Choosing Justice – An Experimental Approach to Ethnical Theory: 22 (California Series on Social Choice and Political Economy) - Hardcover

Frohlich, N

Hardcover

ISBN 10: 0520072995 ISBN 13: 9780520072992

Publisher: University of California Press, 1992

Synopsis

This book presents an answer to the question "What is fair?" In their approach to ethics, the authors argue that much of the empirical methodology of the natural sciences should be applied to the ethical questions of fairness and justice. Moving ethical theorizing out of the armchair and into the laboratory, the authors test John Rawl's theory of distributive justice with cross-national experiments from Canada, Poland and the USA.

About this title

About the Author

Norman Frolich was Professor Emeritus at the Asper School of Business at the University of Manitoba. Joe A. Oppenheimer is Professor Emeritus in the department of Government and Politics at the University of Maryland.

Excerpt. © Reprinted by permission. All rights reserved.

Choosing Justice: An Experimental Approach to Ethical Theory

By Norman Frohlich and Joe A. Oppenheimer

University of California Press

Copyright 1992 Norman Frohlich and Joe A. Oppenheimer
All right reserved.
ISBN: 0520072995
Chapter One
Empirical Considerations Concerning
Impartial Reasoning

Where does empirical inquiry fit into a quest for a theory of distributive justice? One answer to that question is direct and simple: it stems from the role of impartial reasoning in determining rules for just distributions. Specifically, we advocate empirical work because it is difficult to determine the conclusions of impartial reasoning. To show why empirical methods might be reasonable and useful, we investigate some of the bases of moral reasoning related to our approach.

The Tradition of Impartial Reasoning in Ethical Theory

One hallowed tradition of ethical inquiry is the invocation of impartial reasoning as a basis for achieving moral knowledge. As far back as the first century B.C. , Publius Syrus of Rome noted that when disputes arise, there is an inevitable problem of bias. His dictum was "No one should be judge in his own case."1 The intuition here is that we are likely to assign undue weight to our own interests. To obtain impartiality, third parties, not involved, should be called on to judge. In the days

Maxim 545, as quoted in Bartlett (1980, p. 111).

when wool was symbolic of commerce and wealth, members of the British Parliament had to declare their financial interests in a bill being debated. Members with such interests were required to retire to the door of the Commons and sit on a wool sack. Any arguments these members might offer were presumed to be unduly colored by self-interest. Sitting on "the wool sack" made this partiality concrete.

Indeed, the most famous moral dictum, the Golden Rule (do unto others as you would have them do unto you), arguably "the common moral denominator of all the world's major religions" (Gewirth 1978, p. 133), is an application of impartiality. It calls for each of us to lend equal weight to the concerns of our fellows and even offers a simple method for invoking the requisite impartiality. "The Golden Rule is predicated upon an exchange of viewpoints, as are maxims such as that urging us to stand in another man's shoes" (Henberg 1978, p. 723).2

As M. C. Henberg also notes (1978, p. 723), others have sketched ways in which this impartiality might be accomplished. "C. I. Lewis, for instance, suggests that impartial valuations are best rendered by imagining that the experience of all concerned persons were one's own, 'as, for example, if you had to live the lives of each of them seriatim.' Similarly, Richard Hare suggests that people who render differing moral judgements should imagine that their desires and inclinations are exchanged for the desires and inclinations of their antagonist."3 The intuition behind these and many other similar positions is simple: to be fair, we are to project ourselves into the position of the relevant others and give equal weight to their concerns. The judgment about what is fair must be based on an impartial point of view.

Note that even in such technical works as those of Sen (1973, pp. 1415) where he discusses equity his stance (which he refers to as the Weak Equity Axiom) on interpersonal comparability is not far removed from this position.

The internal citations are to Lewis (1946, p. 547) and Hare (1963, p. 123).

