Looping learning: why real school improvement is recursive

More systems thinking for school leadership

Everyone wants school improvement, but few seem prepared to ask what it actually means—or, more importantly, how it happens. We pour time, money, and energy into strategies, frameworks, and interventions, all designed to drive progress. And yet, so much of what we do feels oddly repetitive. There’s a constant nagging deja vu inherent in so much of what schools do. Haven’t we been here before? Initiatives come and go, language changes, posters are updated, but underneath it all, the same problems resurface. The curriculum narrows. Morale dips. Pupils disengage. Improvement, it seems, is often little more than performance.

One of the bitterest ironies in education is that although schools are supposed to be places of learning, they often struggle to learn as organisations. And not because they lack intelligence. On the contrary—schools are full of highly educated professionals, many of whom are among the brightest and most committed in the public sector. But as Chris Argyris pointed out over three decades ago in Teaching Smart People How to Learn, intelligence is no guarantee of growth. In fact, it can be a serious obstacle.

Argyris’s core insight is unsettling: the smarter and more successful we are, the less likely we are to question our own reasoning. We’ve learned to win praise for being right, not for being curious. And when things go wrong - when lessons flop, results dip, or behaviour unravels - we tend to look outward, not inward: it’s the kids, it’s the curriculum, it’s the leadership. Anything but us.

This kind of reasoning can easily become institutionalised. Schools, like many professional organisations, reward - and therefore prioritise - surface-level fixes. We engage in what Argyris calls single-loop learning: we tweak the timetable, rewrite a policy, add a new behaviour system. But we rarely ask the deeper questions. Why do we think this approach works? What assumptions underpin our interventions? What might we be getting wrong? These are the hallmarks of double-loop learning, and they are surprisingly rare.

[A] thermostat that automatically turns on the heat whenever the temperature in a room drops below 69°F is a good example of single-loop learning. A thermostat that could ask, "why am I set to 69°F?" and then explore whether or not some other temperature might more economically achieve the goal of heating the room would be engaged in double-loop learning.

A school that responds to declining exam results by increasing revision sessions or introducing new intervention groups is engaging in single-loop learning; it’s correcting the error without questioning the underlying approach. But a school that asks, Is prioritising internal data the best way to improve outcomes? or Is our curriculum design contributing to a shallow understanding of content? - and, crucially, has systems in place to ensure these questions are asked and answered - is engaging in double-loop learning. It’s not just fixing the problem, it’s interrogating the assumptions that define the problem in the first place.

The purpose of double-loop learning is not just to tackle problems as they arise but to change our mental models about what those problems are. It is designed to add to, to complicate, our mental models of reality, to give us the wherewithal to change our minds.

So, if double loop learning leads to schools designing better systems for improving, why is it so uncommon? Because questioning ourselves is uncomfortable. It’s far easier to manage optics than to interrogate reasoning. Far easier to appear as if you’re improving than to confront the possibility that what you’re doing isn’t working. In a high-stakes environment - one where judgment is swift and accountability often punitive - it’s no wonder schools opt for safety. Too often, the unspoken rationale is that we make decision for our own convenience rather than on what’s best for the children and the communities we serve.

Single-loops are more liekly to lead to repition than to genuine improvement. We cycle through the same ‘new’ initiatives every few years, each one repackaged with fresh acronyms and glossy posters. Deep down, we know it’s all a bit performative. But the system rewards performance, so we comply. We write beautiful policies that look good to inspectors and other extenrnal audiences but never really reach the classroom. We rehearse answers to anticipated questions rather than reflect on uncomfortable truths. We busy ourselves with producing the illusion of progress and improvement, rather than going through the pain required to actually improve.

If this never worked we’d soon stop. The fact single loop solutions do sometimes appear effective fools into doubling down our investment into them. When they appear to work it as often despite rather than because of our actions. We may hear that they worked there (without understand everything that contributed to their success) and assume they will work equally well here. We things we work out we ignore the role of chance and assume success is due to our efforts but when they don’t we look for other factors to blame. We insulate ourselves from reality.

To shift this, we should move from thinking of schools as machines to be tuned and start thinking of them as learning - self-improving - organisations. That means embedding the habits of double-loop learning: honest self-inquiry, critical reflection, intelligent accountability processes and the courage to change when the evidence says we should.¹

It also means rethinking professional development. Despite the wild popularity of instructional coaching models currently sweeping schools, most CPD is still top-down, content-heavy, and assumption-laden. I’d go further, most instructional coaching is also top down and assumption-laden. Unless our CPD programme helps staff explore how they think about their own thinking - unless it challenges the mental models that guide practice - it won’t make much difference. Focussing on helping teachers become a bit better at teaching is insufficient. We also need systems that focus on recursion. Or, to dredge up and repurpose a tired old formulation, learning how to learn.

What if school improvement were recursive?

