Decision Making
R. J. Nelson*
Environmental ethics tends to be dominated by the idea that the right environ-
mental actions require a change in the value systems of many people. I argue that
the "rebirth" approach is perverse in that moral attitudes are not easily changed
by moral suasion. A properly ethical approach must begin where we are, as
moderately moral people desiring the best for all. The real ethical problem is to
develop procedures for collectively defining environmental ends that will be fair to
the parties participating in the decision process. This idea is essentially utilitarian,
and depends on the maximization of expected social utility. This type of environ-
mental ethics is contrasted with current theories of social choice in welfare
economics and with Rawls' theory of justice as fairness.
Scoby (Minneapolis: Burgess Publishing Co., 1971). Almost every paper in this collection of
essays claims that no lasting changes can be made until our ethic is revised.
263
264 ENVIRONMENTAL ETHICS Vol. J
11
If a certain strategy fails to bring about a desired end, then it would seem
good sense to abandon it. Furthermore, if there is good evidence that it will
never succeed even though the end is ethically compelling, then holding to
that strategy is in a certain sense morally wrong. For a consequentialist at any
rate (and I suspect that in environmental affairs most of us are consequen-
tialists) an act is right if it has the desired end as a consequence. The act is
surely not wrong if it intends the end but misses. But it is wrong if it always
fails to generate the end and is known to fail. A self-defeating act ofthis kind is
procedurally perverse. Yet, such acts which are known to be ineffectual are
often launched with the best of intentions. On my view, preachment and
moral suasion, whether in spiritual or mundane affairs, are quite often pro-
cedurally perverse, though not immoral. Attitudinal reform (a valued end) is
sought, but for the most part fails. This is one of the least dubitable facts of
human existence.
Those who advocate ethical reform, especially of others, seek a general or
global resolution ofthe environmental crisis: once attitudes are changed, they
say, specific local attacks can be launched using whatever resources the
reborn have at hand. An alternative strategy which is available to those of us
who are skeptical of ethical rebirth, for reasons adumbrated in the foregoing,
is to take up each problem or problem type in its particularity and deal with it
as best we can given our differing attitudes. The driving idea here is that an
environmental problem should be approached locally, not globally, and that
increased information about local conditions imparted to those directly con-
cerned will tend to induce a disposition to do what is right. 2 This is an up-to-
date version of the essentially classical ethical position that the more a person
2 Local may in same cases imply geographically regional but need not. What is intended is the
Fall 1979 ENVIRONMENTAL DECISION MAKING 265
development of a good strategy to handle cases, not the environmental problem in one fell
swoop.
3 Cf. J. S. Mill, Utilitarianism, ed. Samuel Gorovitz (New York: Bobbs-Merrill Co., 1971), p.
21. In this passage Mill explicitly includes all sentient creatures in the good.
4 M. D. Mesarovic, J. G. Klabbers, and J. M. Richardson, "An Examination of Alternative
Strategies for Control of Eutrophication in the Lake Erie Basin: Application of Regional Multi-
level Model," Modeling and Simulation, Vol. 5, Pt. 5, Proceedings ofthe 5th Annual Pittsburgh
Conference, 1974, pp. 195ff.
266 ENVIRONMENTAL ETHICS Vol. J
affected flora and fauna. This done, ifthe possible environmental states under
which we decide are risky, the decision maker can maximize expectations,
thus following quite rational procedures for the best interests of alle
Yet I say this is "unfortunate" for two reasons. (1) The more detailed and
refined the analysis of outcomes of a spectrum of alternative actions is, the
more likely is disagreement among those of us, including legislators and our
agents in regulatory offices, who have to decide what action to choose. Unlike
the business establishment there is not one decision-making person, but
many. Paradoxically, the more information the harder the choice. Indeed,
one of my central theses is that the local utilitarian, one who wishes to
improve the environment in ways best for all, always starts from a vaguely
perceived, commonly held goal which, when expressed as specific alternative
objectives, is in some sense blocked by collective disagreement; thus arises the
need for generating a collective or social set of utilities. We must emphasize
that in environmental affairs at least utilitarian goals are far from clear, and
there is little or no initial collective agreement on acts. (This idea is developed
in the following pages and receives a technical formulation in n. 10.) (2) The
harder the choice, the more likely that we will resort, after all, to suasion and
eventually to instituting programs to alter values, perceptions, and moral
attitudes. Combining (1) and (2), the greater the information the more we are
compelled, lacking some method of social decision, to persuade others to
adopt our own value schemes, and hence to lapse into procedural perversity.
