This dissertation addresses a fundamental problem in the Artificial Intelligence field of qualitative physics: extending qualitative simulation and analysis to biological systems, and complex systems in general. A major problem in analyzing complex systems is the plethora of individual components and behaviors, which need to be organized into some coherent description of the working system. One way to conceptualize the myriad behaviors of a complex system is by applying knowledge of teleology (i.e. purpose).

Using the domain of circulatory physiology as an example, this dissertation integrates knowledge of structure, behavior and teleology to extract comprehensive explanations of complex systems. Circulatory physiology is a fruitful domain for teleological investigation for two important reasons: (1) unlike engineered artifacts, which generally have a single, well-documented purpose, biological systems admit many teleological interpretations, all of which may simultaneously apply, and (2) biological systems exhibit purposeful behavior when viewed from both the steady-state and dynamic perspective.

This dissertation formalizes both of these perspectives in a manner indigenous to computer simulation. It develops the Bipartite Teleology Model (BTM), which consists of four generic teleologies: transport, conservation, dissipation, and accumulation, along with a set of heuristics for evaluating a circulatory system's ability to satisfy these purposes during both steady-state and dynamic conditions. Furthermore, BTM provides guidance in the explanation of circulatory behaviors in terms of circulatory structure.

This dissertation describes The Biology Critic (BIOTIC), an implemented computer system which analyzes circulatory systems using BTM along with standard qualitative-simulation algorithms and a well-defined critical context. BIOTIC evaluates a circulatory system's ability to satisfy teleologies such as oxygen transport or heat conservation while the organism performs various activities in various environments. When run on qualitative models of reptilian and mammalian circulatory sytems, BIOTIC produces critiques similar to those published by expert physiologists. Furthermore, BIOTIC can criticize a wide variety of existing and novel circulatory systems relative to many different contexts. These results support my central tenet that the combination of qualitative simulation and teleological interpretation suffices to produce intelligent analyses of complex biological systems.