Ethics of Risk PDF

Ethics of Risk Author(s): J. E. J. Altham Source: Proceedings of the Aristotelian Society , 1983 - 1984, New Series, Vol. 84 (1983 - 1984), pp. 15-29 Published by: Oxford University Press on behalf of The Aristotelian Society Stable URL: https://www.jstor.org/stable/4545003 JSTOR is a not-for-profi...

Ethics of Risk Author(s): J. E. J. Altham Source: Proceedings of the Aristotelian Society , 1983 - 1984, New Series, Vol. 84 (1983 - 1984), pp. 15-29 Published by: Oxford University Press on behalf of The Aristotelian Society Stable URL: https://www.jstor.org/stable/4545003 JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at https://about.jstor.org/terms The Aristotelian Society and Oxford University Press are collaborating with JSTOR to digitize, preserve and extend access to Proceedings of the Aristotelian Society This content downloaded from 213.55.225.:ffff:ffff:ffff on Thu, 01 Jan 1976 12:34:56 UTC All use subject to https://about.jstor.org/terms  II*-ETHICS OF RISK by J. E. J. Altham I The ethics of risk is that branch of moral philosophy that enquires into the principles that should morally guide choice in circumstances where the outcome of some possible course of action is not known. Two general classes of such circumstances are usually distinguished. One is the class in which a decision is e really taken under risk, the other that in which a decision is taken operated under uncertainty. In a problem of decision under risk, the outcome of some possible course of action is not known in unity advance, but the range of possible outcomes of each available course of action is known, and the probabilities of all the outcomes in each range are also known. A decision is taken under uncertainty when even less knowledge is available than in a case of decision under risk. In particular, even where the possible outcomes are known in advance, the probabilities of each of these are not. The expression 'ethics of risk' is intended to cover both classes of problem; it may be thought of as short for 'ethics of risk and uncertainty'. The subject covers an enormous field in both private and public life. The major decisions of our lives are made without the knowledge necessary for certainty about the outcome. When a man and woman decide to get married, they do not know that it will be a success. If they decide to have children, they do not know that the baby will be born safely without serious defects, nor do they know what their own competence will be as parents. The choice of occupation and the choice of where to live are also made with only imperfect information. We do not usually know when we will die, nor when we shall be incapacitated by illness or accident; if we did, it would make a great difference to many decisions. It is even more obvious that major decisions of governments are taken under risk or uncertainty, whether the * Meeting of the Aristotelian Society held at 5/7 Tavistock Place, London WCI, on Monday, 24 October, 1983 at 6.00 p.m. This content downloaded from 213.55.225.15 on Sat, 15 Apr 2023 14:04:42 UTC All use subject to https://about.jstor.org/terms 16 J. E. J. ALTHAM problem be one of responding to the Argentine invasion of the Falklands, of making plans to provide enough but not too much electricity-generating capacity in several years time, or fixing standards of safety in the chemical industry. Not surprisingly there are many disciplines that in one way or another study risk and help us to understand and to cope with the problems that arise because information is imperfect. Actuaries, decision analysts, game theorists, economists, legal theorists are among those who have to give advice on such problems. But considering the pervasiveness of human ignorance, one is surprised that questions concerning risk and uncertainty have not usually been near the centre of discussions by moral philosophers. It seems to me that there is something of a gap in moral theory here. One possible explanation of the comparative lack of attention paid to the ethics of risk by moral philosophers may be a belief that no special principles or considerations come into play in situations of risk, and that consequently there is no such subject as the ethics of risk. One point of view that leads to this conclusion is the following. What values there are is decided, in so far as it can be decided, by moral philosophy that does not essentially involve risk or uncertainty. This philosophy also decides, again to the extent that this can be decided, on the relative weights to be given to the various values it discerns. Decision theory, on the other hand, tells us what it is rational to do in given circumstances of risk or uncertainty, given relevant information about our values. There would then be no distinct ground for the ethics of risk to occupy. It would have only the role of putting together and applying the results of other branches of enquiry, and would have no principles of its own. Utilitarianism for example leaves no place for an ethics of risk with principles of its own. In a standard form it copes with risk and uncertainty as follows. Everything of value is interpreted in terms of, and measured by, utility. There are different views about what utility itself is, but they do not now concern us, except to note that the determination of utility does not have to involve considerations of risk. The single ultimate injunction is to maximise the net balance of this utility. In circumstances of risk the agent does not know what will maximise utility. However, given