1. What We Talk About when We Talk About Injustice with Children

Talking to our children about injustice, if done well, will contribute to many different parenting tasks. Consider just a few. Since injustice is complex, it will help our children to understand complex ideas. Since what counts as injustice is a deeply contested question, it will help our children to think critically about the world. And since justice and morality are intimately related, it will help our children to become morally better people. But notice two things: first, all these tasks could also be accomplished without talking about injustice. Teaching children science will help them to grasp complex ideas; teaching them logic will help them to think critically; reading them stories will help them to become morally better people. And second, notice that these tasks would be just as necessary in a fully just world as in an unjust one. Complex ideas, critical thinking, and empathy are crucial tools for children in even the best world.

But while talking about injustice contributes to many different parts of parenting, it also plays a more unique role: it prepares our children to be good citizens in an unjust world. As John Rawls reminds us, a just society will only ever be possible if citizens acquire the appropriate sense of justice in childhood (Reference Rawls1999).Footnote ³ If instilling a sense of justice in children is necessary simply to maintain an already-just society, then it will be even more important in an unjust one. In a society that is already just, citizens need only appreciate and support the status quo; in an unjust society, however, moving towards justice will require a citizenry that not only recognizes instances of injustice, but can imagine more just alternatives and be motivated to work towards them. And this requires specific discussion of injustice. Especially in wealthy, white families, actual children often spend their childhoods insulated from and ignorant of many of the concrete injustices in the world. If left up to chance, these children might never develop a sophisticated understanding of injustice at all, or – even worse – they might absorb ideas that legitimize the unjust status quo. Being able to respond to injustice responsibly requires a thoughtful and morally defensible understanding of injustice, and responsible parents will need to help their children to develop it.Footnote ⁴ But while I’ve argued for these points elsewhere, here I will just assume them for the sake of argument. What I am interested in here is not whether parents should talk to their children about injustice, but rather how they can use the tools of the non-ideal theory debate to do it well.

If the aim of talking to children about injustice is to prepare them to fulfil their duties in an unjust world, then, responsible conversations about justice will need to go beyond a focus on complexity, critical thinking, and empathy. Children will need to understand injustice in particular, and they will need to understand it at a level that lets them do things with that understanding. That is, they will need to be able to judge circumstances they encounter as just or unjust, and to know what to do in response.

While there may be many methods of helping children to achieve this kind of understanding, I’m going to focus here on one: that is, on helping our children to develop informal mental models of injustice. I’ll do so for two reasons. First, children already utilize models to understand the world, and recognizing this puts us in a position to help them to do it better. I am not, of course, talking about the formal models employed by economists or computer scientists. Rather, by “model” I mean the more general thing that Nancy Nersessian means: a representation of a system with interactive parts that represents the interactions between those parts and thereby helps us to understand them (Reference Nersessian and Vosniadou2013: 4). Models in this general sense are fit for use by even young children, since they can be relatively simple, purely mental, and can model all sorts of situations and interactions, including the kinds of relationships in the social world that characterize injustice. And indeed, the use of mental models for children is not only possible, but extremely beneficial. What models of all sorts do for us is help us to arrange and systematize what we know about the world in a way that lets us zoom in on a problem that concerns us, while also taking advantage of the wider information that situates and constrains it. In Nersessian’s words,

Just as we can switch our attention in a photograph between the face of a loved one and the landmark they stand in front of without losing awareness of either, our mental models let us efficiently focus on and manipulate different of their parts depending on our purpose.

Having the capacity to develop mental models in this way is clearly adaptive for human beings, since it allows us to anticipate our environment and potential outcomes of actions more easily. In fact, the development of informal mental models is so evolutionarily adaptive that we all invariably develop and rely on them whether or not we are aware of it. The content of our particular mental models, however, is by no means invariable. Instead, our models will be significantly influenced by both our own perceptions and the explanations of the world that we receive from others (Nersessian Reference Nersessian and Vosniadou2013: 22), and it is this development of domain-specific knowledge that parents can help their children with. Children seem from young ages to have a keen natural interest in justice (Abbad Reference Abbad1988; Rutland and Killen Reference Rutland and Killen2015), and they will therefore unavoidably represent the world to themselves in ways that track and explain which things count as injustices. But parents can influence which things count as injustices on the models they end up developing.

I am not suggesting that parents ought to aim to impose models of injustice on their children and encourage them to accept those models unquestioningly. I take it that Rawls is right that there is significant reasonable disagreement about what justice requires, and I am compelled by the general idea that we owe our children an open future in which they can use their moral powers to decide what they believe and who they want to be (Feinberg Reference Feinberg1992 [1980]). But try as we might, it is impossible to leave a child’s future wholly open (Mills Reference Mills2003; Millam Reference Millam2014). As Cass Sunstein and Richard Thaller remind us, our choices are not made in a vacuum, and there is no such thing as a neutral choice circumstance. Any possible set of background conditions will nudge us in some directions rather than others (Sunstein and Thaller Reference Sunstein and Thaler2008). No matter how hard we try to ensure that our children are left free to develop their own sense of justice, then, both the messages they receive from the world about injustice, and the conversations we have – or choose not to have – with them, will play important roles in the sense of justice they go on to form. The best we can do is to offer them a range of models of injustice, point out strengths and weaknesses of each, and actively encourage them to critically develop their own model in response.Footnote ⁵ In doing so, we should be attendant to the strong framing effects exerted by the status quo, and we can certainly take advantage of the careful reasoning that we ourselves have done about injustice.Footnote ⁶ For instance, the models that children build of who ends up suspended in school might take many forms, and if parents recognize that the status quo is likely to support a model that explains suspension via disrespect and violence, they can encourage their child to consider an alternative that incorporates the ways that racial biases lead us to interpret the same behaviours positively, negatively, or neutrally in different populations (Skiba Reference Skiba, Bangs and Davis2015). Recognizing that children form and utilize these mental models allows us as parents to strengthen and improve the models they land on by teaching them to think critically, and by encouraging them to actively consider models that address the deficiencies of the status quo.

