The task requires time, effort, and patience to wait well beyond the limits of one person's life. Balance of, on one hand, reflecting, discussing, and perfecting methods of decision-making and, on the other hand, of decision-making itself and action is not easy to find. Often it is even difficult to believe that market and, especially, democratic principles have anything in common with rationality. For example, Feynman (1988) describes a discussion of defects the day before the tragic launch of Space Shuttle. The committee of about twelve people could not come to agreement about the question how dangerous was an impact of low temperature on seals of hard fuel tanks. Because voices divided in half, the head of the committee made decision. Yet if instead of democratically looking for an opinion of majority, the committee paid attention to two experts on seals that were present and objected the launch, the tragedy would not happen.

Because values of rationality, of competition, and of reward for performance are considered in accord with both a market economy and a democratic political system, they are loudly praised and placed at the core of positive values of modern societies.

The main thesis of this essay is that the resulting cultural homogeneity is a major obstacle to learning. To demonstrate this point, notions of rationality and motivating impact of indirect reward systems are discussed and several distinctions are made to suggest how core values may be and need to be supplemented to promote learning. In conclusion I speculate about possible ways to create a learning-friendly environment. I will not touch upon a topic of influence of competition on learning. It is very popular now and is thoroughly discussed elsewhere (Kohn 1986).

The paper was written to stimulate a discussion of the questions that are listed below. Constructive criticism of assumptions and inferences is very welcome. But first let me define several terms, which are introduced in the study or whose meaning varies from author to author. The "definitions" do not claim to be rigorous operational procedures or logical constructs.

The questions are:

Although the essay is relevant for understanding problems with learning in educational and scientific institutions (Becker 1986, Ch. 6), it grew from contemplation about forces which promote and hinder introduction of disciplined problem solving approach (DPSA) for improving quality in manufacturing and service industries. To make the further analysis more specific, the next section describes the essence of DPSA and links it to the broader concept of learning.

DPSA is a major part of Total Quality Management, which is, in its essence, an attempt to combine productivity achieved through division of labor characteristic to a modern industrial or service organization with quality typical for an artisan (Herrington 1992). This goal can be achieved, figuratively speaking, by making separate departments and individuals collaborate as left and right hands of a craftsman.

According to the Total Quality Management, DPSA performs an important tripartite role: motivating, uniting, and directing behavior of individuals. First, DPSA helps to motivate employees by enriching their jobs. Now employees in addition to performing tasks also have tools and methods to improve their work routines. Second, it helps to resolve questions and come to consensus. Instead of being purely emotional, arguments may be supported by evidence, and there are commonly accepted decision-making procedures and terminology. Conflicting opinions are formulated as testable hypotheses and carefully collected facts help to resolve the disagreement. Third, DPSA allows to unearth problems, to analyze them emphasizing underlying processes, and as a result, to find more optimal ways of doing things.

DPSA looks like a modified trial-and-error experimentation--one of two major mechanisms of organizational learning from direct experience identified by Levitt and March (1988). They view organizations "as learning by encoding inferences from history into routines that guide behavior" and add:

The main advance of DPSA over the trial-and-error experimentation is an attempt to use quantitative techniques to compare the outcomes of experimentation with targets. DPSA also has three other distinctive features: emphasis on documenting history from which lessons are drawn, search for causes of problems, and insistence on individual as well as organizational learning. These three features are interconnected.

Search for causes permits to organize documentation, helps to separate what is relevant from what is not. Documentation allows an individual to learn about experiments conducted in past, to compare a new problem with similar problems taken in their context, and to find out which of the current beliefs were tested and under what assumptions.

Thus DPSA represents a significant change in traditional organizational learning defined by Levitt and March. Actually, it does not fit this definition anymore. Because DPSA becomes more and more widespread, it is necessarily to relax the definition of organizational learning by dropping the statement that it does not make history from which lessons are learned accessible to members of organization. From now on the following definition of organizational learning will be used in this study:

To reap the benefits and advantages of DPSA organizations have to engage into the systematic search for interrelated phenomena, which is commonly known as search for root causes, develop documentation, and create conditions for individual learning. All these activities require additional time and resources. For this reason they are often overlooked or deliberately omitted. In a short run it leads to no negative effects. At the beginning a small part of possible problems is covered by documentation. Almost all problems are encountered for the first time. With time the situation is changing. Knowledge contained in documentation accumulates and provides an insight into increasingly large number of problems. Savings of time and resources are proportional to easiness of learning from documentation as compared with learning from trial-and-error experimentation. To enable effective and continuous learning documentation must be rational. This is a crucial requirement and what it means 'rational' will be discussed at length in the next section. A process of developing such documentation will be further called a rational learning.

