Organizational Learning: What is New?1
Edgar H. Schein
MIT Sloan School of Management
Table of Contents
In tackling this question, it is important to note at the outset that we have neither a very
good understanding of the word organizational nor of the word learning. We talk glibly
about organizational learning without, for example, taking into account the useful
distinction made by Craig Lundberg between "Organizational Learning" (OL) by which we
typically mean learning by individual and groups IN the organization vs. the "Learning
Organization" (LO) by which we mean learning BY the organization as a total system. We
also need to distinguish both of these concepts from the concept of "Organization
Development" (OD) or as Golembiewski prefers "Organization Development and Change"
(ODC).2
Furthermore, most of the writers and practitioners in all three of these fields throw around
the concept of "Culture" (CU) as if we understood well what that concept means. I am
especially struck by the glibness of those who call for the creation of "learning cultures" or
"cultures of openness and trust," as if culture could be ordered up like an item on a
restaurant menu.
In trying to focus on what is new it became apparent to me that I had to decide first which
of these various concepts to tackle. I knew that a literature review would be relatively
useless because the literature is all over the map and even more confusing about
what OL, LO, ODC, or CU are really all about. One reason for this confusion is that we
have so many methodologies and paradigms for looking at these phenomena.
I have reached the conclusion over the last several decades that the only way out of this
confusion is to go back to real data, based on intensive observation of real phenomena. I
call this the clinical approach and will begin this paper by attempting to characterize its
unique features (Schein, 1987).
The clinical approach can be described in terms of the six points on
Chart 1:
a) The data to be analyzed must be based on intensive indepth observations of real
phenomena, preferably in crucial cases of learning or change that are based on real
problems to be solved in the "here and now".3
b) The data always include the effects of interventions even when the intervention is as mild
as the observer showing up at the scene of the action and asking a question or two. In fact,
without intervention you cannot decipher how the system really works, as Kurt Lewin
reminded us long ago. On the other hand, the best data on how the system works is often
revealed in how it reacts to our interventions.
c) The clinical approach always assumes a better or a worse state and is, therefore,
intrinsically concerned with both pathology and health. Even if the observer attempts to be
neutral and nonevaluative, it will be apparent that the system being studied is concerned
about better and worse states, so the observer must have some model of pathology and
health in mind. One crucial way of gathering data, then, is to conduct post-mortems after
the members of the system have taken some action.
d) Instead of looking for statistical regularities the clinical approach assumes that one's
theoretical model must take into account all the deviant cases and be able to explain them.
This position focuses one on puzzles and anomalies as the potentially most productive loci
for insights.
e) Concepts and theories that arise out of the observations must deal with the real dynamics
of the system and must, therefore, be process oriented as well as structure oriented.
f) Where individuals, groups, and larger organizational systems are involved, one must
focus on the systemic dynamics and not get caught up with oversimplified linear causal
models.
The clinical perspective grows out of work with clients who need help. This initial bias
toward the client is a disadvantage for knowledge generation because it limits what the
clinical observer can observe, but it has the great advantage that if the client wants help, he
or she de facto licenses the observer to dig deeper, to ask embarrassing questions, and to
delve into areas that would ordinarily be concealed. If one is looking for the "real"
dynamics of what is going on in the system, one is therefore better off with the clinical
perspective.
As I have applied this perspective to my own experience with OL, LO, ODC, and CU I
have reached the conclusion that we have spent far too much time on learning by
individuals and groups, what is typically the approach of OL and ODC, and far too little on
understanding the deeper dynamics of the Learning Organization (LO) and the role of
Culture (CU) in those dynamics. I will therefore focus the rest of this paper on the latter
two domains.
Organizations are complex systems. Before we can decide how organizations learn, we
must define what we mean by systemic health and learning, as defined in Charts
2 and 3.
Health is a difficult concept, whether we apply it to individuals, groups, or larger systems,
but efforts over the last 50 years have yielded some general definitions that seem to hold up
(Jahoda, 1958; Bennis, 1962).
The first important point to note is that systemic health can only be understood as a
combination of four factors, each of which must be present to some degree: 1) a sense of
identity, purpose, or mission; 2) a capacity on the part of the system to adapt and maintain
itself in the face of internal and external changes; 3) a capacity to perceive and test reality;
and 4) some degree of internal integration or alignment of the sub-systems that make up the
total system.
