For the people around
The blind spot in diversity programmes
The programme has been running for years. There are trainings, a charter, metrics and a named contact. And still, neurodivergent employees say they are not included in it. That can sound ungrateful, but it is precisely observed: diversity work and accessibility for neurodivergent employees act at two different points, and the second one is missing from most programmes.
This article frames three answers from Autistic Mirror in outside-view mode for the surrounding people, with the employer role. It explains mechanisms and does not replace legal, organisational or therapeutic advice.
The answers below were generated live in the app on 29 September 2026 in German and taken over unabridged and unedited. This English version is an idiomatic translation of those same answers.
1. Why the programme does not land
The first answer separates two levels of intervention that programmes often treat as one.
User question
Our diversity programme has been running for three years with training on gender, background and physical disability. Neurodivergent employees still say they do not feel included in it. Why does a programme like that not land with them?
Autistic Mirror (outside view for the surrounding people, employer role)
Diversity training is conceptually built around visible characteristics and social categories. Neurodivergence works differently, and that structural incompatibility explains why the programme does not land, even though the intention is there.
Predictive coding is the central mechanism here. Autistic processing works with a different balance between prediction and incoming signals. Neurotypical cognition filters environmental input heavily through expectations, so the brain mainly processes deviations. Autistic processing weights the incoming signals more strongly and the predictions more weakly. That sounds like a technical footnote, but it is central to the training question: a session built on "taking a different viewpoint" and "shifting perspective" addresses a cognitive architecture that is set up differently in autistic employees. The content does not structurally fit the way of processing.
Concretely: training on gender and background addresses social categories that are negotiated through language, representation and implicit attitudes. The remedies on offer are typically empathy exercises, perspective taking and emotional reflection. That assumes neurotypical social cognition as the standard. Neurodivergence, however, is not a social marker in that sense, but a different processing architecture. What creates accessibility is not a changed attitude among colleagues, but concrete structural adjustments: communication formats, sensory environment, clarity of expectations. That level almost never appears in standard DEI training.
Monotropism intensifies the problem. Autistic attention bundles intensely onto one channel instead of spreading broadly. A training session that switches quickly between topics, combines small group exercises with unstructured social interaction and implicitly expects everyone to participate in the same way is cognitively expensive for autistic employees, regardless of the content. The format of the training is itself a barrier.
Also involved: the double empathy problem. When neurodivergent employees say they do not feel included, that is precisely observed. Training on neurodivergence developed by neurotypical trainers, without involving autistic employees in the design, reflects the neurotypical outside perspective, not the actual experience. The double empathy problem describes how misunderstandings between autistic and non-autistic people arise bidirectionally, so not through a deficit on one side. Training that frames neurodivergence as a "particularity to be taken into account" rather than as a valid way of processing in its own right reproduces exactly that imbalance.
The core of the gap: DEI programmes address discrimination through changes in attitude. Accessibility for neurodivergent employees arises through changes in the environment. Those are two different levels of intervention. As long as your programme only works on the first level, the second stays structurally invisible.
2. Why communication training changes little
The second answer explains why a measure targets one person although the mechanism sits between two sides.
User question
Our inclusion measures almost always come down to autistic employees getting communication training. Why does that change so little about how the team works together?
Autistic Mirror (outside view for the surrounding people, employer role)
Because the measure is applied in the wrong place.
Communication training for autistic employees assumes that the problem lies in their communication. The actual mechanism is different: when autistic and non-autistic people misunderstand each other, that does not arise from a one-sided deficit, but from two different processing styles meeting. That is what the double empathy problem describes. Both sides leave the conversation thinking the other expressed themselves unclearly or reacted inappropriately. The gap in understanding is mutual.
The empirical counter-check makes this clear: among autistic people, communication flows smoothly, without what neurotypical observers would describe as a communication problem. That means the problem does not lie in the person, but in the interface between two different communication styles.
Another mechanism sharpens this: masking. Automatic autistic reactions are suppressed prefrontally in social situations while the actual work content is being processed at the same time. That runs as a dual task and permanently consumes cognitive capacity. Training aimed at imitating neurotypical communication patterns increases that masking load. It does not train better collaboration, it trains more concealment. The short-term effect can be that meetings run more quietly. The medium-term effect is usually that exhaustion increases and actual work performance drops.
