Chapter 2

Bioethics
Utilitarianism

vs.

Bioethics is a recent phenomenon. It is an attempt to develop institutions that help people make difficult decisions about health care, health
research, and other research and applications in biology, such as genetic
engineering of crops. It draws on moral philosophy, law, and some of its
own traditions. These traditions take the form of documents and
principles that grew out of the Nuremberg trials after World War II. In
some cases, these documents have acquired the force of law, but in most
cases the principles are extralegalthat is, enforced by general agreement. Bioethics now has its own journals and degree programs. You
can make a living (with difficulty) as a bioethicist.
As a field, bioethics plays three roles somewhat analogous to
medicine. It is an applied discipline. People trained in bioethics work in
hospitals and other medical settings, sometimes for pay, to help staff and
patients think about difficult decisions. I was, for several years, a
member of the Ethics Committee of the Hospital of the University of
Pennsylva- nia. This committee had no full-time paid staff but it did
draw on vari- ous administrative resources of the hospital. We had
monthly meetings that consisted of case presentationsusually reports
of cases that were settledand sometimes more general presentations.
Members of the committee were on call for quickly arranged consults,
in which four or five members would meet with medical staff and
patient representa- tives (and rarely the patient herself) over some
difficult decision. The most common involved pulling the plug, but
other decisions involved

10

Chapter 2

matters such as whether to turn a patient out of the hospital, given that
she was no longer in need of hospital services but still unable to live on
her own because of a combination of incompetence, poverty, and lack of
others who might help her. Some members of this committee had taken
courses in bioethics, but most had no special training. Nonetheless, the
committee saw its work as informed by the tradition of applied
bioethics as I discuss it here.
The second role is academic. As I noted, bioethics has its own journals and societies, and some universities have departments of bioethics,
or parts of departments devoted to it. I have almost nothing to say
about the academic side of bioethics. The literature is huge, and it
would take me too far afield. Some contributors to this literature would
agree with things that I say here; others would dispute them. Many are
utilitarians (although few of these are involved with decision analysis).
There is no consensus. The consensus arises when the bioethics
literature is distilled into its final common paththat is, the actual
influence of bioethical dis- cussion on outcomes, and that will be my
focus.
The third role is in the formulation of codes of ethics. Here the situation is somewhat unique, since most codes of ethics are written by the
practitioners to whom the codes apply. When it comes to the ethics of
research, in particular, a certain tension arises between researchers and
those nonresearchers who write the rules. The tension plays out largely
in the review boards that examine research proposals (chapter 7).

2.1

History: Nuremberg and Tuskegee

The history of bioethics follows a pattern that may explain many

human rules, both social and personal. Some catastrophe happens, then
people look for some rule that would have prevented the catastrophe if
the rule had been in effect. Then they implement the rule. The problem
is that they do not think much about whether the rule may also prevent
other things that would not be so bad in its absence, or whether the
rule will prevent the next catastrophe.
Modern bioethics began largely with the discovery of Nazi abuses
committed in the name of medical research prior to and during World
War II. Prisoners in concentration campswho were imprisoned
largely

because of their ethnic background or their political viewswere subjected to horrendous and harmful procedures. They had no choice, but
of course their options had already been sharply limited by their imprisonment. (Similar studies went on with conscientious objectors in
the United Statessuch as the starvation studies of Keys and Keys
but most of these subjects had more options.)
These abuses were uncovered at the Nuremberg war-crimes trials
af- ter Germanys defeat (19451946). The court, as part of its verdict
against several physicians involved in the experiments, proposed a set
of prin- ciples (Trials of War Criminals . . . 1949), which never acquired
the force of law, but which has been the basis of all subsequent codes.
The first principle of the Nuremberg Code was:
The voluntary consent of the human subject is absolutely essential. This means that the person involved should have
legal capacity to give consent; should be so situated as to
be able to exercise free power of choice, without the intervention of any element of force, fraud, deceit, duress, overreaching, or other ulterior form of constraint or coercion;
and should have sufficient knowledge and comprehension
of the elements of the subject matter involved as to enable
him to make an understanding and enlightened decision.
This latter element requires that before the acceptance of an
affirmative decision by the experimental subject there
should be made known to him the nature, duration, and
purpose of the ex- periment; the method and means by
which it is to be con- ducted; all inconveniences and hazards
reasonable to be ex- pected; and the effects upon his health
or person which may possibly come from his participation
in the experiment.
Although this principle sounds reasonable, its subsequent
interpreta- tion has led to rules that have made things worse, such as
the rules that prevented infants from being subjects in the experiment
that killed Jesse Gelsinger, and a three-year moratorium on much
emergency research in the United States (section 6.4).
The Nuremberg Code inspired later codes, in particular the Declaration of Helsinki of 1964 (World Medical Organization 1996) and the Belmont Report (National Commission for the Protection of Human Sub-

