A preliminary typology of organizational learning: synthesizing the literature

A preliminary typology of organizational learning: synthesizing the literature

Danny Miller

Organizational learning processes are being explored with increasing interest and vigour. But it remains unclear just what learning is, how it takes place, and when, where and why it occurs. Part of the problem is that learning, as portrayed in the literature, is a haphazard and eclectic notion. Researchers lump together processes that are strikingly different in their causes, effects, and domains. The thesis of this article is that we can only begin to understand learning after we have made some essential distinctions among its many varieties.

Based on two fundamental contrasts among the dominant organizational paradigms: voluntarism vs. determinism, and methodical vs. emergent thought and action, we have identified six common modes of learning. These manifest contrasting cognitive and socio-political processes, engage different parties, and have varying spheres of influence. We shall argue that the nature, contexts and typical outcomes of the different learning modes will be very distinct.

What is Organizational Learning?

The definition of learning remains somewhat obscure, in part because the process has been described so differently in the literature. Some authors view learning as a change in behavior in response to a stimulus (Cyert & March, 1963; March, 1989). But this seems mostly to be a description of reaction or adjustment, which may be blind, automatic, and productive of no new knowledge. Other scholars suggest that learning requires some conscious acquisition of knowledge or insight on the part of organization members (Argyris & Schon, 1978; Hedberg, 1981; Huber, 1991). But were this knowledge to be unrelated to organizational action or decision making, it would be relevant only to individual learning, not to that of organizations.

Fiol and Lyles (1985) noted that “organization learning means the process of improving actions through better knowledge and understanding.” We agree that it is important to include both cognitive and behavioral elements in a definition of organizational learning. However, we are uncomfortable with any normative overtones. Given that some kinds of learning may ultimately do more harm than good, and that learning can help one objective or manager at the expense of others, we propose a different definition: organizational learning is the acquisition of new knowledge by actors who are able and willing to apply that knowledge in making decisions or influencing others in the organization.

Learning is to be distinguished from decision making. The former increases organizational knowledge, the latter need not. Learning may in fact occur long before, or long after, action is taken. Of course, methods of decision making may well influence learning processes, and vice versa.

Dimensions of Contrast: Two Paradigmatic Distinctions

Some of the most influential paradigms of organizational research can be contrasted along two important dimensions. The first dimension is the extent to which human or organizational action is constrained by cognitive, political and resource factors. Compare, for example, voluntaristic theories of business strategy that envision few such constraints, with more deterministic bureaucratic or ecological theories that anticipate many such limitations. A second dimension is the mode of administrative thought and action. Compare economic theories that emphasize methodical and intentional behavior, with institutional theories that envision emergent and spontaneous behavior far less driven by technical or economic norms.

Voluntarism vs. Determinism: The voluntarism-determinism axis gauges the extent to which people and their institutions are deemed intelligent and autonomous actors rather than entities severely restricted in cognition and action (Astley & Van de Ven, 1983; Berger & Luckmann, 1966; Burrell & Morgan, 1979; Giddens, 1979; Hrebiniak & Joyce, 1985). Strategic choice theorists, for example, adopt voluntaristic perspectives, allowing much latitude for free choice by decision makers (e.g. Andrews, 1971; Ansoff, 1979; Porter, 1985). Bureaucratic and neo-bureaucratic theorists are more deterministic and view cognition as being channeled and behavior as being constrained (Crozier, 1964; March & Simon, 1958; Cyert & March, 1963). Population ecologists see almost no latitude for strategic choice – or at least for strategic choice that increases the chances of organizational survival (Hannan & Freeman, 1977).(1)

These paradigmatic distinctions suggest an important difference in the way organizations can learn. We may contrast voluntaristic learning that is relatively free in thought and action, with that which is bound by cognitive (March & Simon, 1958), resource (Cyert & March, 1963), political (Lindblom, 1959) or ideological strictures (Berger & Luckmann, 1966). Global analysis performed by top managers is a relatively unconstrained kind of learning. Ideally, it surfaces critical assumptions about an organization and its markets, evaluates a wide range of strategic options, and affords much latitude for taking action (Ansoff, 1980). Some scholars, however, see firms as acting and learning in a more incremental way by performing small experiments that only slightly modify existing methods or offerings (Cyert & March, 1963; March & Olsen, 1976). A still more constrained kind of learning takes place within existing routines and programs: for example, employees may learn how to adjust or modify their equipment based on feedback from their quality control monitors (Nelson & Winter, 1982).(2)

Method vs. Emergence: A second dimension that contrasts theories of organizations gauges whether decisions are based on methodical analyses and concrete standards versus emergent intuitions and subtle values. Transaction cost theorists (Williamson, 1975), industrial organization economists (Porter, 1980, 1985), and proponents of corporate planning (Steiner, 1980), view managers as intendedly rational actors who make decisions by systematically analyzing hard information about competitive options and costs. Norms of economic or technical rationality predominate. By contrast, institutional theorists (DiMaggio & Powell, 1991; Meyer & Rowan, 1977) and those who have studied entrepreneurs (Collins & Moore, 1970; Miller, 1990b) emphasize how organizational actions are more spontaneous and emergent. They are driven by subtle normative considerations, fads and rituals, and even personal hunches and motives.

Again, these paradigmatic distinctions suggest a rather striking difference in the way organizations learn. Methodical inquiry is analytical and prizes “objective” facts; it is systematic and often tests notions deductively. Objectives are clear, and facts are gathered and evaluated in an orderly way, and with explicit purpose (Ansoff, 1965). Emergent rationality, on the other hand, is more spontaneous and intuitive, and it centers on instincts and impressions. Intuitive managers learn tacitly and inductively, garnering insight by interpreting unstructured impressions and streams of data. These administrators may not have precise objectives, may gather facts only implicitly, and might make choices quite unconsciously (Miller, 1990b; Mintzberg, 1989).

Examples of methodical learning include systematic analyses of competitive markets and business strategy, experiments with products or technologies that aim to reduce costs or increase profits, and the use of statistical quality control procedures to identify problems or opportunities in process design. By contrast, emergent learning may be evidenced by conceptual insights that provide executives with a new vision for their enterprises. Emergent learning can also come from the social and political interchanges managers have with parties inside and outside the organization.

