Supported Voting: A How-To Guide

This article takes as its starting point the moral requirement to include persons with serious cognitive impairments in democratic decision-making. That said, including such persons poses particular practical challenges to effective democratic participation. Nussbaum has set out the most extensive proposals for inclusion based on a model of guardianship, but we ﬁ nd they fall short due to not suitably respecting and facilitating the subjective decision-making of impaired persons. Instead, we argue for a model of co-constitution, whereby aides work within a supported decision-making paradigm to arrive at political choices in collaboration with impaired persons. To ﬂ esh out this model, we propose collaboration be based on three ideal types of interaction – expert, Socratic, and deliberative – designed to accommodate dif-fering degrees and types of cognitive impairments. We further argue that the aide should have a relationship of detached professionalism with the impaired person and therefore should not be the person ’ s carer. Finally, we conclude that this separation be combined with transparent and regular invigilation of the aide ’ s decision-making process to best safeguard the process from abuse or error.


Introduction
Persons with serious cognitive impairments (PCIs) have long been considered a challenging 'edge case' for political theorists, problematising the centrality of cognitive capabilities to theories of justice and citizenship. John Rawls, for instance, labels the status of cognitive impairments one of his three unsolved problems of justice. 1 Nevertheless, recent work in political theory from Eva Kittay, 2 Martha Nussbaum, 3 and Nicholas Munn 4 (among others) has forcefully made the case for the full inclusion of PCIs across society, including in democratic decision-making. It is therefore increasingly recognised that the reduced capacities of PCIs are no longer acceptable grounds on which to exclude them from democracy. Indeed, such capacity restriction arguments face slippery slopes towards excluding gullible or foolish noncognitively impaired persons and risk-harming outcomes by narrowing the range of perspectives available to inform the political process.
Nevertheless, exactly how to include PCIs in the political process remains a controversial question. Nussbaum has set out the most extensive proposals, but they face normative difficulties. In particular, she proposes a system of guardians voting in the best interests of those that cannot form their own coherent set of political preferences. 5 However, we contend that this recommendation is problematically paternalistic. This is reflected by the fact that it contravenes the United Nations Convention on the Rights of Persons with Disabilities which specifically sets out a requirement to replace surrogate decision-makingincluding the type of guardianship proposed by Nussbaumwith supported decision-making. 6 In this article, we aim to move beyond the guardianship paradigm proposed by Nussbaum towards a co-constituted model of supported voting appropriate for a significant number of PCIs. Instead of a third party deciding on what constitutes the independent interests of the PCI, our alternative approach proposes working with PCIs to help both develop and identify their subjective political preferences. To help distinguish our approach from Nussbaum's, we use the term 'aide' instead of 'guardian' to describe the individual supporting the PCIreflecting the distinct aim of arriving at a decision that arises from the PCI's own preferences and values. Although we argue that this model best supports the democratic inclusion of PCIs, it too raises normative questions. Aides are a way to enable PCI participation in light of the latter's particular circumstances, but they are also a further potentially fallible intermediary step in the democratic decision-making process. The aim of this article is to properly theorise the implementation of this extra step. Thus, we raise and offer tentative solutions to three questions: how should the aide approach collaborating with the PCI, what relationship should the aide have with the PCI, and how to oversee the process to ensure proper representation?
The answer to these questions illustrates a unique intersection between the fields of care ethics and what might be thought of as more overtly political theories. 7 Working closely with PCIs as aides requires many traits emphasised by care ethics including empathy and understanding. However, we find that the goal of democratic inclusion of PCIs stands in a degree of tension with a quintessentially caring relationship, in particular the emphasis on enabling the PCI's political independence irrespective of caring concerns for other aspects of their wellbeing interests. We argue that an aide playing the role of a detached professional, who draws from three defined interactive methods and is invigilated regularly, can best construct a preference and vote on behalf of a PCI who is incapable of voting for themselves. No doubt there are further normative questions raised by the use of co-constituted supported voting. We hope, nevertheless, to lay the groundwork for further fruitful research on theorising the practice of supported voting along the lines of inquiry we have outlined even if others disagree with the substantive conclusions we have reached here.

Section 1: Co-constitution
To begin with it is important to set out what kinds of persons with impairments we are addressing. We start with Nussbaum's proposals as she offers both a useful categorisation of persons with cognitive impairments and accompanying voting proposals via a three-level typology of how they impact on voting. 8 Inevitably these categorisations are generalisationsmost human characteristics display a scalar quality, and cognitive impairments are no different. Nevertheless, this is a helpful, if slightly blunt, tool for understanding the varying challenges involved in ensuring the participation of persons with cognitive impairments.
