ABSTRACT
The weakest aspect of individualist theories is the fact that individuals do not think the way these theories presume they do. And, if one describes how individuals really do think, it is by no means clear how these muddy and fallible thought processes can be used to explain social events. Two of the most significant attempts to describe how people really do think and to come to terms with it are those of Simon, on the one the hand, and that of Tversky and Kahneman on the other. Although each effort points to deviations in the ways people in fact thinkor are able to think-from the ways rational choice theory presumes they think, neither theory can overcome the difficulties this causes for the construction of social scientific explanations. They both measure real thought processes against the ideal of rational choice theory and find the gap between the ideal and reality to be significant. But both conclude that the gap must be closed, rather than that the ideal must be revised. One of the most important defects of rational choice theory is the fact that how one perceives situations is a crucial part of rationality and rational choice theories cannot explain how that is done. Esser tries, but fails, to use rational choice theory to make good on this defect of rational choice theories. Simon’s bounded rationality remains utopian Herbert Simon not only pointed to false assumptions about how individuals think found in rational choice theories, but also sought an account of the way humans really do think. Such a theory, he said, had to take account of the limitations of human rationality, above all of their limited capacity to find the right information and to process enough of this information in the amount of time available, but should still explain how individuals could make correct, or any rate good, decisions in spite of these limitations. Theories of social processes, he thought, could then take account of the difficulties individuals face in making (rational) decisions. He came upon these problems as he tried to develop a general theory of organizations, one which would explain how various types of organizationsbusiness, military, governmental-worked well or poorly. He later tried to develop an alternative framework for such studies which included a psychological theory and a theory of rationality which complemented it. He wanted to explain how individuals in fact processed information and came to (rational) decisions. If he knew that, he thought, he could develop a theory of how organizations could be
structured to provide individuals with the right-or at least the best-information at the right time. In this way the organization would function well. It too would obtain and process information in limited amounts of time in effective ways. In his later work he sees a key to this theory of rationality in computer theory. Computer theory, or programming, could offer accounts of how information could be processed to produce the correct results. This would also, he thought, be at least a model for a description of human rationality. Simon’s criticisms of rational choice theories are regularly convincing and often devastating. He points out, for example, that any attempt to explain how individuals should behave rationally in competing situations will always involve an infinite regress: if I take account of some expected response of my competitor when I make a decision, he takes account of my taking account of his decision, and so forth. His observation that decisions are based on small amounts of information and/or false assumptions and that they cannot be made in accord with some comprehensive process due to limitations of time are also correct, as is his appeal to take account of institutions in the decision making process. But he nevertheless accepted too much of the traditional view of rationality he criticized so well. He still presumes that a decision procedure for reaching good decisions is possible, if sufficient information is presented properly and processed in the correct way in short enough time spans. His ideal model of rationality is the computer, which is provided with the right information at the right time. His program was to explain how it was possible to approximate this ideal. But the ideal is mistaken, because it presumes that there can be a solution to the problem of induction-he implicitly seeks one in computer theory without explaining how and why that should be possible-and because he ignores the epistemological problem already pointed out by Frege: the kind of reduction he seeks leads to idealism and then to solipsism-Frege pointed this out in regard to Wundt (Wettersten 1995a). Simon wants to be a realist just as much as Wundt did, but Simon’s program leads him away from it just as Wundt’s did. Simon is impatient with those of us who doubt that the problems which he seeks to solve by developing a program which thinks in all significant respects just as man does will not be obtained. He points out-quite correctly-that current difficulties and lacunae are no reason to presume that full success will not be obtained some time in the future. What he fails to note, however, is that his own view is just one further metaphysical view-or program-which cannot be refuted, but which, for that reason, is not necessarily very good. He ignores both the powerful criticisms which have been given of all attempts to solve the problem of induction, on the one hand, and Frege’s explanation of why a physiological theory of how knowledge is obtained is not possible, on the other. Simon’s ideal cannot be approximated Before turning to the difficulties Simon’s program encounters, it will be useful to look more closely at Simon’s ideal. We can then better see why the problems posed by criticisms of induction raise the difficulties his theory faces. Simon
argues forcefully and correctly that individuals do not and cannot attain that degree of rationality attributed to them by rational choice theories. He insists that, if social theory is to provide knowledge of human affairs, it should correctly describe how individuals think. He needs, then, an alternative theory of rationality which does that. Simon turned to psychology to find descriptions of thought processes. At first, in, for example, Administrative Behavior, he found guidance in Tolman’s Purposive Behavior in Animals and Men. Tolman studied how rats reacted to stimuli in an attempt to find laws describing the relations between stimuli and reactions. His book appealed to Simon, I presume, because he talked about how the attainment of knowledge was related to the pursuit of goals. Tolman’s book offered Simon a formulation of problems within learning theory. On Tolman’s view the central problem of learning theory consists of finding the laws which connect stimuli and responses to them. Learning is an information processing system in which knowledge consists of establishing new connections between stimuli and the appropriate responses, where appropriate responses are determined by purpose, which in the case of the rats is primarily to find food. This is a version of the traditional associationist program which saw learning as the building of associations between sensations, but it is ‘externalized’ in the sense that learning does not deal with the private associations of sensations and perceptions but with the observable associations of stimuli and responses. Following the lead of Tolman and other psychologists, Simon posed his problems of learning theory as those of the reception of information and the proper processing of it. Learning concerned, on the one hand, the amount of information and the nature of the information received and, on the other hand, the way in which connections between the reception of such information and the appropriate responses were formed. These were also interpreted as the central problems of organization theory, since the central problems