Welcome to Political Not Metaphysical, a blog devoted to key texts and issues in political philosophy. Below you will find an (almost) comprehensive list of blog posts by category for your browsing pleasure. See a text or an issue that’s not covered? Email me at reasonablepluralism(at)gmail(dot)com, and maybe it will appear on the blog.

Blog List by Category:


John Rawls, A Theory of Justice: Reader’s Guide

As devoted readers know, I am currently in the process of writing a comprehensive reader’s guide to John Rawls’s A Theory of Justice. Part I of the guide is now complete! In order to make the guide easier to use, below are links to all the posts so far. This will be updated as I add more:

Part I: Theory

  1. Prefaces, Ch. 1, §. 1-4: Core Ideas
  2. Ch. 1, §5-9: Moral Theory
  3. Ch. 2, §10-17: Principles of Justice for the Basic Structure
  4. Ch. 2, §18-19: Principles for Individuals
  5. Ch. 3, §20-25: The Original Position: Set Up
  6. Ch. 3, §26-30: The Argument for the Two Principles

Part II: Institutions

  1. Ch. 4, Part II (Intro) + §31: Part II (Structure) + How to Apply the Principles
  2. Ch. 4, §32-35: Equal Liberty of Conscience
  3. Ch. 4, §36-38: Political Justice, Participation, and the Rule of Law
  4. Ch. 4, §39-40: Priority of Liberty Defended + The Kantian Interpretation of JAF
  5. Ch. 5, §41-42: Theory of Justice and Political Economy
  6. Ch 5, §43: Institutions for Distributive Justice
  7. Ch. 5, §44-45: Intergenerational Justice (Justice and Time)
  8. Ch. 5, §46-50: Restatement of the Principles, Comparison with the Precepts of Justice, Desert-Based Theories, Mixed Conceptions, and Perfectionist Views
  9. Ch 6, §51-54: Natural Duty and The Principle of Fairness
  10. Ch. 6, §55-59: Civil Disobedience

Japa Pallikkathayil, “The Possibility of Choice: Three Accounts of the Problem with Coercion”

Coercion is both presumptively wrong and undermines the responsibility of the person coerced. How do these two features of coercion hang together, and what explains why? In this paper, Japa Pallikkathayil offers a rich and original account.

She focuses on cases of “volitional” coercion (hereafter, simply coercion)—where one agent tries to alter another person’s choice—rather than on physical coercion (brute force). Typically, philosophers have focused on trying to define coercion. Pallikkathayil instead tries to explain what’s wrong with the paradigm cases of coercion—for example, the Mugger’s threat, “your money or your life”—rather than on definitional matters. Such cases typically involve “contingency announcements”: speech acts of the form ‘I will do x if and only if you do y”. Only a small subspecies of contingency announcements is morally problematic—the prospective trade, ‘I’ll give you my pretzel’s for your chips’, isn’t, for example—because these announcements unacceptably constrain the victim’s choice. So part of the puzzle is to explain, not only (1) what the moral objection to the mugger’s announcement is and (2) why it diminishes the victim’s responsibility, but also (3) to do so in a way that does not overgeneralize to other cases (the lunchtime offer above, e.g.).

Pallikkathayil considers three accounts:

  1. The impaired action account [IA]
  2. The impaired consent account [IC]
  3. The impaired normative authority account [INA]

She rejects the first two (though the second, sympathetically) and defends the third.

  1. Impaired Action

The impaired action account models volitional coercion on physical coercion. It claims that coercion is wrong because it induces a defect in the action’s relationship to her action or agency. In order for the account to work, “action” must be understood in a rich sense, one connected to autonomy, for clearly the mugger’s victim still does something. But autonomy can be understood in two senses: (1) as a relation between an agent and her physiological states, or (2) as a relation between an agent and other agents. The impaired action account takes the first sense.

To undermine its appeal, Pallikkathayil undercuts the two features of the mugger-victim interaction that seem to support it: (1) the victim’s diminished responsibility and (2) the victim’s experience of the interaction as an unwilling participant.

First, responsibility. The connection between coercive contingency announcements and responsibility seems to motivate IA. But, Pallikkathayil contends, this rests on a conflation between different sense of responsibility. Consider four ways of parsing “is A responsible for Φing in situation S”:

  • May A’s behavior be attributed to the agent (for moral appraisal)?
  • Is the action permissible or impermissible?
  • Is the action praiseworthy or blameworthy, etc.?
  • Should we let the benefits and burdens of the action fall on the agent who performed it?

Coercion can effect (2)-(4) without altering our judgment about (1). The fact that the action is attributable to an agent does not settle how we ought to evaluate it. The agent’s responsibility might still be mitigated. So, IA isn’t needed to explain the agent’s diminished responsibility

Second, unwillingness: coercive interactions are one’s which the agent does not typically engage in willingly. Yet, there are different senses of unwillingness.

  • Coercion places its victims in circumstances that are hostile to their aims.

But, (1) doesn’t undermine an agent’s capacity to act.

  • Coercion shapes the agent’s alternatives such that the agent’s reasons point in one direction.

However, (2) only undermines an agent’s capacity to act if one conflates having meaningful options with acting. You can choose without meaningful options: suppose your only job offer is your dream job. What makes the negative case seem different? Perhaps, Pallikkathayil conjectures, it concerns how one experiences the force of reasons: as if the verdict comes from outside. But, still, in both cases the agent can deliberate and act.

  • The action is the result of deliberative confusion
  • The action is akratic

These sense may affect the status of a behavior as a genuine action. But these aren’t necessary effects of the mugger’s announcement.

  1. Impaired Consent

The impaired consent account suggests that the problem lies in the relation between the victim and the coercer. This seems right. But it suggests that the impairment takes a specific form: the victim’s response does not constitute consent to participate in bringing about the coercer’s end.

To unpack the suggestion, we need to analyze consent. To (validly or successfully) Consent (in the sense relevant here) is to “exercise a normative discretionary power”; that is, to do something which is morally transformative and reshapes the landscape of permissions and obligations.

In the case of coercion, this moral transformation does not occur: the mugger can’t keep the money, e.g. And the impaired consent account claims this is because the recipient is unable to consent to the mugger. But, for the account to work, we need an explanation of why this is so.

The Kantians who advanced this account have paid insufficient attention to this task. Pallikkathayil tries to help them out. The standard story comes from Kant’s formula of humanity on which one must never treat others as a mere means. One way of spelling out the idea of a mere means is through “possible consent”: A treats B as a mere means just in case B couldn’t possibly consent to that treatment.

Christine Korsgaard develops this story via the case of deception. If I ask you for a loan that I never intend to repay, you think my end is different than it is. Since this is so, consent to furthering my true end (taking your money) is allegedly impossible. On this view, consent seems to require choosing an action under a description that makes reference to the other’s true ends. Kantians claim the same problem occurs in the case of coercion: the victim can’t consent to the mugger’s end of obtaining the money. But, why? The victim, after all, gives the money knowingly and intentionally. What more is required for consent?