that utility is the value, it is rational to maximise expected utility, and this is the most straightforward This content downloaded from 213.55.225.15 on Sat, 15 Apr 2023 14:04:42 UTC All use subject to https://about.jstor.org/terms  ETHICS OF RISK 17 utilitarian injunction. Coping with uncertainty is trickier. If even the probabilities of outcomes are unknown, expected utility does not have any immediate meaning. The utilitarian can, however, resort to subjective probabilities, and maximise expected utility on that basis. No new moral principle is involved in either case; utility and rational decision are the tools needed. Utilitarianism has the merit of seeming to make sense of some commonsense rules in this area. For example, the more harm something might do if it goes wrong, the higher must be the standards of inspection. A motor car has to have an MOT test after three years, and yearly thereafter. Much more stringent and frequent tests are required for a nuclear power station. Again, a dangerous activity may be permitted if the utility to be gained from it is sufficient, but is forbidden if its objects are trivial. People are permitted to drive motor cars, but if it could be shown that soap operas on television cause as much damage (brain damage presumably) as is caused on the roads, soap operas would be banned. Facts like these fit utilitarianism well. This essay, however, is written from a standpoint of dissatisfaction with utilitarianism, based on grounds that I cannot argue here. Its object is to explore how two other frameworks of ethical thought might try to deal with problems or risk, namely natural rights theory and contractualism. Both of these run into serious problems, and there remains a serious task for those who wish to reject utilitarianism to provide an alternative theory in this field. I can take only a few tentative steps, but they will I hope indicate that if an alternative to utilitarianism is to be provided, it will have to give a distinctive place to the ethics of risk, which would have principles of its own, and not merely apply the results of other branches of enquiry. II The first task is to consider natural rights. The putative right to focus on is the right not to be put at risk of harm by the activities of others. Nozick has discussed this in a passage that is a fine example of the candour with which he brings out difficulties in his own position. My discussion is closely based on his.1 If natural rights theory of risk is to be allowed to get off the ground then the theory must be permitted the assumption, at least for This content downloaded from 213.55.225.15 on Sat, 15 Apr 2023 14:04:42 UTC All use subject to https://about.jstor.org/terms 18 J. E. J. ALTHAM the sake of argument, that there is a natural right not to be harmed. If there is a right not to be put at risk of being harmed, it seems that we can only be brought to see that there is, and what its extent is, by being shown its relation to the right not be harmed. Let us then start by accepting that ifJohn fires a loaded gun at Mary, intending that the bullet shall harm her, and it does, then barring some pretty special circumstances he has violated a right of hers. But suppose that he plays Russian roulette with her as target. He has five chambers with blanks in his revolver, and one with a live bullet. He spins and fires. She is hit by a live bullet, and is severely wounded. It remains plausible to say that he has violated her rights. But suppose that she is lucky, and only a blank is fired. No actual harm is done. Has John violated Mary's rights? The natural rights theorist must say that he has. For he sets rights as prior to duties. If his position is to have any plausibility, he must hold that John is wrong to play Russian roulette with Mary as target, that it is his duty not to. This duty must be derived from her right. So whether or not her right has been violated is independent of the actual outcome. The conclusion is acceptable. The problems begin when we consider cases where the probability of harm is much lower. The theory loses all plausibility if it claims that nobody must ever do anything that has any positive probability, however small, of resulting in harm to someone else. This rule would paralyse human life. So at least a small probability of harm to others must be acceptable. The task is to say how small the probability must be for the act to be allowable. Harms as well as probabilities vary in magnitude, which prompts the question whether the threshold probability above which a risky action is prohibited should be the same for all harms or not. There are problems either way. The idea that it should be the same seems to lead to absurdity. Consider some violation of rights of great triviality, such as dirtying your handkerchief without your permission. Although hard to take seriously, this is a genuine violation of rights. Soiling your handkerchief without your permission is a worsening of the condition of your property without your allowing it, and this is prohibited. However, if we were to try to fix an appropriate threshold probability of soiling your hand- kerchief, below which it would be all right to perform an action This content downloaded from 213.55.225.15 on Sat, 15 Apr 2023 14:04:42 UTC All use subject to https://about.jstor.org/terms  ETHICS OF RISK 19 which risked having this result, we should not wish to make the didn't threshold very low. If we are prepared to go along with this exercise, we might settle on a probability of about one in fifty. If oder