Chris Argyris:

Theories of Action,
Double-Loop Learning and
Organizational learning
 The work of Chris Argyris
(1923-) has influenced
thinking about the
relationship of people and
organizations,
organizational learning and
action research. Here we
examine some key aspects
Theories of action: theory in use and
espoused theory
Our starting point is Argyris and Schön’s (1974)
argument that people have mental maps with
regard to how to act in situations.
This involves the way they plan, implement and
review their actions. Furthermore, they assert
that it is these maps that guide people’s actions
rather than the theories they explicitly espouse.
What is more, fewer people are aware of the
maps or theories they do use (Argyris, 1980).
One way of making sense of this is to say that
there is split between theory and action.
However, Argyris and Schön suggest that two
theories of action are involved.
A theory of action is first a theory: ‘its
most general properties are properties
that all theories share, and the most
general criteria that apply to it – such as
generality, centrality and simplicity - are
criteria applied to all theories’ (Argyris and
Schön 1974: 4).
The distinction made between the two
contrasting theories of action is between
those theories that are implicit in what we
do as practitioners and managers, and
those on which we call to speak of our
actions to others. The former can be
described as theories-in-use. They govern
actual behaviour and tend to be tacit
structures.
Their relation to action 'is like the
relation of grammar-in-use to
speech; they contain assumptions
about self, others and environment -
these assumptions constitute a
microcosm of science in everyday
life' (Argyris & Schön 1974: 30).
The words we use to convey what
we, do or what we would like others
to think we do, can then be called
espoused theory.
When someone is asked how he
would behave under certain
circumstances, the answer he usually
gives is his espoused theory of action
for that situation.
This is the theory of action to which
he gives allegiance, and which, upon
request, he communicates to others.
However, the theory that actually
governs his actions is this theory-in-
use. (Argyris and Schön 1974: 6-7)
Making this distinction allows us to
ask questions about the extent to
which behaviour fits espoused
theory; and whether inner feelings
become expressed in actions.
In other words, is there congruence
between the two? Argyris (1980)
makes the case that effectiveness
results from developing congruence
between theory-in-use and espoused
theory
For example, in explaining our actions to a
colleague we may call upon some
convenient piece of theory. We might
explain our sudden rush out of the office
to others, or even to ourselves at some
level, by saying that a 'crisis' had arisen
with one of 'our' clients.
The theory-in-use might be quite different.
We may have become bored and tired by
the paper work or meeting and felt that a
quick trip out to an apparently difficult
situation would bring welcome relief.
A key role of reflection, we could argue,
is to reveal the theory-in-use and to
explore the nature of the 'fit'. Much of the
business of supervision, where it is
focused on the practitioner’s thoughts,
feelings and actions, is concerned with the
gulf between espoused theory and theory-
in-use or in bringing the later to the
surface.
This gulf is no bad thing. If it gets too
wide then there is clearly a difficulty. But
provided the two remain connected then
the gap creates a dynamic for reflection
and for dialogue.
To fully appreciate theory-in-use
we require a model of the
processes involved.

To this end Argyris and Schön

(1974) initially looked to three
elements:
Governing variables: those dimensions
that people are trying to keep within
acceptable limits. Any action is likely to
impact upon a number of such variables –
thus any situation can trigger a trade-off
among governing variables.
Action strategies: the moves and plans
used by people to keep their governing
values within the acceptable range.
Consequences: what happens as a result
of an action. These can be both intended -
those actor believe will result - and
unintended. In addition those
consequences can be for the self, and/or
for others. (Anderson 1997)
Three Elements
Where the consequences of the strategy used are
what the person wanted, then the theory-in-use
is confirmed. This is because there is a match
between intention and outcome. There may be a
mismatch between intention and outcome.
In other words, the consequences may be
unintended. They may also not match, or work
against, the person’s governing values. Argyris
and Schön suggest two responses to this
mismatch, and these are can be seen in the
notion of single and double-loop learning.
Single-loop learning

