Evolvability

Evolvability is the capacity of a system to generate heritable variation that can be acted on by natural selection, enabling adaptive evolution and the emergence of novel traits. It concerns the potential for change over generations, not any single instance of fitness improvement.

In biological systems, evolvability is shaped by genetic and developmental architecture. Factors such as mutation rate, recombination, and gene regulation influence the amount and usefulness of variation. Modularity and redundancy in genetic networks can limit harmful pleiotropy and promote future innovation. Phenotypic plasticity, canalization, and robustness can also affect evolvability by allowing organisms to explore new phenotypes without fatal disruption. Population-level properties, including size, structure, and mating patterns, further affect the availability of heritable variation and the path of evolution. Whether evolvability itself is selected for remains a topic of debate.

In artificial life and computer science, evolvability refers to a system's ability to produce varied and meaningful offspring representations. Evolutionary algorithms and digital organisms are used to study how encoding schemes, variation operators, and fitness landscapes influence long-term adaptability. Researchers aim to design representations that bias variation toward useful changes, for example through modular encodings or mechanisms that favor incremental improvements.

Measuring evolvability is challenging. Common approaches assess the size or connectivity of neutral networks, the degree of redundancy or modularity, and how readily new functions arise under environmental change. The concept is distinct from current fitness and from short-term adaptation, focusing on the system's potential to generate adaptive novelty across generations.