Structuring an Impartial-Reasoning Device

One traditional proposal for determining the content and implications of impartiality is to adopt the perspective of an ideal observer. Such an observer would have to be endowed with certain powers to be capable of making the appropriate moral judgments. Analysis of these properties has varied, but their definition is always designed to ensure that such observers would exercise impartial reasoning in their judgments.4 Such a proposal clearly requires that the observers have considerable knowledge plus exquisite powers of reasoning. No mere moral philosophers could claim, even in their wildest dreams, to approach these qualities even though they argue that such an ideal observer might be required to identify right principles.

Were it a straightforward matter to exercise this device to discover ethical principles, we would be making observable and indisputable progress in this area. The absence of unequivocal headway in moral theory likely reflects the difficulty of identifying and implementing a state of impartiality (in fact and principle) from which to reason. No real observer can be ideal. Real-world deviations from the idealizations of philosophers can be expected to lead to deviations in the acceptability and quality of their conclusions.

Thus, one reason for the failure to make headway in the development of theory and in the discovery of principles about justice and fairness is the inability of real people to be consistently impartial. Two clearly visible barriers are central to this failure. Individuals are unable to shed their own interests, and they have their unique perspectives, which are products of their cultural heritage, life experience, and genetic endowment. These factors inevitably lead the theorist to import biases into any attempt to reason as an impartial observer. In the end,

For example, Firth (1952) has specified that an ideal observer needs to be omniscient with regard to relevant nonethical facts, omnipercipient (able to empathize perfectly) with all the relevant parties, disinterested and dispassionate as among the parties and toward the issues involved, consistent (over time), as well as normal in other respects.

these shortcomings lead to ambiguity, confusion, and dispute over the fundamental issue at stake: the nature of the alternatives that would be selected. Evidence that the problem is susceptible to these difficulties includes the lack of consensus among philosophers who have used this approach to generate ethical principles. Thus, imagining an ideal observer is easier than finding one. Perhaps for this reason other proposals for identifying fair outcomes via impartial reasoning have sprung up. Some of these proposals are for devices that are disarmingly intuitive, simple, and effective. Indeed, some have been widely adopted.

Cutting Cakes

An example of a device or procedure that generates an impartial outcome is ready at hand. For two people to divide a piece of cake fairly, one person cuts and the other chooses. The principle underlying this device is elementary. The job of setting up the alternatives is separated from the job of making the choice. The general acceptability of this process is obvious. The slicer knows that any inequity between the portions is likely to work to the disadvantage of the slicer. Thus the person who is setting up the payoff structure (by slicing the cake) has every incentive to be as fair as possible in the division. So "you cut, I choose" is a universally acceptable procedural solution to the problem of two parties fairly dividing a fixed quantity.5 This procedure has a major implication: some ethical problems have universally acceptable solutions and implementable devices for identifying them.

Although extensions of this notion to more than two persons are possible, it becomes ever more complicated as the problem of fair division becomes more complex. When the division is to be among workers who toil together to make a joint product, the procedure is inadequate, for then the quantity being divided is not fixed but is a function of the work done by

Some extraneous considerations such as role or status differentiation could lead the device to be rejected in particular instances.

each participant. The amount received today could be a factor determining the team member's effort tomorrow. Subtle adaptations might be required to structure a fair procedure in such complicated cases.

Imperfect Information

As mentioned in the Introduction (note 1) the device of imperfect information may be useful in dealing with complex problems of fair division. Rawls articulated a particular set of conditions of imperfect information and called them a "veil of ignorance" (1971, pp. 136137):

The idea is to set up a fair procedure so that any principle agreed to will be just. Somehow we must nullify the effects of specific contingencies which put men at odds and tempt them to exploit social and natural circumstances to their own advantage. To do this I assume that the parties are situated behind a veil of ignorance. They do not know how the various alternatives will affect their own particular case and they are obliged to evaluate principles solely on the basis of general considerations.