Case A individuals are de facto prevented from exercising their right to vote because of stigma or a lack of social adjustments. This includes people with limited ability to read or who have severe anxiety. Case B individuals are capable of communicating preferences to another individual but cannot seem to exercise the franchise themselves. Here Nussbaum specifies that she is concerned with people who are only capable of communicating in ways that are comprehensible to a small group of trusted individuals. Case C individuals, on the other hand, are incapable of either forming or communicating coherent views on their own.
We agree with Nussbaum that Case A individuals primarily require accommodations in terms of investment in voting infrastructure, akin to reasonable adjustments in the workplace. Case B individuals on Nussbaum's categorisation also seem to require limited helptheir views are fully formed and coherent but require translation in order to be implemented. Our concern is primarily focused on Nussbaum's category of Case Cwhere apparently neither voting infrastructure accommodations nor dedicated interpersonal translation will be enough to enable exercise of the franchise. When we use the term 'PCI', we refer primarily to individuals who approximate this Case C category. The inability to independently form views, let alone communicate them, leads Nussbaum to her substantive guardianship proposals. 9 Nussbaum's view is that we ought to give the voting rights of PCIs to trusted guardians who will exercise these votes in their interests. However, although well intentioned, her proposal's narrow focus on representing interests leads her astray. This can be grasped if we begin with a potential ambiguity in her account regarding what it means to exercise votes in the interests of PCIs. For example, one guardian might think that it is in the PCI's interests to vote for a party that promises greater funding for projects which support all PCIs, while another might think it is in the PCI's interests to vote for a party promising tax cuts so as to increase economic growth and benefit the wider economy including the PCI. Yet another guardian might think that the PCI would benefit from a generous spending programme, but that the party promising it has a poor record at managing public finances and so should not be trusted to build the scheme sustainably. In leaving it up to guardians to interpret the interests of PCI we may end up misrepresenting their interests or, worse, due to lack of oversight, enable the guardian to vote in their own interests twice.
Nussbaum argues that guardians casting votes on behalf of PCIs badly is not a unique problem, as many people make poor political decisions, sometimes under the influence of others, 10 but this analogy is inappropriate. Nussbaum's proposed solution to the ambiguity problem highlights the fact that there are in fact two distinct issues which are run together here: poor decision-making and appropriate regard for, and representation of, the PCI. Nussbaum treats a guardian voting without proper regard for the PCI as equivalent to nonassisted individuals making decisions for poor or random reasons. So understood, misinterpretation by a guardian is simply a variation on a common democratic problem of ignorance or mistake.
As Nussbaum herself acknowledges, the right to vote has both a substantive and a communicative value. 11 Substantively, ensuring the right to vote enables all citizen's interests to be expressed politically, and, communicatively, the right to vote represents equality of status. However, Nussbaum's own proposal of a guardian determining what is in the best interests of a PCI does not suitably respect this second status equality. This can be illustrated with the classic example of enlightened philosopher monarchs. While such monarchs may account for the interests of their subjects through their enlightened decision-making, and may even weigh them equally such that the interests of two PCIs are equal to two non-PCIs, there is a clear status inequality present in the monarch having the sole decision-making power. This issue is obscured by Nussbaum's focus on voting as having one's interests counted. Voting itself, particularly in mass democracies, is in fact arguably a poor method of advancing a person's interests it is well established that the material impact of any single vote prior to voting tends to approach zero. 12 Rather than being solely about interests per se, political equality entails individuals' equal rights to be heard and influence the political process through their voice and vote. 13 When a nonassisted person makes a poor political decision, whether down to lack of information or poor decision-making, they are still making the decision for themselves. 14 PCIs already face barriers to exercising their autonomy across a range of life activities including their living arrangements, their ability to travel, and in their interactions with medical professionals. Such restrictions are compounded by the harm they receive when their trusted intermediary in the political process does not give them proper regard and influence in making decisions on their behalf.
There are a number of established positions in the literature to specify the exact grounds of this harm. We do not develop a full analysis here, but at least two grounds seem highly plausible. The first is an excessive paternalism which harms PCIs through impinging on their agency, even if it is of a supported kind, when they could be exercising it. 15 The second is a denial of proper access to their democratic rights. 16 The same motivations for ensuring PCIs are democratically represented in the first place hold true for ensuring that they are properly consulted in the exercise of their rights.
