Health of the nation: a philosophical approach to health insurance and the welfare state

In this blog, Rob introduces the ideas of one of his favourite philosophers, Joseph Heath, and shows how they can be applied to contemporary debates about health care and the welfare state.

Introduction: Joseph Heath, market failures and the welfare state

Joseph Heath is a Canadian philosopher who has written about a variety of philosophical topics. One of his major contributions is to apply economic ideas to moral and political philosophy. He has argued that the concept of market failure should be central to how businesses and governments make decisions.

Specifically, Heath argues that, to act ethically, businesses should refrain from taking advantage of market failures, while governments should focus on resolving them. This might sound anodyne, especially to economists, but it is actually an original and distinct philosophical view.

Market failures come in many shapes and sizes. I have identified some of the main ones below. (For an accessible introduction, I would recommend this free online course).

They have in common that they prevent buyers and sellers from carrying out trades which they would otherwise voluntarily make. In economic language, they prevent Pareto improvements from taking place – trades where at least one individual is made better off without making another worse off – and result in inefficient outcomes.

Type of Market FailureDescriptionExampleSolution
ExternalityWhen the individual reaping the benefits from an economic activity does not pay the full costs associated with itA factory polluting the local environment but not paying to clean it upTax the individual or business so they do pay the full costs for their activity
Collective Action Problem
When a group of individuals would collectively be better-off if they cooperated, but they fail to do so because they reason that they will be individually better-off if they defect 
Depletion of shared natural resourcesPrivatise the shared resource or regulate its use
MonopolyWhen are too few sellers in a market or inadequate competition between them, enabling them to price gougeUtilities like electricity, sewage, or water Regulate prices, break up large companies into smaller ones (competition policy), or take them into public ownership
Asymmetric InformationWhen one party in a trade has more information about a good or service than another, enabling them to take advantage of the otherMarket in second hand goodsThird-party accreditation of product quality

Heath takes these concepts and applies them to business ethics and political philosophy.

In business ethics, he disagrees with the two most popular views: that the primary responsibility of corporations is to maximise profits for their shareholders or, alternatively, to balance the interests of their stakeholders. Instead, he argues that the chief responsibility of business is to ensure there is an efficient allocation and use of resources in society1. They should do this by refraining from exploiting market failures. For example, a business should not pollute the local environment without sufficiently compensating those who will be affected (not necessarily something that should be taken for granted!).

1Heath’s “market failures approach” to business ethics builds on the empirical claim that capitalism and free markets have, overall, served us well for the last two hundred years and are demonstrably the most effective way of allocating resources in society to their best (most efficient) use.

In political philosophy, Heath argues that the strongest normative justification for the welfare state – the public provision of health care, pensions, unemployment benefits and so on – is that it solves market failures and provides public goods, which are valuable for everyone in society but would be underprovided if left to market forces alone.

This is distinct from rival views which claim the normative basis for the welfare state is either that it ensures everyone in society’s basic needs are met or, alternatively, that it increases fairness by redistributing wealth and income from rich to poor. To be clear, Heath is not saying these are not legitimate goals, just that they are not the strongest justifications.

I am not going to rehearse Heath’s arguments in this blog (but I have noted his key essays at the bottom). Instead, I aim to show their power and relevance by applying them to a live debate: the public provision of health care.

Specifically, I will respond to the case made by Alex Voorhoeve (another moral and political philosopher with economic inclinations) that it is permissible for a government to mandate individual citizens to purchase more comprehensive health insurance than they themselves want.

Inspired by Heath, I will attempt to show that the two arguments Voorhoeve makes for this position are unconvincing. I will conclude by suggesting that concerns that motivate Voorhoeve’s argument are not as troubling as they initially appear. We can support the public provision of health insurance – and be confident it can function effectively – without being overly worried about some citizens opting out.

What does public provision of health care mean?

It will be useful to clarify what I mean by public provision of health care before getting into the debate. The meaning of public here is probably obvious: the government rather than the market.

The meaning of provision, however, is more subtle2. It refers both to the demand and the supply side of health care, though these are distinct. The demand side essentially means the funding. The options are a single payer public system (where everyone pays the same for insurance), a fully competitive market (where individuals take out private insurance policies), or something in between.

2Or, at least, its meaning is lost on certain commentators on both sides of the political spectrum, who alternatively invoke the privatisation or socialisation of the health system without being clear on what they believe is being either privatised or socialised.

The supply side means the health care professionals (along with the equipment and facilities they use) who deliver health care services. The options are salaried employees of the state, private practitioners and companies, or something in between.

The demand and the supply side can be joined in different combinations. For example, funding of the health system could be via a single payer public system but care itself could be delivered by private practitioners and companies who are contracted by the government.

In this blog, I am focusing the question of how the demand rather than the supply side of the health system should operate. Specifically, this relates to the extent to which governments should be involved in the design and provision of health insurance.

May a government mandate more comprehensive health insurance than (some) citizens want for themselves?

Voorhoeve frames his argument in the context of the Affordable Care Act in the US. This landmark act – more commonly known as Obamacare – provided affordable health insurance for millions of Americans who were not previously covered.

It was also extremely controversial and much maligned. One particular criticism was that the Act increased the insurance premiums for some US citizens against their will, because their coverage was judged to be insufficiently comprehensive – not providing them with enough protection against the risk of ill health3.

3As a more recent example, we could consider the UK government’s increase of the rate of National Insurance in 2023 to fund adult social care. It is possible some UK citizens who will be required to pay the increased amount would prefer not to do so, even if this meant running the risk that they will not be able to access social care that they need later in life without paying very high costs.

Voorhoeve illustrates the moral quandary with the words of one US Representative:

“What do you say to Mark and Lucinda in my district who had a plan, they liked it, it was affordable, but it is being terminated [because it is not sufficiently comprehensive to meet the requirements of the Affordable Care Act]? It was what they wanted and I will remind you: some people like to drive a Ford, not a Ferrari. Some people like to drink out of a Red Solo cup, not out of crystal. You are taking away their choice.”
Representative Marsha Blackburn, 2013

This feature of the Act seems objectionable because it is an intrusion into an individual’s choice or, in other words, is paternalistic4. It requires an individual to buy more protection against risk than they may themselves be willing to.

4There is a presumption in political philosophy that paternalism is bad and that it is only permissible for governments to behave in paternalistic ways in extreme cases. For more on arguments for and against paternalism, see this free article.

Voorhoeve attempts to show why this feature is justifiable despite its seemingly paternalistic nature. He provides two arguments for this position: (a) the adverse selection argument; and (b) the social egalitarian argument.

(a) The adverse selection argument

Voorhoeve first argues that a mandatory minimum package of health insurance is justified because otherwise the health insurance market could collapse due to an adverse selection mechanism.

Adverse selection is a type of market failure that results from asymmetrical information. Consider a competitive insurance market, where an individual wants to protect themselves against the risk of some significant harm. They therefore pay a sum of money to an insurer, who promises to pay out should this harm befall them.

The insurer will want to know the probability that this harm will occur and, in the event that it does, the amount that they will need to pay out. This will enable them to offer as good a deal as possible to the individual (while still making a profit) and win their business in a competitive market.

This means that the insurer is incentivised to find out as much relevant information as they can about the individual, so they (or, more likely, the actuary working on their behalf) can calculate the risk of harm as precisely as possible. If the individual does not face a significant risk of harm, this does not necessarily pose a problem to them. They can share information about themselves freely with the insurer, without being too concerned about having to pay high insurance premiums as a result.

If they do have a specific risk factor, however, which leads them to face a risk of harm that is statistically higher than the average, this may not be the case. They could want to conceal their risk factor from the insurer so that they do not have to pay high premiums. (Importantly, we are assuming that an individual is likely to have much better information about their own health – and hence their risk of harm – than a third party).

The scene is now set for an adverse selection mechanism to emerge. Imagine a stylised insurance market where two people, Bob and Jill, want to insure themselves against the risk of ill health. The most likely scenario is that one of them will become very ill, in which case their care will cost $10,000, while one will not be ill at all, in which case their care will not cost anything.

As a result, the insurer charges Bob and Jill $5,000 each in premiums, because this is the average they will have to pay out for the two of them. Significantly, however, Bob is already an extremely unhealthy person, while Jill is extremely healthy. The insurer, on the other hand, has no idea about this. While Bob may be very happy to pay the $5,000 premium, Jill is likely to be far less eager. She therefore decides to take her business elsewhere, leaving Bob to pay the elevated $10,000 costs alone.

This is the essence of adverse selection. If we were to add more people to the insurance market, the same mechanism would lead the (relatively) healthy to drop out of market given the high costs they face, while the (relatively) unhealthy face steadily increasing premiums. A death spiral sets in and the insurance market collapses.

To avoid this situation, Voorhoeve argues, there is a strong case for a government to mandate all citizens to buy health insurance that will provide them with a standard of care that is above a certain threshold. In effect, this prevents the healthy from dropping out of market and a death spiral from setting in. Importantly, this mandate can be justified to the healthy as in their own interests. Everyone – even the healthiest – face some risk of serious ill health at some point and therefore would not want all health insurance markets to collapse.

(b) The social egalitarian argument

Voorhoeve’s second argument is that, without a required minimum level of health insurance, social egalitarian public goods, which everyone has reason to value, would be underprovided.

