Dr. Andrew Wilson: "Perception, Action & Dynamical Systems"

Date: Thursday, 28 April 2011
Time: 11:00
Location: 6th floor meeting room, DNC Huset, Aarhus University Hospital
44 Nørrebrogade, Aarhus

Dr. Andrew Wilson studies the perceptual control of action, with a special interest in learning. He is a lecturer in Motor Control with the Institute of Membrane and Systems Biology at University of Leeds.

Abstract:

Perception, Action & Dynamical Systems

Dynamical systems have seen widespread adoption in psychology. It is a useful formal framework for describing the behaviour of the kinds of complex systems we seem to be. One of the strengths of dynamics is that it can provide an abstract description of a system’s behaviour without any reference to exactly how the system has been built. This allows us, for example, to borrow the description of a simple mass-spring system and use it to describe all kinds of behaviour (e.g. steering, obstacle avoidance, rhythmic limb movements). This strength is also a weakness, however; the model may successfully describe behaviour but it does not help usexplain it unless the components we use refer to the actual composition of the specific dynamical system of interest. Dynamics is therefore a methodological tool, and not a theory of actual human behaviour.

 

The perception-action approach is such a theory. It is inspired by ecological theories of perceptual information (e.g. James J Gibson) and the understanding of the complex but highly organised nature of the action systems (e.g. Nikolai Bernstein). This theoretical approach meaningfully constrains the composition and organisation of the dynamical systems models we create so that they can describe task specific mechanisms to explain performance in a perception action task. We can then test the predictions of the model experimentally.

 

A classic experimental task for this kind of work is coordinated rhythmic movement and it has recently been explicitly modelled as a perception-action dynamical system. In this talk I will describe the model and the perception-action modelling strategy which led to its current form. This strategy relies on a broad range of experimental techniques, many of which we developed to test the model. I will also discuss recent data which have revealed how the basic task dynamic alters with learning, and discuss how we are applying the strategy to expand the model without losing the explanatory power.

 


A CFIN & MINDLab event. ALL ARE WELCOME -- No registration necessary.