The rational organizational learning in contradistinction to organizational learning described by Levitt and March (1988) has one more advantage in addition to a higher efficiency in solving problems after enough knowledge is accumulate d. It allows individual and organizational learning to go hand in hand.

The situation when individual learning is lagging behind is fraught with detrimental consequences. When an organization becomes more and more complex and precisely tuned but not understood, most improvement attempts will produce numerous side-effects and, usually, will not bring the desired result. This makes organizational members feel helpless. If their needs are not satisfied, a frustration follows. If organization creates welfare of its members in some incomprehensible mystic way, they will venerate and worship it. In both cases experimentation will be inhibited and organizational learning will cease.

The current interest in DPSA was born after realization that this picture corresponds to reality in many companies. Yet if this approach is implemented within the old culture, it may divert from rational learning despite of or, more exactly, because of cultural stress on virtues of rationality. To understand how it may happen we need to analyze a notion of rational learning juxtaposing it with an intuitive one.

The difference between two approaches is recognized from time immemorial. A distinction between heart and mind, feelings and reason is being made, at least, since the argument about superiority of Truth over Good in Dialogues of Plato (Pirsig 1984). It looks that everyone knows what I am speaking about. Yet I have not found any conceptual definition and cannot come up with one. The research done on the issue (Genter and Toupin 1986; Bakken et all 1992) simply describes an experimental situation in minute details. All these are indicators of a basic phenomenon, which cannot be reduced to other terms.

The best illustration of the difference I found in Pirsig's book (1984:46-49). The author describes how he offered to repair his friend's motorcycle--"his new eighteen-hundred-dollar BMW, the pride of a half century of German mechanical finesse"--with a shim cut out of an aluminum beer can. Author is sure that "any true German mechanic, with a half-century of mechanical finesse behind him, would have concluded that this particular solution to this particular technical problem was perfect." But to the author's surprise his friend didn't see the cleverness of this at all. In fact he got "noticeably haughty" and "was actually offended". Pirsig comes up with the explanation:

On the next page Pirsig continues:

Good examples of the intuitive vs. rational dichotomy from business setting may be found in Senge's work (1990) where he employs a metaphor of seeing leaves and seeing a stream carrying these leaves. The difference between the two problem solving approaches may be clearly seen when designing an essentially new product by building its prototype and creating its mathematical model.

The prototype is a real thing painted with, let say green color, standing in the particular room of the particular building and the particular car passes down the particular street exerting some particular influence on the prototype. Prototype and its surroundings are inseparable and consist of infinite number of parts, whose interrelations and whose impact on the prototype's work are neither known nor understood. Certainly, some conditions are specified and controlled because their impact is known. But there exist an infinite number of others, which are unknown yet considered irrelevant.

A problem associated with this assumption is familiar to all engineers. Sometimes weird things start to happen with a prototype. Nothing helps. And in some time the prototype "takes care of itself," and nobody knows what has happened and will it happen again or not. This leads to superstitious beliefs in possible influence of personal traits of colleagues or to fear of anyone moving the equipment (Jordan and Lynch 1992). Prototype's complexity causes uncertainty which impedes learning.

Uncertainty is effectively reduced by building a model. One starts from selecting relevant phenomena on the basis of observations of the prototype or knowledge of how the prototype operates. The decisions about what is relevant are, usually, of yes-no type. This purposeful, conscious and systematic simplification allows us to restrict analysis to only a few components and their relationships that can be explored in depth. It means that the shape of these relationships may be derived from underlying mechanisms. The nature of operations with mathematical symbols propels this undertaking and sometimes requires it. It also requires us to assign numerical values to variables and parameters. Development of clear operational definitions becomes inevitable. In one word, the fine structure of mathematical language helps to reveal many vague assumptions about what is relevant and what is not, as well as about possible values of variables and the shape of relationships between them. Cutting a model out of context makes a researcher to explore a context instead of taking it for granted.