Recursion, in systems thinking, is the idea that patterns repeat themselves at different levels of a system. A cell resembles an organ, which resembles the organism. A microcosm mirrors the whole. In a recursive model of school improvement, the processes that drive change at the top must also exist at every layer beneath. What happens in a classroom mirrors what happens in a department, which reflects the school as a whole. If leadership is reflective but classrooms are performative, we have a break in the chain. If teachers ask students to be critical but leaders demand compliance, we’re modelling contradiction. If leaders value appearance over substance, so will middle leaders. If departments avoid hard questions in meetings, teachers will do the same in lessons. If students are taught to “get the right answer” without reflecting on their thinking, don’t be surprised when staff default to compliance in in their own ‘professional development’. Culture isn’t declared, it has to be enacted, recursively.

And if learning is to happen - substantive learning that questions assumptions, proposes new questions and offers templates for evaluating answers - it has to happen at every level. The school must become a learning organisation not by sending teachers on courses, but by building structures where reflective inquiry is the norm. The headteacher who reviews behaviour data without interrogating the school’s assumptions about discipline is no different from a pupil who asks for the mark but never reads the feedback. Both are trapped in single loops with no change to mental models.

To break out of this cycle, for school improvement to be more recursive and reflective, we should consider making three changes.

First, we need feedback loops that run in both directions. Not just leaders evaluating teachers, but teachers feeding back on policies. Not just schools measuring student outcomes, but students reflecting meaningfully on the processes they experience. (You might think your retrieval routines are great, but if students are able to tell you they aren’t learning anything, they’re definitely not.) Hierarchical, top down structiures are great for imposing compliance but fatally flawed when it comes to generating creativity and criticality. Recursion requires that the same questions are being asked at every level: Is this working? How do we know? What are we assuming?

Second, we need a shift from performance to substance. Many schools have become expert at simulating the appearance of improvement: branded lesson slides, neat curriculum maps that artificially impose the appearance of coherence on a random progression from one damn thing to the next, rehearsed inspection responses about intent, implementation and impact. But all these are single-loop artefacts. A double-loop school would ask harder questions: Why do we teach this? Why in this way? What actually needs to be on a slide to promote the kind of thinking we want students too enage in? How will leaders actually know whether the curriculum is effective? A school obsessed with polishing its proxies might produce a gloss of competence, but it’s unlikely to be one that’s focussed on students - especially those from the most disadvantaged backgrounds - achieving real success.

Third, we need leaders who embrace uncertainty and model fallibility. The recursive model demands psychological safety: the freedom to question, to admit ignorance, to change one’s mind. If senior leaders can’t do this - if they treat dissent as disloyalty and feedback as threat - then the rest of the school will follow suit. Reflection will be replaced with compliance. Inquiry will give way to performance. And the system will lose the capacity to learn.

All this demands a kind of intellectual humility that’s rare in school leadership. It means accepting that you don’t know best, and that what matters most might be happening at levels of the system you never see. It means recognising that some of the most powerful improvement work doesn’t actually look like improvement. When we’re in it, we more likely to see uncertainty, questions, even mess. But recursion isn’t chaos. It’s gradually converging coherence. If we can trust that designing a double loop process that explicitly builds in the assumption checking ‘how do we know?’ questions and commit to holding ourselves, publicly, to account, then we increase the probability that we get to where we hope to go.

Accepting and embracing uncertainty is deeply uncomfortable. We’re trained in the belief what’s needed is firm, decisive action and unhesitating confidence. This might make those around feel reassured, but it will not lead to a culture where schools can improve beyond the most superficial levels. We can challenge certainties about uncertainty by asking whether school leaders should encourage dissent, or suppress it? Should they reward honesty, or compliance? The answers might seem obvious but they are in direct opposition to what we usually experience. To repeat, a culture of learning requires the psychological safety to admit mistakes, ask naïve questions, and experiment without fear of being labelled as ineffective.

Recursion reframes culture. We often speak of embedding values like high expectations, curiosity, or respect. But values aren’t embedded by posters or assemblies, they’re embedded by repetition. If senior leaders treat middle leaders with cynicism and box-ticking evaluation process, it should be no surprise when middle leaders treat teachers the same way, with teachers, in turn, modelling the same to students. Culture is recursive. The way we behave at the top echoes all the way down the system. It has to be instantiated in every conversation, every meeting, every corridor.

This doesn’t mean lowering standards or abandoning rigour. Quite the opposite. It means having the confidence to say, “We don’t know if this will work so we’re going to find out.” It means privileging what’s true over what’s popular, and what’s effective over what’s expedient. That takes intellectual humility and moral courage. But it also takes structures that support learning over time: coaching, inquiry cycles, curriculum reviews grounded in actual student work, not just glossy overviews.

If schools want to be better - not just look better - they have to take organisational learning more seriously. That starts with recognising that being smart is not the same as being right, and that improvement begins not with answers, but with better questions.

Until then, we’ll continue to mistake motion for progress and compliance for professionalism. We’ll keep fixing problems without asking whether we’re solving the underlying causes. And we’ll continue to run schools that teach children to learn, while quietly forgetting how to do it ourselves.

1. Quick and shameless plug for my book, Intelligent Accountability: creating the conditions for teachers to thrive