2007 Fiscal Year Final Research Report Summary
A Conslrudive Approach tovards Emagence of the Theory of Mind
Project/Area Number |
18500112
|
Research Category |
Grant-in-Aid for Scientific Research (C)
|
Allocation Type | Single-year Grants |
Section | 一般 |
Research Field |
Intelligent informatics
|
Research Institution | Nagoya University |
Principal Investigator |
ARITA Takaya Nagoya University, Graduate School of Information Science, Professor (40202759)
|
Project Period (FY) |
2006 – 2007
|
Keywords | Theory of mind / Evolutionary Psychology / Cognitive Science / Genetic Algorithm / Artificial Life / Baldwin Effect / Multilevel Selection |
Research Abstract |
This study focuses on the understanding and the application of the evolutionary basis of the emergence of the theory of mind. This study focuses on the understanding and the application of the evolutionary basis of the emergence of the theory of mind The results can be summarized as follows. (1) We constructed an agent-based model with a framework composed of two stages, negotiation using signals and subsequent strategy decision. It was shown that two types of communication emerged in the evolutionary simulations. In one type of communication, agents coordinate by the difference in the timing of sending a concessionary signal. In the other type of communication, they coordinate by resonating their signals. (2) We explored the emergence of communication based on gaze control as a result of the coevolutionary dynamics of "mind reading' and "manipulation of others". It was shown that a kind of communication based on gaze movement has emerged although agents have implicit and explicit cost
… More
s of taking their eyes off their directions of movement. (3) We constructed a simple fitness function that represents a multi-modal fitness landscape in which there is a trade-off between the adaptively of individual and the strength of epistatic interactions. It was shown that the Baldwin effect repeatedly occurred through the evolutionary process and facilitated its adaptive evolution. (4) We constructed an agent-based model of multilevel selection and introduced migration and evolution. It was shown that positive assortment was generated especially when a migration strategy was adopted in which individuals respond specifically to bad environmental conditions. It was also shown that the evolution could further facilitate positive assortment by working with migration. (5) Embodied Evolution (EE) can solve serious issues in conventional evolutionary robotics including the gap between the performances in simulation and the real world. We introduced pre-evaluation into the EE framework so as to restrain robot behavior with low fitness. Apre-evaluation mechanism was adopted as a co-evolutionary system in the proposed architecture. Less
|
Research Products
(11 results)