These points deserve further elaboration and demonstration.
(1) If we were to take a poIl to determine attitudes toward environmental
conditions, I expect that nearly 100 percent of those polled would show a
preference for pure air over polluted, pure water over poIluted, abundant
energy over little, and the wilderness protected from the other fellow forever .
There would no doubt be a scattering of misanthropes, revolutionaries, and
twisted visionaries who would drive the percentage slightly below 100, but let
us factor them out. Most ofthem do not want to participate in a social collec-
tive to decide anything whatever, while we do, even if we disagree among
ourselves on some points.
Now everyone knows that pure water comes at a cost, and that a more
realistic and useful poIl would ask for a preferential ordering ofvarious mixes
of water at different quality levels with corresponding costs. So let us assurne
that we have available a composite quality index that aggregates concentra-
tions of carbon, phosphorus, heavy metals, coliform bacteria, and so forth.
You as poIlster now invite us to order our preferences for water at index i at
cost c. Presumably c is a trade-off function of i and there will be quite a large
number of alternatives i generated by assigning values to c. Let us say that
there are ten which are discriminably different to those of us polled, that is to
say, that differ in ways significant enough so that a preference ordering is psy-
chologicalIy feasible. Now my claim is that there will be a distribution of
Fall 1979 ENVIRONMENTAL DECISION MAKING 267
percentages across the alternative mixes, and that it is highly unlikely that a
majority will prefer some one over all others, much less a consensus even for
those of us (excluding the crazy ones) who are determined to make a reason-
able decision. If you believe this to be overly pessimistic, then add more
factors to the aggregated indices and include not only water, but air, litter,
animallife, etc., forming a multiple objective including lumped costs. And
finally, include in your considerations some measure of all the known inter-
actions and ecological trade-offs among the elements of the objective func-
tion. A good computer simulator will yield just such highly refined alterna-
tives as outcomes to a wide range of policies and actions which one might
specify.
It seems clear apriori that the more information which is accumulated and
absorbed into our considerations in the ways roughly indicated, the greater
the ground for rational individual choice (for example, by maximizing ex-
pected utility over the alternatives). At the same time, paradoxically, the
lower is the likelihood that the individual evaluations underlying each per-
son 's preference ordering of the myriads of alternatives will be compatible
enough to ensure a rational collective choice, even if we are one and all the
most benevolent and altruistic ofutilitarians. 5 The added information and in-
creased refinement of our objectives have produced a situation where the
group capacity for concerted action appears to be completely blocked.
It is worth noting that this phenomenon, which is manifested whenever we
proceed from initial broad agreement to specification of details necessary for
action, does not occur in labor disputes. Indeed, where there is initial broad
disagreement, if not conflict, refinement of objectives seems to enhance the
likelihood of settlement of some kind. We all support pure water, for
example, so long as we are sufficiently vague about costs and specific aspects
of quality. But we do not all support higher wages for labor; and, whereas
refinement of objectives inhibits social choice of water alternatives, the cor-
responding refinement of employment condition objectives (generating
"packages") always seems to improve the prospects for successful arbitra-
tion . 6 So the principle that increased information tends to stymie collective
decision can be said to apply (so far) only to environmental problems,
although a similar phenomenon appears to occur in all collective situations
where there is initial agreement on indefinite goals.
(2) The principle that increased information tends to facilitate individual
5 Of course, a lot depends on how our calculus is set up: what are the intervals of definition of
the quantitative variables, what interactions are really known and can be represented mathemati-
cally, and so forth. I suspect, however, that no matter how the details of a model vary, one from
the other, the values incompatibility among individuals will manifest itself strongly as a direct
function of informational detail.
6 This point is discussed in some detail in areport of mine to the American Arbitration Associ-
ation, "Dispute Settlement Using a Technique of Social Decision Theory," J une 1977.
268 ENVIRONMENTAL ETHICS Vol. J
111
7 I am indebted to Mortimer Kadish for the happy locution moderately moral. A more precise
pollutes a stream running through his property and eonsiders the poor state
of the water to be an "externality." Yet, in his role as a deeiding party he
honestly believes that everyone is on the whole better off if he is allowed to
eontinue his present praetiees. He disagrees with Z who also wants the best
for all; but Z thinks he ean aehieve the happy state only with better water,
even if jobs be lost. Yet they are both reasonable persons and, when it comes
down to it, might aeeept aresolution oftheir eonflieting views ifthe proeedure
for doing so were just.
