The veil of ignorance, proposed in John Rawls’ 1971 book, A Theory of Justice, is a key concept in the liberal pursuit of a just society. Rawls suggests that, when building a society, we should place ourselves behind a veil of ignorance, which prevents us from knowing our place in that society: as far as we can tell, we will be randomly assigned a position in the society we build. Rawls asserts that, when comparing two societies, the more just society would be the one into which we are more willing to be placed without knowing our position within it beforehand. A just society is “a society that if you knew everything about it, you’d be willing to enter it in a random place.”
This, Rawls argues, inevitably leads self-concerned (and, according to Rawls’ more utilitarian followers, rational) beings to construct a welfare state, in which individuals cannot fall into poverty, even if that means that it is harder to amass extreme wealth. However, humans are not rational, self-concerned beings, and it is not at all clear that the veil of ignorance would lead people to construct an egalitarian society. Key insights from decision science can help us understand how people might think under a veil of ignorance, and why.
This article will rely on prospect theory, for which Daniel Kahneman and Amos Tversky won the Nobel Prize in economics in 2002. This theory came to replace the previously widely held view of human beings as rational economic agents (Homo economicus), who deal with probabilities and consequences in a mathematically consistent and value-maximizing way. While prospect theory has been falsified, in the sense that there are economic behaviors that deviate from its predictions, its insights into people’s understanding of extremely high and low probabilities, and the importance of reference points, are well established.
The Probability Weighting Function
First, let us imagine a situation in which a loved one takes a flight, on which there is a 1% chance that the plane will crash, killing said loved one. How much are you willing to pay to remove the 1% chance of a crash, thus increasing the probability of a safe flight from 0.99 to 1? And, in a different scenario, how much would you pay to increase the probability of a safe flight from 0.5 to 0.51? If you are like most people, you are willing to pay quite a lot to reach certainty, but you don’t value the change from 50% to 51% nearly as much. This is hardly rational, in a mathematical sense, since a change from 0.5 to 0.51 has the same value as a change from 0.99 to 1. Even economists tend to overvalue certainty. This is known as the certainty effect.
The flipside of our tendency to overvalue certainty is our tendency to overvalue low probabilities, a phenomenon known as the possibility effect. For example, how much are you willing to pay for a probability of 0.001, as opposed to 0, of winning a million dollars in the lottery? And how much are you willing to pay to change the probability from 0.5 to 0.501? The way that people value a 0.001 change in probability is not constant: many of those who regularly bet on the lottery assign too much value to the mere possibility of winning.

So, how do people’s evaluations of probabilities affect their decisions under a veil of ignorance? Let’s imagine we are asked to choose between two societies in which 1000 utility points are distributed among 100 people. In one society, all the points are distributed roughly equally, whereas the other society roughly follows a Pareto distribution, in which many people have few or no utility points, while a few people have most of the points. Rawls would expect people to prefer the egalitarian society, since they would not wish to have few or no utility points, i.e. to live in extreme poverty. However, people might overvalue the possibility, however unlikely, of having many utility points—of being extremely wealthy—because they put too much value on the mere possession of this possibility.
The Reference-Based Value Function
Further complications arise from the role of reference points in people’s understanding of economic consequences. Rawls assumes that people are concerned about becoming impoverished, which should lead them to prefer a welfare state. Alternatively, the utilitarian analysis of economic consequences, adopted by some of Rawls’ followers, follows a diminishing utility pattern, meaning that the more you have of something, the less you are affected by a unit change in that thing. Put more concretely, a penniless individual values an increase of 50 dollars to his wealth much more than a billionaire would. As such, rational individuals should prefer avoiding poverty (where every utility point counts) to having the chance of becoming extremely wealthy (where utility points have highly diminished value). By contrast, prospect theory suggests that people’s preferences are relative, not absolute—we don’t just evaluate the sum of utility points, and then apply a diminishing value function; rather, we compare consequences to a reference point, and evaluate them in different ways, depending on how they deviate from that reference. A reference-based analysis from beneath the veil of ignorance might diverge from Rawls’ emphasis on the fear of becoming impoverished.
To illustrate the importance of reference points, let’s consider a classical experimental paradigm. Imagine I give you 50 dollars and then I ask you to bet on a coin toss: if the coin comes up heads, you give me 25 dollars, whereas you will earn a certain amount of money if the coin comes up tails. What is the minimum amount of reward money it would take for you to accept the bet? Now imagine instead that I told you that I am going to give you 50 dollars in five minutes time, but first I give you the chance to take the same bet. How much reward money would it take for you to accept the bet?
If you randomly assign people to two groups, and ask one group of people to bet under the first circumstance, and the other to bet under the second, you get very different patterns. People who bet money they see as their own are much more hesitant to lose it, and need a much bigger incentive to be persuaded to take the bet. This is because they frame their current financial situation as a reference, from which a lost bet would lead to a loss, whereas winning the bet would lead to a gain. By contrast, those who are told they will be given 50 dollars in the future are already in a framework of gain, whether or not they win the bet. This matters, because people are much more sensitive to losses than to gains. This is called loss aversion, and it is the main reason for the importance of reference points in prospect theory. (add graph of the value function?)