Social egalitarian public goods can be roughly defined as an enabler of trust: the glue that holds the fabric of society together. They are social egalitarian because they allow everyone, regardless of their rank or station, to engage with each other on a basis of equality. This is an important pre-condition for a liberal society.

They are public goods because, in keeping with classic public goods like security or clean air, they are non-excludable and non-rivalrous. This means that for anyone to be able to consume them, everyone must be able to consume them, and their consumption by one person does not affect their accessibility to others.

Public goods are liable to be underprovided by market forces alone. Since everyone can access them without restriction, individuals will be tempted not to contribute to their provision and free ride on the contributions of others. (In other words, collective action problems often emerge around the provision of public goods).

Voorhoeve lays out four distinct social egalitarian public goods related to health, which I have summarised in the table below.

Social egalitarian public goodDescription
Absence of domination and exploitation in private lifeThe less well-off are not highly dependent for their wellbeing on the powerful
Absence of domination in political lifeThe interests of the less well-off are effectively represented
Social bases of self-respectEveryone is able to participate in public life with dignity
Attitudes required for social cooperationEveryone is willing to work with others for mutual benefit, enter reasoned discussions and abide by fair arrangements

His claim is that, if adequate health care is not guaranteed to everyone in society, these social egalitarian public goods will be underprovided. This generates a prudential reason (a reason motivated purely by self-interest) as to why a healthy person should desire other people to be healthy too. On a societal level, this benefits the healthy, enabling them to live in a trusting, well-functioning community.

Importantly, if each healthy individual were left to choose whether to contribute extra resources to provide these social egalitarian public goods, they may decide not to do so. They would be tempted to free ride on the contributions of others, leading to a breakdown of trust and a collective failure to produce the social egalitarian public goods.

In turn, this justifies an external enforcement agency – such as a government – obliging everyone to contribute a minimum amount to the collective scheme. One way of achieving this would be through a mandatory minimum level of health insurance.

Why I think Voorhoeve’s arguments are unpersuasive

I have doubts about the success of both of Voorhoeve’s arguments. First, the adverse selection mechanism is a theoretical risk to the effective functioning of the health insurance market, but in practice less of a concern. Second, even if we accept Voorhoeve’s claim that adequate health care for everyone in society is necessary to the provision of social egalitarian public goods, it is not clear that the best way to provide these is through health insurance. Ideas from Heath provide the basis for both responses.

Adverse selection is more a theoretical than a practical problem

Insurance is a way for an individual to reduce the risks of harm they face in an uncertain world. I do not purchase insurance as an altruistic act. I do it because, by pooling the risk that I face with others, I make it more statistically stable and therefore predictable. This helps me to calculate how much I should spend on protecting myself against risk relative to other things.

Nevertheless, insurance does not have to be perfect, just good enough that it is worth me buying it rather than bearing the risk on my own. Heath presents this distinction in a more technical way in a paper about the insurance industry. Drawing on game theory, he distinguishes two types of equilibrium that could be reached in a cooperative arrangement such as an insurance market:

  1. The feasible set: “the set of possible cooperative arrangements that offer individuals expected payoffs higher than those that could be obtained through the non-cooperative strategy of self-insurance”
  2. The core of the game: “an arrangement from which no individual or coalition of individuals has an incentive to defect”

The feasible set compares a sufficiently good outcome to a bad outcome – the difference between having a decent package of insurance versus none at all. The core of the game compares an optimal outcome to all other outcomes – the difference between having the perfect package of insurance, tailored to the specific risks that an individual faces, versus any other package (or none at all).

The reason why adverse selection is more a theoretical than a practical problem is because, in the real world, we are more interested in ensuring we are in the feasible set than in the core of the game.

If an individual does not have the perfect insurance package, that does not necessarily mean they will drop out of the market. They will only do so if the costs involved in finding a better alternative – such as gathering information about different options and coordinating with others – are sufficiently low and the benefits are sufficiently high that it will be worth their time and effort. This provides an in-built brake on the emergence of adverse selection mechanisms.

(b) The best way to provide social egalitarian public goods may not be through health insurance

Why do governments get involved in the provision of health care? As I noted in the introduction, Heath suggests the most powerful argument is that governments are able to provide services like health care more efficiently than market forces (in the economic sense of producing Pareto improvements). This is because governments are able to create – or facilitate the creation of – large risk pools that distribute the costs of health care more evenly among the population than would otherwise be possible.

Perhaps the more common argument for government involvement in the provision of health care, however, is that this promotes equality. Yet, as Heath points out, the best way for a government to promote equality is usually through the tax system, specifically through progressive taxation which can be used to redistribute wealth and income from rich to poor.

Granted, Voorhoeve’s social egalitarian argument is distinct from other egalitarian arguments in its claim that healthy people have a prudential reason for wanting others in society to be healthy too. Promoting health among everyone in society produces social egalitarian public goods which, in turn, are valuable for all. The normative basis for this argument is efficiency, rather than equality.

I think, however, that the link Voorhoeve makes between social egalitarian public goods and efficiency is tenuous. It would be far more natural to see the value of social egalitarian public goods as consisting in the fact that they promote equality. They are desirable more because they make society a fairer place than because they make each individual better-off. If this is the case, then, as Heath suggests, the best way to provide social egalitarian public goods may be through progressive taxation rather than health insurance.

This intuition can be motivated by a simple thought experiment. Recall unhealthy Bob and healthy Jill. Imagine also that Bob is wealthy, while Jill is poor. Mandating a minimum package of health insurance in this case could actually be regressive, with poor Jill subsidising rich Bob. Jill might reasonably object if we tried to justify this using Voorhoeve’s social egalitarian argument.

“You might have increased equality in one way, through the creation of social egalitarian public goods,” she could say. “But you have decreased equality in another way, through a forced transfer from someone poor to someone rich”.

If, instead, the government were to use progressive taxation to provide social egalitarian public goods (topping up individual insurance contributions), this situation would not arise. A greater proportion of the social egalitarian public goods would be paid for by rich Bob than by poor Jill.

There are, moreover, precedents for governments using progressive taxation in this way. For example, the subsidisation of transport in hard-to-reach areas and public investment in digital connectivity are usually funded in this fashion rather than via public insurance schemes. These are policies, furthermore, which can be justified (at least in part) by an appeal to the fact that they provide social egalitarian public goods.5

5This is related to a broader critique of a tradition in egalitarian political philosophy, which views health as an area where government involvement is especially important. Arguably this tradition pays too much attention to health and not enough to other important issues like transport or communications. Government involvement in these areas can be justified using similar arguments to the ones which this tradition puts forward in relation to health.

Conclusion: public health insurance without mandates?

Voorhoeve gives two arguments as to why it is permissible for the government to mandate more comprehensive health insurance than some citizens want for themselves. I have argued these are both unpersuasive.

Where do we go from here? I see three options.

  1. Try a different route. Find other arguments that show why it is permissible for the government to mandate more comprehensive health insurance than some citizens want for themselves. I will not consider this option further.
  2. Bite the bullet. Accept it is paternalistic for the government to mandate more comprehensive health insurance than some citizens want for themselves but maintain this is the best option anyway.
  3. Reverse course. Hold that the government should refrain from mandating more comprehensive health insurance than some citizens want for themselves.

Biting the bullet is an unattractive option, given it would mean the government adopting a paternalistic policy. Yet there are some things to be said for it. This could build on findings from behavioural economics. The behavioural economics literature identifies a number of cognitive biases which cloud people’s judgements when they make decisions about how to navigate risk. These are not just one-off events, but systematic (and therefore in theory predictable) diversions from rationality.

For example, people tend to discount the value of the future in a hyperbolic way – seeing a positive experience in the present as dramatically more valuable than a positive experience a few years in the future. Moreover, when it comes to assessing small risks of significant harm, people can be innumerate – treating a low probability as zero probability.

Such cognitive biases could justify the government making some decisions on behalf of citizens, particularly in relation to risk, on the grounds that it is in citizens’ own interests. Yet the paternalism objection to this approach remains.

The more appealing option could be to reverse course. On the face of it, this may appear strange. If the government refrained from mandating more comprehensive health insurance than some citizens want for themselves, this could prevent those who opt out from accessing important but expensive health care when they need it and threaten the effective functioning of the health insurance market for those who remain.

For two reasons, however, I think these concerns are less troubling than they initially appear. The first is the argument I gave above suggesting that the risk of an adverse selection mechanism emerging in the health insurance market is quite low. An in-built brake on the emergence of adverse selection mechanisms means we do not need to worry as much about the effective functioning of the health insurance market in the absence of a mandatory minimum package.

The second is that there are ways to encourage citizens to buy a reasonably large package of health insurance without resorting to coercion. Most obviously, we could explain to citizens the power of pooling their risk of ill health with others and emphasise that, as they grow older and their chance of ill health increases, they will benefit from having paid relatively higher insurance premiums when younger.

More interestingly, we could appeal to citizens’ feelings of solidarity with others in society. Health insurance, risk pooling, and the nuts and bolts of the welfare state may sound technical and boring. Yet in large, modern societies, they are one of the most significant ways in which we are connected to others with whom we live and work. The stranger in the crowd, the person we pass in the street – simply by living alongside each other and pooling the risks that we face, we benefit each other. Communicated effectively, this reciprocal advantage may be enough to persuade each person to invest in a sufficient quantity of insurance – to the benefit of all.