The second reason to think about injustice conversations in terms of models, is that it lets us take advantage of the philosophical work on non-ideal theory that already thinks about justice in terms of acceptable and unacceptable idealization of models. One of the primary ways of understanding the point of contention in the non-ideal theory debate is to focus on how, when, and to what extent our work on justice ought to idealize away from actual features of our messy and unjust world. Do we best understand the requirements of justice by abstracting away from the real world and imagining what justice would require if everyone behaved as they should? Or should we understand the requirements of justice by attending directly to the clear injustices that confront us in the world? In other words, which features should be built into our model of justice, and which should not?

Charles Mills has helpfully distinguished between two different kinds of idealization our models of justice might employ. We’ll call them “descriptive model idealization” and “ideal model idealization” (Reference Mills2005). Descriptive model idealization involves leaving out some features of the world due to the sheer complexity of the world we are trying to model. Imagine that we are trying to model the way that traffic flows through a traffic circle, and consider everything that is true of that circle: its latitude and longitude, the colours of passing cars, the number of passengers in each car. If we tried to include all of this information in our model, we would be so quickly overwhelmed by the sheer volume of data that we would struggle to develop a model at all. But luckily, most of that information also makes no difference to our model’s ability to function appropriately. Generally, longitude and car colour simply won’t matter for our purposes. So in order to build a model that we can use, we can and should engage in descriptive model idealization by leaving all of that extraneous information out.

The abstraction involved in ideal model idealization is different. Here we idealize away from certain features not because they make no difference, but because they ought not occur when the process works properly. Think of modelling the best set of chess moves from some given point in a game. The outcome of any real game will surely be affected if Player 2 cheats, or if Player 1 is fall-down drunk – and these kinds of facts will therefore be relevant to our chess model in a way that longitude and car colour were not relevant for our traffic circle model. But they will be irrelevant in a different sense: if one plays as they should – that is, thoughtfully, carefully, and according to the rules – then there should be no reason to consider drunkenness or cheating in our model.

The question for Mills is whether we ought to engage in the second kind of idealization, and, if so, what role those idealized models ought to play in our theorizing. And both thinking about injustice conversations in terms of model-building, and thinking about the place of these two types of idealizations, will also be helpful for parents deciding which models to offer their children as starting points. Which extraneous details can they leave out in order to make a complex world manageable to model? And to what extent should their models build in or idealize away from the failures of real people to play by the rules and the failures of real structures to deliver justice?

But parents’ particular project of talking to their children about injustice will also require them to ask further questions about idealization that political philosophers need not ask. In other words, parents’ project requires methodological tools beyond those that have currently been developed by political philosophers. I’ll call these questions of “educational model idealization”.

To see why this additional framework is needed, let’s consider the difference in purpose between the political philosopher and the parent. When professional philosophers use models of justice, they tend to use them to solve a problem that follows from and assumes knowledge of a vast and highly sophisticated body of knowledge. With that context in mind, philosophers’ models are generally intended to lay out either a procedure or a conceptual map that is novel and (hopefully!) superior to the established alternatives. This is what I call a model-as-solution: it aims to be a novel solution to a problem, that perfects or presses the limits of human knowledge in the area. When professional philosophers consider whether to engage in either descriptive model idealization or ideal model idealization, they are deciding whether to do so while developing a model-as-solution.

Parents deciding how to talk about injustice – especially with younger children – are doing something very different when they decide whether to leave out or abstract away from features of the world. They are aiming to provide models that help their children to simply understand the world, in ways that need not be novel and that often should not be complex. This purpose will be familiar to most professional philosophers in their capacity as teachers. When a biology professor provides students with a simple 4×4 Mendelian square to explain the role of dominant and recessive alleles in determining eye colour, they are presenting a solution that leaves much out. Since so much is left out, the professor offering it knows it won’t work in practice to make sense of our reality. But they offer it in order to help their students to make sense of an initial set of ideas that they can use later to make sense of more complex related ones – so that, if they progress far enough, they might even be able to propose novel models-as-solutions that do stand up to the task of allowing us to accurately explain, effectively manipulate, or provide appropriate prescriptive recommendations for the real world. For the moment, however, they are consciously offering models that aim to succeed as educational tools but that they recognize fail as solutions to the problem at hand.

