Cohen on Justice and Equality reading group (3)

by Ingrid Robeyns on February 10, 2009

So we’ve finally arrived at Chapter 3 in Cohen’s Rescuing Justice and Equality. In Chapter 3, ‘The Basic Structure Objection’, he aims to show that principles of distributive justice apply to people’s legally unconstrained choices. Rawls, whose theory of justice is his main target in this book, has famously argued that the primary subject of justice is what he calls ‘the basic structure of society’, being “the way in which the major social institutions distribute fundamental rights and duties and determine the division of advantages from social cooperation” (TJ Rev Ed p. 6). Cohen’s critique on Rawls is that the principles of justice should also apply to choices which are left open by the rules of those institutions.

As with other chapters in this book, this chapter is also more or less a reprint, from his 1997 article ‘Where the Action is: on the site of distributive justice’. So the claims Cohen makes here are not new, and have already been debated. So let me just pick up a few points (hopefully not too idiosyncratic) that either struck me or particularly interested me.

Cohen’s ‘fundamental concern’ is ‘the pattern of benefits and burdens in society: that is neither a structure in which choice occurs nor a set of choices, but the upshot of structure and choices alike’ (p. 126). He believes that ‘there is scope for relevant personal justice and injustice within a structure, and, indeed, that it is not possible to achieve distributive justice by purely structural means’ (127). So far I agree. My own interest in issues of gender justice fuel my concern here, since as feminists have said for decades (and as Cohen reminds us in the opening of this chapter), ‘the personal is political’. “Harry’s recent post”: about gender injustice makes it very vivid that the distribution of burdens and benefits, and the distribution of valuable opportunities, are to a considerable extent affected by how other people, and within households in particular one’s partner, behave. Rather than just focussing on institutions, we should also focus on people’s choices, and thus we also need an egalitarian ethos.

Cohen argues his case by developing what he regards as a fatal ambiguity in Rawls’s specification of the basic structure. Which institutions are part of the basic structure? Do only legally coercive institutions belong to the basic structure, or can noncoercive institutions also be part of it? Is the basic structure just about formal structures or are informal structures also included? Cohen believes that we should have a broader interpretation of the basic structure, since Rawls’s argument for his restricted focus on the basic structure is that “it’s effects are so profound and present from the start” (TJ Rev ed p. 7). Then the family should clearly be part of it (we have had many posts on CT about this issue over the years, most recently “this one”: Yet if we endorse this broader specification of the basic structure of society, then we must include personal choices that are not legally prescribed, since noncoercive choices are the sustances of these noncoercive institutions that would be included in a broader specification of the basic structure.

I am definitely sympathetic to Cohen’s point that the site of justice should not be restricted to the basic structure, but should also include choices and individual actions. I am not entirely sure how much Cohen wants to say about the place of social norms in Rawls’s work — I can only observe that I have been rather confused trying to find out to what extent social norms and their effects are part of Rawls’s work on justice.

Yet Rawlsians have of course responses to Cohen’s type of argument. They argue that there are good reasons to restrict our focus to the basic structure. One possible response is that Rawls aims to develop an account of political justice, hence does not talk about justice within what Rawls calls ‘voluntary associations’ (such as the family) (the word ‘voluntary’ seems rather unfortunate, since for children it certainly is not voluntary, and if a partner gradually moves into a positon of dependency, it also no longer can be regarded as voluntary, since there is no way out. But I’m putting that quible aside now). So Rawls does say, in later work, that his principles of justice would need to be supplemented with accounts of ‘local justice’, which would apply to these ‘voluntary associations’. Rawls holds that the principles for the basic structure may not be the principles we choose for those ‘voluntary associations’, so that we would first have to work out those principles of local justice (I can find the exact references in Rawls’s work tomorrow if anyone wants to have them). If the Rawlsian principles of justice that apply to institutions, and these accounts of local justice, taken together, would be consistent, it remains to be seen how much of Cohen’s critique would still bite.

Second, Rawls’s notion of justice is less ‘pure’ than Cohen’s; for example, Rawls is concerned with constraints such as not having to monitor or judge the pattern that emerges (in his discussion of pure procedural justice). This is a good reason to focus on institutions, since once they are set up, and delivering the most just outcome, we only have to administer the working of the institutions, and not the actions and levels of well-being/advantage of all individuals. Rawls is also concerned with other issues which for Cohen are no longer about justice but about considerations that do not reflect fundamental principles. So a Rawlsian could argue that Cohen may be right as far as his project is concerned –the search for the notion of pure justice, uncontaminated by other values and feasibility considerations–, but that the Rawlsian project is a different project.