an action had a less than one in fifty chance of resulting in stand dirtying your handkerchief, and there was nothing else to be said against it, then it would be all right to perform it. But if the threshold probability is the same for all harms, then if it is all right to take a one in fifty chance of soiling your handkerchief it is also all right to take a one in fifty chance of killing you. But that is absurd. Adjusting the threshold probability, while keeping it the same for all harms, does not enable any escape to be made from the absurdity, as a little reflection will show. So let us consider the consequences of the alternative, that the threshold varies with the magnitude of the harm. Under this supposition it would be all right to do something with a comparatively large probability of causing loss to somebody else, if that loss were sufficiently slight. But if the possible loss or harm is great, then the threshold probability would be correspondingly small. That looks more sensible than the previous suggestion, and makes it appear that the rule should be framed in terms of a specified value of the expected harm from an action. An action will be prohibited, then, if the expected harm from its performance exceeds the specified value. Actions with small values of expected harm will be permitted. But as Nozick is quick to point out, permitting actions with small but still positive values of expected harm does not fit the character of at least his version of natural rights theory. For consider an action which is certain to cause some very slight harm, a harm so small as to be less than the specified threshold value. According to Nozickian theory, it is not all right to perform it. 'De minimis non curat ius naturale' is not a principle that holds in his theory. Certainty of outcome shades insensibly into probability of outcome, and so the specified value of the threshold expected harm must match the threshold of certain harm. The trouble is that where harm is certain, the threshold is zero. One is prohibited from causing any harm. Consequently, one must be prohibited from doing anything that has any positive probability of causing any degree of harm, however slight. This rule is, however, quite clearly impracticably restrictive. This content downloaded from 213.55.225.15 on Sat, 15 Apr 2023 14:04:42 UTC All use subject to https://about.jstor.org/terms 20 J. E. J. ALTHAM These considerations bring out clearly that a strict absolutist theory of natural rights is unworkable in the context of risk, which in turn throws considerable doubt on whether absolutism is tenable where risk is not involved. For if it is permissible to impose some risk of harm on others, then if enough people take advantage of such permission sooner or later the harm in question will eventuate. It would be implausible to permit a general practice which occasionally results in harm, where it is known that it will, and yet forbid absolutely every particular action that will certainly result in harm. In both cases some attention must be paid to that for the sake of which the actions in question are done. This is the other factor missing from Nozick's discussion. Just as, where risk is not involved he does not allow any trade-off between a violation of rights and any good to be achieved, or evil averted, by doing so (except if the evil to be averted is catastrophic), so no similar trade-off is permissible where risk is involved. It is not necessary to be a utilitarian to see that this is unrealistic. Absolutism is at least definite. Once it is relaxed in cases involving risk it is difficult to see how natural rights theory can attain any definiteness about the extent of the relaxation. Practicability requires that we say that people at most have the right not to have risk imposed on them without their consent except when some sufficient value is attainable only by such imposition of risk. But it is hard to see how natural rights theory can remove the vagueness in this formulation, so as to specify what values are sufficient for the imposition of what degree of risk. No one answer is self-evident. We can ask what the constraints represented by rights are based on, but since in this type of theory 'right' is the fundamental moral term, our answe cannot be based in other moral terms, but must rather appeal to such fundamental ideas about man as that he is a self- determining rational agent, that he is a being with the capacit for a meaningful life, or that he is an end in himself. The prospects of arguing from conceptions of such generality to a determinate answer to questions in the ethics of risk look bleak. Rather than pursue these ideas further, it is more promising to look elsewhere. This content downloaded from 213.55.225.15 on Sat, 15 Apr 2023 14:04:42 UTC All use subject to https://about.jstor.org/terms  ETHICS OF RISK 21 III If we now turn towards contractualist theory, we have as usual to say something about Rawls.2 It is one of the best known and most striking features of Rawls's theory that it makes use of a model involving choice under uncertainty. It might thus at first seem to be well-placed to treat the ethics of risk and uncertainty, but it turns out that this is not so. Rawls argues that his original position, in which parties choose the principles to govern the basic structure of society, has special features that make it rational for them to follow the maximin rule of choice in making their decision. Three standard objections are first that it is irrational to use maximin, secondly that if it