Optimization of radiation protection as a special case of decision theory

Differential cost-benefit optimization is presented as a special case of more general procedures of decision theory that use probabilistically defined utility functions. These more general procedures can also be applied in cases where radiation exposures are only potential, to optimize the required level of safety features. It is shown that, for disruptive events of low probability and large resulting consequences, the concept of 'expectation' of consequences cannot be used in decision making, but that the use of probabilistically based utility functions can conceptually assure a consistent approach in deciding the required level of safety. The paper explores the use of utility functions of logarithmic form to assign weights to consequences involving different losses of life. (author)