In a capitalist democracy there are essentially two methods by which social choices can be made: voting, typically used to make ‘political’ decisions… - Kenneth Arrow

Various medical groups are continually putting out advice as to health-inducing behavior (diet, exercise, and so forth). Individuals who follow this advice will reduce their demand for medical services. An extreme version is the campaign of the American Dental Association for fluoridation of water, a measure designed to reduce cavities and so the demand for dental services.
There is also a logical problem. How does the relatively small group of physicians acquire power? In a democratic system, political power derives from numbers or from wealth. While physicians are certainly relatively well off, they do not have the concentration of wealth of many industries. Their power, it seems to me, is derived from moral authority. It is derived from the same professional respect that we are trying to explain. Indeed, when the American Medical Association came into conflict with either the retired or with the business interests that bought medical plans, the weakness of its power base quickly became apparent.

If the state of information (the set of signals received) is given and constant, then optimal choice is a problem of decision making under a given uncertainty, a situation that has been the subject of considerable analysis in the last thirty years. The problems of the economics of information proper arise when the probability distribution of states of the world is a variable. In the language adopted here, the signals received can vary. The existence of signals creates two important possibilities for the improvement of decision making. The first is taking advantage of the existence of signals. If the individual knows that a signal will be received before the decision has to be made, his optimal choice should be a function of the signal. We can think, alternatively but equivalently, of making the decision after the receipt of the signal and basing it on the probability distributions of consequences conditional on the signal, or of making the decision in advance for all possible values of the signal.

Learning, as studied by psychologists, closely resembles sequential analysis in some aspects. Learning experiments usually consist of a series of trials in which the subject’s choices are sometimes rewarded and sometimes not. The individual, after making many choices, eventually begins to discriminate between the proper response and the improper one. At some point, presumably, he could terminate the experiment, at least in the sense of disregarding the further observations and making the same choice each time.