Korsgaard claims what is needed is “some power over the proceedings”. This might mean something like control over the circumstances of choice. It is true that the victim seems to lack control over how her options have been shaped. But, the suggestion raises three questions:

  • Why is a say in the shaping of one’s options a precondition for consent?
  • What kind of say is required?
  • Does the mugger prevent his victim from having this kind of say?

Pallikkathayil works backward, beginning with three. Pallikkathayil argues that there is only one kind of control which is necessarily denied in the mugger-victim interaction. So, this has to be the relevant sort of control, if the explanation is to work.

But she begins by distinguishing some various kinds of control. There are two broad approaches. On nonmoral approaches, the requisite kind of control is to be determined without any moral evaluation of what the parties are doing—just by looking at how the parties take each other’s plans (whatever they are) into account. This approach is appealing because consent seems to be a basic wrong (a wrong we can identify without having to identify any other wrongs). But, Pallikkathayil argues, the approach fails, and this suggests that coercion isn’t a basic, but rather a parasitic wrong (a wrong that depends on other wrongful aspects of an interaction).

Here’s her argument. She begins by asking what, on the nonmoral approach, power over the proceedings might require. She considers various interpretations.

The first is a weak interpretation, on which all that is needed is to give another adequate control is “a willingness to revise one’s plans in light of the considerations the other raises. This, however, is too weak: if the victim can negotiate (I’ll give you my wallet, but not the pictures), that’s fine, on this view.

So we might prefer a second, strong interpretation. On this view, one’s option can’t be altered by another without one’s agreement. Each has a veto over the interaction. But, if everyone (including the mugger!) has a veto, conflict is simply irresolvable.

A third, bilaterial interpretation has it that “the parties are required to come to a shared view of how the interaction will proceed”. This doesn’t seem much better. The parties could come to a shared interaction by (e.g.) a fair coin toss. So, the view doesn’t explain how the mugger makes consent impossible.

So, we need to consider a moral approach. Pallikkathayil’s thought is this: a contingency announcement constrains the recipient’s options only when (1) the announcement has the power to influence the receipt (2) only if the recipient is denied veto power over the intention announced. She calls this the veto conception of constraint. The first part of the definition is added in order to accommodate the lunchtime case: here, even if I’m set in my intention, you can escape that intention by declining the trade.

This can then explain the mugger case in the following way: if the mugger gave the victim veto power over the intention, the victim would have a costless way of avoiding death. So, the announcement only makes sense if one assumes the victim is not given such power. Of course, the announcement does not guarantee the mugger’s preferred outcome: (1) the victim may not regard the reason as sufficient (say, in the case of ‘a beating or your life’, she’d accept a beating to keep the money, (2) the mugger may misjudge the costs in question.

The veto conception does not make reference to the impermissibility of the intention announced. It holds that an intention’s impermissibility is neither necessary nor sufficient to make an offer constraining. To make the case that it is not necessary, Pallikkathayil considers the case of an employer who says to her employee: I will fire you if and only if you are late for work again. Here, the employers options seem constrained because, if the employed had control over the intention, the mode of influence would be ineffective. The contingency announcement here is permissible, but constraining.

So, in what sense does the mugger impermissibly constrain the victim’s options? To make sense of this, some terminology. A legitimate demand is the counterpart of an obligation. If A has an obligation to treat B in a certain way, then B can legitimately demand that she be treated in that way. In the mugger case, the victim can legitimately demand that the mugger abandon his intention because it is an intention to violate an obligation he has to the recipient. Hence, veto power.

But this won’t work for all cases. Suppose A kidnaps B to extract ransom from C. B has a legitimate demand here, but why C? Pallikkathayil suggests that, by linking C’s action, A confers standing to B. [I’m not sure how this explanation works. See page 12].

Let us say that someone who has standing to legitimately demand that another abandon her intention has a moral veto over that intention. This then suggests an account of impermissibly constrained options:  “a contingency announcement constrains the recipient’s options when the announcement has the potential to influence the recipient only if the recipient is denied a moral veto”. This then furnishes the relevant sense of “power over proceedings” need to make consent impossible. It is a moralized conception of this power because it relies on a moral assessment of the announcer’s intention.

And this proposal makes sense. Consent, after all, supposed to be morally transformative. And its moral powers can only work if one presupposes a background of moral norms that specify the prohibitions which giving consent cancels. This set of prohibitions carves out a ‘sphere for each person’ within which she is solely entitled to determine what happens. This then suggests the following interpretation of the power of proceedings: in order to have this power, one’s option must not be impermissibly constrained.

One objection is this. Suppose the mugger threatens to smash the victim’s windshield if she doesn’t fork over the cash. But, suppose the victim’s car (unbeknownst to the mugger) is already destined for the junkyard, and so the victim is not moved by the offer. Here, it seems like the victim can exercise her discretionary authority despite her impermissibly constrained options. So, one might add the condition that one lacks power over proceedings only if one has been deprived of options that one regards as deliberatively significant.

Pallikkathayil sets aside the issue of which of these two views one should assume and simply states that disjunctive view that: in order to have the power over the proceedings that is needed to make consent possible, a person’s options must not be impermissibly constrained, at least not in ways that are deliberatively significant, and perhaps not at all. (She highlights two features of the view, which I ignore—see p.14).

Now return to the impaired consent account. On this account, the mugger’s action is impermissible because: (1) it impermissibly constrains the victim’s options and (2) this constraint has the further effect of preventing the victim from exercising her power to consent, which she is entitled to exercise. So, it explains the wrong. Moreover, it explains why the victim is not responsible: while the victim’s action is still a proper object of appraisal, her action does not have the morally transformative licensing effect it otherwise would. Her action does not count as consent to participate in the mugger’s plan.

To see how this is relevant, here’s another case: suppose A is a terrorist and B is a subway engineer. A threatens B to give up the subway plan. Even if B acquiesces, she “does not count as working” with B to bomb the subway” (in contrast to a co-conspirator).   There are various complications here which I’ll set aside (see p. 15-16).

So, the impaired consent account is plausible. But, Pallikkathayil thinks, it still has a problem. The main problem is that consent is not the only kind of morally transformative action. There are other normative discretionary powers, such as the power to promise. Promising generates obligations to perform the actions promised. Coercion may disrupt the exercise of these powers as well (e.g., promise me X or I’ll kill you). So, the account needs to be generalized.

  1. The impaired normative authority account [INA]

This brings us to Pallikkathayil’s own preferred view: the impaired normative authority account. Normative discretionary powers allow agents to change the landscape of permissions and obligations simply by choosing to do so. More generally, impermissibly constrained options impede the exercise of other forms of basic normative authority as well.

Thus, INA builds on IC. A certain contingency announcement impermissibly constrains a persons options, and this has the further effect of prohibiting the victim from exercising her normative authority.

One objection to both INA and IC is that coercion seems to be a basic wrong, rather than a parasitic one. Consider this case: an employer (who may otherwise permissibly fire her employee) says: I will fire you unless you have sex with me.  INA suggests this is not wrong in the same way the mugger’s offer is. Pallikkathayil offers some arguments who way the offer might be wrong that strike me as very unsatisfactory. And she concludes that, in absence of any explanation of the kind she canvasses, the announcement does not impermissibly constraint the employee’s options. More broadly, the range of blackmail cases seem to be counterexamples, which Pallikkathayil recognizes and address in a footnote (see p. 19 for her discussion).