Presupposing this somewhat fietional state, the ethieal situation is not one
that is pereeived to require reform of the value system of the hunter or of the
businessman or of anyone else, but rather one that demands exereise of
rational and fair proeedures for making a soeial deeision. In looking at the
situation in this way we still aeknowledge that our problems are essentially
ethieal, but in the sense that we desire ethieally correet proeedures or means
rather than the eorreet apriori pereeption of ends; for as a group we do not
know what those ends are beyond the initial ethieal position we postulated,
whieh is utilitarian. Loeal ends emerge after the soeial ehoiee meehanism has
operated on individual preferenees, and not before. A vague global utilitarian
end plus immediately feIt need motivates our deliberations, while the right
loeal end is generated by soeial ehoiee. The position is still eonsequentialist:
the right aet is the one that intends the good; but the good is what ordinary,
moderately moral persons in the deeiding group determine it to be by a eor-
reet proeedure under the guidanee of their initial utilitarian eonvietions.
I have already made the central claim ofthis paper that ethieal problems of
the environment should be handled loeally, not globally. These ends must be
determined by sonle kind of soeial ehoiee meehanism and not by ethieal
rebirth, whieh I hold is procedurally perverse. What we have to find are ethi-
eally eorreet or just procedures. I now want to put down a few remarks whieh
I hope might guide such a seareh.
As just explained, we are not simply eoneerned with how oneought to aet
in order to obtain eertain ends, but with how one ought to deeide what a
proper loeal soeial end iso More speeifieally, we assurne that eaeh agent
(member of the deeiding party) preferentially orders the eolleetion of alter-
native aetions regarding the loeal situation aeeording to both his evaluations
of the eonsequenees of eaeh poliey and the morality of the aet itself. This is
just to aeeept the fact that the moderately moral person does not always
deeide on the basis of his own interest or the general weal but also takes into
aeeount what he thinks is morally obligatory on hirn. Thus, Xmight prefer A
to Bon the grounds that A represents the best environmental state, while yet
finally ranking B over A because he thinks that the aetual produetion of the
state A might involve deeeption or broken eommitments of some kind. (I do
not mean to imply here that X is formalist, and not utilitarian even in the
restrieted loeal sense; he might take moral eommitments to be part of the
Fall 1979 ENVIRONMENTAL DECISION MAKING 271
good. The issue is not relevant here.) Supposing, then, that each agent has
ordered the alternatives, the question is, what is an ethically correct way of
combining preferences to a group or social preference ordering? 1fthis can be
done, the highest in the order is the local proper end.
The similarity of the decision situation thus hypothesized to that consid-
ered in models in welfare economics and the theory of elections is quite strik-
ing. 8 As in those cases, the problem may be stated as one of aggregating
individual preferences to a social preference ordering; however, our interest
here is not in constructing formal models meant to show the possibilities of
aggregation under certain assumptions, but rather is frankly prescriptive in a
certain metatheoretical sense. I do not want to attempt anything so ambitious
or presumptuous as prescribing an ethically correct procedure, but only to
sketch what seem to be basic principles. The kind of prescription I have in
mind is already illustrated in the assumption that each agent when ordering
alternatives takes his evaluations of the consequences of an act and his judg-
ment of the morality of the act itself both into account. My theory does not
tell hirn what that evaluation and judgment should be; it only prescribes that
both be considered. The assumption might be an accurate description ofhow
certain individuals actually do decide, but that is a question to be left to social
science. The welfare economist might not regard the outcome of social choice
(for example, the impact of collective choice on the market) as an ethical
end - there is far from wide agreement on this score - while an ethically cor-
rect decision procedure does, as already asserted: our problem is to generate
an ethically satisfactory environmental end via social choice.
Unlike the model considered by Arro\\' and those scientists directly influ-
enced by his work, environmental decision making by a group involves riskin
an essential way. In the welfare model it is assumed that decisions are made
under conditions that involve no risk. Environmental conditions, on the
other hand, are saturated with risk, namely with climatic, seasonal, eco-
nomic, and demographic circumstances which we are not able to predict with
certainty, and hence over which we are able to exercise relatively little con-
trol. Hence environmental computer analysis (which I am also hypothesiz-
ing, but which is not essential) takes into account both command or control
variables - the constituents of policy - and exoganous or environmental vari-
ables - which define various states ofthe environment to which we can assign,
at best, various probabilities of occurrence. Two examples will perhaps make
the distinction clear. In the market or in voting (two cases Arrow specifically
has in mind) the act and the consequence of an act are identical, although the
same entity is no doubt described in different ways in psychological contexts,
on one hand, and economic ones, on the other. For this reason Arrow
8 Kenneth J. Arrow, Social Choiceand Individual Values, 2nd ed. (New Haven: Yale Univer-
assurnes that the decision problem is one under certainty, not risk: there is no
range of outcomes of the act as in cases where the state of nature makes a dif-
ference, and there is no possible lapse between individual act and outcome.