There is much more to say about the basic structure objection, but I’ll trow this in the virtual group since I’ve been holding up this discussion already for too long by taking too long to write this post.



Jon Mandle 02.11.09 at 12:54 am

I was struck by Cohen’s recognition of the following point: “To be sure, so Rawls acknowledges, people’s choices can themselves be assessed as just or unjust from a number of points of view. Thus, for example, capriciously appointing candidate A rather than candidate B to a given post might be judged unjust” even if it is permissible according to the rules of a just institution. (p.124) So Cohen recognizes (here at least) that Rawls does not exclude individual attitudes – or the ethos of a society – from evaluation from the point of view of justice. It seems to me that this undercuts an awful lot of Cohen’s critique – or at the very least, the rhetoric that he often uses. It turns out that Rawls’s focus on the basic structure does not preclude us from criticizing the “high fliers” who extract great wealth from their rare and valuable talents. As Ingrid says, for Rawls, the principles for evaluating the basic structure may not be the same as those for evaluating local justice, or individual actions – but there are such principles. Cohen, in contrast, insists that they are the same – if the difference principle applies to the basic structure, it should apply to individual actions, too, on pain of hypocrisy.

Here’s why: “My own fundamental concern is neither the basic structure of society, in any sense, nor people’s individual choices, but the pattern of benefits and burdens in society: that is neither a structure in which choice occurs nor a set of choices, but the upshot of structure and choices alike…. My root belief is that there is injustice in distribution when inequality of goods reflects not such things as differences in the arduousness of different people’s labors, or people’s different preferences and choices with respect to income and leisure, but myriad forms of lucky and unlucky circumstance. Such differences of advantage are a function of the structure and of people’s choices within it, so I am concerned, secondarily, with both of those.” (p.126)

This account starts by identifying the pattern of benefits and burdens that would be just, and then evaluates institutions, individual actions, a society’s ethos, or anything really, in terms of their contribution to or interference with achieving that pattern. But this is exactly what Rawls rejects when he says we must rely on pure procedural justice. It’s pure because there is no way to identify what a just distribution would be except identifying a just procedure and actually carrying it out. “If it is asked in the abstract whether one distribution of a given stock of things to definite individuals with known desires and preferences is better than another, then there is simply no answer to this question.” (TJ, rev.ed. p.76) (In context, I assume “better” means “more just.” And, of course, in some local contexts, we can identify what pattern would be just.) Without such a single, general pattern that serves to evaluate both institutions and individual actions, there is no reason to assume that the same principle is appropriate for both.


Chris Bertram 02.11.09 at 7:09 am

I can’t see that Jon’s point at #1 is right.

First, the quote on p. 124 links to footnote 19. What Rawls is saying is that of course we use the language of justice (and fairness!) in evaluating all kinds of circumstance, and he doesn’t want to say (dogmatically) that such evaluation is out of place. Now that’s definitely an improvement on the attitude of some of his epigones who do, it seems to me, want to engage in such dogmatic restriction. But what Rawls is saying is that such evaluation is outside the purview of a proper _theory of social justice_. And, of course, Cohen disagrees with him about that, because he disagrees that the institutional structure plays the unique and privileged causal role that Rawls attributes to it.

Further, Jon writes:

But this is exactly what Rawls rejects when he says we must rely on pure procedural justice. It’s pure because there is no way to identify what a just distribution would be except identifying a just procedure and actually carrying it out.

Except …. that this is a bit of a fix in Rawls, isn’t it? Because part of the identification of a just set of procedures is based on their statistical propensity to generate a distributive outcome that (inter alia) is to the greatest benefit of the least advantaged. At which point Cohen is going to point out that the formal rules constitutive of institutions are not the only things conspiring to generate outcomes, but that attitudes, choices etc can also play a systematic role and that their playing that role really ought to be part of a decent theory rather than being arbitrarily excluded from it.

So I think that when someone like Annette Lareau points out that Bourdieuan “habitus” does a lot to explain patterns of relative advantage and disadvantage (and their intergenerational reproduction), it is natural to think, “hmm, Rawls left that bit out, didn’t he?”


Chris Bertram 02.11.09 at 7:12 am

btw: when we decided to do an online reading group for Cohen, we were pretty clear about the terms for participation. I’ve just deleted 3 off-topic comments.


David Wright 02.11.09 at 8:24 am

From my intellectual vantage-point, this kind of critique of Rawls goes back to Nozick, if not earlier.

