Visualizing evolutionary activity of genotypes

We introduce a method for visualizing evolutionary activity of genotypes. F ollowing a proposal of Bedau and Packard [11], we define a genotype's evolu tionary activity in terms of the history of its concentration in the evolvi ng population. To visualize this evolutionary activity we graph the distrib ution of evolutionary activity in the population of genotypes as a function of time. Adaptively significant genotypes trace a salient line or "wave" i n these graphs. The quality of these waves indicates a variety of evolution ary phenomena, such as competitive exclusion, neutral variation, and random genetic drift. We apply this method in an evolutionary model of self-repli cating assembly language programs competing for room in a two-dimensional s pace. Comparison with fitness graphs and with a nonadaptive analogue of thi s model shows how this method highlights adaptively significant events.