Genetic algorithms, teleological conservatism, and the emergence of optimal demand relations: The case of stable preferences (Q1341876)

Mainstream economics requires in the theory of demand that individuals maximize a given utility function under a budget constraint (with given prices and fixed income). Does the demand relationship which is generated from such a procedure require ``too much rationality'' from the human being? The author writes that ``real rationality consists in behavior according to heuristic rules which, while not optimal, give satisfactory results in a wide range of cases'' (p. 187). The paper distinguishes between and discusses various more ``cognitively realistic'' demand rules. Simple groping is a process of trial and error choice which is very rudimentary. The genetic algorithm is a method based on random variation and selection in which the random variation mostly takes the form of random recombinations of given elements. In consumer demand, both the mathematical specification and the parameters of the demand rules are coded as inputs to a genetic algorithm. The idea is that consumers periodically reconsider their demand routines and adopt in part the routines of other consumers who are more successful in matching their purchases to their needs. This can be characterized as a process of emulation. The third scheme is called groping over heuristics or teleologically conservative algorithm. This procedure takes the lottery element out of the genetic algorithm and emphasizes the element of selection. The latter method turns out to be particularly successful in the case that individuals differ over their utility functions. Simple groping performs rather poorly when the commodity price varies. The genetic algorithm produces an optimal demand relationship in the case of variable prices but is less efficient when consumers differ in terms of preferences. The reviewer would have preferred a more carefully written paper. Demand rule (3) does not follow from utility function (1). Why was utility function (1) chosen? Later on, the author worked with a different utility function (5b), which is not motivated either. In (6) optimal demand is independent of income, contrary to optimal demand in (3). Also, (6) contains an error. What happens to the results and findings in the case of more than two commodities?.