Modules, layers, hierarchies, and loops where artificial intelligence meets ethology and neuroscience - In context of action selection

Abstract

The paper presents arguments for why AI methodologies should be informed of both behavioural science and neuroscience studies, and argues why this is possible. Through identifying the resemblance points, we will discuss whether findings of ethology and neuroscience can be used in the process of design and development of non-classical AI systems. To this end, we focus on a specific example that has long been investigated in all the concerned disciplines: the action selection problem. The paper overviews action selection mechanisms in behaviour-based AI and neuroscience in order to identify the commonalities that underly the understanding of action selection in different disciplines, and that may constitute the pieces of a common language with which AI can communicate ideas and findings to and from ethology and neuroscience. © Springer-Verlag Berlin Heidelberg 2005.