Learning, adaptation, and the complexity of human and natural interactions in the ocean

Here, I explore the system-level consequences of learning and adaptation among fish and fishers. The fundamental idea is that the cost of acquiring the knowledge needed to resolve uncertainty is the principal driver of social and spatial organization. This cost limits agents’ actions and leads them to prefer relatively persistent associations with familiar agents and places. When all agents act in this way, the regularity and self-reinforcing nature of familiarity leads to the emergence of a self-organized system. Systems like this are characterized by diverse, place-based, and relatively durable groups, groups of groups, and rough hierarchical structure. This occurs in both the natural and human parts of the system. The costs of resolving uncertainty also determine the interactions of fish and fishers. The uncertainty of search leads fishers preferentially to target older fish and aggregations of fish. These are the repositories and mechanisms for the replication of the knowledge needed for self-organization. The loss of this information selectively, but unintentionally, disrupts the behavioral regularity that organizes the natural system, leading eventually to its disorganization. From this theoretical perspective, sustainable fishing requires conservation of the knowledge in DNA and memory because this is the fundamental basis for the self-organization of the natural system. Collective action is also subject to the costs of resolving uncertainty. In complex systems, these costs are minimized at the local level in the system, where the most direct, but not the only, feedback occurs. This implies the need for multiscale governance with an emphasis on collective learning through localized science and user participation. Finally, the complexity of ecosystem interactions argues for qualitative harvesting rules governing how, when, and where fishing takes place. These rules are most likely to generate a persistent signal and rapid learning, but only when combined with effective governance.