In a sense these four conditions are a prerequisite for learning or can be thought of as the
basic of "capacity to learn." When we apply the learning concept to any complex system we
note a very important distinction that has been made by most theorists, the difference
between single loop and double loop learning (Argyris and Schon, 1996), or what Senge
(1990) has called the difference between adaptive and generative learning, or what others
have called the difference between 1) maintenance and growth vs. 2) transformation.
When one applies this notion to complex organizations that are systems composed of many
sub-systems each of which is composed of many individuals, one can see that the total
organization's capacity to maintain itself and grow, to continue to act effectively in the face
of changing circumstances, depends upon the creation of a set of shared assumptions that
cut across the subsystems and that survive in spite of changes in the individual membership
of the sub-systems, i.e. the culture.
In other words the culture (CU) of the organization is both the consequence of the
organization's prior experience and learning, and the basis for its continuing capacity to
learn. What the organization can or cannot do will depend very much upon the actual
content of its culture and how that culture aligns or integrates the various sub-cultures of its
sub-systems.4 And the long-range adaptability of the organization will depend upon its
ability to perpetuate the core elements of its culture through socialization processes, while
maintaining enough slack to allow for the evolution of new cultural assumptions to take
into account new ideas.
What then do we mean by generative learning or transformation of a system? If the
organization's "knowledge" both explicit and tacit ("know-how") is embedded in the
culture and in the alignment of its sub-cultures, then it follows that transformation is
tantamount to a change in the culture itself--a change in the organization's sense of identity,
its goals, its core values, its primary ways of working, and so on (Schein, 1985, 1992).
5
Cultural assumptions provide stability and meaning to our daily life. They structure our
perceptions and thoughts, and they tell us how to evaluate and feel about things. It follows,
therefore, that if some of those assumptions need to change because they are out of line
with new data about the external or internal environment, such change will be preceded by
a period of anxiety, and that anxiety will produce denial and various other kinds of
defensive resistance to change. But this resistance to change is normal and must be
sensitively dealt with.
In another context I have argued that the management of this kind of "normal" resistance to
change involves at the individual level the management of two kinds of anxiety: l)
survival anxiety or the anxiety that if I do not change I will no longer be able to get
along, or keep my job, or maintain my sense of identity and competence; 2) learning
anxiety or the anxiety that if I do attempt to learn or change I will lose my
identity and sense of competence.6
For learning to occur at the individual level, then, survival anxiety must be higher than
learning anxiety, and this can be achieved by one of two ways. The leader as change agent
can 1) escalate the survival anxiety which, however, risks even more denial and
defensiveness and ultimately rejection of the leader, or the leader can 2) lower the learning
anxiety on the assumption that there is already sufficient survival anxiety in the system, or
3) the leader can do both at once.
These are difficult change management issues and require a deeper understanding of what
is involved in large complex systems change. Let us next try to shed some light on these
learning and change processes ir. the LO by analyzing several puzzles that the clinical
observer will become aware of.
Any department or group within the organization operates by the same systems dynamics
as the total organization. One of the most fundamental of those dynamics is the avoidance
of entropy, that is, to hold the system together to fulfill its functions in the face of entropic
forces on the part of the sub-systems within it. What is locally rational for a sub-system is
not necessarily aligned with the goals of the larger system so there will always be
disintegrative forces operating that must be managed by the larger system.
Given this dynamic, if a fundamental change occurs in one sub-system, e.g. a production
unit discovers that it can be much more effective by having employees manage themselves,
such a change will threaten the equilibrium in all of the neighboring systems. To avoid the
anxiety and possible upsetting of their equilibria, the neighboring systems will defend
themselves against the implications of the change.
To complicate matters, the source of stability for the subsystem cultures is not necessarily
within the total organization. Some of the sub-cultures have their reference base in the
occupational community from which their members have come. The professional
salesperson may identify with salespeople worldwide to a greater degree than with the sales
organization of his or her immediate employer. And the same is potentially true for each of
the functional units of an organization.