It is this kind of model-as-educational-tool that parents are aiming to develop for their children, and the type of idealization they ought to engage in will be determined by this purpose. But parents must consider three additional unique features of talking to their children about justice, that academics offering models-as-solutions need not consider, and that professors offering models-as-educational-tools also do not encounter (or encounter to different or lesser degrees). First, young children lack relevant knowledge, both of the highly sophisticated and technical literature on justice which political philosophers assume and build on, and of most of the empirical and sociological facts about the world that are relevant to discussions of justice. To the extent that this kind of information is relevant to the models parents might offer, then, it will need to be either abstracted away from or introduced. But second, children also lack the fully developed brains that allow them to manipulate concepts as complex as theories of justice, and this will require parents to consider in their model development not just what children do not yet know, but also what they cannot yet grasp. Sophisticated conceptual and empirical information can be learned, but given the nature of brain development, that acquisition will of necessity proceed slowly. The models of the world that parents offer children for consideration must therefore take seriously where their particular children are currently in those processes. (I’ll address this question in greater detail below.)

Consider, now, a third difference between political philosophers developing justice-models-as-solutions and parents developing justice-models-as-educational-tools. Once again, parents stand in a similar kind of intermediary role to their children that professors stand in to their students – and that is absent in the context of the political philosopher developing a model-as-solution. Political philosophers in search of solutions are in charge of the course of their own research. They may pursue at their own discretion whatever paths interest them most, and at the end of the day they are free to waste their own time. Professors developing models-as-educational-tools for their students do not have the same broad discretion. Instead, by virtue of the relationship they stand in to their students, they have a moral responsibility to choose the models that they judge to be most important for their students to know, and to present them alongside other models and theories in a way that allows them to present a coherent body of knowledge offered for a justifiable purpose. Similarly, parents offering models-as-educational-tools to their children will be constrained by the responsibilities they have by virtue of the relationship in which they stand to their children. (I’ll return to these constraints below, including the special constraints that this relationship yields when the subject matter is justice.)

So parents talking to their children about justice will need to think – as political philosophers do – about what they take the content of justice to be, and will accordingly need to ask themselves which features of the world are irrelevant to justice, and whether to focus on modelling compliance with justice or deviations from it. In other words, they will need to ask themselves questions about descriptive and ideal model idealization. But they will also need to ask themselves further questions about the constraints introduced by their special relationships to their children and their children’s limited but developing capacities.

Note that the further constraints on idealization that I discuss here are not specific to talking to children about injustice. Instead, they will be relevant to many questions about the education of children (and indeed, to questions about education more generally). Nonetheless, developing them in our discussion of injustice conversations is crucial for two reasons. First, parents who try to use the methodological questions in the non-ideal theory debate to guide their injustice conversations will fail if they do not also bear in mind the idealization questions involved in developing models as educational tools. So questions about models-as-educational-tools are crucial for our purposes even if they are not unique to them. But second, the questions about models-as-educational-tools will sometimes require different answers when asked in the context of injustice than they will in other educational contexts. As we’ll see below, the kinds of “caricatures and white lies” that might be appropriate in other educational contexts will often need to be avoided when questions of justice are at stake.

Let’s turn now to seeing how holding this specific purpose and these specific questions in mind can help to more straightforwardly and productively answer some of the questions asked in the traditional ideal/non-ideal theory debate.

2. Must We Begin with an Ideal of Perfect Justice?

Perhaps the central question in the traditional ideal/non-ideal theory debate has concerned the extent to which we require ideal theory as a prerequisite for doing non-ideal theory. It was famously Rawls’s position that we must begin with ideal theory, claiming that it played a crucial orienting role for non-ideal theory. His claim is not, to be clear, that ideal theory is more important than non-ideal theory. To the contrary, he holds that the questions of non-ideal theory are ultimately “the most urgent and pressing questions”. We must begin with ideal theory, on his account, because without it, non-ideal theory would have nothing at which to aim (Rawls Reference Rawls1999: 8).

Rawls didn’t develop this argument for the orienting role of ideal theory in great depth, but John Simmons has taken up where he left off. For Simmons, ideal theory doesn’t just play an orienting role for non-ideal theory, telling us what non-ideal theory should ultimately be aiming for. It also plays an evaluative role, letting us determine which proposals in non-ideal theory are defensible and which are not. In his words, “[a] good policy in nonideal theory is good only as transitionally just – that is, only as a morally permissible part of a feasible overall program to achieve perfect justice, as a policy that puts us in an improved position to reach that ultimate goal” (Simmons Reference Simmons2010: 22). So for him, it is not enough that a policy aims at perfect justice – it must rather be a coherent part of an overall scheme that is both feasible and that moves us towards perfect justice. And this, he claims, requires a robustly developed account of ideal justice.

Consider an orienteering example. If you are in North America, and you aim to reach the tallest mountain around, it is not enough to see Denali and orient yourself towards it. Rather, to develop a defensible course towards your goal, you need to take into account many other things: Are there sheer cliffs or large canyons between you and the mountain that you would do better to navigate around? Is there a body of water between you and the mountain that would make it worth the resources required to bring a boat? Are there delicate ecosystems nearby that you would harm by passing through? All of these questions and more will be relevant to your journey, and failing to consider any of them and simply making for the mountain could leave you choosing a route that is imprudent or even impermissible.