jects 1979). The last of these was in part a response to the Tuskegee
study of syphilis, in which 600 black men, 400 of whom had syphilis,
were monitored from 1932 to 1972, without treatment, to observe the
natural course of this lifelong, debilitating disease, even though an
effec- tive treatment (penicillin) became available in the 1950s. The
experiment ended as a result of press coverage. In 1974, partly as a
result of this coverage, the United States Congress passed the National
Research Act, which, among other things, appointed a commission to
produce a report. The resulting Belmont Report has been the basis of
United States policy, although, like the Nuremberg Code, it never
acquired the force of law (National Commission for the Protection of
Human Subjects 1979). The 1974 act also mandated Institutional
Review Boards (IRBs) for review- ing human subjects research in
institutions that received United States government funds.1
Many national governments, and the United Nations, have established formal bodies to deal with bioethical questions. Some were created because of some immediate concern but then went on to address
other issues. For example, in the United States, President George W.
Bush created the Presidents Council on Bioethics largely to help him
decide what to do about stem cell research, but the Council has now
produced several reports dealing with other issues, such as the use of
biotechnology to increase happiness. The United Nations Educational,
Scientific and Cultural Organization (UNESCO) has a bioethics section
that coordinates the activities of member states, most of which have
their own bioethics advisory committees or agencies. In the United
States, in- dividual states have bioethics panels. Universities run degree
programs in bioethics, and their graduates are employed in hospitals,
government agencies, and professional societies.

2.2

Principles of bioethics

Most introductions to bioethics provide an overview of various philosophical approaches to morality but then conclude with some sort of
list of basic principles. Typical are those listed in the Belmont Report
(Na1

Fairchild and Bayer (1999) discuss critically the extensive use of Tuskegee as a source
of analogies in bioethics.

tional Commission for the Protection of Human Subjects 1979): respect

for persons, beneficence (or nonmaleficence), and justice. 2 Other lists of
principles are often provided, but these three seem to capture a great
deal of what might be called the standard approach, even though the
Belmont Report applies them only to questions about research.
Most of these principles arose in moral philosophy. For example, the
idea of respect for persons was emphasized by Kant, beneficence by the
utilitarians, and justice by Aristotle (and many others). But they were
incorporated into bioethics often in response to specific violations, as
in the case of the Nazi crimes. The response is of the form, Can we
find a principle that this particular crime or abuse violates? Thus,
most of the abuses violated respect for persons because the subjects did
not freely consent (in the case of the Nazis) or were not informed (as in
the Tuskegee study); they also violated nonmaleficence because harm
was done.
The principles in question are surely good ideas when other considerations are equal. In real cases, they conflict with each other and sometimes even with themselves. I present them to illustrate how this might
happen. Decision analysis is designed to resolve such conflicts by quantifying the relevant considerations: outcomes and their probabilities.

2.2.1 Respect for persons

Respect for persons incorporates at least two ethical convictions: first,
that individuals should be treated as autonomous agents, and second,
that persons with diminished autonomy are entitled to protection. The
principle of respect for persons thus divides into two separate moral requirements: the requirement to acknowledge autonomy and the
require- ment to protect those with diminished autonomy. We might
call these principles autonomy and paternalism, respectively. The
principle of au- tonomy implies that people should be able to make
choices for them- selves, after being fully informed.
2

The quotations that follow are from the Belmont Report, Part B, Basic ethical principles.

