¹ I have in mind here the sorts of experiments reported in Time 13 12 1982.Google Scholar

² Urmson, J. O., ‘Saints and Heroes’, Essays in Moral Philosophy, Melden, A. I. (ed.) (Seattle: University of Washington Press, 1958), 198–216.Google Scholar

³ In discussion of ‘death with dignity’, perhaps this is precisely what is at issue: living like that is worse than not living at all.

⁴ My argument against holding out hopes of heroic medical interventions thus parallels arguments against holding out hopes of miracles: even if miraculous cures do sometimes happen, we do a disservice to those who are afflicted to hold out any real hope that each of them will be cured in this way.

⁵ ‘Standard practice’ is defined in that way, at least for purposes of defining malpractice and medical negligence—see Fletcher, George P., ‘Legal Aspects of the Decision Not to Prolong Life’, Journal of the American Medical Association 203 (1968), 65–68.CrossRefGoogle ScholarPubMed

⁶ Williams, Bernard, ‘Deciding to Believe’, Problems of the Self (Cambridge University Press, 1973), 136–151CrossRefGoogle Scholar, Pears, David, Motivated Irrationality (Oxford: Clarendon Press, 1984).Google Scholar

⁷ Sometimes, of course, hope is itself therapeutic, i.e. a patient's belief in the likely success of the treatment causes it to succeed. There, it is not only pragmatically but epistemically rational to believe the treatment will succeed, for believing will make it so.

⁸ Kahneman, Daniel, Slovic, Paul and Tversky, Amos (eds) judgment Under Uncertainty (Cambridge University Press, 1982).CrossRefGoogle Scholar

⁹ This is formalized in, e.g., Akerlof, George A. and Dickens, William T., ‘The Economic Consequences of Cognitive Dissonance’, American Economic Review 72 (1982), 305–319.Google Scholar

¹⁰ It would hardly have been good practice for doctors to have offered a lowprobability treatment where there was some higher-probability one available; and the aim of this paper is to discover what might be said against even ‘good medical practice’ in this regard, rather than merely to score easy points indicting bad practice.

¹¹ Another way of finding preferences for realistic preferences implicit in people's first-order preferences seems less promising. We might point to the undeniable psychological fact that people generally adapt to circumstances: if it is clear that they cannot have something, they usually come not to want it, or at least not to pine for it; they resign themselves perfectly well, if not perfectly happily, to doing without it. But we cannot infer from this that people have a meta-preference to learn to do without the therapy in view, for while the cure is improbable it is not impossible. In these circumstances the psychological tendency toward wishful thinking has as much claim to represent people's deeper preference ordering as does the psychological tendency to resign oneself to one's fate.

¹² We would not call them ‘heroic’ if they did not. But semantics aside, we would have no grounds for taking an action with such low chances of success unless the change it might make would be substantial. All heroic measures make a big difference, but not all that makes a big difference is necessarily ‘heroic’, in the sense of being rare or unusual. Hay fever injections for serious sufferers might be an example. The objections here lodged against heroic measures are usually waived in the case of such non-heroic measures precisely because, being not uncommon, they usually do not interrupt people's life plans in the same way as heroic measures do. But if it should happen that, despite their being not uncommon, these non-heroic measures nonetheless did somehow cause people to get stuck at a fork in their lives, then the same objections would apply.

¹³ Economists appreciate that, in an uncertain world, it is valuable to have options; and irreversible choices, because they foreclose options, carry a special penalty in those terms. See, e.g., Arrow, Kenneth J. and Fisher, Anthony C., ‘Environmental Preservation, Uncertainty and Irreversibility’, Quarterly Journal of Economics 88 (1974), 312–319CrossRefGoogle Scholar. Again, this penalty is not an infinitely large one, so it may well be right that we should often pursue irreversible courses of action despite this consideration that always militates against them. Of course, heroic measures themselves are often option-preserving strategies, on the part of doctors presented with, e.g., a comatose patient brought into the casualty ward; and once hooked up to life-preserving equipment, it may be hard to get a patient unhooked. But insofar as they can anticipate that the latter proposition is going to be true, doctors are wrong about the former position. That is, if they are going to be locked into continuing treatment, once started, then starting treatment forecloses options just as much as not starting it.

