Evolution, Plasticity, and Forgetting: Adaptation in a Simple Dynamic Environment

Abstract

This record contains a small exploratory computational research project investigating the interaction between evolution, lifetime plasticity, and forgetting in a dynamically changing environment. The experiment uses a 2–6–2 neural network and compares three conditions: Evolution only (E), Evolution with Plasticity (EP), and Evolution with Plasticity and Forgetting (EPF). The environment undergoes periodic reversals every 20 generations over a 100-generation experiment. The results indicate that lifetime plasticity improves adaptation compared with evolution alone. The overall comparison between EP and EPF does not show a statistically significant improvement. However, EPF exhibits several descriptive patterns associated with adaptive robustness, including rapid recovery during later environmental reversals and a more concentrated final-fitness distribution. These observations suggest that forgetting may potentially influence adaptation by regulating the persistence of previously acquired information. This interpretation is exploratory and should not be considered a confirmed causal effect. The uploaded materials include the research paper, experimental source code, recorded experimental history data, and summary results.