is rational to do so it is not uniquely rational, and thirdly that even if it is rational it is irrelevant because the original position is an inappropriate model. But in saying that Rawls does not really help with the present problem of the ethics of risk I am not thinking of any of these objections, but rather that his strictly limited defence of maximin gives us very little guidance as to what principles about acceptance and imposition of risk we ought to adopt. The maximin rule is claimed to be applicable when three conditions in particular hold. One is that the probabilities of outcomes cannot be computed on an objective basis. The second is that it is not worthwhile taking an unknown additional chance for the sake of extra goods beyond the minimum one can be sure of attaining. The third is that rejected alternatives have some possible outcomes that are unacceptable. The original position is defined so as to satisfy these three conditions. Now suppose the parties in the original position try to extend their purview beyond the consideration of principles for liberty, and the distribution of income, wealth and opportunity, and ask themselves to what extent, and for the sake of what, it is permissible to engage in activities from which others might suffer. This is plausibly represented as a question ofjustice, and indeed of justice as fairness. What risks is it fair to impose on others? Faced with this question, the parties in the original position would quickly realise the impossibility of eliminating risk altogether. Since each is bound to be subject to some risk, there will be no definable minimum that each can be sure of attaining. However, they can be more or less cautious. The analogue to the This content downloaded from 213.55.225.15 on Sat, 15 Apr 2023 14:04:42 UTC All use subject to https://about.jstor.org/terms 22 J. E. J. ALTHAM conclusion using the maximin rule would be to make the risk as small as possible. To do this, however, involves them in enormous cost, and is prima facie irrational. Their position is not one that satisfies Rawls's second condition for the rationality of maximin. For it is worthwhile taking an extra chance for the sake of extra goods, if the cost of an additional factor of safety is sufficiently high. Therefore, on good Rawlsian grounds it would be reasonable for the parties to be less than maximally cautious. But if we ask how much risk beyond the irreducible minimum they should accept, the device of a Rawlsian original position does not seem to give us much guidance. It would be quite out of keeping with his general position for the parties to go for maximising expected utility, since this principle suffers from insensitivity to distributional considerations in just the same way as utilitarianism generally, and as Rawls objects to the latter on these grounds, so he would object to the former. The best that the Rawlsian original position can do in the ethics of risk is to regard the absence of risk as one of the primary social goods. Absence of risk is then one of the bundle of goods to which the difference principle applies: the long-run prospects of a representative of the least advantaged group must be maximised, and one element in these prospects is the degree of security from risk of harm that is offered. But this treatment of the question evades rather than solves the problem I am addressing. For this problem is what risks it is fair to impose for the sake of other goods, or what is the fair weight to be given to a given degree of risk in compiling an index of goods. The difference principle limits the inequality in the distribution of allowable risk, but gives little guidance as to what general level is allowable. The existence of risk, and the manifold ways in which it arises from human activities, also provide a striking illustration of one of Hart's main criticisms of the Rawlsian principle of the priority of liberty. '... whether or not it is in any man's interest to choose that any specific liberty should be generally distributed depends on whether the advantages for him of the exercise of that liberty outweigh the variousdisadvantages for him of its general practice by others.'3 The general exercise of some liberties may appreciably raise the general risk-level. Sometimes the risk-level may be raised so far that it is clear that the disadvantages of permitting the activity outweigh the This content downloaded from 213.55.225.15 on Sat, 15 Apr 2023 14:04:42 UTC All use subject to https://about.jstor.org/terms  ETHICS OF RISK 23 advantages. Drunk driving can serve as an example. Harsanyi also uses the device of a veil of ignorance in formulating an 'original position' from which moral principle are to be chosen. Because of this feature, it is appropriate to something about his views at this point, although he is not a contractualist. In his model, a chooser knows how each person will be affected by action in accordance with each principle he might select, but the chooser does not know which of the affected individuals he or she is. But whereas in Rawls a party in the original position cannot assign any probability to his being any particular individual, in Harsanyi the chooser is supposed to choose as though there were an equal probability of his being any of the affected parties. He argues that in this position it is rational for the chooser to maximise expected utility. This leads Harsanyi to a form of utilitarianism.4 Taking the perspective of the ethics of risk helps us to see that the use of this equiprobability model of