Max Planck Institute for Dynamics and Self-Organization -- Department for Nonlinear Dynamics and Network Dynamics Group
Personal tools
Log in

BFNT/BCCN Sonderseminar

Thursday, 26.05.2011 16 c.t.

Sequence Generation in Dynamic Field Theory: From Serial Order to Behavioral Organization

by Dr. Yulia Sandamirskaya
from Institut für Neuroinformatik, Ruhr-Universität-Bochum

Contact person: Florentin Wörgötter

Location

Faculty of Physics, Friedrich Hund Platz 1, Seminarroom SR18 (E1.112)

Abstract

In my talk, I will introduce the basic concepts of the Dynamic Field Theory (DFT) -- the neural-dynamic framework for modeling cognitive functions and behavior. Stability of behaviorally relevant neural states is the core property of the dynamic neural fields (DNFs), which enables coupling of the DNFs to noisy and unreliable sensors and motors, as well as building complex DNF architectures that model perceptual, cognitive, and motor processes. In order to generate sequences of stable states that control actions in an environment, a neural-dynamic representation of a condition of satisfaction is introduced in this framework. Activation of the condition of satisfaction system triggers a cascade of instabilities leading to a sequential transition. Based on this mechanism, arbitrary fixed sequences may be learned and generated using the DFT framework, but also rules of behavioral organization may be incorporated. The robotic implementations of the sequence generation architecture demonstrate how the neural-dynamic model may guide behavior of an embodied agent situated in a physical environment.

back to overview