Clearly, these two principles conflict. People should free to decide,

but the freedom can be overridden if their capacity is diminished. But
they cannot be coerced either. Thus, the working out of these principles
requires various subsidiary principles for deciding who is diminished
and who is free from coercion. For example, children below a certain
age or demented adults are assumed to have diminished capacity.
Prisoners are sometimes considered to be in a coercive situation
because they do not have the freedom to voluntarily consent to a
research study, therefore they must not be given the choice of whether
to consent.
Further efforts are required to define capacity when it isnt clear
whe- ther people have it. Does psychological depression count as
diminished capacity? What about mild schizophrenia (or severe
schizophrenia, for that matter)?

2.2.2Beneficence
Persons are treated in an ethical manner not only by respecting their
decisions and protecting them from harm, but also by making efforts
to secure their well-being. . . . Two general rules have been formulated
as complementary expressions of beneficent actions in this sense: (1) do
not harm and (2) maximize possible benefits and minimize possible
harms (Belmont Report, Part B, National Commission for the
Protection of Hu- man Subjects 1979).
The do no harm maxim, originally from the Hippocratic oath, is
often interpreted as meaning that one should not injure one person regardless of the benefits that might come to others, but the Belmont Report argues that it is acceptable to increase the risk of harm to someone
in order to help someone else (e.g., in research). Even the law considers
an increase in the risk of harm as a harm in itself. The report seems to
be attempting to balance risk of harm and risk of benefit, but it does
this in a crude way.
The principle of beneficence also creates conflict: A difficult ethical
problem remains, for example, about research that presents more than
minimal risk without immediate prospect of direct benefit to the
children involved. Some have argued that such research is inadmissible,
while others have pointed out that this limit would rule out much
research promising great benefit to children in the future.

2.2.3 Justice
An injustice occurs when some benefit to which a person is entitled is
denied without good reason or when some burden is imposed unduly.
Another way of conceiving the principle of justice is that equals ought
to be treated equally. As the Belmont Report (Part B, National Commission for the Protection of Human Subjects 1979) notes, this
statement begs the questions of what benefits or burdens are due or
of what the measure of equality is (contribution, need, effort, merit, and
so forth).
In practice, considerations of justice come up when, for example, research is done on poor subjects, who bear the risk, while the benefits of
the research often accrue to rich patients who can pay for the resulting
new technology. Thus, justice in practice is often a way of limiting further harm to those who are already down and out.
These three principles generate their own internal conflicts, but they
also conflict with each other. Justice may demand punishment, but beneficence may demand mercy in the same case.

2.3 Bioethical
utilitarianism

principles

vs.

The situation in applied bioethics is much like that in the law. We have
superordinate principles either in the common law or a constitution.
These principles are then made more specific by legislatures and
courts, and more specific still by case law. Applied bioethics does not
seem to have an analogue of case law.
More broadly, the same sort of cognitive processes are involved in
the development of religious rules. Some case arises, judgments are
made by applying known principles. These are weighed against each
other somehow. Ultimately a new precedent is set. If a similar case uses
the precedent, then a new, subordinate principle emerges (Hare 1952,
ch. 4). Much of the academic literature concerns the reconciliation of
prin- ciples, but I shall deal with practice. In practicefor example,
when review boards decide cases or when government agencies write
regula- tionsthe process is more intuitive. When principles conflict,
people differ in their belief about which principle dominates. People
may be especially attached to one principle or another, or chance
factors may

bring one property of a case or another to the front of each judges

mind. People may even consider consequences in the case at hand,
weighing one consequence against another. The principle they then
formulate namely, When we have this dilemma, horn A takes
precedence over horn Bmay create worse consequences when the
same dilemma arises in a different case.
In some cases, people consider more than one example before they
at- tempt to state a general rule. For example, when stating the do-noharm rule, the authors of the Belmont Report clearly considered the
possibility that it might be good to harm some a little in order to help
others a lot. They attempted to draw the line by saying that this was
acceptable if the harm was uncertain. (The benefit of research is
typically uncertain.)
Yet, uncertain harm to each of many research subjects amounts to
almost certain harm to at least one subject. Does it matter that the identity of this person is unknown? Is it more acceptable to shoot someone
if the shooter does not know the victims name? Psychologically, the
identified-victim effect is salient (Jenni and Loewenstein 1997), but this
is easily understood in terms of the way we process information, which
need not correspond either to moral rightness or best consequences
(even assuming they arent the same).
Principles of this sort do have one thing going for them, just as the
law itself does: it is better to have some method for resolving disputes
than none, and it is better if people can predict what judgments will be
made about their behavior than if they cannot. Clear, public, principles
are a lot better than nothing, even if the principles themselves are arbitrary.
A utilitarian approach can do what the law doesthat is, provide a
largely (but not perfectly) predictable way of settling difficult cases. It
also is likely to produce better consequences. To put it simply, suppose
that a utilitarian analysis says one thing and the application of bioethical principles, in the way usually done, says another. For example,
suppose the issue is whether unconscious patients coming into an
emer- gency room could be subjects for testing a life-saving procedure,
given that they cannot consent. A utilitarian analysistaking into
account the probability that the procedure is effective, the (low but
nonzero) proba- bility that a person would not consent to it if given the
chance, and the degree of resulting harm or benefit (utility) from the
use of the procedure