For Argyris and Schön (1978: 2) learning

involves the detection and correction of
error. Where something goes wrong, it is
suggested, an initial port of call for many
people is to look for another strategy that
will address and work within the governing
variables.
In other words, given or chosen goals,
values, plans and rules are operationalized
rather than questioned. According to
Argyris and Schön (1974), this is single-
loop learning.
Double-Loop Learning
An alternative response is to question to
governing variables themselves, to subject them
to critical scrutiny. This they describe as double-
loop learning. Such learning may then lead to an
alteration in the governing variables and, thus, a
shift in the way in which strategies and
consequences are framed.
Thus, when they came to explore the nature of
organizational learning. This is how Argyris and
Schön (1978: 2-3) described the process in the
context of organizational learning:
When the error detected and corrected
permits the organization to carry on its
present policies or achieve its presents
objectives, then that error-and-correction
process is single-loop learning.
Single-loop learning is like a thermostat
that learns when it is too hot or too cold
and turns the heat on or off. The
thermostat can perform this task because
it can receive information (the
temperature of the room) and take
corrective action.
Double-loop learning occurs when
error is detected and corrected in
ways that involve the modification of
an organization’s underlying norms,
policies and objectives.
This process can be represented
quite easily by a simple amendment
of our initial representation of
theory-in-use.
Single-loop learning seems to be present
when goals, values, frameworks and, to a
significant extent, strategies are taken for
granted.
The emphasis is on ‘techniques and
making techniques more efficient’ (Usher
and Bryant: 1989: 87) Any reflection is
directed toward making the strategy more
effective.
Double-loop learning, in contrast,
‘involves questioning the role of the
framing and learning systems which
underlie actual goals and strategies’
(op. cit.). In many respects the
distinction at work here is the one
used by Aristotle, when exploring
technical and practical thought.
The former involves following routines and
some sort of preset plan – and is both less
risky for the individual and the
organization, and affords greater control.
The latter is more creative and reflexive,
and involves consideration notions of the
good. Reflection here is more
fundamental: the basic assumptions
behind ideas or policies are confronted…
hypotheses are publicly tested… processes
are disconfirmable not self-seeking
(Argyris 1982: 103-4).
The focus of much of Chris Argyris’
intervention research has been to explore
how organizations may increase their
capacity for double-loop learning.
He argues that double-loop learning is
necessary if practitioners and
organizations are to make informed
decisions in rapidly changing and often
uncertain contexts (Argyris 1974; 1982;
1990). As Edmondson and Moingeon
(1999:160) put it:
The underlying theory, supported by years
of empirical research, is that the reasoning
processes employed by individuals in
organizations inhibit the exchange of
relevant information in ways that make
double-loop learning difficult – and all but
impossible in situations in which much is
at stake.
This creates a dilemma as these are the
very organizational situations in which
double-loop learning is most needed.
The next step that Argyris and Schön take is to
set up two models that describe features of
theories-in-use that either inhibit or enhance
double-loop learning. The belief is that all people
utilize a common theory-in-use in problematic
situations.
This they describe as Model I – and it can be said
to inhibit double-loop learning. Model II is where
the governing values associated with theories-in-
use enhance double-loop learning.
Model I and Model II

Argyris has claimed that just about all the

participants in his studies operated from
theories-in-use or values consistent with
Model I (Argyris et al. 1985: 89).
It involves ‘making inferences about
another person’s behaviour without
checking whether they are valid and
advocating one’s own views abstractly
without explaining or illustrating one’s
reasoning’ (Edmondson and Moingeon
1999:161).
The theories-in-use are shaped by an
implicit disposition to winning (and to
avoid embarrassment). The primary action
strategy looks to the unilateral control of
the environment and task plus the
unilateral protection of self and others.
As such Model I leads to often deeply
entrenched defensive routines (Argyris
1990; 1993) – and these can operate at
individual, group and organizational levels.
Exposing actions, thoughts and feelings can make
people vulnerable to the reaction of others.
However, the assertion that Model I is
predominantly defensive has a further
consequence:
Acting defensively can be viewed as moving away
from something, usually some truth about
ourselves. If our actions are driven by moving
away from something then our actions are
controlled and defined by whatever it is we are
moving away from, not by us and what we would
like to be moving towards.
Therefore our potential for growth and
learning is seriously impaired. If my
behaviour is driven by my not wanting to
be seen as incompetent, this may lead me
to hide things from myself and others, in
order to avoid feelings of incompetence.
For example, if my behaviour is driven by
wanting to be competent, honest
evaluation of my behaviour by myself and
others would be welcome and useful.
(Anderson 1997)
“It is only by interrogating and
changing the governing
values, the argument goes, is
it possible to produce new
action strategies that can
address changing
circumstances.”
Model I theory-in-use
Characteristics