Creating a Typology From the Two Dimensions

The voluntarism/determinism and methodical/emergent dimensions in combination evoke six common types of organizational learning: types that vary in their approaches, outcomes and contexts. We propose three kinds of methodical learning and three kinds of emergent learning, each distinguished by the level of voluntarism: i.e., by the constraints on thought and action (see Table 1).

A typical variety of methodical learning is the systematic analysis that some managers do when formulating or implementing strategy. Such administrators are said to scan the environment for information, to analyze that information along with data on internal resources and processes, and to generate numerous action options. They select an outstanding alternative from among these options. Then they methodically evaluate achievements against predetermined objectives (Andrews, 1971; Ansoff, 1965; Fredrickson, 1986; Hart, 1992).

Other researchers, while endorsing such methodical inquiry, have called attention to the time and resource limitations that constrain its thoroughness. They describe learning that is characterized by bounded, problem-driven search, incremental experimentation and satisficing behavior (Braybrooke & Lindblom, 1963; March & Simon, 1958). Managers are said to engage in little reflection, to gather only the most accessible information, and to consider only a few alternatives on the way to an “adequate” solution. The latitude for dramatic action is thus limited.

Another methodical mode that allows still less scope for voluntarism is structural learning, which is driven by routines that standardize information processing and behavior. Routines specify what data managers must gather and attend to, and they guide how managers interpret that information (Levitt & March, 1988; Nelson & Winter, 1982; Starbuck, 1985; Starbuck & Milliken, 1988).

Analysis, experiment and structural learning are all methodical and deliberate. They test ideas by systematically gathering factual information and changing behavior. Results are then monitored, and the cycle begins again. However, as we move from analysis to routines there is a progression towards less voluntarism. Analysis places few restrictions on learners, small experiments limit action, and routines channel both thought and action.

Three emergent forms of learning are also quite common, and again, they may be distinguished according to their degree of voluntarism. Some scholars, for example, have pointed to synthesis – learning that is intuitive and holistic (Miller, 1990b; Mintzberg, 1989; Palmer, 1969). Such learning represents an instinctive form of pattern recognition an ability to generate global insights about issues facing an organization. It is internal to a manager, usually mysterious, and may even be unintentional.

Table 1. Modes of Learning

Mode of Thought and Action

Methodical Emergent

Constraints (Voluntarism)

Few constraints Analytic Synthetic

Action constrained Experimental Interactive

Action and thought constrained Structural


A less voluntaristic form of emergent learning takes place via interaction and occurs when there is much social and political activity. Contrasting ideas and conflicting objectives among members of an organization may cause search and choice to be based on – and limited by – a complex forcefield of different aims and issues (Cyert & March, 1963; March & Olsen, 1976). Actors infer a course of action by discovering patterns in the pressures and opportunities facing them as they interact with a host of allies and rivals.

Institutional learning is determined by ideologies – by institutional forces such as laws, social norms or personal values that shape managerial thinking (DiMaggio & Powell, 1983). Here learning is a product of indoctrination, either subtle or direct (Selznick, 1957; Scott, 1995).

Again, as we move from synthetic to institutional learning there is a progression from more voluntaristic modes of emergent learning to more deterministic ones. Synthesis places few restrictions on actors, political interaction limits action, and institutional learning channels both thought and action.

The six kinds of learning are presented in Table 1. It is interesting that as we move down the table there is a tendency for learning to take place at lower levels of the hierarchy – analytical and synthetic learning, as will be described, are often the province of upper echelons, experimental and interactive learning are engaged in frequently by middle managers, and structural and institutional learning take place even at lower levels. The scope of learning also alters as we move down the table – from a selective to a broader influence. We will have more to say later about these tendencies.

Common Modes of Learning

Our six modes of learning each represent a common “configuration” of mutually supportive elements. The significance of these modes is not that they are exhaustive but that they reveal the wide range of common learning processes identified by the literature. Also, we will argue that they produce disparate outcomes and must occur in distinctive contexts. Although one or two learning modes may dominate in many organizations, several modes may easily co-exist.

In presenting each mode, we will first describe search and choice behavior, then address the locus and diffusion of learning with some typical examples, and finally, discuss the potential strengths and weaknesses of each mode. These descriptions are followed by hypotheses concerning the contexts, outcomes, and connections among the modes. The summary Table 2 relates the six modes of learning to their sources in the literature; Table 3 then compares the modes along some salient dimensions of process.

Analytic learning

Rational analysis is a well-known, perhaps overly idealized, mode of methodical learning (Allison, 1971; Ansoff, 1965). Learning occurs via intensive and systematic information gathering both from within and from outside the firm. Operations [TABULAR DATA FOR TABLE 2 OMITTED] are analyzed and the environment is scanned to discover key problems and opportunities. In making decisions, top managers and their staff of planning analysts evaluate a wide variety of alternatives. Careful choice is made of the alternative that maximizes one or two key objectives (Grandori, 1984), usually ones having to do with profitability and growth. Ultimately, feedback from performance is scrutinized and used to adjust tactics or strategies, thus beginning another learning cycle (Steiner, 1979). Much of the information gathered is quantitative and is monitored via formal systems. The emphasis is on hard intelligence data, deductive logic, numerical calculation, and even optimization (Ackoff, 1971).

The proactive and far-reaching analysis we refer to here generally can only take place at upper echelons – by executives and their staff departments. Grand plans have little chance of being accepted if they are formulated by those with [TABULAR DATA FOR TABLE 3 OMITTED] little power, especially if they challenge elite beliefs (Staw, 1977). Also, the mandates of lower level managers are too specialized to provoke analyses of fundamental scope.

Most analytical learning is diffused only selectively to lower levels of the organization. Although the results of learning are implemented via detailed plans, programs and routines, lower levels may not always be made aware of underlying rationales.

Applied to the learning of strategy, the analytical mode might be used to do competitive analyses of markets, to develop and monitor strategic marketing plans and new product strategies, and to make diversification and acquisition decisions (Porter, 1985; Schoemaker, 1993).

