A safety that most autistic people arrive at work already without
Three weeks ago I covered the Neurodiversity in Business survey: three years of tracking, and the one metric its own research identified as decisive — psychological safety — hadn’t moved. Elizabeth Follmer and Jennifer Spoor, at the University of Washington Bothell and La Trobe’s Olga Tennison Autism Research Centre, have now published a paper in Organizational Dynamics that supplies the mechanism: why that metric is so hard to shift for autistic people specifically, and what actually moves it.
Psychological safety, in Amy Edmondson’s formulation, is the condition in which people feel able to take interpersonal risks, make mistakes, exchange feedback, and be their authentic selves without being penalised for it. Follmer and Spoor’s opening observation is that autistic people do not arrive at work with a neutral level of it. They arrive depleted — after a lifetime of schools and other neurotypical spaces that left them with substantial anxiety and the settled expectation of being misunderstood and rejected. The workplace, and above all the relationship with the person who manages you, then either rebuilds that safety with deliberate effort or drains what remains. Almost by default, it drains it.
The paper’s frame is the double empathy problem, and it does the work throughout: cross-neurotype interactions break down for both parties, but the autistic party is blamed for the breakdown, and the neurotypical party rarely reflects on their own contribution. That asymmetry, the authors argue, is what makes autistic psychological safety structurally fragile — every misunderstanding gets filed as the autistic person’s deficit, so every interaction carries the risk of being read as failure. From that base they name four barriers. Each is worth walking through, because each is a place where the difficulty gets located in the wrong person.
The numbers behind the argument are grim and stable: employment rates for autistic adults of around 30% in the UK in 2024, largely unchanged since 2019, and roughly 14% in the US and Canada — figures that have persisted straight through a decade of corporate “neurodiversity hiring” programmes.
Stigma, and the mask that proves the safety isn't there
The first barrier is stigma, and the paper’s treatment of it turns on a single reversal: masking is not a coping strategy autistic people happen to use at work. It is a measurement. People mask because being themselves would be punished. So wherever you find autistic employees suppressing stims, forcing eye contact, softening direct phrasing, sitting through sensory discomfort without asking for anything — you have found the absence of psychological safety, made visible.
Follmer and Spoor trace where the stigma comes from: media portrayals that reduce autism to a narrow formula (white, male, socially inappropriate, redeemed by an extraordinary niche talent), which then shape neurotypical colleagues’ expectations of what an autistic person is and what they must compensate for. The corroborating literature is blunt. Lacy Poiencot and colleagues’ scoping review of stigma and employment, out this month in JADD, quotes an autistic employee saying it is “almost easier to disclose that I’m gay or trans than to say, hey, I’m autistic,” and another whose request for a written task order was met with “How ridiculous, of course that’s not possible.” Disclosure of neurodivergence, in that review, is repeatedly followed by increased scrutiny, dismissal, or job loss.
Late diagnosis compounds this in a way the paper is unusually alert to. Many current employees grew up when their traits were filed as quirks or other conditions entirely; they receive their diagnosis in adulthood, often mid-career, sometimes in a leadership role. For them a new autistic identity has to be integrated with a professional identity that may be high-status and built on years of successful masking. Late-diagnosed leaders who treat the diagnosis as learning may grow from it; those who suppress it pay in mental health. Either way the workplace they return to has done nothing to make the disclosure safe — which is why so many make it to no one.
Norms and miscommunication — the clarifying question read as defiance
The second barrier is neurotypical norms, and it has two faces. One is sensory: workplaces built for neurotypical comfort are, for many autistic people, too bright, too loud, too smelly, too distracting — a level of discomfort that for some precludes employment altogether and for others silently taxes performance all day. The other is the spiky profile: autistic people are more likely to be exceptional at some tasks and below average at others, where most jobs assume moderate competence across the board. A colleague who is brilliant at pattern recognition and confused by an unexplained change of routine gets read as inconsistent, or “not a team player,” and receives repeated negative feedback about things beyond their control — which the paper names as exactly the signal that says you don’t belong here.
