Andreas Teuber
Department of Philosophy
Brandeis University

OUR atttitudes toward risks vary, depending on what has happened to us, what we expect, how we feel, what we know, and what matters to us. We ignore some risks; overestimate others. Our perceptions are selective. They are influenced by the forms of social life we inhabit and they change as social life changes. Nothing illuminates this last point so well as the origin of our present understanding of life insurance, the earliest inklings of which can be found in the prospectuses of the first life insurance companies in the latter part of the eighteenth century in Europe.

In the 16th and 17th centuries, life insurance was conceived primarily as a wager. Short-term bets were taken out on the lives of a third person, usually a prominent figure like the Pope. The custom seems to have originated in 15th century Spain and spread successively to Italy, Holland, France, and England. The 1570 Code of the Low Countries lumps life insurance together with other "abuses, frauds, and crimes," and by the end of the 17th century it was illegal almost everywhere in Europe except Naples and London. Throughout the early modern period, life insurance was "betting on life," not planning for it.

Changes in the concept had to await the arrival of a new group of individuals, those of newly acquired wealth, who had something to lose were their fortunes to reverse themselves. An early prospectus of the Equitable Society for the Assurance of Lives, founded in London in 1762, extols the benefits of life insurance taken out on the head of the household, by painting a vivid portrait of the "indigent and deplorable situation" that could be brought about by his unexpected early death. Playing upon the anxieties of family members, it makes much of the "catastrophe" which would occur, were they suddenly "catapulted out of the middle ranks of society" - which by dint of their own "industry and thrift" they had struggled so hard to build up - "to the very bottom."1

The transformation in the conception of life insurance from "betting on life" to "security against its loss" illustrates dramatically that our attitudes towards risk are bound up with what else we believe and value. They can be seen to make sense only against a certain social background without which they would lose much of their point.

I. Risk and Consent

We live in a culture that prizes autonomy and it is against the background of our belief in its value that our responses to risk acquire their meaning and life. I do not want to explore the precise shape and contour of the moral importance that we attach to autonomy or to defend its special significance, only to point out that we respond differently to risks that we choose for ourselves and over which we believe we have some degree of control and to risks to which we are involuntarily exposed. In technical discussions of risk it is frequently assumed that risks are negativities and, in a calculation of costs and benefits, are consigned to the "costs" side of the ledger. To a mountaineer, however, the risk of a climb or a particular route up a rock face is seen as part of what makes that climb a challenge, even worthwhile. Mountaineers will rank different climbs in terms of their DP or "death potential." The risk is seen as partly constitutive of the climb, part of its good, and does not fall outside the means-ends continuum. In this sense, risk is not merely some cost to be offset by overall benefit.

Part of the reason for the mountaineer's attitude towards risks is tied, surely, to the degree of control he believes he has over the risks he must bear. Richard Solomon has reported that parachutists enjoy a high that lasts twenty-four hours after their jump.2 Hearing of these findings, Thomas Schelling commented: "I wonder whether someone thrown from an aeroplane, saved by a parachute that he didn't know he had, would be exhilarated by the experience," 3 - a remark that reflects his suspicion that sensations of control make a difference to our risk responses. Indeed the experiences of mountaineers and sky-divers show the extent to which not only perceptions but evaluations of risks are shaped by the degree to which the risk-bearer feels in command.

Our judgments, too, about when it is permissible to expose a person to risk or injury are tied to notions of individual responsibility and human agency. The source of an injury is often as important to us as its extent. It makes a difference to us whether a harm was due to natural causes or brought about by the action of another person. If it was caused by another person, we usually want to know if he knew what he was doing, and if he did, whether he acted deliberately and our preoccupations are reflected in our legal codes and moral practices. If, as Annette Baier has suggested, morality is nothing more or less than the culturally acquired art of selecting those harms that make us mad4, the key question to ask from a moral point of view is not "what should we do about risk?" but "which risks do we want to have on our conscience?"

In the following pages I shall focus on those risks imposed on us by others, on what might be called, "risk-impositions." Which risk-impositions are morally acceptable and which are not? Philosophers have paid surprisingly little attention to the permissibility of risk-impositions, having devoted almost exclusive attention to the conditions under which it may or may not be permissible to cause another person harm.5 If, as we seem to believe, it is wrong to cause another person harm without that person's consent, is it wrong to impose a risk of harm without consent?6

In light of our commitment to autonomy, it is perhaps not unreasonable to think that consent might play a role in the legitimation of risky activity. It plays an important role in the political process which relies on the participation of the governed for its sense of legitimacy and in the marketplace where the ideal transaction is represented as one of voluntary exchange. Consent too is thought to be necessary in cases of medical experimentation. Detailed procedures have been developed for the use of consent forms and for obtaining a patient's "explicit, informed consent."

Given that our intuitions are fairly well grounded about the conditions under which we think it is permissible to expose patients to the risks of experimental drugs or surgery, it is tempting to take such cases as a model for the justification of risky activity in general. This, in turn, suggests a minimal first principle:

Imposing risks on people is justified if, and only if,
it is reasonable to assume that they have consented to those risks.7

II. Problems of Consent

But as workable as such a principle appears to be in the context of doctor and patient, can it be extended to less clearly circumscribed and more problematic risk-impositions: exposure to pesticide drift, the venting of radioactivity from a nuclear power plant, carbon monoxide fumes from congested traffic, acid rain from coal-burning power plants located many miles away, or chemicals in the water supply from an abandoned toxic waste site? In nearly all of the more problematic cases the risks to which people are exposed are not distributed singly from one individual to another. Risks are distributed in small increments over a very large population or as the result of the actions of a great many individuals, each of whose action, taken alone, is likely to cause little harm but when combined pose a very grave threat indeed. The business of identifying and distributing consent forms to all those who might possibly be at risk is surely an insuperable task, to say nothing of the difficulties of measuring the small magnitudes of risk to each person who may be affected. To complicate matters still further, many risks have multiple sources and may inter-relate in synergistic fashion so that it is difficult, if not impossible, to figure out who is doing what to whom, let alone draw up contracts among the relevant parties.

Even if the practical difficulties of obtaining people's consent could be overcome, it is widely reported that people are notoriously poor judges of risks. People's perceptions frequently fail to match up with the actual dangers risks pose and few people have a "feel" for what a chance of dying, say a chance of one in a million, really means. Research by psychologists Amos Tversky and Daniel Kahneman has shown that we are regularly led astray in our assessments of probabilities by rules of thumb. Faced with a judgment that requires even a minimal familiarity with statistics, we frequently avoid the statistical information and rely instead on a description or heuristic which feels less strange. 8 We tend to overemphasize low probabilities and underestimate large ones. We have to struggle to resist the gambler's fallacy: the belief that after a series of losses the odds must favor a win. We are also poor judges of outcomes. We appear to be more concerned to avoid a loss than to receive an equivalent gain, and this asymmetry can be exploited in the way choices are presented.9 Retailers, for example, know enough about our suceptibility to the way options are framed to represent a surcharge for credit card customers as a discount to those who are willing to pay cash.10

The influence of framing on judgments about risk is systematic and pervasive, and shows up at all levels of education. Health care professionals are no less susceptible to the effects of framing than their patients who have less experience and lack their expertise. The following hypothetical case was put to a group of physicians:

Imagine that you have operable lung cancer and must choose
between two treatments: surgery and radiation therapy.
Of 100 people having surgery, 10 die during the operation,
32 are dead after one year, and 66 after five years.
Of 100 people having radiation therapy, none die during treatment,
23 are deadafter one year, and 78 after five years. Which treatment do you prefer?11

Given these options, fifty percent of the physicians said they preferred radiation treatment. However when the same options were presented in terms of survival rates rather than mortality rates, 84% said they would prefer surgery.