The guardianship model, therefore, fails to ensure PCI's status as equal cocitizens in the political process. A supported decision-making paradigm in contrast does not require third parties to simply identify and interpret PCI interests. Rather, the objective of supported voting is for the PCI to be the nexus of decision-making: to ensure that the vote cast represents their appraisal of the relevant political voting options. However, the particular condition of PCIs means they will need assistance to carry out such an appraisal.
For PCIs to properly exercise their political rights requires collaboration between the aide and the PCI. Supported decision-making as a process is not fully captured by a myopic focus on either the political outputs of the process or what interests might be said to preexist it. 17 Leslie P. Francis and Anita Silvers have laid conceptual groundwork for this model in their work on PCIs collaborative scripting ideas of the good. They centre their account on the varying expressions and communicative methods of PCIs and how 'the collaborator's role is to attend to these expressions, to fit them together into an account of ongoing preferences that constitutes a personalised idea of the good, and to work out how to realise this good under existing circumstances'. 18 The aide, in engaging in this process, does not simply identify or interpret a set of interests but builds, in collaboration with the PCI, a coherent set of political values and preferences. From this conception, the aide can then cast a vote on the PCI's behalf.
Francis and Silver's model has faced criticism from David Wasserman and Jeff McMahan on the grounds that the overwhelming contribution of the collaborator in cases of severe cognitive impairment means that anything constructed will be owned by the former, not the latter. Analogously, they argue that, however tightly fitted a suit is, it is the work of the tailor not the wearer. 19 In the context of forming a political decision, however, this is not a devastating criticism. Just as a tailor uses specialist skills the customer does not have to craft a suit that fits the customer's particular needs, so too does the collaborator use skills the PCI does not have to help craft a political decision that fits them. In both cases the bespoke end product depends on suitable collaboration with the client and their particularities. Although general suit measurements may be objectively identifiable, crucial decisions such as the desired fabric and the number of buttons, vents and pocket styles all require subjective dialogue with the particular customer. If the right to vote were predicated on the idea of respect for our unaided decision-making capacities, then we might argue that a co-constituted vote is disrespectful. However, we, alongside Nussbaum, believe that the right to vote is part of what is required to respect each other as citizens whose voices deserve to be heard. A custom-made political decision does communicate this respect, in much the same way as a bespoke suit communicates sartorial eleganceso to speak.
This collaborative model requires knowledgeable and attentive aides equipped particularly with relationship building and interpretative skills. This is inevitably a complex process which will depend heavily on the idiosyncrasies of the PCI and their relationship with their aide. However, we can identify at least three ideal types which their relationship might follow: the expert, the Socratic, and the deliberative approaches.
The expert approach has the aide take a position of partial epistemic authority over the PCI. In this role, the aide uses their expertise in the domain of political decisionmaking to convey important information to the PCI, helping them develop their own understanding and views. This may involve either presenting novel material or breaking down and explaining existing material in a way which meets a variety of criteria including accuracy, breadth, impartiality, clarity, and accessibility. 20 The aide, of course, must sometimes defer or be responsive to the PCI's epistemic authority in particular areas, such as the experience of cognitive impairment. Nevertheless, it is the aide's expertise on politics and current affairs that guides the PCI's preferences, values, and experience toward a political decision. In this sense, this is similar to a person taking expert advice in a principal-agent relationship, such as a lawyer advising their client. The expert brings their own knowledge to bear but tries to do so in a way which the client can appreciate and understand such that the latter can make an informed decision through the aide. Clients, whether they are PCIs or nonexperts, are likely to find aspects of the relevant world confusing, and so the aide in this role aims to translate and explain relevant aspects of the world into terms they can understand and upon which they can make a decision.
Such a model is better suited to those with particularly severe cognitive impairments where there are preexisting preferences, such as a love of music or compassion towards animals, that do not cohere in a way that suggests they are capable of holding clear political preferences. It is important to note the expert is only necessary in cases in which there is both a communicative and an internal ordering impairment. Therefore, it is not enough for the person to be unable to communicate a complex political view; they must not have the capacity to order their preferences in such a way as to constitute a political view. In many cases, the expert may have to play a more interpretive role, seeking information on PCI preferences that can be used as relevant data in constructing a political preference. The more severe the impairment, the more the expert is likely to need to rely on these interpretive skills over their advisory skills.
