Daily Life

Promises and Decisions - Why They Are Non-Negotiable for Autistic People

When you make a promise, it is not a statement of intent. It is a fact that gets built into your internal model of reality. The same applies to decisions. Once made, they often act like a fixed anchor. What is interpreted from the outside as stubbornness can be explained by two mechanisms that shape many autistic brains. Both are models from research. How promises and decisions are processed in autistic people has not been studied directly so far, and not every autistic person experiences it this way.

Why Promises Are Not Social Rituals

Monotropism describes a common autistic attention profile (Murray, Lesser & Lawson 2005). Many autistic brains do not distribute attention broadly. They channel it into few but intense streams. Whatever enters this attention tunnel is processed very thoroughly.

A promise given by an autistic person enters this tunnel. It is not registered as a social convention but as binding information that gets integrated into the internal model of reality. The promise becomes part of the data structure the brain uses to predict the world.

This is why a broken promise can generate a strong error signal. Not because autistic people hold grudges. But because information classified as reliable turns out to be false. The brain has to update its model around that person, and that costs energy.

The phrase "I didn't really mean it" is not an apology in this context. It is a retroactive devaluation of information that has already been processed. The autistic brain asks the logical question: Then why was it said?

Decisions as Neurological Anchors

The predictive coding model describes brains like this: they constantly generate predictions about the world and compare them against incoming data (for autistic perception: Pellicano & Burr 2012). Discrepancies between prediction and reality produce prediction errors: error signals that consume neural resources.

Applied to decisions, this means: for many autistic people, a decision does not act like a tentative assessment but like a fixed component of the internal model. The brain builds its predictions on top of this decision. Behavioural patterns, expectations, plans: everything gets organised around it.

"Think about it again" costs autistic people more than neurotypical environments assume. It does not mean changing an opinion. It means dismantling a complete internal model and building a new one. Every branch that depended on the original decision has to be recalculated. This is mainly a cognitive process.

What Can Overturn a Decision

Autistic decisions are stable but not immutable. The difference lies in the revision mechanism.

Neurotypical revision often works through social persuasion. Someone argues emotionally, applies persuasion, appeals to group conformity. Many autistic people barely respond to this, or experience it as manipulation.

What can actually overturn an autistic decision: new data. When information emerges that disproves the internal model, a reassessment begins. Not because someone was persuasive, but because the data landscape has changed. The difference between "I was convinced" and "I processed new information" is fundamental for many autistic people.

The second mechanism is a values conflict. When a decision violates a core value (and this discrepancy becomes conscious), a revision can begin. Autistic value systems are often rigid, not because they are unreflective, but because they function as internal axioms on which other decisions are built.

This applies to all decisions involving people with deep attachment: partnerships, close friendships, parent-child relationships, working relationships built on trust. The bond itself is a data source. As long as the data that led to the decision has not been disproved, the decision stands. Emotional pressure from the outside does not change the data.

When Feelings Are Built on Decisions

Feelings built on a data-based decision follow the stability of that decision. As long as the foundation has not been disproved, the feeling remains anchored, even when it temporarily feels different.

This becomes particularly visible with experiences for which no frame of reference exists. A first job in a new field. A first deep friendship after years of isolation.

Or a first relationship where communication works without the effort of translation. The mind cannot keep up because there is no template against which the experience can be measured. Everything is new, and the predictive coding system has no stored models to fall back on. This creates not just uncertainty. It creates stress, because the brain is trying to classify an experience for which it has no category.

External stress sources (work, social demands, sensory overload) add to this processing load. When the total load exceeds capacity, the state tips into overload. And under overload, the brain falls back on old models: models based on past injuries, not the current reality. The result can be a resolution that matches neither the real feeling nor the data behind the underlying decision. The resolution feels compelling in the moment because the overloaded system has no capacity to check it against the actual data.

What is needed in this moment is not an argument and not a solution. It is space. Space to process, to classify, to build a new frame of reference, without pressure, without a time limit.