Simmons takes this to be an argument for the primacy of ideal theory in political philosophy. And as a matter of the role of political philosophers I’ll leave open for the sake of argument the possibility that he is right. But let’s turn instead to how we should resolve this debate in the context of models-as-educational-tools. Does this argument transfer, and should we also focus first or primarily on perfect justice when we’re talking to our children?

Note two things about Simmons’ argument, and the orienteering example that goes with it. First, determining the ultimate aim in the mountain example is far easier than determining the ultimate aim when it comes to justice. In the mountain case, it’s been decided in advance that the tallest mountain is our goal, and Denali unambiguously satisfies that goal. All that is left to determine is how to get there. But “justice” is a much more convoluted metric than height. Justice is an essentially contested subject: we don’t all agree on what justice requires, and well-developed and defensible theories of justice are a dime a dozen.Footnote ⁷ Asking ourselves to aim at what perfect justice requires isn’t like asking us to aim at the tallest mountain in North America – it’s like asking us to aim for the most important cultural site. Perhaps no one will end up at the Burger King in Bimidji, MN, but we’ll end up in an overwhelmingly high number of different places even if we all make the judgements about our routes impeccably.

The second thing to note is how difficult it is to make judgements about our route even if we have decided on a destination. Even if we limit ourselves to the three questions asked above (and there would surely be many more in the case of determining routes towards justice), consider how much knowledge we require to reach our mountain: we need to know the layout of the terrain in a way that lets us determine the follow-on effects of any navigational choice we make; we need to know what kinds of costly resources might be useful on each separate route; we need to know precisely how useful they might be, and how to accurately make trade-offs between the usefulness of those resources and the costs of attaining them; we need to know what counts as a fragile ecosystem, what kinds of interactions would disrupt each one, where those ecosystems lie, and how morally important it is to maintain them. We need, in other words, a staggering amount of knowledge in a staggering number of domains even if we do agree on our destination.

As I said, I’ll leave open for the sake of argument whether political philosophers must both resolve the essentially contested nature of justice and master the full set of normative, technical, and empirical expertise required before they can defensibly make proposals for reforming our shared social and political world. But it would be ludicrous to suggest that we needed to do the same before helping our children figure out how to understand and act in our world. Perhaps the aim of philosophy is ultimate truth, and we fail ourselves when we give up before we achieve it. But parents talking to their children about justice are helping their children to figure out how to live in the world – how they should want to remake their world, and what their obligations in bringing that world about are. So parents will, of necessity, need to adopt a course of action that lets them say something substantive now, rather than waiting until all of the difficult and contested questions about justice have been resolved. They will need to use and introduce values, of course, as no proponent of beginning with non-ideal theory denies. But in using those values, they will do better to help their children to make more local judgements about intuitively obvious cases of injustice, and about the comparative goodness of possible alternative courses of action.Footnote ⁸

Having said that, while Simmons’ requirement shouldn’t count as a strict constraint for our purposes, it can still serve an orienting role. We don’t want what we say to our children to be either piecemeal or – even worse – conflicting, and so we should help them to think about each judgement they make about justice in light of the other judgements they’ve made. We should ask them to treat like cases alike, and to regularly step back and reconsider both where they want to go and whether the judgements they’ve made about the world and what to do in it are going to be good ways of getting us there.Footnote ⁹ What we cannot do is ask them to develop a complete picture of ideal justice before they make moral judgements about our world and how to act in it. If scores of professional political philosophers over the course of several generations of careers haven’t succeeded, we shouldn’t ask a child to try.

Note that the discussion in this section, unlike the last, does not seem unique to the way in which we should talk to our children. When I discussed models as educational tools above, I considered things like children’s level of cognitive development and parents’ obligations of care to their vulnerable young children – each of which places explicit constraints on how our conversations with our children should go in the moment. Parents’ duties of care become much weaker as their children age, and the cognitive capacities of a seven-year-old are very different from those of a teenager. The particular models-as-educational-tools that a parent offers her child will need to be tailored to that child’s age, and the concept of models-as-educational-tools is therefore directly relevant to how to talk to our children about injustice.

The lessons we take from our discussions of Simmons, on the other hand, seem as relevant to how adults should think about injustice as they do to discussions of injustice with children. Indeed, these lessons might seem even more relevant to adults, since adults will have both greater capacities and greater opportunities for acting to bring about justice in the world. In some sense this is correct. But as I said in the Introduction, a large part of parenting is preparing children for the adulthoods that await them, and in this way the lessons we take from Simmons are relevant to parents talking about injustice. Furthermore, as I’ll go on to discuss below, there are some important reasons for parents to bear these lessons in mind even in their earliest conversations with their children. To best appreciate those reasons, however, we should turn to the other ways in which the non-ideal theory debate and our purpose of talking to children can inform each other.

3. Should We Fear Local Optima?

The conclusions of the last section leave us with a worry that we must address: namely, won’t this method of identifying and moving towards comparative improvements in justice risk leaving us stuck at local optima? The concern here is about the effectiveness of making a string of comparative judgements in an uneven or “rugged” terrain. Again, the mountain metaphor is useful. Imagine that I take off from my home, making for Denali, but using the decision procedure that at every juncture, I will take the path that leads uphill. After all, my aim is to end up at the highest point on the continent, so why shouldn’t I simply always aim to get higher? The answer is obvious: there are smaller hills and mountains near me, and once I end up at the top of one of them, I will no longer be able to use my method to progress. Every step I could take will take me lower, but without backtracking and taking a lower route, I will never make it to Denali.