In “Anarchy, State, Utopia,” Nozick outlined a parable that went something like this: Pick your favorite theory of distributive justice and imagine that a society’s outputs are actually distributed in accordance with it. Now imagine that someone in that society is particularly good at playing basketball, and lots of people in that society like to watch well-played basketball, and are happy to each give the really good player $25 for the privlege of seeing him play. Afterward, everyone is happier but distributive justice is no longer satisfied, since resouces are no longer distributued according to the theory.

Nozick’s point was that justice should be judged on the basis of process, not outcome. Choen seems to approach the same Rawlsian fizzure from the other side, arguing that despite having all been made better-off, the people in the parable should all feel bad for having violated the egalitarian ethos of the ruling theory of distributing justice.

Cohen can be commended for his logical commitment. But to my petite bougeois mind, people who don’t shirk back from the practical application of such an idea are frightening, intellectual enablers of the kind of authoritarian utopian projects that so plagued us in the 20th century.


Chris Bertram 02.11.09 at 8:45 am

David, you haven’t read Cohen’s _Rescuing Justice and Equality_ have you? Doing so is a prerequisite of participation on these comment threads.


ingrid robeyns 02.11.09 at 8:47 am

Chris wrote:
“So I think that when someone like Annette Lareau points out that Bourdieuan “habitus” does a lot to explain patterns of relative advantage and disadvantage (and their intergenerational reproduction), it is natural to think, “hmm, Rawls left that bit out, didn’t he?”
Yes, I agree here. I spent some time over the last years trying to see whether Rawls can satisfactorily deal with issues of gender justice. ‘Gender’ is one of those phenomena where one needs to look at processses like the gradual development of a habitus or gendered social norms (these things are closely related, in my view). And I fail to see how and where Rawls has space for them.
Now, as I try to say in the post, a Rawlsian would have an answer to such a critique, since she could say that one should work out the principles of justice that apply to these non-formal-institutional subjects of justice such as social norms. But then this reply can, I believe, only work if these two exercises are ‘additively seperable’, that is, they show no spil-over effects. In other words, the domain of institutions and the domain of non-institutional practices (norms etc.) are two strictly seperate domains; justice requires that principles for each domain are met. That seems an untenable position, since non-institutional practices are pervasive and interact with institutions. So this is why in the end I probably side with Cohen on the question of the site of distributive justice.


Andrew Lister 02.11.09 at 12:47 pm

Regarding Jon’ s comment that “there is no way to identify what a just distribution would be except identifying a just procedure” (i.e. just institutions), Chris said:

“Except …. that this is a bit of a fix in Rawls, isn’t it? Because part of the identification of a just set of procedures is based on their statistical propensity to generate a distributive outcome that (inter alia) is to the greatest benefit of the least advantage”

The problem in a nutshell is that just distributions are ones generated by just institutions, but just institutions are defined in part by the distributions they generate, which seems circular. The problem disappears, however, if we distinguish between distributions across social positions and distributions across individuals. Institutions are just if, among other things, any inequalities between social positions raise the lowest position (as measured by situation of the typical fully cooperating member of that position). But looking at any particular distribution across individuals, one can’t say whether it is just without knowing what institutions generated it.

If it’s true that social and cultural norms reproduced in daily interactions largely independently of laws and institutions are very powerful in determining who will occupy what position, from one generation to the next, then that would look like a threat to equal opportunity. So if there is a prior principle of equal opportunity, we might be required to design our laws and policies to counteract these norms, in some way, even though the norms aren’t primarily generated by institutions. The Rawlsian schema can accommodate this possibility, I think, even if Rawls didn’t recognize the factual premise.


Ingrid Robeyns 02.11.09 at 1:30 pm

Andrew Lister:
“…we might be required to design our laws and policies to counteract these norms, in some way”

But what if laws and policies have only very limited effects on these norms, and the real changes to undo the injust effects of acting on social norms must come from individual and groups and practices that fall outside the legitimate scope of government intervention? In that case both the causes and the sites of rectification are not institutions but rather non-institutional practices, and hence they would not fit in the Rawlsian framework.


Rob 02.11.09 at 2:55 pm

“But what if laws and policies have only very limited effects on these norms, and the real changes to undo the injust effects of acting on social norms must come from individual and groups and practices that fall outside the legitimate scope of government intervention? In that case both the causes and the sites of rectification are not institutions but rather non-institutional practices, and hence they would not fit in the Rawlsian framework.”