Rawls proposed, in short, a kind of thought experiment. We are to imagine individuals (representative of classes and other social positions) deprived of all particular information about their tastes, talents, dispositions, and so forth, and asked to play the role of judge regarding rules to govern a society that they and their progeny are to inhabit. The veil prevents them from knowing the particular role they are to play in that society. This imperfect information induces conditions that generate impartial reasoning.

The conditions of Rawls (and Harsanyi) give all individuals an equal stake in every possible payoff because they do not know who they will be, and, therefore, their interest is to be fair to all. In addition to this incentive to be evenhanded, the individuals can be presumed to have the other characteristics of the ideal observer.6 Rawls, who develops the argument more

Hare (1973), in his incisive review of Rawls, notes a correspondence between Rawls's veil of ignorance and ideal-observer theory. The reader can compare Rawls's list with Firth's (see footnote 4 above). But note that there are some significant differences (including the fact that an impartial observer would have to take into account interacting utilities) and that Rawls's and Harsanyi's derivations posit (perhaps counterfactually) that people are self-interested.

fully than Harsanyi, assumes that these individuals have knowledge of the general laws governing society, can project themselves into all the possible states they might occupy, do not have any particular attachment to any given individual, and deliberate calmly until they reach a reflective equilibrium. He concludes that, as a group, such individuals would unanimously agree on a single principle.

It is possible to characterize the generic argument that lies behind this methodology of Rawls and Harsanyi. Both argue that impartial reasoning establishes the validity of a particular principle of justice. The syllogism to that effect might go as follows:

C1 ,, Cn are the ideal conditions of impartiality. Any principle unanimously accepted under ideal conditions of impartiality is a valid principle of justice. Under C1 ,, Cn principle P would be accepted unanimously. Therefore P is a valid principle of justice.

Harsanyi and Rawls then use thinking about how such individuals might choose as a basis for determining a just organization of society. No doubt this device casts a powerful light on the issues at hand, but it is clearly filtered through the eyes of the beholder. So it is not surpising that they reach different conclusions about which principle will receive universal support.

The reasons for the divergence are not hard to identify. For individuals behind the veil to make judgments about principles of distributive justice, they must have at least two things: some residual preferences regarding alternative states of the world and some notions of how different choices and behavior relate to achieving preferred states.7 These preferences can vary.

They must, in other words, have some knowledge of causal relationships about empirical matters.

Rawls tries to confine this variance by limiting the objects to be considered. The only explicit notions of "the good" that Rawls (1971, p. 142) allows behind the veil are "that they would prefer more primary social goods rather than less." Here Rawls (p. 92) defines primary social goods as "rights and liberties, opportunities and powers, income and wealth." Thus a rule of distributive justice is reduced to a rule for dividing primary social goods. But even a short list of primary social goods leaves open the question of potential tradeoffs between quantities of the items on the list. Even if individuals prefer, as posited, more to less, how much gain in power should be traded for how much decrease in wealth or income, and so forth? These questions are important because the agents must live with their choices.

Agents would later gain further information and occupy and judge the states they have created. That post hoc judgment would determine the acceptability of the decisions taken behind the veil. If application of a procedure leads to general post hoc dissatisfaction, the result is not likely to be stable. But how can individuals, behind a veil of ignorance, know how they will feel with their recovered tastes? In the absence of that knowledge, there is no way of specifying ex ante what their choices might be. Thus, two items hamper reasoning about what might be concluded under conditions of impartiality:

1. the difficulty of specifying the content of individuals' knowledge in that state 2 the difficulty of predicting the acceptability of the choices after the individual exits that state of ignorance

In other words, theorists seem unable to provide a rich enough texture for the hypothetical individuals who are making the decisions. Thus, reasoning about what might happen under ideal conditions of impartiality is fraught with peril: the employment of the mere procedures as a guide leaves one with ambiguous conclusions. As Stephen Darwall (1983, p. 94) notes in discussing the question of the motivating power of preferences: "If we put so much distance between ourselves and the objects of our intrinsic

preferences that what initially attracted us about them no longer does so at all, then we are left with no basis for choice." To the extent that Harsanyi and Rawls cannot be precise regarding those tastes that remain vested in individuals behind the veil, their conclusions are subject to biases in their projections. And any individual can conceivably be subject to some distortion of perspective when viewing the original position through the lens of his or her own experience, knowledge, and intuition. As Hare puts it, "The truth is that it is a wide open question how the POPs [people in the original position] would choose; he [Rawls] has reduced the information available to them and about them so much that it is hard to say what they would choose, unless his own intuitions supply the lack" (1973, p. 250).