and the violation of the subjects wishessays that the

procedure should be tested, but an application of the rules (let
us suppose) says it cannot be tested. The utilitarian analysis is
by definition consistent with what each person would want,
without knowing who he would be, but taking into account the
probability that he would benefit from the treatment and the
probability that he would object to the study. Thus, behind this
sort of veil of ignorance, the utilitarian conclusion yields the
best for everyone. Thus, a situation could arise in which the
overriding of a utilitarian conclusion makes everyone worse
off. This is a simplistic example, but it illustrates a more
general problem (Kaplow and Shavell 2002). The problem is,
when we are in a situation like this, why should we make
everyone worse off?
In more typical situations, we are viewing the situation ex
post, after we know who is whom, so we are trading off the
benefits/harms to some against the benefits/harms to others.
How can we justify preventing a large harm to some people,
for example, just because doing so will avoid a smaller harm
to others? What can you tell the ones who would lose out? At
least, in the utilitarian outcome, you can say, I understand
that you will lose here, but someone else will lose more if
you dont. That is at least something to say. This is not a
decisive argument for utilitarianism, of course. But it does
point out the natural question that arises when any rule or
principle goes against the utilitarian conclusion. The second
principle of beneficence, that we should maximize ben- efits
and minimize harms, comes close to utilitarianism itself. The
dif- ference is that utilitarianism holds that benefit and
harm are rela- tive terms, which can be understood as
upward or downward changes, respectively, on the same
continuum of good (i.e., goodness or utility).
Thus, harms and benefits can be compared with each other just
as easily as benefits can be compared with benefits and harms
with harms.
The first principle, do no harm, is more difficult to justify,
except for two things. First, most applications of this principle
allow us to weasel out of the implication that harms of action
are never justified by avoid- ing much greater harms of
omission. The Belmont Report, for example, accepts the idea
that the risk of harm is acceptable, even though a risk
becomes a certainty.
Second, the modified principle (in which risks are allowed)
might be a reasonable guide to action even though there are
many cases in which

it leads to the worse outcome. It could be seen as a kind of warning

against overconfidence, such as might be exhibited by surgeons who
de- ceive their patients into undergoing an experimental procedure that
they deceive themselves into thinking will save many lives. Hare (1981)
ar- gues that such intuitive principles are often a good guide to action
in everyday cases. Thus, to be a good utilitarian is sometimes to go
against our own judgment of what would produce the best outcome,
because we know that we must correct that judgment for systematic
biases such as overconfidence.
I tend to think that this sort of argumentmade extensively by
Sidg- wick (1962)is too often used by utilitarian writers to justify their
own conservative intuitions. Yet, whether this conservative argument is
cor- rect or not in a given case, my point here is that it serves as a
utilitarian justification of the do no harm part of the principle of
beneficence. And my general argument here is that all of the basic
principles used in applied bioethics can be understood as having
utilitarian roots.
Consider now the competing principles of autonomy and paternalism, both part of respect for persons. Utilitarianism not only tells us
why each of these is a good idea when other things are equal, but it
also tells us how to think about the conflict between them. Autonomy
is good because it allows people who are well informed and rational to
make choices that maximize their own utility. In particular, each person
tends to know her own valuesher own utility functionbetter than
others know it.
But some paternalism, at least in the form of expert advice, is
justified when the person cannot get the relevant information about the
relation between decision options and the utility of outcomes. Experts
can also help by setting default options that are best for most people
(Camerer et al. 2003; Sunstein and Thaler 2003). More extreme
paternalism is war- ranted when the person will not follow good expert
advice and when the effect of such advice is great. The utilitarian
criterion here is whether we can expect the person (and anyone else
affected) to be better off making her own judgment or having it made
for her.
One other consideration is education. We sometimes allow people
to make their own decisions even when we think they will not choose
optimally because we want them to learn from experience. We must
balance their future utility against their present disutility. This requires

