Complexity and social decision rules

This paper attempts to show how a particular concept and measure of complexity, as derived from automata theory, can be meaningfully interpreted in a program of 'limited rationality' regarding individual or social choices. The complexity measure appears to be a natural consequence of looking at a decision rule as a finite-state machine that computes preferences bounded by computational constraints. By factoring the social decision process into component processes it is demonstrated how searching for improvement depends on 'structural' and 'computational' limitations.