If then the production organization discovers that selfmanaged teams using group incentive
systems are more effective, this does not necessarily impress the sales organization whose
reference group is deeply embedded in the assumption of individual competition and
individual incentive systems. This line of thinking reveals the operation of three particular
sub-cultures that must be taken more seriously if we are to understand the dynamics of the
LO (See Chart 4).
Every organization has various sub-systems whose function it is to deliver the products or
services that derive from the organization's basic mission or primary task. These sub-
systems develop their own cultures and it is those cultures that often become the primary
target of organizational transformation efforts.
In most complex organizations the operators have learned that the world is systemically
inter-connected and that it takes cooperation and teamwork to increase effectiveness. The
operator culture takes it for granted that people make the difference and are the
organization's ultimate asset. When we see production units that are dysfunctional it is
usually because these insights are used to defeat management goals, i.e. that operators use
their teamwork skills to subvert and defeat management rather than work on its be ha If .
It is this culture that is typically the target of ODC and other change efforts such as total
quality and re-engineering. But paradoxically some of the most dramatic changes in this
culture do not diffuse to other parts of the organization or up the hierarchy. To understand
this lack of diffusion we must consider two other powerful cultures at work in
organizations.
In every organization there will be one or more groups whose job is to design the various
processes by whiCh the organization delivers its products and services, and by which it
rnaintains itself. Thus we have engineers or designers of production processes, sales
processes, financial processes, and so on. The members of these groups have received
their education outside the organization and they identify themselves on a global basis with
their professional reference groups to a greater extent than with their colleagues inside the
organization.
If we examine the essence of their culture we observe that its primary assumption is that
technical elegance and simplicity of solutions is a primary value and that solutions must be
efficient and error free. Since human are the most common source of errors, the best
solutions should be free of humans altogether. I remember vividly the two engineers sitting
in front of me on a flight to Seattle pointing out to each other as we were landing how
redundant and expensive the Cockpit crew was since the plane could be landed perfectly well by
computer.
What I have observed in a number of organizations is that when the operators begin to tout
more teamwork training and more support for teamwork, the "engineers" propose instead
to develop technical solutions for the problems that the operator team is trying to address.
We find then two sub-cultures that are not aligned and that, in fact, speak different
languages, have different values, and are oriented toward totally different goals.
Furthermore, we find that the ODC and LO oriented change agents tend to side with the
operators and label the engineers as not being humanistic enough, forgetting that it is the
engineering community that is, in a larger sense, the engine of major innovation in most
industries.
Instead of figuring out how to increase mutual understanding between engineers and
operators through creating real dialogues 7 between
them, we all too often call for the
"humanization" of a community whose core assumptions state that humans are the source
of error, noise, and messiness in operations. But these two cultures are not the
whole story.
Organizational survival and growth usually boils down in the end to an economic issue,
and the custodian of that issue is ultimately the CEO of the organization. CEO's the world
over live in a financial environment in which their attention is focused primarily on the
financial well being of their organization. The capital structures, the financial markets, and
the concerns of the stockholders all force attention to financial matters.
One exception to this focus can be observed in founders of organizations who hold an
ownership position or in CEO's whose
power base is in family or personal ownership rather than their Board of Directors.
8 But
CEO's who are accountable to their boards are likely to have learned from their own
experience and from fellow CEO's that they alone are accountable, that they cannot trust
information coming up through the system, that both the operator and engineering cultures
are too indifferent to costs and must therefore be controlled financially, and that people are
a cost rather than an asset. In this sense they collude with the engineers in preferring to run
their operations with the smallest number of people possible.
The impact of the CEO culture is that if some of the requirements for enhancing learning or
improving effectiveness increase costs or require some time off from daily operations they
will be refused on financial grounds. I remember vividly the reaction of a group of CEO's
to the proposition that for organizations to learn they have to create some "slack" to allow
people to learn new skills. In today's economic environment, argued the CEO's, one must
be lean and mean, and the very idea of "slack" is unthinkable. Yet learning is undermined
by that very attitude.9
Here again I observe the ODC and OL communities calling for the "humanization" of
CEO's instead of acknowledging that this powerful global cultural community is far too set
in its ways to really pay attention to such a call. What we need is better understanding of
the CEO culture and the ability to create a real Dialogue between them, the engineers and
the operators. Each of these cultures has a valid set of assumptions from their own point of
view and we as change agents and agents of learning must help each of these communities
to understand themselves and each other so that they can become better aligned in any given
organization.
