An analysis in the light of Habermas’s general Action theory
We assume that Habermas’s theory of action rationality offers an apt theoretical frame for a systematic comparison of two acquired rationality -conceptions in economics: the conception of consumer behavior and of the older choice and decision theory. In our opinion that frame suits as well for a critical evaluation of the two conceptions.
As point of departure for this analysis we have chosen the address of R.March to the AEA in 1978: ‘Bounded rationality, ambiguity, and the engineering of choice’ (March:1978). In this address March criticizes, on empirical and theoretical grounds, the commonly accepted concept of calculated rationality. The theoretical grounds on which March builds his argument presuppose, we believe, a theory of action rationality that outlines the relation between acting and rationality. In his text such an outline is lacking. We will show how Habermas’s writings do provide the rationality theory wanted.
As we will make clear, March’s critique can also be applied to the theory of consumer behavior. For in the two theory traditions human behavior is stripped of its action characteristics. They do not fit into the conception of science adhered to in these theory traditions.
In the first paragraph we will present in two sections the rationality conceptions of the two theory traditions we have selected. In the second paragraph we introduce Habermas’s conception of action rationality. And we will sketch what his theory has to offer in terms of an analysis of a plurality of rationality aspects. The third paragraph is divided into two sections. In the first we develop the critique that from the action theoretical perspective of Habermas can be formulated concerning the theory of consumer behaviour. And in the second we give arguments why March’s critique on the rationality conception of the choice and decision theory cannot be justified without such an action theory. In conclusion of the chapter we make some general notes on the implications that, according to us, the action theoretical perspective of Habermas has for the positioning and assessment of (the concept of ) economic rationality.
1. Economists on rationality
1.1 The rationality of preferences in the theory of consumer behavior.
Robbins compares in his ‘An Essay on the Nature and Significance of Economic Science’ (Robbins:1935 ) two definitions of economic science. The first is called a material definition as it delineates, in line with the classical tradition, the class of phenomena which are the object of economic science. From this perspective economic science could be designated as the study of the causes of material wealth. Thus, economics pertains only the forms of behavior which influence the development of wealth. The second definition is determined by the concept of scarcity. And it is this definition that, in Robbins’s view, is the superior one. According to this definition the science of economics is not focused on certain forms of behavior, but it studies one specific aspect of behavior in general: the aspect of scarcity (Robbins:16-7). The scarcity concept refers to the relation between a given complex of needs and the limited, alternatively employable means for fulfilling those needs. Robbins himself does not explicitly assume that a subject has ordered the complex quantity of dissimilar needs, e.g. to the extent of urgency or preference. Though he is of the opinion that such an assumption is not contrary to his preferred definition (Robbins:15,note 1). For example, if it is assumed that subjects do construct an ordinal utility scale regarding their actual needs, then one can state that there is one need and one goal for behaviour: the maximalization of one owns utility, instead of fulfilling the many needs that function as different goals of behavior. When we add that assumption to the object definition of Robbins, his view can be called representative for the older theoretical approach of consumer behaviour.
This theory takes as its point of departure that a consumer through his behavior maximizes a given individual utlity function under for him actual price and budget restrictions. Not only is it implied that a consumer has ordered in a consistent way his preferences into a utlity function, but also that he has chosen and applied the means that serve optimally the realization of the maximal utility. These assumptions mark how in this theory rationality and rational action are understood. This view can be explicated as follows: rationality is the characteristic of behavior that fulfils in an optimal way the requirements of efficiency and effectivity.
It is this older theory of consumer behavior that P.Samuelson as well as Stigler and Becker will modernize and improve. Robbins’s definition of the science of economics in terms of scarcity is for them in itself beyond dispute. The objections that they bring forward relate to specific presuppositions that, in their opinion, are an unnecessary burden on the older theory of consumer behaviour. They want to remove from this theory all elements that in their view are not observable and cannot be objectified. In particular the way in which consumers themselves handle their preferences, reach decisions and have expectations about the utility of certain actions fall under this verdict. Such elements should not be taken into consideration in an analysis of consumer behavior. The authors mentioned will base their analyses exclusively on observations of buy transactions and of the economic variables (prices and incomes) involved.