A similar worry holds for a piecemeal comparative methodology in pursuing justice: by comparing immediate options in terms of how well they approximate justice, I may pursue the best of the local bunch – but I might therefore take myself further away from the ultimately just solution (Gaus and Hankins Reference Gaus, Hankins, Vallier and Webber2017). Consider the fight for same-sex marriage in the United States. Let’s grant for the sake of argument that a world in which same-sex couples can marry is more just than one in which they cannot. In that sense, the successful fight to legalize marriage in many states, and ultimately to overturn the Defense of Marriage Act, was a step towards justice. But some critics argue that this fight ultimately set back queer rights, since it relied upon emphasizing the ways in which some same-sex couples successfully lived up to the ideals attached to heterosexual relationships. While the fight to overturn the Defense of Marriage Act yielded the right for same-sex couples to marry, it did so at the cost of further delegitimizing many of the things that are distinctive of queer cultures (Rimmerman Reference Rimmerman2015).

This is a serious worry about the comparative approach to justice, and parents who are going to pursue it as they develop models of justice and injustice for their children must have a reply. Parents and philosophers alike should, I think, grant that this possibility is real: using the comparative approach may lead us to locally better solutions that take us further from ultimate justice – and indeed, they might even lead us to locally better solutions that make the achievement of perfect justice impossible. Think of a society that practices chattel slavery, but is at the brink of outlawing it. Imagine that what perfect justice requires is an integrated society of equals in which one’s race has no functional influence on one’s life chances. Imagine also that the society in question is deciding between a Jim Crow system in which slavery is outlawed but rights and life chances remain fundamentally unequal, and a Bantustan system in which formerly enslaved persons are relegated to local autonomous territories that they legally control. It seems at least plausible both that the Bantustan system is more just than the Jim Crow system, and that the Bantustan system – by creating separate legal systems and physical and territorial separations between the two populations – makes an integrated society of equals ultimately harder to achieve. Which of these scenarios we end up in is deeply consequential from the viewpoint of justice, and parents helping their children to understand which path they should fight for will need to be able to address this kind of worry.

Once again, however, parents must remember that they are not raising political philosophers aiming to capture the ultimate truth about justice. Instead, they are raising moral agents and members of political communities who have obligations to their fellow citizens and human beings. And remembering this difference in purpose should help us to respond to the local optima concern.

Let’s give ourselves a framework to help to clarify the problem (Figure 1). We are at point A, and are choosing where to go. From our local position we can aim for either point A1 or point A2, each of which is more just than our current point. We do not know it yet, but if we move in the direction of A1, we will be moving towards point O, which is both a local and global optimum. If we move in the direction of A2, we will be moving towards S, which is a local but not global optimum. According to the worry at hand, we had better not move towards either A1 or A2 until we know where each will ultimately take us, because making a move without that knowledge risks moving us towards a local optimum and away from the global optimum.

Figure 1. Two paths away from injustice.

While I grant that partially-informed movement has this drawback, I’m going to argue that it is nevertheless justified for our purposes, for two reasons. First, our children have obligations to their fellow humans, who actually suffer from injustice in the moment. As we have seen, what justice ultimately requires is an extensively if not essentially contested question, and even if we could reach agreement on its requirements, comprehensively tracking the possible routes between it and our own world would be a herculean empirical, moral and theoretical task. That task may well never be achieved, and if they want to fulfil their duties to their fellow human beings, our children cannot simply sit back and wait for the full set of relevant questions to be settled. Instead, they must work in the course of their lives to make changes in the political community they actually find themselves in, for the sake of the political fellows they actually have.Footnote ¹⁰ And as long as A1, A2, O and S are all more just than our current world, they cannot justify it to unjustly harmed members of their actual political community to hold off on making a move until we have the full picture in place. Even if we get stuck in S and can never reach O, at least we are no longer in A.

Second, even if reaching a local optimum ends up taking us further away from perfect justice, reaching it may not preclude reaching perfect justice. If we reach a local optimum and then realize that it is local rather than global, we may be able to backtrack in the direction of justice, or forge a new path that will take us there. And if we continue with our orienteering example, reaching a local optimum may well help us to see more clearly where we are. When hiking, finding higher ground can give you a more comprehensive view of where you are and where you might go, even if the point you have reached ends up not being your final destination. And there is reason to think that the same will be true of our political movements towards justice. We human beings are not especially good at predicting the full range of effects that our actions will have, and if standpoint epistemology has taught us anything, it is that lived experience is a uniquely valuable source of knowledge about the moral features of our world (Fricker Reference Fricker2007). It’s very plausible that it will only be by finding ourselves actually in a political community more just than the one we have now that we can effectively theorize the finer points of the injustices that remain.

So for both of these reasons, we should not hold off on offering our children models of a more just world until we can determine what the model of a perfectly just world would look like. A range of more just alternatives will help our children to satisfy their duties to their fellows by working towards something better than we have now – and if we can achieve one of those alternatives we may well be in a better position to see more clearly what justice requires going forward.

