We hear a great deal these days about the difference between “equality of opportunity” and “equality of outcome.” The former, it is suggested, is an uncontroversial goal. The latter is now commonly attacked, often on the ground that it involves a denial of individual responsibility. Behind this attack is the assumption that much of the inequality that we see around us results from differences in the choices people have made about, for example, how careful they will be, what chances they will take, and how hard they will work. Those who demand greater equality are seen as denying that individuals should be held responsible for these choices.

John Roemer offers a version of egalitarianism that he believes is not vulnerable to this charge. According to this version, opportunities for a certain good are equal when people who are alike in all qualities that are beyond their control have equal chances of achieving that good. There may of course be some disagreement about which factors affecting a person’s success are “beyond his or her control.” Roemer does not argue for a specific answer to this question. But he clearly believes, not implausibly, that equality of opportunity as he understands it would demand far more equality of outcome than we now experience, even if the idea of what is “beyond a person’s control” is understood in a narrow and uncontroversial sense.

Roemer’s very interesting and well-argued paper raises a number of questions. First, it is important to be clear about how his idea of equality of opportunity is related to the notion that commonly goes by that name. Equality of opportunity, as most people understand it, is violated when some candidates for a position are preferred to others on grounds that are not relevant to that position. It is also violated when unfortunate circumstances, such as lack of education, prevent some people from competing fairly on the basis of their relevant abilities. But a system for choosing among applicants to become opera singers, basketball players, or mathematics professors would not normally be said to violate equality of opportunity just because it made success depend on differences in innate ability that are beyond the applicants’ control. It seems that Roemer would object to such a system, however. He writes: “…an equal opportunity policy must equalize outcomes in so far as they are the consequences of causes beyond a person’s control, but allow differential outcomes in so far as they result from autonomous choice.”

So we have here what appear to be two different ideas of equal opportunity. Call the first “relevant competition” and the second “autonomous competition.” Either of these ideas might be expressed in terms of the metaphor of “a level playing field,” because they involve different ideas about when a competition is fair. Roemer appeals to the idea that a competition is unfair when competitors have unequal chances of winning, due to factors beyond their control. “Relevant competition” appeals to the idea that competition is fair when it tests for relevant abilities (as well as for personal qualities such as determination and discipline that are displayed in competitors’ “autonomous choices”).

It might be argued that these two notions are only superficially different, because anyone who believes in relevant competition is actually committed to autonomous competition instead. The argument would be that the reason for objecting to discrimination on the basis of race or sex, or to the unequal starting points that relevant competition rules out, is simply the fact that these introduce inequalities based on factors that are beyond the competitors’ control. But this is not the only reason. The “relevant competition” version of equality of opportunity is more plausibly based on the quite different idea that the social reasons for having a position often entail relevant grounds for selecting individuals for that position. It follows that it is an abuse of power to choose among applicants on some other basis, such as racial prejudice or friendship, and that applicants are treated unfairly when unequal starting points prevent them from competing “on their merits.” I would agree with Roemer that this version of equality of opportunity is not acceptable as a complete conception of justice; for one thing it says nothing about the size of the rewards that can justly be attached to various positions. But it does capture a distinct moral idea which, many would think, must play at least a subordinate role in any acceptable account of justice.

Roemer’s version of equality of opportunity is appealing because it is more egalitarian than this alternative, while at the same time recognizing that inequalities are acceptable when they result from choices for which the affected individuals are responsible. My second point concerns Roemer’s interesting proposal about how to determine individuals’ relative degrees of responsibility for their fate. This proposal is independent of his interpretation of equality of opportunity. Even those who reject his idea of equality as an overall standard of justice may nonetheless believe in specific claims to equal treatment which are sometimes modified by considerations of individual responsibility. They may believe, for example, that people’s prima facie equal claims to health care on the basis of medical need may be modified by considerations of individual responsibility, so that higher premiums could be demanded of those who smoke or engage in other forms of unhealthy behavior.

But when are people responsible for these decisions, and in what degree? Roemer proposes that we answer this question as follows. First we are to list those factors influencing decisions of the kind in question (say a decision whether or not to smoke) that we regard as beyond an agent’s control (factors such as age, sex, race, social class, etc.). People alike in these factors are said to belong to the same “type,” and a person’s degree of responsibility (the amount of self-control and discipline that he or she exercised) is said to be reflected by their position in the distribution within their type: if almost everyone in Andrea’s type smoked more years than she did, then she is less responsible for smoking than George, who smoked more than almost anyone in his type. In the limit, Roemer says, “if all 60 year old steelworkers smoked for thirty years, I would say that the choice of ‘not smoking’ was not accessible to 60 year old steelworkers: as a 60 year old steelworker, one would have had effectively no opportunity except to smoke for thirty years.”

