The Methodology of Political Theory

Christian List and Laura Valentini*

Abstract: This article examines the methodology of a core branch of contemporary political theory or philosophy: “analytic” political theory. After distinguishing political theory from related fields, such as political science, moral philosophy, and legal theory, the article discusses the analysis of political concepts. It then turns to the notions of principles and theories, as distinct from concepts, and reviews the methods of assessing such principles and theories (e.g., the reflective-equilibrium method), for the purpose of justifying or criticizing them. Finally, it looks at a recent debate on how abstract and idealized political theory should be, and assesses the significance of disagreement in political theory. The discussion is carried out from an angle inspired by the philosophy of science.

Political theory, sometimes also called “normative political theory”, is a subfield of philosophy and political science that addresses conceptual, normative, and evaluative questions concerning politics and society, broadly construed. Examples are: When is a society just? What does it mean for its members to be free? When is one distribution of goods socially preferable to another? What makes a political authority legitimate? How should we make collective decisions? What goals should our policies promote? How should we trade off different values, such as liberty, prosperity, and security, against one another? What do we owe, not just to our fellow citizens, but to people in the world at large? Is it permissible to buy natural resources from authoritarian governments? Can war ever be just?

Political theory is a long-established field. Its questions have animated thinkers since Classical Antiquity. Consequently, the methods of theorizing and substantive conclusions are diverse. In this article, we review the methodology of a core branch of contemporary political theory: the one commonly described as “analytic” political theory.

Given space constraints, we are not able to cover the history of political thought, the study of ideologies, the comparative study of political thought across cultures, and “continental”  political  theory,  including  “hermeneutic”,  “post-structuralist”,  and “post-modernist” approaches (for a more comprehensive discussion, see Leopold and Stears 2008). Nonetheless, the label “analytic” should not be interpreted narrowly. It is meant to refer to an argument-based and issue-oriented, rather than thinker-based and exegetical, approach that emphasizes logical rigour, terminological precision, and clear exposition. The term “analytic” is not intended to refer only to the logical and linguistic traditions of philosophy associated with the Vienna Circle and philosophers

in Oxford and Cambridge in the first half of the 20th century. Contemporary analytic political theory goes well beyond conceptual analysis. At least since the publication of

This review is structured as follows. In Section 1, we briefly demarcate the scope of political theory. In Section 2, we comment on the analysis of political concepts. In Section 3, we introduce the notions of principles and theories, as distinct from concepts. In Section 4, we discuss the methods of assessing such principles and theories, for the purpose of justifying or criticizing them. In Section 5, we review a recent debate on how abstract and idealized political theory should be. In Section 6, finally, we discuss the significance of disagreement in political theory.

One clarification about the nature of this review is needed. Although we cover established ground, we do so from an angle that will be somewhat unfamiliar to at least some political theorists – namely an angle inspired by the philosophy of science. We have chosen this angle with a view to systematizing the activity of analytic political theorizing so as to make its connections with other fields of philosophy and positive  science  more  transparent.  This  seems  appropriate  in  the  context  of  a handbook on philosophical methodology.

1. The scope of political theory

To demarcate the scope of political theory, it is helpful to distinguish it from its most closely related neighbouring fields: political science, moral philosophy, legal theory, normative economics, and social ontology. We also offer some comments on the use of the label “political theory”, as opposed to “political philosophy”.

1.1 Political theory and political science

Political theory can easily be distinguished from (positive) political science. Political science addresses empirical and positive questions concerning politics and society (for an overview, see Goodin 2009). It seeks to describe and explain actual political phenomena,  such  as  elections  and  electoral  systems,  voter  behaviour,  political- opinion formation, legislative and governmental behaviour, the interaction between the legislative, executive, and judicial branches of the state, and the stability or instability of different forms of government. Political theory, by contrast, addresses conceptual, normative, and evaluative questions, such as what a democracy is, how we ought to organize our political systems, and how to evaluate the desirability of policies.

Political theory and political science can complement each other. Normative recommendations and evaluations of policies or institutional arrangements often rest on empirical premises. It is hard to arrive at a blueprint for a just society, for example, without understanding how society actually works, since normative recommendations may have to respect feasibility constraints (e.g., Gilabert and Lawford-Smith 2012). Thus political theory requires political science and the social sciences more generally. Similarly, when political scientists investigate, for instance, whether democracy promotes economic development or whether free societies are more politically stable and less corrupt than unfree ones, they need to know what counts as a democracy or how to define freedom. These questions require the conceptual input of political theorists.  Finally,  among  the  large  number  of  empirical  questions  that  political

scientists could investigate, some are undoubtedly more interesting, relevant, and pressing than others, and political theory can help shape the research agenda.

1.2 Political theory and moral philosophy

While there is a natural division of labour between political theory and political science, the distinction between political theory and moral philosophy is subtler. Some scholars view political theory as a subfield of moral philosophy, in which the concepts and principles from moral philosophy are applied to political – and, more broadly, social – problems. If one views political theory in this way but also does not want it to collapse completely into moral philosophy, one must give a clear criterion of when a problem counts as “political”.

Unfortunately, this is not straightforward. For instance, saying that a problem is political if it involves multiple people or their living together may seem plausible, but is  too  inclusive.  Many  problems  in  personal  ethics,  such  as  how  to  treat  one’s relatives or friends, would count as political on this criterion, even though they are not generally classified in this way. On the other hand, saying that a problem is political if it involves the state or government is too restrictive, because a number of problems outside the sphere of the state or government narrowly construed might still be seen as political.

Consider, for instance, gender relations in civic life or in the workplace, which many people regard as a political issue. The question of whether, and how, a distinction between the private or personal domain and the political or social one can be meaningfully   drawn   is   controversial.   Some   feminist   thinkers   have   famously challenged the possibility of drawing any such distinction and have endorsed the slogan “the personal is political” (for an overview, see Baehr 2013). In sum, if one wanted to distinguish political theory from moral philosophy by referring solely to the substantive domain  of  problems  addressed,  one  could  at  best  use  some  heuristic criterion to capture a conventional distinction, but this would yield no principled line.

Another way to distinguish political theory from moral philosophy is to invoke the conditions of theorizing in each field. The aim of moral theorizing, one might say, is to come up with the correct solution to any given moral problem simpliciter – the solution that, based on the theorist’s comprehensive moral view, is right. The aim of political theory, one might argue, is different. The political theorist, at least under modern conditions, is engaged in problem solving under a particular constraint: the presence of pluralism and disagreement about how to solve the problem at hand (see Rawls  1996;  Waldron  1999).  Thus  any  compelling  solution  to  the  problems  of political theory, such as how to define justice or how to design a legitimate procedure for collective decision-making, must appeal to people with a variety of (reasonable) viewpoints, precisely because those solutions are meant to apply to, and be acceptable in, pluralistic societies.

