model not only explains the time delay of Libet’s experiment, but also is positioned as an improved model of Global Workspace Theory.

Dependence of load imbalance on data allocation among distributed processing modules in an advanced intelligent network

This paper considers advanced IN systems in which the data for each customer are allocated in a distributed way among multiple processor modules. A method is presented in which the traffic imbalance for customer data is represented. Based on such a representation, a method to evaluate the load imbalance among the processor modules, as well as a method to estimate its upper limit, are presented. Using the presented method, even if customer data composed of number conversion and screening information, and with a large traffic imbalance factor of several thousand to several tens of thousands, are allocated at random to the processor modules, the load imbalance can be reduced to 10 to 20% at the present level of processor performance and number of dynamic steps in the transaction. It is shown that dynamic load balancing can be achieved for this kind of imbalance by combining the processing that can be executed independently of the customer data, such as directory retrieval, which appears in ordinary distributed processing.