Logic, Reasoning and A Programming Language for Simulating Economic and Business Processes with Artificially Intelligent Agents

5 Modelling Agents with Limited Rationality

A second issue is how agents adapt their knowledge and information in a world where Bayesian methods are inappropriate. A natural approach to dealing with this issue is to assume that agents reason about their environments and seek to identify the reasons why they perform well or badly in those different circumstances.

The use of declarative programming languages enables us to model this reasoning process when agents have limited computational and information-processing capacities. This is because instead of modelling only numerically defined conditions and outcomes, we are given a suitably expressive language in which to model the computational, learning and decision-making processes of agents. Credible internal states, which represent the structure of knowledge in relation to its environment, can be specified and modified as the agent develops. Since declarative programming languages are also easier to relate to formal logics, they enhance the prospects for developing rigorous models of processes involving intelligent agent behaviour.

If such models are not subsets of known formal logics, it might at least be possible to identify the logical properties which such models lack and then to decide whether the missing properties are important or even appropriately absent.

Logic, Reasoning and A Programming Language for Simulating Economic and Business Processes with Artificially Intelligent Agents - 12 APR 96

[Next] [Previous] [Top] [Contents]

Generated with CERN WebMaker