This seems like the place to push Pallikkathayil’s account, and I think Pallikkathayil needs to say more here. But, she concludes, that “the temptation to understand coercion as a basic wrong may reflect a failure to consider the many grounds on which it might be impermissible for an agent to act on the intention he is announcing”.

Though I am skeptical about some of the implications of the view, Pallikkathayil’s account is interesting and worth taking seriously, and breaths much needed life into the philosophical discussion of coercion.

David Estlund, Utopophobia

Is it a defect of a normative political theory that the requirements it recommends are not likely to ever be met? In “Utopophobia”, David Estlund argues that it is not. The mistake that those who think it is a defect make is to confuse a standard of justice with an appropriate practical goal.

To begin, Eslund distinguishes different kinds of realism. The crudest—complacent realism—says everything is as it ought to be. Normative theories ought not to be complacent. Some, though, are more or less realistic. There are, then, a range of “Noncomplacent nonutopian” or “aspirational” theories, and these are the ones worth taking seriously.

Some theories are, however, too unrealistic. For example, one might argue that a conception of society that posits standards for people or institutions that are “beyond their abilities ever to live up to” falsely imposes those standards.

Can, Estlund grants for the sake of obvious, does imply ought. This concession, he insists, is not obvious. Cohen, for example, has argued that standards of justice might be “normatively fundamental”—we ought to bring them about if we can. But, he accepts it for his purposes. Estlund’s key point is that ought does not imply reasonably likely.

To bring this out, he draws a distinction between ability and probability. Even if an action has a probability of zero, that does not mean the action is impossible in the relevant sense. It may be within the agent’s ability, but simply unlikely that she will do it. So, the fact that an agent certainly won’t comply with moral standards doesn’t mean she isn’t obligation to.

This, however, raises the question: should normative theory limit itself to setting standards that, not only can possibility be met, but are likely to be met in practice? Estlund thinks not.

First, though, Estlund explores the dangers and prospects of “hopeless aspirational theories” (that is, theories that will certainly not be complied with, although it is within the ability of agents to do so). Working towards such theories can also have positive effects: it may motivate people to do better than they otherwise would. However, such theories can be dangerous if people work towards them: actions towards ends that will never be achieved can be wasteful.

Estlund discusses a particular set of cases in which this danger obtains, concerning institutions and personal duties. Suppose it should be the case that people should be much more impartial and institutions should compensate those who are wronged by unjustified partiality. But, if people do not leave up to their duties, having such an institution may be harmful. But this, Estlund thinks, is not a defect of the theory: the theory may be the truth about justice, but not answer what we ought to do.

One objection here is that the theory of justice is normative and normative theories concern what we ought to do. Estlund has two replies. First, the theory is still evaluative. Second, the theory does council action: it says that society should: (build such institutions and be less impartial). It is not practically idle just because it won’t be done.

A second objection is this: aren’t hopeful theories (that is, theories which aren’t hopeless) better? These theories recommend standards that it is both possible and reasonable to believe people will meet. Estlund thinks not. It would be better if the best theory turned out to be one that people complied with, but it is not a defect of the theory that they won’t.

Many moral theories are alleged to suffer from a different problem: they are too demanding, in that they demand compliance with moral requirements that people cannot be reasonably required to bear. Estlund dubs such theories harsh. Harsh theories, he thinks, are false. But this is a separate issue from improbability: a theory can be improbable without being harsh.

Estlund does think nonhopeless theory is important. We should try to find out what we ought to do given what people and institutions are likely to do. Estlund calls this concessive theory. There is room for both hopeless and concessive theory.

Estlund considers the role of concessive theory in moral philosophy. Consider the case of Professor P: P has been asked to review a paper, but knows that he likely won’t do so even if he accepts. There are two questions here: (1) is he obligated to review the paper? (2) what should he do given that he knows he likely won’t. It is both the case that he ought to accept and review, and that given that he won’t do both, he ought to decline. This is not a contradiction: S ought to do A and B does not entail S ought to do A and S ought to do B. In politics, we ought to build and comply with certain institutions; given that we won’t comply, we ought not build. But, the latter claim does not cancel the obligation to do both.

Estlund then turns to a different difficulty: if may be that P can perform some action, but it is quite another question as to whether Q can get P to do it. Suppose a theory requires a high level of political participation and involvement. It might not require getting anyone to participate. Instead, it requires direct of people that they participate. Another case is mass-scale social compliance: suppose it is morally required not nepotistically hire one’s children. Even if people won’t comply, there isn’t anyone who is required to get them to comply.

Some oughts are essentially collective. Suppose the choir ought to sing the G chord. This does not imply that each member of the group ought to do so, if the others won’t. There is still a question about doing one’s part, and one’s part might not be equivalent to the group action.

Is Estlund’s argument so far a defense of “ideal theory”? Ideal theory, he thinks, can mean different things. Consider two idealizations associated with it: full compliance and complete virtue. Neither is realistic. But ideal theory so construed is not the same as aspirational theory. Consider an aspirational theory which admits that not everyone will comply with what is morally required of them. The theory doesn’t assume full compliance. It says that it is morally required to comply.  It does not falsify the facts. Aspirational theory is not conditional on fictional assumptions: its conclusions apply to society as it is.

The interaction between concessive and aspirational theory is complicated. Suppose that in a fully just society there would be no need for penal institutions. Since we are not in such a society, we need such institutions. So we need to ask the concessive question: what should they be like? But we made need further levels of concessive theory: for example, a theory of how prison guards should behave, given that they are unlikely to be fully just. It might look like we may end up conceding almost everything. But there is a limit. This brings out that it is no good objection to a theory that it is hopeless.

But what is the point of aspirational theory? Suppose some aspirational theory is true. It may have some (oblique) practical value. But even if it does not, the truth may be itself important, because it is a truth about justice.

Holly Lawford-Smith, “Ideal Theory—A Reply to Valentini”

Holly Lawford-Smith, “Ideal Theory—A Reply to Valentini”

Holly Lawford-Smith’s brief article, “Ideal Theory—A Reply to Valentini”, is exactly what it sounds like: a concise reply to Laura Valentini’s “On The Apparent Paradox of Ideal Theory”.

Valentini, as I summarize elsewhere, outlines a paradox consisting of three premises: Any sound theory of justice must be (1) action-guiding and (2) ideal, but (3) any ideal theory fails to be action guiding. Valentini then argues that the paradox can be overcome by disambiguating (3), and showing that some ideal theories can be action-guiding.

Lawford-Smith agrees that the paradox can be overcome, and aims to strengthen Valentini’s argument. She does so in three ways: she argues against premises (1) and (2), and then she adds some additional considerations that bolster (3).