The norms and the effects of the norms are two different things. Consider Cohen’s two feminist examples, of families favouring sons over daughters in the provision of higher education and of wives doing more domestic labour where both spouses work outside the home. What it is right to oppose, I think, is not these norms in the abstract, but these norms when they have effects which disrupt attempts to ensure the provision of the various goods Rawlsian principles of justice direct us to provide. If daughters are not unduly disadvantaged by their parents’ favouring of their brothers in the provision of higher education, then I don’t think we should worry about it (in the same way as about, say, the effects of the general outlines of tax law). Some children fall out with their parents and don’t get any support from them for higher education. Of course, this should not seriously impact on their ability to enter and succeed in higher education, but the tendency of children to fall out with their parents is not something political institutions should be attempting to regulate. Similarly, as long as when couples choose to distribute the domestic labour in certain ways, they are only choosing to distribute domestic labour in certain ways, then I don’t see why political institutions should concern themselves with that. The norm is not the problem; it is that its effects spill over into other areas – so that for example jobs get structured around the expectation of a partner who does the majority of the domestic labour. Legislation can prevent that spill-over, but absent the spill-over, it’s not clear that it is properly a concern of social justice.

Another way of putting this is to make precisely the Wilt Chamberlain point. Cohen and his epigones are walking slap-bang into having to confront Nozick’s critique of patterns, without the Rawlsian get-out of pointing out that whatever principles we might want to use to regulate contracts with no tomorrow, they’re not going to be the same as the principles we might want to use to regulate massive coercive systems extended both over time and territory.


Chris Bertram 02.11.09 at 3:14 pm

Hmm, Rob, I may be misunderstanding what you take the Wilt Chamberlain point to be, but if it is the claim that any distribution that results from the free choices of individuals from a just starting point is itself just, then you’ll remember that Cohen devoted most of a book to rejecting that very principle.


Andrew Lister 02.11.09 at 4:11 pm

Thinking about Ingrid #8’s idea that laws might have little effect on norms, and Rob #9’s point that what matters is not the norms themselves but their effect on some other variable of interest, let’s take an example. Suppose that the preferences about childrearing, work, and parenting that Harry described in his earlier post turn out to be very stable, even after the inegalitarian institutions that formerly helped instill those preferences are gone. Suppose also that there is no way to change these preferences by law, policy or institution that would be consistent with respect for basic liberties. Now we face a choice about how to organize childrearing and work. These are basic social functions, upon which we all depend. We can decide to leave them up to the market, but not if this will lead to systematically unequal life chances for that half of the citizenry that is able to bear children. So the question would be whether there are laws and policies that can rectify the inequality that will otherwise occur. Laws and policies can do nothing about culture, we’re assuming, but maybe they can mitigate or counteract the effect culture has on life chances. I suppose it is possible that there would be no policies consistent with basic liberties that could counteract the inequality in life chances, and that therefore our only alternative would be to say that individuals have a duty to participate in the development of an ethos of egalitarian parenting. I guess it depends what exactly our first principle of basic liberties contains.


Chris Bertram 02.11.09 at 4:40 pm

#11 Well that’s one way of looking at things, but it seems to me to collapse the evaluation into prescription in a characteristically Rawlsian way. Isn’t Cohen going to say (wouldn’t he be entitled to say) that, given the facts, and the conflict with other values (liberty) there is nothing we should do to try to make that society more just. But we still shouldn’t say that a society like that is a just society, because of the pervasive and ineradicable inequality it contains.


Rob 02.11.09 at 5:10 pm

Chris # 10, so that’s not the point I’m trying to make. The point I’m trying to make is that given that we have strong intuitions – or at least I do, and given the way Cohen takes a response to Nozick to be absolutely necessary, I assume he does too – about the Wilt Chamberlain case, you’re going to want to avoid endorsing the claim that that intuitive judgment about that case is just flat-out wrong. I think that’s going to be difficult if you don’t grant a special role for the basic structure, since it seems to me that the sensible way to explain away the force of that example is by pointing out that it’s not about a basic structure. I don’t think, either – although I may have misremembered – that Cohen’s arguments in ‘Self-Ownership etc…’ actually commit him to the view that what Nozick says about the Wilt Chamberlain example itself is wrong, although obviously Cohen rightly thinks it’s not generalizable in the way Nozick claims.


Pete 02.11.09 at 5:52 pm

To respond to Ingrid’s concern that there is no place in the Rawlsian system for addressing gender inequality within the family:

In the case of systematic gender discrimination within families, note that we are probably not dealing with the ideal situation involving reasonable and rational citizens. Insofar as they think gender rightly affects economic and political power, these persons are unreasonable: remember that all reasonable persons accept the conception of citizens as free and equal. So, we have the non-ideal case of dealing with unreasonable persons, and unreasonable comprehensive doctrines. Rawls says that in non-ideal cases like this, measures that are unjust in the ideal case can be justified.

