One of the primary areas of scientific research is understanding how complex systems work, both structurally and functionally. In the natural world, complex systems are very high dimensional, with many interacting parts, making studying them difficult and in some cases nearly impossible. Due to the complexity of these systems, a lot of modern research focuses on studying these systems from a computational viewpoint. While this necessarily abstracts away from the true system, we attempt to represent the salient aspects of the system in order to better understand it. Results from such computational studies can yield insight into the natural system, and actively constrain the research space by suggesting hypotheses that can be tested.In this work, I investigate the structure and function of two seemingly disparate complex digital systems. I begin with an investigation of the structure and function of an evolved cognitive architecture, and look at how this structure is affected by environmental changes by developing some new metrics to classify cognitive systems. I then look at the structure of the primordial fitness landscape in a different digital system, and use techniques inspired by information theory to understand the structure of this landscape. I first look at the role of historical contingency in the evolution of life by studying how the structure of this fitness landscape affects the evolutionary trajectories of life. I then investigate how information is encoded in the primordial fitness landscape. I then extend this analysis by developing a general approach for calculating the information content of individual sequences, and use them to analyze the primordial landscape. Finally, I validate this information-theoretic technique by predicting the effects of mutations on the function of a specific protein, and show that this technique can outperform the current state of the art approaches.
Read
