Madhusudhan Venkadesan

Postdoctoral Associate of Mathematics

Web Site www.math.cornell.edu/~madhu Contact Information Office:  411 Malott Hall Phone:  (607) 255-7244 Fax:  (607) 255-7149 Email:  madhu@math.cornell.edu Courses & Office Hours

Education

Ph.D. (2006) Cornell University

Research Area: dynamical systems, biological motor control

The core of my research interest is to develop experimental and theoretical capability to understand "how" animals coordinate their body using muscles to stably and robustly interact with the world.

How do the "brain" and "body" interact with the world to yield the quintessential robustness and versatility of animal behavior? This question dates back to at least 2300 years as seen from Aristotle's arguments in De animalium partibus (ca. 340 BCE). Attempts to answer this question have resulted in remarkably detailed knowledge of the constituent sensorimotor elements in animals ranging from relatively simple nematodes to extraordinarily complex humans. However, we remain unable to predict or mimic the overall behavioral dynamics of most animals under most circumstances, with added difficulties arising from "noise" and "time-delays" that are ubiquitous at almost every level in biological systems.

I developed an experimental-mathematical technique in my doctoral dissertation to understand how multiple and redundant sensory inputs are used by the nervous system during dexterous manipulation tasks. A simple phenomenological model explained that the everyday and experimental observation of context-dependent use of vision in manual tasks is a natural consequence of noise and time-delays when combining redundant sensations task-optimally. Through collaborative projects, this method of characterizing dynamical dexterous manipulation has also provided new insights into the cortical networks responsible for sensorimotor integration in the hand and the effects of treatment methods for thumb osteoarthritis.

Currently, my work (jointly with John Guckenheimer) is to design experimental methods and data analysis techniques for creating data-driven low order dynamical models. Most dynamical systems encountered in biology or engineering are often too high-dimensional to model from first-principles. By being able to generate data-driven low-dimensional models of their behaviour, we will create the capability to mimic, predict, or control these otherwise intractably complex systems.

Selected Publications

The strength-dexterity test as a measure of dynamic pinch performance (with F. J. Valero-Cuevas, N. Smaby, M. Peterson and T. Wright), Journal of Biomechanics 36 (2003), 265–270.

Dynamic dexterous manipulation: Benefits of the edge of instability in exploring complex dynamical behavior, Ph.D. Thesis, Cornell University (2007).

Manipulating the edge of instability (with J. Guckenheimer and F. J. Valero-Cuevas), Journal of Biomechanics 40 (2007), 1653–61.

Last modified: September 5, 2007