Explication and conscious simplification characteristic to models reduces but does not exclude influence of a context completely. When the model is numerically solved on computer, the computer becomes a model's context. Yet these artificial surroundings are better controlled. Even if the model is not executed on a computer--and there is no possibility of hardware breakdowns, software errors, and of problems caused by instability or rounding mistakes of numerical methods--as a rule a researcher will make one or several mistakes. Therefore of crucial importance become methods allowing to check for such mistakes. This and determining limits of a model's applicability, which is related to its accuracy, is at least as complex and time consuming as deter mining properties of components and their interrelationships and writing them down in a formal mathematical language.

Ideally, development of a prototype goes hand in hand with building its model. Using a prototype one can conduct highly accurate experiments, but only in a limited number of infinitely complex cases. For example, one can check influence of temperature on efficiency of air conditioning in five temperature points under impact of lighting, humidity, fluctuations of net voltage, specific air pollution, and of infinite number of other known and unknown impacts from environment. A model may account for lighting, humidity, and some kind of fluctuations of net voltage only. But it will provide us with a curve describing a relationship between temperature and efficiency for all temperature points in the range of the model's applicability. Although du e to simplifications, these results will contain some error, they allow optimization without a danger of mistakes related to interpolation between the five points measured on the prototype. Understanding of underlying processes may legitimate the interpolation. Resulting benefits may be essential in the multivariate optimization. Also the model often reveals conditions under which the prototype is prone to weird behavior described above.

Prototype and model are fundamentally complementary and this holds true for intuitive and rational modes of problem solving in general. An engineer building a prototype uses--consciously or not--mental models created by all previous experience and constantly refers to intuition and sensory information especially on early stages of designing a model. Learning is impossible without theorizing as well as without practicing (Senge 1990). But under some conditions intuitive and rational modes of problem solving may combine in the way that forms a major block to individual learning.

One of the possible origins of this problem is overemphasizing virtues of rational thinking and underestimating its difficulty. Two possible scenarios come to mind.

First, one does not pay attention to his or her intuition because of being blinded by reverence to rationality of mental models one presently has. For example, users of photocopying equipment respond to the question whether they are willing to pay additionally for nicer design with an unequivocal "no." All of them justify their decision in the same way: "We do not buy equipment for looks." When shown pictures of two machines that differ only aesthetically, they unanimously select one of them, agree to pay more, but with a strange persistence they try to make inferences about functional differences from differences in decor.

Second, when people are not sure about how mental models were acquired--how they have learned what they know now--they may silence a voice of intuition because of a fear to introduce a cognitive dissonance, which they do not know how to deal with (Festinger 1957). Even when only a little learning and re-consideration is needed, one does not know that. If the way knowledge is arranged in one's head is not systematic and does not allow a conscious search of elements which should be corrected, one will feel that all understanding built throughout one's whole life is endangered and may crumble leaving one as helpless as a newborn child.

Cramming by going many times but quickly and uncritically through the same material leads to such "untouchable" knowledge. The behavior usually stems from underestimating time and effort needed to learn rational ways of problem solving. As it was mentioned earlier, methods serving to secure against mistakes and to find limits of model's applicability are complex and time consuming. Mastering them requires understanding of how the problem solving methods were derived. That is usually left out because of lack of time. When an instructor, who finds mistakes for them, guides students and matches problems with appropriate methods, an illusion of understanding is easy to achieve.

From my experience of teaching statistics, I would distinguish several levels when students feel that they understand material:

I would say that half of students in a regular college course in statistics for not-exact science majors usually could reach the third level. It requires 14 weeks with 3 hours of in-class and the same amount of independent work--84 hours total, some previous exposure and ability to abstract reasoning, and a high-school preparation in basic mathematics which includes some algebra.

