Implementing Double-Loop Learning in Organizations: Lessons from Chris Argyris

Most organizations are reasonably good at fixing problems. A department misses its targets, so processes are adjusted. A policy fails to deliver, so a new circular is issued. A service complaint comes in, so a protocol is updated. This is the natural, comfortable rhythm of institutional life – and for routine challenges, it works well enough. But what happens when the same problems keep returning, no matter how many times they are “fixed”? What if the real issue lies not in the solution, but in the assumptions behind the question itself? This is the uncomfortable territory that Chris Argyris explored through his concept of double-loop learning – a framework that asks organizations to stop fixing symptoms and start questioning their own foundations.

Who was Chris Argyris? 🔗

Chris Argyris (1923-2013) was an organizational theorist and professor at Harvard Business School whose work fundamentally changed how we understand why organizations succeed or fail at learning and change. Drawing on earlier intellectual traditions from psychologists like Kurt Lewin and John Dewey, Argyris spent decades studying real organizations – observing how smart, well-meaning people consistently created conditions that blocked genuine improvement. His key insight was deceptively simple: learning involves increasing the effectiveness of our actions so as to produce intended results, but we are often unaware of how our own behaviors get in the way. From this foundation, Argyris – working closely with his colleague Donald Schön – developed the theory of single-loop and double-loop learning, which remains one of the most influential frameworks in organizational behavior and public administration.

The core idea: two kinds of learning 🔗

To understand double-loop learning, it helps to start with what it is not. The distinction between single- and double-loop learning draws on cybernetic theory, and the classic illustration is a thermostat. A thermostat monitors room temperature and switches heating on or off to maintain a set point. That is single-loop learning – detecting a deviation and correcting it within an existing framework. When someone physically changes the temperature setting, however, the system is being questioned at a deeper level. That is the double-loop moment.

Applied to organizations, single-loop learning addresses problems by adjusting actions within existing norms, while double-loop learning is required when those problems cannot be resolved through small adjustments because they involve the organization’s governing variables themselves. Governing variables are the underlying values, norms, and assumptions that shape how an organization defines its goals and measures its success. Double-loop learning asks: are these governing variables still valid? Are they producing the outcomes we actually want?

Single-loop learning in practice 🔗

Consider a government scheme to improve school enrolment in rural areas. Year after year, enrolment figures fall short of targets. The single-loop response would be to redouble implementation efforts – more field officers, more awareness camps, stricter monitoring. These are adjustments made within the existing framework without questioning whether the framework itself is sound. The assumption that attendance is the primary lever for educational outcomes goes unexamined. Single-loop learning is not inherently bad – it is efficient for routine, well-defined problems. But it becomes a trap when it is applied to complex, systemic challenges that actually require a different kind of inquiry.

Double-loop learning in practice 🔗

A double-loop response to the same enrolment problem would pause and ask harder questions: Why are families not sending children to school? Is the curriculum relevant to the lives of rural communities? Are the metrics we are using – enrolment numbers rather than learning outcomes – capturing what actually matters? Is our assumption that centralized delivery of education is effective still justified? Double-loop learning involves questioning the very framing and learning systems that underlie actual goals and strategies. It is more creative and more risky, but it is also the only kind of learning capable of addressing deep, recurring organizational failures.

Theories of action: what we say vs. what we do 🔗

One of Argyris’s most powerful contributions is his distinction between espoused theory and theory-in-use. Espoused theory is what an organization – or an individual – says it believes. Theory-in-use is the implicit set of assumptions that actually governs behavior, and it can only be inferred from observing actions, not from reading mission statements or policy documents.

Most organizations espouse values like transparency, participation, and continuous learning. Yet the same organizations may routinely suppress dissent, punish those who raise problems, and reward the appearance of competence over genuine inquiry. There is often a deep disconnect between what people say they believe – their espoused theory – and the beliefs that are actually implied by their behavior – their theory-in-use. This gap is not simply hypocrisy. It is, according to Argyris, a structural feature of how most people and organizations operate – and it is one of the primary reasons double-loop learning is so difficult to achieve.

Model I and Model II: the two operating systems 🔗

Argyris and Schön described two contrasting patterns of behavior – Model I and Model II – that either block or enable double-loop learning.