If we use such a methodological, rather than substantive, criterion for distinguishing political theory from moral philosophy, we need not worry about identifying a particular domain of problems that counts as political. Rather, we can say that the hallmark of political theory is its mode of theorizing, against the background of (reasonable) pluralism. We return to this idea in Section 6, where we discuss the significance of disagreement in political theory. Of course, substantive and methodological criteria for distinguishing political theory from moral philosophy can be combined.

1.3 Political theory and legal theory

We now turn to the relationship between political theory and legal theory. The two overlap, and it is difficult to draw a sharp distinction between them. We might again arrive at a rough distinction by using some heuristic criterion to identify what counts as “legal” or “related to the law”. As with the attempt to distinguish the “political” from the “private” or “personal”, however, we cannot expect any such criterion to yield a definitive distinction.

Alternatively, we might try to distinguish political theory from legal theory by identifying different modes of theorizing associated with each field. For instance, an argument to the effect that justice simpliciter requires respect for human rights and certain universal welfare protections is distinct from an argument to the effect that a particular constitution or kind of legal system, properly interpreted, requires them. One might say that the former argument belongs to political theory, the latter to legal theory (see the discussion of the nature of legal interpretation in Dworkin 1986).

More generally, one might say that the deontic concepts used in legal theory, such as legal permissibility, legal rights, and legal obligations, are different from their counterparts in political or moral theory and therefore require a different analysis.1

Still, political theory and legal theory are best seen as overlapping fields of enquiry.

1.4 Political theory and normative economics

Political theory also overlaps with normative economics, especially with social choice and welfare theory. Social choice and welfare theory is the formal, but also normative and evaluative, study of (i) collective decision-making procedures, (ii) mechanisms for allocating benefits and burdens in society, and (iii) methods by which a social planner, policy maker, or institutional designer can assess the goodness or desirability of different social states, policies, or institutions. Normative economists investigate these – (i), (ii), and (iii) – by introducing desiderata that any reasonable procedure, mechanism, or method is required to satisfy and then asking which procedures, mechanisms, or methods, if any, satisfy the given desiderata. (For a survey, see List

2013.)

The substantive questions addressed in social choice and welfare theory are similar to some of those addressed in political theory. For example, questions such as how to distribute benefits and burdens are addressed by political theorists and normative economists alike. Indeed, Rawls’s Theory of Justice was, in part, influenced by the normative works of economists such as Kenneth Arrow (1951), John Harsanyi (1955), and Amartya Sen (1970). Similarly, normative economists frequently draw on moral and political theory. For example, John Roemer’s (1998) formal work on distributive

justice is influenced by G. A. Cohen’s (1995) work in political theory; and the work on  variable-population  social  choice  by  Charles  Blackorby,  Walter  Bossert,  and David Donaldson (2005) is influenced by Derek Parfit’s (1984) work on population ethics.

Arguably, the main difference between political theory and social choice and welfare theory is not a substantive one (although the former is broader than the latter), but a methodological one. Mainstream political theory is a non-formal discipline, making at most limited use of formal methods from mathematics, logic, and economics, while social choice and welfare theory is predominantly formal.

1.5 Political theory and social ontology

A less well-known but growing field in the neighbourhood of political theory is social ontology. Social ontology investigates the nature of phenomena such as joint intentions, collective actions, social norms and conventions, group agency, and institutions (e.g., Gilbert 1989; Pettit 1993; Searle 1995; Bratman 1999; Tollefsen

2002; Tuomela 2007; List and Pettit 2011). Its central question, roughly speaking, is: What are the building blocks of the social world, and how are they related to one another, to the individuals involved, and to the physical world?

The substantive questions in social ontology are distinct from those in political theory and in some respects prior to them. Social ontology is primarily a positive and explanatory field rather than a normative or evaluative one. Its relevance to political theory lies in the fact that political theory cannot get off the ground unless we are clear about what entities and properties exist in the social world. For example, before we can answer questions about rights, obligations, and responsibilities, we need to know what entities can be the bearers of rights, obligations, and responsibilities. Are these restricted to individuals, or do they also include certain collectives? Should we regard the state as an agent, as a mere collection of individuals, or as some kind of special fiction? Similarly, before we can answer the question of what is or is not socially desirable, we need to know what the possible objects of value might be. It is difficult to determine, for instance, whether there should be any cultural rights or state subsidies for cultural activities unless we can give at least a partial answer to the question of what we mean by “culture”.

Since different social-ontological theories give us different accounts of what entities and properties there are in the social world, they can, in turn, impose constraints on what the possible objects of analysis in political theory might be.

1.6 Political theory and political philosophy

Although some scholars distinguish between political theory and political philosophy, this is mainly a conventional distinction. It refers, roughly, to the different styles of political theory conducted in political science and philosophy departments, respectively, especially in North America. (In the UK, much of what is conventionally called “political philosophy” is traditionally conducted in political science departments.) Arguably, “political theory” is the slightly more inclusive and interdisciplinary label, referring not only to philosophical work but also to a variety of

other approaches. As mentioned, here we focus specifically on the analytic branch of political theory.

2. Concepts in political theory

A long-standing concern in political theory is the analysis of political concepts: freedom, equality, justice, authority, legitimacy, democracy, welfare, and so on. Each of these has been interpreted and defined in numerous ways, and political theory can help us clarify the advantages and disadvantages of different interpretations and definitions. The bulk of political theory in the decades prior to the publication of Rawls’s A Theory of Justice was conceptual analysis.2

The analysis of political concepts is relevant not only to normative theory building (e.g., any theory of liberalism must employ some concept of liberty) but also, as already noted, to positive work in political science. Thus conceptual analysis is, in some sense, the least normative or evaluative part of political theory.

Of course, when we analyze concepts such as freedom and democracy, we are usually interested  in  the  kinds  of  freedom  and  democracy  that  we  find  valuable  or normatively required. Hence the ultimate motivation for our analysis may well be a normative  or  evaluative  one:  we  may  wish  to  clarify  these  concepts  in  order  to express normative or evaluative principles in terms of them. Logically, however, the question of how to understand freedom and democracy is distinct from the question of whether  freedom  and  democracy  so  understood  are  valuable.  Indeed,  political scientists may sometimes be interested primarily in whether freedom or democracy can serve as independent variables in explanations of political phenomena (e.g., when they investigate whether freedom promotes prosperity or whether democracies are less prone to fighting wars against one another), irrespective of any considerations of

value.3 In  this  section,  we  introduce  some  key  ideas  relevant  to  the  analysis  of

political concepts.

2.1 What is a concept?

We  use  concepts  to  categorize  or  classify  objects.4 The  concept  democracy,  for example, may help us distinguish between those forms of government that are democratic and those that are not. The concept legitimacy, in relation to acts of state coercion, may help us distinguish between those acts of state coercion that are legitimate and those that are not.

