AUTHOR=Egbert Matthew D. , Barandiaran Xabier E. 

TITLE=Using enactive robotics to think outside of the problem-solving box: How sensorimotor contingencies constrain the forms of emergent autononomous habits

JOURNAL=Frontiers in Neurorobotics

VOLUME=Volume 16 - 2022

YEAR=2022

URL=https://www.frontiersin.org/journals/neurorobotics/articles/10.3389/fnbot.2022.847054

DOI=10.3389/fnbot.2022.847054

ISSN=1662-5218

ABSTRACT=Biologically inspired robots have often taken inspiration from natural
  behaviour generating mechanisms and strategies to solve robotic engineering
  problems. At a more abstract level biologists have sought to understand living
  systems as a set of responses (traits) directed to solve the problems posited
  by the environment (adaptation). We here suggest that biologically inspired
  robotics could go much further. An enactive approach to robotics states that
  i) robots must first \emph{have their own} problems instead of solving those
  posited by external observers; ii) in doing so, they must assert a form of
  life whose structure and topology is the object of study.
  
  We take habits as the building block of an enactive inspiration of living principles (not biological mechanisms) to build robots. In order to make habits emerge out of sensorimotor interactions we implement an Iterative Deformable Sensorimotor Medium and leave agents to freely navigate a changing environment. We observe the spontaneous emergence of different habits, their re-enactment and the organization of an ecology of habits within each agent. We also explore the differences between agents endowed with different sensorimotor modalities and observe the formation of distinct topological invariants.
  
  We discuss the results in the wider context of: a) enactive approaches to life and mind, b) sensorimotor contingency theory, c) adaptationist vs. structuralist explanations in biology and d) the limits of functionalist problem-solving approaches to (artificial) intelligence.