What Autistic Mirror does for learning and development

An autistic employee spends three hours in a group workshop. He says little, joins the role play round hesitantly and takes away hardly anything. To learning and development, this quickly looks like a lack of interest. Autistic Mirror starts somewhere else. The app asks what the format costs neurologically.

This article is not a guide to careers and visibility. The article Best work, no promotion covers that side. Here, three unedited answers show what the app can do for learning and development and where its limit lies.

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

1. Why group workshops do little

User question

Why do group workshops do little for some autistic employees?

Autistic Mirror (outside view/environment, role employer)

Group workshops bundle several demands at once: social processing, sensory input, unpredictable conversation and the switch between listening, thinking along and speaking. For autistic employees this is not a format problem but a capacity problem.

Monotropism is the central mechanism here. Autistic attention gathers like a tunnel on one channel instead of spreading widely. Workshop formats demand exactly the opposite: following the facilitator, placing other people's contributions, formulating your own thoughts and judging when an interjection fits, all at the same time. These parallel channels compete for the same limited attentional resource. The content that is actually meant to be processed often arrives last.

Predictive coding adds to this. The autistic brain weights the actual sensory input more strongly than its own prediction. Group workshops are full of the unpredictable: others talk over each other, the conversation jumps, facilitation decisions come without warning. Each of these deviations from the expected course creates a prediction error that the nervous system has to process with effort before it can turn back to the content. That costs capacity which is then missing for the actual learning goal.

Finally, masking works as a silent constant load. Autistic employees who do not want to stand out in group formats suppress a range of automatic reactions on the side: stimming, different eye contact, pauses that last longer than expected. This is a dual task that runs prefrontally and draws on capacity that should belong to the workshop content.

In concrete terms: an autistic employee can leave the room exhausted after a three-hour group workshop and take away only fragments of the content, not because the topic was difficult, but because the format was neurologically expensive. From the outside this sometimes looks like disinterest, but it is not.

Formats that make the content more accessible without changing the person are often simple: written advance materials with questions and time, allowing asynchronous input, smaller groups or individual work phases within the workshop. Which option fits is something the autistic person knows best, and a direct conversation about it is always the more precise route than a blanket solution.

2. Role plays and spontaneous exercises

User question

What happens during role plays and spontaneous exercises?

Autistic Mirror (outside view/environment, role employer)

When a role play or a spontaneous group exercise is suddenly announced in a meeting or training, many autistic employees face several demands at once that reinforce each other.

Predictive coding is the first mechanism involved. This means: the brain constantly runs a predictive model of what happens next. Role plays and spontaneous exercises break this model, because the further course suddenly becomes unstructured and socially open. The more precisely someone otherwise predicts (and autistic processing is often especially detail-oriented and reliant on predictive stability), the more expensive this break is. This is not an oversensitivity problem but computational effort: the old model has to be discarded and a new one built while the exercise is already running.

At the same time masking comes into play. In social situations, autistic employees often automatically control their reactions prefrontally, that is, with the part of the brain that is also responsible for planning and attention. This runs as a dual task alongside the actual content. A role play now adds several new demands: taking on a role, improvising socially, being observed at the same time. The capacity that would be needed for the exercise itself is already tied up by the masking load.

The third factor is phatic communication. Role plays often demand exactly this: social improvisation without an information goal, ease with vague expectations, responding to mood rather than content. For an attentional profile that processes language mainly referentially (that is, towards content and clear meanings), this mode is not automatically available. It has to be consciously analysed and served, which again costs capacity.

What the surroundings then see is often hesitation, withdrawal, silence or a result that seems "too serious". This is not resistance to the exercise and not a lack of engagement, but the visible result of this capacity situation.

For the design this means: announcing it in advance helps, because it keeps the predictive model stable. A short written description of the task before starting helps, because it allows entry without pressure to improvise. An optional observer role as an equal form of participation helps, because it allows involvement without forced social improvisation.

3. The question about training design

User question

How should I design trainings?

Autistic Mirror (outside view/environment, role employer)

For many autistic employees, trainings are a particularly dense combination of sensory load, social demand and information processing at the same time.

Predictive coding explains a central mechanism here: the brain continuously constructs predictions about what comes next and compares them with the actual input. Autistic processing weights sensory input signals more strongly and predictions more weakly than neurotypical processing. This has a concrete consequence: surprises, unclear structures and format changes within a training create more processing effort, because every deviation from the expected has to be actively compensated. A training that feels "lively" and "varied" to neurotypical participants can cost autistic participants a considerable part of their capacity for handling the format changes instead of the content.

Monotropism adds to this: autistic attention gathers intensely on one thing instead of spreading out. This means that depth works well, while breadth with frequent changes of topic and method is expensive. Long single inputs with a clear focus are often processed better than a short rotation of different activities.

On the sensory side it is worth keeping an eye on the training environment: room brightness, acoustics (echoing vs. dampened), seating arrangement (free choice is better than assigned seats at group tables), the rhythm of breaks. The sensory budget is limited. It describes the total capacity for processing stimuli, social demands and information together. If a large part of it is spent on environmental stimuli, less remains for the content.

Concretely helpful are:

  • An agenda in writing in advance, with time blocks and a clear indication of when group work, input and breaks come
  • Clear facilitation rules for contributions, so that no implicit expectation arises about when someone should speak
  • Written documentation of the core content, not only spoken delivery
  • Free choice in interactive formats, meaning no compulsion to spontaneous group work or role plays without lead time
  • Follow-up material, because information processing often continues after the training and only then becomes complete

In all of this: the autistic person knows best which formats work. A short conversation beforehand about what they need is more precise than any general design rule.

What the app does

The answers separate content from format. Someone who takes little away from a workshop has not necessarily failed at the topic, but at the number of channels running at the same time. Monotropism, predictive coding, masking and phatic communication explain why group rounds, role plays and frequent changes of method tie up capacity that is then missing for learning.

For learning and development this adds a further view: what does a format cost before any learning happens? The app explains this individually and takes the question, the conversation and the role profile into account.

What the app does not do

Autistic Mirror does not diagnose, does not assess performance and does not design a finished training concept. The app does not replace integration services, occupational health or the expertise of learning and development.

On the third question, how to design trainings, the app names environmental adjustments such as a written agenda, clear facilitation rules or free choice in interactive formats. These lie within the product limit because they concern the environment and are not meant to change the person. The app also points out that the autistic person knows best which formats work.

Context

The cited works treat monotropism, predictive coding, masking and phatic communication each as separate research fields. How they work together in workshops and trainings is a mechanistic interpretation, not a statement about every autistic person. Some autistic people learn well in groups.

Statements about prefrontal capacity, the weighting of predictions and the sensory budget are models. They make processing tangible but do not measure individual ability to learn. Applying phatic communication to role plays is the app's interpretation, not a study of its own.

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 an occupational health assessment.

Sources
  1. Murray, Lesser & Lawson (2005). DOI: 10.1177/1362361305051398
  2. Pellicano & Burr (2012). DOI: 10.1016/j.tics.2012.08.009
  3. Hull, Mandy, Lai, Baron-Cohen, Allison, Smith & Petrides (2019). DOI: 10.1007/s10803-018-3792-6
  4. Coupland, Coupland & Robinson (1992). DOI: 10.1017/s0047404500015268
Book a first call (20 min)

Or try it yourself for free