It is perhaps not completely surprising to learn that people are poor judges of probabilities, but "we want to give [people] credit for at least knowing their own minds," as one report puts it, "when it comes to assigning values to the outcomes of their choices."12 Apparently, very little credit is due, as experiment after experiment reveals:

Imagine that the United States is preparing for the outbreak
of an unusual flu epidemic which is expected to kill 600 people,
unless action is taken. Two alternative programs to combat the disease are proposed
If program A is adopted, 200 people will be saved. If program B is adopted,
there is a 1/3 probability that 600 will be saved and a 2/3 probability that no one
will be saved When the alternatives were posed in these terms in a test survey,
72 percent of the respondents opted for program A, only 28 percent for program B.
A second group was given the same options, but re-described (re-framed) in this way:
If program A is adopted, 400 people will die; if program B is adopted,
there is a 1/3 probability that nobody will die, and a 2/3 probability that 600 people will die
This time only 22 percent opted for the first program, while 78 percent opted for the second.13

It is generally believed that consistency in judgments is a minimal condition of rationality. Since our judgments about risk are apparently inconsistent, it is hard not to draw the conclusion that our attitudes towards risk are also irrational. These findings have disturbing implications for public policy, especially in a society like our own which relies on a democratic process. If we are irrational in our judgments about risk, the policies we enact will reflect a similar bias. Given our untrustworthy attitudes, a consent-based approach to legitimating risk-imposing activities can only lead to irrational public policies.

III. Cost-Benefit Analysis as a Substitute for Consent

But if we cannot rely on a consent-based approach because it is unworkable or not to be trusted, where are we to turn to find a way to legitimate decisions that expose us to risks to which we have not been able to give our actual consent? How should such decisions be made?

Some believe that the way out of this apparent cul-de-sac is for us to rely on cost-benefit analysis to tell us what to do in our public policies involving risks. But cost-benefit analysis is regarded with no small amount of suspicion by the general public. No matter how efficient such analytic methods may be, they are believed to require the technical expertise of well-trained professionals for their operation. Therefore, policies which emerge from their application are likely to seem less legitimate than policies that evolve, however clumsily, from a more participatory political process. Indeed, cost-benefit analysis appears to be as far removed from a consent-based approach as one could find.

A powerful case, however, in favor of cost-benefit analysis, put forward recently by policy analysts Herman Leonard and Richard Zeckhauser, suggests that we may have less to fear from such methods than we imagine 14 They admit without hesitation the moral appeal of consent-based solutions and go on to argue that - contrary to popular opinion - cost-benefit analysis can be grounded in a form of consent after all. Assumptions about its narrowly technocratic character are misplaced.

Their argument, in brief, runs as follows. Cost-benefit analysis incorporates into its calculus the criterion of economic efficiency as well as an assumption of rationality. Since people, by and large, want to act rationally and prefer efficiency to inefficiency, they would consent to public policies concerning risks to health and safety that were based on cost-benefit analysis, because the outcomes generated by the analysis are the same as the outcomes that they would choose, if they could go through the process themselves. Leonard and Zeckhauser:

If...benefits and costs were fully and appropriately measured
and equally distributed over individuals,we would expect
that the benefit-cost choice criterion would yield the same result
as universal individual consent. In this polar case, consent
and the benefit-cost criterion are equilvalent, and in this sense cost-benefit analysis
can be thought of as a form of "hypothetical" consent by the community.15

Reasonable people should accept the results of cost-benefit analysis applied to risks because the method measures and compares all relevant factors, incorporating all our values and assigning them weights that we would give to them were we making the decision ourselves, or more exactly, were we making the decisions on our rationally best behavior. In this sense risk cost-benefit analysis is our most reasonable self writ large. It is not an impersonal, narrowly technocratic method for calculating outcomes. It identifies choices from a range of alternatives that an idealized rational actor would choose. Reasonable people ought to rely on cost-beneft analysis to identify solutions to complex risk decisions because that is the rational thing to do. What we lose by a lack of opportunity for actual consent we gain in the concept of rationality. A government agency that relies on cost-benefit analysis to arrive at a public policy decision about risk ought not to be seen as imposing its will on an unsuspecting populace since all that cost-benefit analysis does, we would do.

Cost-benefit analysis applied to risks fills a vacuum created by the unavailability of market solutions and the impracticality of obtaining the explicit, informed consent of all parties to a risk decision. And not only does it fill a void, it fills it, according to Leonard and Zeckhauser, in an ethically appropriate way, since the ethical underpinnings for the defense of cost-benefit analysis ultimately rest, if not on actual consent, on a form of consent nevertheless, albeit hypothetical. What are we to make of this? Do we have in cost-benefit analysis a solution to our problem?

a. Commensurability of Risks

For cost-benefit analysis to do its work, it is essential that a number of widely divergent alternatives be comparable. Part of the strength of the technique is its claim to be able to order and rank a variety of options. To perform its task, it is necessary to find a measure that is sufficiently abstract and general so that the costs and benefits of various alternatives can be lumped together and added up. However the homogenization that takes place often results in the glossing over of differences and the production of outcomes that offend common sense.

Of course common sense itself is not always a reliable guide. Indeed the failures of common sense in making judgments about risks are part of the rationale for wanting something like risk cost-benefit analysis in the first place. But the measure which cost-benefit analysis adopts to represent all factors on a single scale must yield comparisons that are in some way recognizable to us and not so counter-intuitive as to appear arbitrary. Some such requirement must be taken seriously especially in light of a defense of cost-benefit analysis on grounds of hypothetical consent, on the grounds of its being able to produce results to which people would consent. As Leonard and Zeckhauser themselves make clear, for their defense of cost-benefit analysis to work, benefits and costs must be "appropriately" measured. Part of what will determine that appropriateness will be the intuitive appeal of the comparisons made under the technique, in their connecting up, however dimly, with common sense. But this feat is much less easy to accomplish than might be thought. Imagine that we are faced with having to compare the expected health risks of the following two options:

Option #1: the death of two children
with an average future life expectancy of 70 years.

Option #2: the incapacitation (say by a serious bout of influenza)
of 16,000 people for an average period of 4 days.

To compare the costs of the two options, assume that a common measure of "work-days lost" is chosen, giving the following results:

For Option #1 the total work-day loss will stand at
2 X 7 X 365 or 51,000 work-days lost.

For Option #2 the total work-day loss
will be 16,000 X 4 or 64,000 work-days lost.16

On the basis of these calculations we should prefer Option #1. This defies common sense: four days of discomfort even for a great number of workers, seems preferable to the deaths of the children. Perhaps the wrong measure of comparison was chosen, and another measure other than work-days lost would yield a more intuitive result. Perhaps. But for a comparison to yield the right intuitive "feel" the common measure we choose will more than likely have to take some of the particularities of the cases into account. Although we may be able to come up with a common measure in a specific case that will yield intuitively acceptable results, this works against the grain of the analysis as a useful tool at the level of public policy where there is incentive to come up with more and more general measures in order to compare the greatest possible variety of options. It is hard to avoid the conclusion that no matter what scale of measurement is chosen, any general measure of comparison will inevitably produce intuitively unacceptable results.The technique also does not pay sufficient attention to different human responses to risk, treating them instead as if they all belonged on the same level, although perhaps varying in intensity. Individual responses to risks that are treated as numerically equivalent sometimes turn out to be significantly different despite their numerical equilvalence. Too often these differences are allowed to disappear in some general formulation or get written off, if they cannot be made to fit, as simply irrational.

b. The Distribution of Risks

For cost-benefit analysis to be of use in the formation of public policy some way of assigning a value to the risks themselves is necessary and so some sensitivity to the ways in which risks are distributed throughout a population is in order. But there are a number of different ways to identify the magnitude of risk to a population. So, for example, we can estimate the chance of dying from a nuclear accident either as the likelihood of the number of deaths throughout a given population or we can spread that risk of death equally over individual members of the same population and estimate the average risk per individual.

