Can Mutual Imitation Generate Open-Ended Evolution?
Takashi Ikegami [1,2],
Atsushi Masumori [1,2],
John Smith [1],
and Norihiro Maruyama [1,2]
1-Department of General Systems Science., University of Tokyo, Meguro, Tokyo 153-8907 Japan
2-Alternative Machine Inc., 6-23-4 Jingumae, Shibuya, Tokyo 150-0001, Japan
corr: ikeg@sacral.c.u-tokyo.ac.jp
We only find open-ended evolution (OEE) in the development of human technology or in the evolution of life itself. The research on OEE at ALIFE aims to discover a mechanism that generates OEE automatically in a computer or machine. A potential mechanism and the conditions required have been discussed in three previous workshops. In this study, we propose and discuss man–machine interaction experiments as a new OEE mechanism. The pertinent definition of OEE here is whether we can continue to create new movements that are distinguishable to us. We consider the development of body movement patterns generated when Alter3 androids imitate each other and when Alter3 androids and humans imitate each other. We use UMAP contraction and transfer entropy to measure these changes and demonstrate that man–machine communication is far more dynamic and complex than the machine–machine interaction. We discuss how human subjects can engender OEE via communication with the android.