¹⁴ Nozick, Robert, Philosophical Explanations (Cambridge, Mass.: Harvard University Press, 1981), 403–450Google Scholar; Wollheim, Richard, The Thread of Life (Cambridge University Press, 1984)Google Scholar. Those arguments characteristically appeal, in the first instance, to external standards of ‘the good’, it is true. But those standards will prove sufficiently compelling in terms of a sufficiently large portion of possible life plans that most people will embrace them as internal standards for judging the success of their own lives, as well.

¹⁵ Their only alternatives are to proceed on the improbable assumption that the heroic measures will succeed (in which case they will face great costs in reversing those choices when they fail, if they can reverse them at all), or else to proceed on the more likely assumption that they will fail (in which case they will face great costs in reversing those choices should they succeed, if they can be reversed at all, and enormous regret at their choices if they cannot be reversed).

¹⁶ It varies, among other things, with the time it takes for those uncertainties to play themselves out. If there is always ‘one more possibility’ just around the corner, then people's life choices may be suspended virtually indefinitely. If there are few possibilities and quick resolutions, the interference might be slight.

¹⁷ That is to say, I see no grounds for supposing that the various factors bearing on this decision vary systematically with one another, in either direction. The only moderately plausible argument along these lines I can see for heroic interventions is to say that the less difference it will make to your life the less paralysing it will be on your future planning. But that is not to say that hedging and backtracking is more possible (or even less costly) in those cases; it is merely to say that it is less worthwhile there. The more nearly indifferent we are between the alternative paths before us, the less we mind irrevocably committing ourselves to one and forsaking the other forever.

¹⁸ More formally, heroic measures should be rejected if the opportunity costs of waiting to see how they turn out exceed the utility of success, discounted by its improbability, less the utility of the best alternative life available absent the heroic measure. Ideally, this is a calculation that rational individuals should perform for themselves, in deciding whether to seek or forswear heroic measures. But the well-established psychological dynamic of ‘responsibility avoidance’ may prevent them from acting rationally in this regard, and thus constitute yet another reason for public policies against heroic measures (see Kahneman, , Slovic, and Tversky, , op. cit., for the psychometric evidence).Google Scholar

¹⁹ ‘Good life choices available to them’ is intentionally equivocal, as between our objective and their subjective assessment of those options. There is a case, in terms of efficiency, for allocating scarce or heroic treatments to paraplegics who simply refuse to reconcile themselves to life in a wheelchair as a subjectively unacceptable alternative. But that is inequitable, in the sense of penalizing the virtuous who come to terms with their condition; and we may, for that reason, prefer to make policy on the basis of an objective assessment of how well off people are rather than responding to their subjective assessments.

²⁰ Also involved in such calculations is the consideration of just how much of an interruption to people's lives the prospect of heroic treatment might be. If they can get on with a large part of their lives, even pending resolution of the uncertainty—or if they can make contingent plans for the future, depending upon how it turns out, that do not have to be implemented until after the uncertainty has played itself out—then the costs would once again be cancelled. The two conditions in the text are more appropriate to the concerns of policymakers than this one, only because most of the people offered any given heroic intervention will usually find themselves in much the same position with respect to those matters, whereas this third consideration is much more contingent upon the particulars of each person's own life plan.

²¹ ‘Informed consent’ is certainly among them—though here, as with many medical experiments, informing people that the treatment probably will not work may well compromise the success of the experiment. Just as a scientifically sound drug trial may require giving people in a control group placebos without telling them, so too might we have scientific grounds for withholding information about the realistic chances of success with heroic interventions. How these scientific considerations are traded off for ethical ones is, of course, a notoriously difficult problem. It is not, however, one unique to the cases I am here considering.

²² I am grateful to audiences at the Universities of Stockholm and York for comments on earlier versions of this essay.