What is important is the kind of justification that must be given: basic liberties can be limited only in order to better ensure and expand the basic liberties; fair equality of opportunity can be violated only in order to better establish fair equality of opportunity, and so on. So, for example, while affirmative action is strictly speaking unjust in the ideal case, it can be justified in historical and cultural conditions in which one group is systematically denied fair equality of opportunity, in order to move society towards this very ideal of fair equality of opportunity. Similarly in the gender case. However, as Rob in #9 says, how any particular couple decides amongst themselves to distribute the burden of domestic labor is not within the purview of the principles of basic social justice (this is, I think, the kind of thing that should also fall within Cohen’s “personal prerogative,” and I’m looking forward to his account of it).

I also want to emphasize Rawls’ (and others’, of course) insight that institutions can affect culture. Who we come to be depends greatly upon the background conditions we find ourselves in. While private citizens acting in association with each other managed to get civil rights legislation passed, this legislation has in turn had a profound effect on the overall culture of the United States (though to be sure much more must be done). We can discourage theft by paying a living wage; we can reduce sexual harassment with legal prohibitions, both because they make it known publicly that such behavior is unacceptable, and because it can help victims to feel empowered to fight back against it. I take it that this is the kind of thing Rawls is interested in when he considers the stability of the society well-ordered by a conception of justice. The question is whether citizens will over generations continue to accept and be motivated by the conception of justice given the embodiment of the conception in the basic structure, and this depends on how they and the public culture are affected by this basic structure itself.


Ed 02.12.09 at 11:44 am

(If this interests anyone ) In Cohen’s Political Philosophy Class at Oxford last year he set his paper along with A. J Julius’ “Basic Structure and the Value of Equality”
(Philosophy & Public Affairs – Volume 31, 2003, pp. 321-355) – which I guess shows he thought this was the most sophisticated challenge of his view . I think its fair to say we were all pretty confused with Julius, as its really hard, so if anyone can decipher the main point Julius makes we may understand Cohen better.


AdamH 02.12.09 at 3:47 pm

Something that puzzles me about this Cohen chapter (and the paper on which it is based) is this. Cohen sometimes seems to be making purely ad-hominem points agaist Rawls. But he also sometimes seems to want to show that certain Rawlsian theses are plain false, independently of their consistency with other parts of the Rawlsian view. In particular, it seems he wants to show that we should not restrict the application of principles of (social) justice solely to the coercive institutions of the state (let’s suppose that’s what “the basic structure” means). But I’m not sure how he shows this:

It is true that the main reason Rawls mentions for applying principles of justice solely to coercive institutions is that their effects are “profound and present from the start”. And Cohen is surely right that if that’s the only reason for focussing on the coercive institutions, then they same principles should apply to various non-coercive structures, choices and so on.

But maybe Rawls is wrong about the reasons for focussing on the coercive institutions – maybe there are other reasons for thinking that distinctive principles apply to them. For instance, most obviously, the fact that they are coercive, but presumably there are others (perhaps Julius’s ideas about “framing”.) For Cohen’s conclusion to go through, doesn’t he need to survey, and reject, all possible reasons for thinking that the coercive structure are distinctive? Or at least, all the plausible contenders?

[I haven’t finished the whole book yet, but this seemed like the chapter that raises the issue most fully, so if my concern is addressed elsewhere, I’ll be delighted to hear it, and I apologise.]


Pete 02.13.09 at 2:16 pm

AdamH, I want to pick up on your point about coercion. Cohen says some interesting things about coercion in the first appendix to Chapter 3. Discussing the idea that coercion is central to the question of Political Liberalism, “We might say that justice is what warrants coercive imposition where coercion is necessary for it to be observed. This would be a stronger and more principled reason for putting coercion at the center of the matter…” (p.148). The idea, I believe, is that justice concerns those just states of affairs that require coercion in order to be achieved. The just states of affairs warrant coercion because coercion is necessary.

What is most interesting to me about this comment is that this “stronger and more principled reason” for taking coercion to be important to justice is not available to Rawls due to his emphasis on pure procedural justice, a disagreement between the two that Jon elaborated on above in the first comment. Since Rawls thinks of distributive justice as a matter of pure procedural justice, he does not have recourse to a state of affairs that can be described independently of the institutions through which it is brought about.

Instead, I think the justification of coercion in Rawls’ system must lie in the answer to the question of how free and equal persons can be free together. The right way to evaluate candidate coercive structures is by principles that would be chosen by free and equal persons, in their capacity as free and equal (and not due to their private interests). Then, this structure does not undermine freedom (as some think any coercion at all does), but rather implements conditions of equal freedom.

Comments on this entry are closed.