For  the  purposes  of  this  article,  we  assume  that  any  concept  has  a  domain of application. This is the set of objects of which it is meaningful to ask whether they fall under the given concept or not. We might say, for example, that the domain of the concept democracy is the set of all systems of government or decision-making. For any object in that set – i.e., any system of government or decision-making – we can meaningfully ask whether it is democratic or not. By contrast, for objects outside that domain, it is not meaningful to ask whether they are democratic or not. It makes no sense to ask, for instance, whether an equilateral triangle or a mountain is democratic or not (cf. Dworkin 1986, 75). Note that the domain of the concept democracy – or justice, or freedom, and so on – can be variously specified; we return to that point below.

Further, any “classically well-behaved” concept has defining conditions. These determine, for any object in the concept’s domain, whether that object falls under the concept (“satisfies it”) or not. In our example, the question of whether a particular system of government or decision-making, say the political system of Iceland, is democratic or not depends on a variety of features of that system: how decisions are made, who participates in those decisions, how the participants provide their input, how the decisions are implemented and by whom, and so on.

Philosophers are divided over the extent to which concepts in general have defining conditions (for a review, see, e.g., Margolis and Laurence 2012). Many of our common-sense concepts arguably lack such conditions. We may be able to pick out some paradigm instances (or “prototypes”) of redness or beauty, but may be unable to arrive at clear-cut necessary and sufficient conditions that an object must satisfy to be red or beautiful (for a related discussion, see Dworkin 1986, chap. 2). By contrast, in theoretical work, it is usually desirable to look for concepts with defining conditions. Occasionally, however, some theoretical concepts may be regarded as undefined “primitives” or as characterizable only through “prototypes”.

Finally, a concept’s extension is the subset of the domain consisting of precisely those objects that fall under the concept (“that satisfy it”). If the concept has defining conditions, these determine the extension. The extension of the concept democracy is the set of all those systems of government or decision-making that, according to the concept, count as democratic. More generally, there can be “non-binary” concepts, which do not subdivide objects into those falling under the concept and those not falling under it, but which instead classify objects on one or several dimensions that may  each  admit  of  degrees.  For  example,  equality  and  welfare  are  non-binary concepts. The level of equality or inequality in a particular distribution of goods is a matter of degree, and different interpretations of equality give us different accounts of when one distribution counts as more equal than another (see, e.g., Sen 1980). Similarly, a person’s welfare is a matter of degree and, on some accounts, even given by a vector of multiple attributes, representing different dimensions or aspects of welfare. Sen (1987) has argued for the “constitutive plurality” of the concept standard of living; there are multiple dimensions on which a person’s standard of living can be categorized.

For practically any salient concept in political theory, there are debates about what the domain of application is, what the defining conditions, if any, are, and which objects belong to the concept’s extension and which do not. Just think of the many different

ways in which the concept democracy may be understood. The domain of application may be specified in a variety of ways: for example, as a set of voting procedures, as a set of decisions, or as a set of entire political systems. Similarly, we may be divided over the defining conditions that determine whether something is democratic or not: for example, do only formal, constitutional features of the political system matter, or are features of actual political practices relevant as well, and if so, how? And is democracy precisely definable at all? In consequence, we may end up with different extensions of the concept democracy (see, e.g., Christiano 2008; List 2011). Similar considerations also hold for other key political concepts, such as justice, freedom, equality, and legitimacy. Indeed, many political concepts are what political theorists call “essentially contested” (Gallie 1955).

2.2 Concepts and conceptions

Political theorists, following Rawls, who in turn follows H.L.A. Hart, sometimes distinguish between concepts and conceptions (Rawls 1971/1999, 5). Concepts, in that terminology, are less fully specified than conceptions. For example, we may have a  broadly  outlined  concept  of  freedom  as  the  absence  of  constraints  on  agents’ actions, which still leaves open what kinds of constraints, agents, and actions matter (MacCallum  1967).  A  full  specification  of  those  constraints,  agents,  and  actions yields a precise conception of freedom. Different conceptions can thus be compatible with the same broad concept.

We can translate the distinction between concepts and conceptions into our earlier terminology by defining a conception exactly as we defined a concept in the last section, where the domain, defining conditions (if any), and extension are fully specified, and redefining a concept as a broader family of such conceptions, with some aspects of the domain, defining conditions, or extension left open.

2.3 Desiderata on concepts

The following is a list of desiderata that systematize requirements often implicitly employed by political theorists engaging in conceptual analysis:

Respecting our intuitions: We may want to interpret a concept, such as freedom or democracy, in a way that is broadly in line with our intuitions, especially if this concept has a common-sense interpretation. If we arrived at an interpretation of freedom that classifies some intuitively clear cases of unfreedom as instances of freedom, or vice versa, this would be suspect. For example, G.A. Cohen has famously criticized Robert Nozick’s “moralized” conception of freedom (according to which freedom is, roughly, the absence of rights-violating interference) on the grounds that it delivers counterintuitive judgments, such as that a justly imprisoned criminal is not unfree (G. A. Cohen 1988). Similarly, we would be disturbed if our interpretation of democracy classified some clear-cut cases of tyranny as democratic. In such cases, we would either have to give strong reasons for overruling our intuitive judgments or search for a better interpretation of the concept. Later, we discuss Rawls’s method of reflective equilibrium, which may help us adjudicate cases in which our theoretical conclusions conflict with our intuitive judgments (Rawls 1971/1999; Daniels 2013).

Playing the right normative, evaluative, or descriptive role: We may be interested in a concept because we wish to use it in some normative or evaluative principle or in some explanation in political science. A good interpretation of the concept is one that successfully plays the intended normative, evaluative, or descriptive role. Different roles might require different interpretations of the concept. For example, if we want the concept of justice to offer a comprehensive picture of how society should be organized, we are likely to develop a thicker account of it than if we understand justice as one value among the many that should guide institutional design (Rawls

1971/1999; G. A. Cohen 2008, 271–272). However, to avoid a proliferation of rival interpretations  of  the  same  concept,  we  might  also  be  looking  for  a  single interpretation that can successfully play multiple roles.

Standing in the right relationship to other concepts: Since we typically employ concepts not in isolation but in connection with other concepts, we may require these concepts to be related to each other in the right way. If we take rights to entail obligations, for example, our interpretation of rights may constrain the way in which we can consistently interpret obligations. Sometimes we may wish some concepts to be directly inter-definable. It is often held, for instance, that the concept permissibility must be definable in terms of the corresponding concept obligatoriness and vice- versa:  it  is  permissible  that  p  if  and  only  if  it  is  not  obligatory  that  not  p. Consequently, our joint analysis of permissibility and obligatoriness must respect this constraint. Finally, we may wish some given concepts to be sufficiently “differentiated” from one another, in order to avoid redundancies or confusions. (For a related discussion of concept formation in the social sciences, see Gerring 1999.)