First, the argument against (1), the premise that theory of justice must be action guiding. This requirement must be a requirement that stems from “something distinctive about justice” (against the point that normative theory just means a theory that is action guiding, Lawford-Smith notes (1) that there are other aims of normative theory—e.g. explanation and justification—and (2) notes that the question can be reposed as: why must a theory of justice be normative?).

Some normative claims—‘unowned oughts’ (‘life ought not to be so unfair’) [note: this seems like an evaluative rather than a normative claim per se) and ‘past oughts’ (‘WWII ought not to have happened’)—seem legitimate but not action-guiding. Perhaps they are explained by general principles that are action-guiding? Not always, and not in the case of unowned oughts, argues Lawford-Smith. So, (1) isn’t obvious.

But, more generally, Lawford-Smith argues that normative theories have a different task besides guiding action: namely, helping us to understand “the nature” of things—in this case, the nature of justice. A theory of justice might tell us what justice is, and doing that might be helpful independent of whether it is action guiding. Perhaps one might respond that such theories are “indirectly” action guiding. But this, Lawford-Smith thinks, trivializes the premise.

So, she concludes, either action-guidance means something trivial, and so any theory meets it. Or, it means something more robust, in which case the premise is false.

Second, Lawford-Smith takes on (2). Valentini seems to think that an ideal theory is necessary to provide a practical standard for evaluation as well as to provide direction for how we should push institutional reform. This, Lawford-Smith claims, presupposes a false view about ideals. It assumes there is a “single ideal” and once we know it, we can know how to act. But, Lawford-Smith claims, it is more plausible that there are multiple ideals—ideals of justice, equality, welfare, etc.—that must be traded off. And, following Sen, she claims that we can know the better without knowing the best.

Lawford-Smith then makes a second, more interesting argument: many standard ‘ideal theories’ aren’t ideal at all. She offers a reconstruction of Rawls’s and Dworkin’s theories as counterfactuals. In the Rawls case, the theory goes: if people are in these circumstances and had these interests, they would choose these principles. Rawl’s doesn’t, however, assert that the antecedent of the condition obtains. Instead, he takes it that the counterfactual as a whole to have “some normative force” for what principles should be in place. [Note: she doesn’t argue why it has these implications. It is hard to see why it does. Her reconstruction here seems particularly vulnerable to objections similar to those about the relevance of hypothetic consent.] Thus, she concludes, Rawls’s and Dworkin’s view aren’t (or need not be seen as) ideals. And though this move won’t work for all theories, it proves again that ‘ideal theory’ isn’t necessary.

Lawford-Smith then turns to (3). She argues, contra Valentini, that even theories which idealize “their subject” can provide guidance. We can understand theories in science, like e.g. Galileo’s, as specifying an ideal, whose assumptions we can relax gradually. The same, she argues, can be said of Rawls’s theory of international justice. This doesn’t hold for all theories. Some might presuppose an ideal which can never be reached. But it does hold for many. A theory build on falsehoods might specify a standard which we can gradually relax, and so provide useful guidance.

Thus, Lawford-Smith concludes, ideal theory is in better shape than Valentini supposes.

Laura Valentini, “On the Apparent Paradox of Ideal Theory”

Laura Valentini, “On the Apparent Paradox of Ideal Theory”

Many critics of ideal theory contend that there is a gap between ideal theories and our non-ideal circumstances—a gap which renders ideal theories ill-suited to guide action in the real world. Valentini calls this the guidance critique. But yet, ideal theory seems inescapable. This, Valentini generates a paradox, which can be stated as follows:

  • Any sound theory of justice is action-guiding.
  • Any sound theory of justice is ideal.
  • Any ideal theory fails to be action-guiding.

Valentini’s article first articulates, and then responds to, this paradox. In outlining Valentini’s argument, I’ll focus on her treatment of Rawls (ignoring her interesting discussion of Dworkin and Kymlicka)

She begins by explaining each of the premises.

  • Any sound theory of justice is action-guiding.

A theory of justice is a normative theory. And normative theory aims at guiding action. This seems obvious, but some question this claim. For example, G.A. Cohen argues that justice is a fundamental moral value which is wholly insensitive too social facts. Fact insensitive theories ask: What is justice as such, in its pure, fact-free form? As such, these theories can avoid the paradox by rejecting (A).

But most theories don’t reject (A): Rawls, for one, doesn’t. These fact sensitive theories ask: What principles of justice should govern the exercise of political power for it to be justified? These theorists intend for their answers to that question to be helpful in guiding action in non-ideal circumstances.

  • Any sound theory of justice is ideal.

Ideal theory can mean at least two things. First, it might mean a theory which aims at “an ideal of a fully just world towards which we should aim. Second, it might mean a theory that involves idealized (i.e., false) assumptions. For Valentini (for purposes of this article) an ideal theory is a theory which meets both these conditions. Rawls’s is a canonical example. Rawls’s theory involves false assumptions—full compliance, for example—and we are supposed to strive to bring about a society in which its principles of justice are operative. It is hard to see how a theory of justice could avoid being ideal: a theory of justice, after all, needs to abstract away from imperfections and injustice.

  • Any ideal theory fails to be action-guiding.

In order to assess this claim, we need to distinguish between three senses in which a theory might be action-guiding.

C1: A theory is appropriately action-guiding only if it motivates existing agents.

But, Valentini retorts, the fact that people lack motivation hardly counts against (the truth of) conception of justice. As long as it is reasonable to expect people to be motivated by it (under favorable circumstances), there’s no trouble,

C2: A theory is appropriately action-guiding only if it is immediately applicable to current political decisions.

Valentini rejects this requirement of immediacy. There will always need some translation needed between theory and practice. That’s unavoidable. But ideal theories might guide indirectly—providing principles that make us sensitive to some important considerations.

C3: A theory is appropriately action-guiding only if it is, if we were to apply its principles, we would obtain morally adequate results.

This interpretation better captures what critics of ideal theory mean when they claim that ideal theory is useless or morally problematic. An ideal theory is useless if there exists an unbridgeable gap between the theory and the non-ideal circumstances. A theory is morally problematic if, were we to apply the theory’s principles, it would generate morally bad outcomes.

So, C3 is what we should have in mind. The next step, Valentini thinks, is to see whether C3 is true. She argues that while C3 is true of some ideal theories, it is not true of ideal theories as such.

To do so, she considers two explains of ideal theories that can guide action, and two that cannot. I’ll sketch one of each.

First, the good case: Rawls’s theory of domestic justice. Rawls’s principles of justice idealize away gender discrimination. But, in doing so, he arrives at principles that can be used to judge if a society is gender just, and to constrain institutions (like the family) which perpetuate gender injustice. Of course, what one should do in response to this gender inequality depends on social context.  But, Rawls’s theory gives us resources to determine whether and when a society is gender just. Rawls’s theory (like any ideal theory) is thus incomplete. But it is not irrelevant or counter-productive.

Now the bad case: Rawls’s theory of international justice. Rawls’s theory (presented in The Law of Peoples) aims to detail the “foreign policy of a reasonably just liberal people”. But, in so doing, it idealizes away crucial issues like migration, the regulation of trade. How this is misleading can be seen from Rawls’s discussion of how rich states should interact with poor ones.

