Objections are frequently raised to current decision-making theory on the grounds that almost exclusive attention is given to the systematic and “logical” aspects of the process without taking sufficient account of the large role that is played in these matters by human intuition and emotion. The dispute about “logical” versus “intuitive” decision-making goes back many years, antedating the first edition of Administrative Behavior. One of its starting points was a well-known essay by Chester Barnard on “Mind in Everyday Affairs,” which was published in 1938 as an appendix to his Functions of the Executive. More recently, the supposed neglect of intuition in the “logical” approach attributed to the decisionmaking framework has been the object of strong criticism by, among other writers, Chris Argyris and Henry Mintzberg.48
1. Barnard’s Thesis of Non-logical Decision Processes
The Barnard essay will provide a good starting point for our own discussion. Its central motif was a contrast between what Barnard called “logical” and “non- logical” processes for making decisions.
By “logical processes” I mean conscious thinking which could be expressed in words or by other symbols, that is, reasoning. By “non-log- ical processes” I mean those not capable of being expressed in words or as reasoning, which are only made known by a judgment, decision or action.
Barnard’s thesis is that executives, as contrasted, say, with scientists, do not often enjoy the luxury of making their decisions on the basis of orderly rational analysis, but depend largely on intuitive or judgmental responses to decision-demanding situations. No neglect of the non-logi- cal or intuitive here! Although Barnard did not provide a set of formal criteria for distinguishing between logical and judgmental decision-making, he did provide a characterization of the two styles that makes them easily recognizable, at least in their more extreme forms. In “logical” decision- making, goals and alternatives are made explicit, the consequences of pursuing different alternatives are calculated, and these consequences are evaluated in terms of how close they are to the goals.
In “judgmental” decision-making, the response to the need for a decision is usually too rapid to permit an orderly sequential analysis of the situation, and the decision-maker cannot usually give a valid account of either the process by which the decision was reached or the grounds for judging it correct. Nevertheless, decision-makers may have great confidence in the correctness of their intuitive decisions and are likely to attribute their ability to make them rapidly to their experience.
Most executives probably find Barnard’s account of their decision processes persuasive; it captures their own feelings of how the processes work. On the other hand, some students of management, especially those whose goal is to improve management decision processes, have felt less comfortable with it. It appears to vindicate snap judgments and to cast doubt on the relevance of management science tools, which almost all involve deliberation and calculation.
Barnard did not regard the non-logical processes of decision as magical in any sense. On the contrary, he felt they were mostly grounded in knowledge and experience:
The sources of these non-logical processes lie in physiological conditions or factors, or in the physical and social environment, mostly impressed upon us unconsciously or without conscious effort on our part. They also consist of the mass of facts, patterns, concepts, techniques, abstractions, and generally what we call formal knowledge or beliefs, which are impressed upon our minds more or less by conscious effort and
At the time I wrote Administrative Behavior (1941—42) I was troubled by Barnard’s account of intuitive judgment (see footnote 10 of Chapter III), largely because he left no clues as to what subconscious processes go on while judgments are being made. Wholly persuaded, however, that a theory of decision-making had to give an account of both conscious and subconscious processes, I finessed the issue by assuming that both these processes were essentially the same: that they draw on factual premises and value premises, and operate on them to form conclusions that become the decisions.
Because I used logic (drawing conclusions from premises) as a central metaphor to describe the decision-making process, many readers of Adminis- trative Behavior have concluded that the theory advanced here applies only to “logical” decision-making, and not to decisions that involve intuition and judgment. That was certainly not my intent. But now, the ambiguity can be resolved, because we have acquired a solid understanding of what the judgmental and intuitive processes are. I have already given a brief glimpse of them in the previous section of this commentary. I will take up the new evidence in a moment; but first, a word must be said about the “two brains” hypothesis, which argues that rational and intuitive processes are so different that they are carried out in distinct parts of the brain.
