Paper 162 Details

Title

Analysis of causality network from interactions between nonlinear oscillator networks and musculoskeletal system

Abstract

In order to understand the interactions between the body, brain and environment that generate various behaviours, it is necessary to consider the network structure that dynamically emerges from interactions among the brain regions even though the brain has a fixed anatomical structure (Bullmore and Sporns, 2009). Kuniyoshi and Suzuki (2004) proposed a model in which adaptive behaviours emerge through body constraints as chaotic itinerancy that is induced by coupled chaotic elements. Moreover, Yamada and Kuniyoshi (2012) revealed the influence of embodiment in nervous system by embodied network. They constructed an embodied network using transfer entropy based on motor information and found that the embodied network had the properties of a complex network. However, they did not specify the structures of the network and dynamic changes in the network structure caused by different movements.

In this paper, we address the network structure relationship that dynamically emerges through interactions between the network, body and environment. We conducted a physical simulation using a snake-like robot with a nonlinear oscillator network (Mori et al., 2013) and estimated the network structure based on transfer entropy for each different movement. We defined a wired network for the physically embedded network and a causality network for the estimated network structure. In order to understand the relationships of the oscillators in the emergent casualty networks within the periodic behaviours by the robot, we extract the causality subnetworks by Infinite Relational Model (IRM) (Kemp et al., 2006) and analyze the networks by the complex network theory. Moreover, we measured average transfer entropy between body and network to know relationship between body and the causality networks.

Video Abstract