Imagine now that we are faced with estimating the risks of radiation-related deaths from a nuclear accident of the same order of magnitude that occurred at Chernobyl to the population surrounding a nuclear power plant in New England. To those within the plant and in the area immediately surrounding it there is a danger of losing their lives instantly, whereas there is a considerably lower risk of delayed cancer deaths to those living further away. If we translate the expected number of deaths likely to occur throughout New England from such an accident into average individual risk, we seriously underestimate the risks to those who reside in the immediate vicinity of the plant. If we choose to estimate the average individual risk to members of the population within a two-mile radius of the plant, we ignore the risks to the larger population. Which population we choose as the population at risk is not merely a matter of convenience. If we know, for instance, that the chances of dying from a nuclear accident are close to 100% for those working within the nuclear plant; whereas those who live furthest away face no risk at all, it is misleading to spread this risk over the entire population as an average risk per individual. Since cost-benefit analysis will coincide with universal individual consent only when all members of a community are affected equally under each alternative entertained by the analysis, there is a strong temptation to ignore distributional considerations completely in favor of an average risk per individual spread over a population as large as possible. As Douglas MacLean has suggested "we might speculate that some intuitive sensitivity to distributional issues helps to explain why most hazard comparisons have so little impact on public opinion and concern."17

c. The Value of a Human Life

One possible explanation for the counter-intuitive result in the earlier comparison of health risks in Options # 1 and #2 above is that many of us feel that human lives ought to be treated differently from other costs. A human life has a special value in our culture and ought to be treated separately from other things we value, even severe but non-fatal injuries. To some policy analysts it has, therefore, seemed useful to try to come up with a single, determinate value for a human life.

There is little agreement, however, among policy analysts over how a figure for the value of a human life ought to be set. Since there is no market in human lives, inferences are frequently drawn from situations in which people reveal what they are willing to pay for their own safety. But the conclusions of these indirect tests vary so widely that the most determined policy analyst is likely to despair of ever being able to put a single, determinate value on a human life. An investigation into consumer purchases of smoke detectors, for example, estimates the value of a life at $175,000;18 whereas a study that looked at the decisions by drivers to wear seat belts puts the figure closer to $593,000.19 Yet another study analyzing wage differentials received by workers on dangerous jobs gives the estimate of $3,200,000.20

Whatever tendencies may underlie consumer preferences or might serve to explain the different levels of consumer concern for their own safety, the variation in estimates drawn from such indirect tests puts the policy analyst who is eager to establish a single, determinate monetary value for a human life in an awkward position. In the end he may simply opt for a figure which is exactly what Dan Usher does in a recent article on "The Value of Life for Decision-Making in the Public Sector," where he recommends an amount of $750,000 "increased each year in accordance with the rate of inflation" and provided that "discretion [is] exercised when the victim of a calamity can be identified and in decisions involving the old and the infirm."21

Even more troubling than the difficulty of obtaining a value for human life, however, is the assumption by policy analysts that their calculations bear some, let alone any, relation to what we - those of us living (and dying) within the culture - might understand that value to be. Commenting on the current recipes for coming up with a monetary value for a human life, the economist E. J. Mishan remarked that they make him think what it would be like "to calculate the value of a two-week honeymoon for a loving couple by reference to their anticipated outlays (including foregone earnings) plus perhaps an allowance for the probability of non-consummation owing to the frigidity in either."22

Putting a price on life is sometimes seen as an example of the large scale computations that have become so much a part of modern politics which "bring with them," as the philosopher Stuart Hampshire has written, "a coarseness and grossness of moral feeling , a blunting of sensibility, and a suppression of individual discrimination and gentleness."23 But to the charge that putting a price on life shows "a coarseness of moral feeling," the policy analyst can reply that governments cannot avoid the trading of lives for money since they are responsible for programs in public transportation, national defense, environmental protection where lives are saved or lost at a price. Since governments cannot avoid trading lives for money, they might as well set the terms (the value of a human life) under which the trading takes place. Indeed we need not get all worked up about putting a price on life because the lives for which we seek to find a value are statistical lives. As Thomas Schelling has explained, "what the government buys if it buys health and safety is a reduction in individual risks. The lives saved are usually a mathematical construct, a statistical equivalent to what, at the level of the individual, is expressible as a longevity estimate but not a finite extension of life."24 By deciding to adopt this or that public program, by deciding, say, to raise or lower auto-emission standards, a decision is made about the number of lives that are likely to be saved or lost, but no decision is made, so the claim runs, about the life or death of identifiable individuals. Again as Schelling has put the matter, life and death decisions for specific individuals do indeed involve "anxiety and sentiment, guilt and awe, responsibility and religion . . . But most of this awesomeness disappears when we deal with statistical deaths, with small increments in a mortality rate in a large population."25

And to the charge that putting a price on life allows government agencies and their officials to "play God," the policy analyst has this reply:

The specification of the value of life does not give powers over life and death
to the administrators of public sector projects. It takes such powers away.
Having placed the responsibility for road-building and medical care in the public sector,
we have, knowingly or not, compelled ourselves to decide between your life or mine.
We can allow administrators to act as they see fit. . . Or we can bind these decisions
with rules that make those decisions less arbitrary, less subject to political influence,
and less tragic. Specification of a uniform value of life constrains the choices of administrators
of public sector programs and avoids many situations where one man must give
or withhold the life of another.26

d. Risky Situations of Relative Urgency

Still many are made nervous by the thought of a "price" on life. It is difficult to get rid of the impression that "putting a price" on life somehow misses the more fundamental point that life is "beyond price." Faced with the imminent death of persons whose actual (non-statistical) lives are in danger, we respond as if we believed that life was sacred and not just another commodity. We have an extraordinary determination to engage in the rescue of individuals whose lives we know to be in danger, a determination that is apparent in efforts to save a small child caught in a well, miners trapped in some passageway far beneath the surface of the earth, or passengers lost at sea. In these cases people are willing to go to almost any lengths, without much concern for "cost," to save the lives of those at risk.

Imagine that we are faced with the business of determining policy in the long term for, let us say, mining safety. We might very possibly conclude that "in some cases, for the price of a rescue, we could save a much larger expected number of lives through various safety measures."27 From the perspective of cost-benefit analysis, "fewest lives would be lost if the entire budget is spent on prevention of accidents and on easy rescues, and no heroic or extraordinary rescues are attempted in case of disaster."28 But now imagine that we adopt such a policy. Any attempt at heroic rescue of miners will divert resources away from the installation of safety equipment that will save more lives in the long run. But what would it be like to stand by while several miners are lost who might otherwise have been saved but at a cost that has been calculated to be not worth the price?

Here the prescriptions of cost-benefit analysis diverge from what we urgently feel is the right thing to do. "Put a price" on life and include that "price" along with other costs and benefits in a total risk cost-benefit package and surely a result will emerge - even if the price on human life is set very high - that on some occasions at least the rescue of a person stranded at sea or buried in a mine will be "so expensive as not to be worth its cost."29Yet we appear to be willing to accept a considerable amount of inefficiency or cost-ineffectiveness in pursuit of what we believe is more important: saving the lives of those who are immediately at risk. Why not then include in a cost-benefit analysis, not only lives saved and lost, but also some measure of the psychological effects and mental disturbances that persons are likely to experience as a result of being barred from attempts at heroic rescue? There are at least three objections to be made to this proposal.