So what should happen after the standard 16-hour two-day course in statistical process control conducted to a company's employees? If they try to use the learned methods, they, most probably, will produce meaningless results. If one cares to check them against intuition, a feeling of mess and confusion will follow. Sometimes an attempt to use formal methods is abandoned. Sometimes an organizational culture forces one to persist using statistical tools. Another 16-hour course may be conducted. Employees are brought several times through the same material on a high speed--the surest way to produce "untouchable" knowledge. Now management, which also was not given enough time to realize all subtleties of just-in-timeliness of training, wants to see results of training put to work. Yet gaps in understanding of formal logic of statistics cancel its disciplining and guiding impact. People try to close these gaps with intuition. But intuition has developed when dealing with sensory information and is of small help in ordering abstract symbols. Then opinion of authorities is sought and is accepted in a hurry. The problem of distinguishing what people say and what they mean creates a fundamental difficulty for analytical separation of a model from its context (Archer 1988:127-142). The resulting shapeless conglomerate covered with patches of superstition is not amenable either to intuition or logic. People become suspicious to both of them and stick to what is authorized by superiors or to what is already known, especially if this is legitimated by "use" of formal problem solving methods. In this context a notion that rational substantiation hinders formation of consensus in negotiations and only leads to "a spiraling series of arguments" (Weingart et al 1993) becomes understandable. Individual learning becomes increasingly impeded.

Let me call this stage the "Hell road" stage, because it is well known that the road to Hell is paved with good intentions. This stage is an outcome of attempts to use knowledge for solving practical problems when it is not mature yet. Here comes a problem.

The maturation requires a close contact between theory and practice. It is guided, especially at initially stages of theory building, by striving to be practically relevant and to solve practically important questions. As far as I see it, the only criteria of success at this stage, is a capability of making realistic predictions. Yet predictions made by using immature formal theoretical methods usually will be inferior to those resulting from intuition of experienced men of practical affairs. Under conditions of institutionalized rationality it is difficult to be engaged in research and simultaneously admit that decisions are still being made on the basis of intuition and often go against research findings. So, organizations are apt to fall into one of two extremes: complete neglect of disciplined methods of inquiry or uncritical acceptance of any approach that claims to be scientific. The case of oscillation between these two extremes is not the best environment for learning, but it looks like this is almost the only opportunity for it. Enthusiasm and expectations that "scientific" method will lead us leads only to bitter disappointments and abandoning all efforts. Yet after the lesson is forgotten the cycle repeats itself. It is interesting to observe a struggle of Dr. Deming to maintain the balance. Alas, as far as I can see, his efforts lead to one more, so to say, "flip of faith". Control charts are worshiped, performance evaluations--blasphemed.

In an informal discussion one makes a point saying: "If all diagonal members of a matrix are equal to zero, it will lead to disaster." No further explanation is provided or requested.

Under conditions of institutionalized rationality the fallen emerging understanding usually claims to be a rational one. The following interrelated characteristics of the former help to distinguish between the two: reliance on word of authority, simplified patterns of thinking, emphasis upon style rather than content, use of jargon in official communication and of emotional language in informal discussions.

Because neither intuition, nor reasoning is of help for analyzing statements made on this stage, mentioning an authority is accepted as a proof of correctness. Personality and status of a proponent of an idea become more important than the idea itself. Consequently, a challenge to an idea is perceived as an attack on a person. How something is said is more important than what is said. Use of jargon and emphasis on style are two sides of the same coin--use of metaphors and superficial analogies when describing relationships. Patterns of thinking are employed also as metaphors and their intuitive appeal is more important than validity. A list of the most widespread simplifications that are rarely corroborated includes:

- use of yes-no classification instead of a continuum;

- search for a root cause instead of cycles with feedback (Senge 1990)

- confusing necessary and sufficient conditions;

- use of only finite quantities precluding comprehension of limits;

- search for one main cause instead of a number of partial causes;

- no discrimination between casual and correlational relationships;

- disregarding sample size (Kahnemann and Tversky 1974);

- assuming normality of distributions;

- restricting analysis to monotonic relationships of "the more, the more" or "the more, the less" kind;

- assuming that monotonic relationships are linear.

On the "Hell road" stage a number of rituals increases. They are not so obviously irrational as in the case of intuitive problem solving because they look for legitimization under disguise of rationality (Meyer and Rowan 1977). Rituals play the same role as assumptions for rational thinking: to reduce anxiety caused by uncertainty. The difference between two approaches is in the number of assumptions, their explicitness, and how threatening it is perceived to question them.