judgment, and that fact means that utilitarianism does not dictate the
answer mechanically. But at least it tells us what judgment we ought to
try to make.
Finally, consider justice. Its concern for the worse-off is rooted in
util- itarianism combined with the declining marginal utility of goods.
Most goods have less utility, the more of them we have. In the case of
money the reason for this is simple. As we gain in a certain period of
time, we rationally spend the money first on the things that are most
important. Food before fashion. Thus, the first $100 gives us more
utility than the last $100. Other goods show the same decline because
of satiation. The first apple eaten is better than the fifth. There are
exceptions to this rule; some goodssuch as peanuts, perhapscreate
increased desire for the same goods up to a point. But these are
interesting only because they are rare, and they do not affect the
general conclusion about the declining utility of money.
The principle of declining marginal utility by itself would prescribe
equality in the allocation of money and other goods. If we start from
equality and increase the wealth of one person at the expense of another, the gainer would gain less than the loser loses, in terms of what
the money can buy. On the other side, if all money were distributed
equally, then money could no longer provide an incentive to work. The
amount of productive work would be reduced. Thus, the other side of
justice is roughly the idea of equity as distinct from equality, that is, the
idea (as stated by Aristotle in the Nichomachean Ethics and by many
others since) that reward should be in proportion to contribution. The
precise formula of proportionality is not necessarily the optimal
incentive, but it is surely a reasonable approximation. Ideally, to
maximize utility across all peo- ple, some sort of compromise between
these two principles is needed, as is discussed at length in the theory of
optimal taxation. For example, one common proposal is to tax income
at a fixed percentage after subtracting some minimum amount.
Although the idea of equality and the idea of equity are both rooted
in utilitarian theory, people may apply these ideas even when the
utilitar- ian justification is absent. Greene and Baron (2001), for
example, found that people want to distribute utility itself in the same
way they would distribute money. They preferred distributions that
were more equal. The subjects judgments were internally inconsistent,
because the sub-

20

Chapter 2

jects themselves judged that the utility of money was marginally declining, hence they should have taken this decline into account, if only a
little, in their judgments of distributions of utility.
On the other side, Baron and Ritov (1993) found that people wanted
to penalize companies for causing harm even when the size of the
penalty would not affect compensation to the victim and when the
penalty would provide no incentive for anyone to change behavior
(because the penalty would be secret and the company that did the
harm is going out of busi- ness anyway). The idea of penalties for
causing harm is an application of the equity principle to the cases of
losses. Possibly the same sort of result would be found for gains.
In sum, the basic principles of traditional ethics look a lot like
heuris- tics designed to be rough guides to utility maximizationin
other words, rules of thumb. When their utilitarian justification is not
understood, they take on a life of their own, so that they are applied
even in cases when the fail to maximize utility.

2.4 Rules and flowcharts

Many of the applications of bioethics result in rules that people are
sup- posed to follow. The rules may concern when it is appropriate to
waive informed consent to research, when to use a placebo control,
when to ac- cede to a patients wish to die or forgo treatment, and so
on. The rules are typically stated in a format that is supposed to be easy
to implement that is, a series of questions, each question yielding one
of a small set of answers so that they may be put into a flowchart.
Figure 2.1 shows a good example, one section of the rules that the
University of Pennsyl- vania uses to decide whether a research proposal
is exempt from formal review. (The term exempt is something of a
euphemism, since a mem- ber of the IRB must review and approve the
request to be exempt from review.)
This is a good example because the decisions at each step are fairly
clear. (But what exactly is an educational survey? an observation of
pub- lic behavior?) The use of rules in this form becomes more difficult
when some of the categories require judgment about where the case
falls on some continuum. The law is full of such examples, such as the
standards

Bi
oe
th
ic
s
vs
.
Ut
ili
ta
ri

21

Figure 2.1: Partial flowchart for exemption from IRB review.