Samuelson pleads in his ‘Note on the Pure Theory of Consumer’s Behaviour’ (Samuelson:1938) to remove from this theory all ‘introspective-psychological’ remainders. Concepts as need and utility should be done away with through transforming the theory of consumer’s behavior into a ‘revealed preference theory. This theory makes it namely possible on the basis of only one ordering postulate to deduce from repeated observations of the behavior of a consumer his preference curves. When a consumer buys in a given situation a parcel of goods A, then, according to Samuelson, allows this one postulate to deduce that the consumer concerned apparently did prefer A above any other one alternative parcel Bi that he, given the available budget, could have bought. With sufficient repetition of observations of consumer’s buying behavior the preference ordering that the consumer apparently is using , can be determined.
The thus briefly summarized view of Samuelson builds on the standpoint, that a persistently maintained observing and objectifying attitude of the economist towards consumer’s behavior is the only scientifically justified way to explore individual preference orderings. The ‘revealed preference theory’ of Samuelson provides no room for the manner in which consumers themselves think about their preferences and order them, and about the role played by their preferences in their actual behavior.The inner motives of the consumer are considered to be scientifically unreliable. The scientific method for exploring the preferences of a consumer consists, according to Samuelson, in a behaviouristic analysis of his buying behavior.
In the same sense as Samuelson, Stigler and Becker intend to improve the older theory of consumer behaviour (Stigler and Becker:1977). Their main objection is, that economists, when predictions about behavior in the future not come about, always can invoke non-economic factors that have changed the preferences of consumers. They will not accept such ad hoc adaptations. Instead they promote as research hypothesis the presupposition that the preferences of consumer households as they manifest themselves in the preference functions, are stable over time. (Lonely individuals are treated also as consumer households.) They add explicitly the presupposition that each consumer household maximizes its utility. Taking into consideration the delineation of rationality Becker has given elsewhere (Becker:1976), it is quite plausible that Stigler and Becker assume that consumer households behave rationally. For, today, according to Becker, everybody agrees more or less that behavior can called rational if it corresponds with the solution of the optimalization problem: maximize under given restrictions a well-formed function as an utility or profit function (Becker: 153).
It is quite clear that Stigler and Becker do not claim that their presupposition of stable preferences is realistic. It is also their opinion that tastes and preferences will change over time They believe, however, that the presuppositions of a theory should not be judged in terms of their being realistic, but exclusively in terms of their capacity to predict. This is also true for the other presuppositions we have mentioned. They have not to be taken literally. They have to be understood ’as if premises’: consumer households are supposed to behave as if tastes do not change and as if they maximize their utility (cf. Friedman:1966). By making the choice of the presuppositions we have been talking about Stigler and Becker have developed a theory that allows making testable predictions about future consumer behavior based on observations of economic variables only (Stigler and Becker:89). Statements about changes in the behavior of a consumer can be deduced from observable changes in prices and incomes. Following a different path than Samuelson, Stigler and Becker too eliminate from theory formation all aspects of choice behavior that are not objectifiable and observable from within a behavioristic conception of economics.
1.2 Rationality in choice and decision theory.
In his address of 1978 March offers an overview of developments in the field of choice and decision theory that have occurred in the foregoing two decades (March:1978). He focuses his analysis on the changes of the rationality concept.
As to the analysis of the economic behavior of subjects the starting point has been for a long period a conception of rationality that March characterizes as unlimited He describes it as follows. The subject was supposed to be able to gather and digest all information relevant for his choice (concerning the means, the alternatives, the action consequences and so forth), and to make on this basis the optimal choice. The subject was esteemed as ‘omniscient’. More and more often since the middle of the fifties voices from the circle of choice theoreticians were heard stating, that in the theory should be accounted for the limitations of the subject. After some time the conception of unlimited rationality became exchanged for more mitigated conceptions of rationality among which the conception of bounded rationality. For the development of this concept March points to the important contributions of J.H.Simon, especially his (Simon:1955,99-118). In the conception of bounded rationality one starts not any more from the assumption that the subject can know all relevant information. Particularly the subject cannot have all the information required to ascertain the precise action outcomes.
March considers this shift in the thinking about the rationality of choice behavior, how interesting it may be in itself, not as a fundamental change. For the conception of bounded rationality has at its base the same conception of human behavior as has the conception of unlimited rationality. In his view both conceptions are variants of the more comprehensive conception of calculated rationality. Both conceptions namely presuppose that the subject relates his actual actions consciously to the knowledge he has about his personal goals and future action outcomes (March:592).The subject will choose and execute those actions from which he expects results that satisfy the most his individual preferences.