The Socratic approach sees the aide seeking to develop the PCI's views and understanding but not necessarily from a position of epistemic authority. As in a classic Socratic dialogue, the aide is not here introducing or explaining new material but probes and questions the PCI to stimulate their critical thinking so as to help them to develop their own views. In a PCI context, this requires a particular sensitivity to the cognitive capabilities of the PCI to understand the type of questioning which will prove fruitful to developing their views.
This approach would be more appropriate for PCIs who hold a set of preferences but who would not be able to exercise the franchise even with reasonable adjustments, perhaps due to problems with memory and attention or an inability to make use of insights gained through conversation in the political arena. Here we have in mind PCIs who have a more coherent set of internal preferences, but whose primary impairment is an inability to weigh competing concerns and demands to come to a final decision, such as that which can arise with ADHD or Down's syndrome. 21 By stimulating these preexisting preferences, the Socratic aide can construct a communicable political preference that represents a set of deeply held values.
The deliberative approach sees the aide engaging in a persuasive exchange of reasons with the PCI. This is modelled on the deliberative practices which deliberative democrats see as central to democratic legitimacy. 22 Here the aide and the PCI are seeking to respond to each other's reasons with their own in an iterative fashion, building a coherent dialogue. This role bears a strong resemblance to that of the Socratic approach, but it differs in that the aide is not required to conceal their epistemic authority. Instead of offering prompts to help the PCI order their own preferences, this approach sees them using their expertise and reasoning skills to offer new information and arguments to which the PCI can respond.
This approach may be more appropriate in cases where the PCI has a set of coherent internal preferences, but they are unable to convert them into a political preference alone. Someone living with Alzheimer's disease, for instance, may have a strong set of values and preferences that are expressed clearly, but they may be unable to retain information regarding candidates for election and the policies they have proposed on a long-term basis. By engaging in debate and discussion, the aide can continually prompt and remind the person of this information, presenting them with opportunities to express long-held views and values in a way that is relevant to the present political context. The key point to recognise here, however, is that the aide is not seeking to win the argument and, thus, replace the PCIs view with their own, but to use deliberation to get a clear sense of how their values interact with current affairs. 23 Francis, in other work, makes use of the notion of a cognitive prosthesis in her discussion of collaboratively scripted conceptions of the good. 24 Much in the same way that someone with a physical impairment may use a stick to walk or glasses to see, the PCI uses their aide to assist their political decision-making. Viewed in this light, we can think of the three different approaches as providing different types of assistance for decision-making. In the same ways that some people with ambulatory impairments use sticks whilst others use wheelchairs, the PCI needs different kinds of cognitive prosthesis dependent on the needs of a given impairment.
The implementation of this framework would represent an improvement on present arrangements in most (if not all) jurisdictions. Nevertheless, there are two potential objections to our arguments which must be considered to clarify the scope and strength of the claims we present here. Firstly, some may note that in each approach we require at least a minimal level of engagement from the PCI. However, as presented by Nussbaum, there may be some people categorised under Case C who are entirely incapable of engaging in this process. Although we have doubts regarding how widespread this category is, we do concede that it is likely that such PCIs exist. The inability to respond to prompts, however, is an important conceptual distinction which should be reflected in the framework. Insofar as they exist, we propose categorising such individuals as a new separate Case D. In such cases, guardianship of the kind Nussbaum suggests may be appropriate, but we will not argue to that effect here.
Secondly, while Francis is right that cognitively able people rely on devices to aid their mental processing too, such as notebooks and calendars, some may object that there is a key difference here; notebooks and walking sticks are not persons with agency of their own. However, the role of the aide as a cognitive prosthesis need not differ in essence from this prosthesis. Many if not most social endeavours require a division of labour and teamwork. Different individuals bring together their varying skills and knowledge to create an outcome. In hierarchical organisations, decisionmaking is taken at the top and a fair, if somewhat unusual, characterisation could be that analytical or research teams beneath key decision-makers function as a cognitive prosthesis to help them make their decisions. The central point here is for the aide to see themselves as a tool to assist the PCI's own thought processes. It may be impossible to entirely eliminate the influence of the personal views and subjectivity of the aide in assisting and interpreting the PCI's thought processes. Nevertheless, the prosthesis analogy provides an ideal to aim towards.