In summary, one major reason why innovations in the operator culture do not diffuse is
because the engineering and CEO culture are fundamentally oriented toward other kinds of
concerns--technological elegance and financial viability.
Two new ideas are involved in answering this question. First, we must take more seriously
the growing evidence that learning is ultimately a social process that occurs in a community
of practice.10 Ideas are not enough. Until those ideas are embedded in the daily routines of
practitioners they have not really been "learned." And the evidence is mounting that this
final embedding occurs best "on the job" so to speak and in the actual social context in
which work is done. In a sense, one can say that apprenticeship is THE fundamental way
of learning, and mentoring rather than being an option in career development may be a
necessity.
This argument applies in particular to transformational learning where the new practices are
based on new cultural assumptions. It is only in the group context of working with fellow
learners that one can create enough psychological safety to permit the learners to overcome
their learning anxiety. And it is only in this group context that coaches can show learners
the new practices that are called for by the new cultural assumptions.
Consider, for example, how the new idea of self-managed groups would translate into
daily practices. Not only would the members of the work group have to learn to trust each
other and to communicate more openly with each other, but they would have to give up
notions of individual competitiveness and rewards in favor
of much vaguer concepts of team accountability and shared rewards.
Managers then become resources, consultants, and coaches and the
managerial job itself may come to rotate among group members
according to who has what knowledge and skill. Describing this
hardly tells one how to do it, especially if one has grown up in an
individualistic competitive culture.
The second part of the answer has to do with conceiving
organizational learning as a three stage process. In the MIT
Organizational Learning Center, Senge has put considerable emphasis
on creating a consortium of representatives of the sponsoring
companies. I have observed that some of the critical learning stages
occur in this consortium, as symbolized by the following event.
During one of the quarterly meetings of the representatives one of
them asked another whether they would be willing to come into
their company and do a workshop on systems dynamics. What struck
me about this was the fact that they did not ask anyone from MIT or
from any consulting company who are in this business, and that
made me realize that with most great ideas some version of the
following three step process takes place.
Step 1. The idea is articulated by academics. Usually such ideas are
not invented in academia but are the abstracting from observed data
about new practices in organizations of the essence of those new
practices. For example, McGregor's Theory X and Theory Y were
descriptions of and abstractions from the observed practices of more
and less effective managers.11
Academics write about and lecture
about these ideas and a small number of practitioners catch on or
recognize what is being talked about. But most of them
misunderstand or misapply the ideas until they are ready to embed
them in their organizations.
Step 2. The commercial consulting community recognizes the
potential of the idea and, if they sense practitioner interest, they develop
educational and training programs to embed the ideas in
their client organizations. But in the process of commercializing the
ideas they tend to standardize them which makes them to varying
degrees inapplicable to a particular organization's circumstances.
The programs are purchased by management, mandated for their
organizations and applied in a routinized fashion which leads to some
learning but no real embedding because of the irrelevance of the
ideas to the local situation. Furthermore, management becomes
dependent on the consulting organization and abdicates its
responsibility for determining the actual relevance of the ideas to
their organizational situation. That attitude leads the learners in the
operator culture to treat the new ideas and implied practices as the
latest "management fad" rather than something fundamentally
important.
If the operator culture recognizes the need for real change in their
operations, if they recognize that they need to operate from
fundamentally different assumptions to remain effective, they will
attempt to learn from the consultant programs but will discover that
the consultants do not have enough knowledge of the local situation
to be genuinely helpful in translating the ideas into new practices. It
is at that point that step 3 or the creation of communities of practice
comes into play.
Step 3. A learning consortium emerges. The learners in the local
community of practice recognize the need for new learning but must
find teachers and coaches who can appreciate the nuances of their
situation, yet who are not caught up in their local operator culture.
For this purpose they must find others who are in a similar
predicament yet who are outside their own organization.
The role of the academic and/or consultant could be to create and
nurture such consortia but put the emphasis on creating processes
that allow the members to learn from each other instead
of trying to teach the new practices directly. They could coach the
members on how to be effective teachers and coaches. The emphasis
for the "outsider" is to create learning processes and to function as
process consultants (Schein, 1987) to the consortium and the
community of practice.