The organizational culture that overemphasizes virtues of rational thinking and underestimates its difficulty is not enough for halting the emerging understanding for a long time. As a mode of individual learning it is inferior to a combination of intuitive and rational modes described at the beginning of this section. To explain its endurance we need to take into consideration the motivational aspects of organizational culture.

Two reasons to perceive learning as risky were mentioned. Both of them are of psychological nature: cognitive dissonance aggravated by lack of understanding how to escape from it, which is characteristic for the "Hell road" stage of understanding, and discomfort caused by improvement attempts, which mostly fail when individual learning lags behind the organizational one. But what may prevent majority of people from learning is of social origin. Influence of culture on learning is not limited to creating unrealistic expectations to rational problem solving. The personal discomfort of failure is magnified and criteria of success and failure become loosely coupled to learning in the culture of reward systems based on indirect motivation.

In this culture understanding and satisfaction of intellectual curiosity are not valued per se. Indeed, learning is not among direct means to what is considered an end in itself--monetary rewards, which are inseparably linked to possibilities of consumption, recognition, and a mix of recognition and material rewards called prestige.

Possibilities to exchange results of their work for compensation are limited for employees. Job market is even further than product and service market from the one guided by classic economic rationality. Therefore, it is widely perceived that to get all these rewards one needs to be nice to those who may influence his or her remuneration. Indeed, being nice becomes a good to be exchanged (Mills 1963). It is revealing that a professional journal for engineers (it seems me it was EE ) recommends to divide time "fifty/fifty" between working and socializing with your boss and colleagues. An author explains that this is better than to spend 70% of your time working and the rest suffering that you are overlooked for a promotion. Certainly, people often discuss work during their lunch. Being nice helps to get things done. Yet to get things done is, actually, linked to rewards through being nice. To get things done becomes just one of the ways to be nice. Most of people in most situations when there is a conflict between getting things done and being nice, will select the latter (Argyris 1990).

Only to the extent that learning may promote an ability to get things done, it becomes of any importance. Thus, the following diagram may represent the structure of a commonly shared image of reward systems.

This is why, to engage in learning is a pretty risky and indirect way to go for desired rewards. Actually, the chance that people who are really hooked on learning will be rewarded is infinitesimal.

An aim to get things done does not necessarily propels and even may be in conflict with learning. To understand how things work one may need to break them. Mistakes are necessary to learn about boundary phenomena, i. e. to answer questions like: how much is not too much? The best way to learn about a bankruptcy is to drive a company into a bankruptcy (Bakken et al 1992). On the other hand to get things done one may need to change them. Yet observations often require having things in undisturbed state.

People for whom learning is the main motivator may disregard deadlines, pay no attention to others, be forgetful about everything what is not directly related to the problem they are solving. Formulating a question makes them happy. Yet others may perceive that as uncovering new problems. The natural analysts constantly and almost unconsciously build models of everything in their surroundings. This is how they make sense of their environment. And they may automatically evaluate how everyone they encounter fits their models. Such evaluation may be annoying for most of people. The natural analysts, in turn, may be angry with others because their behavior does not optimize learning. White lies and cover up of mistakes are serious sins for them. In one word, people devoted to learning often are not nice in the general meaning of this term.

People whose utmost desire is to understand are efficiently screened out of the culture where being nice is more direct way to goals of consumption and recognition than learning. A usual reward system does not work in the case of such learnaholics. Pay raises or public recognition of their merits may not stimulate them. That may irritate their bosses. Problems will become even worse, if a natural analyst's attention is attracted by a reward system itself. In such circumstances they may lose both income from work and any possibility to learn provided at work. If a person is not successful according to the commonly accepted criteria of success, whatever a person does tends to be disregarded. A question: "If you are so smart, why aren't you rich?", is a rhetorical one.

There are many disparities between a dominant reward structure and what can motivate natural analysts. They are well known from research on reward structures in scientific institution. I was not able to find only one demonstrated by professionals who are offered a position where they will be underutilized, but prefer an unqualified job which pays considerably less and has nothing in common with their profession.

