Dementia, Sex, and the Capacity to Consent
It’s the start of Dementia Action Week 2026, which runs between the 18th and 24th of May. The purpose of the week is to raise awareness about dementia in all its forms, share stories by people living with dementia and their caregivers, and support new research.
We will be posting throughout the week to share the work being done at the IDEA Centre on dementia and applied ethics.
In our first post, Danielle Bromwich considers whether people living with dementia can consent to sex. Caregivers and adult social care practitioners have long had to navigate how to support and protect those living with dementia in sexual relationships. Danielle argues that all parties have been let down by accounts of decision-making competence that are unfit for purpose. This has left many to come up with ad hoc frameworks of their own. Drawing on her recent work with Joseph Millum in their new book, The Ethics of Consent, and their new publication, The Separate Grounds of Competence, Danielle sketches a novel view of competence that not only provides principled and practical guidance, but enables caregivers to keep respect for autonomy separate from the duty to protect those with marginal decision-making capacities from harm.
Sandra Day O’Connor, the first woman appointed to the US Supreme Court, was familiar with Alzheimer’s long before she died of complications of the disease. She’d left the court nearly two decades earlier to care for her husband, John Day O’Connor, who’d been diagnosed with Alzheimer’s in his late fifties. Eventually, caregiving at home became unsustainable, and 17 years after being diagnosed, John was moved to a specialist care facility. The move was hard on him. His physical health had declined significantly, and he was experiencing severe anxiety and suicidal depression. But within days, he met Kay, a fellow resident, and the pair fell in love. “Forty-eight hours after moving…he was a teenager in love,” his son, Scott said. “He was happy.” As, indeed, was his wife of 55 years. Sandra Day O’Connor supported her husband’s relationship and would regularly visit the couple. By all accounts, she was relieved and delighted that the man she loved had found peace and happiness.
The WHO estimates that over 57 million people across the world live with dementia with a further 10 million new cases added each year. As various types of dementia are notoriously underdiagnosed, the real number is likely much higher. With an aging global population, these numbers are set to surge. Researchers predict that 1.7 million people in the UK will be living with dementia by 2040. That’s double the number in 2023. And yet, despite these numbers, there’s surprisingly little guidance for families, caregivers, and adult social care practitioners looking to support and protect people living with dementia in romantic relationships. Sometimes these relationships predate a diagnosis, other times people living with dementia fall in love; what happened to the Day O’Connors is not that uncommon.
When appropriately safeguarded, romantic relationships can be tremendously beneficial to people living with dementia. The detrimental effects of loneliness, isolation, and lack of meaningful connection are well established, and many working in adult social care settings can attest to the physical and psychological benefits of intimacy, loving touch, and even sex. In this case, John was transformed by the euphoria of love. However, the risks of exploitation, non-consensual sex, and other harms are high, and the costs of allowing or preventing these relationships can be devastating.
The permissibility of sexual relations involving people living with dementia doesn’t just pose hard questions for family members, caregivers, and adult social care practitioners; it also challenges consent theorists. Were a couple like John and Kay to have sex, they may do so willingly with neither illegitimately controlling the consent decision of the other. They may also still understand what sexual intercourse involves. The ethical and legal issue is whether they are still capable of consenting to sex. Whether, that is, they still retain the ability to exercise their autonomy rights and grant the other permission to have sex with them. If they are, then they appear free to pursue their desire for intimacy. But if they’re not, then it seems like they ought to be prevented from engaging in sexual relations even if that causes them distress and further mental and physical decline.
In recent work, Joseph Millum and I argue that such cases pose two questions for consent theory:
- Can persons with dementia give valid consent to sex?
- Are their caregivers—family members, nursing home staff, legally designated third parties—permitted to prevent them from having sexual relations?
These questions are rarely, if ever, posed separately. Existing views of competence assume that the capacities a person needs to give valid consent are the very same capacities they need to have a claim against other people interfering with their consent decisions. We call these Common Grounds views. They imply that the answer to the first question determines the answer to the second: if the couple can consent to sex, then third parties have no right to interfere with that consent decision. We think that this assumption is false. Instead, we think that the kind of competence implicated in consent transactions involves two separate rights: the power to waive rights and give consent and a separate right against paternalistic interference. On our alternative view—the Separate Grounds view—the capacities that ground the first right do not ground the second.
