Years in pupil support work have shown that the role is, more than anything else, the work of holding a relationship steady while a young person figures out how to be in school. Some weeks that means sitting next to a Year 2 — sorry, an S2 — pupil for ten minutes during an English lesson, saying very little, while they decide whether they are going to pick up a pen. Some weeks it means catching a pupil at the door before assembly because their support assistant told me they had a rough morning at home and I want to be the first adult who notices. Some weeks it means standing outside a classroom door with a colleague, both of us holding a difficult conversation about a young person we have, between us, known for four years and still cannot quite reach.

The cognitive-instructional research base, which is where the original Learning Lens taxonomy was grounded, has remarkably little to say about any of this.

That silence is a feature of what the research set out to measure, not a failing of it. Rosenshine, Hattie, Black and Wiliam, Alexander, the EEF guidance reports, the cognitive-load tradition that runs from Sweller through Mayer to the modern working-memory researchers — this is some of the most carefully accumulated evidence in education, and it tells us, with a precision that very few human enterprises can match, what works in classrooms when classrooms are functioning. The research is not silent on the relational layer because the relational layer doesn't matter. It is relatively quiet about it because the cognitive layer is what the research methods of the last fifty years have been most successful at measuring. Effect sizes for retrieval practice, for formative assessment, for worked examples — these are stable, replicated, and operationally specific. Effect sizes for "the teacher noticed that the pupil was anxious before he picked up the pen" are harder to come by, not because the practice doesn't matter but because it is a different kind of evidence.

When I built the original taxonomy of fifty-three classroom practices across ten categories, I made a deliberate decision to anchor it on the cognitive-instructional layer. The reasoning was straightforward: the evidence was deepest there, the observational signal was clearest there, and a pupil-support leader trying to build a tool that schools across Scotland would use — and would trust — could not afford a research base that was thinner than the one practitioners were already familiar with. There was a second reason too, less articulated at the time. I knew, having spent years sitting in case conferences and reviewing pastoral plans and reading Scottish Government guidance on inclusion, how much harm can be done by an evaluative gaze trained on the relational work of teaching. The relational work is the most easily misread part of what teachers do. A quiet voice from a teacher to a heightened pupil can be co-regulation; it can also be giving up. A teacher staying late to talk to a child in the corridor can be relational repair; it can also be a boundary failure that the teacher is going to pay for, in their own wellbeing, six months down the line. Without a careful frame, observation evidence on this layer is not just unhelpful — it can actively harm professional trust.

So I left it out. I told myself, and I told other people, that the taxonomy was a cognitive-instructional taxonomy and that the relational and regulatory layer would have to be addressed elsewhere — in pastoral systems, in CLPL, in the wellbeing module that I was planning but had not yet built. I made the case, and I think the case was right at the time, that a taxonomy that pretended to capture everything would capture nothing well, and that the discipline of restraint was what made the cognitive layer credible.

But the longer I have used the product — the more I have watched colleagues use it, the more I have read what schools say they want from a self-evaluation tool, the more I have thought about how the cognitive moves land in real classrooms with real pupils — the less defensible the deliberate omission has felt. Not because the original reasoning was wrong, but because the reasoning has run out of ground to stand on. The relational layer is not separate from the cognitive layer. It is the precondition for it. A pupil who is dysregulated cannot retrieve, model, extend, or analyse. A pupil who arrives at a lesson with the relationship to their teacher unrepaired from yesterday's rupture is not in a position to access any of the practices the rest of the taxonomy describes. The cognitive layer assumes the relational layer is intact. When the relational layer is fragile, the cognitive moves do not land.

There is a particular kind of conversation, familiar to anyone who has worked in pupil support, that runs roughly like this. A teacher is concerned about a pupil. The pupil is bright. The pupil's prior data is good. The pupil's coursework is solid when they engage. But the pupil is not engaging, or is engaging erratically, or is engaging in some lessons but not others, and the teacher is asking what is going on. The honest answer, almost always, is that something has happened in the relational fabric of the school day that has made the pupil unable to bring their thinking brain to the lesson — a falling-out with a friend, a difficult morning at home, a rupture with another teacher that they have not been able to get past, a sensory environment that is overwhelming them, a transition from one phase of the day to another that they cannot make. None of this is captured by the cognitive-instructional taxonomy. All of it is captured, in some form, by the relational and regulatory practices that pupil-support leaders watch for, name, and try to teach to colleagues every day of their working lives.

So the taxonomy now includes a Relationships & Regulation category. Five practices: named greeting and active engagement at lesson start; co-regulation of dysregulated pupils; low-arousal response to heightened pupils; naming and validating emotional states; active repair after rupture. They sit alongside the cognitive practices, not above or below them. They are observable — visible, in the room, in the first ten minutes of a visit — and they are research-grounded, drawing on the developmental science (Tronick on rupture and repair, Shanker on co-regulation, Perry on the neurosequential model), the practitioner-research literature (Bomber, Geddes, Treisman), the low-arousal tradition from McDonnell and Studio III, and the converging evidence from the EEF's behaviour and SEL guidance reports. The research base is structured differently from the cognitive layer — more practitioner-research and case-study, less RCT, but mechanism-grounded and convergent across sources. I have written elsewhere, in the evidence base document that accompanies the lens, about exactly where the evidence is strong and where it is thinner. I am not interested in claiming a certainty the evidence does not support.