Most of the above problems with a person who is directly motivated by learning are the same as for anyone who really cares about what he or she does. A custodian may work too good and because of that too slowly. He may become angry with his boss who walks on a wet floor. Indeed, it is possible to turn any work into creative activity with a large learning component. It depends on a person. People who care to learn are different in the way that they are interested in a possibility to get things done rather than in getting things done. Also they may be more sensitive to cognitive dissonance because of refined mental models they possess.

When most people whose motivation is shaped under the present culture try to learn, they quickly learn one thing: how risky it is. Can this situation be alleviated? Works by Argyris (1990) and Senge (1990) are examining this topic in depth. I will select only those aspects of the problem which are touching upon cultural issues. Also I will propose some actions which, it seems me, are inseparably linked and will go hand in hand with cultural change.

Relax institutionalization of rationality. To perform this task researchers and people of practical affairs need to join their efforts.

Researchers modestly recognize that they cannot provide directions for what to do. At most they can tell what will happen, if this or that is done. When social phenomena are involved even this is usually impossible at the present stage of development of social sciences. Direct benefits of research in this case are almost always limited to developing intuitive understanding that may help in making better decisions.

People of practical affairs do not use formal problem-solving methods and scientific studies for legitimization of their actions. They accept recommendations only after checking them against their intuition. If intuition says them nothing they disregard the study's recommendations (but not the study itself!).

Word of caution. The above qualifications will immediately reduce demand for research and training.

Recognize differences in needs of individuals. The recognition should lead to more diversified reward systems. I am not an expert in this area. My feeling is that the present reward systems emphasizes remuneration for activities one is told to do and does not value an effort to figure out what is to be done. The underlying assumption under designing reward systems is: employees must be motivated to do what they would not do otherwise. It is supposed that needs of employees always differ from the needs of a company. It is still considered that the best what can be achieved by remuneration is to create a zone of indifference: after employees have sold their "eight hours a day", they are ready to do whatever they are told to do and what, supposedly, corresponds to the company's needs. May be this is a valid approach to making majority of people to work. Yet a saying goes: four people are enough to bring a horse to a river, but even four times four cannot make her drink. I think that the same is true for learning and agree with Senge (1993) who says:

Word of caution. When some employees are allowed and even supported "to learn what they want to learn" and to learn is what they want to do, while other are coerced by the reward system to do what they would not do otherwise, a contradiction in a very sensitive place of Cultural System is created. This may have major consequences for stability of Socio-Cultural interaction (Archer 1988). It is difficult to imagine what will follow, if some organizational members are remunerated to do what they want to do and they are urged to trust what their feelings tell them, when the rest, who constitute a majority, do not have this privilege. By whom and how these people will be selected? It is easy to say that they should be appropriately socialized, but also not easy to find by whom and who will decide what is an appropriate socialization. The problem is well known, but solution is not (Scott and Hart 1979).

Recognize differences in individual cognitive styles. Different individuals use different proportion of intuitive and rational learning. It is quite widely accepted that this distinction corresponds to different levels in organizational hierarchies (Argyris 1990; Kim 1992). On the bottom people are technical and concrete, when on the top they use an abstract language of money. Middle managers have a role of translators in this framework. Argyris compares information which is relevant on the level of first-line supervisors and of senior executives. Former "usually deal with information that is concrete, unique to the group, subjective, and implicitly logical (p. 159)." Later "require information that is more abstract, objective, and explicitly logical and that is trendable and can be compared (p. 160)." There is a need to build a system that will enable translation and screening of information to provide organizational members with data relevant for making effective decisions. Capacities and constraints of the human mind require that different kinds of information are matched with appropriate cognitive styles of decision-makers. This matching will be done within the limits imposed by the requirement that an information of adequate kind and complexity is used to make decisions.

Word of caution. First, phenomena that are easy to measure, for example financial results, are given more weight in making decisions (Simon 1987/c1978). This is one of the reasons why human factor and theories we use to create meanings, which are at the core of the double-loop learning (Argyris and Schon 1978), are often overlooked. Second, translation brings in distortion. Third, organizational systems tend to settle into procedural routines and stabilize themselves. To confront the tendency, a learning mechanism is to be built in the system that will adjust it to new sources, nature, and quantity of information.