Having defining conditions that are neither too “thick” nor too “thin”: Even when we have settled the domain of a given concept (e.g., we focus on a concept of freedom whose domain is a set of acts, as opposed to a set of agents or institutional arrangements), we might still be divided over the concept’s defining conditions. In the case of freedom, a huge variety of different defining conditions have been proposed (for overviews, see Carter 2012; Lovett 2013). To narrow down the range of possibilities, we might require that the defining conditions be neither too “thick”, nor too “thin”: they should not refer to any “irrelevant” facts about the objects to be categorized, but refer to all “relevant” facts. There can then be debates about which facts are or are not relevant and what counts as too “thick” or too “thin”. For example, a concept is moralized if its defining conditions refer to some normative or evaluative

facts. A concept is non-moralized otherwise.5 Nozick’s above-mentioned concept of freedom as the absence of rights-violating interference is moralized in this sense. Political theorists are often divided over which concepts in political theory should be

moralized. Similarly, a concept is modally demanding if its defining conditions refer not only to facts about the actual world but also to facts about other possible worlds –

i.e., facts about what would be or might be, not merely about what is. A concept is

modally undemanding otherwise. Pettit, for instance, argues that freedom is modally

demanding: a slave with a benevolent, non-interfering master still counts as unfree, because there is a nearby possible world in which the master interferes (Pettit 1997). Similarly, concepts such as security and peace are arguably modally demanding, referring not only to the absence of relevant harmful actions or military conflicts in

the actual world but also their continued absence in a range of nearby possible worlds. Political theorists are divided over which, if any, concepts should be modally demanding (see, e.g., Pettit 2011; Southwood 2013).

Having defining conditions that are epistemically accessible: Depending on the intended use of a concept, we may require its defining conditions to be such that it is possible, at least in principle, for us to know whether an object meets them. For example, a concept of welfare whose defining conditions refer to certain kinds of mental states that are inaccessible to any observer would be of little practical use. Similarly, the defining conditions of justice under Robert Nozick’s account, which refer to the entire history of transactions leading to the current distribution of entitlements,  fail  to  meet  this  desideratum  and  thus  render  Nozick’s  account somewhat practically inert (Nozick 1974). Of course, the context and intended use may determine what counts as epistemically accessible.

2.4 What concepts are not

Concepts should not be confused with principles or theories – the topic of Sections 3 and 4. In particular, principles and theories have propositional content and may be true or false, while concepts, by themselves, cannot be true or false. They only categorize objects and, in doing so, can be more or less useful, more or less plausible, and more or less in line with established use or with our considered judgments.6 To give a simple illustration from outside political theory, the concepts red, green, or triangular are neither true nor false. Only statements in which they occur can have truth-values,  such  as  “tomatoes  are  red”  (true),  “snow  is  green”  (false),  and

“rectangles are triangular” (false).

Even if we have a full account of when a political system is democratic or what it means  for  someone  to  be  free,  this  still  leaves  open  the  question  of  whether democracy or freedom are desirable and whether we ought to promote each of them. We need principles or theories – making statements such as “we ought to respect freedom” or “we ought to make decisions democratically” – to address the latter questions. Such principles or theories are then capable of being true or false. Simply put, concepts can serve as building blocks of principles, which can serve as building blocks of theories.

Still, people sometimes say things such as “freedom as non-interference is the true conception of freedom”. Assertions of this kind are best interpreted as abbreviations for claims such as “freedom as non-interference is the conception of freedom that is, in some relevant sense, most appropriate”, which, in turn, could mean that it best captures our established use of freedom or alternatively our considered judgments about what counts as free. Literally, however, the claim that one concept of freedom is the true one is not meaningful, since only things with propositional content can have truth-values. Concepts do not have propositional content: the extension of a concept is not a set of possible worlds (which is what the extension of a proposition is), but a set of objects.7

Even on the above-mentioned charitable interpretation in terms of appropriateness, the assertion “freedom as non-interference is the true conception of freedom” is ambiguous. As noted, it would have to be understood as implying that, among the many different interpretations of freedom, one stands out as “most appropriate”. But, unless   we   specify   a   criterion   of   appropriateness,   there   is   no   unique   such interpretation. For example, one interpretation may be most in line with our common- sense use of the word “freedom”, another most in line with our considered judgments about what counts as free, a third most suitable for playing a particular role in a theory of justice. Different criteria of appropriateness may diverge, and there is no application-independent criterion. Thus expressions such as “the true conception of freedom” are, at best, shorthand for more elaborate expressions involving criteria of appropriateness. To avoid ambiguity, it is best to spell those out explicitly.

3. Principles and theories in political theory

While analytic political theory until the early 1970s was primarily concerned with the analysis of concepts, John Rawls’s Theory of Justice (1971/1999) invigorated the quest for theories and the principles underlying them. Rawls formulated some principles of justice, which are the basis of his theory of how we should organize the “basic  structure  of  society”,  namely  its  main  political,  legal,  and  economic institutions.

In this section, we discuss the notions of principles and theories. Although these are widely used in political theory, they are seldom carefully defined. We hope, therefore, that our discussion will be clarifying.

3.1 What is a principle?

A principle is a statement – a proposition expressed in language – that applies, at least potentially, to more than one case. Usually, this is marked by the occurrence of expressions with appropriate quantifiers, such as “for all X, subject to certain conditions, Y is the case”. A principle is evaluative if it has evaluative content; for instance, it includes evaluative predicates or concepts such as good or bad, better or worse, desirable or undesirable. A principle is normative if it has normative content; for instance, it includes deontic operators such as ought, may, permissible, obligatory, right, or wrong. A principle without any evaluative or normative content is positive. Such principles are common in the sciences; think of the principle of conservation of energy in physics.

Classic examples of normative principles are the Ten Commandments from the Bible, the Golden Rule (“You should treat other people in the way in which you would like them to treat you”), Bentham’s principle of utility (“An action is right if it maximizes total utility”), and more recently Rawls’s principles of justice. Roughly speaking, these state that each person is entitled to the most extensive system of individual

liberties, compatible with a similar system for everyone else (the “equal-liberty principle”); and socio-economic inequalities are permissible if and only if they are compatible with a system of fair equal opportunities (the “fair equality of opportunity principle”) and benefit the least well-off members of society (the “difference principle”) (Rawls 1971/1999).

The propositional content of a principle is the set of all its implications. Sometimes this propositional content may depend on auxiliary assumptions. As already noted, their having propositional content sets principles apart from concepts, which merely offer categorizations.

3.2 What is a theory?

The word “theory” is commonly used in two distinct senses. First, it can refer to an entire academic field or area of enquiry, such as when we speak of “political theory” or “economic theory” as general areas to which curricula or scholarly journals are devoted. Second, the word can refer to a specific theory within such an area, such as Rawls’s theory of justice, the theory of the firm in economics, or Newton’s theory of physics. Our focus here is on theories in this second, specific sense. Surprisingly, there exists no canonical definition of a theory in that sense in political theory.