choice is unpersuasive. In ordinar we generally think that there is a great difference, morally speaking, between choosing to undergo some risk and choosing an action that imposes an equivalent risk on others. It may be all right for me to practise tricky manoeuvres while hang-gliding, but wrong for me to do something as dangerous to others as my hang-gliding stunts are to myself. Similarly, if asked to choose what should be done on the assumption that I have an equal probability of being any of the parties affected I may make a different choice from the one I should make if asked to choose on the assumption that some other person for whom I have disinterested sympathy has a similarly equal probability of being any of those affected. If there is one person who comes off particularly badly, I might be prepared to run that risk for myself, but think it wrong to choose that someone else should run it. The equiprobability model in effect assumes an analogy between individual and collective rationality that ordinary moral thinking rejects in the case of risk as elsewhere, and that is often seen by critics of utilitarianism as a source of its weakness. This makes it quite unpersuasive. IV If a contractualist theory is to provide a satisfactory framework for thinking about the ethics of risk, it will need to take serious This content downloaded from 213.55.225.15 on Sat, 15 Apr 2023 14:04:42 UTC All use subject to https://about.jstor.org/terms 24 J. E. J. ALTHAM account of the asymmetry between accepting risk for oneself and imposing it on others. The sketch I offer in the remainder of this paper does this. It starts, however, from a different point. One of the more evident facts in this area is that people's attitudes towards risk and uncertainty vary greatly. To some people, in some circumstances, the consciousness of risk has positive value. To others in the same circumstances, the consciousness of risk may be unpleasant. What gives one some pleasurable excitement gives another the pain of anxiety. A person may be tempera- mentally risk-averse or risk-loving. Similarly a person may relish or detest the experience of uncertainty, where objective probabilities are not assignable. Let us call the man who dislikes the experience of uncertainty cautious, and the one who relishes it adventurous. Now while some extreme degrees of caution or risk-aversion, and some extreme degrees of love of risk or adven- turousness seem absurd, there is a range of each temperament within which any position cannot be condemned as irrational. The White Knight in Through the Looking Glass is the symbol of absurdity here: his horse had anklets round its feet to guard against the bites of sharks. On the other hand it is neither absurd to wear a crash helmet on a push-bike, nor absurd not to. Such differences in legitimate attitudes cannot be ignored in the ethics of risk, but they create serious problems, particularly in the context of contractualism. For if we are looking for a suitable original position in which individuals are to come to an agreement on principles governing how much risk to allow, and for the sake of what, these individuals must either be supposed to know their own attitudes in the face of risk and uncertainty or not. If they do know them, then either they all share the same attitudes, or they do not. If they share the same attitudes, and arrive at principles on the basis of them, then the result will seem arbitrary, since other attitudes, presumably leading to different conclusions, are no less rational. If they do not share the same attitudes, then it is hard to see how they could agree. On the other hand, if they do not know their own attitudes, then they will have to decide in conditions of risk or uncertainty about what their own attitudes to risk and uncertainty are. Some principle of decision in such conditions will have to be ascribed to them, and arbitrariness reappears again, arising from the fact that there is no one principle of decision which is This content downloaded from 213.55.225.15 on Sat, 15 Apr 2023 14:04:42 UTC All use subject to https://about.jstor.org/terms  ETHICS OF RISK 25 uniquely correct. Principles risk, so that the initial exclusion of these from the original position is in vain; they reappear by the back door. A further difficulty is that attitudes towards risk and un- certainty cannot be isolated from other attitudes, tastes and values. Different people have desires for different kinds of life, and some kinds of life demand a form of society in which the general level of risk is higher than it is in others. A frugal, regulated life in small communities, based on the application of selected safe technologies, is advocated by some members of the environmental movement. For its advocates, the alleged relative absence of risk in such a society compared with the liberal high technology society we live in in the West is only one of its attractions. For its detractors, such a society would be insufferably boring, and that would be only one of its defects. It is far too ambitious to hope to settle such conflicts of ideals. The most that can reasonably be attempted is to see what kind of modus vivendi might be arrived at, in so far as the differing preferences are rooted in differing attitudes towards risk. These differences in value and attitudes give rise to intractable problems, and the few positive moves I am able to make here enable only very limited progress to be made. But by breaking the problems up it is possible to become clearer about where the primary