This is the conventional wisdom, which is not based on fact. Brain cell death is thought to occur when blood glucose drops below 1 mmol/l. Most hypos are treated well above this level. People with hypoglycemia awareness can function normally at 2 mmol/l with full cognitive function and no brain damage. That 3.5 mmol/l is an arbitrary number below which there may be stress response induced symptoms, like functional brain failure. Brain failure is fully recoverable and goes away when normal blood glucose returns.
The point at which brain failure and brain damage happens depends on various factors, including diet, activity levels and the amount of injected insulin. The most elegant solution is to wean the brain off glucose and get it to use ketones of energy. Blood glucose becomes much less of an issue for the brain. It is done by eating ketogenically and becoming fat adapted.
According to this
NCBI article, brain cell death occurs at under 1.9 mmol/l but they don't say how they figured this out. If blood glucose drops very and low and very fast, it can cause coma. Then brain damage definitely occurs, but a
2007 study found that brain cells were killed by oxidative stress caused by glucose used to treat the hypoglycemia, not the coma itself. This study was done on rats so the analysis is more thorough, but the findings may not apply to humans.
The takeaway for me is that there are no convincing answers to the question of at what point hypos cause brain damage (brain cell death). It seems to be much lower than what we are told. Much of the narrative is speculative and used encourage responsible blood glucose control by T1s. I guess that is fair enough, but I really don't like scare tactics. As someone dealing with the harsh realities of T1 and everything that goes with it, I want to know the truth.