Note that again these lessons might seem more relevant to adults than children, since the political and social obligations of children (as well as their capacity to make and learn from gradual social change) are always more limited than those of adults, and at younger ages are plausibly non-existent.Footnote ¹¹ Throughout the next section I will address why these considerations nevertheless bear directly on injustice conversations between parents and children.

4. What Kinds of Constraints Should We Build into Our Models?

Turn now to a separate way of understanding the distinction between ideal and non-ideal theory, which takes the crucial question to be the extent to which constraints are built into our theory.Footnote ¹² Here we understand all political theory as occurring on a spectrum. At one end of the spectrum, we build absolutely no constraints into the requirements of justice, counting as irrelevant all concessions to reality. On the other end, we build in every constraint that might stand in the way of justice, requiring our theories to give full weight to and account for each obstacle. On this way of understanding the problem, then, the problem with both extreme ideal and extreme non-ideal theory should be clear. In Collin Farrelly’s words,

Indeed, even in an ideal political world the most extreme ideal theory might not be particularly helpful, if we assume that some psychological facts (say about motivation or at least limited self-interest) will always be true of human beings. And clearly a theory on the extreme non-ideal end will not be helpful either if it takes people’s unwillingness to be motivated by or comply with the demands of justice exactly as it is and says that the requirements of justice must accept and allow for them.

All really interesting and useful political theory, then, will happen somewhere in the middle: it will take some facts (about, say, human psychology or moderate scarcity of resources) as hard constraints, but recognize some others (say, certain inequalities in the status quo, or some people’s unwillingness to do what justice requires) as features that our theory of justice should help us to make up for rather than simply accept.

A full accounting of which substantive constraints the models we give our children should take as given and which they should work to reform is clearly beyond the scope of this paper. As it would for any political philosopher aiming to develop a model of justice, the project of cataloguing and justifying the inclusion and exclusion of the full range of relevant constraints on justice would be a huge one. But for our purposes, it is worth addressing one significant constraint that must be built into our models-as-educational-tools that need not be built into the political philosopher’s model-as-solution. And this is the constraint of the cognitive developmental capabilities of our children at the point at which we offer them models of injustice. Note that while most questions about appropriate constraints in the current literature focus on ideal model idealization and on models-as-solutions, our question concerns models-as-educational tools, and will involve descriptive model idealization as much as ideal model idealization. Both questions, however, involve thinking carefully about the place of constraints in our models, and using the traditional debate as a starting place will therefore be helpful in addressing our related problem.

Thinking well about this set of constraints will require careful interdisciplinary work done with developmental psychologists, but at this initial point I’ll offer two points to take away. First, we must recognize the obvious but important point that the question about model selection for our children is not one asked at a single time, but one that must be asked over and over as our children grow. Clearly, the cognitive capabilities of a 4-year-old are very different from the cognitive capabilities of a 16-year-old, and the models we offer them must be attentive to this fact. In figuring out how many descriptive features of the world to idealize away from, we need to take into account what level of complexity and interplay between variables our children can developmentally handle. The descriptive models we offer our older children will be able to incorporate a level of subtlety and nuance that has no place in the models we offer to our youngest children.

This extremely obvious point is worth noting not only for its basic importance, but also because it helps us to think about a second equally important but much less obvious point. This is that, in offering our children models of injustice, we must be attentive to the ways in which the earlier and later models will interact. What we are able to offer our older children will depend on what tools and models they already acquired at younger ages, and we as parents must be alive to the ways in which the early models we offer both facilitate and constrain the later models we can provide.

The clearest take-away here is the way in which early models provide a knowledge base that can be assumed and built upon by later models. There is a clear parallel here between discussions of injustice and other areas of progressive skill acquisition, in that material that has not yet been introduced cannot be assumed or built on. For instance, learning algebra will be impossible without first having learned multiplication, and learning to conjugate verbs in a foreign language will be impossible without first having learned relevant vocabulary. Similarly, parents will need to provide earlier models of injustice that incorporate concepts that they will later want to draw on.

But what goes into our early models can also actively constrain what we can effectively offer in our later models. Appreciating this point fully will require much more engagement with the literature in developmental psychology and learning theory than this paper can accommodate, but let’s consider here just one way in which our earlier choices constrain the effectiveness of our later choices: confirmation bias. Confirmation bias is the phenomenon according to which we seek out and are more inclined to believe evidence that supports beliefs we already have, and are resistant to recognizing or believing evidence that contradicts our current beliefs (Nickerson Reference Nickerson1998). It is a widespread and well-established psychological tendency, and it matters for our current purposes because it shows the importance of the content of our early models.