Section 2: The aide and the PCIa professional relationship
PCIs often require support across a range of life activities, not just with voting. This is generally provided by carers who assist PCIs on a day-to-day basis. The role of carer is usually taken up by either a family member or a care-staff professional who aims to develop a close relationship of understanding and emotional and physical support for their charge. It therefore seems logical for the carer to be the person who supports the PCI in their political decision-making, as they do across a range of other aspects of their life. This obvious 'default' position has been adopted by Nussbaum and others who suggest that PCIs should have close guardians act on their behalf. 25 We deviate from this approach here and instead propose that an aide who assists with political decision-making should not be the PCI's carer. The overriding reason for this is that the care relationship is governed by a set of norms which are in tension with those required by the supported decision-making paradigm.
The relationship of political aide to PCI is necessarily one requiring significant investment in developing an interpersonal relationship of mutual understanding. It is also a relationship of dependency and support and so therefore seems an apt sphere for an ethic of care. However, the object of the relationship is facilitating the PCI's independence and political agency. Rather than the aide 'taking care' of the PCI, their role is to serve and support them in their independent decisionmaking. This distinction arises in at least two waysa concern for best interests and degrees of closeness.
In a relationship of care such as between a carer and their PCI charge, the carer's role is to understand and ensure that they act in their charge's best interests. 26 This is a task which requires a deep level of communication and understandingno two human's precise needs are identical, and so to appreciate a particular PCI's requirements is a labour of care. Ultimately, however, the carer's overriding obligation is to follow their own understanding of what constitutes their charge's best interests. In this sense the relationship is an asymmetric one in favour of the carer's decision-making. If their charge decides they want to undertake dangerous activities such as drinking a noxious substance or going out onto a busy road, it is up the carer to exercise their best judgement as to whether to permit this or not. The carer is the ultimate decisionmaker. Part of the carer's role is to protect the PCI in light of the PCI's potentially impaired judgement through the usage of their own.
It is this point which highlights the central difference between the caring role and the supported decision-making one. As emphasised in the above discussion of the various support approaches, the aim is for the aide to absent their personal judgement of the issues themselves as much as possibleand instead to enable the PCI to make their own political decision. In some sense, a decision entirely independent of outside influence is an impossible idealan aide will inevitably need to make some judgements if only to decide how to productively pose questions or inform the PCI to help them make their own judgements. However, the difference in emphasis between this and the carer's role should be clear. An aide's purpose is explicitly not to ensure the PCI make a political choice in accordance with what the aide believes is in their best interests but to act as a procedural support to the PCI making their own decision whatever it may be.
The second difference relates to the first and is that of the closeness of the relationship between the PCI and their aide or carer. One of the insights of care ethics has been to identify the positive role of emotions such as sympathy in relationships of care. 27 In explicit contrast with alternatives such as Kantian and Utilitarian ethics which emphasise universalizable abstract rules, care ethics has sought to refocus on the value of close interpersonal relationship and their attendant feelings: The preferred way of relating to one another morally can be called natural caring. By 'natural' I mean a form of caring that arises more or less spontaneously out of affection or inclination. In natural caring the phenomenological features described earlier do not require a special ethical effort; they arise directly in response to the needs of the cared-for. No mediating ethical-logical deliberation is required. 28 This type of affective bond is part of what helps sustain carers in their role. Caring work can often be deeply draining, both physically and emotionally, before one even considers the deleterious effects of societal conditions such as lack of financial support and status respect for caring labourers. 29 Under such conditions care ethics identifies that affective bonds both constitute and help motivate caring relationships. What we say here is not to disparage or undermine the importance of this aspect of care. However, the role of the aide requires an emotional distance which is in tension with this type of affective bond. As an extension of the PCI's own decision-making process, the aide must set aside their own personal feelings as well as judgements on the substantive issues. By way of comparison, doctors as a matter of professional ethics avoid treating people they have personal relationships with such as friends or relatives. The motivations and feelings which accompany such relationships may impair their professional judgement. In the aide-PCI relationship, such distance is even more important. Doctors are still mandated to act in the best interests of their patientsinformed consent notwithstanding, their overriding obligation is to care for their patients. The aide's overriding obligation is to facilitate the PCI's autonomous political decision. This may entail helping them reach political decisions which the aide finds noxious or not in the PCI's own best interests. However, the role of the aide is to support them in working through such decisions by providing them with information and helping them to consider their reasoning.
