From Embodied Cognition to Free Will - Humana.Mente

S. Torrance & T. Froese – An Inter-Enactive Approach to Agency 29
narrative manner and to affirm in an intuitive way the personal sense of being
enabled or constrained in order to establish a science of social or interactive
situations. What is additionally required is a basic proof of concept, which
demonstrates that these narrative and phenomenological descriptions of the
efficacy of interaction processes in certain social situations are not merely
metaphorical embellishments of what is essentially a sum of individual actions.
To show that the social interaction process itself can play a constitutive role for
the actions of individual agents, we need to be able to show this process at work
in a concrete model that allows for systematic exploration of the essential
parameters.
One suitable way of satisfying this additional requirement is to take
advantage of recent work in Evolutionary Robotics modeling. Since its
beginnings in the early 1990s, the Evolutionary Robotics approach has
established itself as a viable methodology for optimizing dynamical controllers
for physical robots (Nolfi and Floreano 2000), as well as for synthesizing
simulation models of what has become known as „minimally cognitive behavior‟
(Beer 1996). The idea here is to set up an evolutionary algorithm that can
automatically shape the dynamical system of a model agent so that it performs a
given task in the simplest possible way, while still raising issues that are of
genuine interest to cognitive scientists (Beer 2003, Harvey et al. 2005, Froese
and Ziemke 2009). In the last decade there has been a growing interest in
using this kind of Evolutionary Robotics approach to investigate the dynamics
of social interactions (e.g., Iizuka and Di Paolo 2007a, Froese and Di Paolo
2008, Di Paolo et al. 2008), and the methodology has accordingly been
extended to include „minimally social behavior‟ as well (Froese and Di Paolo in
press).
We will draw on some of our own modeling work in this area (Froese and
Di Paolo 2010, in press), because the specific aim of these models is to serve
as proof of concepts which demonstrate that interaction processes themselves
can play a constitutive role in shaping individual actions over and above the
sum power of the individual agents. In fact, as we have already argued
extensively elsewhere (Froese and Gallagher 2010), the methodology of
Evolutionary Robotics is well suited to complement phenomenological
investigations in the cognitive sciences.
3.2 THE DYNAMICS OF PERCEPTUAL CROSSING
Froese and Di Paolo (2010) used an Evolutionary Robotics approach to