Karl-Dieter Opp recently published a programmatic article in this journal to demonstrate the plausibility of an individualistic explanation of social evolution. 1 I shall try to show that this programme is realizable only in a misleading and partially trivial sense.
In Opp's opinion any individualistic explanation is intimately bound up with the possibility of deducing statements describing changing states of systems from a set of statements formulated in terms of "properties of individuals". 2 As the basis of explanation Opp proposes a theory which includes elements of learning theory as well as elements of utility theory. The central thesis T which takes the main burden of explanation amounts to this: "If the reward value of the consequences of an activity A is greater than the punishment value of the (perceived) consequences of A and if this resultant value is greater than the respective value of any other activity, then A will occur". 3 Stable states of systems are then identified with the fact that all members of a society accept certain rules of behavior, because they believe them to be rewarding. Correspondingly societal evolution is defined as the change of the societal system-state resulting from the acceptance of a different set of rules as more rewarding or more profitable.
The theoretical point of this argument seems to be this: The state of a system is in the abstract determined as "the distribution of values of a set of those variables the system was described by". 4 Consequently the possible change of a system will depend on the very fact that "the value of at least one variable had changed". 5 As the application of Opp's version of learning theory requires that "if we want to explain the properties of social systems by individual sentences then the properties must be constructed from the properties of individuals ''6 any acceptable explanation of a system-state results from the validity of a theory like T and the specific distribution of perceived rewards among the members of that system. This differential distribution of reward values represents the set of initial conditions within Theory and Decision 14 (1982) 427-434. 0040-5833/82/0144-0427S00.80.