2. Split Brains and the Forms of Thought
Physiological research on “split brains”—brains in which the corpus callosum, which connects the two hemispheres of the cerebrum, has been severed—has provided encouragement to the idea of two qualitatively different kinds of decision-making—the analytical, conesponding to Barnard’s “logical,” and the intuitive or creative, corresponding to his “judgmental.” The primary evidence for this dichotomy is that (in right-handed people) the right hemisphere plays a special role in the recognition of visual patterns, and the left hemisphere in analytical processes and the use of language. Other evidence in addition to the split-brain research also suggests similar hemispheric specialization. EEG techniques, for example, can be used to measure relative activity in different parts of the brain. For most right-handed subjects, when the brain is engaged in a task involving recognition of visual pattern, activity is stronger in the right than in the left hemisphere; with more analytical tasks, the pattern is reversed.34 This specialization is supported also by more recent evidence derived with the addition of CAT scans and MRI techniques.
The more romantic versions of the split-brain doctrine extrapolate this evidence into the two polar forms of thought labeled above as analytical and creative. The analytical left hemisphere, so this story goes, carries on the humdrum, practical, everyday work of the brain, while the creative right hemisphere is responsible for those flights of imagination that produce great music, literature, art, science, and great management. The evidence for this romantic extrapolation cannot be derived from the physiological research, which has demonstrated only some measure of specialization between the hemispheres—in particular that the right hemisphere plays a special role in recognizing visual patterns (but another part of the brain in recording their locations in the visual field).
The physiological evidence does not in any way imply that either hemisphere is capable of problem-solving, decision-making, or discovery independently of the other. The real evidence for two different forms of thought is essentially that on which Barnard relied: the observation that, in everyday affairs, men and women often make competent judgments or reach reasonable decisions rapidly—without any overt indication that they have engaged in systematic reasoning, and without their being able to report the thought processes that took them to their conclusions. There is also some evidence for the very plausible hypothesis that some people, when confronted with a particular problem, make major use of intuitive processes in solving it, while other people make relatively more use of analytical processes.
For our purposes, it is the differences in behavior, not the differences in the hemispheres, that are important. Reference to the two hemispheres is a diversion that can only impede our understanding of intuitive, “non-logical” thought. The important questions for us are “What is intuition?” and “How is it accomplished?” not “In which cubic centimeters of the brain tissue does it take place?”
3. New Evidence on the Processes of Intuition
We have already seen that in recent years a great deal has been learned about the processes human beings use to solve ill-structured problems, and even to create works of art and science. This knowledge has been gained in the psychological laboratory, by observing the behavior of people who are demonstrably creative in some realm of human endeavor, and by using
underlie intuitive judgment, even though most of these processes are not within the conscious awareness of the person using them.
Cognitive science and artificial intelligence have devoted a great deal of attention to the nature of expert problem-solving and decisionmaking in professional-level tasks, in order to gain an understanding of the differences in behavior between experts and novices and to learn more about how novices can become experts. The goal of the cognitive science research has been to model the human processes; of the artificial intelligence research, to build “expert systems.” Both lines of research have greatly deepened our understanding of expertise.35
Intuition in Chess-Playing. Chess may seem remote from management, but it is a game requiring a high level of intellect and careful thought, and grand masters are normally full-time professionals who have devoted many years to acquiring their mastery. Much research has been done to discover the basis for expertise in chess and the nature of the intuitive judgments that appear to play such an important part in the game.
Chess might also seem an unlikely domain in which to study intuition. Chess-playing is thought to involve a highly analytical approach, with players working out systematically the consequences of moves and countermoves, so that a single move may take as much as a half hour’s thought, or more. But chess professionals can play simultaneous games, sometimes against as many as 50 opponents, and exhibit only a moderately lower level of skill than when playing under tournament conditions. In simultaneous play, the professional takes much less than a minute, often only a few seconds, for each move. There is no time for careful analysis.