Model I: the defensive default 🔗

Model I represents the actions of almost all human beings, and its four key governing values – being in control of one’s environment, seeking to win and not lose, suppressing negative feelings, and acting as rationally as possible – do not vary significantly across age, culture, or organizational context. When individuals operating under Model I join an organization, they create what Argyris called defensive routines – habitual patterns of behavior that protect people from embarrassment or threat, but in doing so, systematically block honest inquiry.

Defensive routines are patterns of interpersonal interaction that people create to protect themselves from embarrassment and threat – and they manifest in many recognizable forms. Mixed messages (“be innovative, but don’t make mistakes”), undiscussable topics that everyone knows about but no one will name, and what Argyris called fancy footwork – elaborate rationalizations used to avoid confronting inconsistencies in one’s own behavior – are all symptoms of a Model I organization. Model I behaviors create organizational environments where defensive routines become entrenched, preventing genuine learning and adaptation.

Argyris also identified a striking paradox he called skilled incompetence: the more experienced and senior a person is, the more skilled they tend to be at defending their position and avoiding uncomfortable truths – and therefore, paradoxically, the less capable they are of genuine learning. As he observed, smart people often do not learn because they have too much invested in proving what they know and avoiding being seen as not knowing.

Model II: the productive alternative 🔗

Model II is not simply the opposite of Model I. Its key characteristics include balancing advocacy with inquiry by clearly stating one’s position while genuinely seeking to understand others’ perspectives, testing assumptions publicly by making one’s reasoning explicit and open to examination, taking joint responsibility for organizational challenges, and focusing on common interests rather than narrowly defined self-interest.

In a Model II environment, people are encouraged to surface disagreements, test their assumptions openly, and treat error not as something to be hidden but as a source of learning. When advocacy is combined with inquiry, there is a shift from unilateral attempts to defend one’s view to collaborative efforts to explore multiple perspectives and learn. This creates the conditions necessary for double-loop learning to occur.

Why organizations resist double-loop learning 🔗

Argyris identified several reasons why many organizations resist double-loop learning, including resistance to change, fear of failure, and an overemphasis on control. But the resistance runs deeper than organizational policy – it is embedded in the behavioral routines of individuals, reinforced by institutional culture, and often invisible to those who perpetuate it.

A particularly insidious form of resistance is what Argyris called the O-I learning system – an organizational environment dominated by Model I behaviors. O-I systems involve feedback loops that make organizational assumptions and behavioral routines self-reinforcing, inhibiting the detection and correction of error and giving rise to mistrust, defensiveness, and self-fulfilling prophecy. In other words, the very mechanisms an organization uses to manage itself can progressively undermine its capacity to learn.

Another structural barrier is what might be called the management trap: leaders who ask for honest feedback but respond to it with defensiveness or blame. Over time, this teaches employees that candor is risky, which in turn ensures that leaders never receive the accurate information they need to make good decisions. When these attempts at honest communication do not succeed, the most common responses are more single-loop learning and blaming others or the environment.

Implementing double-loop learning: what it actually requires 🔗

Achieving genuine double-loop learning in an organization is not a matter of running a workshop or drafting a new policy. Behavioral double-loop learning entails changes in values and frames governing how people interact – and this is a slow, difficult, and often uncomfortable process.

Stage one: awareness 🔗

The first requirement is developing awareness of one’s own defensive routines and the gap between espoused theory and theory-in-use. This sounds straightforward but is extraordinarily difficult in practice. To promote learning, defensive routines must be interrupted and undiscussables must be made discussable. Individuals need practice to develop the capacity for double-loop learning. Tools like the Ladder of Inference – which maps the mental steps between observable data and the conclusions we draw – and the left-hand column technique, which surfaces what people are thinking but not saying in conversations, are practical instruments Argyris developed to help individuals begin this process.

Stage two: new behavior under pressure 🔗

Knowing what Model II behavior looks like and being able to produce it under stress are very different things. Initial attempts to produce new behavior often result in what Argyris described as “gimmicks” – where the seemingly new behavior is used in the service of the old values. For example, a manager who asks for team input but has already decided on the answer, and is simply going through the motions of consultation, is producing a gimmick. Others in the organization almost always recognize it as such. Genuine change requires not just behavioral rehearsal but a real shift in the underlying values that drive behavior.