Analytical learning gives managers a better idea of the critical forces that they have to deal with in their environment. It also identifies practices and areas of operation that need improvement. At its best, this kind of learning even helps managers think critically about their assumptions regarding markets, practices and success factors (Ansoff, 1979; Senge, 1990). But global analyses may fail to be informed by the experiences of lower level managers. They may also be excessively general, mechanically rigid, and devoid of vital “soft” information (Aguilar, 1967; Halberstam, 1986; Mintzberg, 1979; Wilensky, 1967).

Synthetic Learning

Compared to analysis, synthesis is a less systematic but more emergent, intuitive and holistic mode of learning. It combines different bits of knowledge in a new way so that novel relationships or patterns are revealed – so that “the whole becomes greater than the sum of the parts.” Concepts may be “reconfigured” to display harmony, consistency, and fit (Mintzberg, 1989). Focal themes, critical relationships, pressure points, or systemic properties may be identified to reveal new possibilities (Nonaka, 1988).

True synthesis calls upon the creative capacity to detect systems and configurations where others see only a jumble of elements. Managers engage in a hermeneutic process of interpreting organizational “texts” streams of decisions and events – to uncover key patterns, assumptions, hidden meanings, and subtle dynamics (Palmer, 1969; Radnitzky, 1968; Kets de Vries & Miller, 1987). This elusive pursuit is beyond the ability of many managers, and even those who rely on it can only do so sporadically. Moreover, whereas analysis employs deductive logic and can be approached methodically, synthesis is inferential and is not amenable to any step-by-step procedure. Indeed, the notion of method is to a degree alien to this kind of learning, which may take place quickly and without warning, but which also might not be achieved even after months of effort.

The evaluation standards used in synthesis are largely aesthetic and subjective. Managers, rightly or wrongly, “recognize intuitively” the importance and value of their emerging insights (Palmer, 1969), and may engage in more deliberate assessment only later (Radnitzky, 1968).

Synthesis is normally a product of a single creative mind. Although insight may come suddenly, it often happens only after managers have grappled long and hard with a messy assortment of issues, problems, and opportunities. Hence, as with analysis, the most poignant insights from synthetic learning may remain with a few managers. It is mostly the implications of that learning – in the form of new goals, decisions and prescribed procedures – that are infused throughout the organization.

Two organizational phenomena best illustrate synthesis: configuration and systems thinking. Synthetic learning can pull together information in a way that gets rid of extraneous details and converges on what is most important. It identifies a “configuration” – a theme that reveals how the parts of a problem, strategy, or even organization interrelate (Miller, 1987, 1990b). This may help managers discover a core competence or critical resource, or identify a competitive advantage (Peteraf, 1993; Prahalad & Hamel, 1990). Synthetic learning may also find synergies among organizational departments or skills, or it may discover a desirable nexus between organizational competences and challenges (Hambrick, 1994; Miller, 1987; Mintzberg, 1973, 1989).

Another application of synthetic learning is systems thinking (Beer, 1981). Some executives are able to view their organizations as dynamic, evolving configurations: they not only see how the parts fit together, they also develop a feel for the dynamics of the system. They begin to understand networks of causality, to discern complex problematic relationships, vicious circles, emerging opportunities, and shifts that signal special challenges (Senge, 1990; Stata, 1989; Steinbrunner, 1974).

There are many strengths associated with synthesis. It can discover powerful synergies and great opportunities, and it may give managers a more profound understanding of the underlying dynamics of their organizations (Mintzberg, 1989; Senge, 1990). But synthesis may create a vision that is so integrated and compelling that it defies necessary alteration (Miller, 1990b, 1993). Also, effective synthesis is beyond the intellectual capacity of many managers.

Experimental Learning

Scholars of organizations have noted that the analytical model ignores limitations on decision makers (March & Simon, 1958). They believe that voluntarism is bounded by intellectual, temporal and economic constraints (Grandori, 1984). They also suggest that overcoming these constraints requires simpler, more incremental approaches to learning – for example, performing small experiments and monitoring the results (Quinn, 1980). Managers can explore their complex environments in a gradual, piecemeal way instead of making grand long-term plans (Burgelman, 1990; Staw, 1977).

Like analysis, experimentation is an intendedly rational, methodical approach to learning. There is a deliberate attempt to systematically gather and interpret information in the hope of improving the behavioral repertoire of the organization. But now action sometimes precedes analysis in the learning cycle (Weick, 1979). Experimental learning is also more spontaneous than analytical learning as it is not so governed by detailed plans. Search occurs by conducting experiments remedially, opportunistically, and in many areas (March, 1989, 1991). Choice – the decision of how far to go with experiments – is generally determined by feedback about results (Lant & Mezias, 1992).

Experimentation reduces the burden on top managers by occurring at many levels and places of an organization (Hart, 1992). In some cases, “atomistic” experimentation takes place – small projects and adaptations that are done almost independently by operating units (Garud & Van de Ven, 1992). Elsewhere, experiments are performed higher up and are more coordinated and interrelated (Pascale, 1989). Although participation in experimental learning is sometimes quite broad, what is learned may be used only locally. Thus, experiments are more likely to influence situational tactics and practices than global strategy. Occasionally, however, experiments will produce dramatic local victories or blunders that will be noticed throughout the system and serve as seminal strategic lessons (Lant & Mezias, 1992).

Experimental learning is especially likely to occur during attempts at adaptation or renewal, as changes to product lines and methods are made to discover better ways of doing things (Zaltman, Duncan & Holbek, 1973; Garud & Van de Ven, 1992; March, 1991).

Experimentation often has the advantage of limiting risks, reducing the cognitive burdens of top managers, and exploiting the capacities for learning that exist throughout an organization (Huber, 1991; Quinn, 1980). But it usually results in knowledge that is more local, more fragmented, and thus harder to integrate than that which derives from most other modes of learning (Miller, 1990b; Staw, 1977).

Interactive Learning

Like experimentation, interactive learning involves learning-by-doing, which occurs simultaneously in many parts of an organization. But instead of systematically experimenting with practices and offerings, managers learn in a more emergent and implicit way: by bargaining and trading with each other and with external stakeholders (Cohen, March & Olsen, 1972). Learning is by the exchange of information and the evaluation of transactions that reveal the motives, resolve, and resources of rivals and allies within and outside the organization. It is the product of a spontaneous process of local adjustment to local opportunities and political forces (MacMillan & Jones, 1986; Pfeffer, 1981).

