Issues of Observability

Issues of Observability
Because decision scientists cannot perform dissection, they are subject to a greater scientific discipline than that required of anatomists.
If a decision scientist tried to account for an individual’s purchases
of bananas as the direct result of something like an “appetite for fruit,” we would not know how to test this hypothesis—that is, we would not know how to independently “look for” such an appetite, even if we had a scalpel and an open, anesthetized brain. Such unob20
Machina
servable constructs like appetites, utility, and preferences can—and do—play a role in scientific decision theory, but only as inside links in a causal chain that ultimately starts with fully observable independent variables and ultimately ends with fully observable dependent variables.
For example, given the joint hypothesis that well-defined commodity
preferences exist and are also stable from day to day, standard consumer theory allows us to infer enough information about these preferences from an individual’s past demand behavior to be able to make refutable predictions about their future demand behavior, even for some combinations of prices and income never before observed.
In the following sections, we shall see that in passing from choice over certain commodity bundles to choice over uncertain prospects (either “objective lotteries” or “subjective acts”), hypotheses involving the unobservable constructs of commodity preferences and utility functions
can be replaced by hypotheses involving the unobservable constructs
of risk preferences and beliefs, which also link observable independent to observable dependent variables. Whether the notion of “states of nature” can similarly serve remains to be discussed.