Traditionally, cognitive science has focused on symbol-manipulation tasks in isolated problem domains. Implicit in this approach is the identification of intelligence with high-level, algorithmic, and uniquely human abstract reasoning. Arguably more essential, however, is the coordinated perception and action required of even the simplest organisms, whose survival depends on reacting appropriately and in real time to the constantly changing conditions of a messy and often hostile environment. Embracing this latter view, some modern approaches to cognitive science identify intelligence with adaptive behavior: an intelligent organism is one whose actions support the survival of the individual or species.
In this dissertation, I seek to characterize the neural basis of intelligence by investigating the neural correlates of adaptive behavior. This work is performed in the context of Polyworld, an agent-based artificial life model. In Polyworld, a population of agents evolves in a simulated ecology. Each agent is endowed with a rudimentary sense of vision and is capable of a set of simple actions controlled by a neural network. The population's long-term survival depends on the evolution of adaptive behavior via rewiring of these networks over successive generations. I identify the neural mechanisms underlying adaptive behavior by investigating trends in neural properties, both over evolutionary time and across experimental conditions.
Applying the tools of graph theory, dynamical systems theory, and information theory, I analyze Polyworld's neural networks using metrics with demonstrated relevance in other studies of complex systems. Many properties of interest are amplified over evolutionary time: the emergence of a successful survival strategy is accompanied by trends toward small-world neural structure, critical neural dynamics, and effective neural information processing. These trends are statistically significant when compared to a neutrally evolving null model, suggesting correlations with adaptive behavior. More direct evidence for these properties' adaptive significance comes from correlations with environmental complexity: as experimental conditions become increasingly difficult, each property is further amplified, fostering the increasingly complex adaptations required for survival.
|Advisor:||Beer, Randall D., Rocha, Luis M.|
|Commitee:||Beggs, John M., Sporns, Olaf, Yaeger, Larry S.|
|School Location:||United States -- Indiana|
|Source:||DAI-B 81/4(E), Dissertation Abstracts International|
|Subjects:||Artificial intelligence, Computer science|
|Keywords:||Adaptive behavior, Agent-based modeling, Artificial life, Artificial neural networks, Complex networks, Evolutionary algorithms|
Copyright in each Dissertation and Thesis is retained by the author. All Rights Reserved
The supplemental file or files you are about to download were provided to ProQuest by the author as part of a
dissertation or thesis. The supplemental files are provided "AS IS" without warranty. ProQuest is not responsible for the
content, format or impact on the supplemental file(s) on our system. in some cases, the file type may be unknown or
may be a .exe file. We recommend caution as you open such files.
Copyright of the original materials contained in the supplemental file is retained by the author and your access to the
supplemental files is subject to the ProQuest Terms and Conditions of use.
Depending on the size of the file(s) you are downloading, the system may take some time to download them. Please be