When we ask the grand master how good moves can be found under these circumstances, we get the same answer that we get from other pro- fessionals when they are questioned about rapid decisions: It is done by “intuition,” by applying professional “judgment” to the situation. A few seconds’ glance at the position suggests a good move, although the player has no awareness of how the judgment was evoked. Even under tournament conditions, good moves usually come to a player’s mind after only a few seconds’ consideration of the board. The remainder of the analysis time is generally spent verifying that an apparently plausible move does not have a hidden weakness.
We encounter this same kind of behavior in other professional domains where intuitive judgments are usually subjected to tests of vari-ous kinds before they are actually implemented. The main exceptions are situations where the decision has to be made before a deadline or almost instantly. Of course we know that under these circumstances (as in pro- fessional chess when the allowed time is nearly exhausted), mistakes are sometimes made.
How do we account for the judgment or intuition that allows the chess grand master usually to find good moves in a few seconds? A good deal of the answer can be derived from an experiment that is easily repeated. If a grand master and a novice are shown, for five seconds, a position from an actual but unfamiliar chess game, and asked to reproduce the position, the grand master will usually do so with about 95 per cent accuracy (23 or 24 out of 25 pieces), while the novice will average about 25 per cent (5 or 6 pieces). Does this mean that chess expertise is based on superior visual imagery? No; because if we now present boards with the same pieces placed at random, the novice will still replace about 6 pieces, but the grand master only about 7. The difference lies not in imagery, but in the expert’s knowledge. For the grand master, a position from a well-played game is not a clutter of 25 pieces, but an arrangement of a half dozen familiar patterns, recognizable old f riends. On the random board there are no such patterns, only the 25 individual pieces in an unfamiliar arrangement.
The grand master’s memory holds more than a set of patterns. Associated in memory with each pattern is information about its significance— what dangers it holds, what offensive or defensive moves it suggests. Recognizing the pattern brings to the grand master’s mind at once moves that may be appropriate to the situation, and it is this recognition that enables the professional to play very strong chess at a rapid rate. Previous learning that has stored a large indexed chess encyclopedia in the expert’s head makes this performance possible. This, then, is the secret of the grand master’s intuition or judgment.
We mentioned earlier in this commentary the estimates that have been made of the number of familiar patterns in the expert’s memory— estimates in the neighborhood of 50,000. The natural language vocabularies of college graduates have been estimated to be in the range of 50,000 to 200,000 words— nearly the same range. Recognizing a word accesses our memory’s store of its meanings, in the same way that recognizing a chess pattern accesses knowledge of its chess significance.
Intuition in Computerized Expert Systems. A growing number of successful expert computer systems are capable of matching professional human that take the form of “if-then” pairs. The “if” is a set of conditions or patterns to be recognized; the “then” is a body of information associated with the “if” and evoked from memory whenever the pattern is recognized in the current situation.
In medical diagnosis, where there has been much study of both human intuition and expert systems, diagnosis systems like CADUCEUS and MYCIN consist of a large number of such if-then pairs, combined with an inference machine of modest powers. These systems are now capable of medical diagnosis at a good clinical level within their respective limited domains. Their recognition capabilities, the if-then pairs, account for their intuitive or judgmental ability; their inferencing processes account f or their analytical ability.
Medical diagnosis is just one of a number of domains for which expert systems have been built. For many years, electric motors, generators, and transformers have been designed automatically by expert systems developed by large electrical manufacturers. These computer programs have taken over from professional engineers many standard and relatively routine design tasks. They imitate rather closely the rule-of-thumb procedures that human designers have used, the result of a large stock of theoretical and practical information about electrical machinery. Recognition plays a large role in these systems. For example, examination of the customer’s specifications reminds the program of a particular class of devices, which is used as the basis for the design. Parameters for the design are then selected to meet the performance requirements of the device.
