Who am I when the mask falls away?

Autistic Mirror (inside view)

That is grief and disorientation at the same time, and both can be explained by the same process.

Years of masking mean the brain has built a parallel prediction model. Predictive coding describes how the brain constantly stores models of what comes next and what is expected. With long, intensive masking, the brain does not only learn which behaviour fits externally, it starts storing that model as a self-description. The mask is not encoded as a costume, but as a prediction about who you are.

When masking then falls away, predictive coding reports: pattern missing, prediction failed. That feels like emptiness, but it is not proof that nothing is there. It is a prediction error without immediate resolution, because the new model has not been built yet.

The autistic self was present the whole time, it just was not encoded as self, but as disturbance, as the thing that had to be masked. The sensory reactions, the way of thinking, the things that create energy, the responses to injustice or inconsistency, the special interest, all of that runs without a mask. That is not emptiness. That is what remains when the translation work stops.

Autistic inertia also plays a role here. Inertia means that a state, once started, is neurologically hard to switch. Decades of masking is a deeply grooved state. Switching to an unmasked self takes time, not because of willpower or a lack of it, but because the brain has to build new patterns while it is decoding old ones at the same time.

The question "who am I without the mask" cannot be answered by thinking, only by observing. What remains when nothing is performed? What creates energy instead of costing it? What was always there but got explained away?

Autistic Mirror (inside view)

The fear behind this is real: what if dismantling the mask means less of you remains?

But the distinction matters mechanistically.

Masking is sustained prefrontal control over automatic processes: suppressing stimming, adjusting facial expression, regulating speech rate, filtering answers before output. That costs energy because it is a dual task. Processing content and performing your own output at the same time.

The abilities that became visible in the process come from a deeper level, though. Precision with detail, pattern recognition, linguistic accuracy, systematic thinking, intense focus on a problem. Those are not masking products. That is autistic cognition that was visible through the mask.

What falls away when masking decreases: the performance layer. The smooth facade, the social timing, the automatic adjustment of output to neurotypical expectations.

What remains: the mechanism that produced the actual performance.

There is, however, a transition phase that can feel like loss. If masking ran for a long time, the whole system was calibrated for a dual task. When the control layer falls away, the neural system needs time to reorganise. Abilities can temporarily seem less accessible, not because they are gone, but because energy is being redistributed right now.

The praise applied to something real. The mask only packaged it in a particular form.

Autistic Mirror (outside view)

When your partner masks less after late identification, his nervous system goes through something that could be described as restraint collapse, meaning a decline in a continuously maintained suppression effort. But there is more to it than that.

Masking is not a decision made in the morning. It is a learned system of compensation routines that keeps prethalamic and prefrontal tissue permanently under load. Your partner has probably spent decades automatically and unconsciously dampening sensory reactions, delaying visible body signals, holding eye contact, adjusting tone, suppressing stimming and generating answers from social scripts. All of that costs glucose and cognitive capacity that is then missing elsewhere.

When that suppression effort decreases, what appears is not "a different person", but the same person without the translation layer. What you experience as unfamiliar is most likely his actual communication and regulation pattern: more direct reactions, visible stimming, less adjusted facial expression, different speech rhythms, clearer refusals of stimuli. That is not deterioration. That is the end of a metabolically expensive effort.

There is a second mechanism that is almost always involved here: predictive coding. Over years, your brain learned a model of your partner, meaning expectations about how he will speak, react and behave. That model was based on the masked version. When he now reacts differently than expected, your brain reports a discrepancy before you have consciously thought about it. The feeling of unfamiliarity is not a sign of an actual change in the relationship. It is the adjustment period of your prediction model to a person you have in fact known all along.

And a third aspect: many late identified autistic people go through a kind of retroactive processing in this phase. Behaviours they suppressed with effort for years are tried out or return. That can seem intense. Special interests take up space. Sensory limits are communicated more clearly. Moods are less smoothed over. That is not a relapse, it is often the first time his nervous system is allowed to regulate instead of control.

The unfamiliarity you notice is, paradoxically, a sign of safety. He is showing you a version of himself that previously cost too much to show.

Autistic Mirror (inside view)

An autistic mentor has lived experience. She knows the feeling from the inside, not just the mechanism behind it. She can say "I know that, for me it was like this" and create a recognition effect that does not replace an explanation, but does something else: it anchors.

The app explains the neurological mechanism without subjectivity. No "that is how it was for me", no filter of experience, no story of its own that could overlay yours. That is an advantage and a disadvantage at the same time.

Advantage: the explanation does not apply to one person, it describes the mechanism as such. No double empathy problem between you. The double empathy problem describes how communication between differently neurodivergent people, or between neurotypical and autistic people, is misread on both sides, because both use different prediction models for social signals. Two autistic people can also have very different profiles. The app has no profile that gets in the way.

Disadvantage: experiential knowledge cannot be represented. A mentor can notice that a question is really a different question, because she has taken the same detour. The app only recognises what is explicitly stated.

What the app specifically does differently from a mentor on mediation questions:

• It gives no assessment of the situation, only the mechanism behind it.
• It recommends no strategy from its own experience ("this is how I solved it").
• It is consistent, regardless of the time of day, exhaustion or its own regulation needs.
• It has no memory of the last conversation, no relationship dynamic, no implicit expectation.

What a mentor can do and the app cannot: judge whether the question fits the situation, ask follow-up questions without a protocol, answer from her own lived reality, and notice in conversation when someone is in shutdown and needs silence rather than knowledge about mechanisms.

Autistic Mirror (inside view)

When two autistic people interact, the translation effort that otherwise costs most of the energy falls away.

Masking works like this: in neurotypically dominated social contexts, the autistic brain permanently runs a dual task. It processes the content of the conversation and simultaneously controls its own performance, meaning facial expression, timing, volume, eye contact, phrasing. That is prefrontal control over processes that would otherwise run automatically, and it measurably costs glucose and cortisol, even though it is invisible from the outside.

At the meeting with the other woman, your brain apparently recognised that the prediction model "situation X requires response Y" was not firing. Predictive coding means your brain constantly compares expectations with incoming signals. In neurotypical social contexts, the expectation "I will be misread or judged" automatically triggers control mode before the interaction has even begun. If that signal is missing, control mode stays off.

On top of that there is the double empathy problem: communication difficulties between autistic and neurotypical people are bidirectional, not a one-sided deficit. Autistic-to-autistic communication works differently because both use the same prediction model for social signals. Directness is not read as rude, pauses are not read as disinterest, depth of detail is not treated as exaggeration. The translation work falls away on both sides.

What you experienced as "not having to perform" is the state when the prefrontal control layer falls away and only the actual content of the conversation is processed. That is neurologically cheaper and explains why these encounters are often less exhausting than shorter interactions in other contexts.