22

Chapter 2

relevant to various kinds of legal cases: beyond a reasonable doubt,

clear and convincing evidence, preponderance of the evidence,
arbitrary and capricious, and so on. In the case of beyond a reasonable
doubt, for example, more certainty is surely better than less, but how
much is re- quired to meet that criterion? The criterion is vague. In the
rules that result from bioethics, criteria are often stated in terms of risk
and benefit, for example: high risk, minimal risk, substantial chance to
benefit, and so on. The definition of minimal risk used at the University
of Pennsylva- nia is the probability and magnitude of harm or
discomfort anticipated in the research are not greater than [sic] in and
of themselves than those ordinarily encountered in daily life or during
the performance of routine physical and psychological examinations or
tests. The problem with this definition is that what is routine for one
person is high risk for an- other. Do we consider this? If so, minimal
risk for someone who lives in a high-crime neighborhood is much
higher than that for a suburbanite.
To take an example of where flowcharts can lead to trouble, consider
a simple rule such as: The study can be done in the way proposed if
the risk is minimal and there is a substantial chance of benefit for the
sub- jects. We could imagine a flowchart with two questions, one about
the risk and one about the benefit. Both criteria would have to be met.
Now suppose that we quantify risk as the probability of some
particular bad event, such as a side effect from an experimental drug,
and we quantify benefit as the probability of a good event, such as a
cure of a serious dis- ease. Suppose that our judge defines minimal risk
as less than 5%, and substantial chance of benefit as more than 20%.
This means she would approve a study with 4% risk and 21% chance of
benefit, but disapprove a study with 5% risk and 90% chance of benefit.
In contrast, decision analysis would quantify the utilities of the two
events. If the disutility of the bad event were, say, four times the utility
of the good event, then we would want to approve the study when the
probability of the bad event is less than 25% of the probability of the
good event. Thus, we would just approve it if the probabilities were,
respectively, 5% and 21%, or 20% and 81%, or 0.1% and 0.5%.
It seems arbitrary to set the risk criterion in the same place
regardless of the potential benefit. The trouble with categorical rules is
that they seem to require this. In real life, people fudge. They let their
criterion for minimal risk be influenced by other things they know.
So the rule

Bioethics vs. Utilitarianism

23

does not help. In fact it hinders, because some judges will want to
follow it literally. Thus, the attempt to mechanize judgment fails. People
must fall back on trading off quantitative attributes even while trying to
follow rules that prohibit such trade-offs.
The utilitarian alternative I advocate is to make the trade-offs and
judgmentswhen they are involvedexplicit. Given that judgments
are involved anyway, the introduction of judgment is no loss. It is an
explicit acknowledgment of what happens anyway. Making the process
explicit, in the form of decision analysis, may lead to better judgments.
At the very least, explication serves the purpose of describing more
accurately how decisions are made, for the benefit of those affected by
them. In the next chapter, I begin the discussion of the basis of decision
analysis.
The effort to make the trade-offs explicit need not always involve
numbers. It could simply involve bearing in mind a rule that permits
them. For example, weigh the probability and magnitude of harm
against the benefits. Such a rule, although vague, would be as easy to
apply as the (equally vague) rules now in use, but it would also focus
judges on the issues that matter most in terms of consequences.
It may seem that decision analysis involves false precision. It is true,
of course, that when we judge a probability to be .43, we might actually be happy with a judgment of .38 or .50. Most numerical judgments
are soft, and some are softer than others. However, most of the time
the results of analysis will not depend on which reasonable numbers
we choose. When they do, the decision is truly a close one, and we
cannot go far wrong with either choice. True precision is not required.

2.5

Conclusion

In this chapter, I have tried to contrast the utilitarian approach I will

take with the tradition arising from bioethics as it is practiced. Bioethical practice is concerned with the application of principles. When this
practice leads to outcomes that are worse on the whole than could be
achieved some other way, we are led to wonder whether we could find
a way to achieve better outcomes consistently. Some may say that bad
outcomes are just the price of morality, but what is this morality that
allows us to make things worse for someone else?

24

Chapter 2

A more serious question is whether, perhaps, the principles lead to

better outcomes than the direct attempt to get good outcomes. This is
an issue I shall address in later chapters. But the general answer is that
we can incorporate this possibility into decision analysis itself, as a
correc- tion for the possibility of error in the analysis.
The next chapter will go further into the analysis of decisions.