When an Autistic Decision Is Made, It Stands Whole

Autistic Mirror (inside view)

What you describe is not a personality trait and not stubbornness. It is predictive coding combined with a specific autistic tendency that can be called "autistic commitment".

Predictive coding means: your brain constantly builds a model of the world and updates it when new information comes in. In autistic brains, explicit, consciously made decisions are stored as facts in the world model, not as tentative probabilities. The underlying mechanism is a stronger weighting of bottom-up information. Bottom-up processing means: perception is built from concrete detail up to context, not the other way around. A decision that has been made is a concrete fact. It is stored as such, not as "the most likely option, revisable at any time".

The result: your internal model now contains "I live on Amrum" as a baseline state. The question "but when will you move back" produces a prediction error, an inconsistency between the stored model and what someone else apparently sees as open. That inconsistency does not feel like an inviting reflection question. It feels like an error in a system that is actually closed.

Non-autistic brains weight top-down predictions more strongly. Decisions stay stored longer as "high-probability forecasts", not as facts. That makes them easier to revise, but also less stable. Keeping the question "and what if you don't like it?" open is not doubt for a non-autistic brain, it is normal model maintenance. They continuously simulate alternative futures as a planning resource.

Your brain does not need to do that here, because the decision is no longer in the "to be processed" area. It is closed. Open loops are the opposite. The autistic brain keeps unresolved processes active and energy-intensive. Closed decisions close that loop. That is not rigidity, it is efficiency: a closed loop no longer needs to be managed.

That is why the question about moving back does not exist internally. Not because you suppressed it, but because the model no longer contains that branch. Amrum is the baseline. Hamburg is past, not option.

You can find sources for the mechanisms mentioned in the glossary.

Autistic Mirror (inside view)

The question goes deeper than noise level or population density. There are at least four structurally different neurological factors.

Environmental predictability

In Hamburg, predictive coding fires constantly at full load. The autistic brain weights sensory input more strongly than predictions, and a large city delivers a constant stream of prediction errors: unexpected movement at the edge of vision, sudden jumps in loudness, changing faces, unpredictable traffic flows. Every error costs processing energy, because the brain has to resolve it. On Amrum, the prediction model is radically simpler. The sea makes sounds constantly, but the same ones. The wind comes from one direction. Horizons are empty. The brain can settle on a stable model and stay there, without constantly correcting it.

Structure of sensory stimuli

City stimuli are high-frequency, discontinuous and socially coded: sirens, snippets of conversation, doorbells, advertising, glare, smells with unclear source. Sensory gating (the pre-attentive filtering meant to suppress background stimuli) delivers less automatic suppression in autistic people. That means: all stimuli stay in the foreground, including irrelevant ones. In a city, input is therefore permanently maximal.

Nature stimuli are structurally different. They are spectrally broader, more repetitive, more continuous. Wind, water, sand under the feet are sensorially rich but not discontinuous. The nervous system does not have to keep classifying them as "new". The gating mechanism can stabilise, because nothing keeps demanding reclassification.

Absence of social decoding work

In the city, social space is densely populated, and every encounter implicitly activates the double empathy problem: two different prediction models meet, and the autistic brain carries the higher translation load in neurotypical-dominated spaces. Subway, supermarket, café, sidewalk, everything is potentially socially coded space. On a small island, that drops. Not because the people are friendlier, but because the density of implicit social reading demands per hour falls dramatically.

Monotropism and openness

Monotropism describes how autistic attention concentrates: intensively on one thing, not spread across many at once. In a large city, dozens of stimuli compete simultaneously for the same attention tunnel, which produces chronic splitting. An open horizon, a dyke, an empty beach line are environments in which the monotropic focus can settle on something singular (the horizon, the water, the clouds) without being constantly interrupted. That is not a metaphor for calm, it is structural compatibility between attention architecture and environment.

That you are calmer with tourists during peak season than you were in Hamburg points to the fact that the decisive factor is not crowd size, but the character of the background noise. Tourists on a beach make noise in an open, acoustically diffuse space without reverberation. Hamburg makes noise in street canyons with reflections, surprise elements and social density in tight quarters.

You can find sources for the mechanisms mentioned in the glossary.

Autistic Mirror (outside view, close circle)

What is happening with her has several layers.

Autistic decisions often come about differently than neurotypical ones. Predictive coding describes how the brain builds prediction models and raises alarms when there is deviation. Autistic brains weight sensory and factual input more strongly than social expectations. That means: when an autistic person makes a far-reaching decision (apartment gone, car gone, island), it is usually the result of intensive, deep processing. Not an impulse. A closed internal process.

When that process is now called up again from the outside with "But really? But really?", the following happens: predictive coding has to generate and resolve the same prediction error over and over. That costs real cognitive energy. The question does not trigger curiosity, it triggers a computation.

On top of that comes autistic inertia. Once a state transition has been completed, it runs stably at the neurological level. Repeated questioning acts like an attempt to destabilise that state from outside, even when it is meant kindly. The brain responds with resistance, because the query is registered as an inconsistency with an already stored factual state: "decision made. query does not fit."

The growing irritation is not a hypersensitivity problem. It is the sign that the same processing load has been repeated too often.

What works differently, without being patronising: you do not have to comment on the decision. If you ask "how is it there?" instead of "are you really staying?", you signal that you treat the decision as a fact and not as a tentative state that still needs confirmation. The difference lies in the starting point of your question: is the decision already contained in your question, or are you implicitly putting it up for a vote?

You can also say directly to her what you observe, that you notice the repeated questions from others are costing her, and that you do not intend to repeat them. That is not being patronising. It is explicit communication that you respect her processing state.

You can find sources for the mechanisms mentioned in the glossary.