In summary, if learning ultimately only occurs in a community of
practice, and if transformational learning involves changing of some
cultural assumptions, it must be mediated by a consortium of
practitioners who provide to each other the support and insight that
only a fellow practitioner could provide and, at the same, an outsider
perspective that permits local cultural assumptions to be surfaced
and examined.
Procter and Gamble started to transform their production systems
more than 30 years ago and announced a few years back that the
last of their unionized plants had finally adopted the new system,
after some 20 years of change efforts. Turnaround managers who
have gotten involved in major organizational transformations that
involve new cultural assumptions talk in terms of 10 to 15 year
programs. Why should these programs take so long?
To answer this question we must again go back to systems theory
and note that even if the "executive system," the sub-system with all
the formal power evolves new cultural assumptions, each of the sub-
systems must go through its own learning process before the entire
organization can be said to have learned. In fact, CEO's complain
bitterly about how little power they actually have to create major
change in their own organizations.
In many organizations one can see a complex scenario like the
following: the CEO has a conversion experience about how to
run an organization involving more delegation, team work,
selfmanaged groups, participation, open communications, horizontally
linked networks, and the like. He or she discovers that neither his
executive subordinates who are living out the CEO culture nor the
designers and engineers in the organization have any genuine insight
into the new ideas. In fact, both of these groups may regard the CEO
as having gone off the deep end and will subvert the new ideas to
the best of their ability.
How then should we conceive of the learning process in the LO? I
propose that we go back to some old ideas about how systems cope.
12
Chart 5 summarizes what I called an adaptive coping cycle that
applies to all kinds of systems and both kinds of learning, adaptive
and transformational. The argument is that not only the total
organization, but each sub-system within it, including each individual
learner within a sub-system must go through this cycle before they
can said to have learned, changed, or been transformed.
The steps in this cycle remind us that learning is a complex multi-
stage process and that it can be undermined at any stage by one of
the steps not being negotiated successfully. Or, to put it another way,
each step highlights certain kinds of organizational pathologies that
can arise, and for each of these pathologies one can also conceive of
the various remedies that the ODC and OL practitioners can apply.
Possible pathologies:
Possible Remedies:
Possible Pathologies
Possible Remedies
Possible pathologies
Possible Remedies
Possible pathologies
Possible Remedies
Possible pathologies:
Possible remedies
Possible pathologies:
Several implications derive from this form of analysis. First of all, it
is clear that every stage has to be successfully negotiated before one
can say that the organization has learned (LO). In ODC and OL we
often take credit for sub-group or individual learning without even
investigating whether or not the total organization has changed or
learned.
Second, it should be clear that each stage or step has its own
particular pathologies requiring its own particular remedies. In fact
the broad range of ODC and OL techniques can usefully be sorted by
their relevance to each step.15
Third, in complex systems composed of multiple inter-acting and
inter-dependent sub-systems the overall learning process will take
time because it must occur in each sub-system and must then be
integrated in the total LO process.
Fourth, the CEO and engineering cultures will hamper the integration
of learning within the organization because the CEO's will be
disproportionately obsessed with just financial
considerations and the engineers will be disproportionately obsessed
with just technical solutions. Mechanisms will have to be designed to
help CEO's and engineers to think more integratively about the health
of the systems in which they function.
Fifth, because the sub-systems of the organization develop cultures
of their own, transformational learning will always involve culture
change. Processes of diagnosing and evolving culture are therefore
central to any organizational learning processes.
In other words, the LO is a complex beast consisting of many systems
whose separate learning and change efforts must be coordinated and
integrated. It is time to accept the reality of this complexity and stop
oversimplifying systemic learning processes by touting particular
remedies like leadership, vision, re-engineering, total quality,
customer focus, systems thinking, and the like. Ultimately what is
new in this field is the recognition that transformational learning,
however necessary it may be, will require patient and careful
research before we can advocate any particular learning mechanisms
of how to do it.
Argyris, C. & Schon, D. A. (1996) Organizational Learning 11, Reading,
MA.: Addison Wesley.
Bennis, W. G. (1962). Toward a "Truly" Scientific Management: the
Concept of Organizational Health. General Systems Yearbook, 7, 269-
82.