First, the significance to oneself of being the kind of person who cares enough in situations of extreme urgency is not the sort of thing that can be captured by the narrowly economic values employed by cost-benefit analysis. To do so the method would have to expand its conception of value. Second, being the kind of a person who comes to the rescue of someone in distress tends to be experienced as part of what makes life itself worth living, as valuable for its own sake, apart from what good it may produce. Whatever value assigned to it by cost-benefit analysis, there will always be something more of value, i.e., the intrinsic worth of being this kind of person, that the analysis fails to capture. Finally, when we respond to others in situations of distress, we do so without too much thought about how else we might spend our time. We respond sympathetically and spontaneously. Stepping back to make these decisions by cost-benefit analysis may endanger these more or less spontaneous acts of human fellowship, "may profane things that must be held sacred," as Alan Gibbard has suggested, "things that, because of who we are, we must hold sacred if our lives are to be deeply rewarding."30

No doubt there are some policy analysts who may simply wish to dismiss our feelings of urgency in cases of heroic rescue as another example of our unsophisticated judgments about risks. On this view we should neither be surprised nor overly concerned that the analysis points one way and our feelings another: "there is nothing about risk cost-benefit analysis, even in an idealized version, that guarantees the humane result."31 We should expect it to deviate "from moral judgments in which we place great confidence,"32and in those situations where it does, we should be open to reforming or replacing our moral judgments, which are in any event hopelessly confused, with the prescriptions of the analysis.

But anyone who wishes to defend cost-benefit analysis as a reliable guide to policy concerning risks with an argument based on the idea of hypothetical consent cannot simply dismiss cases of heroic rescue as one more example of the failings of common sense. Indeed, Leonard and Zeckhauser look for ways to accommodate our feelings of urgency, acknowledging that "some social values will never fit in a cost-benefit framework and will have to be treated as "additional considerations" in coming to a final decision."33 But since they offer no principled way of how to bring in our feelings of urgency, their solution is inadequately ad hoc. If some values cannot be handled by the cost-benfit criterion itself but are still to be given some weight as "additional considerations," the method begins to look less and less like a neutral procedure that can be applied more or less automatically. Leonard and Zeckhauser seek to console potential critics who worry that cost-benefit analysis is liable to be open to abuse with the thought that "its ultimate grounding in analytic disciplines affords some protection."34 But to the extent that some social values have to be taken into account on an ad hoc basis, such grounding affords that much less protection.

e. Risks and Rights

Individual rights make up another area with which cost-benefit analysis is ill-equipped to deal. There are, of course, many decisions we make where rights play no role at all. But in the arena of risks to life, health and safety this is not so: rights play a very critical role indeed.

Rights constitute what might be called a minimal respect for persons and constrain persons from treating others in certain ways. Rights do not allow trade-offs. They cannot be brought together into one single homogeneous total. Each person's rights count separately and have their own independent moral value. My rights are, therefore, incommensurable with your rights; each placing limits on the actions of the other.35

Imagine that we are faced with the task of developing a national policy concerning secondhand cigarette smoke in enclosed public spaces. If we think of the effects on smoker and non-smoker alike, there are a clear set of costs and benefits on both sides. There is the irritation from the tobacco smoke felt by the non-smoker as well as the discomfort and frustration experienced by the smoker who is asked to refrain from smoking. If we assume that a link between lung cancer and secondhand smoke has been established, there will be costs associated with the longer term health effects to both the smoker and non-smoker.36 In an enclosed space, such as the cabin of a commercial airline, it may not be easy for a non-smoker to escape the effects of secondhand smoke. Recently smoking has been banned by several commercial airlines on short-term flights. It is possible to imagine that some calculation of costs and benefits went into these decisions: the time and savings from no longer having to maintain separate sections for smoker and non-smoker is set against the minimal discomfort the habitual smoker will have to endure on a short flight. For transcontinental flights it may be believed that the costs of discomfort to smokers are too great to be outweighed by other benefits.

But now we live in a culture that takes rights seriously, in particular rights to persons and property. If a person is in an enclosed space from which she cannot easily walk away, the tobacco smoke she experiences from someone else's cigarette is not only a cost, but a violation of her right not to be harmed without her consent. As Peter Railton has put it: "each time I send some of my tobacco smoke your way, you suffer some small physical change, not for the better."37 On a rights-based view, each person's rights set up a boundary across which others are not permitted to go. And although the change that a person suffers from the tobacco smoke imposed upon him may be very small, on a rights-based view what matters is "whether or not a boundary was crossed, not the magnitude of the consequences: shoplifters have been prosecuted for stealing a pack of chewing gum."38 Of course, smokers can claim that they have rights too and they do, only each person's rights are brought up short when they bump up against the boundaries of somebody else's rights. Smokers are free to smoke so long as they respect the boundaries of others. They are free to smoke in the confines of their own room. In an enclosed public space, the smoker's right to smoke stops, like the right to swing one's arm, at the tip of somebody else's nose.39

A public policy based on rights will mean either outright prohibition of smoking in enclosed public spaces or the provision of segregated smoker and non-smoker sections where this is feasible. The provision of smoker sections would appear to balance rights against the discomfort to non-smokers of being asked not to smoke and has the look of a cost-benefit solution. But it is important to recognize that the rights of non-smokers set the bounds within which different strategies of how to implement a policy that observes those rights might be entertained. Rights are not "additional considerations" to be taken into account in some ad hoc manner as a supplement to maximizing total benefit. Rights enter into the making of policy at the foundational level, as it were, and other considerations such as the comfort of smokers can only be entertained after the rights of non-smokers have been taken into account.

Workplace air pollution presents similar problems to the risks of secondhand smoke only here there are obvious benefits to be gained from the polluting activity. There is great incentive, no doubt, to keep production costs as low as possible. Nonetheless there appears to be general agreement that "workplace air pollution violates workers' rights."40 Current public policy reflected in the Occupational Safety and Health Act is rights-based: permissible exposure levels for any toxic substance must ensure that "no employee will suffer material impairment of health or functional capacity."41

However if we extend a rights-based solution into other areas, such as outdoor air pollution, most risk-imposing activities would have to be prohibited. The extremity of rights-based public policies concerning risks to safety and health have inclined some policy analysts to disregard rights considerations altogether. Since cost-benefit analysis is unsuited to handle these sorts of considerations anyway, such an inclination may be more self-serving than considered. Nonetheless it is true: rights leave little room for balancing or compromise and perhaps in the arena of outdoor air pollution we arrive at the limits of their usefulness in setting policy concerning risk. But if rights place an unusually strict limit on the levels of risk to which citizens may justifiably be exposed, it is also true that risks to which people are exposed are permissible if it can be shown that the bearers of those risks have given their consent. Only consent can turn a potential violation of a right into a permissible act: "Consent, as it were, opens up boundaries so that a border-crossing is no longer a moral offense."42 For this reason it makes sense to reconsider the role actual consent might play in the legitimation of risky activity.43

IV. Bringing Consent Back In

Because of the moral appeal of the notion of consent, proponents of cost-benefit analysis have tried to seek a justification for the technique in a form of consent, albeit hypothetical. On this interpretation it is legitimate to impose risks on people without their actual consent so long as the benefits are sufficiently great, so long as the benefits of so doing outweigh the benefits of alternative courses of action. Risk-impositions proposed by the analysis are justifed because reasonable people prefer alternatives where benefits exceed costs and so would choose the same outcomes as the analysis recommends if they could go through the process themselves. But this defense of the technique only works if the costs and benefits that are the input of the analysis are appropriate and comprehensive. Since they are neither, a defense of cost-benefit analysis as a form of consent does not succeed.