On the other hand, the decision to ban phosphorus fertilizers can have radi-
cally different consequences depending on future clin1atic, flood, erosion,
etc., phenomena; hence any decision one way or the other is risky.
The two problems of social decision making, Arrow's and ours, thus are
nontrivially different, although similar in respect to the problem of aggrega-
tion. The differences deepen when we consider procedures for fashioning a
social decision out ofthe individual preference orderings. The Arrowian style
is to find, ifpossible, a function whose domain is the set ofpossible individual
orderings (an element of which will consist of a particular ordering for each
individual) and whose range is the possible social orderings. For exan1ple, if
there are three alternatives, X could prefer A to Bto C, or A to Cto B, etc., as
could likewise Y: there are six possible orderings for each. Hence, the domain
of the function has thirty-six elements for the two person group with three
alternatives. One of them might be A to B to C for X, and C to B to A for Y.
The welfare function sought allows us to map these orderings into an aggre-
gate or social ordering, and similarly for all other thirty-five possible pairs.
When the decision has to be made under risky conditions, definition ofthe
problem appears to be far more complex. A possible method for environ-
n1ental decision is to treat the problem as one of maximizing expected social
utility. In order to sketch the idea, I have to assurne here that the reader is
familiar with the procedure of solving decision problems under risk by the
method of maximizing expected utility. As abrief refresher, we suppose that
the individual is confronted with a set of alternative actions each one ofwhich
generates several outcon1es depending on various possible states of nature.
These outcomes are ordered preferentially by the individual, and he assigns a
real nurrlber to each element ofthe ordering in any way he chooses under the
single restriction that the magnitudes ofthe numbering reflect the ordering of
the alternatives. This assignment is called a utility junction. The order-
preserving character of this function assures that the considerations that led
to the original ordering - in the environmentally relevant case, subjective
evaluations of outcomes together with judgments of morality-are exactly
reflected in the associated numbering. We assurne that the several states of
nature have probabilities of being realized which total unity. The individual
now computes the sum ofthe products ofnumerical outcomes and associated
probabilities for each alternative acL The one which yields the largest sum is
the best act in the sense that it maximizes expected utility. 9
In extending this procedure to an analogous social decision method we
9 R. Duncan Luce and Howard Raiffa, Games and Decisions, lntroduction and Critical Survey
note, first, that the ordering of outcomes by the individual goes just as in the
Arrowian case with the added move of assigning numbers. It is at this point
that we face the comparable problem of aggregation. Arrow seeks a welfare
function directly on preferences as previously described, while we proceed (in
this one of several possible methods) by aggregating corresponding numbers
to yield composite evaluated outcomes. Then we nlaximize over these out-
comes using the same probabilities as in the case of the individual decision
maker. This is maximization ofexpected social utility. This process yields a
"best" -namely the alternative act (and afortiori policy) that will yield the
desired local end, depending, of course, on what state of nature actually
obtains. 10
We must now face the question ofhow such a procedure is to be secured as
ethically correct. 1 1 In section I, what I argued for in effect was the replacement
of moral suasion by a correct procedure for defining an environmental end. 12
So what we must seek is a procedure of aggregating individual utilities which
is in some sense fair. The apriori guiding idea is that everyone's preferences
should be treated on a par in the aggregating process, and that the resulting
composite evaluated outcomes must fairly represent the collective. I am going
to introduce two principles which have been recognized elsewhere as part of
the possible method of social choice I have been outlining. 13 The first is a kind
of equality principle which states that the function used to aggregate individ-
ual utilities must be symmetrical. What this means is that it makes no differ-
ence whose utilities are substituted for which independent variables in the
aggregating function - every individual 's utilities have the same significance
as every other. It is an equality principle in the sense that persons are to be
treated equally as decision agents, and not that they get equal benefits. An
altruist could in certain cases participate eqUallY in adecision in which he
came out on the low end either with respect to welfare or to rights, and accept
the outcome. The second is a kind of Pareto principle which says that if at
least one individual ofthe deciding group prefers outcome A to Band the rest
10 The characterization of our agent as "utilitarian" can now be made more precise. He is a
kind of act utilitarian (or even rule utilitarian if one chooses to count metarules as rules) who
acquiesces in a method that maximizes social utility over consequences of acts that reflect both
his appraisal of ends and his judgment of the morality of the act. 1 do not claim that this is a
suitable, univer5al ethical stance, but only a plausible one toward what 1 have termed local
environmental problems. It may be adequate for many other complex situations which have
value or ethical overtones.
