Every decision you make, every plan you design, every intervention you implement, is based on a model of how the world works. Not the world itself. A model. The model lives in your head. It is built from experience, education, culture, and the particular lens your professional training has given you. And it is, inevitably, wrong in ways you cannot see, because the nature of a mental model is that it filters out what it has not been designed to notice.
Gregory Bateson, whose work sits behind much of systems thinking, used a phrase that has stayed with me: the map is not the territory. The map is useful. The map is necessary. But the moment you mistake the map for the territory, the moment you treat your model as though it is the situation itself, you stop seeing what the model was not built to show you.
What Mental Models Do and Don't Do
A 2025 paper in Systems Research and Behavioral Science defines systems thinking as a combination of perception, prior knowledge, and reasoning processes for guiding decision-making in complex, dynamic situations. Crucially, it notes that people with different levels of systems thinking knowledge produce different models of the same situation, and therefore make different decisions, even when the reasoning process is identical. The model determines what is visible. And what is invisible in the model is usually where the problem lives.
In my experience across training and education, the most common form of this is the expert's blind spot. The experienced practitioner has a model built from extensive exposure to similar situations. That model is often correct. It is also, in novel situations, the thing most likely to produce the wrong diagnosis, because the expert pattern-matches to familiar categories rather than looking fresh at what is actually present.
“The problem with a good mental model is that it makes you faster at reaching wrong conclusions.”
The Assumptions Inside Your Assumptions
When I use the Pattern Before the Incident framework from Designing Behaviour in the Classroom, one of its functions is precisely to disrupt the mental model. The six diagnostic questions are designed to produce information that the practitioner's existing frame might not prompt them to seek, questions about time of day, subject, social configuration, what preceded the incident by hours rather than minutes. These redirect attention from the behaviour itself, which the model has already categorised, to the conditions that precede it. Which is where the useful information almost always lives.
The Integrative Failure
A 2025 paper published in PMC on integrating systems thinking and behavioural science makes a point that I find consistently underappreciated: traditional approaches to behaviour change have focused on education and awareness, assuming that increased knowledge leads to better decisions. The evidence does not support this assumption. Much of behaviour is driven by automatic, intuitive processes that educational interventions typically overlook. This is a mental model failure at the level of policy design. The model says: people behave sub-optimally because they lack information. The intervention gives them information. The behaviour does not change. The model is rarely revised.
Expanding the Frame
Expanding a mental model requires exposure to perspectives that sit outside it. This is one of the reasons I believe strongly in cross-disciplinary reading, not as a productivity strategy but as a corrective to the inevitable narrowing that comes from expertise. Polyvagal theory changed how I understood behaviour in classrooms. Somatic experiencing changed how I understood burnout. Systems dynamics changed how I understood why interventions fail. None of these were in the professional training for the roles I have occupied. All of them changed what I could see. The question worth asking regularly is not 'am I right?' The question is 'what would I be missing if my current model is incomplete?'
Bart Reevers