difficulty lies. First there are cases where a number of people all think it valuable to engage in some activity, where that activity involves some risks to persons other than the ones engaging in it. If one person agrees to its being permissible to engage in it, he must think that the value to him of being allowed to engage in it outweighs the disvalue of the extra risk he bears through others engaging in it. The activity in question may unavoidably involve risk to others, but the degree of risk involved may depend, and usually will, on how it is carried on. It can be done with more or less stringent precautions. Parties can then agree if there is a level of stringency of precaution at which it is worthwhile for each to be allowed to engage in the activity despite the additional risk from the others' doing so as well. Such agreement is possible even among people whose attitudes towards risk are not all the same. It may be worthwhile for the This content downloaded from 213.55.225.15 on Sat, 15 Apr 2023 14:04:42 UTC All use subject to https://about.jstor.org/terms 26 J. E. J. ALTHAM more cautious to put up with more risk than he would ideally like for the sake of being able to engage in an activity, and worthwhile for the more sanguine to put up with a more stringent level of precautions than he thinks is really necessary in order to secure the agreement that would allow him to engage in the activity in question. Provided that the positions of the parties are reasonably equal, so that no group has the power to impose unfair terms, the outcome in such cases may be left to bargaining. Secondly, there is little problem if those who wish to carry on some activity that carries risk to others are able, at not too great cost, to isolate the others from the risks so that they do not need to bear them after all. Thirdly, there are possibilities of trade between people all of whom want to do some things that carry risks to others, but who do not all want to do the same risky things. There may be possibilities of putting together a package of activities and associated precautions that permits each to pursue his ends subject to constraints, and that each can agree on as making him better off. The possibilities gestured towards here help us to see where the really difficult area lies. This is where a person cannot be isolated from the risks occasioned by desired activities of others, but does not have enough desires of his own for activities that are risky to others for a satisfactory bargain to be possible. Suppose Jack's satisfactions come from eating nuts, practising yoga, talking to friends and watching the sunset, while Norman's vast desires require nuclear power stations to provide the energy to enable him to satisfy what he thinks of as his needs. How can there be a reasonable agreement between them? Jack is not impressed by Norman's offer to compensate him for any harm he suffers as a result of Norman's activities. His difficulty is that if he is conceived as merely concerned to advance his own interests, he has a reason for agreeing to a risk on condition compensation is paid if he subsequently suffers from the risk's eventuating only if the compensation can make him at least as well off as he would have been without it and without the risky activity. But this severely limits the role of compensation. Many harms cannot be compensated for in this full sense. Further, ifJack is a nervous person, he may be made This content downloaded from 213.55.225.15 on Sat, 15 Apr 2023 14:04:42 UTC All use subject to https://about.jstor.org/terms  ETHICS OF RISK 27 anxious or even frightened by knowing that he is subject to certain risks that result from the activities of Norman and others who share Norman's tastes. It is very difficult to compensate for fear. If the cause of fear is short-lived, then it is possible that those made afraid might be made an award sufficient to convince them that overall they are as well-off as they would have been had they never had to suffer the fear in the first place. But here we are speaking of compensation for a continuing fear. There is no real substitute for a sense of security. If Jack feels fearful, it is no use offering him some other good than safety to make up for his fear. For his fear will pervade his life, and poison his enjoyment of such other goods. We may then envisage the more cautious arguing as follows with the less cautious: 'We want a safer environment. We do not think that we shall gain significantly from further activities that increase the risks to us. If the level of risk goes up, some of us will feel insecure, and any of us might be harmed. We can indeed reasonably believe, since we are speaking of cases where we cannot be isolated from the risk, that some of us will be harmed in the extra accidents that will occur if these extra activities go ahead. What reason can you give us for agreeing that you should go ahead and increase the level of risk? It is no use your offering us some other advantages to compensate us for undergoing the additional risk. The more nervous among us will feel fear if the risk goes up as you propose, and do not wish to accept any advantage in exchange for their sense of security. Our knowledge that some of the harms that sooner or later some of will suffer from these risky activities are uncompensable also makes us unimpressed with your offer.' On the other side peop complain in their turn that if the cautious veto their activities i this way, their desires will be unable to find expression, and the prospects