The conception of calculated rationality starts from an implicit, and according to March unjustified, assumption. It assumes that a subject not only sees his personal preferences at a given moment as to be realized goals towards which he directs his actions, but that he also uses those preferences as measuring standard for the evaluation of action results. In March’s opinion, however, it is quite possible that the preferences of a subject change during the action, and that, due to it, a subject valuates the results of the action differently then he did before the action. With a well-chosen series of examples of everyday choice behavior March shows how subjects in different ways, that have to be characterized as intelligent, go about with developing, clarifying , refining etc. their individual preferences. In such cases, in choosing their actions, subjects take into consideration eventual developments in their preferences. This way of doing would not be rational measured against the rationality criteria inherent in the conception of calculated rationality. March opposes this view. Such ways of doing which may be called intelligent, are for him a reason to challenge the starting basis of the above outlined conception of calculated rationality. In what follows we will present four of the examples he has given to sustain his critique.
(a) March calls attention to the possibility that actions that a subject has tuned on the realization of his preferences, may turn out to have changed his individual preferences. In such cases his preferences cannot be taken to be an exogenous variable, as happens in the conception of calculated rationality. They have to be seen as an endogenous variable, that is, a variable that can change in the choice and action process itself. We think that such a case can be illustrated with the following statement of someone who is fond of, say, rhubarb: ‘Whenever I begin eating it, I can not stop overindulging it, and this spoils my taste for other dishes’. When this consumer goes shopping and makes his choice of goods, he reckons with the effect that the eating of rhubarb has on him. For this reason he might consider not to buy rhubarb, but not because his preference goes to other vegetables he actually buys.
(b) It is also possible that a subjects aims with certain actions to the refinement of his taste. In such a case the subject may act in a manner that not suits his already present preferences, but instead is intended to a change in, a refinement of, his preferences. It means that a consumer may have good reasons to choose a product of which he, on the moment of his buy, knows not at all if he likes it better then the alternatives known to him.
(c) A similar situation occurs in circumstances in which a subject by acting tries to get a better idea of his preferences. The subject expects to know more about his preferences only after the action, dependent on the experiences that go with it.
(d) Beside the possibility that subjects consciously try to get a clearer picture of their preferences, it can happen that preferences unexpectedly become clear. In the light of a certain interpretation, for example, the executed actions and their outcomes may get afterwards a different meaning. Thus a consumer who has spent a lot of money in a delicatessen, may justify it for himself : ’Altogether it is still much cheaper then going out for dinner to a restaurant’. Only after the buy he legitimizes his buying behavior.
We have presented two conceptions of rationality that have been quite common among economists. A couple of things merit attention. At the one hand the development of the theory of consumer behavior has gone in the direction of removing from theory formation all elements of behavior which are held not to be observable and not objectifiable. This has been done with an appeal upon the requirements that a modern conception of science imposes on (economic) research. It has led to the situation that the behaviour of economic subjects is deduced from a behavioral model that the researchers, on the basis of theoretical presuppositions, have construed concerning registrable aspects of behavior. Through following this path consonance has been realized between the rationality conception of the theory of consumer behaviour and the conception of calculated rationality that March speaks of. At the other hand the same author brings to the fore developments of the choice and decision theory that in his view demand to break away of a too narrowly conceived conception of rationality. For this appeal he refers to the aspects of intelligence unmistakingly present in everyday choice behavior. As we will see, March is of the opinion that a conception of science that forbids to take serious these aspects, should be challenged And we hope to have made plausible that his verdict is valid not only for the older choice and decision theory, but for the older theory of consumer behavior as well.
We believe that the general action theory of Habermas offers a framework and a level of analysis that are appropriate to handle the problems March has drawn our attention to. Therefore we will continue with a sketch of Habermas’s theory.
2.1 Habermas’ general theory of action
It is Habermas’s conviction that a pertinent explication of societal phenomena can only be given on the basis of a theoretical design. He presents such a design with his general action theory. That theory connects in a very precise manner action and rationality. The very succinct introduction thereof in this paragraph draws on (Van Doorne:1982).