Stage three: institutionalizing new norms 🔗

The third stage of behavioral double-loop learning is to integrate the new behavior, as informed by new values and frames, into group norms and relationships so that it becomes the new normal. This means building organizational structures that support rather than undermine open inquiry – flattening hierarchies where possible, creating genuine feedback mechanisms, rewarding those who surface problems rather than those who conceal them, and ensuring that leaders visibly model the behavior they expect from others. If executives do not model Model II behaviors, the organization will not follow.

Double-loop learning in public organizations 🔗

The relevance of Argyris’s framework extends well beyond private sector management. In public sector organizations, learning cannot be separated from questions of power and influence, and different groups with different interests will attempt to shape the development and implementation of policy. This makes the collaborative inquiry that double-loop learning requires both more necessary and more politically charged.

Research on continuous improvement in public service organizations has found that the prevailing mindset in public sector organizations operates largely within a single-loop model, whereby insufficient reflection on governing logic occurs – and as a result, genuine learning and re-evaluation of the principles behind policy interventions does not take place, causing improvement efforts to eventually succumb to recurring problems.

The traditional bureaucratic structure – characterized by hierarchical authority, procedural compliance, and rule-bound decision-making – is in many ways a textbook Model I environment. Questioning established norms is institutionally risky. Admitting that a policy’s underlying assumptions may be flawed can feel politically and professionally dangerous. Double-loop learning has been used in leadership training, which prepares public servants to lead in their organizations by helping them dig deeply into the assumptions of their employees and of the institutions they serve.

There are, however, meaningful entry points for double-loop learning in public administration. Policy review mechanisms that go beyond compliance monitoring to examine whether policies are solving the right problems; citizen feedback systems that inform not just delivery but design; and inter-departmental learning forums that encourage structured reflection on governing assumptions – all of these create conditions in which double-loop learning can begin to take root. The important prerequisite is that senior leadership must genuinely want to hear what they may not want to know.

The continuing relevance of Argyris 🔗

In an era of rapid technological change, mounting public expectations, and increasingly complex governance challenges, the capacity to question one’s own assumptions is not a luxury – it is a survival skill. Argyris’s insights reveal why some organizations evolve while others stay stuck – and how genuine transformation requires examining the governing beliefs and rules behind actions, not just fixing visible problems.

Argyris identified defensive routines that inhibit learning at the individual, group, intergroup, and organizational levels, and he promoted a shift from defensive Model I reasoning to productive Model II reasoning as the path toward institutions that can genuinely learn and genuinely serve. The framework he built with Donald Schön remains one of the most rigorous and practically useful tools available for understanding why organizations fail to change – and what it would actually take to change that.

For students of public administration, Argyris offers more than a theory of organizational learning. He offers a mirror – one that asks whether the institutions meant to serve the public are capable of questioning their own assumptions with the same rigor they apply to the problems they are trying to solve.

What do you think? When a long-standing government policy continues to underperform despite repeated adjustments, at what point should administrators stop refining the approach and start questioning the assumptions behind it? And what conditions – in terms of culture, leadership, or institutional design – would need to exist for that kind of deeper questioning to actually be possible in a hierarchical public organization?

References

1. https://infed.org/dir/welcome/chris-argyris-theories-of-action-double-loop-learning-and-organizational-learning/
2. https://hbr.org/1977/09/double-loop-learning-in-organizations
3. https://scholar.valpo.edu/mssj/vol24/iss1/6/
4. https://actiondesign.com/resources/readings/double-loop-learning
5. https://en.wikipedia.org/wiki/Double-loop_learning
6. https://www.antoinebuteau.com/lessons-from-chris-argyris/
7. https://www.resilientminds.co.nz/wp-content/uploads/2024/06/ThoughtLeader_Argyris.pdf
8. https://thesystemsthinker.com/overcoming-defensive-routines-in-the-workplace/
9. https://banotes.org/administrative-thinkers/chris-argyris-organizational-behavior-learning-pioneer/
10. https://scholar.valpo.edu/cgi/viewcontent.cgi?article=1042&context=mssj
11. https://umbrex.com/resources/frameworks/organization-frameworks/argyris-single-and-double-loop-learning-model/
12. https://link.springer.com/article/10.1007/s11115-015-0303-5
13. https://link.springer.com/chapter/10.1007/978-3-030-92934-3_1
14. https://www.ebsco.com/research-starters/education/double-loop-learning
15. https://kanbanzone.com/2026/chris-argyris-organizational-learning-ii-double-loop-learning/