As a general claim about responsibility, this does not seem very plausible. If all 70 year old citizens of Wisconsin with incomes over $150,000 per year consistently vote for candidates who have taken a position favorable to them on Social Security and Medicare, we would not normally conclude, on this basis alone, that they had “effectively no opportunity” but to do so. When factors “beyond their control” give people in a given class strong reasons for acting a certain way, a uniform pattern behavior may result, but these people may still be fully responsible for what they do.

We need to be careful here, however, not to mislead ourselves by failing to distinguish between different contrasts that the phrase “should be held responsible for” can be used to mark. On the one hand, to say that an agent can be held responsible for an action may be to say that he or she is properly subject to praise or blame for performing it. It is this sense that is called to mind by the contrast between results that are due to an agent’s “autonomous choice” and those that are due to factors “beyond his control.” The citizens of Wisconsin in my example are clearly responsible in this sense for their voting behavior, and this makes it reasonable to argue about whether that behavior shows them to be greedy or just reasonably prudent.

On other occasions, when we say that certain people “should be held responsible” for their conduct, what we mean is that it would be justifiable policy to make them bear the consequences of that conduct. This use of the phrase does not mark the same distinction as the previous one. A bank teller who, faced with a credible threat of violence, hands over the drawer of money, is responsible (and may be praised) for acting coolly and avoiding a bloodbath, but should not be “held responsible” for the loss of the money. It is easy to see the distinction between the two senses of “responsibility” in this case, since obviously a person who has acted well should not be made to bear costs that are brought on by someone else’s acting badly. But the same distinction exists with regard to an agent who has not acted well. Bankruptcy law presumably reflects the view that unsuccessful entrepreneurs, even if they have acted unwisely, should nonetheless not be held responsible (i.e. required to pay) for the full consequences of their unwise choices.

More generally, we all act unwisely from time to time — smoking, drinking, eating too much, exercising too little, and so on — and it is an important question how far a humane and justifiable social policy would “hold us responsible” by making us bear the costs of our unwise behavior. Roemer’s proposal is addressed to this question, not to more general issues about “responsibility.” It therefore presupposes a situation of a particular kind, in which people are subject to certain pressures and temptations and the question is: how severe must these pressures be before we should say that the individuals in question should not be made to bear the costs of their unwise choices?

I believe that the coincidence between the statistical frequency of a form of self-restraint and the justifiability of making people bear the costs of not exercising it could not be more than rough and ready. Nonetheless, I think there is something importantly right and salutary about Roemer’s proposal. In some cases in which we are initially inclined to “hold people responsible” for their choices by denying them aid, we may rightfully be deterred by reflecting on the fact that these are choices that most people would have made if placed in similar circumstances.

There is, however, something potentially misleading about the terms in which Roemer’s proposal is put. Given the age at which his 60 year old steelworkers started smoking, the limited foresight one has at that age, and the difficulty of quitting smoking once one has begun, no humane policy would restrict their medical benefits now, or insist now that they pay higher premiums, because they have been smokers. But this is not because the alternative of not smoking was “not accessible to them” or because they “had effectively no opportunity except to smoke.” Both of these claims strike me as false. There is, however, room for saying that even though they are responsible for making certain choices (they are, for example, properly criticized for so doing), we ought to help them nonetheless.

This brings me back to the political debates about individual responsibility with which Roemer and I both began. It is often said that there are two approaches to phenomena such as crime and teenage pregnancy: they can be seen as forms of misconduct for which individuals should be held responsible, or as “social problems” whose causes should be alleviated. Conservatives, so the story goes, take the former line while liberals take the latter, thereby denying individual responsibility. The mistake, I believe, is to see these as alternatives between which we must choose. This mistake is fostered by a failure to distinguish between the two senses of “hold responsible” that I distinguished above. Even if everyone in a certain “type” engages in a given form of unwise or reprehensible conduct, they may still be blamed, or even punished, for doing so. But this is no reason to take finger-wagging or longer prison sentences as the only appropriate response to their conduct. We should also ask how we can help them, and especially how we can change things so that fewer people are put in circumstances where they are likely to make bad choices of this kind.

I do not think that John Roemer would disagree with this. After all, the idea of equality of opportunity calls our attention first of all to the need to improve the conditions in which people’s choices are made. But his emphasis on the dichotomy between those consequences people are responsible for and those they should be protected against pushes him to say, implausibly, that those whom we think should be helped are not responsible for their actions. I worry that this may leave him open to conservative objections of the kind that he very properly sought to avoid.