"LTN

2o

ON RATIONAL DECISION MAKING

C. WEST CHURCHMAN
University of Berkeley

I Reprinted
Vol.
Management Technology
No. 1962
Printed in
Reprinted Management Technology
Vol. No. 1962
Printed in

ON RATIONAL DECISION MAKING 1

»
C. WEST CHURCHMAN

University of Berkeley

1. What is rational behavior?

The occasion of this meeting marks one more incident in peaceful international
existence. The very spirit of the meeting itself, in which intellectuals and man-
agers of different nations meet to discuss their mutual problems, is a sign of our
times. We of Canada and the USA have learned to develop sound international
relationships just as we are learning to establish sound relationships between
science and management.
Management science and the profession of operations research are both based
on the general supposition that sound relationships between different parties in a
decision making situation do exist and can be found by diligent search and re-
search. We follow the pathway of a great historical precedent, which generally
goes under the name of rationalism. The precedent says that sound relationships
between different parties in a decision making situation can be established by
means of reason. It goes on to say that reason is something that all men share,
and that when men come to understand clearly, they inevitably will decide in
the same ways.
It is worthwhile exploring the concept of reason now and again, because as we
learn more about our world, we learn to define reason in better ways. In societies
with powerful ruling classes it was easy to define reason. Reason was the set of
principles that kept the ruling class in power, much as reason in any patriarchal
household is the principle that "father knows best." But as exploitative ruler-
ship dwindles in a society, it becomes more difficult to say exactly what reason
is supposed to be. We sense that its meaning changes as we learn more from
experience, and therefore it is important to reassess its meaning from time to
time.
Suppose we look at three different attempts to define reason. The first is in
♦ terms of a set of fundamental precepts that are invariant over all behavior. The
second is more modest, and simply says that reason is understanding man. The
third tries to take a quite different approach, in which the intellectual sacrifices
»
some of his value in order to get along with the problem.
1Delivered before the joint meeting of the Canadian Operational Research Society
and The Institute of Management Toronto, May, 1962.
71
72 CHURCHMAN

It is important to discern the spirit of this discussion, which is frankly intel-

lectual. The intellectual is the fellow who wants to understand objectively. He's
very proud of the concept of objectivity and willing to defend it at all costs. He
certainly tolerates managers and others who make non-objective decisions, be-
cause he sees that often they can't do much else. But in the end he'd like to see
all decisions made objectively, just as he'd like to have us all reach an objective
understanding of our environment.
Hence, the purpose of our exploration is to arrive at an objective basis for
resolving conflicts between parties, and more generally, an objective basis for
reaching correct decisions.
*
2. First answer: rational behavior is based on fixed
principles of conduct
Exploration number one says that reason is a set of basic principles of conduct.
It means by this that all conduct can be examined and classified under the head of
either the rational or the non-rational. The classification, furthermore, never
changes—i.e., rationality is invariant with respect to time, place or situation.
Three examples will suffice to illustrate this meaning of rationality. An ancient
example is a set of moral precepts such as one finds in the Ten Commandments.
"Thou Shalt Not Lie" is a precept that clearly classifies all behavior of a certain
type as irrational. The Ten Commandments are ten necessary conditions for
determining whether conduct is rational. Such precepts are still with us today.
They tell us how decent citizens ought not to cheat on their income taxes, ought
not to fix prices in collusion, ought not to entertain Communists.
However, the intellectual has trouble with these precepts. He doesn't object to
people holding strong views on this or that aspect of conduct. But he can't under-
stand how the precepts can be justified objectively. He finds, for example, that
cheating is perfectly all right in some societies (even our own in some situations),
that price fixing is standard operating practice in some places, and that Com-
munists like to entertain Communists.
Hence, the cautious intellectual tries to find a more satisfactory set of fixed
principles. This has led him in the past to find some "reason" for all the moral
precepts that men in different societies hold so dear. Thus he came up with a
marvelous way to explain thedivergencies of belief. All rational men seek to max-
imize their "happiness." This, he said, is the basic precept of all conduct.
This second example of a moral precept sounds very plausible, but it's forever
running the risk of toppling into an utterly vague platitude. What is happiness,
after all? If we're not careful, we'll end up by saying that it's what all men seek to
maximize. This is the reason that economists found it more satisfactoryto modify
the precept to read "all men seek to maximize profits." But then the precept lost >
its moral tone altogether. A prophet uninterested in profits could hardly be ex-
pected to comply with the command; the more holy a man, the less likely would
he be to assert that the economist's precept was rational. *
The third example is a very important one in contemporary thinking. It is the
attempt to find rules of rational behavior in situations when the conflict between
ON RATIONAL DECISION MAKING 73