A Separate Grounds view of competence has important implications for people with marginal autonomy and their caregivers. It rejects and then replaces an assumption that explains why existing theories of competence struggle to offer plausible or principled guidance about whether and why the consent decisions of people living with dementia should be respected.
What do existing views of competence imply for the Day O’Connor case?
Let’s assume that as John and Kay’s relationship developed, so did their mutual desire for sexual relations. Existing views of competence assume that there’s only one right exercised in consent transactions: the right to consent. The only issue, then, is whether a person with dementia retains the power to exercise their rights through consent. If they do, they should be permitted to have sex. If they don’t, any attempt to engage in acts of physical intimacy should be prevented. All this is simple enough. The complicated issue—the issue over which consent theorists disagree—is what capacities a person needs to consent.
Some views are demanding. They tell us that the power to consent requires the capacity to (i) communicate a choice; (ii) understand relevant information; (iii) appreciate the situation and the consequences of different choices; and (iv) reason about the available options. The couple are unlikely to be found capable of consenting to sex on such a view. Based on the media reports, John would often fail to recognise his wife. It is therefore unlikely that he could also recognise, let alone reason through, the consequences of pursuing intimacy with someone other than his wife. And, since he was incapable of altering his claim against interference, his legal caregivers would’ve been permitted to prevent him from having sexual relations. On this view, even though his family were motivated to prioritise his wellbeing, they ought to have prevented him from engaging in acts of physical intimacy simply because he was incapable of waiving his rights against such acts.
Other views are more liberal. They tell us that the power to consent requires only the capacity to value, and that does not require a person have especially sophisticated cognitive capacities. If John and Kay could still access normative standards that they had set for themselves, they may be perfectly capable of still consenting to sex. And, as a result, it might be paternalistic to prevent them from pursuing their relationship. Of course, the all-things-considered judgement here might not be so straightforward. It’s possible that John’s current value-based desire to be in an intimate relationship with Kay is in tension with his life-long value-based desire and promise to be faithful to his wife. However, if he cannot recall that earlier commitment or value-based desire, then it seems like it would be a mistake to hold him hostage to it especially when it stands in the way of a relationship that he currently values. Given their impairments, proponents of liberal views generally agree that couples like John and Kay need help translating their values into actionable plans. This view gives us the opposite verdict to a demanding one: his caregivers were likely justified in scaffolding and supporting his romantic relationship with Kay even if that relationship included sexual relations.
Then there are risk-adapted approaches to competence. These help us see the problems with the liberal and demanding views. After all, there are variants on the Day O’Connor case in which the risk of harm associated with sexual relations is very high. It seems implausible to think, as the liberal view implies, that having the capacity to value means that John can consent to sex with, say, an exploitative suitor and that entails a duty on his caregivers not to paternalistically interfere to prevent him from harm. Of course, demanding views protect John from paramours of this sort, but they also imply that he should be prohibited from becoming intimate with someone he loves, who loves him, and poses him no risk of harm. Risk-adapted views can generate intuitive verdicts because they tell us that the capacities required for consent vary with the risks or stakes of the activity: low-risk sex only requires minimal capacities whereas high-risk sex requires more cognitive ability. The problem with these views, as we have argued at length elsewhere, is that they are unprincipled. Risk does not reliably track complexity any more than complexity tracks risk. There’s no reason to accept that the cognitive capacities required to consent should increase with the risks or stakes of the activity.
In sum, existing views of competence face a dilemma: they either generate too many implausible verdicts, or they generate intuitive verdicts that they are unable to justify.
Still, you might think, what makes the Day O’Connor case so hard is not the risk of harm or the fact that the couple have dementia, it’s that they met and fell in love while living with dementia. Suppose that before John was moved to a specialist care facility, he and his wife were still in love and wanted to continue to have sex. Some think that it is possible that John could have given prior valid consent—a type of advance sexual directive—that would permit the couple to remain sexually active as his disease developed. This, of course, is not possible in the actual case because John and Kay met while cognitively impaired. Others think that the validity of concurrent consent is easier to establish when the person with dementia is giving it to a person with whom they have a long-standing relationship with no history of abuse. His wife, unlike his new partner, is more likely to have developed a reliable communicative shorthand with John that would result in valid tokens of consent.
