[Next] [Previous] [Top] [Contents]
Modelling Bounded Rationality In Agent-Based Simulations using the Evolution of Mental Models
If you seek to model real economic agents then, unless you make some very sweeping assumptions, the entities in your software model will also need the broad characteristics of the real agents. This is in contrast to main-stream economics where, by and large, the agency nature of the agents is ignored, in favour of trying to capture their behaviour en masse.
The purpose of an agent in such a model is also different from agents that are designed with a particular purpose in mind or for exploration of the most effective and flexible algorithm for a set of problems. In such modelling one seeks for as much veracity as is possible given the usual limitations of time, cost and technique and one does not necessarily look to design them to be efficient, general, or consistent in their beliefs.
In particular we are interested in agents who:
- do not have perfect information about their environment, in general they will only acquire information through interaction with the dynamically changing environment;
- do not have a perfect model of their environment;
- have limited computational power, so they can't work out all the logical consequences of their knowledge;
- have other resource limitations (e.g. memory);
In addition to these bounds on their rationality, other characteristics are included, namely:
- the mechanisms of learning dominate the mechanisms of deduction in determining their action;
- they tend to learn in an incremental, path-dependent [3] (or "exploitative" [26]) way rather than attempting a global search for the best possible model;
- even though they can't perform inconsistent actions, they often entertain mutually inconsistent models.
There are several possible ways of using evolving populations to simulate a community of economic agents:
- each member of the evolving population corresponds to one agent;
- each agent could be modelled by a whole evolving population;
- the whole population could be modelled by the whole evolving population but without an individually intended agent <-> gene correspondence.
Method (1) has been used in several models of agents which evolve (e.g. [15, 29]), here the genetic development has nothing to do with the nature of an agent's cognitive processes but helps determine its goals or strategies. Method (3) above is popular in economics (e.g. [2, 4]), but unless such a model predicts pertinent properties of real populations of agents, it is a bit of a fudge, and means that the observable behaviour and content of individual entities in the model do not have a clear referent in what is being modelled. This makes it far less useful if one wants to use such models to gain a detailed insight into the internal dynamics of populations. Method (2) actually addresses the cognitive process as the agent corresponds to a population of mental models. This has been done before in a limited way in [25], but here agents have a fixed menu of possible models which do not develop.
Modelling Bounded Rationality In Agent-Based Simulations using the Evolution of Mental Models - 17 MAR 98
[Next] [Previous] [Top] [Contents]
Generated with CERN WebMaker