In the interactive mode, aims are modest. Learners attempt to understand just enough about the diverse pressures on them to discover some areas of freedom and opportunity. They seek out moves that are “good enough” to incrementally advance their agendas. Then they quickly re-evaluate and adjust their actions in response to feedback. This entire process is essentially one of “muddling through” (Braybrooke & Lindblom, 1963; Lindblom, 1959).

Interactive learning is more intuitive and inductive than methodical. It relies on hunch, political instinct and “reading people.” Managers learn about their surroundings by interpreting their many encounters and bouts of negotiation. And the many competing stakeholders usually limit the scope for action (Eden, 1992).

The unit of learning is typically an individual or department trying to achieve local objectives (Cyert & March, 1963; March & Olsen, 1976) – aims that may have more to do with personal socio-political agendas than with organization-wide economic or service goals (Georgiou, 1972; Silverman, 1970). Although scores of managers participate in interactive learning, learning outcomes tend to have only a local effect. But collectively, these outcomes, agreements, side-payments, and incremental adjustments can very much influence the evolutionary path of the organization. Many organizations are profoundly shaped by these multitudinous adaptations to a wide variety of constituencies (Cohen, March & Olsen, 1972; Chaffee, 1985; Hart, 1992; Pettigrew, 1973; Shrivastava & Grant, 1985).

Interactive learning is especially common in public organizations where goals are vague and power is broadly distributed. In modifying college curricula, for example, there may be a protracted series of bargaining sessions among different departments, each trying to increase the level of representation of its discipline. In the process, professors learn about each others’ preferences and agendas, resources and skills, and areas of compromise. This knowledge affects the subsequent bargaining tactics of each actor. Learning continues until some agreement is reached after a series of proposals and counterproposals (March & Olsen, 1976). Similar processes may take place in inter-organizational exchanges, as networks of companies negotiate and carry on joint ventures or projects (Harrigan, 1986).

A strength of interactive learning is that it allows managers to exchange a good deal of information with one another, and this fosters more realistic collaboration. Also, like experimentation, interactive learning simplifies organizational adaptation by breaking it down into local agreements and accommodations (Cyert & March, 1963). This is especially useful when numerous stakeholders with different values are involved. Unfortunately, political opportunism may run rife, and global objectives may be sacrificed to parochial political considerations (MacMillan & Jones, 1986; Williamson, 1975).

Structural Learning

One of the most pervasive forms of methodical learning occurs via organizational routines (Cyert & March, 1963; Nelson & Winter, 1982; Perrow, 1986). Routines are products of analytical learning and design; they codify prior learning by specifying how to carry out tasks and roles efficiently. Thus, they are perceived to be instruments of rationality. But routines also guide learning, both implicitly and explicitly (Nelson & Winter, 1982). They spell out methods for improving efficiency, correcting errors, or refining existing processes. This invokes a programmed kind of learning in that routines normally delineate standards and rules of conduct that are supposed to ensure reliable performance (Grandori, 1984; Hannan & Freeman, 1984). In effect, search behavior is guided by routines, and choice criteria are specified (Herriot, Levinthal & March, 1985).

Routines also teach by directing attention, institutionalizing standards and assumptions, and even creating basic vocabularies (Hedberg, 1981; Meyer & Starbuck, 1991; Starbuck, 1985). They control what information people receive, and also influence how managers will view that information, thereby imposing consistency in thought and action (Clegg, 1989).

The notion of structuration as developed by Giddens (1979) provides some insights not only into how routines shape learning, but how they shape themselves. Over time, there are reciprocal processes in which actors design routines, which then channel their behavior and perceptions, which in turn influence the way in which future routines are designed (Ranson, Hinings & Greenwood, 1980). In this cycle of meta-learning, the routines come very much to be modified by the information they produce (Whittington, 1992). Although routines are usually designed high up in the organization – by managers and analysts who adopt elite perspectives – their effects are broad and pervasive.

There are many examples of routines. Quality control routines allow employees to discover why there are flaws in their output and show managers how to take corrective measures. Marketers learn about their customers by following market research routines. Engineers develop better products using design routines.

Routines foster habit and redundancy; they constrain the range of experiences, and so are more likely to foster learning that refines existing practices rather than discovers new ones (Argyris & Schon, 1978; March, 1991). But some routines may be used to systematize the process of innovation itself (Nelson & Winter, 1982). Unfortunately, like other forms of bureaucracy, routines can create inertia, tunnel vision, rigidity, and alienation from work (Perrow, 1986).

Institutional Learning

Institutional learning is an emergent, inductive process by which organizations assimilate values, ideologies and practices, either from their environments or from their elite members (Scott, 1995). Learning is by a very large group of organizational participants, so that knowledge is widely diffused. It is the broad membership that learns, and the environment or elite members that “teach.” Moreover, the content of what is taught is often ideological. Given that learning may be through subtle indoctrination, socialization, or even coercion by powerful parties, it allows for little voluntarism on the part of learners.

Indoctrination may occur subtly, as when professional or societal values are infused throughout an organization via the efforts of influential members (Clark, 1956; Fligstein, 1985; Meyer & Rowan, 1977). For example, hospitals might learn their values from a powerful group of physicians, all of whom have been trained and socialized in a similar way (DiMaggio & Powell, 1983). Such values, many of them implicit, are reinforced by role models, status rituals, special procedures, vocabularies, and the like (Perrow, 1986; Scott, 1995). This kind of learning may go on constantly as new members of an elite enter an organization and convey the evolving values of a group or profession.

Indoctrination may also occur more overtly, as a leader preaches his or her vision of the mission of the organization and diffuses it to members via rituals, symbolism, charismatic exhortation, or example (Berger & Luckman, 1966; Deal & Kennedy, 1982; Selznick, 1957). CEOs might, for instance, preach the doctrine of quality because this is a key aspect of their vision, or because that is what they believe is most important to customers (Pascale, 1989). Or executives might establish organizational myths and legends that celebrate the achievements of a founder or valued employee. These instruct by example (Feldman & March, 1981; Peters & Waterman, 1982).