the parties is governed by rule. This is the so-called "game-theory" approach to

rationality. In a constant sum two-person game, neither party has any "right"
to expect more than the "value" of the game to him. He may get more if his
opponent acts stupidly, but he must admit that a min-max strategy is therational
choice of both players. It is, in fact, "fair."
Having found a criterion of rationality that seems to work in a simple situation
(the two-person, constant-sum game), game theorists are anxious to find similar
criteria in more complicated and "realistic" situations. For many of us, it is
doubtful whether Nash equilibrium points and the like are rational strategies,
M not because some "better" strategy clearly exists, but because the whole approach
seems defective. The defects occur at the very outset, in the so-called simple two-
person, constant-sum games. The rational strategy in this case is rational only
because it must be by definition. Given that there are only two players, given the
exact pay-offs, given that the result of any game is constant, given that the pay-
offs represent the rational objectives of the players, it follows that minimax is the
rational strategy. But if one player wants to lose, or to give the other something,
or wants only to be ahead, shall we say that he is thereforeirrational? The answer,
as everyone knows, is rather subtle. In effect, game theory does insist that ra-
tional players follow a rational structure in their choices, but not in theirutilities.
In other words, it makes good sense to say that if choice A is preferred to B and
B to C, then A will be preferred to C by a rational person. But to many game
theorists, it makes no sense to assert that it is "rational" to prefer A to B. For
example, it makes no sense, they say, to assert that it is "rational" to prefer
peace to war, or honesty to dishonesty. Hence, minimax strategies at best deter-
mine a set of means given the ends.
On these grounds, it is simply a mistake to think that game theory, or much of
so-called decision theory, is an analysis of rational behavior. The work in these
fields is undoubtedly very important, but it has very little to do with our learn-
ing more about rationality. This is because the problem of rationality is not to
define rules of behavior, given the goals, but rather to define rational goals. To
relegate rationality to the study of means only is to trivialize it. It is to lose the
whole traditional spirit of the concept of rational behavior to say that a man may
"rationally"murder his friends in cold blood, as long as he structures his choices
according to "rational" rules.

3. Second answer: rational behavior is based on the

evolution of Nature
Suppose we say that the search for rationality in a set of fixed moral precepts
'.
fails. Hence, we try a second exploration, also based on a long history of thought.
This approach argues that Nature is in some way fundamentally rational, and
that reason itself is an evolving concept. The most primitive amoeba show a
skeleton of rationality in their methods of nutrition and reproduction. Higher
living forms may display quite elaborate rationality in their struggle to survive.
So man, if he survives, will develop more and more elaborate and satisfactory
concepts of rationality. The earliest rational goals were survival in any form.
74
CHURCHMAN

Then came comfortable survival, so that it was irrational to getwet or cold when
one didn't have to. Then came intellectual survival, so that it was irrational not
to understand when one could understand.
This approach is very appealing on a number of grounds. It is more modest
than the moral precept. It doesn't say, "We have the final answer to rationality,"
but rather "This is rational as we see it today." It allows for a constant re-exami-
nation of the goals of man. Furthermore, in principle it permits objectivity, for
it allows us to ask an empirical question : how is man evolving? If we can under-
stand his evolution, we can understand what his rational goals really are. For
example, if democracy is at a more advanced evolutionary stage of man's social ..
development than totalitarianism, then we can say that the advocates of totali-
tarianism are irrational in today's world.
This theory of rationality made a lot of sense to the biologists of the nineteenth
and even the twentieth century. It also makes sense to today's evolutionary
industrial theorists. Modern industry began with very crude machinery, crudely
operated. After awhile men learned how to build better machines, but they
neglected the living standards of the worker. After awhile they were forced to
recognize a worker'sclaims, but theycouldn't figure out how to use him efficiently.
Along came industrial engineering, and efficiency went up. Along came automa-
tion, and it went up even more. Along came operations research, and even greater
refinements were introduced. At each stage we redeveloped our notion of rational
industrialization. Today we don't hesitate to say that a management that ignores
worker rights or uses old methods of manufacturing is "backward." We think
it is backward because it comes earlier in the evolutionary phases ofindustrializa-
tion. Those of us that are honest about it expect that we will look backward to
the industrial theorist of two decades from now. God knows how irrational our
methods may appear to be to the inhabitants of the twenty-third century.
It's all very happy thinking, this evolutionary concept of rationality. But it
also has much of the feeling of naivete about it. At times it seems to be saying
that any change is a good change, even if automation leads many citizens into
economic disaster, even if technology destroys individual creativeness, even if
science blows us all to our doom. The next stage of industrial evolution may be
1984, and therefore 1984 is rational! Get on with it at all costs; if change is pos-
sible, do it !
If we want to be honest, we have to admit that people are mean, arrogant, and
downright evil. Worst of all, they are stupid. They don't listen to good advice.
They don't want the other fellow to put anything over on them. They all want
to be politicians; big, important politicians who make the important decisions.
Some of the more shrewd want to be big scientists who will really make the im-
portant decisions. It can't be my decision if it's made in accordance with a strict
plan of development. It can't be very much at all if everyoneaccepts it as rational.
There is some comfort, it is true, in placing the responsibility for rationality
squarely in the hands of collective mankind, rather than in the individual con-
science. In a large world view, I can forget the crudities of my behavior and that
of my neighbor, because in the long sweep of things these crudities mean nothing.
ON RATIONAL 75