The governing Values of Model I are:

Achieve the purpose as the actor

defines it
Win, do not lose
Suppress negative feelings
Emphasize rationality
Primary Strategies are:
Protect self and others unilaterally
Usually operationalized by:
Unillustrated attributions and evaluations
e.g.. "You seem unmotivated"
Advocating courses of action which
discourage inquiry e.g.. "Lets not talk
about the past, that's over."
Treating ones' own views as obviously
correct
Making covert attributions and evaluations
 Face-saving moves such as leaving
potentially embarrassing facts unstated
Consequences include:
Defensive relationships
Low freedom of choice
Reduced production of valid information
Little public testing of ideas

Taken from Argyris, Putnam & McLain Smith

(1985, p. 89
Chris Argyris looks to move people from a
Model I to a Model II orientation and
practice – one that fosters double-loop
learning. He suggests that most people,
when asked, will espouse Model II.
As Anderson (1997) has commented,
Argyris offers no reason why most people
espouse Model II. In addition, we need to
note that the vast bulk of research around
the models has been undertaken by
Argyris or his associates.
Significant Features of Model II
The significant features of Model II include the
ability to call upon good quality data and to make
inferences. It looks to include the views and
experiences of participants rather than seeking to
impose a view upon the situation.

Theories should be made explicit and tested,

positions should be reasoned and open to
exploration by others. In other words, Model II
can be seen as dialogical – and more likely to be
found in settings and organizations that look to
shared leadership.
It looks to:
Emphasize common goals and
mutual influence.
Encourage open communication, and
to publicly test assumptions and
beliefs.
Combine advocacy with inquiry
(Argyris and Schön 1996; Bolman
and Deal 1997: 147-8).
We can see these in the table below.
Model II characteristics

The governing values of Model II include:

Valid information
Free and informed choice
Internal commitment
Strategies include:
Sharing control
Participation in design and implementation
of action
Operationalized by:
Attribution and evaluation illustrated with
relatively directly observable data
Surfacing conflicting view
Encouraging public testing of evaluations
Consequences should include:
Minimally defensive relationships
High freedom of choice
Increased likelihood of double-loop
learning

Taken from Anderson 1997

As Edmondson and Moingeon (1999:162)
comment, employing Model II in difficult
interpersonal interactions ‘requires
profound attentiveness and skill for human
beings socialized in a Model I world’. While
they are not being asked to relinquish
control altogether, they do need to share
that control.
Organizational learning

Chris Argyris and Donald Schön suggest

that each member of an organization
constructs his or her own representation
or image of the theory-in-use of the whole
(1978: 16).
The picture is always incomplete – and
people, thus, are continually working to
add pieces and to get a view of the whole.
They need to know their place in the
organization, it is argued.
An organization is like an organism
each of whose cells contains a
particular, partial, changing image if
itself in relation to the whole.
And like such an organism, the
organization’s practice stems from
those very images. Organization is
an artifact of individual ways of
representing organization.
Hence, our inquiry into organizational
learning must concern itself not with static
entities called organizations, but with an
active process of organizing which is, at
root, a cognitive enterprise.
Individual members are continually
engaged in attempting to know the
organization, and to know themselves in
the context of the organization. At the
same time, their continuing efforts to
know and to test their knowledge
represent the object of their inquiry.
Organizing is reflexive inquiry….
[Members] require external
references. There must be public
representations of organizational
theory-in-use to which individuals
can refer.
This is the function of organizational
maps. These are the shared
descriptions of the organization
which individuals jointly construct
and use to guide their own inquiry….
Organizational theory-in-use,
continually constructed through
individual inquiry, is encoded in
private images and in public maps.
These are the media of
organizational learning.