To provide a starting point, we propose a simple definition inspired by the philosophy of  science  but  adapted  to  the  present  context.8 We  define  a  theory  as  a  set  of statements – propositions expressed in language – which is a candidate for playing some theoretical or practical role and which is, ideally, representable as the set of all

implications of some underlying principles. The set of principles from which the theory can be derived – if there is such a set – is called the theory formulation.

Although loose and abstract, this definition has some merits. First, it allows us to view positive theories (in the sciences) and normative or evaluative theories (in moral philosophy or political theory) as instances of the same general category. Second, it makes transparent the differences between them. For example, the roles played by theories  can  range  from  descriptive,  explanatory,  and  predictive  (in  the  case  of positive  theories)  to  evaluative  and  prescriptive  (in  the  case  of  evaluative  or normative theories). Third, the definition allows us to identify the special challenges that arise when we construct and assess normative or evaluative theories.

Paradigmatic examples of theories according to our definition are Newton’s theory of physics and Rawls’s theory of justice. Each can be viewed as a set of statements, entailed by some underlying principles, which can play a descriptive, explanatory, or prescriptive role. Newton’s theory is the set of all statements entailed by Newton’s principles of physics, perhaps together with some empirical premises about the solar system or other physical systems of interest. It can be used, for instance, to explain and predict the trajectory of the planets around the sun and to guide such engineering

projects as travelling to the moon and safely back. Rawls’s theory is the set of all statements entailed by Rawls’s principles of justice, perhaps together with some empirical premises about relevant social conditions. It can play a prescriptive or normative role, guiding us in the design of social institutions (for an earlier discussion of the relationship between normative and positive theories, see McDermott 2008).

For our purposes, the biggest structural difference between Newton’s and Rawls’s theories is that one is positive and the other normative. Indeed, the principles underlying Newton’s theory are positive principles, while those underlying Rawls’s are normative ones. Generally, a theory is positive if it has no evaluative or normative content; it is evaluative or normative if it has such content. Evaluative theories that offer evaluations of “goodness” or “betterness” are also called axiological.

Earlier, we associated evaluative content with the occurrence of evaluative predicates or concepts, such as good, better, desirable, and so on, and normative content with the occurrence of deontic operators, such as ought, may, and so on. Different accounts of what qualifies as evaluative or normative content can be given, and we need not commit ourselves to one such account here. On any reasonable account, Rawls’s theory will come out as normative and Newton’s theory as positive.

While Newton’s and Rawls’s theories are paradigmatic instances of our definition, a theory need not be self-consciously theoretical. A set of rough and informal principles describing how ordinary objects behave when pushed, dropped, or thrown can constitute a “folk” theory of motion that is predictively useful in everyday contexts. Similarly, a set of basic principles describing how animals respond to noise, movement, and the presence of humans may constitute a simple predictive theory of animal behaviour that members of hunter-gatherer societies might have used to guide their actions. We also routinely employ normative theories without self-consciously doing so. For example, a set of simple principles specifying how we should or should not treat others may constitute a simple “folk” theory of personal ethics.

3.3 The distinction between a theory and the support for it

Our simple definition of a theory is silent on whether the theory is true, useful, or good in some sense. A theory that is false, irrelevant, or superseded still counts as a theory. Describing something as a theory carries no assessment of its truth or acceptability. Thus the most far-fetched and implausible conspiracy theory – for example, in science or in history – still qualifies as a theory.

Similarly, the attempt to downgrade a set of statements by asserting “it is only a theory”, as critics of evolutionary theory or global-warming sceptics sometimes do when they describe their target, makes little sense. Calling something a theory is consistent with its being well-supported and true just as it is consistent with its being speculative and even false.

An important distinction is that between a theory and what is offered in its support. A theory, as we have defined it, is distinct from any arguments, evidence, or justification given for it. For example, Newton’s actual or hypothetical experiments – such as how an apple fell onto his head (though supposedly a myth) – are not part of his physical theory itself; they are part of the evidence he had for that theory. Similarly, Rawls’s

arguments in support of his principles of justice, such as his original-position thought experiment (a hypothetical choice situation in which the parties to the social contract must agree on principles governing the basic structure of society), are not strictly speaking part of his theory of justice itself. The original-position thought experiment, like  any  physical  experiment  or  any  social  scientist’s  observation,  is  offered  in support of the theory in question. Thus Rawls’s book, A Theory of Justice, consists of the theory itself (everything entailed by his principles of justice), together with arguments in support of it (most notably, the original-position thought experiment and reflective-equilibrium considerations), comparisons with rival theories (such as utilitarianism), and a fair amount of commentary on the theory’s interpretation and its applications. This distinction, between a theory and the argument or evidence in support of it, will help us clarify some recent methodological debates in political theory, about abstraction and idealization, which we discuss in Section 5 below.

An important task, separate from defining a theory, is therefore to identify the requirements for a good, or acceptable, theory, and to spell out how we can assess it.

4. The assessment of principles and theories

There are two kinds of criteria that we may use to assess – especially to justify or to criticize – principles and theories: “internal” and “external” criteria. The former concern the way the principles or theory are formulated and their internal logical structure. Criteria such as consistency and parsimony fall into this category. The latter concern the relationship between the principles or theory and what these are “about”: their normative or evaluative content, in analogy with the empirical or descriptive content of a scientific theory. Criteria such as truth or normative adequacy (in analogy with truth or empirical adequacy in science) fall into this second category. In this section, we discuss the two kinds of criteria (“internal” and “external”) in turn.

Throughout this discussion, we focus on theories, rather than principles, as the units of  assessment.  This  is  no  loss  of  generality.  We  are  usually  interested,  not  in individual principles in isolation, but in sets of principles that we wish to assess together. If we wish to assess a principle by itself, we can view it as a special case of a theory, namely a theory that consists just of the principle and its implications.

4.1 Internal criteria

Although we have kept our definition of a theory deliberately thin, defining it simply as a set of statements that may play some theoretical or practical role and that is, ideally, derivable from some underlying principles, we usually want theories to satisfy some further requirements. We now discuss several common criteria for assessing a theory’s internal structure. Like our definition of a theory, they are inspired by the sciences, but apply to normative and evaluative theories as much as they apply to positive ones. Some of the criteria are so obvious that they are often left unacknowledged; nonetheless, it is useful to make them explicit.

