Date of Award

2012

Degree Type

Report

Degree Name

Master of Science (MS)

Department

Computer Science

First Advisor

Nicholas Flann

Abstract

Decision making in natural systems, such as the body's immune response to a potential pathogen or a bacterial colony's initiation of fruiting due to food scarcity, is distributed over many cells that posses only local information, and not determined globally. Understanding how accurate decisions can be made in such systems where no individual decisions maker has complete information has important implications in distributed software and can provide insights into the biological evolution of complexity. In this work, the process of distributed decision making is modeled using the majority problem in cellular automata, and information theoretic measures of Kolmogorov complexity are applied to quantify information flow during the decision making process. Results show that (a) when the decision making process converges the information content of the dynamics quickly reaches a peak then decays to near-zero; (b) if the process does not converge and becomes chaotic, information content oscillates over a large unstable range; (c) extensive statistically significant differences exist in information flow dynamics between convergent and chaotic outcomes; and (d) there are small, but statistically significant differences in information flow dynamics between convergence to the incorrect answer. This last result supports the hypothesis that correct decision making maximized information flow among agents in distributed decision making.

Comments

This work made publicly available electronically on November 5, 2012.

Share

COinS