With young children, we will have to make quite a few descriptive model idealizations, since their ability to understand complex theories will be so limited. And this might tempt us towards ideal model idealization as well, since models of ideal justice are often more straightforward than models of actual, messy injustice. For instance, it is easier to explain that everyone ought to have an equal chance at a good life, and that where people end up should be a function of their preferences and desire to work hard, than to explain the complex and interacting ways in which gender, race, class, citizenship, etc., in fact affect people’s life chances. But if we begin with this kind of model, intending to offer the more complex but accurate model later when our children are better able to understand complexity, confirmation bias means that we risk inclining our children to develop a colour-blind, meritocratic comprehensive world view. Parents are not the only source of children’s beliefs about justice and injustice, and children will encounter many, many sources of information that take this normative world view – that is, that the world should be a meritocracy and that skin colour should not matter – and offer it as a descriptive one – that is that the world is an effective meritocracy, and that race, gender, etc. are irrelevant to life outcomes. If the normative view is all that children have been offered by their parents as a way of conceptualizing justice, then confirmation bias means that they will be inclined to seek out and accept evidence that fits with this worldview, unfortunately including the evidence that the view is descriptively as well as normatively accurate.

This is not, of course, a comprehensive account of the ways in which children’s capacity should encourage us to build constraints into the models we offer them. But this example should illustrate for us two important things. First, it highlights the ways in which models-as-educational-tools must be different from models-as-solutions. Models-as-solutions can aim to build in the appropriate level of complexity from the jump, since they are meant for those whose cognitive capacities are as highly developed as they will ever be and who already possess a firm grasp of the relevant information. Models-as-educational-tools cannot; instead, they must be developed with careful attention to the ways in which capacity development and knowledge acquisition proceed over time.

But second, we also start to see here how beginning with a specific project – for us, talking to children about injustice – allows methodological discussions to start to offer more substantive advice. Considering confirmation bias isn’t on its own enough to tell us what precisely to tell our children about sexism or racism, but it does push us to consider the kinds of messages they are likely to get in the world, and so develop models that explicitly reject those messages, thereby hoping to nullify confirmation bias that counts in favour of them. In this way, models-as-educational-tools are different when they model injustice than when they model other sorts of complex ideas acquired over time. As Adrian Currie and Kirsten Walsh argue, the use of what they call “caricatures and white lies” often plays an extremely useful role in early stages of education for facilitating the future acquisition of more complicated ideas (Currie and Walsh Reference Currie and Walsh2015). In domains like theirs – that is, education about early modern philosophy – this may well be true. But in contexts like the discussion of injustice, where recognition of injustice can be painful, and where we have strong incentives to believe that we live in a just and justified world (Jost and Hunyady Reference Jost and Hunyady2002), we should be careful not to begin with white lies or caricatures that are easy to understand but that justify the status quo. If we do, confirmation bias and the attraction of a justified status quo that requires no work from us will both make it harder for our children to develop a better and more accurate understanding of injustice later on. For a variety of reasons the substantive models that different parents offer to their children will not and should not all be the same, but this discussion should remind us that each parent involved in talking to their children about injustice is ultimately involved in a substantive and not merely methodological project.

At this point it should be much clearer why the considerations from the previous sections are directly relevant to injustice conversations with children. Although they concerned the ways in which adults should ultimately think about injustice, the discussion in this section puts us in a better position to appreciate the role that childhood conversations about injustice play in our adult beliefs and judgements. Given that the broader goal of this paper is to show how the non-ideal theory debate and the purpose of talking to children about injustice can inform each other, the argument in this section is suggestive rather than complete. However, I have developed the account in much greater detail elsewhere (Terlazzo manuscript b), and insofar as it succeeds, we have a strong case for thinking that the range of questions in the non-ideal theory debate bear directly on our conversations with children.

5. What Kind of Action Guidance Should Our Models Aim to Provide?

Yet another way of thinking about the distinction between ideal and non-ideal theory is to focus on whether and what sort of action guidance it offers.Footnote ¹³ Zofia Stemplowska invites us to think about non-ideal theory as theory that offers viable action guidance, while allowing the possibility that theories of justice might have functions beyond action guidance (such as, say, evaluation of circumstances) that make ideal theories a complement to non-ideal theories. Since one of the most fundamental purposes of talking to our children about injustice is to help them to figure out what to do in the world and how to respond to the injustice that they face, thinking about what makes for viable action guidance will also be a useful framing for our conversations – and once again, doing so with our purpose firmly in mind can help us to develop more substantive recommendations for the models that we offer to our children.

Her definition helpfully highlights several things that make it appropriate for helping to develop models for our children. The first is that in order to be action-guiding the recommendations required need to be at least in large part locally focused – they need to be focused on problems that are actually being faced (or likely to be faced), and they need to direct courses of action rather than merely providing alternatives to current circumstances. The second is that the recommendations need to be both achievable and desirable only as far as we can judge. This allows us to accommodate the concerns about incomplete theories of justice and local optima raised above. Building both of these features into the requirements for viable action guidance recognizes the central way in which the models of justice we offer our children are meant to facilitate their action in the service of justice in the epistemically limited position and the unjust world in which they find themselves.

What I want to briefly develop further is what it is for action guidance to be desirable in our context. And I’d like to develop it at two different levels: first, in terms of the content of the action guidance offered by the models-as-educational-tools we provide to our children; and second, at the level of the action guidance that parents need when developing those models.