Consistency: We require a good theory to be logically consistent. Formally, the set of statements constituting the theory must be capable of being simultaneously true. An obvious reason for requiring consistency is that anything follows from an inconsistent set of statements (ex falso quodlibet). Thus an inconsistent set, such as one containing

both “p” and “not p”, is of little use, whether for explanatory, predictive, evaluative, or prescriptive purposes. By entailing everything, it is too indiscriminate.9

Deductive closure: We require a good theory to be deductively closed. This means that any statement that is logically entailed by the theory also belongs to the theory.10

The idea underlying deductive closure is that we want to be able to identify a theory

with everything to which the theory is logically committed. If the theory asserts “p” and “if p then q”, for example, then it should also assert “q”. We would consider a theory defective if it were committed to the first two statements, but not to the third. The way we characterized Newton’s and Rawls’s theories had deductive closure built into it, since we characterized each as the set of all statements that are logically entailed by the relevant principles. Deductive closure is easy to achieve even when a theory is initially given in a form that violates it: we can re-define the theory as the set

of all statements entailed by the original, non-deductively-closed formulation.11

Axiomatizability: This requirement is implicit in the final clause of our definition of a theory, which says that a theory should ideally be representable as the set of all implications of some underlying principles. Formally, a theory is axiomatizable if there exists a finite set of principles such that the entire theory can be expressed as

their body of implications.12 It should be evident from our earlier discussion that

Newton’s and Rawls’s theories are usually presented in axiomatized form, as the bodies of implications of Newton’s and Rawls’s principles, respectively, perhaps together with some auxiliary assumptions. An axiomatizable theory can be presented in an informative manner, simply by specifying the set of principles from which it can be derived. If the theory could only be presented by brute enumeration of all its implications – typically infinitely many – there would be no succinct way of summarizing its content. A good theory illuminates its subject matter by giving us a manageable set of principles – a manageable theory formulation – that encodes the theory’s entire propositional content.

Parsimony: We require a good theory to avoid any unnecessary complexity, and to be as simple as possible, in an appropriate sense of simplicity (on simplicity, see, e.g., Baker  2013).13 Scientists  commonly  care  about  parsimony,  often  under  the  label “Occam’s razor”, and Rawls, for instance, also emphasizes simplicity as a virtue of a normative theory.14 What counts as “simple” may be different from context to context

and may also depend on what the theory is about. We usually want to find the simplest theory able to account for its subject matter. To be illuminating, the theory ought to be simpler, for example, than the target phenomenon it seeks to account for, as well as simpler than its rival theories. A scientific theory, for instance, should be simpler and more succinct than an enumeration of all the empirical facts it seeks to explain; otherwise it cannot play any explanatory role. Likewise, a good normative theory should be simpler than an enumeration of all case-specific normative judgments. The relevant bar of simplicity may be adjusted depending on the theory’s subject matter.

The present list of criteria for the internal assessment of theories is only illustrative, not exhaustive, but given space constraints, we now move on to external criteria.

4.2 External criteria

We defined a theory as a set of statements that is a candidate for playing some theoretical  or  practical  role.  Implicit  in  this  definition  is  the  idea  that  there  is something the theory is about: any theory is intended to represent, summarize, or capture something “outside the theory”. It may capture this correctly, in which case the theory is true, correct, or externally valid, or it may fail to do so, in which case it is false, incorrect, or externally invalid.

What exactly a given theory is intended to represent needs to be spelt out further. In the case of a physical theory, the answer is relatively straightforward, especially if we accept “scientific realism”: it is intended to represent certain physical facts about the world, such as facts about how physical objects behave in response to each other (see, e.g., Chakravartty 2013). In the case of a normative or evaluative theory, the picture is more complicated. If we are realists about normative or evaluative matters, we may say that the theory is intended to represent some theory-independent moral facts. If we are not realists about normative or evaluative matters, it is harder to specify what a normative or evaluative theory is intended to represent (on moral realism and anti- realism see, respectively, Sayre-McCord 2011; Joyce 2009).

Yet the very idea of a theory breaks down unless we assume that there is something potentially representable by it, however observer-dependent or socially constructed it might be.15 If we were nihilists, to take an extreme example of the denial of any normative or evaluative facts, we would not be able to engage in normative or evaluative theorizing in earnest.

Thus, in this section, we assume that normative or evaluative theories are truth-apt: it makes sense to ask whether they are true – or, some might prefer to say: correct or externally valid. We thus accept a form of “cognitivism” about such theories. This

assumption is still compatible with a variety of views about the “meta-ethical” status of normative or evaluative judgments. We return to some of these issues in Section 6. Granting, then, that there is some standard of correctness by which we can assess normative or evaluative theories (an ontological assumption), we still need to know how to do this assessment (an epistemological question). We now review several methods of testing a theory for external validity.

Taking intuitive judgments as strict evidence: According to this method, the test for a normative or evaluative theory is whether it fits our intuitive judgments about the relevant normative or evaluative matters. On this approach, our intuitive judgments have the same status as empirical observations in science. In science, a theory is empirically adequate if it entails the correct observation statements (see, e.g., Quine

1975; van Fraassen 1980). Similarly, in moral and political theory, we might call a theory normatively or evaluatively adequate if it entails the correct normative or evaluative statements. According to the strict-evidence method, these are precisely the normative or evaluative statements supported by our intuitive judgments (cf. the discussion in Dworkin 1975). Although simple and analogous to familiar scientific methods, this method has some problems. First, while we may be confident in some of our normative or evaluative judgments, other judgments may be more tentative, and in some cases – especially when the issue is less familiar – we may not have any firm intuitions at all. Second, our intuitive judgments may be subject to biases and framing  effects,  which  may  cast  further  doubt  on  their  reliability.  Third,  our normative or evaluative judgments may not be consistent with one another, or they may entail other judgments that we reject on reflection; in such cases, the strict- evidence method provides no guidance at all.

Reflective equilibrium: A more refined method of theory testing is the reflective- equilibrium method. It does not treat our intuitions as independent evidence, prior to the theory in question, but requires us to reach a “mutual fit” between the theory and our considered judgments. This works as follows. We begin with some initial theory, perhaps inspired by our initial intuitive judgments or given by some prima facie principles, then consider the implications of the theory and ask whether they are also in line with our judgments. If those implications fit our judgments, the process stops. It is more likely, however, that only some of the theory’s implications fit our judgments, while others do not. We then reassess both the theory and our judgments. In some cases, we may decide, on reflection, to revise the theory by changing some of the constituent principles, so as to bring the theory in line with the judgments we are unwilling to give up. In other cases, we may decide to overrule our judgments and embrace the theory’s implications as our new considered judgments. A reflective equilibrium is reached when the implications of our possibly revised theory are in line with our possibly revised judgments. At least since Rawls’s Theory of Justice, reflective equilibrium has been one of the most widely used methods in political theory (Rawls 1999, 15–18, 40–46; Daniels 2013). Its details can be spelt out in a variety of ways. For example, we may choose the units of assessment more narrowly or  more  broadly.  We  can  either  search  for  a  reflective  equilibrium  involving  a narrowly specified theory, constituted by a small number of principles together with very  few  auxiliary  assumptions,  or  we  can  search  for  a  reflective  equilibrium involving  a  more  broadly  specified  theory,  constituted  by  a  larger  number  of principles and further additional assumptions – which in turn may be interpreted as a conjunction of multiple theories, covering a wider domain of issues. Political theorists

