Two draft discussion documents are available.
1. Do we need to depend on emergent features of dynamical system models
in explaining how understanding of spatial structure, motion and
causality develops?
See this discussion
of two concepts of causality (Humean and Kantian) and related topics
including the thesis of the 'Primacy of Dynamical Systems' (PDS).
2. Can we design a domain in which to explore the design of a 'child
robot' learning about Kantian causation and related things?
The polyflap
domain seems to have some of the required features.
Does the process of exploring and learning about domains of manipulable
physical objects, such as objects in the polyflap domain, depend on the
truth of PDS, or could such exploration and learning occur under the
control of cognitive mechanisms that do not have the fast fluent control
of actions on which PDS concentrates?
Clearly both are important and need to be studied.
A.S.
(Last modified 23 Aug 2005)