Institutional learning harmonizes the values of a leader, community, or stakeholder with those of the broader membership of an organization (Clark, 1956; Selznick, 1957). It also creates coherence among the beliefs of employees, making it easier for them to work together (Whitley, 1991). Such learning tends to legitimize an organization’s efforts in the eyes of powerful external parties. Of course, indoctrination can lead to blindness and intolerance (Janis, 1972), and it can enforce a cultural homogeneity within an organization that reduces the chances of survival (Clegg, 1989; Miller, 1993).


Having described the modes of learning, we can begin to examine the contexts in which they are apt to occur, the types of outcomes they are most likely to produce, and some of their interrelationships. Table 4 summarizes most of our hypotheses. It should be noted that the hypotheses only call forth general tendencies and assume ceteris paribus conditions. They are merely speculative starting points intended to launch discussion and inquiry.


According to Thompson and Tuden (1959) and Grandori (1984), two aspects of context can have an especially profound influence on learning and decision making: uncertainty concerning means, and disagreement about goals. The former concerns the degree to which there are known, reliable ways of achieving goals; the latter refers to the level of consensus among managers about the goals themselves. A third aspect of context, the complexity of an organization, may also have a beating on a few modes of learning. Complexity is reflected by an organization’s size and the diversity of its products and markets.

H1: Analytical learning will be most common where there is: a) modest uncertainty about means, and b) little conflict about goals.


High levels of uncertainty would make methodical, intensive analysis and long-term planning very difficult (Fredrickson, 1986; Simon, 1947); very low levels might make it unnecessary. Disagreement about goals among executives would place political obstacles in the way of comprehensive rational analysis (Grandori, 1984; March & Olsen, 1976).

H2: Synthetic learning will be most common where there is: a) much uncertainty about means, and b) modest conflict about goals.

The above hypothesis is especially speculative. Synthetic learning is a mysterious process that may be called upon when there is a need to pull things together in a new way. It may thus be particularly likely when there is growing uncertainty about means (Senge, 1990). In addition, conditions of moderate goal conflict may create dissonance and thereby motivate high level re-appraisals of an organization’s overall orientation. Intense goal conflicts, however, might cause severe tensions that distract managers and thus impede creative synthesis (Thompson & Tuden, 1959).

H3: Experimental learning will be most common where there is high uncertainty about means.

Experimentation is an incremental process that reduces cognitive overload under conditions of means uncertainty. It allows the gradual exploration of complex environments and permits learning to take place locally (March & Simon, 1958; March, 1991). Because it is decentralized, experimentation can occur even if there are goal conflicts; these, however, are by no means intrinsic to the experimental mode (Cyert & March, 1963; Grandori, 1984).

H4: Interactive learning will be most common where there is: a) much uncertainty about means, b) much conflict about goals, and c) significant organizational complexity.

Interactive learning occurs when it has to: when a high level of uncertainty forces decision making down to lower levels, and where much goal conflict engenders fragmented parochial perspectives (Cyert & March, 1963; Grandori, 1984; Lindblom, 1959; Thompson & Tuden, 1959). These conditions are especially common in large, complex organizations that act as political force fields.

H5: Structural learning will be most common where there is: a) little uncertainty about means, and b) little conflict about goals.

Routines require programmable and therefore clear goals and well-known means of attaining them. Their explicit and structured nature makes routines impractical when there is much goal conflict or means uncertainty (Thompson, 1967). Routines are also most attractive to larger organizations that have to do the same tasks many times over (Perrow, 1986).

H6: Institutional learning in the form of indoctrination will be most common where there is: a) modest uncertainty about means, and b) modest conflict about goals.

By the time managerial elites begin to concentrate on indoctrination, they normally feel confident that they know what has to be done but want others to be motivated to work out exactly how to do it (Miller, 1990b). Indoctrination may also help overcome mild goal conflicts at lower levels by motivating employees to do what is necessary (Selznick, 1957). Strong goal consensus, on the other hand, would make indoctrination superfluous; extreme disagreement might make it futile. Turning to a different form of institutional learning, organizations may be more open to learning from their institutional settings – for example, from evolving industry or social norms or from new recruits – when means and goals are open to discussion (Clark, 1956; Scott, 1995).

Match and Performance

A seventh hypothesis might be elicited from the preceding six: namely, that the natural matches between learning and context postulated above often will be functional, and that deviations from these matches may sometimes hurt performance. For example, using routine-driven learning in uncertain contexts might result in dangerous narrowness and inflexibility (Miller, 1993). Moreover, institutional learning would be impractical where there is much goal conflict (March & Olsen, 1976). Interactive learning, on the other hand, might be unnecessarily inefficient in stable settings characterized by goal consensus because such contexts normally permit more economical kinds of learning (Grandori, 1984). Hence, as a rule:

H7: The further the learning modes stray from their typical contexts, the more organizational viability is likely to suffer.


Although many different things can be learned from each of the modes, we believe that some outcomes may be especially typical: strategies are shaped by analytical learning, configurations by synthesis, procedures by structural learning, tactical accommodations by interaction, values by indoctrination, and innovations by experiment. An important distinction among these outcomes is the centrality of what is learned – the extent to which it underlies a good deal of what organization members think and do (Tushman & Romanelli, 1985). Another distinction is the radicality of what is learned the degree to which it promotes a fundamental change in basic assumptions or orientations (Argyris & Schon, 1978). A final outcome distinction concerns the relationship between knowledge and behavior. Learning always involves a continual interaction between these two aspects, but for some modes of learning, changes in behavior usually precede changes in knowledge. For others it is the reverse, and for still others, the two are too closely intertwined to know (Weick, 1979).

H8: Under analytic learning: a) centrality tends to be high, b) radicality tends to be moderate, and c) knowledge tends to influence behavior more than vice versa.

Analytic learning is often used by upper echelons in strategic planning and priority setting exercises, which can have an important impact on many aspects of organizational functioning. But the use of methodical, deductive systems and the reliance on hard, quantifiable data may cause tunnel vision and leave fundamental assumptions unquestioned, thereby limiting the potential for radical change (Halberstam, 1986; Starbuck, 1985).(3) The deliberate gathering and analysis of information precedes action, but feedback can then be used to adjust behavior.