sometimes speak of “narrow” reflective equilibrium in the first case, and “wide” reflective equilibrium in the second (Daniels 2013). Similarly, we may take different views on which kinds of judgments – especially whose judgments – should serve as input to this method; more on this below. Although the reflective-equilibrium method is consistent with the idea that we arrive at our normative or evaluative theories through careful deliberation, we may be worried about the possible arbitrariness of its outcome, since there may not always be a unique equilibrium. In some cases, we may not be able to reach any equilibrium at all (the non-existence problem), such as when we theorize about genuine moral dilemmas and vacillate between different theories that each fit only some of our judgments while conflicting with others. In other cases, there may exist more than one equilibrium, in that we can arrive at different “packages” of revised theories and judgments that each have the required “mutual fit” (the non-uniqueness problem). Arbitrary factors such as framing effects or the order in which we consider different implications of the theory may then affect which reflective equilibrium we end up with (the path-dependence problem).

Thought experiments and intuition pumps: Whether we opt for the strict-evidence method or the reflective-equilibrium method, we may sometimes wish to sharpen or clarify our intuitions or judgments. Thought experiments and real-world cases can serve as useful “intuition pumps” (Dennett 2013; Brownlee and Stemplowska forthcoming). Here, we consider some hypothetical or actual scenario that prompts strong normative or evaluative judgments. In the much-discussed “trolley problems”, for example, we are asked to judge what actions, if any, would be permitted to prevent a run-away trolley from crashing into, and killing, a larger group of people, at the expense of leading it to crash into, and kill, a smaller group (Thomson 1985). We then use these judgments to test our relevant normative or evaluative theories, following either the strict-evidence method or the reflective-equilibrium method. The usefulness of intuition pumps, especially ones involving highly idealised, counterfactual scenarios has recently been the object of considerable controversy in political theory (Elster 2011). We return to this issue in Section 5.

The relevant judgments: Both the strict-evidence method and the reflective- equilibrium method raise the question of which kinds of judgments, and whose judgments, to use in testing our theories. Should we test our theories on the basis of relatively spontaneous judgments or on the basis of suitably “filtered” judgments, and how  should  that  filtering  take  place  (Rawls  1999,  42)?  And  should  we  use  the political theorist’s judgments (which might be affected by his or her ideological views) or society’s (which might similarly be affected by biases), and in the latter case, which society should we focus on (Miller 1992; Walzer 1983)? For example, while  the  Rawls  of  A Theory of Justice  arguably  followed  the  former  approach (relying  on  the  political  theorist’s  judgments),  the  Rawls  of  Political Liberalism subscribed to the latter (looking at society) (Rawls 1971/1999; Rawls 1996). Specifically, the later Rawls re-interprets his theory of justice as an articulation of the ideas implicit in the public culture of liberal democratic societies. The building blocks of his account of justice are explicitly “drawn” from, and supposed to be widely acknowledged within, the society for which that account is designed. Relatedly, political theorists disagree about whether the judgments to which they appeal in theory testing should “fit” the particular practice the theory is meant to regulate. If the answer to this question is positive, then the exercise of theory construction is best seen as an attempt to offer what Ronald Dworkin calls a “constructive interpretation”

of  existing  political  practices  (Dworkin  1986,  chap.  2;  James  2005;  Sangiovanni

2008). If the answer is negative, then the exercise of theory construction is best

understood as an attempt to “discover” particular normative and evaluative truths, independently of existing social practices (on interpretation versus invention, see Walzer 1987). To illustrate the difference between the two approaches – at least on one  reading  of  what  sets  them  apart  –  consider  the  following  judgment,  made explicitly in relation to the practice of camping with friends: “we should institute a regime of shared ownership” (see G. A. Cohen 2009). On a practice- dependent/interpretive approach, this judgment should count as relevant evidence only in the construction of a “normative theory of camping”, not in the construction of normative theories of other practices, such as socio-political relations within the state. For proponents of practice-independence, by contrast, all normative or evaluative judgments, including the one in question, have cross-contextual validity in the identification of what justice or other moral values demand (this point is made in

Ronzoni 2012; see also Miller 2002).16

The applied-moral-philosophy method:  A  final  method  of  justifying  a  theory  in political theory is to show that it can be derived from some independently accepted moral principles or theory. A committed utilitarian or Kantian, for example, may regard a normative or evaluative theory as justified if and only if it can be derived from utilitarian or Kantian principles, which are treated as independently given. In this vein, Nozick says: “Moral philosophy sets the background for, and boundaries of, political philosophy. What persons may and may not do to one another limits what they may do through the apparatus of the state, or do to establish such an apparatus” (Nozick 1974, 6; see also Otsuka 2003, 3). Those theorists who view political theory as a subfield of moral philosophy will find this method appropriate. (Works in which political theory is conducted – at least to some extent – as applied moral philosophy include Singer 1972; G. A. Cohen 2008; Fabre 2012.) By contrast, if we consider the activity of political theory to take place against the background of (reasonable) pluralism about moral matters, the applied-moral-philosophy method is problematic, since it relies on the acceptance of a specific moral theory (for a critique of the “applied moral philosophy” approach, see Williams 2005; see also Galston 2010).

5. Abstraction and idealization in political theory

In political theory, as well as in other disciplines, theories are often abstract and/or idealised in certain respects. In this section, we explain what this means and discuss some methodological issues raised by abstraction and idealization.

5.1 Defining abstraction and idealization

Broadly following Onora O’Neill (1996, chap. 2), we say that a theory is abstract with respect to an issue – represented by a set of statements – if it is silent on that issue; formally, it has no implications at all for the given statements, implying neither any of these statements nor any of their negations. Newton’s theory of physics, for example, is abstract with respect to the colours of the physical bodies whose motion it represents. A theory is idealized with respect to an issue – again represented by a set

of statements – if it entails some (simplifying or limiting) falsehood about that issue; formally, it has a false implication for some of the given statements, implying one or more false statements among them or the negations of one or more true statements. A simple Newtonian theory of mechanics, for instance, may be idealized with respect to friction,  entailing  the  absence  of  friction  in  the  physical  systems  it  represents, although friction is present in the real world.

Since theories are meant to simplify the world, most political theorists agree that abstraction is virtually unavoidable in theory construction, and an innocuous intellectual exercise (O’Neill 1996, chap. 2; see also the discussion in Stemplowska

2008). Idealization, unlike abstraction, is looked at with greater suspicion in political theory and is considered potentially problematic (O’Neill 1996, chap. 2).

