| Agents are of central importance to cognitive science, but research usually takes them as pre-given and proceeds to study some of their particular aspects, often without awareness of or a definite answer to the question, "what is an agent?" The conceptual framework of autopoiesis and enaction provides a foundation that defines agents as emergent individuals that act in an environment to fulfill their physiological needs.;To establish this definition in concrete examples, this dissertation introduces computational models and analyses that demonstrate several properties of agents. It examines an artificial chemistry that supports the emergence of minimal protocells. These protocells have a metabolism made of autocatalytic components, and an external boundary that self-assembles and encapsulates the metabolism, keeping it from diffusing into the environment. Metabolism and boundary are mutually enabling processes, which together counter the effects of diffusion and decay. When their symbiosis is broken, the protocell disintegrates.;Systematic analysis reveals the rich consequences of protocellular organizations. Ontogenies are mapped as network structures, with the networks' nodes as reachable protocell morphologies and its edges as the transitions between morphologies. Analyses of these networks reveal properties such as irreversibility (some changes cannot be reversed under any circumstance) and branching (unfolding down one ontogeny excludes morphologies accessible by other ontogeny). Viability is quantified as expected lifespan, and measured across different protocell configurations. This provides a basis for measuring agents' basic goal of adaptivity --- to increase their viability in a given environment through internal restructuring or environmental change.;The cellular Potts model (CPM) framework is examined to study structural coupling (the bi-directional interactions between an agent and environment). The network-based methodology for analyzing ontogenies is extended to incorporate a local environmental state and is demonstrated in a CPM. This reveals several interesting features, such as a divergence in the space of possible ontogenies when placed in different environments, and that niche construction can increase an individual's viability. |
