Picture the document on the kitchen table. A woman in her late sixties, recently diagnosed, still able to explain her own reasoning better than most people half her age, writes down that she wants to be enrolled in Alzheimer's research after she can no longer say so. She signs it. Ten years pass. She does not remember the table, the pen, or the reasoning. A researcher arrives with a protocol. Whose permission is in that envelope?

That is the question at the centre of a paper published in the Journal of Law, Medicine & Ethics (Volume 53, Issue 4, Winter 2025, posted online on October 16, 2025) by Dean Evan Hart of Global Public Health at NYU. It is an argument, not an experiment: no patients, no trial, no data collected. Hart's starting claim is a practical one, that a large number of people with Alzheimer's are ruled out of clinical research because they are judged unable to give informed consent, and that research advance directives (RADs) offer a way through, letting a person specify in advance what research they will permit and what limits apply. The philosophical machinery behind that is Ronald Dworkin's precedent autonomy: the recorded will of a competent person carrying force after the capacities that made it competent are gone.

Then Hart does the honest thing and states the objection at full strength. Rebecca Dresser's "changing selves" argument holds that the neurodegenerative course of Alzheimer's ruptures the psychological continuity that we usually treat as the ground of personal identity, so that, as the paper puts it, the person who writes the RAD is not the same person as the one it is later used on. If that is right, the signature is not authority, it is a stranger's instruction. The paper also notes a related position it attributes to Wilkinson, in which decisions should instead be made by clinicians with an eye to the well-being of the person who is actually present. This is not a debating point invented for a journal. It is the same argument that has been fought over for three decades in the case of the contented patient whose earlier self would have been appalled by her current life.

Hart's answer is to refuse the premise that personhood reduces to psychological capacities, and to argue from the empirical literature on what persists in Alzheimer's rather than from what he calls philosophical preconceptions. He concludes that there are strong enough reasons to reject the claims of Dresser and of Singer that personhood is ruptured or lost, while conceding that parts of self-identity are indeed damaged, and that his paper does not offer "a conclusive answer about personal identity loss in Alzheimer's cases." That concession matters more than the conclusion. What is on the table is a considered position in a live dispute, not a verdict, and readers should treat any account of it that sounds settled with suspicion.

Meanwhile the world has not waited for the philosophers. The nearest thing to a public verdict came from a national survey of Americans aged 51 and older, published in Neurology on January 13, 2009 by Scott Kim and colleagues, which asked about family surrogate consent rather than written directives: between 67.5% and 82.5% said society should allow families to make that decision depending on the scenario, and nearly 68% said so even for a first-in-human gene transfer study. Between 54.8% and 66.8% would grant their own future surrogate some or complete leeway. An interview study published in Ethics & Human Research in 2023, with 25 adults aged 55 to 83 who had an interest in dementia research, found appetite for advance research directives alongside a shortage of usable documents for anyone who wants to write one. Public opinion, in other words, is well ahead of the paperwork, and the paperwork is well ahead of the philosophy.

For a family reading this, nothing here changes what is legally possible in your jurisdiction, and nothing here says what anyone should sign. What it does is name the thing that sits unspoken in a lot of clinic rooms. When a relative who no longer recalls the diagnosis pulls their arm back from a needle, is that a person withdrawing consent, or distress that comfort can settle? The RAD case says the earlier, deliberate self should be heard. The changing selves case says the person in the chair is the one who counts. Both sides agree that visible objection in the moment must be respected. Everything else is unsettled, and it is being settled in practice, protocol by protocol, while the argument runs.