In chemistry, reaction paths for synthesizing organic molecules can be designed by expert systems. These chemical synthesis programs employ the same mixture of intuition and analysis that is used in the other expert systems, and by human experts as well. Other examples of expert systems can be cited, all of them exhibiting reasoning or analytic processes combined with processes for accessing knowledge banks with the help of recognition cues. This appears to be a universal scheme for the organization of expert systems— and of expert human problem-solving as well.
Notice that there is nothing “irrational” about intuitive or judgmental reasoning based on productions. The conditions in a production constitute a set of premises. Whenever these conditions are satisfied, the production draws the appropriate conclusion—it evokes from memory information implied by these conditions, or even initiates motor responses. A person learning to drive a car may notice a red light, be aware that a red light calls for a stop, and be aware that stopping requires the process—or inversely, how automatic the response is—may differ, but the one response is not more “logical” than the other.
Intuition in Management. It seems important to present this evidence, much of it from professional domains remote from management, because the notion has such wide currency that “intuitive” judgment has quite different properties (mostly thought to be wholly unknown) from “logical” judgment. The evidence indicates strongly that the intuitive skills of managers depend on the same kinds of mechanisms as the intuitive skills of chess masters or physicians. It would be surprising if it were otherwise. The experienced manager, too, has in his or her memory a large amount of knowledge gained from training and experience, and organized in terms of recognizable chunks and associated information.
Marinus J. Bouwman, for example, has constructed a computer pro- gram capable of detecting company problems from an examination of accounting statements.36 The program was modeled on detailed think- ing-aloud protocols of experienced financial analysts interpreting such statements, and it captures the knowledge that enables analysts to spot problems intuitively, usually at a very rapid rate. When a comparison was made between the responses of the program and the responses of expert human financial analysts, a close match was usually found.
In another study, R. Bhaskar gathered thinking-aloud protocols from business school students and experienced businessmen, who were all asked to analyze a business policy case.37 The final analyses produced by the students and the businessmen were quite similar. What discriminated most sharply between the novices and the experts was the time required to identify the key features of the case. This was done very rapidly, with the usual appearances of intuition, by the experts; it was done slowly, with much conscious and explicit analysis, by the novices.
Some Conclusions. The description, in detail, of the use of judgmental and analytical processes in expert problem-solving and decision-making deserves a high priority in the agenda of management research. However, on the basis of the research that has already been done, it appears exceedingly doubtful that there are two types of managers (at least, of good managers), one of whom relies almost exclusively on recognition ( alias, intuition), the other on analytic techniques. More likely, there is a continuum of decision-making styles involving an intimate combination of the two kinds of skill. We will likely also find that the nature of the problem to be solved will be a principal determinant of what mix will be most efficacious.
With our growing understanding of the organization of judgmental and intuitive processes, of the specific knowledge that is required to per- form particular judgmental tasks, and of the cues that evoke such knowl- edge in situations in which it is relevant, we have a powerful new tool for improving expert judgment. We can specify the knowledge and the recognition capabilities that experts in a domain need to acquire, and use these specifications for designing appropriate learning procedures.
We can also, in more and more situations, design expert systems capable of automating the expertise, or alternatively, of providing the human decision-maker with an expert computerized consultant;.: Increas- ingly, we see decision aids for managers that are highly interactive, with both knowledge and analysis shared between the human and the auto- mated components of the system. A vast research and development task of extracting and cataloging the knowledge and cues used by experts in different kinds of managerial tasks lies ahead. We have seen that in the area of management, the analysis of company financial statements is a domain where some progress has been made in constructing expert sys- tems. The areas of corporate policy and strategy are excellent candidates for early development of such systems.
What about the other aspects of executive work—very central aspects—that involve managing people? What help can we expect in improving this crucial component of the management task? We will take up an important aspect of this question in the next section.
Source: Simon Herbert A. (1997), Administrative Behavior, Free Press; Subsequent edition.