Joseph Heath, ‘A Market Failures Approach to Business Ethics’ (2004)
—, ‘Reasonable Restrictions on Underwriting’ (2006)
—, ‘Three Normative Models of the Welfare State’ (2011)
—, The Machinery of Government: Public Administration and the Liberal State (2020)
Alex Voorhoeve, ‘May a Government Mandate more Comprehensive Health Insurance than Citizens Want for Themselves?’ (2018)

Philosophy Exchange Conference 2021

Recently we had the pleasure of organizing ‘The Philosophy Exchange Philosophy of Science Graduate Conference’ (PSGC 2021). The idea to put on our first conference came about in March 2021, and we got down to planning shortly after. The first step; deciding on a theme. We wanted something that would be accessible and interesting for a large group of attendees and speakers, while still providing a meaningful focal point for discussion. After bouncing around some ideas, we settled on ‘Philosophy of Science: From Theory to Practice’. 

 Our initial plan was for the conference to be a hybrid online/ in-person event, with the in-person component taking place at the London School of Economics. But this became infeasible when LSE announced, due to the pandemic, that attendees who were not LSE students would not be allowed at the event. Moving the conference fully online was a disappointing decision to have to make, as quite a few potential attendees from outside LSE had expressed an interest in joining us in-person. But the conference still ended up producing plenty of rich interaction between members of the philosophy community. 

Lively Attendance

Despite the restriction to online attendance only, we managed to attract a sizeable audience, of somewhere between 15 and 40 people for each talk, who very nicely contributed to some rich Q&A sessions. We were happy to hear from one attendee – a philosophy graduate student who made it to most of the talks – that he felt the format of a graduate student conference made for a comfortable question-asking environment, where attendees from a variety of backgrounds, and levels of expertise, were made to feel welcome. This was a contrast, he suggested, with the typical non-graduate conference, where it can feel that a Q&A is dominated by a small number of insiders to the wider academic discussion that a paper is part of. 

  Some of the discussion also spilled over onto the Discord page we created for the conference. There are various paid softwares and services out there that create platforms for hosting online conferences. But our experience suggests that Discord – which allows creating a forum, with various sub-channels, dedicated to the conference – is an excellent way to do this free of charge. As well as discussions of the talks, the Discord page worked well as a place for networking, banter, and verbal lunchtime discussion between attendees.  

  Another successful part of the conference was the lively Gather Town social. We would recommend using Gather Town for online conference socials, or something similar with games and separate rooms. If the in-person attendees can split off into separate rooms, with two or three people to a laptop, this can even work well as a way to include online attendees in some of the socializing at a hybrid conference. This kind of arrangement led to a surprisingly seamless hybrid social between online attendees and some of the members of philosophy exchange who had gathered in London to attend the online talks together. 

Wide Variety of Topics

We were pleased to see that the student presentations covered a wide range of areas of science, and subfields of philosophy. This included philosophy of maths, statistics, AI, political science, microbiology, physics, theoretical economic modelling, dream psychology, and set theory, as well as work on social epistemology, science policy advising, and the scientific realism debate. This happened despite the fact that our process for anonymously reviewing and selecting abstracts did not involve trying to ensure such a breadth of coverage. 

  The presentations were of high quality, which was not surprising since we received over 30 submission and had to make some tough calls to choose only 14 speakers for the available student speaker slots. All submissions were anonymously reviewed by graduate students from philosophy exchange, who gave them scores from 1 to 5. If the two judges had a ‘substantial disagreement’, of two points or more, they would need to have a chat and come to some agreement. The scoring criterion was, ‘how clear and well-argued a talk would I expect this speaker to give on the day of the conference’. 


  All in all, we felt this system for reviewing worked well, and would recommend it to other conference organisers who have to whittle down their submissions to fit a limited number of slots. Since we are philosophy and science nerds, we wanted to back our impression of how well our procedure worked with some kind of quantitative evidence about its inter-rater reliability (to use a social science term)! The key raw figure for this assessment was that 6 out of our 32 reviews led to a substantial disagreement. This gives a fairly impressive p-value of 0.0035 against the null hypothesis that our reviewers were no better at avoiding substantial disagreement than random chance (i.e. if our procedure for selecting scores between 1 and 5 was completely random, giving a 43% probability of substantial disagreement for each abstract, the probability of as low or lower a number of substantial disagreements as 6 would be 0.0035). Strong evidence that our procedure is at least a tiny bit better at avoiding substantial disagreement than random chance scoring is perhaps hardly a resounding endorsement (there is also modest evidence, with a p-value of 0.06, that it was least 10% better than random chance!). A more serious piece of evidence of inter-rater reliability, or so we would suggest, was that all six substantial disagreements were easily resolved after some brief discussion between the reviewers.  


Quality Talks

In terms of the quality of the talk and discussion, not to mention attendance numbers, the highlights of the conference, unsurprisingly, were our two fabulous keynote speakers; Johanna Thoma and Alistair Isaac. As she said at the start, Johanna’s talk had something for everyone in the audience. Entitled ‘Pure Risk Paternalism: What it is and How to Avoid It?’, it had elements of normative decision theory, philosophy of measurement, philosophy of economic modelling, and moral and political philosophy. It was a superb talk and very nicely set up the lively and engaging Q&A that followed. 

Alistair’s talk also led to an excellent follow on discussion in the Q&A. Entitled ‘High Precision Measurement: Theory-Neutral Knowledge from Theory-Laden Practice’, it was a fascinating excursion into philosophy of measurement, as applied to the history and contemporary practice of physics. We were introduced to the intriguing problem of the ‘bandwagon effect’, and shown how high precision measurement provides a tool to solve it, with all this leading to a novel perspective on the theory laddeness and observer the independent reality of measured quantities in physics.  


  Overall we are very pleased at how well PSGC 2021 has gone. We are extremely grateful to Johanna and Alistair for giving up their time to make this conference happen, and for the excellent work they presented to us on the day. We would also like to extend our thanks to everyone who contributed to the conference by attending, presenting, submitting an abstract, or even just sharing it on social media. 

  We definitely do not plan for this to be the last Philosophy Exchange Graduate Conference. Watch out for our upcoming values in science themed conference, which you will be hearing more about very soon!

Sufficientarianism within Limits

In the second blog in this series, I argued that the currency of sufficientarianism – the metric that can be used to quantify how much is ‘enough’ for each person – should be self-respect. If a person has self-respect, then they have enough.

In this third and final blog post, I explore the implications of this view. Specifically, I explain how it allows the rebuttal of a powerful objection to sufficientarianism: that it is indirectly self-defeating. This provides even stronger grounds to hold that the currency of sufficientarianism should be self-respect. With other currencies, like welfare or resources, the objection cannot be so convincingly answered.

What is an indirectly self-defeating theory?

A self-defeating theory is one which, if sound, would demonstrate itself to be unsound. For example, relativism holds that nothing can be true objectively, for all people at all times. Yet, if true, relativism itself would not be true objectively. Hedonism is a putatively self-defeating normative theory. It holds that people should pursue pleasure above all else. In doing so, however, a hedonist would conceivably receive less pleasure in the long run than if they had pursued another value. Pleasure is not the kind of thing that can be attained through conscious effort.

Derek Parfit made a perceptive distinction between directly and indirectlyself-defeating normative theories. 

  1. Directly self-defeating: where doing what the theory tells you leads to achieve the aims it gives you less successfully than if you did something else (as in the hedonism example).
  1. Indirectly self-defeating: where efforts or attempts to do what the theory tell you leads to achieve the aims it gives you less successfully than if you did something else.

Parfit gives impartial consequentialism as an example of an indirectly self-defeating theory. Impartial consequentialism holds that the morally right action is the one that has the best consequences, impartially considered. An impartial consequentialist might not care much about keeping up with their friends and family. They’d be focused on achieving the best outcomes for people they didn’t necessarily know. Yet if everyone was an impartial consequentialist, then no-one would care about keeping up with their friends and family, and the value of such relationships to society would be lost. 

To achieve the best consequences, impartially considered, it is plausible that it would be best if no-one (or, at least, not everyone) were an impartial consequentialist. Yet this isn’t because of an inconsistency within impartial consequentialism. Rather, it’s because impartial consequentialism neglects important social and psychological facts and so indirectly undermines itself.

How sufficientarianism could be indirectly self-defeating

Directly or indirectly self-defeating theories are unsound. If sufficientarianism were self-defeating, we should reject it and adopt a different principle of distributive justice (such as the difference principle, which I wrote about in the first post in this series).

How could sufficientarianism be self-defeating? Recall that sufficientarianism, roughly put, holds that a just distribution is one in which all members of society have at least enough to live a dignified life. Provided this condition is met, it would be fine, from the standpoint of justice, for individuals to amass as much wealth, power, and influence as they might want.

But now imagine a society where that is taken to extreme limits. A tiny fraction of the population emerges with incredible amounts of wealth, power, and influence, orders of magnitude greater than everyone else. This huge inequality between the 1 and the 99 percent would conceivably be very bad for social and political cohesion. This process could mean that those with the least in society lose out and no longer have enough to live a dignified life. For example, technological progress might drive large-scale automation of people’s jobs, leading to widespread unemployment. However, the increased profits from this automation may be concentrated in the hands of the 1 percent, with the 99 percent left without jobs or compensation.