Rawls constructs his principles under the assumption that there exists a fair background against which states can trade. He asks, then, how just liberal societies should interact with poor ones. This leads Rawls to the conclusion that rich states should provide limited assistance to poorer ones. But, this assumption is deeply misleading. The world is not populated by just liberal states, but by many unjust ones. Accordingly, Rawls’s account of the relation between actual liberal societies and poorer states is problematic: it misses the fact that this relationship is characterized by continuing injustice and exploration. Liberal states thus not only often fail to adequately assist burdened societies, but actively burden them.

Valentini then raises the question: what distinguishes the good ideal theories from the bad or problematic ones? The answer is that while good ideal societies do not idealize their account of the subject of justice, while bad ideal theories do. Thus, good ideal theories tell us what a just set of background conditions ought to be like, while bad ones “surreptitiously assume” that such a background is in place. That is, these bad idealizations build into the theory a normatively false account of a social phenomenon which the theory itself aims to scrutinize: the agents and institutions to whom the principles of justice should apply. In contrast, the “good” ideal theories only articulate “the circumstances under which” it seems plausible to construct a theory of justice: that is, the circumstances under which justice could be achieved.

Valentini then suggests that critics of ideal theory (like O’Neill) are right that, when ideal predicates are assumed as “a condition for the applicability of the theory”, ideal theories face serious problems; but not all ideal predicates function in this way. She then parallels the situation to the issue of good and bad idealizing in science. Valentini contrasts two cases: Galileo’s Law of Falling Bodies and standard economic theory. In the former case, what is idealized away (friction) can be taken back into account at a later stage, while in the latter case, the idealizing assumptions (likely) misrepresent what they intend to describe.

Valentini concludes that though there is nothing wrong with ideal theory per se, idealized accounts can go awry, and ideal theories would do well to keep in might those facts which are relevant to the question they aim to answer.

Ingrid Robeyns, “Ideal Theory in Theory and Practice”

Ingrid Robeyns, “Ideal Theory in Theory and Practice”

“What are the merits and limitations of ideal and nonideal theory, and what is their property role?” asks Ingrid Robeyns in “Ideal Theory in Theory and Practice”.

Normative Social Justice Analysis: A Typology

To answer that question, she begins by distinguishing “three different layers” of “normative social justice research”. (Note in so doing see sets aside the “pure” ideal theory of G.A. Cohen, which we doubts is normative at all).

  1. Ideal Theory

Ideal theory aims to work out principles of justice that should govern a just society. It aims to specify those principles, the compliance with which is required in order for a society to be completely just.

An ideal theory, though, can be comprehensive or partial. A comprehensive ideal theory would “tell us what conditions should be met before each and every instance of injustice is removed”. Most ideal theories are partial in some way, and there is a spectrum here. An ideal theory can be partial in different respects:

  • It may only specify minimal principles (necessary but not sufficient conditions for justice)
  • It may only offer principles for some domain (e.g., gender justice)
  • It may only offer principles for social (rather than global) justice

Ideal theory, so defined, “functions as a mythical Paradise Island”. It sets a goal at which we want to aim, but does not settle how to get there or whether doing so is possible. One can deduce few guidelines about the route from the endpoint.

  1. Nonideal Theory

Since we don’t live in a fully just society, we need nonideal theory.

Robeyns first criticizes Rawls’s understanding on here on two counts:

  1. Rawls includes subjects in nonideal theory that have at least an ideal component (just war and punishment) because Rawls’s understanding of full compliance is too strict (it should mean: under ordinary circumstances, most people comply)
  2. Rawls ignores some of the important issues of ideal theory (such as whether to adapt the principles for nonideal circumstances and how to make comparative judgments)

She then states her own view. Nonideal theory has several main functions: to help make comparisons between different social states (as Sen has emphasized), to guide our actions in moving towards the ideal (what is sometimes called the theory of transition), to help balance various trade-offs (between different domains of justice or between justice and other values), and how to deal with noncompliance.

  1. Action Design and Implementation

Ideal and nonideal theory, together, are still insufficient. In order to know what to do as a matter of individual action or social policy, we also need to take into account feasibility constraints and and unintended consequences. This is the “action design” stage. Feasibility constraints can be of different kinds. Some are “completely unalterable”” (human morality, e.g.) some are highly contingent (social attitudes about some particular policy). There is a continuum here. Unintended conseucnes are “very important”, but Robeyns says little about this.

Once we are past “action design”, we need to think about implementation. Implementation comes with its own set of challenges and difficulties—getting people on board, communicating the policy, handling it democratically, and concerns about legitimacy. There are also issues about short-term versus long-term gains and trade-offs.

Robeyns thus concludes “the road from ideal principles to effective justice enhancing action is long and potentially thorny, and much work is needed before ideal principles can effectively contribute to solving problems of injustice.” Ideal theory is only one step in the larger change, and if philosophers want to contribute to actual justice, it may be worth devoting their intellectual resources elsewhere too.

Idealization Assumptions

Ideal theory is often confused with theory based on idealizations. This, Robyens thinks, is a mistake and she aims to clarify the relationship between idealization and ideal theory.

Idealizations, for Robeyns, are assumptions which “describe certain aspects of a theory differently from how they are in reality”: for example, the character, motives and capacities of human beings, the nature of society, and the like. Idealizations are often simplifications. Sometimes, these assumptions are normatively motivated (they depict agents as better than they really are).

Most, probably all, ideal theorize use idealizations. There are two reasons for this. First, idealizations makes things more manageable and the problems more tractable by reducing complexity. Second, idealizations are used to model desirable aspects of a just society.

It is crucial to understand why any given idealizing assumption is made. For example, we legitimately assume away certain injustices (e.g. slavery) because these ought not to occur in a just society. But, conversely, we can’t assume that everyone will (in an ideally just society) earn the same wage, since it’s not clear why that is an essential part of a just society.

So, some idealizations are in good standing. Why, then, all the resistance to idealization in political philosophy? There are two reasons, Robeyns thinks, that are worth taking serious. The first is that the transition from ideal to nonideal theory is complicated, and ideal theory does not by itself guide action. The second is that some idealizations are bad, and there has been too little discussion of what distinguishes the good one from the bad ones.  To allay worries about idealization as such, Robyens addresses both issues.

With regard to the first, Robyens agrees that principles derived in ideal theory, because they are derived in ways that involve idealizations, cannot “serve as principles for the nonideal world. This is so for a variety of reasons, but Robeyns focuses one in particular: namely, that ideal theory assumes away some important dimensions of injustice.

Take Dworkin’s principles of endowment insensitivity and choice sensitivity. These principles cannot (or better, ought not) be directly implemented since the present world is not as pure as the world in which the principles are derived and defended. For example, people don’t have independent preferences and do base choices on prejudice. So, if we were to implement those principles, they wouldn’t take into account adaptive preferences. Accordingly, the principles need to be altered and the fact that preferences are often based on injustice needs to be taken into account. Dworkin himself has a “theory of improvement”, but that theory is only partial.