It is difficult to project one's own values and expectations into situations that have consequences for others. It is even more problematic when those others have life experiences, values, and goals that we can only dimly imagine. A single philosopher, with the best of intentions, trying to find a fair rule to govern the future lives of diverse individuals can make serious miscalculations. Even the best efforts to be fair and impartial may fail because of an inability to consider properly the real perspectives and needs of others.

Thus, it should not be surprising that different views of human nature lead Harsanyi and Rawls to different conclusions about how people would choose under uncertainty. Again, Hare (1973, p. 249) puts it succinctly:

The POP game is in effect played by imagining ourselves in the original position and then choosing principles of justice. Rawls' POPs come to the decisions that they come to simply because they are replicas of Rawls himself with what altruism he has removed and a veil of ignorance clapped over his head. It is not surprising, therefore, that they reach conclusions which he can accept.

And there is yet another source of difficulty. Although individuals are severely limited in the particular information they have, they are presumed to have certain general knowledge at

their disposal. They are "presumed to know the general facts about human society. They understand political affairs and the principles of economic theory; they know the basis of social organization and the laws of human psychology" (Rawls 1971, p. 137). Alas, given the current state of knowledge in the social and behavioral sciences, it is absolutely safe to assert that there is no consensus on those laws and principles. Nor can there be any assurance that, even were a consensus to exist, those laws would point in a single "correct" direction. Thus Harsanyi and Rawls implicitly rely on unarticulated and potentially suspect scientific principles to reach their conclusions, and, being imperfect, they not surprisingly make different inferences. There is no objective way of identifying, assigning, and hence using a single set of general principles as the basis for reasoning from behind the veil. Any explicit assignment of basic knowledge is sure to be contestable; any implicit assignment, suspect.

Thus, those who argue for imperfect information as a device for promoting impartial reasoning are vulnerable to the assertion that implicit subjectivity colors their conclusions. They are insufficiently explicit about both the personal and the general knowledge held by the uninformed individuals. This lack of explicitness leaves them open to a criticism made by Thomas Nagel (1986, p. 10) in his discussion of progress in philosophy: "We also have to recognize that philosophical ideas are acutely sensitive to individual temperament, and to wishes. Where the evidence and arguments are too meager to determine a result, the slack tends to be taken up by other factors. The personal flavor and motivation of each great philosopher's version of reality is unmistakable." Thus, the deductive and introspective traditions have inherent difficulties in identifying the content of impartiality. Their attempts to strip individuals of their identity to generate impartiality are doomed to yield a vacuous version of impartiality.

We therefore propose to look elsewhere for the content of impartiality, and we suggest that empirical evidence may be generated to identify this content. This is not a great departure from the commentaries on this literature. For example, Hare

has written of Rawls's theory: "Rawls does not conceive of moral philosophy as depending primarily on the analysis of valid moral argument. Rather, he thinks of a theory of justice as analogous to a theory in empirical science. It has to square with what he calls 'facts,' just like, for example, physiological theories. But what are the facts?" (1973, p. 145).8

The Role of Empirical Methods

As we have seen, mere hypothesizing and reasoning about what might happen under conditions of impartiality are doomed to ambiguity. Still, a device producing conditions of imperfect information could be an acceptable generator of impartial reasoning. Although it is clearly impossible to replicate the ideal conditions of impartial reasoning, one can model and approximate them. Were we to do so, the general syllogism above would be replaced by a new one with empirical content. This content provides one possible relationship between empirical inquiry and these ethical issues.9 We begin the new syllogism by maintaining the basic premise:

Any principle unanimously accepted under ideal conditions of impartiality is a valid principle of justice.

and replacing the other statements as follows:

C 1 *, , C* n are experimental approximations to the ideal conditions of impartiality.