In such cases, sufficientarianism would be indirectly self-defeating. By focusing only on ensuring that those at the bottom have enough, sufficientarianism neglects the fact that the sustainability of this situation partly depends on how much those at the top have. Too much inequality between the bottom and the top could threaten the integrity of the sufficientarian distribution. Efforts to achieve the aims given to us by sufficientarianism – that everyone has enough to live a dignified life – could therefore lead us to achieve these aims less successfully than if we followed a different principle of distributive justice, which prevented too much inequality from developing.

Why sufficientarianism is not (always) indirectly self-defeating

I think this objection is unsound. It falsely assumes that the currency of sufficientarianism is one that would allow such huge inequalities to emerge in the first place. Specified correctly, with the right currency, sufficientarianism could prevent this from happening. 

What qualities would such a currency need to have? As well as imposing an explicit lower bound on what people should have, it would have to impose an implicit upper bound on what they can have. As I explain below, I think the currency of self-respect does just this. 

My argument here draws on an insightful paper by Yannic Vitz about the relationship between libertarianism and limitarianism. These are two seemingly competing philosophical theories. Libertarianism is a set of views roughly holding that individual freedom should be prized above all other values. Limitarianism, meanwhile, is the claim that there should be a limit on the amount of resources that any one individual has. We might reasonably suppose that your average libertarian would completely reject this claim. Provided everyone is completely free, then there is no problem with having a lot. Yet Vitz shows that a popular interpretation of libertarianism implicitly does accept there should be such limits. 

This interpretation is one that includes a condition regulating how people acquire resources in the first place. This kind of libertarian might hold, for example, that it is necessary to leave ‘enough and as good for others’, in the words of John Locke. A person couldn’t simply come along and grab all the resources before any else has a chance to. If they did, they would have to pay compensation to others. Moreover, the value of this compensation would have to be greater than that of the grabbed resources, to make up for the harm inflicted as well as the value of the resources foregone. 

Vitz points out that, if one adopts this interpretation, then implicitly one is accepting that there is a theoretical limit on the amount of resources that an individual can have. An individual should not take more resources than they can compensate others for. ‘If one would do so’, in Vitz’s words, ‘one would be impermissibly well off’. 

Similarly, sufficientarianism combined with the currency of self-respect would implicitly impose an upper bound on the amount that an individual or group of individuals could have. This is because, like libertarianism, it would implicitly set some practical conditions for how society is organised. As I explained in the last blog post, these would be that: 

(a) society conforms with some basic principles of equality and tolerance; and 

(b) everyone has an opportunity to achieve reasonably-held goals. 

In practice, condition (a) would directly limit the amount of inequality that could permissibly develop. Condition (b) would do so indirectly. As a society became wealthier, so the expectations of people living in it would become more ambitious and they would need more to achieve their reasonably-held goals. Correspondingly, the level of what constitutes enough would rise, acting as a check on the growth of inequality. In combination, these conditions would arguably prevent the kind of dystopian situation I envisaged above from emerging.

One might object that, even taken together, these conditions would only be a weak check against the growth of inequality. First, even if people’s expectations did rise a society became wealthier, this could be at a much slower rate than increases in levels of inequality. Second, it is possible that people’s expectations would not rise at all (or not very much) as a society became wealthier. For example, there could be a widespread belief that individuals take full responsibility for wherever they end up, because success depends simply on how hard one tries. The fact that some people become wealthier or more powerful may have no bearing on what others reasonably expect. 

In response, I’d note that both objections are empirical in nature. They need to be backed up with evidence before we can judge whether they are persuasive. Moreover, there may be good grounds for dismissing the second objection even if it is empirically valid. This is because it is morally objectionable to hold that individuals in a society must take full responsibility for wherever they end up, regardless of whether a majority believe this is right. This belief is mistaken. It neglects the fact that some people are privileged by the circumstances of their birth; that people have basic needs which others arguably have a duty to meet; and that gross inequalities may be objectionable even if they fully reflect people’s conscious choices.1

 1 There is a parallel here with the problem of ‘adaptive preferences’ explored by Amartya Sen and Martha Nussbaum. The problem relates to situations where people living in deprivation or under oppression lower their expectations of life and need less to satisfy their preferences. This counts against preference-based theories of wellbeing, which hold that wellbeing consists in preference satisfaction. Preferences themselves are not necessarily innocent or value neutral. 

A further defence of the currency of self-respect

If I am right, then sufficientarianism can escape the objection that it is indirectly self-defeating. Yet observe that this is only the case when the currency of sufficientarianism imposes an implicit upper bound on what people can have, in addition to an explicit lower bound on what they should have. How do currencies of sufficientarianism other than the self-respect currency fare in this regard?

First, consider the resources currency. This is clearly vulnerable to the objection. It would require only that everyone has at least a base level of resources, saying nothing against the emergence of extreme inequality in the amounts of resources different individuals have.

Second, consider the welfare currency. This would prevent the emergence of large inequality in the amounts of resources different individuals have, insofar as such inequality would push the group at the bottom of the distribution under the minimum acceptable level of welfare. Yet the size of such an effect may not be large and could therefore be a weak safeguard against the emergence of extreme inequality. More dramatically, we could imagine the wealthy and powerful deliberately maintaining the poor and weak in a state of ignorant bliss in order to keep them pliant, as in Brave New World by Aldous Huxley or The Matrix films. Such a situation may be permissible under sufficientarianism combined with the welfare currency.

Finally, consider the capabilities currency. Depending on the specific capabilities included, this might prevent the emergence of extreme inequality. For instance, capabilities related to political freedoms or economic opportunities might indirectly limit the amount of inequality that could develop. This would be true to the extent that inequality could prevent everyone from having such capabilities. Yet this would not be the case with respect to other, perhaps more characteristic capabilities, such as access to healthcare, housing and education. 

With the possible exception of the capabilities currency, the currency of self-respect emerges as one of the only plausible currencies that can enable sufficientarianism to escape the objection that it is indirectly self-defeating. This provides further grounds to believe that the currency of sufficientarianism should be self-respect.


In this series of blog posts, I have motivated and fleshed out an interpretation of sufficientarianism inspired by the ideas of Rawls. This interpretation may not persuade critics of sufficientarianism that it is the best justified principle of distributive justice. There are counter-arguments that I haven’t addressed. Nevertheless, I hope that my arguments have shown that, starting from a similar position to Rawls (one that is popular with many philosophers), it is possible to arrive at a different destination. If you take Rawls’ arguments seriously, then you should take sufficientarianism seriously too.

With thanks (again) to Yannic Vitz, Cedar Green and Karl Reimer for helpful comments.

Refer to the introductory post in the series here, the first blog here, and the second blog here.

Have some (self) respect- the currency of sufficientarianism

In the first blog in this series, I explained how Rawls’ arguments for the difference principle, his preferred theory of distributive justice, also support sufficientarianism, a rival theory. 

In this blog, I consider a related question: what should the currency of sufficientarianism be? By currency, I mean the metric that is used to quantify what or how much individuals should receive under a sufficientarian distribution. Drawing on the ideas of Rawls, I argue the currency should be self-respect. In other words, provided everyone has self-respect, then they have enough.

Importantly, the debate here is about what currency is of fundamental moral importance, not simply what is an adequate proxy for this currency. We’re trying to work out what, from the perspective of justice, should be distributed fairly between people, even if, practically speaking, we need to allocate something more tangible (e.g. money) to turn the vision into reality. 

What currencies are there?

Currencies of distributive justice fall into three broad categories: welfare, resources, and capabilities.


Making welfare our currency would mean trying to identify and then distribute units of wellbeing. This is motivated by the broadly utilitarian view that wellbeing – rather than, say, rights or virtues – is of ultimate moral importance and that we can accurately quantify it. Accurately measuring welfare might seem impossible, but there are credible ways of doing so. For instance, we could use surveys to measure how happy people are in different situations (or, in more technical language, their hedonic states). 

The advantage to making welfare our currency is that it would enable us to account for the fact that some people might legitimately need more than others. For instance, disabled people might need more than non-disabled people so they could, say, buy a wheelchair or guide dog and attain the same quality of life. By using welfare as our currency, this reality would be reflected in our distributive choices.

However, with welfare as our currency we would not necessarily be able to distinguish legitimate and illegitimate reasons for differential resource allocation. An example of a potentially illegitimate reason is expensive tastes. It does not seem justifiable to give one person more because they need caviar to reach the same level of wellbeing as someone who is happy with a cheese sandwich.


Making resources our currency would mean distributing a good or service that is taken to be valuable for everyone. An obvious example is money. More sophisticated examples could be baskets of different goods and services that plausibly everyone would value, regardless of who they are.

The advantage to making resources our currency is that it would satisfy the intuition that we should be aiming for equality of opportunity rather than outcome. It would give individuals the ability to make choices for themselves about how to use their resources and, in this way, would protect and promote their autonomy.

On the other hand, using resources as our currency conflicts with the fact that different people plausibly need different amounts of resources to achieve equal levels of welfare or standing in society. (This is the same point made above in relation to the welfare currency). 