Thus, more generally, the problem is that we need a story about how to deal with idealization in moving from the ideal to the nonideal level. At the ideal level, it is fine to idealize away alterable causes of injustice which would be eradicated in a fully just society. But what does this imply in the nonideal case?

The second problem is that some idealizations are “bad idealizations”: idealizations which do not serve to model an absence of injustice in a way that is theoretically justified, but rather ignore certain kinds of injustice which do need to be theorized about. An example is assuming that persons are not dependent for care upon others, or not constrained by duties of care. Human dependency is an inevitable fact, one that would arise in a just society. So a theory of justice must give an account of it. In reply to the objection that idealizing away care is not so bad if the theorist is avowedly only offering a partial ideal theory, Robeyns asserts that this idealization will likely have spillover effects to the partial domain in question. The problem of bad idealization, then, also explains why some take ideal theory to have a negative ideological function.

This analysis has three important upshots.

First, theorists should be more explicit about the limitations of their ideal theories, and critics of ideal theories must challenge them at the level of ideal theory.

Second, theorists should consider carefully which idealization can be justified at the ideal level.

Third, nonideal theorizing is extremely important.


Robeyns then concludes that ideal theory plays an important, though limited, role in thinking about social justice. Ideal theorists would do well to acknowledge its limitations. And it certainly seems that the strong priority argument advanced by Rawls is more fraught than it might initially appear.

Robeyns then concludes the debate about ideal/nonideal theory could be made more productive if:

  1. There were more clarity and explicit discussion about the meanings of “ideal theory ” and “nonideal theory”.
  2. “Ideal theory” were not so often confused with “theories based on idealizations”, and more explicit investigation of this relationship occurred.
  3. Political philosophers answered in what sense the role of ideal theory is limited.
  4. Political philosophers worked to understand why and when idealization is useful.
  5. More robust work in nonideal theory were developed.
  6. Students we socialized to take nonideal theory as seriously as ideal theory.
  7. Institutions and philosophers stopped unjustifiably privileging of ideal theory.

Laura Valentini, Ideal vs. Non-Ideal Theory: A Conceptual Map

Laura Valentini, Ideal vs. Non-Ideal Theory: A Conceptual Map

In recent years, political philosophers have started to pay more attention to methodology, largely to due pressure from the charge that political philosophy is too detached to really guide political action  Many of theses methodological debates have clustered together under the heading ideal/non-ideal theory.

In this article, Laura Valentini argues—I think rightly—that the debate about ideal and non-ideal theory is not really one debate. Instead, she argues that there are at least three distinct meanings of ‘ideal’ and ‘non-ideal’, each with a corresponding set of questions:

  1. Full versus partial compliance: what is the relationship between duties and obligations under conditions of full compliance, and duties and obligations under partial compliance?
  2. Utopian versus realistic: which feasibility consideration should constrain normative political theorizing?
  3. End state versus transitional theory: should normative political theory aim at identifying social perfection or on transitional improvements?

In making this argument, Valentini focuses on the case of ideal/non-ideal theory about justice and admits there may be further questions worth identifying in other areas of political philosophy.

  1. Full Versus Partial Compliance

Rawls works out an ‘ideal theory’ which is a theory designed under two assumptions: (i) all relevant agents comply with the demands of justice that apply to them and (ii) the natural and historical conditions in which these agents find themselves are favorable. Non-ideal theory a theory for cases in which (i) and/or (ii) do not obtain.

Many of criticized assumption (i). Although Rawls’s theory may allow us to identify cases of partial compliance, it is not clear that any theory designed for full compliance can tell us what is required of us (what our duties and obligations are) under conditions of partial compliance? Many issues of social injustice exist because of institutional and personal failures to fulfill duties (global poverty, e.g.). And so, we need to know what we ought to do in circumstances in which others fail to do their part?

Assuming one’s ‘fair share’ is defined by what one would be required to do under conditions of full compliance, Valentini argues (following David Miller) there are three broad answers:

  • Do your exactly your fair share
  • Do more than your fair share
  • Do less than your fair share

Each answer might be reasonable in different cases. (1) might be adequate for a case in which further action by a single person would make little difference (e.g., political campaigning for institutional reform, when others are apathetic). (2) might be reasonable when considering very wealthy citizen’s duties to remove poverty in a very unequal society: A should do more than her fair share because doing so would help lift many out of poverty. (3) might be reasonable when even doing one’s fair share would not increase the likelihood of the desired outcome (say, the release of a certain prisoner). Thus, Valentini concludes, we should be skeptical that there exists some “master-principle” which tells us precisely what to do in cases of partial compliance, and ought instead to focus on determining what it is “reasonable” to demand of individuals to do in particular circumstances of injustice.

  1. Utopian Versus (More or Less) Realistic Theories

Should the principles of justice be constructed in a way that takes into account of the facts? Specifically, should they take into account “feasibility constraints”? There are two broad to answers to these questions: ‘yes’ and ‘no’. Those who answer ‘yes’ require some further account of which feasibility constraints and facts they should take into account. Thus, Valentini divides the discussion in two: first, she considers the debate between ‘utopian theories’ and ‘realistic ones’; then, she considers a spectrum of different realistic theories.

2.1: Utopians Versus Realists

G.A. Cohen, in response to Rawls, defends a version of the utopian view. Justice, Cohen argues, is one value among many. So justice itself does not tell us what to do. But justice itself is a fact-free value, and its demands are to be understood independently of feasibility and other facts. Then, in a further step, the demands of justice are to be balanced with other facts and values to yield ‘rules of social regulation’. On this approach, the question of justice is then primarily an evaluative rather than a normative one.

Rawls disagrees. For Rawls, the principles of justice are developed in response to the human condition—a condition which includes moderate scarcity and limited altruism. The principles of justice are thus normative: they tell us what to think about what we ought to do.

What should we make of this debate? Valentini considers two possibilities. The first is that the dispute is merely verbal: a dispute about what the two sides mean by ‘justice’. If this is so, Valentini objects that Cohen’s approach fails to do justice to the important role that ‘justice’ plays in our thought. The second is that the two approaches rest on incompatible assumptions: for Cohen, what matters are principles that are valid across all possible worlds; for Rawls, what matters are principles for this and nearby possible worlds (inhabited by beings like us, in similar circumstances).

2.2: More or Less Realistic Theorizes.

Realism, Valentini notes, comes in degrees. Theorists can agree that facts are crucial to the design of normative principles, but disagree about which facts matter. A common criticism of Rawls’s view is that it is too idealized, in that it assumes away some crucial facts. Valentini considers a spectrum of criticisms.

Waldron sympathetically criticizes Rawls for failing to take seriously the fact that people reasonably disagree about justice. More hostilely, political realists (a la Bernard Williams) argue that perfect justice is imaginable, but not feasible’ so it is unfair to hold actual societies to its standards. We should thus (they conclude) employ principles which take into account the status quo facts of human selfishness, corruptibility and the like.

What to make of this dispute between the ‘realistic utopianism’ of Rawls and ‘full-blown political realism’? To answer this question, we need to distinguish two readings of the realists’ criticism of Rawls. On one reading, the critique is really about Rawls’s obsession with justice, as opposed to other values. On a second reading, the point is that a theory of justice should take into account these further facts about social life.