The third barrier is where double empathy bites hardest: cross-neurotype miscommunication. Neurotypical communication runs on implicit understanding, unstated assumptions, body language and tone. Autistic communication favours directness, weights content over delivery, and treats clarifying questions as expressions of engagement. So an autistic employee who responds to a task assignment with a string of questions intended to pin down the expectation is, in a neurotypical frame, rejecting the assignment or challenging authority — because that is what indirect cultures use questions for. And the autistic response to that misreading is predictable: mask more, ask less, share less. Which stifles exactly the contribution the person could have made.
The point deserves stating in full, because it is the demand-structure argument at the level of a single meeting. Nothing about the clarifying question is deficient. It is a communication style with its own logic — one that, from the autistic side, is more transparent, not less. It becomes a problem only in the presence of a listener operating on different rules who does not know they are rules, and who therefore experiences the difference as rudeness or insubordination. Milton’s original argument was that neurotypical people should take responsibility for their share of the miscommunication. Follmer and Spoor operationalise it: state expectations explicitly, discuss communication preferences early, approach each employee with curiosity rather than assumption. None of it costs anything. All of it requires the majority party to notice they have a style.
The leader relationship that won't form on its own, and what actually fixes it
The fourth barrier is partly a consequence of the first three: low-quality leader-follower relationships. Strong relationships of that kind, in the paper’s framing, form largely through similarity-attraction — people gravitate to those who look and behave like them, and that similarity signals safety and breeds trust. Mixed-neurotype pairs don’t get that instinctive start. So unless the leader makes deliberate effort, the relationship stays thin, and a thin relationship with your manager is where a single misunderstanding turns into a performance review or a stalled career.
The genuinely novel move in the paper is to run this in both directions. Autistic leaders are almost absent from the research — leadership theory, the authors note, privileges charisma and indirect social fluency, and largely overlooks the possibility that autistic people lead at all. But by the same double-empathy logic, an autistic leader who delegates directly and in detail, communicates literally, and misses non-verbal follower feedback will be read by neurotypical followers as micro-managing, blunt, or condescending — and held disproportionately responsible for the resulting friction, exactly as autistic followers are. Same asymmetry, inverted hierarchy. The authors flag this as reasoned inference rather than established finding, and that honesty is worth carrying: it is the plausible extension of a well-documented pattern, not yet a documented pattern itself.
What actually moves the metric, then, is unglamorous and mostly free — and it lands on the organisation, not the individual. Ask every employee, autistic or not, how they work best; that surfaces needs without forcing disclosure and it is universal design rather than accommodation. Manage outcomes rather than policing process, when the role permits it. State norms and expectations explicitly. Lead authentically — self-aware, transparent, honest — which the paper notes may come more naturally to autistic leaders who have spent a lifetime knowing that others think differently. Notice what none of these are: not a training module for the autistic person, not a superpower narrative, not a hiring programme with a press release. They are changes to the environment the person is being asked to survive in. Which is why the metric didn’t move for three years, and why it won’t until the effort goes there.
Citations
Follmer, E. H. & Spoor, J. R. (2026) — Creating inclusive work environments for autistic employees and leaders: The importance of psychological safety — Organizational Dynamics
Edmondson, A. C. & Lei, Z. (2014) — Psychological safety: The history, renaissance, and future of an interpersonal construct — Annual Review of Organizational Psychology and Organizational Behavior
Milton, D. E. M. (2012) — On the ontological status of autism: the ‘double empathy problem’ — Disability & Society
Poiencot, L. M., Vogtle, L. K. & Franc, I. (2026) — Autism Stigma and Employment: A Scoping Review — Journal of Autism and Developmental Disorders
Kilgallon, E., Botha, M., Dwyer, P. et al. (2026) — What the Double Empathy Problem Is (and Is Not) — Autism in Adulthood