Cost-benefit analysis cannot yield the same result as indivdual consent because it fails to respect fully the distinctiveness of people's responses to risks, or to do justice to the morally significant ways risks can be distributed, or to give proper weight to the importance we attach to human life in situations of felt urgency. It fails to capture things of great value to us, such as our concern for autonomy and rights, or captures them in a way that does not reflect the special significance they have for us. For these reasons we should not be persuaded to allow cost-benefit analysis to determine public policy, to do, as it were, our talking for us. What we need are not improved analytic techniques or more finely tuned ways of calculating costs and benefits, but greater opportunities for participation by those affected by public policies concerning risks in the decision-making process.

That process is, of course, a political process and some policy analysts despair of our ever coming to any sort of decision, let alone reasonable ones, by that means. "An unfettered political process," Leonard and Zeckhauser fear, "provides no guarantee that it will seek efficiency or pay attention to the general welfare. It also . . . may not take advantage of the best available information."44 Cost-benefit analysis may not be "perfect," they admit, but even so "we should be able to anticipate better outcomes than we would achieve by 'muddling through.'"45

But "muddling through" may be the only just way to make public choices involving risk because these decisions raise questions of equity to which cost-benefit analysis is blind and about which it is has nothing to tell us. The issues of equity raised by decisions which impose risks on some to the benefit of others, can only be settled in a dialogue among those who are likely to be affected or through their representatives, and perhaps not even settled then. Leonard and Zeckhauser point out that the political process sometimes leads to stalemate,46 and "that's true, too,"47 but we should not then turn to a method that is not equipped to make such decisions in the first place; we should look for ways to unblock the impasse, for better ways to talk with each other, to negotiate and find a compromise.

Our responses to risks presuppose divergent interests and ideals that cost-benefit analysis neither accomodates nor grasps. These divergent interests and ideals are best dealt with by a pluralist approach, not merged into one homogeneous total as they are under cost-benefit analysis.48 A democratic political process works in an essentially pluralist manner and is designed to handle, however messily, the divergent interests and ideals of different people. It is also important to recognize that we do not only want a method that will help us to "anticipate better outcomes," we want a method that achieves its outcomes in appropriate and fair ways.  "It would not do, for example," as Douglas MacLean has observed, "to replace trial by jury or democratic elections with technocratic analytic decision-making techniques, even if these techniques...would achieve better results."49 And it will not do because the legitimacy of the outcome in these cases depends in part on the openness and fairness of the process.

But if people are to become more involved in making public policy decisions concerning risks to life, health and safety, how are we to respond to the psychological studies which show that we are such poor judges of probabilities and that our judgments about risks are frequently "irrational, if not perverse?"50 Two responses at least are worth stressing:

One is that we should make every effort to educate citizens to be better at probabilistic reasoning, and to supply them with relevant information in terms they can understand. If statistical information is required, we should make it more understandable by removing as much of the jargon as possible. We might have reason to hope that over time people will become better at making probabilistic decisions, but given the tenacity of people's inability to assess probabilities, on the job training is not likely to do the trick. Our susceptibility to framing effects might be partially overcome by being careful to frame options in a number of different ways, although a good many framing effects might be given alternative explanations (see below). It may even be wise to hire those very researchers, who study the systematic biases to which we are prone, to serve as counselors in the decision-making process and to share their knowledge to help citizens or their representatives resist some of the more obvious distortions to which they may be inclined but which they may otherwise fail to notice.51

A second possible response is to point out that the findings of the risk perception studies only succeed in showing that people are irrational in their attitudes towards risk by assuming a fairly narrow conception of rationality. Some of our presumed irrationalities and inconsistencies can be explained in other ways, such that on a broader view of the findings our actual choices turn out to be quite reasonable after all. For example, the studies show that "loss aversion" is quite pervasive, but is it irrational? "If your acquaintances, your boss, or your constituents," Douglas MacLean has suggested, "are more likely to hold you responsible for the losses that result from your risky decisions than for the lost opportunities that result from choosing to avoid risks, then choosing to avoid a loss rather than maximize the expected outcome would seem to be an eminently prudent strategy."52

Specific to the conception of rationality presupposed by the psychological studies is the neglect of autonomy. The control people believe they have over risks has sometimes been offered to explain the apparently irrational preference many people show for driving rather than flying. But people are not only concerned about the chance of dying, but about "what life is like while they are living it."53 We may indeed underestimate our chances of survival in those situations where we believe we have some degree of control over the risks we bear, but we also can be seen to value that control. As most of us know too well: "Trade-offs between quality and quantity of life are made all the time in personal decisions about health and safety: Mark Twain told that he could add five years to his life by giving up smoking and drinking, reportedly quipped that five years without smoking and drinking weren't worth living."54

Re-describing an option can lead people to reverse their preferences as the research of Tversky and Kahneman shows. But a number of these preference reversals are also open to alternative explanations, suggesting that there may be more to our ways of thinking and reasoning than is to be found in a conception of instrumental rationality:

A suburban homeowner, let us call him George, is willing to pay his neighbor's son $8 (but no more) to mow his lawn, but refuses to mow his neighbor's (same-sized) lawn for $20. On a narrow economic view of rationality George is not behaving rationally, since he should treat foregone income on a par with out-of-pocket expenses. But now imagine that we re-describe the latter option: George's neighbor will contribute $20 to charity if George will mow his lawn. If George elects to mow his neighbor's lawn, we have induced a preference reversal.55

But for George's decision to mow or not to mow to reveal an inconsistency in his preferences, it is necessary to assume that "mowing his neighbor's lawn for charity," re-describes essentially the same option as the one he was given initially.56 Indeed all cases of preference reversal require the assumption that re-framing does not alter the substantive content of the option. But this is a considerable assumption. Is "mowing for charity" substantively equivalent to "mowing for money?" If George elects to mow his neighbor's lawn under this new description, has he acted irrationally or does his decision reflect a different approach altogether?57 Perhaps, George does not see himself as the sort of person who would do that sort of thing: "he simply does not think of himself as the kind of person who mows other people's lawns for money."58 To do so conflicts with his self-image.

We are not only ends-oriented, but ideal-oriented. We do not just care about where we end up, we care about the kind of people we have to become in order to end up one place or another."Mowing someone else's lawn for charity" is not the same as "mowing someone else's lawn for money." It is a different sort of activity entirely.

Something along these lines helps to explain the apparent irrationality of acts of heroic rescue. When the life of an identifiable individual is in imminent danger and there is some chance of saving that person, however remote, to stand idly by disturbs our sense of self. There is not only value in lives saved; there is value in being the kind of person who will go to heroic lengths to rescue others. To stand idly by may make it difficult for a person to live with himself afterwards, not only because he must live with a death, but because he must live with the sort of person he has become by his failure to rescue. To stand by is not only costly, it is dehumanizing.59

V. The Role and Limits of Consent in Justifying Risk Decisions

Although I am convinced that a consent-based approach is the only fair and appropriate way to justify risky activity, the approach is not without its problems and a number of points stand in need of closer attention and fuller discussion than I have room for here. I would, however, like to mention in closing a few points that will, I believe, require greater clarification, if a consent-based approach is to work.

