- Last Modified: 27/1/12 12:43:28 PM
Real-life Frankenstein stories
The development of SEU theory was a majorintellectual achievement of the first half of this century. Itgave for the first time a formally axiomatized statement of whatit would mean for an agent to behave in a consistent, rationalmatter. It assumed that a decision maker possessed a utilityfunction (an ordering by preference among all the possibleoutcomes of choice), that all the alternatives among which choicecould be made were known, and that the consequences of choosingeach alternative could be ascertained (or, in the version of thetheory that treats of choice under uncertainty, it assumed that asubjective or objective probability distribution of consequenceswas associated with each alternative). By admitting subjectivelyassigned probabilities, SEU theory opened the way to fusingsubjective opinions with objective data, an approach that canalso be used in man-machine decision-making systems. In theprobabilistic version of the theory, Bayes's rule prescribeshow people should take account of new information and how theyshould respond to incomplete information.
Elissa Bassist & Cheryl Strayed
The assumptions of SEU theory are very strong,permitting correspondingly strong inferences to be made fromthem. Although the assumptions cannot be satisfied even remotelyfor most complex situations in the real world, they may besatisfied approximately in some microcosms--problem situationsthat can be isolated from the world's complexity and dealtwith independently. For example, the manager of a commercialcattle-feeding operation might isolate the problem of finding theleast expensive mix of feeds available in the market that wouldmeet all the nutritional requirements of his cattle. Thecomputational tool of linear programming, which is a powerfulmethod for maximizing goal achievement or minimizing costs whilesatisfying all kinds of side conditions (in this case, thenutritional requirements), can provide the manager with anoptimal feed mix--optimal within the limits of approximation ofhis model to real world conditions. Linear programming andrelated operations research techniques are now used widely tomake decisions whenever a situation that reasonably fits theirassumptions can be carved out of its complex surround. Thesetechniques have been especially valuable aids to middlemanagement in dealing with relatively well-structured decisionproblems.