An Interactive View of Rationality Russell L. Ackoff The Journal of the Operational Research Society, Vol. 34, No. 8, Systems in O.R. First, Intemational IFORS and O.R.S. Meeting. Discussion Conference at Henly, U.K. 9-11 May 1983 (Aug., 1983), 719-722. Stable URL: hhup//links,jstor-org/sic?sici=0 160-5682%28198308%2934%3A8%3C719%3A AIVOR%3E2.0,CO" 3B2.W The Journal of the Operational Research Society is currently published by Operational Research Society. ‘Your use of the ISTOR archive indicates your acceptance of JSTOR’s Terms and Conditions of Use, available at hhup:/www.jstororg/about/terms.huml. JSTOR’s Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at butp:/svww jstor.org/journals/ors.huml Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission, ISTOR is an independent not-for-profit organization dedicated to creating and preserving a digital archive of scholarly journals. For more information regarding JSTOR, please contact support@jstor.org. hupslwww jstor.org/ Wed Aug 25 13:51:18 2004An Interactive View of Rationality RUSSELL L. ACKOFF University of Pennsylvania Rationality isa judgment made by one or more persons of one or more others (subjects). I involves the elationship between the researcher's and the subject's models of the really fed hy the subjects) For this reason. the attnbution of irationalty by one to another is offen a consequence of ‘mismatching models. Therefore, tis suggested thal researchers are beter of attributing rationality {o others even i this implies their own irrationality) an trying ta uncover the models) of the others because i leads to better understanding of them and an increased billy 10 serve them, IN FARLIER PaPERS.'? I argued that abjectivity is not a property of individual researchers but is applicable only to scientists taken collectively, to science as @ purposeful system. I also wrote that complete objectivity is a scientific ideal, hence can never be attained but ‘can be approached indefinitely. In this paper I treat rationality ina similar but not identical way. | argue that rationality is not a property of an individual's or collective decision- maker's action, but of the interaction of two or more purposeful systems. My justification for submitting such a treatment to an issue of this Journal that is dedicated to the role of systems thinking in O.R. derives from my criticism of O.R.? for its inclination to exclude significant interactions between systems from consideration. Preoccupation with such interactions is essential to “the systems approach”.” A choice is usually said to be rational if it optimizes some measure of performance that is taken to be appropriate to the situation; for example, maximizing expected value, or minimizing the maximum possible loss or regret. A choice is said to be irrational to the extent that its performance deviates from the ‘best’ choice available, Therefore, the attribution of irrationality requires a comparison of measures of performance associated with possible choices. Such measurement, whether precise or not, requires the use of an explicit or implicit model of the choice situation, The model so used must identify the choices available, the possible outcomes, the probability that each choice will produce each ‘outcome and the relative value of each outcome. Now, when a researcher attributes rationality or irrationality to a subject, he does so using /his model of the subject’s choice situation, not the subject’s model. Obviously, the researcher's and the subject’s models may differ, as may their criteria of choice and the values they assign to variables that are common to their models. Then, which model, criteria and set of variable values should be used in determining rationality? There are other questions raised by the conventional concept of rationality. For ‘example, it follows from this concept that in a choice situation in which all modelled choices yield exactly the same outcome, choice should be arbitrary or random. However, when apparently rational individuals are placed in such situations, many display strong preferences. For example, the value of shoes for walking and protection of the feet is independent of their colour. Nevertheless, most people, when confronted by a choice among shoes that are identical except for colour, display strong preferences. Are they irrational? If we insist that they are, we call into question the concept of rationality rather than the rationality of the subjects ‘One can work out of this dificulty by saying that the subjects attribute a value to colour, and adding this value to the model of their choice situation. Doing so converts their behaviour, previously viewed as irrational, to rational, If we accept this way out, we can always add to, delete from, ot otherwise modify any choice model so as to convert apparently irrational behaviour to rational. Then, if all behaviour can be modelled so as to seem rational, of what use is the distinction between rational and irrational? Moreover, does not such model manipulation rationalize apparently irrational behaviour rather than convert it into rational behaviour? Consider the following two cases in point. n9Journal of the Operational Research Society, Vol. 34, No. 8 In the 1980s emissaries from public and private institutions in the United States went to India to try to induce natives to reduce the rate of growth of their country’s population by the use of contraceptives, which were provided at no cost. They failed and attributed their failure to the irrationality of the Indians. Many abandoned the effort. What alternative did they have? Balakrishnan and Camp* looked into the alleged irrationality of the Indians and found them to be rational relative to their model of their choice situation. The average number of children Indian couples had was the minimum required to assure their survival once they became unemployable. Children were the only type of social security available to most Indians. This ‘discovery’ suggested a completely new approach to population control: that retirement benefits be provided to retired persons and that they be inversely proportional to the number of sons. Consider a laboratory example, Game theorists have told us that the rational choice of both players (4 and B) ina Prisoner's Dilemma (Figure 1) isthe so-called “defecting move’, D. The joint play, D-D, is the ‘solution’ to the game, the ‘equilibrium point’, because if either player unilaterally deviates from it, he will receive a lower payoff, This combination of plays maximizes the minimum gain and minimizes the maximum regret of each player. Nevertheless, in many reported plays of this game under controlled conditions (for example, Rapoport’ and Emshoff"), more players separately and jointly selected the cooperative move, C, than D. Are we to attribute irrationality to them? The fact is they earned more than they would have if they had both selected D. Post-game interviews with some of the players in such experiments have revealed that they were trying to maximize either their joint gain or the cost to the experimenter. They id this despite their inability to communicate with each other. Indeed, in some cases they did not even know who the other player was. These players had used a different criterion of choice from that which some game theorists do. Is this irrational? Nigel Howard” has shown that, in @ model of the Prisoner's Dilemma that differs from the one shown in Figure 1, the joint play C-C is as rational as D-D even if the game theorists’ criterion of choice is used. He argued that from Player 4’s point of view, Player B has four possible strategies: C/C, C/D, D/C and DD, where the first letter designates ‘what B will play if he expects 4 to select C, and the second what he will play if he expects A to play D. When A models B’s choices in this way, he sees himself as having sixteen alternative strategies. These and the associated payoffs are shown in Figure 2. Howard calls this the “'metagame”. He found this model of the Prisoner's Dilemma to have more predictive and prescriptive power than the conventional model. Finally, to cite two ‘extreme’ cases, the eminent psychiatrist Ronald Laing* reported that, when he learned how schizophrenics modelled reality, he found their behaviour rational with respect to that model. This enabled him to develop a new approach to the treatment of this ailment, one in which the analyst entered the world of the schizophrenic rather than drawing the schizophrenic into hi George Calhoun’ showed that the ‘disease model’ of alcoholism, which is used extensively by those attempting to treat this condition, attributes no choice to the alcoholic. Using evidence to the contrary, he constructed a model of alcoholism as an intended solution to problems of stress. This enabled him to make more sense of what is known about alcoholism and to propose a new approach to its treatment: reduction of the stress that ‘causes’ it. This proposal is now under experimental evaluation Player B oD Choices © [3 Player 4 > Fic. L.A pavoff matrix of the Prisoner's Dilemma (the first mumber each ells the payoff to A, the second wo B) nr 4122