Brown, J. S. & Duguid, P. Organizational Learning and Communities of
Practice. Organization Science. 1991, 2, pp. 40-57.
Cook, S. D. N. & Yanow, D. (1993) Culture and Organizational Learning.
J. of Management Inquiry, Vol. 2, No. 4, 373-390.
DiBella, A. J., Nevis, E. S. & Gould, J. M. (1996) Understanding
Organizational Learning Capability. Journal of Management Studies,
33, No. 3, 361-379.
Isaacs, W. N. (1993) Taking flight: Dialogue, Collective Thinking, and
Organizational Learning. Organizational Dynamics, Winter, 24-39.
Jahoda, M. (1958). Current concepts of positive mental health. N.Y.:
Basic Books.
McGregor, D. M. (1961) The Human Side of Enterprise. N.Y.: McGraw-
Hill.
Michael, D. N. (1973) On learning to plan and planning to learn. San
Francisco, CA.: Jossey Bass.
Rashford, N. S. & Coghlan, D. (1994) The Dynamics of Organizational
Levels. Reading, MA: Addison-Wesley.
Roth, G. (1996) Learning Histories: Using Documentation to assess and
facilitate organizational learning. MIT Center for Organizational
Learning.
Schein, E. H. (1965) Organizational Psychology. Englewood Cliffs, N.J.:
Prentice-Hall. (2d Ed., 1970; 3d Ed., 1980).
Schein, E. H. (1983) The role of the founder in creating organizational
culture. Organizational Dynamics, Summer, 13-28.
Schein, E. H. (1987) The Clinical Perspective in Fieldwork. Newbury
Park, CA.: Sage.
Schein, E. H. (1987) Process Consultation. Vol. 2. Reading, MA:
Addison-Wesley.
Schein, E. H. (1992) Organizational Culture and Leadership. 2d. Ed. San
Francisco, CA.: Jossey Bass.
Schein, E. H. (1993) How Can Organizations Learn Faster? The
Challenge of Entering the Green Room. Sloan Management Review, 34,
85-92.
Schein, E. H. (1997). The Three Cultures of Management: Implications
for Organizational Learning. Sloan Management Review, in press.
Senge, P. (1990) The Fifth Discipline. N.Y.: Doubleday, 1990.
l) Emphasis on in-depth observation of crucial cases of learning/change
2) Emphasis on studying the effects of interventions--
"you can't understand a system until you have tried to
change it"
3) Focus on pathology and post-mortems as a way of building a theory of health
4) Focus on puzzles and anomalies that are difficult to explain with current theory
5) Focus on building theory and empirical knowledge through developing concepts
that capture the real
Dynamics of systems
6) Focus on characteristics of systems and systemic dynamics
Type 1: Maintenance and growth--the capacity to continue to act effectively
in the face of changing circumstances
Type 2: Transformation--the capacity to change
fundamental elements of identity and goals, adopt
different basic assumptions and share them
1. Operator Culture. The "line"
2. The engineering community
culture (global)
3. Ceo culture (global)
1. Introduction
2. The Clinical Approach to Organizations, Learning, and Culture
3. The Learning Organization and the Role of Culture
4. Puzzle No. 1: Why do sub-system transformations rarely diffuse to the main system?
Why does culture change in one part of the organization not diffuse to other parts of the
organization's culture?
5. Puzzle No. 2: Why don't great ideas for organizational improvement such as
"empowerment," "teamwork," and "self- managed groups" catch on faster in managerial
practice?
6. Puzzle No. 3: Why does total organizational
learning take so long?
Step 1. Accurate sensing of changes in the external or internal
environment.
Step 2. Getting information to the right place where it can be acted
upon.
Step 3. Digesting and drawing the correct conclusions from the
information available
Step 4. Making internal transformations without creating undesirable
side effects
Step 5. Successfully exporting new products and services
Step 6. Obtaining feedback on whether the new behavior is achieving the
desired results--New sensing activity a la step 1.
Summary and Implications of Adaptive Coping Cycle
References
Chart 1
The clinical perspective on org. change and learning
Chart 2
What is systemic health?
Chart 3
Systemic Learning Capacity
Chart 4
Three cultures of management
Chart 5
The adaptive-coping cycle in complex systems16