H9: Under synthesis: a) centrality tends to be high, and b) radicality tends to be high.

Synthesis may be brought to bear on many kinds of issues, but because it often occurs at the top of an organization and concerns overall strategy, what is learned may be of central importance. Also, the emergent, inductive nature of synthesis may allow creative managers to challenge very fundamental assumptions and practices (Miller, 1990b). Although synthetic insights may be fed by past actions, they can generate cognitive leaps that are not captured by any simple causal connection between knowledge and behavior (Miller, 1990a; Mintzberg, 1989).

H10: Under experimental learning: a) centrality tends to be modest, b) radicality tends to be modest, and c) behavior often precedes knowledge.

Normally, experiments are incremental: they concern competitive tactics and operating practices. Managers discover which of their new efforts produce results, and they tailor their offerings and practices accordingly. The results of most experiments will not cause major shifts in basic goals or strategies (Starbuck, Greve & Hedberg, 1978). Although in the short run, the limited centrality of experiments constrains their radicality, there is at least occasionally a possibility of far-reaching reorientation where results are dramatically positive or negative (Tushman & Romanelli, 1985).

H11: Under interactive learning: a) centrality tends to be low, and b) radicality tends to be modest.

Interactive learning takes place under many constraints. It has local objectives, local effects, and is carried out mostly by middle managers. So its centrality is limited, as is its ability to effect fundamental change. Paradoxically, the political fragmentation that surrounds much interactive learning induces some organizational members to challenge basic assumptions. If these renegades can translate their skepticism into action, they may initiate the first steps towards significant transformation.

H12: Under structural learning: a) radicality tends to be low.

Routine-driven learning may concern local practices such as quality control, or it may pervade more fundamental things such as capital budgeting, innovation, marketing strategy, etc. (Nelson & Winter, 1982). But routines reify assumptions and induce habitual, unreflective behavior, thereby limiting the scope for radical changes (Hannan & Freeman, 1984; Miller & Friesen, 1984a; Tushman & Romanelli, 1985). Paradoxically, “normal organizational routines sometimes provide the equivocal experiences which lead to second order learning and change” (Lant & Mezias, 1992, p. 64). They may, for example, indicate that performance is failing to meet aspiration levels and that some fundamental reorientations are required.

H13: Under institutional learning: a) centrality tends to be high, b) radicality tends to be low, and c) knowledge tends to precede action.

Some forms of institutional learning indoctrinate organization members with basic values that may turn out to be of central importance. But such indoctrination presumes that the elite have already formulated or embraced the values that they will be imparting to others – those who actually do the learning. Hence, values may already be reflected in the basic orientations at the top of the organization even before indoctrination takes place (Selznick, 1957; Scott, 1995).

Mode Combinations and Sequences

We noted at the outset that several kinds of learning can exist in any one organization at any time. But it may be that some combinations are more common than others because they stem from the same basic approaches and worldview. Methodical kinds of learning, for example, may go together: strategic plans formulated by analysis might be implemented by routines and renewed by experiment. All of these modes of learning rely on managerial worldviews that value method, hard information, orderly deduction, and systematic feedback.

Emergent modes sometimes may go together as well since they all favour intuition, induction and subjectivity. Where synthetic learning has resulted in a new organizational vision, for instance, managers may try to indoctrinate members with that vision by fostering institutional learning (Feldman & March, 1981; Scott, 1995). Moreover, the provocative information managers gather via interactive learning may stimulate insightful synthesis.

H14: a) Methodical modes of learning will often occur simultaneously in the same organization; b) emergent modes of learning will tend to occur simultaneously in the same organization.

The six modes of learning may even vary in prevalence at different phases of the organizational life cycle. For example, synthetic or analytical learning may be common during the birth phase as managers try to develop a strategy (Scott, 1971). Institutional learning may then take place to impart elite visions and values to the larger membership (Deal & Kennedy, 1982). Experimental learning then might occur during the growth phase as managers try to develop and extend their strategic recipe (Greiner, 1972). Structural learning also may take place during the later parts of the growth phase and throughout the maturity phases to exploit an established strategy (Miller & Friesen, 1984b). Finally, in maturity and decline, as the organization becomes more politically complex and heterogeneous, interactive learning may become common. Hence, our final speculation:

H15: As organizations progress through the life cycle, they will tend to increase their use of the learning modes in approximately the following sequence: synthetic and analytic, institutional, experimental, structural, and interactive.


The significance of any typology can be assessed in at least three ways: first, a typology must capture important conceptual distinctions that suggest different causes or effects of what is being classified; second, it must derive types that occur very regularly in reality; and third, it must generate testable predictions whose investigation will enlighten academics and practitioners (Miller & Friesen, 1984a). The current typology, although tentative, appears to fulfil the first and perhaps the second of these criteria. But empirical research is needed to assess its adequacy in meeting the third.

Like all typologies, ours has much arbitrariness. It begins with fundamental distinctions among dominant organizational paradigms, it extends and refines these contrasts using the literature on learning and decision making, and it derives six modes of learning. No doubt there are readers who will want to suggest learning modes not included in this typology: for example, learning by organizational populations via natural selection (Hannan & Freeman, 1977), or institutional learning by industries that increases the isomorphism of their practices (DiMaggio & Powell, 1983; Fligstein, 1985). Such modes could extend this typology beyond the organizational level of analysis. Even at the organization level, there are bound to be ways of learning we have not considered. Still, we believe the modes identified here may help researchers to make useful distinctions in examining the sources, facilitators and performance implications of learning, and in relating learning processes to their appropriate contexts and outcomes.

Acknowledgement: The author would like to thank Christiane Demers, Peter H. Friesen, and Jamal Shamsie for their useful comments.


1. It is debatable whether extreme determinism such as that envisioned by population ecologists (Hannan & Freeman, 1977) admits of learning.

2. Given the complexity of the topic, we have chosen to focus on learning inside organizations rather than the collective learning that takes place in industries. Also, because we believe that learning encompasses both new knowledge and adaptive behavior, completely deterministic perspectives are beyond our scope.

