Balancing safety and autonomy: reasonable risk for autistic adults

Devon’s parents had never let him near the stove. Not once, not even supervised, for twenty-six years, because somewhere early on they had decided the risk of a burn outweighed the benefit of him ever learning to cook, and that decision had calcified into a household rule nobody had reexamined since. A support worker, new to the family and unburdened by twenty-six years of accumulated fear, suggested starting with toast. This is the post about what happened after that suggestion, and about the difference between protecting someone and shrinking their life down to the size of your worst fear for them.

Every skill anyone has ever learned came with risk attached

Let me be absolutely clear about what I’m saying, because this topic gets uncomfortable fast, and for good reason. I am not telling you to ignore real danger. I am telling you that zero risk is not the same thing as safety, and a life engineered to eliminate every possible harm is also a life engineered to eliminate cooking, independent travel, internet access, and relationships, because every one of those things carries some risk of something going wrong. That is true for every human being who has ever learned to do any of them. It does not stop being true because the person learning is autistic.

The disability field has language for this: dignity of risk, a term coined by Robert Perske in 1972 to describe the idea that overprotection can be its own form of harm, denying a person the growth, competence, and self-respect that come from managing real risk in their own life. None of this is new thinking. It has simply been slow to move from professional literature into the daily decisions families make in their own kitchens.

Naming the actual risk instead of the imagined one

The fear that kept Devon off the stove for twenty-six years was never named specifically. It was a general, diffuse “he could get hurt” that had never been broken down into what, specifically, and how likely. When his parents finally sat down and named it, the actual risk was a minor burn from touching a hot pan handle, an injury that happens to people cooking dinner in millions of households every single night, survivable, treatable, and, with a specific safeguard, largely preventable.

This is the pattern across almost every domain families avoid out of caution: cooking, community access, internet use, relationships. The imagined catastrophe, unnamed and unsized, justifies a blanket prohibition. The actual, specific risk, once named, usually has a specific, proportionate safeguard that doesn’t require eliminating the activity altogether.

Four domains, four proportionate safeguards

Cooking’s real risk is minor burns and cuts. The proportionate safeguard is supervised first attempts, one task at a time, and safety habits built through repetition rather than a single lecture about hot stoves. Community access’s real risk is disorientation. The safeguard is a practiced route, a communication card, and a check-in plan, widened as competence grows. Internet use’s real risk is exploitation and scams. The safeguard is explicit, ongoing conversation about red flags, not a blanket ban that mostly succeeds at pushing the activity underground where no safeguard can reach it. Relationships carry the real risk of heartbreak and occasional poor judgment in unfamiliar territory. The safeguard is honest, continuing conversation, not surveillance, which teaches secrecy rather than judgment.

In every domain, the through-line is the same: the safeguard should be sized to the actual harm, not to the worst imaginable version of it, and it should shrink over time as competence is demonstrated, not stay fixed forever at its most restrictive setting.

Gradual exposure, not a leap of faith

Devon’s cooking education didn’t start with a full meal alone. It started with toast, under direct supervision, for two weeks. Then scrambled eggs, supervision at a slight distance. Then, six months in, a full simple meal, his parents in the next room for the first few attempts and eventually not at all. He burned himself once, a minor one, and reached for the aloe vera without panicking, because the safeguard included what to do if the risk materialized, not just how to avoid it entirely.

His parents told me afterward that the six months taught them something bigger than how to teach their son to cook. It taught them what he could actually handle, which turned out to be considerably more than either of them had believed for the previous twenty-six years. That gap, between what a person can handle and what a family believes they can handle, is often the real thing being protected when a blanket “no” gets applied to an entire domain of life.

The cost of zero risk is real and usually invisible

None of this was inevitable. Devon could have gone another twenty-six years without touching a stove, and nothing dramatic would have marked that absence. No incident report gets filed for a skill that was never attempted. That’s exactly what makes the cost of overprotection so easy to miss: it doesn’t show up as a crisis. It shows up as a smaller life, quietly, year after year, with nobody ever forced to reckon with what was lost.

The families who eventually widen the circle, one supervised step at a time, usually describe the same thing afterward: relief, more than anything, that the catastrophe they’d been protecting against for years turned out to be manageable, survivable, and worth the risk all along.

Jim Palasty is the founder of OASIS for Autism and a single father of an adult autistic daughter in Michigan.