Furthermore, the evolutionary theory seems to support an objective theory of

rationality, because it is at least conceivable that some day we will have as satis-
factory a theory of social evolution as we have today of biological evolution. But
the theory doesn't answer our most pressing problem: is the way in which man
evolves the rational way? We could say "Yes, by definition," so that no matter
what man will come to be, what he will be will define our rational goals. But this
is an intolerable reply. Men fight to preserve what they take to be rational goals:
freedom, love, beauty, knowledge. If they fail, shall we say they therefore fought
stupidly?
Not quite. There must be a difference between what men are or will be, and
what they ought to be. The lulling comfort of evolutionary ethics is a delusion.
Now one sensible thought is to say that men don't always turn out to be the
people they want to be. Whatever history man may follow, is indeed irrelevant to
the problem of rational goals. But what of the history man wanted to occur?
Could we say that rationality is the foundation of man's deepest hopes and
fondest dreams?
Perhaps we could say this if we knew how to. I said at the outset that there
were threeexplorations we might make in search of rationality. Thefirst — thefixed
moral precept—is impractical. The second— the evolution of man's social state—is
naive. What is the third?
4. Third answer: rational conduct has a universal function
The third is based on an earlier remark I let go by without much comment.
I said that the intellectual loves objectivity most of all. I could have said, just as
well, that he takes objective knowledge to be a supremely rational goal. In a way,
this is our contemporary paradox. Our strong emphasis on a positivist philosophy
of science makes us say that science cannot determine the rational goals of man.
But at the same time the very same philosophy says that objective knowledge
itself is a rational goal of scientists. How did it come to know this?
Now at the end, I can come back to the management sciences and operations
research, to obtain just a hint of the solution of this paradox. Every operations
researcher knows in his heart that the principle theme of his work is compromise :
the compromise between objective knowledge and action. No project ever re-
sults in a system that works as the model says it ought to. It is simply a mistake
to picture the relationship between the researcher and the manager to be one in
whichresearch discovers what ought to be done and "convinces" the manager to
do it. Or that the manager "convinces" the researcher to keep quiet. Successful
operations research, we often say, is the active cooperation of manager and re-
searcher, in which each plays a necessary role in the development. The manager's
i rational objective is control; the researcher's is knowledge. Somehow, the two
sacred goals of each become less important in the successful marriage of managers
and researchers.
v I said I would only end on a hint. I don't like an evolutionary theory of rational
goals, because if matters go on in thefuture as they have in the past, the end of
man's evolution may be the most disastrously irrational state of affairs that is
76 CHURCHMAN

possible. But our concept of management science is that man can work on his
own evolution. He can work very hard on it, if he chooses to do so. One part of
this work consists of the use of research to develop better systems. But the sys-
tems are only better if managers are also involved.
The hint is this: the rational goals of man are those states that would evolve if
manager and scientist were to work together in bringing about change. "Work-
ing together" is an overworked phrase these days. But we have begun to evolve a
very special meaning of cooperation in the management sciences. This can be
expressed most succinctly as follows: management "works with" science when it
discovers how science can become a way of managing. When I say that science
can become a way of managing, I don't imply automation or any other form of
mechanical decision making, because none of these is science. Science is the
creative and systematic discovery of knowledge. In operations research we are
learning how science can be integrated into an organization in such a manner that
it acts as a management function. Operations research is the process of looking
at science as a management function.
The theme being developed here is this: a science that can only be conceived
as a discoverer of knowledge or a satisfier of intellectual curiosity is less rational
than a science that can be conceived as managing as well.
By the same token, we should be able to look at management as a scientific
function. This, indeed, is the manner in which research and development is
evolving in our times. We can no longer think of science as individual behavior;
it has clearly become a managed enterprise. As principles of the management of
science evolve, we can expect that what we have hitherto called "scientific
method" will become the management of science.
This is the hint as to the meaning of rationality: a social institution becomes
rational to the extent that it can be considered to function like some other in-
stitution. The evolution of the rationality of law will include the development of
law as a social science. The evolution of the rationality of politics will include
the development of politics as an educational system.
In other words, it is impossible to determine the rationality of conduct in
terms of one framework alone, as the "fixed principle" theory demands. Nor is
rational conduct simply a development along certain lines, as the evolutionary
theory suggests. The test of the rationality of an institution, or a company, or a
person, is the determination of the manner in which X functions as V, whatever
V may be.
In sum, I'm trying to say that a scientific study of behavior without sound
management can never determine the rationality of the behavior, just as a man-
agement activity without science can never become rational management.
f