Entropy Based Metrics of Sensory Motor Coordination

Abstract

This chapter reviews and summarises the main metrics of sensory motor coordination based on Shannon Entropy. Shannon Entropy and derived metrics provide useful tools for the study of processes that are inherently stochastic, multi-variated and continuous. They allow to associate measures of information and information content changes when the uncertainty associated to a process variates. This is crucial when studying embodied intelligent processes when thanks to mechanisms, such as morphological computation, part of the information processing is outsourced to the systems body dynamics. We are not describing all possible methods to measure sensory-motor coordination. Instead, we focus on the methods that can be applied to the study of the emergence of coordinated ‘intelligent’ behaviours in loosely coupled networks of agents (LCNA). In particular we briefly discuss the methods that allow to study a specific kind of emergent coordinated processes known by the IDSO acronym: Information Driven Self Organization. Moreover, as Information Theoretic metrics are still now not widely known in the Robotics and AI communities we start by shortly introducing the basic related metrics.