here were how the organization could be rational, that is, how it could obtain and process enough of the needed information in the allotted time to make the correct decisions. The method which Tolman and his colleagues then followed was to start with simple situations and to seek to generalize on them. They were not concerned with phenomenological descriptions of thought processes, which would be reduced to some connections. In his later psychological research Simon still follows this rather primitive method in some of his research, but did not stay within its bounds. He looked instead for descriptions of how people really think and then sought to explain it. The most important result he found was Adriaan de Groot’s study of the thought processes of chess players. And this research came out of a completely different tradition in the history of psychology (Berkson and Wettersten 1982, 1984; Wettersten 1985c, 1987f, 1988, 1990, 1992a). Wundt’s student, Oswald Külpe, discovered that his teacher’s attempt to reduce all thought processes to connections between, and/or combinations of, sensations could not succeed. He introduced into scientific psychology the (Kantian) idea that thought processes involved ideas which controlled perceptions and the direction of thought processes. He proposed that one study how this happens. His student, Otto Selz, developed the theory that directed thought processes were given their
direction by tasks. Adriaan de Groot took over Selz’s theory while modifying it in order to develop a theory of thought and learning which could explain the thought processes of master chess players. The small set of mental operations which Selz thought could explain all thought process were not sufficient when applied to chess players and he had no adequate theory of how problems were formed, or in the case of chess players, of how they perceived a situation with its opportunities and dangers. Simon took over de Groot’s study of the thought of chess players as a good phenomenological description of thought processes. But he adopted a different method in psychological research than that used by Selz and de Groot. He sought, namely, to find a description at some deeper level of the thought processes which Selz and de Groot had described at a phenomenological level. In pursuing this goal, he retained the ideal for psychological theory which he had imbibed from the stimulus-response psychology and inductivist methodology. He rejected the nonreductionist, introspective and dualistic approaches of Külpe, Selz, Popper, and de Groot. He sought instead to find the direct connections between stimuli (information) and results (decisions). Such a program accepted the challenge of dualist psychologists to show how apparently dualistic mental processes could be reduced to rules governing the associations between information and conclusions or decisions or responses. If successful, it would unify the two traditions by explaining phenomena discovered by one (Selz and de Groot) with a program proposed by the other (Tolman). Simon’s results break down, however, on just those points where dualist psychologists found stimulus-response psychology inadequate. Selz and de Groot found mental acts which are not determined by the operations of rule governed connections. These mental acts are the setting of tasks and the perception of situations. Simon seeks to unite the program by viewing perceptions as direct response to stimuli, but he has no theory of how these are built. He hoped to describe such thought processes with the help of computer programs. He thought that in this way he could show the rules which governed thought processes. But he knew he had not yet succeeded and the challenge which he still faced. He said that he had no program for the formulation of problems or perceptual capacities. But such a program would be needed to complete his project of describing thought processes with computer programs. Simon faced a problem posed by thinkers such as Popper who claimed that discovery could not be fully explained. That this challenge came from Popper was no historical accident. Popper was a member of Külpe’s Würzburg school and a follower of Selz, whose psychology he translated into a methodology of science. Simon tried to move beyond Selz and Popper by showing that the creation of ideas could also be the product of the use of a program, even claiming he could show how students could repeat Kepler’s discovery of his third law. Such an act would only be possible if one could reconstruct Kepler’s context. And Simon claimed he could do this. Once it had been done, students needed only a few hours to reach the results which Kepler needed years to discover. It is difficult to take this seriously. But just here we find the same gap which Popper, Selz and de Groot discovered. The trick which Kepler accomplished was to find the right problems and the right
alternatives which are to be evaluated. And this is just what Simon could not do. In order to have done that, he would have had to solve the problem of induction; he would have had to show how we could gather the right data, formulate the right alternatives-including the true one-and identify the true one on the basis of our data. Anything less will leave a gaping hole in his theory. Simon poses the wrong questions about institutions Simon’s view of the ideal of rationality as the proper information processing system leads him astray in his analysis of institutions. He seeks to explain their foundation in the rationality of individuals, that is, as information processing systems. In his studies of organizations he asks how they can approach this ideal of rationality. But since this ideal of rationality cannot be realized, he cannot pose the problem of how organizations may approach it in a sensible way. Simon presumes that organizations should be made of programs designed to process information. He presumes that if organizations were provided with all the relevant information and enough time to process it, correct decisions would be reached. Since both the ability to obtain information and the time to process it are limited, the purpose of programs is to select information which allow for satisfactory results in a limited time. But, following the lead of de Groot, he also asks how problems are posed. This is just that area which is the least developed and which he fails to integrate with his view of programs. He thus has no idea how to incorporate it into the other aspects of his theory. Simon’s view of organizations-or any institution-is that they are tools which make decisions on the basis of the way they process information. The first set of questions he poses concern how they in fact use information to make decisions. How does the flow of information lead to the formulation of a decision? How does information flow into the process? Which processes are fast or slow? How does it use expertise? The second set of questions he poses are prescriptive: Does the right information reach the right point in sufficient quantities at the right time? Is the process too slow? Does it take advantage of expertise? The third set of descriptive questions he poses are: How are problems formulated? How are perceptions changed? How are alternatives developed? How should these tasks be carried out? The fourth set of questions is, How should these problems be solved? Simon cannot integrate his answers to these four sets of questions; he can hardly offer any responses to the questions in the last set at all. He sought some definite process for the creation of problems and alternatives but could not find one. His discussion, then, breaks down just when he gets to the most important and interesting questions. In lieu of answering them he wanders into a discussion of whether an organization should use planning or simply rely on the invisible hand in internal as well as in external operations (March and Simon 197ff).