Capabilities can be understood as freedoms or opportunities to do the things we want to do. G.A. Cohen distinguishes capabilities from welfare and resources using the example of nutrition. The capability to be well nourished would mean having access to nutritious food, which allows one to carry out the activities and accomplish the objectives one wants to. By contrast, the welfare currency would focus on ‘the utility [a person] gets out of eating food’, while the resources currency would be concerned simply by a person’s ‘food supply’.

Having capabilities as our currency would involve trying to identify what they should be and then either distributing them fairly or ensuring everyone had access to them. There are a range of plausible contenders for capabilities, such as access to healthcare, housing, education, and employment. The UN Development Programme’s Human Development Index, inspired by the capabilities idea, is an example of what they could look like when operationalised. This index measures countries’ development based on their life expectancy, levels of education, and per capita income indicators.

Making capabilities our currency would arguably combine the best of the welfare and resources currencies. Like welfare, capabilities emphasise what people can actually do with various goods and services, rather than their mere supply. Like resources, they protect and promote individuals’ autonomy because they enable people to make choices for themselves about what to do with their opportunities.

Yet as well as having the advantages of the welfare and resources currencies, the capabilities currency may have some of their downsides. Under the welfare currency, it may be difficult to distinguish legitimate and illegitimate reasons for differential resource allocation. Similarly, under the capabilities currency, it may be hard to specify which capabilities are normatively valid. The resources currency struggles with the fact that different people may legitimately need different quantities of resources. Likewise, the capabilities currency faces the challenge that different people may need different amounts or compositions of capabilities to achieve the same ends.

CurrencyRationaleObjectionAssociated with
WelfareEnables us to account for the fact that some people may legitimately need more than others.Fails to distinguish legitimate and illegitimate reasons for differential resource allocation.Utilitarianism
ResourcesProtects and promotes individuals’ autonomy by enabling them to choose how to use their resources.Doesn’t capture the fact that some people may need more resources to achieve the same welfare level or standing in society as others.Gerald Dworkin
CapabilitiesEmphasises what individuals can actually do with resources, while protecting and promoting their autonomy.May be difficult to specify which capabilities are normatively valid; people may need different amounts or compositions of capabilities to achieve the same ends.Amartya Sen

What should the currency of sufficientarianism be?

In theory, sufficientarianism (along with all other distributive principles) could be combined with any of these three currencies. Welfare sufficientarianism would ensure that everyone in society had a base level of welfare; resources sufficientarianism that they had access to a minimum quantity of goods and services; and capabilities sufficientarianism that they at least those capabilities deemed essential for all. 

Nevertheless, it may be that certain distributive principles are conceptually better suited to one currency or another. For instance, utilitarianism seems clearly suited to the welfare currency.1 Unlike the resources and capabilities currencies, utilitarianism isn’t motivated by appeals to individual autonomy. Instead, like the welfare currency, it is motivated by the idea that we can assess what is good for people from an impartial perspective and make comparisons between how well-off different individuals are.

Similarly, I think sufficientarianism is best suited to a particular currency: self-respect. I won’t comment in detail on how to categorise this currency using the framework above, but I’d suggest it falls somewhere between resources and capabilities.

 1 Utilitarianism is not commonly thought of as a principle of distributive justice, but it can be interpreted in this way. It holds that a just distribution is one that maximises either the average or total amount of a given currency within society. For more on the wide variety of ideas associated with utilitarianism, see this interesting website.

What do I mean by self-respect?

Rawls argued that self-respect comprises two conditions:

  1. ‘A person’s sense of his own value, his secure conviction […] that his plan of life is worth carrying out’


  1. ‘A confidence in one’s ability, so far as it is within one’s power, to fulfil one’s intentions’

I call (1) the psychological and (2) the material condition. While Rawls doesn’t say this explicitly, I think we can take him to mean the conditions are individually necessary but only collectively sufficient for self-respect.

The psychological condition refers to an individual’s belief that their life is important and meaningful. I’d suggest that whether one can coherently hold this belief depends on whether one’s society conforms with some basic principles of equality and tolerance. A society is made of up of individuals pursuing different goals. Nevertheless, if everyone agrees that no-one is automatically more important than anyone else and that there are many legitimate goals one can have, then everyone can appreciate they are important and able to live a meaningful life, regardless of who they are or what goals they have. 

The material condition relates to an individual’s ability to translate their goals into reality. My interpretation is that this does not require that everyone does achieve their goals. People’s goals may be unreasonable; for instance, not everyone is going to be a famous singer or football player. Instead, it requires that have everyone has an opportunity to achieve goals it is reasonable for them to have, such as success in a career. This is essential because a person who doesn’t have such an opportunity cannot genuinely feel their lives are as important or meaningful as others who do. 

One might object that condition (2) is psychological, not material. Logically, a person could be ‘confident in one’s ability […] to fulfil one’s intentions’ even while living in extremely poor material circumstances. Such a person would arguably be delusional though; they would need at least some material resources to achieve their reasonably-held goals. Correspondingly, material resources would still be necessary, even if indirectly, for them to have self-respect.

Why the currency of self-respect?

Sufficientarianism with the currency of self-respect would ensure that everyone in society had enough to secure their self-respect. In line with the interpretation of self-respect I set out above, in practice this would require that: (a) society conforms with some basic principles of equality and tolerance; and (b) everyone has an opportunity to achieve reasonably-held goals.

So why self-respect? I noted above that certain distributive principles seem conceptually better suited to one currency or another. I think this is the case for self-respect in relation to sufficientarianism.

Rawls’ argument that parties in the original position would select a distributive principle that secures the social bases of self-respect gives one reason. (See the blog post before this for an explanation of this argument). What better way to ensure that the ultimate distribution secures self-respect for all than by making this the very currency of the distributive principle?

Yet there are also grounds to support self-respect as the currency of sufficientarianism even if you do not buy Rawls’ reasoning. The concept of sufficiency depends, at least partly, on a social context. Once a person’s basic needs have been met, whether they have ‘enough’ is a function of what others around them have. It plausibly means having access to a quantity and quality of goods and services that a person could reasonably expect depending on the society in which they live.

The currency of self-respect, unlike welfare, resources, and capabilities, is sensitive to social context. It would have different implications depending on the society to which it was applied. For example, in a wealthier society, meeting the material condition may be more demanding than in a poorer society; a person might need more to have to achieve their reasonably-held goals. Accordingly, a sufficientarian distribution with the currency of self-respect would set different levels of ‘enough’ in different societies (varying across both time and place). 


One objection is that the currency of self-respect appears to have regressive implications. We can easily imagine a society in which the quantity and quality of goods and services a person can reasonably expect is very low. The currency of self-respect could legitimise the existence of poverty in such a society, even though we might strongly believe that this poverty is morally bad.

My initial response would be that the currency of self-respect would plausibly imply a high minimum threshold regardless of the society in which it applied. For the material condition to be met, a person would need access to a reasonably high quantity and quality of goods and services. This means the objection has less bite. 

A more profound response is that, in this situation, we could agree that the poverty was morally bad, but not for reasons of justice. Instead, the existence of poverty could be morally bad because it showed that well-off people had not acted upon their obligations to help people in need if it came at insignificant cost to them. We might maintain, though, that the poverty was not unjust. Justice relates to the background rules and principles that should govern society, rather than obligations incumbent on individuals.

A second objection is that the currency of self-respect is insufficiently distinct from another currency of distributive justice: Rawls’ primary goods. Primary of goods are things, divided into the categories ‘natural’ and ‘social’, that Rawls claims it would be rational for everyone to want (see a list in the table below). The specific objection is that Rawls already includes self-respect (referring to it as ‘the social bases of self-respect) within his list of primary goods. So what makes the currency of self-respect different?

Type of primary goodIncluding
NaturalHealth Intelligence
SocialRights and freedoms
Income and wealth
Social bases of self-respect

In answering, I would highlight that other primary goods Rawls identifies are implied by the currency of self-respect. Specifically, the psychological condition would guarantee some basic rights and freedoms, while the material condition would ensure everyone had access to a base level of income and wealth. The currency of self-respect should be preferred to primary goods on the grounds that it is a simpler theory.

A third and final objection is that self-respect is not a workable currency of distributive justice, because accurately measuring whether someone has self-respect is impossible. 

In response, I would emphasise that we’re trying to identity a currency that is of fundamental moral importance and that this may not be easily measurable. Nevertheless, I do think that the psychological and material conditions give us some practical steps to securing self-respect for every individual. Moreover, other currencies, such as the welfare currency, might have appeared impossible to measure, but psychologists and economists have now devised credible methods to do so. This makes it seem more plausible that accurate ways to measure self-respect could be created.

In the third and final blog in this series, I’ll explain how sufficientarianism with the currency of self-respect has some surprising and advantageous features. Specifically, it enables a satisfactory response to the claim that sufficientarianism is indirectly self-defeating. This provides further grounds for holding that the currency of sufficientarianism should be self-respect.

With thanks to Yannic Vitz and Cedar Green for helpful comments.

Refer to the introductory post in the series here and the first blog here.

Same difference: how Rawls’ arguments for the difference principle motivate sufficientarianism

Rawls famously proposed the difference principle as a principle of distributive justice. 

The difference principle: social and economic inequalities are just if and only if they’re part of an arrangement that maximally benefits the worst-off members of society.