How should we think about this case?
What makes this such a hard case is a feature that has gone unnoticed by consent theorists: there are two rights at play in consent transactions. These rights have different cognitive thresholds. In cases of marginal autonomy—cases like those involving people living with dementia—it is possible for a person to possess the capacity necessary to exercise one but not both rights.
The first is the right to consent. What capacities does a person need to be able to exercise their autonomy through consent? The answer is given by considering what makes consent valuable to beings like us. Its value lies in the distinctive good of directing one’s own life. And, to achieve that good, we need a source of direction that gives us reasons for action. This is provided by the interests, projects, or relationships that we value and deem worthy of pursuit. All this suggests, as we argue at length elsewhere, that the capacity to value is what grounds the right to consent.
The second right is the claim against paternalistic interference. What capacities does a person need to be able to demand that others not interfere with their consent decision? It can’t be the capacity to value—in other words, it can’t be the capacity that grounds the right to consent. After all, we often interfere with the decisions of valuers when they make very harmful decisions, and we are justified in doing so. Instead, we think, to be able to legitimately make this demand it must be the case that the person making it can take responsibility for their actions. At that point, they can say to the would-be paternalist, “Don’t interfere with me. I will take responsibility for the harms that might befall me.”
This sketch of an alternative view of competence has three important implications for the Day O’Connor case.
First, it implies that in the actual case, the family were justified in supporting and scaffolding John’s relationship with Kay. It is likely that the couple had retained the capacity to value when they met. And were that the case, they could alter their claims against sexual trespass through consent.
Second, although the analysis implies that the couple may well have retained the right to consent to sexual relations, it does not imply that their caregivers were therefore prohibited from interfering with that decision on paternalistic grounds. It is unlikely that the couple had also retained the capacities to be responsible for their consent decision. And, if they didn’t, they would not have a claim against paternalistic interference. This means that their caregivers ought to take account of their overall wellbeing and consider whether there are any paternalistic reasons to intervene.
Third, while in the actual case there were very few paternalistic reasons to intervene, in variants on the case there might be. However, in order to justify interference, these reasons would need to be weighty enough to outweigh the benefits to both John and Kay of being able to make their own decisions, the considerable detriments to their wellbeing of being separated, and the benefits to their wellbeing of being allowed to pursue their relationship. As described, there do not appear to be any and there are clear costs to separating the couple, and this explains why the decision to support John’s new relationship was the right decision. However, it is not uncommon for people living with dementia to experience anger, aggression, and abrupt mood swings. Suppose that John’s mild manner had become more turbulent. Perhaps he had started shouting at Kay, flying into jealous rages when he saw her with her male carer, and leaving her increasingly distressed by their interactions. What would emerge, in this variant of the case, would be clear reasons to slow the relationship down, possibly even separate the couple, and not allow them to progress to a sexual relationship. And, since neither has a claim against their caregiver’s paternalism, this interference would be justified. Likewise, if John retained moments of lucidity in which he became distressed that he had pursued another woman, there might also be reasons—for his own wellbeing— not to allow his relationship with Kay to progress.
The Alzheimer’s Society estimates that 70% of people in care homes have dementia (or severe memory problems). Just like John and Kay, some of these people will fall in love or want to pursue valued, potentially beneficial intimate relationships. Unfortunately, existing views of competence leave caregivers, family members, and those working in adult social care home settings with inadequate frameworks for safeguarding and scaffolding these relationships. We hope to have offered a view that’s fit for purpose—a view of competence that can explain intuitive, correct verdicts in cases in which we think that people with dementia should be allowed to pursue sexual relationships, that provides guidance in harder, contested cases, and that enables caregivers to keep respect for autonomy separate from the duty to protect people with borderline autonomy from harm.
Danielle discusses competence to consent in more detail on episode 58 of the Ethics Untangled Podcast.