So why does reliance on references matter under the veil of ignorance? Rawls expects people to ignore their circumstances when judging a society, and focus on avoiding the risk of poverty. However, under a veil of ignorance, a poor person is unlikely to judge a society in the same way a wealthy person would, because he does not use absolute metrics, but, instead, his own reference points.
Prospect Theory Under a Veil of Ignorance
Most human societies, both now and in the past, have tended to have a roughly Pareto distribution. Large swathes of a society often possess only a little of its wealth.

One observation that baffles political commentators is that many in the lower classes tend to vote for political parties that maintain the unequal status quo, rather than promote economic equality. This is often described as voting against one’s own interests. However, prospect theory suggests that this is more complicated than it seems. Perhaps people resist high taxes for the super rich because they overestimate their likelihood of becoming rich, owing to the status quo-justifying belief that anyone can make it if they work hard enough. This effect will not be fully undone by a veil of ignorance, since the overvaluing of extremely low probabilities still applies. Furthermore, when a poor person is placed under the veil, and asked to consider being assigned a random position in one of the two societies, he brings his reference with him. As such, he is likely to consider the egalitarian society as a certain, moderate gain, and overestimate the low probability of becoming wealthy in the non-egalitarian society, while remaining relatively insensitive to the prospect of poverty. Under these circumstances, is it really surprising that a poor person under the veil might choose to bet on the unequal society? After all, what has he got to lose? By contrast, a middle-class man would pick an egalitarian society when under the veil, due to loss aversion, evoked by the possibility of living in poverty in an unequal society. This could be exacerbated by his overvaluation of the certainty of avoiding poverty, and the diminishing returns of becoming super-wealthy. Interestingly, this middle-class perspective is quite similar to the loss averse answer Rawls envisages. A very wealthy person might also prefer the unequal society, since, while both evoke strong loss aversion, the unequal society allows for the (overvalued) possibility of maintaining one’s wealth.
Human psychology undermines the key purpose of the veil of ignorance: namely, to allow individuals to judge a society independently of their current situations. Even Rawls might not be able to avoid these effects, since his own perspective may be the result of his socioeconomic background. While the rational, utilitarian choice under the veil of ignorance is equality, it seems unlikely that people under the veil would make rational evaluations. Thus, the veil of ignorance might be mostly effective in convincing middle-class individuals to support welfarism, because of people’s class-dependent reference points, and the way they evaluate extreme probabilities.
[…] — wouldn’t be aware of their future role or position in said society. Therefore, they would need to take an objective perspective to ensure that everyone would have equal […]
“One observation that baffles political commentators is that many in the lower classes tend to vote for political parties that maintain the unequal status quo, rather than promote economic equality. This is often described as voting against one’s own interests.” Consider the statement: Government policies can promote economic equality. Now answer this question: Is the belief that governments can promote economic equality, based on an analytic or synthetic truth? That is, are we dealing with a truth by definition, or a truth by discovery? If the former, how can the effectiveness of a social process be given analytic status? If the later, what is the supporting evidence? In the Western world and richer parts of East Asia, governments are taxing and spending at least a third, and in some cases as much as a half of their GDPs, and yet has economic inequality declined since the mid-20th century, remained approximately… Read more »
Inequality persists through history for a reason: limited resources, after all. If everyone cannot have a spacious house or reproduce, what equality is logically attainable? Existing inequalities have the positive side of resolving visceral survival pressures rather seamlessly, so that we never see those pressures naked. It is not that irrational to gamble for your status.
Very interesting article that raises the broader question about human rationality in sizing up situations and making predictions. It is not just judgment heuristics (Tversky and Kahneman’s work) but many areas in social psychology (confirmation bias, motivated reasoning, cognitive dissonance, fundamental attribution error etc.) and psychodynamic clinical psychology (defense mechanisms) that show how far us humans are from the post-enlightement ideal of the objective observer of reality. How we all imagine ourselves to be objective little scientists and how badly we fool ourselves.
I was simplifying a bit in my first comment; the way Rawls actually writes it up, individuals behind the veil of ignorance have no knowledge of any probabilities about what positions they might end up in in society (see TJ rev. ed. p. 46 — the paper you link to misunderstands this). So have no knowledge about their chance of being rich or poor, for example. And Rawls thinks that when we don’t know the probabilities we’ll aim to maximize the position of the worst-off person in society because we’re maximally risk-averse in those circumstances. It’s your insurance example but taken to the extreme. You’re correct that the Rawlsian veil-of-ignorance ideal doesn’t manifest in the practice. It’s a thought experiment that will never happen, so of course it won’t! But that isn’t a problem with the theorizing behind the veil of ignorance in A Theory of Justice so much as… Read more »
Rawl’s veil of ignorance fails where utilitarian philosophy in general fails: I already have a position and a personal identity that is known to me. Given this obvious reality, why would I judge people for anything other than how their actions affect my interests?
1) The possibility effect as you describe it shows that individuals will sometimes opt for seemingly strange choices in marginal improvements, e.g from a 0 -> 0.001 chance instead of a .5 -> .5001 chance. But that marginal comparison doesn’t show that individuals will make the same choice between the improved options, e.g opt for a 0.001 chance over a .5001 chance — which is what’s relevant to the Rawlsian argument. 2) Your claim about reference points and the perspective of the poor person may be true about how people judge equality or inequality in the real world, but it doesn’t apply to the Rawlsian veil. Remember that the entire point of the veil of ignorance is that a person behind it is ignorant of about his or her personal standing in the world. That includes, among other things, his or her wealth. So no one in the Rawlsian framework… Read more »