Recognize differences in thought processes. According to Sanders (1966) individual learning passes six stages of a hierarchy of thought processes. Each stage requires more complex thinking than the one preceding it, and also builds and incorporates the preceding types of thought. That is, one needs to develop the capacity for lower levels of thought in order to master the higher levels. To ensure continuous learning it is important to link organizational members in the system that will utilize capacities of each level. For this purpose employees should feel free to look for help of the next level of competence, if they feel that they cannot find an adequate answer. Two principles are important here to promote learning through such collaboration. First, do not hand a request up until you are sure that the proper question is formulated. Second, do not hand an response down until you are sure that it has answered the question. This takes time. Both developing a question and finding an answer takes time too. So, enough time is a crucial condition for learning to take place.

Word of caution. It is necessary that each member of organization knows whom to address with a question. For this reason it may have sense to establish a formal status system corresponding to a hierarchy of thought processes. Yet under conditions of institutionalized rationality it may work against the two above principles. Authority of the position alone will make an answer clear and people will hesitate to ask questions. Also classification of individuals as belonging to a particular status may block their learning. The system must provide an opportunity to move up, but should not pull by providing any indirect initiatives. I believe that people will move to the next level, if there is an opportunity, simply because it provides relief from questions which cannot be answered on preceding level. Time is all what is needed.

Total Quality Management movement, which stresses learning as a major condition for quality and recommends allocating adequate time, is elaborated in the emerging philosophy of an agile firm, which agrees that quality is important, but insists that we need to achieve it rapidly (Guaspari 1992). Rapidly means quicker than our competitors. If they started the journey earlier, we need to hurry. To hurry in learning is the surest way to become stuck in its intuitively rational mode. To have enough time for learning, it is better to start earlier.

Keep getting things done and learning coupled but not too tightly. Experimentation in virtual worlds--use of computer games simulating real processes--promises to provide low cost, compressed in time and space and not-threatening environment for learning. MIT's Organizational Learning Center is actively engaged in designing computer simulations of management processes--management flight simulators (Senge 1990). The Center also carefully studies effectiveness of this learning approach (Bakken et al 1992).

Word of caution. Learning in virtual worlds may result in an increasing reliance on symbols and expert systems instead of individual analysis of sensory perception. The consequences may be similar to a skewed perception of a child, who learned karate from a Nintendo game. A deadly strike, which requires years to master and extraordinary energy to carry out, is associated in the child's mind with easily pushing a computer button with one finger. Compression of time and space is achieved at a cost of deceiving individuals' internal clock and feeling of distance. Learning how to feel tenure and dimensions of processes is an important part of learning. Ability to be patient and wait for appropriate time for action is as important as to know what is to be done.

Flight simulator is the next step in the evolution of mediation of experience (Giddens 1991)--the phenomenon that was created by emergence of oral and written languages as well as radio and television broadcasting. An important role of mediation of experience for socialization as well as confusion and anxiety caused by time-space distanciation, which results from mediation of experience, are recognized but not well understood yet.

Argyris, Chris. 1990. Overcoming Organizational Defenses: Facilitating Organizational Learning Wellesley, MA: Allyn and Bacon

Bakken, Bent, Janet Gould and Daniel Kim. 1992. "Experimentation in learning organizations: A management flight simulator approach" European Journal of Operational Research 59:167-182

Barnard, Chester. 1956/c1948. Organization and Management Cambridge, MA: Harvard University Press

Becker, Howard. 1986. Writing for Social Scientists Chicago: Chicago University Press

Bell, Daniel. 1956. Work and Its Discontents Boston: Beacon Press

Bloom, Benjamin. 1956. Taxonomy of Educational Objectives: The Classification of Educational Goals New York: David McKay

Brassard, Michael ed. by. 1985. The Memory Jogger: A Pocket Guide of Tools for Continuous Improvement Methuen, MA: GOAL/QPC

Caulcutt, R. 1990. "More basic tools for quality improvement" Process Engineering, February:50-51

Collins, Randall. 1975. Conflict Sociology: Toward an Explanatory Science New York: Academic Press

Deming, W. Edwards. 1982. Quality, Productivity, and Competitive Position Cambridge, MA: MIT, Center for Advanced Engineering Study

Festinger, L. 1957. A Theory of Cognitive Dissonance Evanston, IL: Row, Peterson.