First, it should perhaps by now be obvious that our minimal first principle cannot hold in the form it was originally stated. If we require actual consent to each and every risk of harm that is imposed, much of what we do in the course of our daily lives will have to be ruled out, since it will be impossible to obtain that consent. Some have argued that many of the risks to which we are exposed are too trivial or at least are not sufficiently important to require consent. Thus, it has been claimed, for instance that "you are not morally required to obtain the permission of everyone who might be in the path of your car should it go out of control on your way to the post office....Nor must you ask the consent of your neighbor before you...make a fire in the wood stove, although the risks to them of fire is increased somewhat."60 But we shall need some way of distinguishing the trivial cases that do not require consent from the nontrivial ones. What's to distinguish, for instance, the (supposedly) trivial case of driving to the post office from the following example: "Jones has a bad case of the flu. When she breathes, the flu virus gets circulated through the ventilation system of the apartment building where she lives. Among the other inhabitants of the building are a number of elderly people in rather [poor] health. By staying in her apartment and breathing, Jones imposes a nontrivial risk of death on them, for the mortality rate of elderly people in [poor] health who contact this kind of flu is, we may suppose, far from negligible."61 The difficulty, of course, is not just that it is not easy to come up with a set of criteria that will pick out the right cases, but that it is far from clear that some of the so called "ordinary" cases are all that trivial.

Second, once we eliminate the trivial cases, if indeed there are such cases, it remains obvious that we shall not be able to obtain the actual consent of persons in most cases of risk-imposition. Many risks, such as the risks imposed on people from the operation of coal-burning and nuclear power plants, involve externalities and public goods. In these cases risk-imposing decisions will be centralized and public policy will partly determine the levels of risk to which people are exposed by setting standards for, say, the venting of radioactivity by nuclear power plants or the release of toxins into the atmosphere by chemical factories. The model of doctor-patient, where a doctor seeks the prior consent of a patient before, say, imposing risks of operating on his left lung, cannot serve as a paradigm for most risk decisions. If cost-benefit analysis is not the answer, we shall need to be clearer about the ways and means people can participate in public decisions that impose risks on them.

For consent to play a legitimating role in the making of these risk-decisions, it clearly cannot come into play only in the final stages of the decision-making process. Potential risk-bearers will have to be involved at every stage of the process, in formulating, implementing, and adopting public policy. This, in turn, will surely require that present opportunities for participation in decision-making be enlarged, and in some situations imaginatively created. Along these lines in the area of workplace pollution, Mark McCarthy has urged that "workers need more than the exit vote of leaving a job with intolerable risks,"62 that they need to be more involved in the decisions by which those risks are controlled. He has suggested not only the creation of labor-management health and safety committees to help to enforce federal standards and to address unregulated workplace hazards but the incorporation of health and safety concerns in the collective bargaining process. He also makes the recommendation that there be greater participation by the affected parties, i.e., representatives of the industries and workers involved, in the drafting of regulations at the federal level.63

Third, in the course of improving opportunities for greater participation, we also need to become clearer about what forms of participation make for morally acceptable decisions. Given the moral pedigree of consent, some have been tempted to attribute its legitimating power to consent in all its many different forms and varieties. Thus, for example, some believe that merely by consenting to work in a factory where one knows in advance that there are airborne hazards, a worker thereby gives his consent to the risks he will bear, or that by purchasing a home near a nuclear power plant, one consents to the risks to which one is exposed in the likelihood of a nuclear accident. No doubt for this sort of tacit consent to be meaningful it must be true that one has had an opportunity to choose among a set of alternatives that are viable and risk-free and that one was really fully informed about potential risks. But even if these conditions are met, it is unlikely that consenting to work at a factory where one knows there are risks constitutes consent to those risks. Judith Thomson has argued,64 persuasively I believe, that in consenting to walk through a park late at night that one knows to be unsafe, one does not thereby consent to being mugged, if that should turn out to happen. In thinking of ways to bring consent back into the process of justifying risky activities, we need to be clear about what counts and what does not count as "consenting to a risk."65

Finally, it is also necessary to acknowledge that in looking for ways for potential risk-bearers to participate more fully in the decision-making process, we may be calling on skills and habits of minds with which citizens have little experience or experience of the wrong kind. Needless to say, our present political institutions and regulatory practices seem to encourage citizens to adopt an adversarial posture. In negotiations interested parties often see themselves in a trial-like process, inclined to assert their positions in the most extreme ways; they may be neither capable or open to arriving at a consensus on public policy concerning risk. No doubt there is some truth to the charge that we live in an adversary culture.66

Insofar as it makes sense to speak this way about us at all, a vision of citizenship as largely a defensive strategy and chiefly, or first and foremost, a matter of rights designed to protect the citizen from all too powerful governors can be traced back to the founding of the Republic. For a consent-based approach to the problem of risk-impositions to work in the ways that I have outlined this vision of citizenship may itself have to undergo transformation. And many of us may have to learn a new set of civic virtues if there is to be any hope of our arriving at a consensus on some of the more pressing public policy questions concerning risks to our health and safety.


I am indebted to all the participants at the Working Session at Georgetown University Medical Center, Washington, D. C., September 21-24, 1989 for their criticisms of an earlier draft and to Sissela Bok, Stephen Graubard, Eli Hirsch, Owen Flanagan, Jerry Samet, and William Talbott for their helpful suggestions.

1. The examples are drawn from Lorraine Daston who offers a brilliant analysis of the attitudes to life insurance, security, and risk in the early modern period in "The Quantification of Risk in the Enlightenment," unpublished manuscript.

2. R. L. Solomon and J. D. Corbit, "An Opponent-process Theory of Motivation," Psychological Review 81 (2), 1974, pp. 119-45.

3. Thomas Schelling, "The Mind as a Consuming Organ," in The Multiple Self, edited by Jon Elster (Cambridge: Cambridge University Press, 1986) pp. 177-95. As Schelling himself notes "in sky-diving and horror movies the glands that secrete euphoriant stimuli are encouraged to be deceived, but not the control centres that would make us sick."

4. Annette Baier, "Poisoning the Wells," in Values at Risk (Totowa, N. J.: Rowman & Allanheld, 1986), p. 49.

5. Robert Nozick who gives a fairly sophisticated account of rights in his book Anarchy, State and Utopia (New York: Basic Books, 1974), p. 75, notes that "no natural-law theory has yet specified a precise line delimiting people's natural rights in risky situations," but does not go on to offer such a theory himself. Samuel Scheffler observes that "while traditional lists of individual rights do typically restrict harmful activity, they are generally silent on the question of risk," in To Breathe Freely, edited by Mary Gibson (Totowa, N.J.: Rowman & Allanheld, 1985) p. 81. Judith Thomson makes a modest, but wonderfully sensitive beginning in her essay "Imposing Risks," also in To Breathe Freely, pp. 124-140. Both Mary Gibson's book and Values at Risk, edited by Douglas MacLean, (Totowa, New Jersey: Rowman & Allanheld, 1986) contain essays that focus on ethical issues raised by government policy concerning air pollution and the meaning and assessment of risk. These two books are among the best treatments of these subjects of which I am aware and I have benefited from them greatly.

6. It would, of course, make matters so much simpler if every question whether it is permissible to impose a risk of harm were always reducible to the question whether it is permissble to cause that harm. But this unfortunately is not likely to be the case as the following example of Judith Thomson brings out: "Here are five, who are starving. They can be saved if and only if Jiggs saves them; but Jiggs will save them if and only if I cause Smith's death by dropping a very heavy weight on his head. Most people are inclined to think I may not do so. But now here are another five, who are also starving. They can be saved if and only if I fly out in my plane and drop them a food parcel. If I do drop the parcel, I impose a risk of death on Smith, for Smith is out in a nearby meadow, tending his sheep, and a sudden gust of wind just might, conceivably, blow the falling parcel onto Smith's head. Is it permissible for me to take that chance? Of course. The risk is utterly trivial. And I think we do feel inclined to go on saying it was permissible for me to take that chance even if, in the event, a sudden gust of wind does blow the food parcel onto Smith's head." To Breathe Freely, pp. 134-35.