The first of these levels is the most straightforward. The models we offer our children are supposed to help them to think through what the unjust features of the world are, what it might look like for those features to be different in both immediate and somewhat longer-term ways, and what kinds of interventions might lead to those kinds of changes. For the models to offer defensible action guidance, then, they need to helpfully pinpoint the most urgent and actionable forms of injustice, and to provide mechanisms for interventions that can work effectively. Thinking about developing models with these focal points is a general helpful exercise, and there are many types of substantive action guidance that we could draw out from developing responses to different actionable and urgent injustices. For now, let me develop just one.

Parents will have a choice about the scope of the models they offer to their children. They could offer, for instance, models that deal with one specific type of injustice, allowing a comprehensive treatment of that injustice and the interventions it might invite.Footnote ¹⁴ Alternately, they could offer an overarching model that aims to identify structural similarities between different types of injustice, or a comprehensive model that aims to incorporate the ways in which different kinds of injustice interact. There are benefits and drawbacks to offering each sort of model, but one of the things that parents should attend to when deciding is the type of action guidance that each model will offer children, and whether that action guidance will do justice to the range of their moral duties.

Thinking about this question of range helps us to notice several significant potential drawbacks of offering models of specific injustices. The first is that offering different models of specific types of injustice can make it hard to compare and prioritize the duties that one might have. A model that identifies specific points of injustice and potentials for intervention likely will not tell us much about the urgency of that intervention relative to the other interventions we might try with regards to other kinds of injustice. Here, something like an overarching structural model might work better, insofar as it offers a common way of thinking about and therefore comparing the harms and mechanisms of a range of kinds of injustice.

A second problem is that looking at injustices individually makes it difficult to see intersectional forms of oppression. That is, this way of modelling injustice can make invisible the ways in which suffering from multiple forms of oppression (as does, say, a Black woman or a disabled gay man) involves a set of qualitatively different harms from those suffered by people who experience only one of those kinds of oppression (Crenshaw Reference Crenshaw1991). Given this, models of justice that attend to injustices individually risk recommending interventions that at best ignore and at worst exacerbate the harms of injustice for those who experience multiple kinds of oppression. And since our children’s moral duties almost certainly rule out exacerbating injustice for more vulnerable parties in order to alleviate it for less vulnerable parties, whichever models we offer to our children must include an awareness of intersectional forms of oppression, and the broader unintended consequences of the interventions possible.

Let’s turn now to the other level at which we might judge the desirability of action guidance. As I said, the models of injustice we offer our children are tools for helping them to eventually fulfil their duties of justice, and the action guidance the models offer them must be desirable in that light. But parents also stand in a moral relationship to their children, and we should therefore also judge the desirability of the models according to the extent that presenting them allows parents to fulfil their own moral duties to their children.

Perhaps the most fundamental moral duty that parents have to their children is to care for and protect them, and we should recognize the harms that can come from introducing our children to the injustice in the world. Many forms of injustice in the world are serious, vast, intractable, and firmly entrenched, and coming to know about them involves a loss of carefree innocence, and the risk of anxiety, depression or even despair (Liu and Lau Reference Liu and Lau2013; Légar-Goodes et al. Reference Légar-Goodes2022). This does not mean that parents should not talk to their children about injustice. Children will have moral duties with regards to injustice, and another of our obligations as parents is to help children to understand and make good on their moral duties. But there are a variety of forms that models of injustice might take, and as parents we can choose the focus that our models have. We might focus on the intractable nature of the problem, or on the kinds of interventions that have been successful in the past. We might focus on the vast number of obstacles to justice, or on the resilience of oppressed people and the coalitions they form with each other and with allies. In order to ensure that we provide models that offer desirable action guidance to our children, we should choose with an eye to injustices that are most important, and the ways in which they are best able to act. But given our own moral duties to our children, we should also choose our models with an eye to helping our children to maintain hope and agency in light of injustice, and to avoid succumbing to the despair that knowledge of the injustice in our world might inspire.Footnote ¹⁵

This discussion of the desirability of action guidance is once again far from complete. But attention to these specific features of desirability should reinforce the points made in the previous discussion. Once again, thinking specifically about our purpose of talking to children about injustice helps us to more concretely and productively answer the questions asked in the broader non-ideal theory debate. And once again, asking the methodological questions with this focus can begin to give us more substantive action guidance in our role as parents – in this case encouraging us to include in our discussions concepts such as intersectionality, resilience and successes in the fight against injustice.

6. Conclusion

This paper in no way constitutes a full programme for talking to children about injustice. It should, however, illustrate the incredible utility of thinking about that project in terms of the non-ideal theory literature. By keeping our purpose firmly in mind, I have shown how we can do three important things: First, we can see when methodological tools that we need are currently lacking in a debate. Second, we can more productively use the extant tools in the debate to answer our methodological questions. And third – at least sometimes – we can begin to see what kinds of substantive recommendations fall out of our use of methodological tools and questions.

As I said, the project of talking to our children about injustice needs more work, and one natural extension of what I’ve done in this paper is to complete that work by investigating more fully the range of methodological tools available, and by developing a more robust set of substantive recommendations for our conversations. But another equally important extension of what I’ve done here is to pursue the same project in other areas of non-ideal justice. While the methodological tools might not be used in the same way, and while the substantive questions and recommendations will almost certainly not be the same, we ought to attend to the ways in which consciously pursuing methodological and substantive projects in parallel can help us to do better philosophy for a non-ideal world.