An action situation has, according to Habermas, the following structural characteristics:
(1) Analytically speaking three basic action dimensions have to be distinguished:
— a dimension of objectivity
— a dimension of subjectivity
— a dimension of normativity
(2) in each action situation there are minimally two actors involved, who relate to each other in conformity with the three dimensions. Thus for each actor it is true that his involvement in an action situation is threefold:
- he is confronted with an existing or a to be realized objective world
- he is related to a subjective world, his own
- he is involved in a social world in the existence of which he himself and. the other take part. (This social world forms the minimum basis of shared norms for action coordination without which the objective and the subjective worlds of the interacting actors would not belong to the same action situation.)
We call the thus defined interrelation of actors within the context of one and the same action situation their action design (conscious or not, verbally expressed or not).
(3) A further part of the structural characteristics of an action situation is that action designs only materialize when the means thereto used are in principle comprehensible for the actors involved. In other words: the intersubjective comprehensibility of the means employed manifests the involvement of actors in one and the same action situation.
(4) Finally an action situation is characterized by the unavoidable fact that the actors mutually make, however implicitly, validity claims regarding their actions (that is, action designs plus their materialization). These validity claims have to be differentiated in conformity again with the three action dimensions:
- the validity claims of truth and effectiveness regarding the objective world
- the validity claims of truthfulness and reliability regarding the subjective world
- the validity claims of rightness and appropriateness regarding the social world.
The validitity claims can be represented schematically as is shown in the following figure.
Figure 1: validity claims from an action perspective
What are the implications of the sketched theory of action for Habermas’s conception of action rationality?
A first implication is, that whenever a situation manifests the structural characteristics (1)–(4), we face an action situation. It is the outcome of the way in which actors (more or less fully) mutually comply with the validity claims mentioned.
A second implication is, that, in accordance with Habermas’s general action theory, all actions respecting validity claims can be called rational. And rational expresses in this context, that actors by mutually taking validity claims into consideration make known that they, may need there be, are prepared and capable to give good reasons for their claims.
A third implication is, that the feature of rationality is no longer, as usual, reserved for the action dimension of objectivity, because validity claims equally play a role in the action dimensions of subjectivity and normativity.
A fourth implication is, that in Habermas’s conception the concepts of action and of rationality are intrinsically tied up with the concept of intersubjectivity. They derive their meaning from the interplay of an actor with one or more co-actors. Thus, the action quality or the rationality of actions can never be determined and judged by one of the parties involved independenly from the other.
A fifth implication of Habermas’s general theory of action is, that the above introduced structural characteristics guarantee a basic similarity of everyday and social-scientific modes of action.
2.2 Habermasian analysis of the two theory traditions
In this paragraph we will re-interpret the behavioristic analyses of the older consumer theory (Samuelson, Stigler and Becker) in the light of the general action theory of Habermas. We have explained that the authors present their behavioristic approach as an alternative for the yet older theory of consumer behaviour. Compared with this theory the newness of their approach consists in completely pass over the question how consumers themselves think about their preferences, how they handle them and which role these preferences play in the determination of their behaviour. The proposed new approach of consumer theory is designed from a purely external perspective. It is a perspective in which gets lost the distinction common to ordinary interaction between behavior and action. This distinction is also a cornerstone of Habermas’s action theory.
It is true that Stigler and Becker presuppose explicitly that a consumer maximizes with his buy behaviour a well-formed utility function. And in this sense his behavior is rational. However, it is altogether not relevant for them to know if the consumer determines indeed his buying behavior on the basis of those considerations. An empirical science as economics has to produce falsifiable predictions. If it does, then the realisticness of the presuppositions of the predicted outcome should not be a concern for the economist. ‘Rationality’ is in their approach not connected any more with what consumers have as motives and reasons for their behaviour.
Let us take an example of consumer behavior: someone buys in the foodstore during the summer season two crops of iceberg lettuce. We will analyse this behaviour first from the perspective of the behavioristic approach of Samuelson, Stigler and Becker. Then we consider what happens from the perspective of the consumer in case. Having done this we reformulate with the conceptual means of Habermas’s action theory these two perspectives.
- The consumer may have purely objective reasons for his behavior. E.g., he can give as reasons that two crops of iceberg lettuce are an important ingredient of a healthy meal; and beyond that the best quality of iceberg lettuce can be bought for the lowest price during the summer season. In terms of Habermas’s general action theory this means, that, concerning his action design and the choice of means, the consumer lays down only the claims on truth and effectivity, that is, focuses on the dimension of objectivity (I,1 and I,2 in the scheme of Figure 1).