In other words, inequality between the richest and poorest is only just if it makes the poorest as well-off as they possibly can be, but not simply because it makes the richest even richer. If implemented, this would have dramatic implications. A billionaire could not justify getting another billion dollars simply by giving some money away to the poorest. They would have to ensure the poorest were as well-off as they possibly could be before helping themselves to more.

Why the difference principle?

Some background is necessary to explain the motivation for the difference principle. It is the second of Rawls’ two principles of justice, which also include:

(1) Equal liberties; and

(2a) Fair equality of opportunity (the difference principle is 2b).

Rawls argued it would be rational for parties reasoning from behind a ‘veil of ignorance’, which deprived them of knowledge of what their lot in life would be, to accept these principles as the standards regulating society. Rawls is not just saying that these are good moral principles on which to base society. He is saying it would be rational – in the best interests – of each party to agree to these principles. 

Rawls could have included a different principle of distributive justice, for instance a sufficientarian principle, in his two principles of justice. So why the difference principle? Rawls provides three separate justifications. 

First, Rawls claims that a person wouldn’t care much about they would gain above the minimum level guaranteed by the difference principle.

Second, Rawls posits that the risks of accepting a distributive principle would simply too high for anyone to rationally accept. For instance, a utilitarian principle could in theory leave the worst-off person with nothing if this somehow maximised the aggregate amount.

The third justification needs a bit more explanation. Rawls invokes Kant’s second formulation of the categorical imperative. This holds that it is morally wrong to treat a person merely as a means to an end; they should always be treated as an end in themselves. Only the difference principle, Rawls claims, would treat persons as ends in themselves when distributing social and economic benefits. By contrast, other principles, like a utilitarian principle, would be ‘prepared to impose on those already less favoured still lower prospects of life for the sake of the higher expectations of others’. 

The problem with this justification is that it involves a moral consideration. Yet Rawls explicitly states that moral considerations are not available to parties in the original position – they are motivated only by rational self-interest. However, the justification could be reconciled with this feature of the original position if we accept that treating persons as ends and not mere means is essential to securing the social bases of self-respect. By self-respect, Rawls means a person’s feeling of self-esteem and confidence that their life’s plans are worth carrying out. Importantly, it would be in the rational self-interest of a person, whoever they are, to desire the social bases of self-respect.

Assuming this is correct, Rawls might appear to have provided solid – although indirect – grounds for why parties reasoning behind a veil of ignorance would choose the difference principle over others. If they chose another, they would risk weakening the social bases of self-respect. The person already worst-off in a distribution could be made even worse-off simply because this benefitted somebody already better-off. The worst-off person would thus be treated as a mere means, suggesting that they are not important at all and thereby undermining their self-respect. Significantly, this would be true even for a sufficientarian principle. Even if the worst-off person’s standard of living was already above a given level, they could still be treated as a mere means to the end of making somebody else better-off.

Why not (only) the difference principle?

I find all of Rawls’ justifications of the difference principle unconvincing. 

The first and second are implausible. It seems perfectly reasonable for a person to desire more than the minimum level guaranteed by the difference principle, so they could go on a nice holiday, buy luxury goods, etc. In addition, it does not seem irrational to make a calculated gamble to secure an amount greater than the minimum level.

The third justification is ingenious but flawed. This is so even if one grants Rawls’ assumption that it would be rational for parties reasoning behind a veil of ignorance to secure the social bases of self-respect.

The reason is that it is unclear that Kantian considerations about treating persons only as ends and not mere means would be relevant to parties in the original position. Importantly, this is not because such Kantian considerations are moral ones; though the parties wouldn’t have moral considerations per se, they might be relevant in that they could be closely connected with securing the social bases of self-respect. Rather, it’s because parties in the original position would be selecting a principle to regulate the distribution of socio-economic positions, which would be anonymous (i.e. unattached to persons).

Kant’s maxim that one should treat persons only as ends and not mere means is grounded in the claim that, as rational beings, persons have intrinsic dignity, which is inviolable. However, since socio-economic positions do not have intrinsic dignity, it is not clear (assuming Kant’s maxim is right) that one should treat them only as ends. Consequently, it is not clear that a person’s self-respect could be undermined simply because they occupy a socio-economic position that has been treated as a mere means. 

One might respond that, since it is inevitable that all socio-economic positions will be attached to persons, treating a position as a mere means is, effectively, to treat the person ultimately occupying that position as a mere means. But, as Rawls stipulates, parties in the original position do not know the probabilities that any particular person will be attached to any particular position. 

This means that, before persons are attached to positions, no person could reasonably claim they have been treated as a mere means. Moreover, provided the distributive principle chosen by parties in the original position is in the ex ante equal interest of everyone, I’d suggest that nobody, even after they have been attached to a particular position, could reasonably claim to have been treated as a mere means.

All this means that Rawls’ third justification fails to show why parties in the original position would choose the difference principle over another that also secured the social bases of self-respect. In other words, it underdetermines the difference principle.

Why sufficientarianism?

If parties in the original position could choose a different distributive principle, what should it be? A plausible contender, I’d suggest, is sufficientarianism.

Sufficientarianism passes both Rawls’ first and second tests. It would give parties in the original position the assurance that they would not end up too badly, along with the option to make a calculated gamble to gain more than the guaranteed minimum level.

It could also secure the social bases of self-respect. Let us assume that a necessary and sufficient condition for securing the social bases of self-respect is that everyone has enough resources (both economic and social) to participate in society – for instance, they can attend and give their views at public meetings. (As I’ll argue in the second blog post, this is a plausible interpretation of what the social bases of self-respect consist of). Provided it is specified in the right way, sufficientarianism would ensure this is the case. In other words, it passes Rawl’s third test too.


Rawls’ three justifications of the difference principle fail to show why parties in the original position would choose it over sufficientarianism. The table below sums this up.

Satisfaction with the minimum levelA person wouldn’t care much about what they would gain above the minimum level guaranteed by the difference principle.Implausible – seems perfectly reasonable to want more than the minimum level.
Less riskThe risks of accepting another distributive principle would simply be too high for anyone to rationally accept.Not necessarily true – for instance if the alternative also guaranteed a minimum standard of living.
Secures the social bases of self-respectOnly the difference principle would secure the social bases of self-respect by treating persons as ends and not mere means in the structure of society. Not true – another distributive principle, such as sufficientarianism, could as well. 

If you buy Rawls’ reasoning about how to identify the right theory of distributive justice, this makes sufficientarianism a plausible contender.

In the next blog in this series, I’ll build on Rawls’ ideas about self-respect to propose a novel ‘currency’ of sufficientarianism. Refer to the introductory post to the series here.

Enough, already! Why Rawlsians could be Sufficientarians

In this series of blog posts, Rob makes three arguments related to sufficientarianism, which is a theory of distributive justice – an area of moral philosophy concerned with the question of how resources should be distributed in a society.


Sufficientarianism roughly holds that a just distribution is one in which all members of society have at least enough to live a dignified life (however that might be defined). Beyond this point, how resources are distributed becomes much less important from the standpoint of justice. This means, from the perspective of justice, it would be okay for there to be billionaires driving round in Ferraris or sailing round in luxury yachts (or whatever they like to spend their money on), provided everyone else has enough for a quality-of-life above a given level. 

I’m not going to rehash arguments for and against sufficientarianism in this series (there’s a good review article on this here). Instead, I’m going to present three distinct but related arguments about sufficientarianism. 

I’ll explain how the arguments made by the philosopher John Rawls for a different principle of distributive justice, the difference principle, also support sufficientarianism. While this doesn’t show sufficientarianism is the correct theory of distributive justice, it does – if you buy Rawls’ reasoning – make it a plausible contender.

I’ll propose a novel ‘currency’ of sufficientarianism. Currency here means the metric that can be used to quantify how much everyone needs to live a dignified life. Inspired by Rawls, I propose this should be enough to secure self-respect for each individual. Everyone should have enough to feel confident in their abilities to carry out their life’s plans.

I’ll defend sufficientarianism from a powerful objection. This claims that sufficientarianism is indirectly self-defeating, since it would allow individuals to amass such large amounts of wealth and influence that they could distort society and prevent a sufficientarian distribution from being realised in the first place. I point out that, with more sophisticated currencies than money or welfare, this is not necessarily true: sufficientarianism imposes an implicit limit on the amount of wealth and influence individuals can amass. 

I’ll conclude that these three arguments motivate a distinctive, Rawlsian-inspired interpretation of sufficientarianism. In other words, they show why Rawlsians could be sufficientarians.

Thought Experiments

Thought experiments are a key tool of analytic philosophy. They involve considering a real-world or imaginary situation which provokes a powerful moral or evaluative judgement. These judgements can then be used to test whether a philosophical theory is right or wrong, much as scientific experiments are used to test whether scientific theories are true or false. If a philosophical theory and our judgements conflict, then that could count as evidence against the theory.

Not everyone agrees on the validity of evidence gathered using thought experiments. Some argue that, because they abstract away from reality (though, as noted above, they can involve real-world situations), thought experiments ignore the emotions, relationships and every-day experiences that shape our actual decision-making. Yet these things, critics claim, are philosophically relevant. More straightforwardly, what happens if two people respond differently to the same situation? Is one correct and one incorrect, as we would conclude in the case of a scientific experiment?

