Exploring Embodied Adaptive Behavior through Evolutionary Computation and Phenomenological Insights from a Consciousness-Inspired Perspective
Year: 2025
Type: dissertation
Abstract: This dissertation integrates phenomenological principles into the empirical study of consciousness and its computational modeling in artificial agents by proposing a theoretical framework and exploring its computational representation. The framework identifies four essential phenomenological properties of consciousness and critiques existing theories, advocating for a more integrative approach. Chapter 2 examines key experiential properties of consciousness through phenomenological practice, while Chapter 3 critically evaluates four major empirical theories, assessing their compatibility with phenomenology and proposing a reconciliatory framework. The dissertation then investigates how computational interpretations of consciousness influence the behavior of simulated artificial agents. Chapter 4 analyzes changes in integrated information in Braitenberg Vehicle-based agents evolved through Genetic Evolutionary Algorithms (GEA). Chapter 5 extends this inquiry with a more complex simulation, where agents dynamically interact with their environment through an evolutionary behavioral selection process, evaluating how integrated information and prediction surprisal minimization contribute to adaptive behavior. These findings provide an empirical grounding for the theoretical framework and open new possibilities for future computational representations of intentionality and agency.
Author: Pae, Hongju
Institution: University of California, Irvine
Language: English
Subfield: Philosophy
Field: Arts and Humanities
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W6991156327