7. See Patrick Derr et al, "Worker/Public Protection: The Double Standard," Environment 23, no. 9, 1981 for a similar proposal.

8. Amos Tversky and Daniel Kahneman, "The Framing of Decisions and the Psychology of Choice," Science 211, 1981, pp. 453-8

9. Amos Tversky and Daniel Kahneman, "Can Normative and Descriptive Analysis be Reconciled" Center for Philosophy and Public Policy project funded by the National Science Foundation, grant number SES-8317726, RR-4, (March 1987)

10. See Richard H. Thaler, "Towards a Positive Theory of Consumer Choice," Journal of Economic Behavior and Organization, 1, (March 1980), pp. 39-60.

11. Originally reported by Barbara J. MacNeil, Stephen G. Pauker, Harold C. Fox, Jr. and Amos Tversky, "On the Elicitation of Preferences for Alternative Therapies," New England Journal of Medicine, 306, (May 1982), pp. 1259-62, described by Amos Tversky and Daniel Kahneman in "Can Normative and Descriptive Analysis be Reconciled" and summarized in Philosophy & Public Policy Report, Vol. 8, Number 1 (Winter 1988), p. 4.

12. Philosophy & Public Policy Report, Vol. 8, Number 1 (Winter 1988), p. 3.

13. Originally reported by Baruch Fischoff, "Predicting Frames," Journal of Experimental Psychology: Learning, Memory, and Cognition, 9, (January 1983), pp. 103-16, described by Amos Tversky and Daniel Kahneman in "Can Normative and Descriptive Analysis be Reconciled" and summarized in Philosophy & Public Policy Report, Vol. 8, Number 1 (Winter 1988), p. 4.

14. Herman B. Leonard and Richard J. Zeckhauser, "Cost-Benefit Analysis Applied to Risks: Its Philosophy and Legitimacy," in Values at Risk, edited by Douglas MacLean, pp. 31-48.

15. Ibid., p. 36

16. The example, modified slightly, comes from Nicholas Rescher, RISK: A Philosophical Introduction to the Theory of Risk Evaluation and Management (Washington, D. C.: University Press of America, 1983) p. 5.

17. Douglas MacLean, "Social Values and the Distribution of Risk," in Values at Risk (Totowa, N. J.: Rowman & Allanheld, 1986), p. 78. The example is a variation of one of several that MacLean gives in his essay in order to show how different risk distributions create a problem of equity for risk cost-benefit analysis. It should perhaps be noted that Samuel Scheffler has argued that a cost-benefit analysis need not rely on a scale of measurement to compare costs and benefits that is purely aggregative and he gives a rough indication in The Rejection of Consequentialism (Oxford: Clarendon Press, 1982) Chap. 4, of how distribution-sensitive principles of comparison might, at least partially, be developed.

18. R. Dardis, "The Value of Life: New Evidence from the Marketplace," American Economic Review, vol. 70 (December 1980), pp. 1077-1082.

19. Glen Bloomquist, "Value of Life-Saving: Implications of Consumption Activity," Journal of Political Economy, vol. 87 (June 1979), pp. 540-558.

20. C. A. Olson, "An Analysis of Wage Differentials Received by Workers on Dangerous Jobs," The Journal of Human Resources, vol. 10 (Summer 1981), pp. 165-185.

21. Dan Usher, "The Value of Life for Decision Making in the Public Sector," in Ethics and Economics, edited by Ellen Frankel Paul, Fred D. Miller, Jr., and Jeffrey Paul (Oxford: Basil Blackwell, 1985), p. 191.

22. E. J. Mishan, "Consistency in the Valuation of Life: A Wild Goose Chase?" in Ethics and Economics, p. 152. There is no hard and fast rule that costs and benefits must be construed in narrowly economic terms. Scheffler has argued this point in The Rejection of Consequentialism. At the same time Scheffler acknowledges that "of course, it is much easier to say this than it is to produce a measure of costs and benefits which is not purely economic, which is sensitive to whatever genuine diversity of value there may be, and which provides a practical basis for social decision making. That, presumably, is one of the reasons why the tendency to rely on purely economic measures of costs and benefits is as great as it is." See To Breathe Freely, pp. 78-9.

23. Stuart Hampshire, "Morality and Pessimism," in Morality and Conflict (Cambridge, Mass.: Harvard University Press, 1983), p.86 quoted by MacLean, in Values at Risk , p. 85.

24. Thomas Schelling, "The Life You Save May Be Your Own," in Problems in Public Expenditure Analysis, edited by Samuel Chase (Washington, D. C.: The Brookings Institute, 1968) p. 161-2

25. Ibid., p. 142

26. Dan Usher, "The Value of Life for Decision Making in the Public Sector," in Ethics and Economics, p. 191.

27. Alan Gibbard, "Risk and Value," in Values at Risk, p. 101.

28. Ibid., p. 97.

29. Ibid.

30. Ibid., p. 102

31. Ibid., p. 98

32. Ibid. The content of cost-benefit analysis is seen differently by different analysts. Some regard it "as a neutral device that merely structures a decision-problem." See Douglas MacLean's introduction to Values at Risk. It orders our wants, looks for ways to make them cohere, and identifies the most efficient means to obtain our ends, whatever those ends may happen to be. This is part of its intuitive appeal: it merely helps us to get what we want, it does not tell us what to want. For Leonard and Zeckhauser the expressed preferences of citizens are the appropriate measure of value. Thus, they say at one point: "risks to lives ought to be assigned the same value by decision-makers that the individuals at risk themselves apply." But they also take the position that cost-benefit analysis is a corrective to our unsophisticated judgments about risks. It seems to me that they cannot have it both ways. There is a tension, perhaps all too obvious, between the intuitive justification of cost-benefit analysis and a defense of it as a corrective to our unsophisticated judgments. This tension spells further trouble for the task of establishing the political legitimacy of the technique. Since its legitimacy hinges in part on its public acceptance, if citizens do not want what cost-benefit analysis prescribes it would be rational for them to want, they are likely to be unhappy with public policies that are based on cost-benefit analysis. This last point is a point made by Amartya Sen and Bernard Williams about utilitarian theory in their introduction to Utilitarianism and Beyond (Cambridge: Cambridge University Press, 1982) adapted to apply to cost-benefit analysis. There is also considerable disagreement amoung utilitarian theorists about what preferences ought to be taken into account by the theory: people's actual preferences or, to take just one example, only their "perfectly prudent preferences" (R. M. Hare's view).

33. Herman B. Leonard and Richard J. Zeckhauser, "Cost-Benefit Analysis Applied to Risks: Its Philosophy and Legitimacy," in Values at Risk, p. 42

34. Ibid., p. 31

35. For a further discussion of some of these issues see the introduction by Amartya Sen and Bernard Williams to their edited volume Utilitarianism and Beyond.