(Argyris and Schön 1978: 16-17)

With this set of moves we can see
how Chris Argyris and Donald Schön
connect up the individual world of
the worker and practitioner with the
world of organization.

Their focus is much more strongly on

individual and group interactions and
defenses than upon systems and
structures.
By looking at the way that people
jointly construct maps it is then
possible to talk about organizational
learning (involving the detection and
correction of error) and
organizational theory-in-use.
For organizational learning to occur,
‘learning agents’, discoveries,
inventions, and evaluations must be
embedded in organizational memory’
(Argyris and Schön 1978: 19).
 “If it is not encoded in the
images that individuals have,
and the maps they construct
with others, then ‘the
individual will have learned
but the organization will not
have done so’ (op. cit.).”
In this organizational schema single-
loop learning is characterized as
when, ‘members of the organization
respond to changes in the internal
and external environment of the
organization by detecting errors
which they then correct so as to
maintain the central features of
theory-in-use’ (ibid.: 18). Double-
loop learning then becomes:
… those sorts of organizational
inquiry which resolve incompatible
organizational norms by setting new
priorities and weightings of norms, or
by restructuring the norms
themselves together with associated
strategies and assumptions. (Argyris
and Schön 1978: 18)
The next step is to argue that individuals
using Model I create Organizational I (O-I)
learning systems. These are characterized
by ‘defensiveness, self-fulfilling
prophecies, self-fuelling processes, and
escalating error’ (Argyris 1982: 8).

O-I systems involve a web of feedback

loops that ‘make organizational
assumptions and behavioural routines self-
reinforcing – inhibiting “detection and
correction of error” and giving rise to
mistrust, defensiveness and self-fulfilling
prophecy’ (Edmondson and Moingeon
1999:161).
In other words, if individuals in an
organization make use of Model I learning
the organization itself can begin to
function in ways that act against its long-
term interests. Indeed, in a very real
sense systems can begin to malfunction.

As Argyris and Schön (1996: 28) put it,

‘The actions we take to promote
productive organizational learning actually
inhibit deeper learning’. The challenge is,
then, to create a rare phenomenon – an
Organizational II (O-II) learning system.
Here we come to the focus of
organizational effort – the
formulation and implementation of
an intervention strategy.

This, according to Argyris and Schön

(1978: 220-1) involves the
‘interventionist’ in moving through
six phases of work:
Phase 1- Mapping the problem as clients see
it. This includes the factors and relationships that
define the problem, and the relationship with the
living systems of the organization.

Phase 2 - The internalization of the map by

clients. Through inquiry and confrontation the
interventionists work with clients to develop a
map for which clients can accept responsibility.
However, it also needs to be comprehensive.

Phase 3 - Test the model. This involves looking

at what ‘testable predictions’ can be derived from
the map – and looking to practice and history to
see if the predictions stand up. If they do not, the
map has to be modified.
Phase 4 - Invent solutions to the
problem and simulate them to explore
their possible impact.

Phase 5 - Produce the intervention.

Phase 6 - Study the impact. This allows

for the correction of errors as well as
generating knowledge for future designs.
If things work well under the conditions
specified by the model, then the map is
not disconfirmed.
By running through this sequence and
attending to key criteria suggested by
Model II, it is argued, organizational
development is possible.