Feynman, Richard (as told to Ralph Leighton). 1988. "What Do You Care What Other People Think?" Further Adventures of a Curious Character New York, London: W.W. Norton & Company

Giddens, Anthony. 1990. The Consequences of Modernity Stanford, CA: Stanford University Press

Giddens, Anthony. 1991. Modernity and Self-Identity: Self and Society in the Late Modern Age Stanford, CA: Stanford University Press

Genter, D. and C. Toupin. 1986. "Systematicity and surface similarity in the development of analogy" Cognitive Science 10(no.3):277-300

Guaspari, John. 1992. It's about time: A Fable about the Next Dimension of Quality New York: AMACOM

Herrington, G. 1992. "Putting Quality Back in the Barrel" Quality Progress, May:63-65

Holmes, F. L. 1980. "Hans Krebs and the Discovery of the Ornithine Cycle" Federation Proceedings 39:216-225

Jordan, K. and M. Lynch. 1992. "The sociology of a genetic engineering technique: ritual and rationality in the performance of the 'plasmid prep'" in The Right Tools for the Job ed. by Adele Clarke and Joan Fujimura. Princeton, N J: Princeton University Press

Kahnemann, D. and A. Tversky. 1974. "Judgment under uncertainty" Science 185:1124-1131

Katz, Daniel. 1975. Bureaucratic Encounters: A Pilot study in the Evaluation of Government Services Ann Arbor, MI: Survey Research center, Institute for Social Research, University of Michigan

Kim, Daniel H. 1992. "Organizational learning and individual learning: where the twain shall meet?" Cambridge, MA: MIT Sloan School of Management, Center for Organizational Learning, working paper D-4114-1

Kohn, Alfie. 1986. No Contest: The Case Against Competition Boston, MA: Houghton Mifflin

Lave, Charles and James March. 1975. An Introduction to Models in Social Sciences New York: Harper and Row

Levitt, B. and J. March. 1988. "Organizational learning." Ann. Rev. Sociol. 14:319-340

March, James G. and Herbert Simon. 1958. Organizations New York: John Wiley

Marcuse, Herbert. 1964. One-Dimensional Man; Studies in the Ideology of Advanced Industrial Society Boston: Beacon Press

Meyer, John, and Brian Rowan. 1977. "Institutionalized Organizations: Formal Structure as Myth and Ceremony" American Journal of Sociology, 83:340-63

Mills, C. Wright. 1963. Power, Politics, and People New York: Oxford University Press

Oakland, J. 1983. "Low usage of SQC and lack of commitment" Quality Today October:31-32

Pirsig, Robert. 1984. Zen and the Art of Motorcycle Maintenance: An Inquiry into Values New York: Bantam Books

Porter, L. and J. Oakland. 1991. "Process capability indices - an overview of theory and practice" Quality and reliability Engineering International 7:437-448

Prietula, Michael and Herbert Simon. 1989. "The experts in your midst" Harvard Business Review Jan.-Feb.:120-124

Sanders, Norris. 1966. Classroom Questions: What Kinds? New York: Harper & Row

Scott, W. Richard and David Hart. 1979. Organizational America Boston: Houghton Mifflin

Senge, Peter. 1990. The Fifth Discipline: The Art and Practice of the Learning Organization New York: Doubleday/Currency

Senge, Peter. 1993. "Internal markets and learning organizations: Some thoughts" Cambridge, MA: MIT Sloan School of Management, Center for Organizational Learning, working paper D-4355

Simon, Herbert. 1983. Reason in Human Affairs Stanford, CA: Stanford University Press

Simon, Herbert. 1987/c1978. "Rational decision making in business organizations" in Advances in Behavioral Economics, volume 1, edited by Leonard Green and John Kagel. Norwood, NJ: Ablex Publishing Corporation

Swanson, Guy. 1980/c1979. "A Basis of Authority and Identity in Post-Industrial Society" in Identity and Authority: Explorations in the Theory of Society ed. by B. Holzner and R. Robertson. New York: St. Martin's Press

Weber, Max. 1968/1924. Economy and Society Stanford, CA: Stanford University Press

Weingart, Laurie, Rebecca Bennett, and Brett Jeanne. 1993. "The impact of consideration of issues and motivational orientation on group negotiation process and outcome" in press: Journal of Applied Psychology