3. This and all following hypotheses require qualification. Centrality and radicality of learning will be a function of how analysis or any other kind of learning is conducted. Where there is much reliance on existing formal systems and past assumptions and worldviews, analysis may lead only to incremental refinements – to practices that serve the same strategies, goals, and power-holders. On the other hand, where challenges are seen as more severe, or when managers see fit to move beyond the existing systems of information gathering and choose to listen to more objective outsiders, the chances are that deeper learning might be able to take place. Assumptions and worldviews might then be challenged by the new information, and a more skeptical light might be cast upon current objectives and strategies.


Ackoff, R. (1971). A concept of corporate planning. New York: Wiley Interscience.

Aguilar, F.J. (1967). Scanning the business environment. Boston: Harvard University Press.

Allison, G. (1971). Essence of decision. Boston: Little Brown.

Andrews, K.R. (1971). The concept of strategy. Homewood, IL: Irwin.

Ansoff, H.I. (1965). Corporate strategy. New York: McGraw-Hill.

Ansoff, H.I. (1979). Implementing strategic management. Englewood Cliffs, NJ: Prentice Hall.

Argyris, C. & Schon, D. (1978). Organizational learning: A theory of action perspective. Reading, MA: Addison Wesley.

Astley, W. G. & Van de Ven, A.H. (1983). Central perspectives and debates in organization theory. Administrative Science Quarterly, 28: 245-273.

Beer, S. (1981). Decision and control. London: Wiley.

Berger, P. & Luckmann, T. (1966). The social construction of reality. Harmondsworth: Penguin.

Braybrooke, D. & Lindblom, C. (1963). A strategy of decision. New York: Free Press.

Burgelman, R. (1990). Strategy making and organizational ecology. Pp. 164-181 in J. Singh (Ed.), Organizational evolution. Newbury Park: Sage.

Burrell, G. & Morgan, G. (1979). Sociological paradigms and organizational analysis. London: Heinemann.

Chaffee, E. (1985). Three models of strategy. Academy of Management Review, 10: 89-98.

Clark, B. (1956). Organizational adaptation and precarious values. American Sociological Review, 21: 327-336.

Clegg, S. (1989). Frameworks of power. London: Sage.

Cohen, M.D., March, J.G. & Olsen, J.P. (1972). A garbage can model of organizational choice. Administrative Science Quarterly, 17: 1-25.

Crozier, M. (1964). The bureaucratic phenomenon. Chicago: University of Chicago.

Cyert, R.M. & March, J.G. (1963). A behavioral theory of the firm. Englewood Cliffs, NJ: Prentice Hall.

Deal, T. & Kennedy, A.A. (1982). Corporate cultures: The rites and rituals of corporate life. Reading, MA: Addison Wesley.

DiMaggio, P.W. & Powell, W.W. (1983). The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48: 147-160.

Eden, C. (1992). Strategy development as a social process. Journal of Management Studies, 29: 799-812.

Feldman, M. & March, J.G. (1981). Information in organizations as signal and symbol. Administrative Science Quarterly, 26: 171-186.

Fiol, C.M. & Lyles, M. (1985). Organizational learning. Academy of Management Review, 10: 803-813.

Fligstein, N. (1985). The spread of the multi-divisional form among large firms. American Sociological Review, 50: 377-391.

Fredrickson, J. (1986). The strategic decision process and organizational structure. Academy of Management Review, 11: 280-297.

Garud, R. & Van de Ven, A.H. (1992). An empirical evaluation of the corporate venturing process. Strategic Management Journal, 13: 93-110.

Georgiou, P. (1972). A re-examination of the goal paradigm. Administrative Science Quarterly, 17: 1-24. Giddens, A. (1979). Central problems in social theory. London: Macmillan.

Grandori, A. (1984). A prescriptive contingency view of organizational decision making, Administrative Science Quarterly, 29: 192-209.

Greiner, L. (1972) Evolution and revolution as organizations grow. Harvard Business Review, (July-August): 37-46.

Halberstam, D. (1986). The reckoning. New York: Avon.

Hambrick, D.C. (1994). Top management groups. Pp. 171-214 in B.M. Staw & L.L. Cummings (Eds.), Research in organizational behavior, Vol. 16. Greenwich, CT: JAI Press.

Hannan, M.T. & Freeman, J. (1977). The population ecology of organizations. American Journal of Sociology, 82: 929-964.

Hannan, M.T. & Freeman, J. (1984). Structural inertia and organizational change. American Sociological Review, 49: 149-164.

Harrigan, K.R. (1986). Managing for joint venture success. Lexington, MA: Lexington Books.

Hart, S.L. (1992). An integrative framework for strategy making processes. Academy of Management Review, 17: 327-351.

Hedberg, Bo L.T. (1981). How organizations learn and unlearn. Pp. 8-27 in P.C. Nystrom & W.H. Starbuck (Eds.), Handbook of organizational design. London: Oxford University Press.

Herriot, S.R., Levinthal, D. & March, J.G. (1985). Learning from experience in organizations. American Economic Review, 75: 298-302.

Hrebiniak, L.G. & Joyce, W.F. (1985). Organizational adaptation: Strategic choice and environmental determinism. Administrative Science Quarterly, 30: 336-349.

Huber, George P. (1991) Organizational learning: The contributing processes and the literatures. Organization Science, 2: 88-115.

Janis, Irving. (1972). Victims of groupthink. Boston: Houghton Mifflin.

Kets de Vries, M. & Miller, D. (1987). Interpreting organizational texts. Journal of Management Studies, 24: 233-248.

Lant, T.K. & Mezias, S. (1992). An organizational learning model of convergence and reorientation. Organization Science, 3: 47-71.

Levitt, B. & March, J.G. (1988). Organizational learning. Annual Review of Sociology, 14: 319-340.

Lindblom, C. (1959). The science of ‘muddling through.’ Public Administration Review, 19: 79-88.

MacMillan, I.C. & Jones, P. (1986). Strategy formulation: Power and politics. St. Paul, MN: West.

March, J.G. (1989). Decisions and organizations. New York: Blackwell.

March, J.G. (1991). Exploration and exploitation in organizational learning. Organization Science, 2: 71-87.