The process entails looking for the

maximum participation of clients,
minimizing the risks of candid
participation, starting where people want
to begin (often with instrumental
problems), and designing methods so that
they value rationality and honesty.
Conclusion

How are we to evaluate these models

and line of argument? First, we can
say that while there has been a
growing research base concerning
the models and interventionist
strategy, it is still limited – and
people sympathetic to the approach
have largely undertaken it.
The process and the focus on
reflection-in-action does appear to
bear fruit in terms of people’s
connection with the exercise and
their readiness to explore personal
and organizational questions.
Second, it is assumed that ‘good’ learning
‘takes place in a climate of openness
where political behaviour is minimized’
(Easterby-Smith and Araujo 1999: 13).
This is an assumption that can be
questioned. It could be argued that
organizations are inherently political – and
that it is important to recognize this.
Organizations can be seen as coalitions of
various individuals and interest groups.
‘Organizational goals, structure and
policies emerge from an ongoing
process of bargaining and
negotiation among major interest
groups’ Bolman and Deal 1997:
175).

Thus, perhaps we need to develop

theory that looks to the political
nature of structures, knowledge and
information.
Here we might profitably look to
games theory, the contribution of
partisan and political institutions
(Beem 1999) and an exploration of
how managers can make explicit,
and work with, political processes
(Coopey 1998).

Perhaps the aim should be ‘to

incorporate politics into
organizational learning, rather than
to eradicate it’ (Easterby-Smith and
Araujo 1999: 13).
Third, and this might be my
prejudice, I think we need to be
distrustful of bipolar models like
Model I and Model II. They tend to
set up an ‘either-or’ orientation.
They are useful as teaching or
sensitizing devices, alerting us to
different and important aspects of
organizational life, but the area
between the models (and beyond
them) might well yield interesting
alternatives.
Fourth, the interventionist strategy is staged or
phased – and this does bring with it some
problems. Why should things operate in this
order.

Significantly, this does highlight a tension

between Argyris’s orientation and that of Schön
(1983). Schön in his later work on
reflection-in-action draws on his pragmatist
heritage (and especially the work of Dewey) and
presents the making of theory-in-action and the
expression of professional artistry in a far less
linear fashion.
Rather than there being phases, we could
argue that intervention of this kind
involves a number of elements or
dimensions working at once.
This said, the theorizing of theory-in-
action, the educative power of the models,
and the conceptualization of organizational
learning have been, and continue to be,
significant contributions to our
appreciation of processes in organizations.
The notion of ‘double-loop learning’ does
help us to approach some of the more
taken-for-granted aspects of organizations
and experiences.

It provides us with a way of naming a

phenomenon (and problem), and a
possible way of ‘learning our way out’
(Finger and Asún 2000). Argyris and
Schön have made a significant
contribution to pragmatic learning theory
(following in the line of Dewey 1933;
Lewin 1948, 1951; and Kolb 1984).
First, by introducing the term
‘theory’ or ‘theory in action’, ‘they
provide the function of abstract
conceptualization (see
experiential learning) ‘more
structure and more coherence’
(Finger and Asún 2000: 45).
Abstract conceptualization ‘becomes
something one can analyze and work
from’ (op. cit.). Second, through the
notion of ‘learning-in-action’ Argyris
and Schön rework the experiential
learning cycle.
Unlike Dewey’s, Lewin’s or Kolb’s learning
cycle, where one had, so to speak, to
make a mistake and reflect upon it – that
is, learn by trial and error – it is now
possible thanks to Argyris and Schön’s
conceptualization, to learn by simply
reflecting critically upon the theory-in-
action.

In other words, it is no longer necessary

to go through the entire learning circle in
order to develop the theory further. It is
sufficient to readjust the theory through
double-loop learning. (Finger and Asún
2000: 45-6)
This is a very significant
development and has important
implications for educators. In the
experiential learning model of
Kolb (1984) the educator is in
essence a facilitator of a person’s
learning cycle.
To this role can be added that of
teacher, coach or mentor, the person
who ‘helps individuals (managers,
professionals, workers) to reflect
upon their theories-in-action’ (Finger
and Asún 2000: 46).
... It is a significant development – but it
has gone largely unnoticed in the adult
education and lifelong learning fields. This
is a result, in part, of rather blinkered
reading by professionals and academics
within that area, and because Argyris and
Schön did not address, to any significant
degree, the arena directly (Argyris’s
continued to focus on organization and
management, and Schön’s on professional
thinking).
THE END