When I first encountered the economic theory of fishing I was astounded to learn that fishermen were assumed to be perfectly knowledgeable about the location of fish (because the fish were assumed to be evenly distributed in space). The reason these assumptions are made, of course, is because to do otherwise would create a very difficult, probably impossible, mathematical problem.

It is extremely difficult to use mathematics to describe the patchiness and diversity of a marine ecosystem, to say nothing of the way fish and fishermen in that system interact. But characterizing fishing so that it is convenient for our mathematical abilities, as economic theory does, completely ignores the fishermen’s problem of learning about and adapting to a complex biological and social environment. As a result economic theory does not have the ability to understand the circumstances in which individual behavior leads to overfishing (or conservation). In a more general sense, the challenge for economic theory is to find a rigorous methodology to analyze the way learning and adaptation affects individual behavior and the organization of economic activity. From this perspective, fishing provides a simple metaphor for normal, knowledge driven, economic processes.

In our paper we describe an evolutionary, computational model (a simulation) in which fishermen are engaged in a competitive search for a valuable resource, lobsters, located in a complex biological and social environment. The evolutionary process in the model concerns the way each fisherman learns about and adapts to his environment, i.e., develops better decision rules to guide his fishing activities. The computational approach we use is termed a learning classifier system⁠. Borrowed from computer science, it mimics a standard Darwinian process and allows us to build a workable model of a complex biological and social environment, in much the same way that individuals can build a workable life in a large and complex environment.

Our virtual fishermen begin the simulation with no useful knowledge (i.e., only random decision rules) and a very small sample of their environment (i.e., a few traps placed at random locations). From that initial sample each fisherman begins to associate certain environmental conditions with better or worse fishing outcomes and uses this very limited knowledge to rearrange his traps, searching for conditions that might yield a better outcome. Repeating this very imperfect process, each fisherman gradually builds a set of ‘fit’ decision rules that tend to yield better results – a higher catch – than other rules.

What fishermen know and the actions they take are strongly dependent on of the costs of acquiring information. Because fishermen have limited time, they can search only a small part of their environment and can communicate with only a limited number of other fishermen. As a result each fisherman learns different things, develops a different history and has different, path dependent, perceptions of the value, scope and direction of the opportunities open to him at any time. In our model we assume fishermen have better, more accurate, communications with fishermen they know well. We can adjust the accuracy (i.e., the opportunity costs) of communications with other fishermen; we use that ability to explore the kinds of individual behavior and social organization that result from different costs of information.

When those costs are close to zero (i.e., when communications among all fishermen are accurate), there are no advantages to knowing someone else well and no self-organized social structure results; in addition, the ease of communication among individuals leads to quick and constructive social adaptation to changing conditions. When information costs are high, however, the value of communicating with familiar people (rather than strangers) is also relatively high, leading fishermen to form persistent individual relationships and equally persistent small groups. However, because they have to compensate for the higher costs of communication, the speed of their adaptive response declines considerably. Consequently, except when the costs of information are close to zero, the model generates an evolving, imperfect system with diverse, imperfect (boundedly rational) individuals.


Read the paper: Wilson, J., Yan, L., & Hill, J. (2013). The Evolution of Self-organization in a Small, Complex, Economy.

Published On: July 3, 2013

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.