March, J.G. & Olsen, J. (1976). Ambiguity and choice in organizations. Bergen, Norway: Universitetsfortlaget.

March, J.G. & Simon, H.A. (1958). Organizations. New York: Wiley.

Meyer, J. & Rowan, B. (1977). Institutionalized organizations: Formal structure as myth and ceremony. American Journal of Sociology, 83: 340-363.

Meyer, A. & Starbuck, W.H. (1991). Organizations and industries in flux: The interplay of rationality and ideology. Working Paper, University of Oregon.

Miller, D. (1987). The genesis of configuration. Academy of Management Review, 12: 686-701.

Miller, D. (1990a). Organizational configurations: Cohesion, change and prediction. Human Relations, 43: 771-789.

Miller, D. (1990b). The Icarus paradox: How exceptional companies bring about their own downfall. New York: HarperCollins.

Miller, D. (1993). The architecture of simplicity. Academy of Management Review, 18: 116-138.

Miller, D. & Friesen, P.H. (1984a). Organizations: A quantum view. Englewood Cliffs, NJ: Prentice Hall.

Miller, D. & Friesen, P.H. (1984b). A longitudinal study of the corporate life cycle. Management Science, 30: 1161-1183.

Mintzberg, H. (1973). Strategy making in three modes. California Management Review, 30(3): 11-24.

Mintzberg, H. (1989). Mintzberg on management. New York: Free Press.

Nelson, R. & Winter, S.G. (1982). An evolutionary theory of economic change. Cambridge, MA: Harvard University Press.

Nonaka, I. (1988). Toward middle-up-down management. Sloan Management Review, 29(3): 9-18.

Palmer, R. (1969). Hermeneutics: Interpretation theory in Schleiermacher, Dilthey, Heidegger and Gadamer. Evanston, IL: Northwestern University.

Pascale, R.T. (1989). Managing on the edge. New York: Simon & Shuster.

Perrow, C. (1986). Complex organizations: A critical essay. New York: Random House.

Peteraf, M.A. (1993). The cornerstones of competitive advantage: A resource based view. Strategic Management Journal, 14: 179-192.

Peters, T. & Waterman, R. (1982). In search of excellence. New York: Harper & Row.

Pettigrew, A.M. (1973). The politics of organizational decision making. London: Tavistock.

Pfeffer, J. (1981). Power in organizations. New York: Pitman.

Porter, M.E. (1980). Competitive strategy. New York: Free Press.

Porter, M.E. (1985). Competitive advantage. New York: Free Press.

Prahalad, C.K. & Hamel, G. (1990). The core competence of the corporation. Harvard Business Review, (May-June): 79-91.

Quinn, J.B. (1980). Strategies for change: Logical incrementalism. Homewood, IL: Irwin.

Radnitzky, G. (1968). Contemporary schools of metascience. Chicago: Regnery.

Ranson, S., Hinings, C.R. & Greenwood, R. (1980). The structuring of organization structures. Administrative Science Quarterly, 25: 1-17.

Schoemaker, P.J.H. (1993). Multiple scenario development: Its conceptual and behavioral foundation. Strategic Management Journal, 14: 193-213.

Scott, B.R. (1971). Stages of corporate development. Case #9: 371-294. Boston: International Case Clearing House.

Scott, W.R. (1995). Organizations and institutions. Beverly Hills, CA: Sage.

Selznick, P. (1957). Leadership in administration. New York: Free Press.

Senge, P.M. (1990). The fifth discipline: The art and practice of the learning organization. New York: Doubleday.

Shrivastava, P. & Grant, J. (1985). Empirically derived models of strategic decision making processes. Strategic Management Journal, 6: 97-113.

Silverman, D. (1970). The theory of organizations. London: Heinemann.

Simon, H.A. (1947). Administrative behavior. New York: Macmillan.

Starbuck, W.H. (1985). Acting first and thinking later: Theory versus reality in strategic change. Pp. 336-372 in J.M. Pennings and Associates (Eds.), Organizational strategy and change. San Francisco: Jossey Bass.

Starbuck, W.H., Greve, A. & Hedberg, B. (1978). Responding to crises. Journal of Business Administration, 9: 111-137.

Starbuck, W.H. & Milliken, F. (1988). Challenger: Fine tuning the odds until something breaks. Journal of Management Studies, 25: 319-340.

Stata, R. (1989). Organizational learning – The key to management innovation. Sloan Management Review, (Spring): 63-73.

Staw, B.M. (1977). The experimenting organization: Problems and prospects. Pp. 466-486 in B.M. Staw (Ed.), Psychological foundations of organizational behavior. Santa Monica, CA: Goodyear.

Steinbruner, J.D. (1974). The cybernetic theory of decision: New dimensions of political analysis. Princeton, NJ: Princeton University Press.

Steiner, G. (1979). Strategic planning. New York: Free Press.

Thompson, J.D. (1967). Organizations in action. New York: McGraw Hill.

Thompson, J.D. & Tuden, A. (1959). Strategies, structures, and processes of organizational decision. Pp. 195-216 in J.D. Thompson et al. (Eds.), Comparative studies in administration. Pittsburgh: Pittsburgh University Press.

Tushman, M.L. (1977). Special boundary roles in the innovation process. Administrative Science Quarterly. 22: 587-605.

Tushman, M.L. & Romanelli, E. (1985). Organizational evolution: A metamorphosis model of convergence and reorientation. Pp. 171-222 in L.L. Cummings and B.M. Staw (Eds.), Research in organizational behavior, Vol. 7. Greenwich, CT: JAI Press.

Weick, K.E. (1979). The social psychology of organizing. Reading, MA: Addison Wesley.

Whitley, R.D. (1991). The social construction of business systems in East Asia. Organisation Studies, 12: 1-28.

Whittington, R. (1992). Putting Giddens into action. Journal of Management Studies, 29: 693-712.

Wilensky, H. (1967). Organizational intelligence. New York: Free Press.

Williamson, O. (1975). Markets and hierarchies. New York: Free Press.

Zaltman, G., Duncan, R. & Holbek, J. (1973). Innovations and organizations. New York: Wiley.

COPYRIGHT 1996 JAI Press, Inc.

COPYRIGHT 2004 Gale Group