On this second—and more interesting—reading, the debate is difficult to resolve. There are good arguments on both sides. On the one hand, the fewer feasibility constraints taken into account at the design stage, the less effective the principles: either because they fail to address the problems we are actually facing or because they will be motivationally defective (because at odds with common sense, and thus controversial). On the other hand, the more factual constraints incorporated, the more likely the resultant principles are to uncritically endorse the status quo or fail to identify some of its salient injustices.

Valentini is thus simple to the suggestion (by Carens) that the appropriate level depends on one’s aims—whether we are trying to critique our existing social assumptions or come up with reasonable policies for our flawed world.  Political theories can be “seen as more or less realistic along a continuum”, and the key is to make sure that the factual assumptions are appropriate to the particular question. Thus, realists and Rawlsians might be simply answering different questions.

  1. End State Versus Transitional Theories

Rawls’s  ideal theory sets a long-term goal for institutional reform, and that ideal theory asks how the goal might be justly and feasibility achieved. Indeed, he suggests that the former task-identifying an ‘end state’ at which we should aim, is a prerequisite the non-ideal task of sorting out how to bring that state about. This priority claim has been disputed by many—most notably, by Sen.

If our concern is to make the world more just, Sen argues, identifying an end state of perfect justice is neither necessary nor sufficient. It is not necessary because we can evaluate whether one state is more just than another even without knowledge of perfect justice. It is not sufficient since the end state itself does not allow us to measure how far any given state is from the ideal.

Responses to this criticism has varied. Some has suggested the importance of ideal theory lies in the importance of knowing what perfect justice consists in itself, and thus, though not prior to non-ideal theory, ideal theory is nevertheless important and respectable. Others (e.g. Simmons) argue that a successful theory of transition does require an ideal theory. We need to know both what counts as success (fixed by the end we want to achieve, given in ideal theory), and also whether individual transitional steps are permissible. This reply, Valentini thinks, points to “an under-explored area in political theory”: the analysis of transitional constraints. These include not only feasibility constraints but also a discussion of the moral costs of transition (acceptable means in moving towards justice. So, Valentini concludes, both Sen and Simmons have a point.


Valentini wraps up the article by noting that the there is probably “no right answer to the question of whether a normative political theory should be ‘ideal or ‘non-ideal’” (i.e., realistic or not). Rather, which idealizations are appropriate depend upon the particular question one wants to answer.

Kwame Anthony Appiah, “Political Ideals: Lessons from John Rawls” in As If: Idealization and Ideals.

Kwame Anthony Appiah, “Political Ideals: Lessons from John Rawls” in As If: Idealization and Ideals.

Anthony Appiah’s recently published book As If: Idealization and Ideals is an insightful and original treatment of the role of idealization in philosophical thought. This has been a hot topic in recent political philosophy. But, part of what makes Appiah’s discussion more interesting than most is that he places his discussion of ‘ideal theory’ within a much broader treatment of ideals in other areas of philosophy. In this post, though, I focus on only Appiah’s explicit treatment of political ideals, rather than summarize the entire book. But I encourage interested readers to have a look at the other two chapters of Appiah’s book.

Ideal Theory

Appiah begins by introducing Rawls’s discussion of ideal theory (for a brief summary, see here; for a different reconstruction , see here). The term is first introduced in A Theory of Justice, but it is not well-defined. Instead, Appiah suggests it should be understood through the idea of a well-ordered society. A well-ordered society, for Rawls, is one is which (1) everyone accepts (and know that others accept) the same principles of justice and (2) basic institutions are governed by those principles. So, an ideal theory is one that is worked out for a society of this kind.

But, Appiah asks, what use is such a theory for thinking about justice in any actual society, since actual societies won’t—for reasons of basic psychology, among others—be well-ordered? One problem is that ideal theories fail to tell us what to do in case where others do not comply with the demands of justice. Similar objections, Appiah thinks, apply to most theories of justice (to Nozick’s (118-120) and Dworkin’s (120-121), e.g.).

To make the basic contours of the problem clearer, Appiah introduces a much simpler case: that of justice for asylum seekers. Suppose (as some argue) in a just world there would be no states. Then, there would be no asylum seeks. So, because there would be no asylum seeks in a just world, an ideal theory tells us nothing about what to do our obligations to them. (Appiah also notes a further problem for Rawls: if parties to the original position know “basic facts” about human psychology, wouldn’t they then assume any society won’t be well-ordered?).

Why, then, is ideal theory—that is, a theory in which full compliance is assumed—the way to proceed in political thinking? (Another way to put the point is why would the parties to the original position assume this? Rawls’s actual argument, Appiah claims, is question begging (124)). The problem is again general. To illustrate why, Appiah considers a mock dispute between a utilitarian (Dr. Welfare) and a defender of partiality (Prof. Partiality). Complying with the latter’s weaker requirement pretty well might be better than complying with the former’s requirement not at all, so if one assumes that, given what people are actually like, these are the two likely outcomes, one can see that the debate here is really about the function of ideals.

Some Basics

So far, Appiah has argued, we need to get clearer about the role of ideals in moral thinking, since many debates in moral theory actually turn on this question. To help do so, Appiah reminds us of two points (discussed earlier in Ch. 1): (a) idealizations are models of something that procedure as if something we know is false were true, (b) we employ idealizations because the resulting model is useful for some particular purpose. What purposes, then, might idealization in normative theory serve?

The answer, Appiah argues, isn’t Rawls’s answer—namely, that ideal theory provides us with the only possible basis for a ‘systematic grasp’ of the urgent problems that arise due to non-compliance. Rather, we need to ask, for each theory: (a) which idealizations does it employ, (b) for what purposes might those idealizations be useful.

Appiah then distinguishes two kinds of idealization: (1) false non-normative assumptions (2) false normative assumption. A case of the latter is (e.g.) where constructs a model in which one assumes that people have no special obligations to their children (Appiah takes it for granted that the assumption is false). Can such an idealization be useful?  Yes, Appiah explains in the next section.


Theories which employ false normative assumptions Appiah dubs counter-normative theories.  Are such theories useful? Well, Appiah thinks, to answer that kind of question, we need to ask: useful for what?

One possibility is that counter-normative theories may help “make the world better” by influencing our dealing with others. This might look question begging: in order to know whether a theory contributes to making the world better, we need to know what making the world better entails. Doesn’t that require a true (not counter-normative) theory?

Appiah thinks not. As we saw in the dispute between “Dr. Welfare” and “Professor Partiality”, the former can agree that, given our limited capacities, it would be better if people acted as if the  latter’s theory were true, since doing so would (by Dr. Welfare’s own lights) make the world better. The latter theory may be strictly false, but better for action.