11 See Paul Kantor and R. J. Nelson, "Social Decision Making in the Presence of Complex
procedure to determine the right course of action, and then not follow it. 1 am assuming (I nlust
assume, because there is little choice) that the moderately moral, utilitarian agent acts according
to the consequences, which is no less than what is taken for granted in any consequentialist
ethical theory.
13 Kantor and Nelson, "Social Decision Making."
274 ENVIRONMENTAL ETHICS Vol.l
of the group is indifferent, then the aggregating function must rank A over
B. 14 There are other conceivable principles which could be used, but these will
suffice here as an illustration of a possible methode
If all of the individual utility functions assign real numbers to the out-
comes, then the only function satisfying the equality principle and the Pareto
principle jointly is the simple sum of utilities. 15 The utilities of each person
assigned to alternative aare added (subject to constraints to be discussed in a
moment); likewise those assigned to b, and so forth. Then the alternative act
corresponding to the highest sum is best. This is the core of the method of
maximizing expected social utility.
In sketching out this method as a possible method for environmental ends I
have deliberately sidestepped two quite technical matters concerning any
method of aggregation, the simple sum or any other. These matters are
among the stickiest in decision theory, and cannot be adequately expressed
much less treated in a short paper which is intended to present an outline of a
certain ethical stance. So I shall only indicate the nature of the questions and a
possible way of handling them leaving their justification (such as it is) to the
references. Both questions have to do with the comparability of the prefer-
ences of individuals.
First, it is necessary that the utilities of each individual lie on a cardinal
scale, which means roughly that certain mathematical operations on the
numbers be meaningful. 16 Second, the utilities of all individuals must be on
the same scale-if they were not, addition would be as meaningful here as
adding two lengths, one given in centimeters and one in inches. The first
requirement is supposed to have been met: it is covered in one plausible way
in the utility theory of Von Neumann and Morgenstern. 17 The other require-
ment calls for some slight comment. Presumably, if utilities of a pair of indi-
viduals are on the same cardinal scale, then it does make sense to say that their
utilities are comparable - that if one sum of utilities is greater than the other,
the first is preferable to the pair of individuals as a pair. For example, if X
rates A four, and B five, and if Y rates A three, and B one, then the composite
evaluated outcomes are seven and six, and we say that as a social group X and
Y prefer A to B. This is reasonable because Y's preference of A over B is
greater than X's of B over A (three minus one is greater than five minus four).
This argument will have to suffice to show that the sum is a reasonable way of
aggregating the utilities to the composite: it is Pareto fair and it represents
differences in degrees of preference.
14 For a good discussion of this type of principle see A. K. Sen, Collective Choice and Social
Welfare (San Francisco: Holden Day, lnc., 1970), pp. 21ff.
15 See Kantor and Nelson, "Social Decision Making." Clearly these conditions are not suffi-
To get the utilities of a group of individuals onto the same scale presents us
with a final problem which has a number of plausible solutions. We suppose
that two alternatives, say A and B, are chosen and assigned the same utilities
for all decision agents. For example, in a pollution problem an outcome en-
tailing complete eutrophication of the lake (the lake becomes a swamp) and
one costing a sum of money beyond the means of any specified economy
might each be assigned extremely divergent utilities which are fixed for all
agents. When these are assigned, the individual scales can then be trans-
formed (normalized) to the same scale by solving systems of equations.
Transformed scales of each individual still represent his preferences since, if
the individual were now to maximize his expected utilities, he would obtain
exactly the same result as by using his original untransformed utilities. It was
in aiming at this result that we required that the individual 's utilities be on a
cardinal scale. Unfortunately this method requires that the two fixed points
be selected either by prior agreement among the agents or by a dictator , a
regulation, a law, or whatever. This is certainly not wholely satisfactory,
although there is reason to believe that extreme situations can often be iso-
lated in local environmental problems, as indicated in the example, although
not in nlore general, global circumstances. 18
The final step of the procedure, which calls for computing expected utilities
in terms of the composite evaluated outcomes, presents us with one more
problem of considerable moment. The assignment ofprobabilities to states of
nature, which is required in decision making under risk, presents theoretical
problems that 1 doubt have been seriously considered. The standard theory,
which goes back to Ramsey, 19 permits one to obtain subjective probabilities
after first determining outcome utilities, using a certain chance mechanism.