The concern regarding closeness also arises from the other directionfrom the perspective of the PCI. To return to the doctor-patient analogy, a patient may feel less able to be forthcoming with a doctor who they know personally due to not wanting to impact their relationshipfor example they may find it difficult to discuss their recreational sex or drug habits, even if such practices may have a bearing on their healthcare needs. PCIs with their relationship of closeness with their carers may find themselves conflicted between developing their own political views and maintaining their relationship with their carer. Colloquially, it is a matter of good manners to refrain from discussing politics and religion at the dinner tablethe principle behind this being to avoid bringing exogenous conflict into these personal relationships. PCIs may similarly feel pressure to avoid conflict to preserve their relationship with their carer, both out of intrinsic desire to maintain good relations and instrumentally in recognition of their degree of dependence on their carer to support them in their life. 30 These concerns are much attenuated when the PCI's counterpart is an aide with whom they do not have the same invested relationship of closeness or reliance as they would with a carer.

Section 3: Dangers and oversight
Many PCIs by virtue of their condition are vulnerable to influence and manipulation. Therefore, an understandable response to supported decision-making is to worry that it is a procedure ripe for abuse. The foregoing has outlined ways in which the supported decision-making paradigm should be approached if it is to work properly. This section is focused specifically on how best to safeguard the process. The first of these methods has already been hinted at in Section 2: separating out roles. Conceptualised as a relationship of power, PCIs are at a disadvantage with respect to both carers and aides with the latter's nonimpaired cognitive faculties. This has wider implications. For example, in the event of a conflict PCIs will find it harder to make themselves understood by third parties, and their testimony is less likely to be taken as seriously as that of their carers and aides due to their impairmenta form of testimonial injustice. 31 Separating out the roles of carer and aide disperses the power that either one might have over the PCI instead of concentrating it in a single individual. This also allows for the PCI to have someone else they can communicate with if one of their relationships is going awry. Similar to the democratic principle of being able to change one's representatives, PCIs could then speak to their carers if they felt that their relationship with their aide was not functioning properly.
Dispersing power is not enough however, another important safeguard is transparency. Given the testimonial difficulties facing PCIs, this is a significant challenge. In the worst-case scenario, the communicative process by which an aide comes to act on behalf of their PCI is completely obscured to outsiders. As a result, it then becomes difficult to verify that the aide has acted in good faith in reaching their decision. One transparency mechanism to help mitigate this is a requirement for a record of the interactions which led the aide to reach a certain understanding of the PCI's preferences. This could be in the form of both the aide's written notes explaining why they reached certain decisions and also video recording of their interactions. This would both act as a sense check for aides to remind themselves of the need to justify their communicative inferences and also as a deterrent to outright fraud. Such records would be subject to strict confidentiality restrictions to help ensure the PCI's privacy and ability to be candid with their aide. They would only be accessible by the PCI, the aide in the course of their work, or by an external inspection or regulator tasked with oversight of the process.
Finally, the most robust way to disperse power and ensure oversight would be to have a situation where the PCI collaborated with more than one aide in the political decisionmaking process. The less labour-intensive version of this would take the form of quality controlspot checks by third-party aides who would review the primary aide's records and also engage in collaborative discussions with a PCI. The primary purpose of this would be to increase the reliability of the collaborative process and to detect and deter fraud. The more intensive version would be for the PCI-aide collaborative process to regularly be run with multiple different skilled aides. These aides would receive a full briefing from the PCI's carers and loved ones but would enact the process of collaboration in private. They would then offer their own interpretations and, via a vote amongst themselves, decide who the PCI would wish to vote for. Pushing the prosthesis analogy further, we might view this as an externalisation of the higher-order process of ordering one's thoughts and managing one's competing considerations before voting. This multimember model of supported decision-making is more likely to catch individual errors, which any aide is prone to, and also to reduce the likelihood of fraud. The downside to this approach would be the additional resources required, but, if these resources can be made available without being overburdensome to other citizens, they would significantly improve the integrity of the supported voting process.

Conclusion
In this article, we have discussed a range of the moral and practical issues involved in facilitating the exercise of the franchise by people with serious cognitive impairments. We have proposed a variety of approaches for aides to adopt under the umbrella of the co-constitutive decision-making model. We have provided arguments against relying on existing carers to provide supported political decision-making, based in part on the need for professional distance as part of the role. Finally, we have given some considerations on methods to increase confidence in the process through various safeguarding measures. In providing this work, we hope to help move the question of whether PCIs should have access to the franchise, which seems well argued, onto how we should ensure they do so in a proper and robust way.