A model of consciousness and self based on simple abstracted brain-like neural network system

Document Type: 
ASSC Conference Item
Article Type: 
Theory of Consciousness
consciousness, concept, self, global workspace theory, neural network
Deposited by: 
Professor Yasuo Kinouchi on 14 February 2009
Date of Issue: 
Yasuo Kinouchi, Katsumasa Masuda, Shoji Inabayashi
Event Dates: 
19-22 June 2008
Event Location: 
Event Title: 
12th annual meeting of the Association for the Scientific Study of Consciousness
Event Type: 
ASSC Conference
Presentation Type: 
A model of consciousness and self that functions corresponding to global workspace theory (GWT) is proposed after investigating a neural network system with a simple abstracted brain-like structure that autonomously adapts to its environment. In this model, the operation of recursive self, nonconscious homunculus, and working memory are explained comprehensively. The system basically learns by trying to avoid unpleasant experiences and repeat pleasant experiences. As a top-level function, a self feels these experiences and consciously decides on actions on the basis of past experiences. The system does these actions faster and more adequately through learning. The system is mainly composed of four functional modules: a concept module, an associative memory module, a core control module corresponding to “self”, and an integration module that connects these modules. The integration module selects the most important combination of concepts in the system at the time on the basis of a mutual vote from the many concepts. The selection result is then sent to the other modules. This function corresponds to GWT. The core control module only has an active control function and decides whether to act as a system. Additionally, it feels the state of the environment at the time, past experiences, and intentional emotion. In our model, “to feel” is executed by only two methods. One is that a core control module perceives “the state of selected micro features in a concept module”, and the other is that a core control module directly perceives “the own state of emotion.” The former corresponds to our daily perception of environment or recall of memories. As direct perception of the own state of the core control module is restricted to emotion, then perception of the declarative content of the own state of the core control module has to go around the other modules. This means that the core control module can only see itself through a mirror and so experiences a recursive self. In addition, the operation of working memories is described as a memory control method applied by the core control module. This model leads us to clear understanding for consciousness and self.
ASSC12Poster20080621(Kinouchi).pdf114.42 KB