These are challenging questions. Something it might help to bear in mind is the distinction philosopher Derek Parfit drew between ‘deep’ and ‘merely technical’ impossibility. A situation is deeply impossible if it violates the laws of nature. It is merely technically impossible if, however implausible it may be, it is at least consistent with reality. We should be more wary about the usefulness of our judgements in deeply impossible cases. So much about our existence would be different if the laws of nature did not hold that the validity of our judgements here seems dubious. Moreover, while it is possible for two people to respond differently to a thought experiment, this might simply call for refining the situation until one of them changes their mind.

Thought experiments can also be strange, shocking and funny. They give philosophers a chance to show off qualities – creativity and wit – that logical reasoning does not. We hear below from different members of Philosophy Exchange about the thought experiments they find most interesting and what we can learn from them.

Singer’s ‘Drowning Child’ Case – Rob

Imagine you’re walking through a park one day. You approach a small pond, where you’re surprised to see a small child splashing about. There aren’t any adults supervising the child and, as you get closer, you realise they can’t swim. If you don’t jump in and rescue them, they will probably drown. You’re wearing an expensive pair of shoes, which you will ruin if you jump in. What should you do? It seems only a moral monster could say that you shouldn’t rescue the child. Doing so will bring about great benefits (saving a child’s life) for minimal cost (ruining your shoes). Most people would probably even say you have an obligation to rescue the child. You would be morally blameworthy (even if not legally culpable) if you didn’t. Yet, Peter Singer argues, we in rich countries have the opportunity to save a child’s life, but many of us effectively let them drown. By giving relatively small amounts of money to effective charities operating in poor countries, we could save a child (even many children) from dying from preventable causes, such as malaria.

I like Singer’s ‘Drowning Child’ case because it’s so simple yet so compelling. Most objections to it focus on whether the analogy between saving the child and giving to charities operating in poor countries holds. This is important but, in my view, beside the point. Yes, we should scrutinise whether charities really are effective. But this objection argues for seeking better evidence, not that we don’t have an obligation to save the drowning child. The thought experiment leads to a radical conclusion: that we should do more, much more (perhaps giving 10% of our incomes to effective charities, as Singer proposes), to help the poorest people in the world.

If you’d be interested in reading Singer’s argument for yourself, you can download his book ‘The Life You Can Save’ as a free pdf here.

Descartes’ ‘Evil Genius’ Case – Cedar

René Descartes’ famous statement “I think, therefore I am” is the result of a thought experiment in radical doubt. Descartes sought truth. But he endeavored to doubt all of his beliefs because he figured that if he could find a belief that he could not doubt, then he could use that belief as a foundation upon which to reconstruct the rest of his belief system, this time with increased confidence in its truth.

In order to explore the bounds of possible doubt, Descartes imagined that an evil genius was tricking him into believing everything that he thought he knew about the world. Descartes found it possible to think that all of his beliefs about the external world, such as his beliefs about colors and shapes and sounds, were merely the result of the machinations of such an evil genius. Then, however, Descartes turned his thoughts to his own mind. At this point, Descartes realized that if he is doubting, then he must exist, since he must be doing the doubting. Thus, Descartes concluded that whenever he thinks, he exists. This is a belief that Descartes could not doubt.

I strongly feel the force of this argument. I have done this same thought experiment in my own head, and I find that I can admit the possibility of the external world not existing, but I cannot doubt that I exist. However, Descartes then attempted to build upon this result in order to provide his other beliefs. I find these later arguments less airtight. That said, one reason why I appreciate this thought experiment is that it provides a clear path which others can take. Subject all of your beliefs to criticism, and see how well they can withstand that criticism. Once you do this, you will have a much better sense of the epistemic value of your beliefs.

You can read more about Descartes’ adventures in radical doubt in his ‘Meditations on First Philosophy’, published in 1641, and available in many pdf forms online. See, for instance.

Schrödinger’s Cat – April

Imagine a cat, a flask of poison, and a radioactive source placed in a sealed box. If an internal monitor detects radioactivity (i.e. a single atom decaying), the flask will be shattered and release the poison, killing the cat. Without external disturbance, the cat is considered to be simultaneously alive and dead, a state known as the “quantum superposition”. Yet, when one looks in the box, one sees the cat either alive or dead, not both alive and dead. Schrodinger’s Cat is a thought experiment devised by Austrian-Irish physicist Erwin Schrödinger during his discussion with Albert Einstein. It seeks to answer the question of when and why exactly quantum superposition ends.

This seemingly abstruse narrative has much more real-world applications than people expect. Principally, it articulates the problem of measurement. A few months ago, I decided to buy a Fitbit, the smart watch that claims to measure all your vitals, including resting heart rates, sleep quality, blood oxygen level, etc. Like many other health-conscious individuals, I was curious to see how I am doing. A few weeks in, I realised that the watch does much more than quietly monitoring my bodily functions. Seeing my own vitals every single day made me exceptionally aware of the discrepancy between what I hoped them to be and what they actually are, and that simple awareness impacted my vitals. I get anxious when I see my blood oxygen level falls (yes, Covid-19 has made us all paranoid about this very particular standard). The anxiety decreased my sleep quality, which in turn, lowers my blood oxygen level even more. The downward spiral continued till one day I got considerably upset and called my mother, only to hear her laughing on the other end and calling me the “Schrödinger’s Cat”.

I love entertaining myself with Schrödinger’s Cat because it opens my mind up to appreciate that everything in the universe is so interconnected that sometimes even the curiosity of a single human being could alter the reality. It makes me feel so powerful and powerless at the same time: Powerful that I could change the objective by one action; powerless because I cannot control the direction of that change in any sense. The dichotomy that true agency only exists in the face of ultimate uncertainty humbles me.

I stopped using the Fitbit a while ago. I have come to term with the fact that I can never “quietly observe” my own vitals, and instead of inevitably altering them with my curiosity, I’d rather just let them be.

To read more about Schrödinger’s Cat, you could start by reading John Gribbin’s 1985 book, In Search of Schrödinger’s Cat, which explains in layman’s term the logic and implications of the famous thought experiment.

Are Humans Exceptional?

This co-authored post examines arguments for and against ‘human exceptionalism’. Human exceptionalism holds that humans possess a special quality which means they are of greater moral value or regard than non-human animals.

Rob starts by sketching out a dilemma – which he calls the ‘evolutionary dilemma’ – that the theory of evolution poses to human exceptionalism. Rob thinks this dilemma motivates the view that sentience grounds moral value. In response, Eva Read presents an objection to Rob’s argument and explores challenges in measuring sentience, a first step to its usability in practical contexts.

Eva is a MPhil/PhD student in the Foundations of Animal Sentience (ASENT) Project at the LSE, a fascinating interdisciplinary research programme attempting to resolve debates about animal sentience. Check out the project’s webpage here!

Human Exceptionalism and the Evolutionary Dilemma – Rob

1) What is human exceptionalism?

There are different versions of the human exceptionalism thesis. A non-exhaustive list includes:

(a) Distinct capacities A human exceptionalist might appeal to the idea that humans have distinct capacities from animals. Such capacities could include greater intelligence, emotion and sociability, along with the ability to reason and use language. Technically, they must add another claim: that these capacities attribute moral status. I think this view is the most naïve of those I canvas. Its empirical premise is doubtful; biological discoveries are demonstrating that some animals, particularly primates, share the putatively distinct capacities. The core problem, as Darwin put it, is that ‘the difference in mind between man and the higher animals, great as it is, certainly is one of degree and not of kind’.

(b) Moral personhood A more persuasive version of the human exceptionalism thesis is that humans have moral personhood, while animals don’t. Moral personhood is taken to be a quality that isn’t reducible to any specific behaviour. The most famous formulation is Kant’s. Kant argued that moral personhood derives from being a rational agent, which means being able to reason and make autonomous choices. This was so important, Kant thought, because morality only existed through the decision-making of rational agents. For example, concepts like duty and responsibility are only meaningful if we assume that they apply to agents who are able to reason and freely make decisions. Since animals are not rational agents, they do not have moral personhood. Kant even said that ‘so far as animals are concerned, we have no direct duties. Animals are there merely as means to an end. That end is man’.

Contemporary Kantians might not go this far, but do accept a significant distinction in our moral duties to humans and animals. For example, Robert Nozick proposed ‘utilitarianism for animals, Kantianism for people’. By this he meant that, when it comes to animals, the morally right thing could be to maximise their wellbeing, even if doing so involves using one as a mere means to this end (such as by sacrificing one animal to save five). Conversely, we should never use one human as a mere means to the end of maximising wellbeing (ruling out sacrificing one human to save five).

(c) The soul Many religious and philosophical traditions, such as Christianity and Platonism, hold that humans have an immortal soul and capacity to have a relationship with God (or gods), while animals have neither. We see this, for example, in Genesis 1, in which God is said to make humans ‘in his own image’. We can reasonably interpret from their omission that animals are not made in God’s own image. While these religious and philosophical traditions might hold that we do owe some moral duties to animals (for instance, based on the concept of stewardship), these are markedly less than the moral duties we owe to humans.

2) The evolutionary dilemma

The core claim of evolutionary theory is that, through natural selection, complex life evolved from simple life. Humans, correspondingly, evolved from more primitive living things. Assuming this is true, the human exceptionalism thesis faces a dilemma concerning the origins of the special quality that morally distinguishes humans from animals. Its two horns are as follows:


(a) The special quality emerged spontaneously and without precedent, which is implausible; or

(b) It evolved, which undermines the idea that it genuinely does morally distinguish humans from animals in the first place.