When this space exists for the first time (an environment where processing is allowed without negative consequences, guilt or shame following), then this itself is an experience without a frame of reference. The brain has no model for safety during processing. It has learned that processing gets punished, that being overwhelmed is weakness, that needs are a burden. A space that does not do this initially generates its own error signal: This cannot be real.

But when the data remains consistent, when the space stays safe, when no consequence follows, when processing remains permitted, then the brain begins to build a new model. Slowly. Data point by data point. It learns: safety during processing exists. It is not temporary, not conditional, not tied to good behaviour. This new model may be the first one that is not based on defence but on experience. And it changes not only the classification of this one relationship. It changes the predictions the brain makes about relationships altogether.

The uncertainty concerns the classification, not the decision. And a resolution under overload is not a revision of the decision; it is an error signal that can be corrected when the system has capacity again.

Difficult phases change the feeling temporarily. But the decision does not waver, because there are no logical reasons to revise it. The autistic brain separates these two layers: the feeling fluctuates, the data does not. And the decision follows the data.

Old experiences create interference signals. Injuries from neurotypical contexts, from childhood, from previous relationships: they produce error signals that do not match the current reality. The autistic brain can recognise this discrepancy: This fear does not belong here. It comes from a different model, built on different data. That does not mean the fear is not real. It is real. But it belongs to a different data set. And the brain can learn to trace the source, if it is given the space and time to do so.

In predictive coding terms: old predictions based on past injuries collide with new data from the current relationship. The system needs time to overwrite the old models. But the conscious decision remains unaffected, because it was made on a separate data foundation.

Even with temporal distance, the decision remains stable when the data supports it. This is not stubbornness. It is integrity of the internal model.

A decision whose data foundation has not been disproved can last a lifetime. This is not clinging to the past. It is the logical consequence of a system that updates decisions based on data, not social trends.

Why Flexibility Means Something Different for Decisions

Neurotypical flexibility is often social compromise. Someone adjusts an opinion because the group expects it, because harmony matters more than consistency, because "not being so rigid" is considered a virtue.

For many autistic people, flexibility feels different, more like a rebuild: not an opinion is changed, but the system built on that opinion needs to be recalibrated. Cognitive flexibility is one of the executive functions where autistic people show more difficulty on average in studies (Demetriou et al. 2018). How demanding it feels and how well it goes in a test can differ (Yeung et al. 2025).

The cost of this process becomes visible through interoception, the perception of internal body signals. Some autistic people report physical symptoms during forced decision revisions: nausea, headaches, exhaustion. This can be understood as the physical side of a cognitive rebuild. The rebuild does not only affect the one decision. It affects every prediction, routine and behaviour that was built on top of it. That is why the energy cost can be so high, and why some need days of recovery after a forced revision.

The solution is not to make autistic people "more flexible." It lies in adapting the environment: fewer arbitrary changes, more predictability. And when a change is necessary, provide data, not pressure.

When the Resolution Does Not Match the Data

Sometimes an autistic person reaches a resolution under overload. The nervous system is at its limit, cognitive capacity is exhausted, and the brain falls back on the oldest available model, even if that model is based on outdated data.

The resolution feels right in the moment. But when the overload subsides and the brain begins to check the resolution against the actual data, a discrepancy emerges. This is not a sign of indecisiveness. It is the brain subjecting its own resolution to a quality review.

This process takes time. Sometimes days. Sometimes weeks. Revising a resolution reached under overload is not backing down. It is the correction of an error signal through better data.

What This Means in Practice

Do not make promises you cannot keep. Not as politeness, not as intention, not as a rough direction. If you are unsure, say so.

Do not frame decisions as stubbornness. The stability of an autistic decision is not a character flaw. It is an architectural feature.

When a revision is necessary: provide new data. Not emotional arguments, not persuasion, not "but everyone else does it this way." Explain what information has changed and why the old decision no longer works under the new conditions.

"I have new information that concerns your decision. Can I show it to you?" works. "Just think about it rationally" does not.

Autistic Mirror explains autistic neurology individually, applied to your situation. Whether for yourself, as a parent, or as a professional.