The probabilities that result, however, are wed to each individual, not to the
collective of agents. The sense in which subjective probability so determined
would have any significance at all for a group is far from clear, although the
method can be applied to the composite preference ordering underlying the
composite evaluations. A way to avoid the problem is by a method, still based
on the equality and Pareto principles, that evaluates acts (not outcomes)
directly taking into account possible outcomes as much as possible (a kind of
procedure under uncertainty, rather than risk) using whatever information is
available, and then to sum over the utilities of each act. The highest is the best
social act, although the end is uncertain. 20 Such a procedure is still, 1 claim,
18 Other methods compel each individual to assign the real number one to his most preferred
and zero to his least preferred alternative. Cf. Sen, Collective Choice. It is not very clear that the
procedure affords comparison of utilities as does our fixed point method.
19 F. P. Ramsey, "Truth and Probability," The Foundations 0/ Mathematics (New York:
Adecision procedure under uncertainty is one in wh ich there are several unpredictable states of
nature to be reckoned with; probabilities cannot be meaningfully assigned in any way.
276 ENVIRONMENTAL ETHICS Val. 1
an ethically correct one as it enables each decision agent to register his evalu-
ations of consequences and of the moral quality of the act itself, and follows
fair principles of obtaining the best social acL
This completes my discussion of ethically correct procedures for the envi-
ronment with an indication along the way of some of the problems that face
any n1ethod of social choice, illustrated in terms of one possible method and
its difficulties. I am less concerned to plump for that particular method than I
am to suggest a territory that might appropriately be called environmental
ethics. Once again, the heart of it is that environmental ethics should not
proceed through suasion or rebirth from someone's correct ethical percep-
tion of environmental ends (whether yours or mine), but from a fair method
of collective choice of specific objectives. For the remaining discussion, in
which I shall return to a further comparison with welfare economics, it will be
useful to have a summary ofthe method ofmaximizing expected social utility:
1. Each decision agent in the face of a local problem preferentially orders
the outcomes or consequences that flow from corresponding actions and poli-
cies. This almost certainly entails quite sophisticated modeling and computer
simulation to generate the alternatives.
2. Each agent assigns real numbers to the elements of this ordering. It is
assumed that these assignments fall on a cardinal scale.
3. All of the individual scales are normalized.
4. Corresponding utilities are summed. This step generates composite
social evaluations. The summing procedure satisfies the fairness principles of
equality and a Pareto condition.
5. Assuming probabilities can be associated with each state of nature, the
expected utilities are computed. The maximum is the best social act (and a
fortiori, policy).
IV
There are several misunderstandings that could arise concerning this pos-
sible method, which I should like to deal with very briefly. (1) Such a proce-
dure will not work in view of Arrow's "Impossibility Theorem." (2) The
procedure is not utilitarian since it rests on a Rawls-Pareto type of difference
principle. 21 (3) It too is procedurally perverse, after all.
(1) Arrow has shown that collective decisions often cannot be rendered.
Specifically his problem, as reviewed above, is to find, if possible, a way of
combining individual preferences to form group preferences. When partici-
pants in a collective decision order alternative ends in a manner that reflects
21 I do not intend to blur distinctions. A Pareto principle says for society to choose A over B if
at least one person prefers it and all others either prefer it or are indifferent. Thus Rawls' indiffer-
ence principle in effect allows A over B even if someone other than X prefers B to A provided that
Xis the least advantaged. I am using "Pareto-type" as a cover for both.
Fall 1979 ENV1RONMENTAL DEC1S10N MAK1NG 277
22 See Arrow, Social Choice. The conditions state that all possible orderings of preferences
must be admissible (domain condition); that if every individual prefers A to Bor is indifferent
between them, so must society (a Pareto condition); collected choice over a set of alternatives
must depend on the ordering ofindividuals only and not on irrelevant orderings (irrelevance con-
dition); collective choice must not be dictatorial.
23 Cf. Sen, Collective Choice, p. 124.
24 lohn Rawls, The Theory o/lustice (Cambridge: Harvard University Press, 1971), p. 302.
25 Ibid., p. 12.
278 ENVIRONMENTAL ETHICS Vol. J
26 This kind of distinction was clarified for me by Mortimer Kadish, although the formulation
here is mine.