What Autistic Mirror does for GP practices

A patient describes abdominal pain calmly, briefly and without visible emotion. In a short appointment, it is tempting to conclude that it is not that bad. Autistic Mirror explains why this conclusion can be wrong for autistic people.

This article does not replace the guide for GPs or medical training. Here, three unedited answers show what the app can do for GP practices and where its boundary lies.

The following answers come from Autistic Mirror, unabridged and unedited. They were generated on 3 October 2026 in German, in outside view/environment mode with the role physician. The English text is an idiomatic translation of those German answers.

1. Flat affect when describing pain

User question

A patient describes pain flatly and without expression. What can that mean?

Autistic Mirror (outside view/environment, role physician)

Flat affect while describing pain is a common diagnostic misunderstanding, because it is read as a sign of malingering or a psychogenic cause. Both can be wrong.

Interoception is the process by which the brain perceives and processes internal bodily states such as pain, hunger or tension. In many autistic people, this signal arrives with lower resolution or in bursts, not as a continuous stream. The pain is physiologically real and measurable, but it is not reliably perceived, categorised and mirrored outwardly as such. The missing affective colouring in expression then does not mean that less is there, but that the channel between bodily experience and its representation is calibrated differently.

Added to this is alexithymia: difficulty assigning bodily signals to a feeling word or an intensity category. Alexithymia occurs in an estimated 40 to 65 percent of autistic people. When someone is asked how strong the pain is or how it feels, this is not a task that can simply be retrieved. The experience is present, but the verbal assignment is missing or costs considerably more cognitive work than the question suggests. The result often looks flat from the outside, because the person is actually searching in real time for what to say.

Together, these two mechanisms also explain why autistic patients sometimes only name pain once it has persisted for a very long time, and why pain intensity is frequently underestimated in conversation. Clinically relevant note: the reported pain level may systematically lie below the actual one. Physical findings and the extent of functional loss are more reliable diagnostic anchors than verbal and facial expression.

2. Why bodily signals are hard to categorise

User question

Why do some autistic people find it hard to categorise bodily signals?

Autistic Mirror (outside view/environment, role physician)

Autistic people often perceive bodily signals such as hunger, heartbeat, exhaustion or tension with lower resolution. The reason lies in the processing of interoceptive signals, the signals that report the internal state of the body to the brain. Interoception refers to exactly this process: the insula, an area deep in the cortex, continuously integrates feedback from inside the body and creates awareness of one's own state from it. In many autistic people, this integration runs with lower precision or at irregular intervals instead of as an even data stream. The signal therefore does not arrive continuously with fine resolution, but in bursts or with delay.

This has a direct consequence for everyday life: hunger, thirst or exhaustion are only noticed when the state is already pronounced. The early, quiet warning that guides action in time with neurotypical interoception is missing.

On this basis, alexithymia, the second level involved, develops. Alexithymia means that the step from a bodily perceived state to naming a feeling is difficult. The insula delivers a signal, but assigning it to a category, "this is fear" or "this is sadness", does not succeed reliably. This assignment is not an automatic process, but runs through learned connections between bodily sensations and emotional concepts. If the interoceptive signal itself is already blurred, this connection also remains unreliable. A large part of what is read from outside as "emotional flatness" is actually this assignment problem, not a lack of feeling.

Important for the practice context: the question "How do you feel?" often falls flat with alexithymia, not because no answer is possible, but because the feeling word finds no anchor in the bodily signal. Body-based approaches, meaning concrete descriptions of physical states, give this step a foundation on which categorisation is then more likely to succeed.

3. The question about history taking

User question

How should I take the history?

Autistic Mirror (outside view/environment, role physician)

The question can be understood mechanistically: what changes in taking a history when the person is autistic?

Interoception is the first anchor. Bodily signals that usually serve as a bridge to emotions in history taking ("How do you feel about that?", "What does that do to you?") are less reliably accessible for many autistic people. This is because interoceptive signals, the perception of internal bodily states such as heart rate, tension or exhaustion, are transmitted atypically and translated less precisely into conscious representations. The bodily state is present, but assignment to emotional categories such as "sad" or "overwhelmed" does not work automatically. This means that affect questions do not yield valid data when asked in an open, introspective way.