On top of that comes a structural level: phatic communication has a relationship-building function in working life, meaning the incidental small talk before a meeting, the implicit signalling in corridor conversations. A monotropic, information-oriented attention profile processes language primarily by content. That channel has to be consciously analysed and served instead of running automatically. That is not a learning deficit. Communication training that drills exactly this increases the effort even further, without changing the environmental conditions that create the effort.
What actually changes collaboration are adjustments at team level: more direct communication standards for everyone, written summaries after meetings, explicit instead of implicit expectations. That lowers the translation load not only for autistic employees, but usually for the entire team. The direction of change is the environment, not the person.
3. Why the wording in your documents counts
The third answer describes what a phrase signals before anyone reads the offer behind it.
User question
Our programme documents consistently say people with autism and impairment. Does the wording have an influence on whether neurodivergent employees feel addressed?
Autistic Mirror (outside view for the surrounding people, employer role)
Yes, the wording has a direct influence on whether autistic employees recognise themselves in your documents or feel excluded.
The majority of autistic adults prefer identity-first language, so "autistic person" rather than "person with autism". That is not a stylistic detail, it is a substantive position: "person with autism" implies that autism is something additive, separable from the actual person, similar to a device you can put down. "Autistic person", by contrast, describes autistic processing not as a property beside the person, but as a fundamental way in which the brain processes, perceives and prioritises information. It is not detachable. Many autistic people therefore experience the phrase "with autism" as a description of a malfunction they carry around, not as a description of their actual neurology.
The second problem lies in the term "impairment". It classifies autism as a deficit before anything has been said about concrete support needs. In the context of programme documents, meaning documents that govern access, support or working conditions, that framing signals: you are here as a recipient of help, not as an employee with specific working requirements. Autistic employees who do not see themselves as impaired, but as processing differently, simply do not find themselves in such documents. They overlook the offers, or they read them and decide against them.
A further aspect is ableist language as a precondition for disclosure. When documents use language that frames autism as a problem, that lowers the likelihood of employees naming their support needs at all. The frame implicitly says: whoever comes forward is admitting to being deficient. That is a structural signal sitting directly in front of the actual offer and blocking it.
Concretely for your documents: "autistic employees" or "neurodivergent employees" instead of "people with autism". Instead of "impairment", name specifically what is meant, for example "different working requirements", "sensory or communicative particularities" or simply "support needs". That is more precise and does not come across as pathologising.
Framing
The three answers show the same gap at three points. A diversity programme works on attitude, representation and language. Accessibility for neurodivergent employees arises at the environment: at communication formats, at sensory load, at the clarity of expectations. Working only on the first level does not make a programme bad, it makes it incomplete.
The answers name concrete adjustments in several places. That is environmental design, not an instruction for individual employees, and not a claim that a particular approach will work in your organisation.
The third answer switches to an informal address because the model picks up the form of address from the question; the German original shows this more clearly than the translation. The statement about the majority preference for identity-first language rests on surveys in the UK autism community; it describes a majority, not unanimity, and individual employees may hold it differently for themselves. Statements about prefrontal capacity and about the weighting of predictions are model accounts, not measured values for individual people. Whether an entitlement to accommodations exists in a specific case is a question for integration services, the relevant inclusion authority and employment law specialists.
Sources
- Milton (2012). DOI: 10.1080/09687599.2012.710008
- Kenny, Hattersley, Molins, Buckley, Povey & Pellicano (2016). DOI: 10.1177/1362361315588200
- Bottema-Beutel et al. (2021). DOI: 10.1089/aut.2020.0014
- Pellicano & Burr (2012). DOI: 10.1016/j.tics.2012.08.009
- Murray, Lesser & Lawson (2005). DOI: 10.1177/1362361305051398
- Hull, Petrides, Allison, Smith, Baron-Cohen, Lai & Mandy (2017). DOI: 10.1007/s10803-017-3166-5
- Coupland, Coupland & Robinson (1992). DOI: 10.1017/s0047404500015268