- However, the consumer may have a very different reason for his action. He wants to express with it something about his subjective world. E.g., he tells that never before he has eaten iceberg lettuce, and that he likes to compare its taste with the taste of common lettuce. He brings to expression his personal intention to experiment with his preferences. (This case is similar to the example of March of which we have spoken above.) In Habermas’s terminology the consumer lays down claims concerning his truhfulness and reliability in the dimension of subjectivity (II,1 and II,2 of the scheme). He does not act conforming an already established preference ordering. And he would not recognize as an adequate analysis of his decision the deductions Samuelson and Stigler/Becker make on the basis of his buying behavior.
- The same would be the case if the consumer brings forward ethical-political reasons to justify his action. E.g., he could argue that in other seasons than the summer iceberg lettuce is imported from South-Africa, and that he will not support in whatever form the Apartheid regime. It is quite possible that left aside the political aftertaste, he could not tell which of the two lettuce varieties he likes better: the imported South-African or the locally grown. In the terminology of Habermas again, in his action prevails the claimson rightness and appropriateness, that is, the action dimension of normativity has the lead (III,1 and III,2).
In conclusion: the consumer could recognize himself in the analysis of Samuelson and Stigler/Becker only in the cases under a.
Samuelson could object that the consumer,in the cases mentioned under b. and c.,precisely make clear why one could not trust what consumers themselves think and say about their preferences. For it is apparent that the consumer had an actual preference for the iceberg lettuce notwithstanding what he himself thinks in those cases. In other words, what counts is not what the consumer thinks and says about his preference, but only the preferences as revealed by the choice he made. But, we repeat, Samuelsons’s position leaves open the question whether the revealed preferences this author talks about can indeed be considered to be the preferences of the consumer. The meaning of the concept ‘preference’ is reduced to the meaning of the concept ’registered choice’. Reduced, because the reasons that play a role in the process of preference formation by the consumers are no longer taken into consideration.
As to the position of Stigler/Becker one could think that only in the case under a. the rationality conditions as they have formulated them are met. Their presupposition that this forms the basis for calling consumer behaviour rational, looks quite different when judged from the viewpoint of an action theory. Seen from this perspective the researchers Stigler and Becker are the actors who determine, independently from the subjects under investigation, that only the claims on truth and effectivity are relevant. And it are the researchers who identify fulfilling the rationality conditions they have formulated, with rationality ‘tout court’. In terms of this conception of rationality buying the iceberg lettuce by the consumer based on the considerations described under b. and c. has to be marked out as non-rational. This stands against the opinion of the consumer himself who thinks that the line of action he has followed is intelligent and reasonable. It is in view of this kind of situations that March pleads in his presidential address for a conception of rationality in which there is room to take serious intelligent forms of everyday behavior.
3. Theoretical implications of March’s critique.
The objections March has formulated against the conception of calculated rationality do not entail, that, according to him, there would be no everyday behavior that comes forth from reasons and motives expressing a purely calculating attitude. He denies not at all the possibility that there are cases in which people in their behavior are led exclusively by efficiency and effectivity considerations. March refuses, however, to accept the presupposition underlying the conception of calculated rationality: that the domain of rationality does not comprehend more than those forms of behaviour which, from the viewpoint of an observer, are comprehensible in terms of effectivity and efficiency. When we characterize his position with concepts of Habermas’s general action theory, we can put it this way: March contests that only the actor’s claims on truth and effectivity (I,1 and I,2) can be honored as claims on rationality. From the examples given under a–d on p. 6-7 becomes clear his conviction that actors can be led by their preferences in other ways that have to be considered as intelligent. The cases mentioned have shown in particular claims on truthfulness. Rightly he has pointed to the fact, that normally the subjectively motivating world is a whole of needs, desires and wishes of different nature. Which means that also for the subjects themselves it hardly occurs that can be spoken of given, unambiguous goals towards which their actions are directed. This being so, it is quite possible that it can demonstrate rationality when a subject is experimenting with his subjective world. And besides that, interwoven with someone’s subjective world play ideals, principles and values an important role in determining a line of conduct. It is the normative side of our actions. According to Habermas, actors can make validity claims also in the dimension of normativity , claims that (in principle) can be sustained with reasons. (Note that it is doubtful if in the choice situations sketched by March there is any that in this sense can be interpreted.)