Any principle unanimously agreed on under experimental conditions C 1 *, , C* n has a claim to be a valid principle of justice.

Because we have constructed a series of experiments that attempt to approximate the ideal conditions, our claim is that:

Any principle unanimously agreed on in our experiments has a claim as a valid principle of distributive justice.

See footnote 11 below, and the surrounding discussion for more on Hare's point.

We are indebted to Thomas Schwartz for suggesting this particular form of the argument.

Yet another, closely related, argument can be made about the rejectability of some candidates for principles of justice. Two supplementary syllogisms could be constructed parallel to the previous syllogisms. The first would be based on the following premise as a substitute for the original confirmatory ideal premise:

Any principle incapable of getting substantial support under ideal conditions of impartiality is presumably rejectable as a valid principle of justice.

Again, because ideal conditions are not realizable, we develop a second argument by substituting empirical approximations of ideal conditions:

C 1 *, , C* n are experimental approximations to the ideal conditions of impartiality.

Any principle incapable of getting substantial support under experimental conditions C 1 *, , C* n can be presumed to be rejectable as a valid principle of justice.

And because, as stated above, we constructed experiments to approximate the ideal conditions, our claim is that:

Those principles identified as incapable of getting substantial support under our experimental conditions can be presumed rejectable as valid principles of justice.

Together the two arguments do not quite make for necessary and sufficient conditions. Yet our claim is strong: (1) Principles that survive with unanimous support have a claim to validity as principles of justice. (2) Those that do not show any strength at all are presumably rejectable. Clearly there are big loopholes in these claims. Some aspects relevant to the analysis of distributive justice might not be properly considered, and the final determination might be mistaken. Some principles weakly affirmed or not strongly rejected might fare better under other approximations of the ideal conditions. Considerable variation in experimental conditions may be required to get a feel for the robustness of any claims.

To make these abstract arguments concrete and to generate

falsifiable conclusions, we begin with the conceptions of impartial reasoning implicit in Rawls's and Harsanyi's models of imperfect information. We develop a laboratory simulation to approximate those conditions and place subjects under those conditions to discover what they choose. This procedure shifts the grounds of the argument from the purely analytical to the empirical. This shift is justified because the central question raised by contractarian theories (such as those of Harsanyi and Rawls) is empirical. The central question is not whether a contract has ever been entered into but whether such a contract would ever be entered into under the specified conditions, and, if so, what its content would be.

There are two traditional ways of evaluating imperfect-information arguments. The first is to question the normative premises that underlie the theory. The second is to examine the validity of the logical inferences. We open a third door. We suggest empirical methods, using the syllogisms above, to illuminate what would happen under ideal conditions of impartial reasoning. After all, a key element of the theories is the presumed individual behaviors under specifiable conditions. One can test such presumptions. In other words, one can evaluate the empirical assertions of the theory.

The theories maintain that individuals would choose one or another principle. But, as shown, the arguments are not definitive. They would be compelling, as ethical arguments, only ifas the ideal conditions were empirically approximatedthe choices of individuals showed, on the average, a tendency to approximate those predicted in the ideal. The conception is similar to a notion of robustness regarding scientific theories. For example, a theory of motion may be defined in the context of an ideal construct, such as a vacuum. Because a perfect vacuum is unattainable, we test the theory under less than ideal conditions. Bridge concepts and supplementary theories about the impact of imperfections in conditions permit an evaluation of the theory in the real world.