More concrete approaches work better. Instead of "How are you right now?", a situational question: "What happened most recently that brought you here?" The person can describe events even if they cannot name the emotional valence.

Alexithymia amplifies this as a precondition. In a considerable proportion of autistic people, estimates vary, but 40 to 65 percent is often cited, emotional states are present but not coded in language. Alexithymia does not mean an absence of emotion, but a lack of verbal availability. This changes what your questions need to do: less "What do you feel?", more "What did you do when that happened?" or "What has changed in the last few weeks?"

A further point concerns predictive coding. Autistic cognition weights incoming sensory and social information more strongly relative to internal predictions. Concretely, this means the history-taking conversation itself is a social stimulus with a high processing load. Eye contact, implicit conversational rules, ambiguous phrasing ("What is on your mind?"), the unfamiliar room, unfamiliar person, unfamiliar procedure. Each of these variables draws capacity that is then no longer available for answers. Predictability at the start, a short description of the procedure and a clear time frame reduce this load.

Directness is not a risk but a relief. Autistic communication is often precise and information-oriented. Indirect questions such as "Perhaps you could tell me what brought you here?" create ambiguities that have to be processed. Direct questions are not impolite, they reduce cognitive effort.

On the double empathy problem: when an autistic person seems monosyllabic, flat in affect or apparently uninterested, this is neither rejection nor a processing deficit. It is a different communication style on both sides. The clinical picture may suggest affective flattening or dissociation although the mechanism is different: speaking in socially unfamiliar situations costs more, the output is reduced, but the inner state is not.

Masking is a concrete diagnostic problem in history taking. Many autistic adults, especially women and people identified late, have decades of learned compensation. They seem socially competent, hold eye contact and answer fluently. This can distort clinical assessment. A direct question about exhaustion after social situations, or about the difference between public and private behaviour, yields more here than observing how the conversation goes.

If written communication is offered beforehand or in parallel, for example a questionnaire to take home, this reduces the simultaneous double load of sensory processing and speech production.

What the app does

The answers separate visible expression from experience. Interoception and alexithymia explain why pain can be real even when it is described flatly, late or as mild. The first answer names physical findings and loss of function as more reliable anchors than facial expression.

For the practice, this adds another interpretation when description and findings do not match. The app explains the mechanism. It does not diagnose and does not assess an individual case.

What the app does not do

Autistic Mirror does not replace a medical examination, clinical guidelines or training. The third answer goes beyond pure explanation when asked about history taking: it names example questions and conversation conditions and switches once to the informal you. It remains unedited for transparency and should be read as a borderline case, not as instructions for history taking.

Context

Interoception, alexithymia, predictive coding and the double empathy problem are separate fields of research. Applying them together to pain history in general practice is a mechanistic interpretation, not a statement about every autistic person.

Statements about the insula and signals arriving "in bursts" are models. The range of 40 to 65 percent for alexithymia given in two answers has no source in the answer. The meta-analysis by Kinnaird, Stewart and Tchanturia (2019) found a prevalence around half, with wide variation depending on the measure.

Sudden pain or a changed condition always needs clinical assessment first, even with a known autism diagnosis. The mechanisms here explain a communication gap. They are not an explanation for symptoms.

With Workplace licences, organisations only see licence and usage volume. Conversation content stays confidential.

The architecture of Autistic Mirror has been published in Autism in Adulthood: Wahl, A. (2026). Autistic Mirror: A Conversational Artificial Intelligence Application for Neurological Self-Understanding in Autistic Adults and Their Support Networks. DOI: 10.1177/25739581261490537.

You can ask your own questions about your situation.

Autistic Mirror is not a medical device and does not replace diagnosis, therapy or medical advice.

Sources
  1. Garfinkel et al. (2016). DOI: 10.1016/j.biopsycho.2015.12.003
  2. Kinnaird, Stewart & Tchanturia (2019). DOI: 10.1016/j.eurpsy.2018.09.004
  3. Bird & Cook (2013). DOI: 10.1038/tp.2013.61
  4. Pellicano & Burr (2012). DOI: 10.1016/j.tics.2012.08.009
  5. Milton (2012). DOI: 10.1080/09687599.2012.710008
  6. Hull et al. (2017). DOI: 10.1007/s10803-017-3166-5
Book a first call (20 min)

Or try it yourself for free