The lens — the way schools select these practices for focused observation — is gated behind training-completion. An observer cannot select the Relationships & Regulation lens until they have completed the structured training package and passed the scenario-based check-for-understanding at the end. This is not because the practices are esoteric. It is because the relational and regulatory work in classrooms is the part of teaching most likely to be misread by an evaluative gaze, and an observer who tags co-regulation when they see something else, or reads the absence of relational repair as a teaching weakness rather than as the absence of a rupture to repair, generates noise rather than evidence. Noise dressed up as insight is worse than no data at all. The training is the protection. The protection is the gate.

There is a wider point here about what observation evidence is for. I have argued, in this blog and elsewhere, that observation should not be evaluative — that the job of an observer is to capture what is happening, descriptively and accurately, and to leave the evaluative work to the school's professional judgement, exercised across many observations and triangulated against other forms of evidence. That argument matters more, not less, in the relational and regulatory layer. A teacher's relational practice, captured across a term of visits and read alongside what their pupils say, what their colleagues observe, and what the teacher themselves reflects on, can tell a school something useful about the climate of its classrooms. The same data, captured by an untrained observer in a single visit and read evaluatively, is a weapon. The architecture of the lens — the gating, the qualifier set, the aggregate-over-visit framing, the teacher-held data — is designed to ensure the first reading rather than the second.

There is also a HGIOS4 dimension that matters for Scottish schools. Until the v3 update, Learning Lens carried the honest acknowledgement that its observation evidence on Quality Indicator 3.1 — Ensuring wellbeing, equality and inclusion — was approach-level only. Schools running structured observation cycles could claim that the way they were running observation spoke to QI 3.1, but the practice tags themselves did not directly evidence wellbeing or inclusion practice in classrooms. With the Relationships & Regulation lens, that changes. The five new practices give direct, contributory observation evidence on QI 3.1.1 (Wellbeing) and QI 3.1.3 (Inclusion and equality) at the level of classroom practice. A strong claim on these QIs still requires triangulation with pupil voice and pastoral data — observation evidence does not stand alone here, and it shouldn't — but for the first time, the observation evidence is not absent. That matters as the new HMIE inspection framework develops and schools think about what evidence they will have available to map their practice into whatever shape the new framework takes.

I want to say something honest about why this addition feels different from the others. Adding a practice to a cognitive category — predictable lesson structure, say, which the v3 update also brings in — is essentially a calibration. It refines what the taxonomy captures. Adding a Relationships & Regulation category is a different kind of move. It says something about what teaching is, what self-evaluation should look at, and where the boundary of the cognitive-instructional research base needs to give way to a different kind of evidence. I am wary of the move, in the sense that any product decision that significantly expands the surface area of what is captured deserves wariness. I am also more confident in it than I have been in any product decision for a long time, because the gap it fills is the gap I have spent most of my professional life trying to close.

A taxonomy that names what teachers do cognitively but does not name what teachers do relationally is a taxonomy that gives you a partial view of the work. The work is bigger than that. The pupils schools teach, support, and (sometimes) fail to reach require both layers, often in the same lesson, often in the same minute. Pretending that a tool built for self-evaluation can describe the cognitive layer cleanly while leaving the relational layer to the pastoral team is the kind of pretence that schools — and their pupils — cannot really afford.

The taxonomy now has fifty-nine practices across eleven categories. The number will keep changing. What matters is that it captures, with as much honesty as the evidence allows, the actual work that teaching is.

Jamie Scobie writes from extensive experience in Scottish secondary education, including pastoral care, data for improvement, and school self-evaluation. This blog is an independent publication: he writes in a personal capacity as the creator of Learning Lens, writing about classroom observation, teaching evidence, and education policy. He speaks here only for himself and for Learning Lens, not for any employer or other organisation. He holds an MSt from Cambridge (Distinction) and a Masters from Stirling.

References

Bomber, L. (2007). Inside I'm Hurting: Practical Strategies for Supporting Children with Attachment Difficulties in Schools. Worth Publishing.

Education Endowment Foundation (2019). Improving Behaviour in Schools: Guidance Report. London: EEF.

Geddes, H. (2006). Attachment in the Classroom. Worth Publishing.

McDonnell, A. A. (2010). Managing Aggressive Behaviour in Care Settings: Understanding and Applying Low Arousal Approaches. Wiley-Blackwell.

Perry, B. D. (2009). Examining child maltreatment through a neurodevelopmental lens. Journal of Loss and Trauma, 14(4), 240–255.

Shanker, S. (2016). Self-Reg. Penguin.

Treisman, K. (2017). A Therapeutic Treasure Box for Working with Children and Adolescents with Developmental Trauma. Jessica Kingsley Publishers.

Tronick, E. Z. (1989). Emotions and emotional communication in infants. American Psychologist, 44(2), 112–119.