Empirical Approaches to Normative Theory

Societies are constantly trying to beat people into shape because they stubbornly fail to conform to some preconceived pattern of human possibility. Political theory is in this sense an empirical discipline whose hypotheses give hostages to the future, and whose experiments can be very costly. (Nagel, 1991, p. 7)Nagel's position has been taken to heart neither by philosophers nor normative political theorists. Both have been reluctant to embrace empirical techniques. Yet, some empirical elements must be understood if one is to tackle the philosophical questions of justice, fairness, and welfare. Here we address extending empirical techniques, such as experiments, ² into social and political philosophy as a means of clarifying the nature of reality about which we ordinarily know too little.

Introduction

The ethical status of a social decision rests on a number of factors, and philosophers differ in the weight they assign to them. For some, process is most important (e.g. Nozick, 1974); for others the outcome or consequences matter most. But there is virtual unanimity in all of modern western political philosophy that for the outcome of a social decision to be judged good, the process must positively reflect at least some of the values of the individuals involved.

Experiments allow one to examine both decision processes and individuals' values, to see how different processes affect the connection between the values and the outcomes and to determine how different processes affect the values of the individuals. We discuss these issues and reach some tentative conclusions about how some processes, specifically ones involving a form of impartial reasoning, can be invoked in the laboratory and can help us understand what constitutes good processes and ethically right decisions.

Understanding the Values of Individuals

Any focus on the welfare consequences of a social policy (and in modern democracies most such policies are evaluated in terms of welfare consequences) involves understanding the values or preferences of individuals. If the values held by individuals were either non-empirical, or as simple to measure as, say, the height of an individual, there would be less for us to concern ourselves with in this essay. But preferences appear to be both hard to measure and extremely fragile. Their instability has been discovered and documented in a set of famous, and seemingly ever expanding, set of laboratory experiments (Kahnemann, 1982; Tversky, 1988; Quattrone, 1988; Grether and Plott, 1979).

Value instability is not a question of under-specification. An individual's values need not be straightforward and simple: They can be conditional (e.g. John prefers sour dough bread to plain white bread unless he is having a peanut butter sandwich or jam with his toast in the morning). The problem is that preferences seem to vary not just with the specifics of the choice situation but rather how the situation is described, or framed. Put another way, John may have several incompatible preference structures in his head regarding a particular situation, that can be evoked by different ways of looking at that situation.

This creates a big problem for the ethical philosopher. If good social policy is to reflect the values of the people in the society, and if we cannot identify a unique set of values for each individual, one is forced to ask a fundamental philosophical question: "Which of the values of each of the individuals should be singled out for representation?" ³ And what determines which values are tapped? When do the expressions of the individual's most egoistic come to the fore? Is that material suitable for the construction of good social policy? Are egoistic desires the basis building ethical theory?

Beyond Egoism

Only a few philosophers argue that egoism should form the basis of ethical theory (see the discussion by Feldman, 1978). A few others object to any role for egoistic values in ethical choice (e.g. Kant). But that position is clearly too extreme for most. Individual values must count for something in determining actual decisions. There must...