for happiness severely damaged in consequence. Two aspects of this difficult issue must be distinguished. On concerns fear, the other the remaining disvalue to the more cautious of a higher level of risk than they prefer. These are importantly different. If you impose a risk on me that is serious enough to make me feel insecure and afraid, then my conscious well-being is affected. If you impose a risk on me which is too slight for me to regard it with anxiety, then although it is against my interest to be subject to this risk, unless I get benefits in This content downloaded from 213.55.225.15 on Sat, 15 Apr 2023 14:04:42 UTC All use subject to https://about.jstor.org/terms 28 J. E. J. ALTHAM exchange that I consider acceptable, my conscious well-being is not affected, unless I should be unlucky enough to be a victim of an event that realises the risk. In the one case the thought of a loss of expected utility gives rise to a loss of actual utility, in the other it does not give rise to any loss of actual utility. Consider the risk of being killed by an aircraft dropping out of the sky. It is a genuine risk, and it is very difficult to protect oneself absolutely from it while leading a normal life. But it is a very small risk indeed. The thought that I just might be killed by an aircraft falling out of the sky causes me no anxiety whatever. It is a thought that fails even to excite my interest. Yet if asked whether it is more in my interest for there to be no chance of being killed in this way I should say that it is, provided that securing this would involve me in no loss of other value. If on the other hand there is a proposal to site a dump for storage of toxic waste near my house, the additional risk I perceive this as creating, particularly if I have reason to believe that the dump will be operated by cowboys, will be enough to make me anxious to a significant degree. Suppose then that there is a prevailing level of risk that does not cause anxiety, even to the more nervous. (There is a degree of nervousness that must be regarded as pathological. Just where it lies is a matter for dispute, but this is not the place to try to fix the point where paranoia begins.) Let us call an additional risk that does not lift the level over the threshold into one that does cause anxiety a subliminal extra risk. Then if an extra risk is subliminal, then although it is against a person's interest to accept it if it brings no benefits, there will not in fact be any strong motive to reject it, since knowledge that it exists does not diminish a person's well-being. But what reason is there for a person to agree to even a subliminal extra risk, if he gains nothing from the agreement? The only resource I can discern here is to assign a desire for agreement to those whom I am imagining to be discussing principles of risk and uncertainty. The basis of such a desire could be either a recognition of the ill consequences of non-agreement, or a recognition of the positive value of agreement itself. This needs more discussion, but if such a desire for agreement can be assigned, then it could outweigh the rather abstract interest there is in not agreeing to an extra subliminal risk that does not otherwise benefit one. The cautious This content downloaded from 213.55.225.15 on Sat, 15 Apr 2023 14:04:42 UTC All use subject to https://about.jstor.org/terms  ETHICS OF RISK 29 could then agree to additional subliminal risks for the sake of agreement itself. It would be implausible to strengthen this line of argument, so as to bring the cautious to agree to extra risks that bring the overall level above the threshold. Their fear creates too great a block. However, the desire for agreement leads the less cautious into incurring the costs necessary to bring their desired activities below the threshold of anxiety of the more cautious. The extra subliminal risk endured by some is thus balanced by the extra costs incurred by others in keeping below the threshold. One feature of common attitudes that this picture matches well is the difference between the level of risk that people put up with in activities they have chosen to engage in, and the very much lower level that is all that they will normally tolerate if the risk is imposed. The risk of smoking twenty cigarettes a day is reckoned to be pretty high. A risk of similar magnitude to the general public from an industrial process, even a very beneficial one, would be regarded as quite intolerable. Even much lower risks are unacceptable. It is difficult for a utilitarian approach to account for the magnitude of this difference, and also difficult to believe that such a difference is not desirable. The difference in acceptability between voluntary and imposed risks makes much better sense from the sort of perspective that I have been sketching, of parties with varying attitudes and interests searching for agreement, and willing to bear certain costs in the cause of such agreement. NOTES what costs 'Robert Nozick, Anarchy, State, and Utopia (Blackwell, 1974), esp. pp. 73-78. 2John Rawis, A Thieory of 3ustice (Clarendon Press, 1972). 3H. L. A. Hart, 'Rawls on Liberty and its Priority', in Reading Rawls, ed. Norman Daniels (Blackwell 1975), pp. 247-8. 4See for exampleJohn C. Harsanyi, 'Morality and the theory of rational behaviour', in Utilitarianism and Beyond, ed. Amartya Sen and Bernard Williams (Cambridge University Press, 1982), pp. 39-62. This content downloaded from 213.55.225.15 on Sat, 15 Apr 2023 14:04:42 UTC All use subject to https://about.jstor.org/terms