Note: One might worry here that Appiah basically concedes the point: yes, in order to know whether a moral theory makes the world better we need to have a true moral theory; but no, we needn’t think that the true moral theory is the best guide to action. In any case, the second point is clearly true—and the one Appiah is interested in. But I do think Appiah needs a better response to the objection about knowledge. One response is this: what our obligations are in the actual world is not determined by what our obligations would be in an ideal world. Those are simply different cases. So, in order to work out what we ought to do in the nonideal case, we do need a true moral theory. But we don’t need an ideal theory.

Or consider a simpler case: although it is not always true that lying is wrong, it might be best to act as if it were true (so too for “torture is always wrong”). Carrying around in my head the “more complex, hedged claim” would require me to check the easy cases against the rule. And given my limitations, I am prone to make mistakes. (These are familiar rule-consequentialist thoughts).

Thus, Appiah thinks, we should distinguish between several questions. First, we should distinguish between:

  1. Should I act as if normative claim X (e.g., torture is always wrong) is true?
  2. Is X true?

Second, we should distinguish between:

  1. Should I act as if X is true?
  2. Should we act as if X is true?

The world might be better if we all acted as if X were true, even if it would not be better if only I alone did so. The “X” Appiah considers is “murder is always wrong”: If everyone acted as if that were true, the counter-example cases to the principle might not even arise. (This might be so, furthermore, whether or not we knew the truth of X.) Another case is trust: it might be better if everyone acted as if people (in general) ought to be trusted, but it might be bad for only me to act this way.

Appiah has thus argued that there may be normative benefits to accepting counter-normative theories.  (He also emphasizes the opposite may be the case too: there may be cases in which A knows that X is true, but acts as if it isn’t (the case of the psychopath, say. This, he notes, raises some issues about internalism and externalism. I’ll set this bit aside).

Identity and Looping Effects

Appiah then turns to the case of theories which make non-normative idealizing assumptions. And the first case he considers is that of identity.  Here, there’s a familiar sort of case. Suppose that races don’t exist, but racism does. In responding to racism we may find (as in the case of affirmative action) that we assign people to racial categories. What’s happening here?

The answer, Appiah argues, is in part that mistaken beliefs can generate social categories and identities. And acting within these categories, even if they don’t reflect some higher objectivity, may be useful for some practical purposes. The case Appiah considers two cases. First, taking up homosexuality as a fiction, rather than as a scientific hypothesis: i.e., not assume that there are really people with fixed sexual identities, but merely noting the utility of acting as if they are. Second, he argues we can racial categories in the same way: educated people know race isn’t real but they behave in many contexts as if it were. Identity, he concludes, are “one of our most potent idealizations”.

Note: I’m a bit unsure about how to take Appiah’s point here because I am unsure what conception of “social construction” he is employing. He seems to imply that if something is socially constructed, it is not “real”. But there are different ways of understanding social construction on which this implication does not hold.

What to Take For Granted

So: some facts can be assumed away for theorizing, if theorizing on the assumption that those facts don’t obtain can be useful. But, Appiah asks, which facts? This, Appiah thinks, is a difficult question. If we go too far—assuming away, e.g., our capacity for pleasure and pain—we end up with a morality that is “rather unhelpful in thinking about life for homo sapiens”. So, Appiah concludes there are “some aspects of human nature” that have to be taken as given. But, on the other hand, theorizing can’t take everything is given—that would just mean that everything that is ought to be. So, Appiah asks: “when should we ignore, and when insist on, human nature?

Staying in Character

As an example, Appiah considers the debate between virtue ethicists and situational psychologists. The former holds that virtues are stable dispositions a person has; the latter claims that there are no such stable dispositions (or very few, anyway). Even supposing the situationists are right, Appiah thinks virtue ethicists can declare that they are engaging in a sort of idealization. But is such idealization useful? Appiah goes through a complex dialectic here: Perhaps “virtue” is a useful ideal to strive for. But, even if so, the resulting “virtue theory” looks rather different than standard approaches to virtue ethics. I’ll leave it to readers interested in virtue ethics to piece through this part of the argument for themselves.

Back to Politics

Appiah then turns back to political theory. The basic problem—what to take for granted—seems “somewhat more tractable” in the case of political theory than it does in other cases. This is in part because theorists have been more explicit about their idealizations: it takes for granted some basic picture of the good and some simplified picture of moral psychology. And, since we do have a quite rich set of psychological theories at our disposal, we know what we are leaving out. So we can consider, I each case, whether adding some extra fact that we do know would be helpful to our theorizing. And, he notes, something that is helpful for one purpose might not be helpful for another.

Example: Migration

As an example, Appiah considers Joseph Carens’s discussion of the ethics of migration. Carens contrasts two approaches. The first, realist approach is concerned with “guiding action in the world in which we currently live”. The second, idealist approach is concerned with “fundamental justification”. These two approaches assume different institutional, behavioral, and political facts. The former may be useful for policy making, but may be vulnerable to the charge that it “misjudges what policies are actually achievable”. To make these possibilities visible, one needs the idealistic approach.

One might complain that only the idealist approach is really moral. The moral answer to the question of what we ought to do is given by the most ideal theory. But one task of political theory is just to figure out what would be best to do given where we are now.

Now Appiah rejects the claim that there should (even in an ideal case) be no sovereign states (or only one state). And in the ideal case each state ought to accept their fair share of refugees and migrants. In the nonideal world, Appiah argues, it would be good if many states did more than their fair share of migrants, given that others work (and also combat the conditions which make migration a necessity for many).

The question of what counts as a fair share is settled by asking for the best possible practices, the full compliance with which would give everyone their do. In deciding whether a practice is possible, we can debate which idealizing assumptions to make. And for this reason, there will be different answers about what counts as a fair share.

The Best and the Better

Rawls’s basic thought is this: a social system is a scheme of cooperation and in order to be entitled support that scheme must give a satisfactory answer to the question of why it advantages some and denies those advantages to others. Rawls’s answer in the case of economic inequality is that you can justify these advantages if you should that everyone would be better off with them because of the incentives they create.

This argument presupposes certain (realistic) motivational structures. Why not be more idealistic here, assumption people work because for the intrinsic satisfaction of it? This question is one Cohen presses against Rawls. Appiah thinks the point is best conceived one about idealization—although Rawls doesn’t think of it in this way.

Indeed, much of Rawls’s argument, Appiah thinks, can be reconstructed without any commitment to the claim that we have a conception of “a perfect society”. And he endorses some version of Sen’s criticism: that we can judge social state A as better than B even if we don’t have a conception of a perfectly just society. Ideal theory, then, isn’t “necessarily the right starting point”.

Often, it is better to start with comparative judgment.  We can and should assume a more modest moral epistemology on which, what we know about justice we learn from negating, as it were, injustice (which we know well). In forming a conception of justice on the basis of negating injustice and oppression, one of course needs to idealize. But the idealization is more tractable and more modest.


Appiah then concludes by discussing the relation between idealization and truth. Truth still matters in normative theory in three ways:

  1. In idealizing, we need to imagine what it is for something false to be true.
  2. A defense of idealization depends on its being true that the model in which that idealization figures is useful for some purpose.
  3. In identifying such a purpose, we need to grasp what it would be for that purpose to be achieved.

So, the truth that matters most is truth about “what is possible”.