The study of evolutionary development (evo-devo) is frequently challenged by the scales of space and time complexity inherent to its study. This has led to the creation of abstract models to allow for the exploration of evo-devo in a manner that is both more computationally feasible and more general, without ties to the specific biological processes of a single organism. Our work expands upon these previous models by introducing an indirect encoding for developmental mechanisms, dynamic fitness landscapes, and a phenotypic structure that allows for the exploration of new interactions between the developmental and evolutionary processes. Introducing these changes allows us to conduct a more thorough study of factors impacting evo-devo. Our experimental results suggest a number of parallels to biological systems. These include representing the synergy of evolutionary and developmental processes, the evolution of adaptable features, and highly conserved regulatory genes. We also discuss the opportunities for exploration opened by this new model. These possibilities include the study of developmental exaptations and the robustness of developmental strategies.

Evolution and development are related processes although their relationship is still not well understood. Attempts to explore their relationship are challenged by scales of time and space, but also by the limitations of studies focused on specific constraints of model organisms. To help gain insight into these phenomena, we create an abstract, general model of a developmental process that guides an agent’s trajectory through a “tunably rugged” NK fitness landscape. The developmental process is represented by a genotype that is evolved and allows us to investigate periods of exploration and exploitation as they relate to periods of an agent’s lifetime and a given landscape’s difficulty. Results show that evolution selects for time-sensitive periods of exploration and exploitation, which vary with the difficulty of the landscapes being traversed. Furthermore, our analysis suggests that phenotypic diversity via random exploration present in both early and mid-life can aid the development of superior phenotypes.

In the spirit of past century evolvable hardware, we explore the application of evolutionary algorithms to field programmable gate arrays and provide an open-source platform for performing intrinsic analog evolvable hardware experiments. We target the reproduction of seminal field experiments that generated complex analog dynamics of unclocked FPGAs which were evolved through genetic manipulation of their binary circuit representation: the bitstream. Further, we demonstrate the intrinsic evolution of two non-trivial analog circuits with intriguing properties: amplitude maximization and pulse oscillation. Lastly, we explore the robustness of evolved circuits to temperature variation and across-chip circuit translation.