As the ideal is approached, the predictions must increase in accuracy or the theory becomes uninteresting. Failure to exhibit

this sort of robustness would constitute a vulnerability or brittleness that would deprive the theory of a considerable degree of attractiveness.10 If one or another theoretical conclusion does not bear out as the conditions are approximated, the weight we ought to give it diminishes. This is so in science, and we believe it should be so with theories of distributive justice.

There are yet other advantages to using actual experiments. Conducted as thought experiments, the arguments of Harsanyi and Rawls invoke the considered judgment of only one observer from the perspective of his or her experience.11 As a real experiment with approximated conditions one can have a selection of participants representing a variety of perspectives and cultures. Testing of this sort also opens the possibility of a wide sampling of the subjective components (for example, relevant knowledge and preference) that are so difficult to identify precisely a priori. In addition, experiments force the subjects to focus on the issue of justice in a direct fashion. And how much better it is to have them contemplating a real (rather than an abstract) decision with concrete implications to concentrate their minds. As Jonathan Harrison notes in commenting on Roderick Firth's notion of an ideal observer: "We do not verify ethical statements by observing the reactions of ideal observers, for either there are none, or we do not have the opportunity

Findings in chaos theory imply that we might not expect uniform increased accuracy in the predictions of the idealized theory as the ideal conditions were approached. Wild gyrations might well occur. Nevertheless, a reasonably wide sampling of instances as the ideal was approached should still reveal some central tendencies. Purely random results or results totally at variance with the predictions of the theory are not to be expected. Were they found, the accuracy of the theory would truly be questionable.

We say only one observer, but Hare says two: the reader and the author. Return to the quote from Hare at the end of the previous section. We ended that quote with Hare's question "But what are the facts?" Hare continues, a few lines later, "Rawls, in short, is here advocating a kind of subjectivism, in the narrowest and most old-fashioned sense. He is making the answer to the question 'Am I right in what I say about moral questions?' depend on the answer to the question 'Do you the reader, and I agree in what we say?' This must be his view, if the considered judgements of author and reader are to occupy the place in his theory which is occupied in an empirical science by the facts of observation. Yet he claims objectivity for his principles" (1973, p. 145).

of observing them. Statements about the reactions of ideal observers would, presumably, be verified by observing the ideal reactions of actual observers" (1956, p. 259).

To those who would object to, or challenge, our conclusions, the following questions could serve as likely jumping off points:

1. Are C 1 , , C n the conditions of impartiality? 2. Are C 1 *, , C* n reasonable enough approximations of C 1 , , C n to serve as an environment for an initial test of the theory?

But objections to the enterprise on the basis that the ideal has not been attained cannot be accepted. After all, the notion of robustness, or approximation of the ideal, as a basis for gaining leverage has a long tradition in political philosophy. It goes back at least as far as Plato. In discussing the ideal state in his Republic , Plato (p. 178) notes that the purely ideal state is unattainable but also that this fact should not prevent one from attempting to approximate it as closely as possible:

Can theory ever be fully realized in practice? (He answers 'No'.) Then you must not insist upon my showing that this construction we have traced in thought could be reproduced in fact down to the last detail. You must admit that we shall have found a way to meet your demand for realization, if we can discover how a state might be constituted in the closest accordance with our description. Will that not content you? It would be enough for me.

Or, to put it differently, because an argument is an idealization does not mean that it is without empirical content. Indeed, one could argue that only insofar as (political) ethics has an empirical element is it appreciably differentiated from logic and mathematics. Thus our interest here is not in what some have called the geometry of morals but in the physics of morals.

Continues...

Excerpted from Choosing Justice: An Experimental Approach to Ethical Theory by Norman Frohlich and Joe A. Oppenheimer Copyright 1992 by Norman Frohlich and Joe A. Oppenheimer. Excerpted by permission.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.

Edition details

• Publisher: University of California Press
• Publication date: 1992
• Language: English
• ISBN 10: 0520072995
• ISBN 13: 9780520072992
• Binding: Hardcover
• Number of pages: 276