Further reaching than the differentiation of the concept of rationality itself as proposed by March, is the theoretical ground on which this proposal is based. The critique he has concerning the conception of calculated rationality rests on critique on the self understood conception of the decision theoreticians concerning what it is to operate scientifically. He reproaches them not to have any more an open eye for the intelligent forms of everyday behavior which he has signalized, because they have to be left aside as being non-rational when measured against the criteria of calculated rationality. They do not recognize any more the constraints that the presence of intelligence in the everyday behavior of people puts on the freedom of construction of the scientist. A theory that completely forgoes such characteristics of intelligence, falls short not only in a descriptive sense, but also in a prescriptive sense. That is why such a theory does not offer sustainement in everyday situations for meaningful changes and improvements of conduct. ‘If behaviour that apparently deviates from standard procedures of calculated rationality can be shown to be intelligent, then it can plausibly be argued that models of calculated rationality are deficient not only as descriptors of human behavior but also as guides to intelligent choice.’(March, 1998, p.593)
On the basis of the critique March has voiced, one could say that he is asking for a better theory. A theory that defines the rationality concept such that it connects to the motives and reasons by which consumers are led, or which they bring forward to justify their behavior. He is conscious of the fact that the hereto-necessary modifications would bring with it a fundamental change of the field of choice and decision theory. Summarizing his critique in our terminology, it would have to be an action theory instead of a behavior theory. It is a pity that he does not succeed in making his fundamental critique productive for the shift demanded for. The main reason for this seems to be, that he has not at his disposal the conceptual means hereto needed.
It goes without saying that nevertheless we consider March as an important exponent of thinking starting to move regarding rationality, in particular in relation with economics. Let us gather from his analysis the several requirements that have to be met by an action theoretic elaboration of the issue of rationality:
- – systematic weight should be given to an analysis of the relation between acting within an everyday setting and acting from a scientific point of view;
- – the relation between the context of everyday activity and the context of scientific activity has to be articulated in terms of a structural relation to avoid that it remains of an ad hoc and arbitrary kind;
- – elaborating that relation should have in any case two sides to it. On the one hand, a search into the meaning and the function that everyday activity has for scientific concept and theory formation. And on the other hand, a search into the meaning and function that the results of scientific activity have for activity within everyday settings.
The question we started with: offers Habermas’s general action theory an appropriate basis for a systematic comparison and a critical evaluation of current conceptions of rationality in economics? We believe that a positiv answering of this question has been justified. We sum up the reasons for this.
We have highlighted two points in the critique formulated by March: (1) on empirical grounds he objects to the current conception of rationality that it neglects the internal perspective on behaviour; (2) his view that the coherence between the internal and the external side of behavior can only be restored in the light of new theoretical assumptions regarding the concept of science itself, specifically with regard to the relation between the everyday and the scientific modes of action. Both the theory traditions we have discussed can be compared regarding these two points. In so far March’s critique can be applied to the two traditions.
We have taken for granted that the empirical critique of March is to the point. Given his competence and his authority within the discipline we do rely on his judgement in this matter. As to the second point: we have articulated the theoretical implications of his critique as the demand for a new action theoretical conception of rationality. And taking into consideration the five presuppositions of Habermas’s conception of action rationality, we consider his general action theory as a serious starting base for attempts to provide a theoretical justification of March’s position.
Is there more, beyond that, for which a new action theoretic conception of rationality can be profitable for economics? We believe that in three other ways this conception can lead to adaptations of the current conception of economic rationality:
- – It makes it possible to correct the one-sidedness of the external perspective;
- – It leads to a differentiation of the rationality concept itself, and thereby leads to the question to be answered whether economic rationality has to be conceived as a partial rationality, namely the goal rationality of the objective action dimension. Given the tradition of economic thinking this objective action dimension may be identified as action under conditions of scarcity;
- – The partial rationality of the specific economic action dimension would then, on theoretical grounds, have to be seen in coherence with the partial rationalities of the subjective and the normative action dimensions. This would entail making economic rationality a relative rationality.
We hope that attention for the problems discussed here will counteract a way of thinking of which the following statement of Arrow testifies: ‘ An economist by training thinks of himself as the guardian of rationality, the ascriber of rationality to others, and the prescriber of rationality to the social world. It is this role I will play’.