36. There is evidence, some of it disputed, that secondhand tobacco smoke increases the risk of cancer. Already in 1980 at the Naval Research Laboratory in Washington, D. C. James Repace and A. H. Lowrey demonstrated that "the smoke pollution inhaled indirectly from cigarettes, pipes, and cigars indoors was not only chemically related to the smoke from factory chimneys, but routinely occurred at far higher levels than did factory smoke or automobile exhaust outdoors." See "Indoor Air Pollution, Tobacco Smoke, and Public Health, " Science 208 (1980). At about the same time J. R. White and F. H. Froeb reported in a study of 2,000 subjects over a ten year period that exposure to indoor tobacco smoke in the workplace caused a reduction in small passageways in the lung to the same extent as smoking one to ten cigarettes a day. A more controversial study by Takeshi Hirayama, the chief epidemiologist of the National Cancer Center Research Institute in Tokyo, reported in a study of 142,857 subjects that nonsmoking women married to men developed lung cancer at nearly double the rate of women whose husbands did not smoke. "Nonsmoking Wives of Heavy Smokers Have a Higher Risk of Cancer: A Study from Japan, British Medical Journal 282 (1981). But about the same time Lawrence Garfinkel of the American Cancer Society published a study that observed lung cancer rates in 176,739 nonsmoking women married to husbands who smoked and found only a small positive result without statistical significance. "Time Trends in Lung Cancer Mortality Among Nonsmokers and a Note on Passive Smoking," Journal of the National Cancer Institute 66 (1981). For a further discussion of these and other findings see James Repace, "Risks of Passive Smoking," in To Breathe Freely, pp. 3-30.

37. Peter Railton "Locke, Stock, and Peril: Natural Property Rights, Pollution, and Risk," in To Breathe Freely, p. 92

38. Philosophy & Public Policy Report, Vol. 5, Number 3 (Summer 1985), p. 2.

39. Ibid.

40. Ibid., p. 3.

41. Ibid.

42. Ibid., p. 2.

43. It is sometimes thought that a violation of a person's rights might be justified if one compensates him afterwards. If this were so, this could become another way to turn a potential violation of a right into a permissible act, implying yet another way to justify risk-impositions. On this view consent would not be required and risk-impositions would be justified so long as one compensated each and every person for any harm that occurred as the result of one's having exposed that person to the risk of harm. But it seems to me that a readiness to compensate a person cannot turn his right not to be harmed without his consent into a morally acceptable act. To harm a person without his consent so long as one is willing to compensate afterwards, pre-empts his right to decide things for himself, treats him as a mere means to one's own ends, and enables those who have sufficient resources to engage in harmful activities more readily and with less compunction than those of lesser means. One way not to deal with the risks imposed on non-smokers from secondhand cigarette smoke is by placing an excise tax or surcharge on cigarettes since it would simply allow those who could afford the increased cost of cigarettes to pay and pollute. An argument for regarding standards of safety and health in the workplace as a matter of workers' rights whose violations are only permissible with their consent is to block the option otherwise open to firms to pay compensation costs rather than making the workplace a safe and healthy environment within which to work. For a related discussion of these issues, see Peter Railton, "Locke, Stock, and Peril: Natural Property Rights, Pollution, and Risk," in To Breathe Freely, especially pp. 112-17. Railton has also expressed surprise that a willingness to compensate could ever, on its own, justify the violation of someone else's rights: "It would seem to be a clear case of using someone - not necessarily misusing him, but using him - if I were to harm or endanger him in order to pursue my own interests, but then made sure to provide after-the-fact compensation at a level he would have to take or leave." p. 115.

44. Herman B. Leonard and Richard J. Zeckhauser, "Cost-Benefit Analysis Applied to Risks: Its Philosophy and Legitimacy," in Values at Risk, p. 47.

45. Ibid.

46. Ibid., p. 37

47. Shakespeare, King Lear, Act V, Scene III, Gloucester: "And that's true too."

48. In their effort to legitimate cost-benefit analysis as a form of hypothetical consent Leonard and Zeckhauser acknowledge that "cost-benefit analysis [only] coincides with universal individual consent when all individuals in the community are affected similarly under each alternative considered. It is much more difficult to determine whether the benefit-cost criterion is a useful means of defining 'hypothetical consent' by the community when individuals have widely different interests." "Cost-Benefit Analysis Applied to Risks: Its Philosophy and Legitimacy," in Values at Risk, p. 36

49. Douglas MacLean, Introduction to Values at Risk, pp. 4-5

50. Philosophy & Public Policy Report, Vol. 8, Number 1 (Winter 1988), p. 2

51. See Baruch Fischoff, "Cognitive and Institutional Barriers to 'Informed Consent,'" in To Breathe Freely, pp. 180-82. Fischoff has ventured the suggestion: "If neutrality is the problem, an obvious possibility is to replace the other party with an actual decision counselor, like a court-appointed lawyer, entrusted with helping the decision maker and enpowered to extract information from the other party or purchase it from other sources (within some budgetary constraint)." p. 182.

52. Douglas MacLean, quoted in Philosophy & Public Policy Report, Vol. 8, Number 1 (Winter 1988), p. 4.

53.. Philosophy & Public Policy Report, Vol. 8, Number 1 (Winter 1988), p. 2

54. Ibid.

55. The example comes from Richard Thaler, "Towards a Positive Theory of Consumer Behavior," Journal of Economic Behavior and Organization, 1, (March 1980), pp. 39-60 and is reported by Jon Elster in The Multiple Self (Cambridge: Cambridge University Press, 1986), p. 27.

56. In order for the two options to be materially equilvalent, it would be necessary to assume that George, having decided to mow the lawn in return for his neighbor's donating $20 to charity, also decides not to contribute $20 that he was planning to make himself to charity.

57. The answer, of course, will depend on how we understand rationality. On a narrow conception we might conclude that George has acted irrationally, but that there is more to life than rational action. Or we may wish to build this something more into our understanding of rationality and conclude that there is more to rationality than is to be found in the narrow, economic conception.

58. This suggestion in fact comes from Amos Tversky in correspondence with Jon Elster. See Elster's Introduction to The Multiple Self, p. 27.

59. Alan Gibbard, "Risk and Value," in Values at Risk, p. 102.

60. Mary Gibson in the Introduction to To Breathe Freely, p. xi.

61.. Samuel Scheffler, in To Breathe Freely, p. 84.

62. Mark McCarthy, "Reform of Occupational Safety and Health Policy," in To Breathe Freely, pp. 201-221.

63. Union officials, as Mark McCarthy is well aware, are not always pleased by proposals for joint labor-management committees and are reluctant to include safety and health concerns in the bargaining process. Their worries are also not completely without reason and need to be taken into account if any restructuring of present relationships between labor and management were to occur. Union opposition is expressed on a variety of grounds by saying that health and safety are rights and workers ought not to have to give up wages or other benefits to pay for them, that safety committees are a way to co-opt workers, that management does not take occupational safety and health seriously enough and ought to be compelled to improve conditions by stricter national regulations, that unions have neither the power nor the legitimacy in this country where only 20 percent of the workforce is unionized to negotiate with management to improve health and safety conditions in the workplace, and that negotiated agreements on health and safety between labor and management open unions up to liability suits from disabled workers. See Mark McCarthy, "Reform of Occupational Safety and Health Policy," in To Breathe Freely, pp. 216-17.

64. Judith Thomson, "Imposing Risks," in To Breathe Freely, pp. 139.

65. Since certain cases of tacit consent may be thought to be morally invalid, and may not count as consent, it should perhaps be noted that there are also circumstancess in which actual consent is thought to be morally invalid. Thus in contractual relationships there are a number of conditions that are generally taken to invalidate the actual agreement among the parties, even if both parties have given their consent, to say nothing of the more complex cases in which courts have refused to uphold workers' agreements to exempt their employer from liability due to negligence. For a potential risk-bearer's consent to justify a risk-imposition the conditions under which both tacit and actual consent are morally vaild will have to be fully and clearly set out.

66.For accounts of the contrast between adversary and consensus culture see, Steven Kelman, Regulating America, Regulating Sweden: A Comparative Study of Occupational Safety and Health Policy (Cambridge, Mass.: MIT Press, 1982) and Michael Thompson, "To Hell with the Turkeys! A Diatribe Directed at the Pernicious Trepidity of the Current Intellectual Debate on Risk," in Values at Risk, pp. 113-135.