Let’s focus on (a) first. Imagine we are present at the birth of the first human. We are about to witness something amazing. Sufficient mutations have occurred in the genes of this new-born baby to tip it from one species into another. Of course, it will be many years until biologists develop the terminology to describe it in this way. However, there is now a difference – even if it’s not perceptible – between the species to which mother and baby belong.

The human exceptionalist might hold that this difference is not only genetic. It is also moral. The baby now has distinct capacities from its mother, which attribute it moral status; or it has moral personhood, while its mother does not; or it has a soul, while its mother does not. These moral differences could then justify treating the baby and its mother in different ways. We might hold that it would be wrong to treat the baby as a mere means to an end, but fine to use the mother in this way. For example, it might be justified to sacrifice the mother to save five people, but not the baby.

This is simply not plausible. The minor genetic differences between mother and baby cannot support such major differences in their moral value or regard. Yet, if the human exceptionalist holds the special quality emerged spontaneously and without precedent, this is what their view entails.

Now consider (b). The human exceptionalist might instead claim that the special quality evolved. While the baby possesses the special quality in its fullness, the mother possesses a proto version of it. The difference in the moral value or regard owed to mother and baby might then be small, perhaps so small as to mean differential treatment of them would not be justified. However, the human exceptionalist is not entitled to this claim without defeating his own thesis. For the special quality that morally distinguishes humans and animals to be genuinely special, a line must be drawn somewhere. It makes no conceptual sense to have, for example, 75% of moral personhood or half a soul.

The evolutionary dilemma fatally undermines the human exceptionalist thesis. The human exceptionalist must either believe something implausible or something self-defeating. They should therefore revise their thesis or adopt a different view.

3) A companions-in-guilt objection

The human exceptionalist could present the following companions-in-guilt objection. Yes, they might say, my thesis is vulnerable to the evolutionary dilemma. However, any view about what it is that attributes moral value or regard (to humans or animals) is vulnerable to a similar problem. This means the evolutionary dilemma that faces the human exceptionalism thesis has less bite.

Assume, for argument’s sake, that what attributes moral value or regard is sentience, by which I roughly mean the ability to feel pleasure and pain. Call this the sentience thesis. This is a natural view for hedonistic utilitarians, like Jeremy Bentham and Peter Singer, who think that the morally right act is the one which maximises pleasure, while minimising pain. A proponent of this view must accept that living things gained sentience at some point of the evolutionary process. We might agree that a single-celled organism isn’t sentient, but what about slightly more complex forms of life? The problem is that whichever point is selected will be somewhat arbitrary. Yet, given that sentience is taken to be the quality that attributes moral value or regard, this seems wrong. A minor genetic difference will be made into a major moral one. We would need to include consideration of the first sentient being in our moral decision-making, but not its immediate predecessor. This seems implausible.

I think this objection can be answered. This is because qualities like sentience are distinct in an important way from the special quality of the human exceptionalism thesis. Consider the difference between categorical and continuous variables in statistics. A categorical variable has a known and fixed number of possible values, for instance the answers ‘yes’ or ‘no’ to the question ‘are you married?’. By contrast, the basic unit of measurement of a continuous variable can, at least in theory, be subdivided without limit. Life expectancy in a given country could be 81 or 58.6 or 67.75. I suggest that the special quality of the human expectionalism thesis is a categorical variable, while sentience is a continuous variable. One living thing can have incrementally greater or lesser capacity to feel pleasure or pain than another.

Assuming this is correct, the problem facing the sentience thesis is much less serious than the evolutionary dilemma facing the human exceptionalism thesis. This is because the minor genetic difference between the first sentient being and its immediate predecessor would not necessarily make a major moral difference. It could instead make a minor moral difference. The ability of the first sentient being to feel a limited amount of pleasure and pain would mean that we would have to factor it into our moral decision-making to some extent. But this could be to a very small extent. This does not seem so implausible.

4) Conclusion I’ve put forward a negative argument against the human exceptionalism thesis, rather than a positive argument in favour of the sentience thesis (or another thesis). Nevertheless, I think that, if successful, my argument does support the view that whatever quality that assigns moral value or regard should be a continuous rather than a categorical variable. In turn, this gives some support to the sentience thesis and, following this, the claim that our moral decision-making should factor in the interests of the animals as well as humans. We might still have good indirect reasons for weighing the moral value of or regard for human life more highly. For example, humans could have a greater capacity for pleasure or pain than animals. Nevertheless, perhaps we should consider becoming vegetarians, or ban factory farming, or try to work out how to reduce wild animal suffering.

Response to Human Exceptionalism and the Evolutionary Dilemma – Eva Read

1) A case for a gradualist move for moral personhood

Reading the above post, it seems that if the gradualist move is available for sentience, then it is not clear why it would not also be available for moral personhood. Accepting that the special quality of moral personhood evolved does not have to be self-defeating for the human exceptionalist. It is not necessarily inconsistent to hold both a) that moral personhood is truly special – in the sense that our species has it and other animals do not, and b) that it was acquired over evolutionary time. We can imagine that whilst Homo sapiens possess moral agency to its full degree, our now-extinct Homo cousins and those who came before them could have possessed it to varying degrees.

This is not so difficult to conceive of when we consider that some levels of ‘proto-morality’ are seen in members of extant animal species today, with behaviours that indicate moral feelings of inequity-aversion, empathy, and altruism. Whilst we might not grant members of these species moral personhood, we might accept that they are moral subjects – individuals whose behaviour can be guided by moral considerations or motivations. The transition from moral subject to moral personhood may well occur on a continuous scale, but the intermediary species between Homo sapiens and our common ancestor with other Hominids are now extinct. This leaves us with a clear categorical distinction between extant species with moral personhood (humans), and those without (other animals).

2) Challenges for the gradualist sentience move

That all being said, my intuitions align with the sentientist. It seems to me that if an individual is capable of consciously experiencing the world – if it feels like something to be that individual – then they are deserving of moral consideration. However, there are significant obstacles to overcome before making the gradualist move in the case of sentience. It seems that rather than existing along a unidimensional scale from less to more experiential capacity, sentience varies along multiple dimensions. The result of this is that it would be a mistake to try to grade different species as more or less sentient than others. A useful analogy here is the concept of some animals being “more evolved” than others, a framework of thinking about evolution stemming from Linnaeus’ laddered hierarchy of evolution. Or, the concept of the Triune brain, with the mammalian brain building upon the paleomammalian brain, which builds upon the reptilian brain. In these examples we underestimated the complexity of the systems in question and enforced an order that is anthropocentric in nature, and not a true reflection of variation across species. The same can be said for sentience. Trying to answer whether a dog is ‘more sentient’ than a salmon makes as little sense as asking which of the two are ‘more evolved’.

3) Dimensions of sentience

Currently, the framework for sentience (see a summary here) considers variation across five dimensions: perceptual and evaluative richness, integration both at one time (unity) and across time (temporality), and selfhood.

Perceptual richness pertains to the detail that an experience is perceived in, and as such relates to the complexity of each sensory system. Each species will have differing levels of perceptual richness for each of their senses. Take, for example, the mantis shrimp who has up to 16 types of photoreceptor cells in their eyes, compared to the three in human eyes. Mantis shrimps are likely to have much higher visual perceptual richness than humans, but we may have higher tactile perceptual richness.

Evaluative richness is the extent to which experiences have an aversive or attractive quality (emotional valence). Emotions have an evolutionary function, with negative states (for example pain, or anxiety) guiding decision-making away from things which are bad for evolutionary fitness, and positive states (for example joy or love) guiding towards those which are beneficial.

Unity refers to the number of conscious perspectives existing within an individual. As humans our experience of the world is highly unified – we have one perspective of the world and how it relates to us. We might assume that within every other individual animal body is one singular agent, but this is not a given. Birds, for example are natural split brains, who lack a corpus callosum which joins the hemispheres of brains in other vertebrates. There are indicators from experiments on pigeons, for example, that there might be two agents acting inside one body.

Temporality relates to whether experience is perceived in a continual flow across time – as is the case for humans. This flow is not a given across other species, instead we could imagine conscious experience happening in snapshots, like in a film montage.

The last dimension is selfhood – whether an individual perceives of itself as one being, separate from but interacting with the external world. This dimension has been of particular interest in the field of animal cognition, with the famous mirror test, for example. Passing the mirror test could indicate one level of selfhood – conceiving of yourself as an individual. Other tests could indicate more complex understanding, for example whether an individual is aware of how much they know, or how confident they should be in their beliefs (metacognition).

4) Implications for moral consideration Let us imagine that we have created a profile for each species with regard to these different dimensions. Trying to find a way to simplify and combine these profiles onto a single sentience scale would be mistaken – how much weight would you give to each dimension? Evaluative richness may have stronger implications for moral consideration than, for example, temporality, but by how much? If we want to use sentience as the basis for grading species on a unidimensional scale of moral consideration, we are left with the challenge of finding an appropriate and reliable way of mapping these complex sentience profiles onto it. Faced with this challenge, perhaps we might do away with a scalar or categorical conception of whether an animal is deserving of moral concern, and instead focus resources into asking other animals what it is that they value and attempting to provide them with such.