5.2 The worry about idealization

The “danger of idealization” in political theory has been discussed primarily in the debate on “ideal versus non-ideal theory” (for an overview, see Valentini 2012). The debate is largely animated by the worry that resort to simplifying assumptions and idealized thought experiments or intuition pumps – which are common in contemporary political theory – will adversely affect the validity of the ensuing theories (see, e.g., O’Neill 1996, chap. 2; Farrelly 2007; Mills 2005). For instance, John Rawls develops his theory of justice assuming full compliance with the resulting principles. This makes critics wonder whether we can trust Rawls’s principles to deliver correct and action-guiding prescriptions for the real world, where many people fail to abide by the demands of justice.

In assessing this worry, two points are worth making. First, like abstraction, some degree of idealization in our theories – in the form of simplifying assumptions – may play an important, and justified, heuristic role. Again, Newton’s theory of physics does not appear to be significantly undermined by its assuming friction away, especially to the extent that information about friction can potentially be re-introduced in refinements or applications of the theory. This suggests that, rather than reject idealizations as problematic from the outset, we must ask whether a theory contains the “right” idealizations, given its purpose (Robeyns 2008; Valentini 2009).

Second, to answer the question of which idealizations are “right” and which are not, we need to distinguish between three possible loci – or levels – at which idealizations can occur: (1) the theory itself, (2) the conditions of application of the theory’s prescriptions (where those prescriptions are of the form “if such-and-such conditions hold, then such-and-such follows”), (3) the justification of the theory. Crucially, idealization at any one of these levels need not entail idealization at any of the others.

5.3 Rawls’s theory as an example

Rawls’s theory of justice, as described earlier, consists of the “equal liberty” (X), “fair equality of opportunity” (Y), and “difference” (Z) principles and their implications. The theory itself would be idealized – a level (1) idealization – if and only if these principles entailed false statements about their subject matter – for instance, if the statement “we ought to re-arrange the tax system so as to benefit the worst off as much as possible” were false. Note, however, that the often-criticized “idealized”

assumption of full compliance is not made within the theory itself, but occurs as part of the justification that Rawls offers for the theory. It is one of the assumptions made by the parties in the original-position thought experiment. So, it is an idealization at level (3). Rawls’s principles, which generate his theory, do not imply any false claims about full compliance; hence we have no idealization at level (1) here. Nor do the prescriptions following from these principles presuppose full compliance for their applicability; hence we have no idealization at level (2) either (see Simmons 2010, 9–

10 for discussion).

Similarly, consider Rawls’s assumption that society exists under favourable historical and social conditions. Relative to existing war-torn, or desperately poor countries, this assumption is clearly false: it is an idealization. Rawls says explicitly that his principles of justice may not apply to societies in which the relevant favourable conditions are absent (Rawls 1999, 216). Does this make Rawls’s theory idealized – an idealization at level (1) – and problematically so? Arguably, it does not, because the favourable-conditions idealization operates at the level of the conditions of application of the theory’s prescriptions, i.e., at level (2). The prescription that is entailed  by  the  theory  (under  a  careful  formulation)  –  namely  “if  favourable conditions hold, justice demands X, Y, and Z” – is still true. Although the conditional nature of this prescription limits the scope of application of the theory, it does not make the theory itself idealized, by generating false prescriptions.

5.4 The worry reassessed

We suspect that, although worries about idealization in political theory are frequently expressed as complaints about “theories being idealized”, they actually tend to target idealizations at levels (2) and (3), rather than (1) – i.e., at the levels of the theories’ conditions of application and the justifications offered for those theories, not at the level of the theories themselves.

Of course, idealizations at levels (2) and (3) often make the target-theories somewhat irrelevant to the real world, by rendering them insufficiently action-guiding in real- world circumstances. While this lack of guidance may be a genuine shortcoming, it is not  accurately  captured  by  the  claim  that  the  theories  themselves  are  idealized. Rather, in the case of a level (2) idealization, the theories entail true prescriptions of an “if-then” sort, whose antecedent conditions – the “if” clauses – do not hold in real- world circumstances. And in the case of a level (3) idealization, the theories may be insufficiently justified, in that the justifications offered for them – such as highly contrived thought experiments – are too idealized to allow inferences for the real world.

Our discussion highlights the importance of clarity about the role that idealizations play in the defence and formulation of one’s theory. Consider a theory of justice prescribing “p”, defended on the assumption that there is no reasonable disagreement about justice within society, and yet formulated in universal terms: “justice always demands that p”. This theory does indeed run the risk of being problematically idealized, if it turns out that the presence of reasonable disagreement makes a morally relevant difference to what justice demands. The theory, in that case, would have false implications   in   a   number   of   situations,   namely   those   involving   reasonable

disagreement. A true principle, by contrast, would only say: “if there is no reasonable disagreement, justice demands that p”.

6. The significance of disagreement in political theory

We noted earlier that, while political theory is sometimes viewed as a subfield of moral philosophy, another view is that political theory is distinct from moral philosophy in that the conditions of theorizing are different (see, e.g., Williams 2005; Larmore 2013). On this view, political theory, unlike – or much more than – moral philosophy, is conducted against the background of (reasonable) pluralism in society. Disagreement about normative and evaluative matters, it is said, has a different status in political theory than in (mainstream) moral philosophy. The nature of disagreement in society constrains what normative or evaluative principles can be defended in political theory. In this section, we explore this view and discuss its methodological implications.

6.1 Disagreement: “constitutive of assertability conditions” rather than “epistemic”

While moral philosophers seek to answer questions such as “what ought we to do in a given situation”, political theorists are faced with questions such as “what ought we do, given that we do not agree about what we ought to do” (see, e.g., Waldron 1999; Larmore 2013). At first sight, we might be perplexed by this alleged difference between political theory and moral philosophy. Are there not equally big disagreements in moral philosophy? Practically every well-known moral dilemma has the property that different people take different views on how it should be resolved, and the question of what one should do in such cases of disagreement arises also in the moral realm.

Yet, the difference between moral philosophy and political theory may be said to be the following. In moral philosophy, we commonly (though not universally) make the assumption that, among the many different rival normative or evaluative theories, one is the independently correct or true theory. The task for the moral philosopher is to identify that theory. If we conduct moral philosophy on this assumption, disagreement is of a “merely” epistemic kind. There is a fact about what the right answer to any normative or evaluative question is; we may just have different beliefs about that fact.

It is less clear – so the argument goes – whether the same assumption can be made in political theory. On this picture, disagreement in political theory may be viewed, not merely as “epistemic”, reflecting different beliefs about the same truth, but as partly “constitutive” of the correctness conditions – or warranted assertability conditions – of normative claims themselves. On this view, whether a normative or evaluative theory in political theory is correct (or assertable) depends, in part, on the society in which the theory is to be applied, and specifically on the level of reasonable disagreement in that society; we define the notion of reasonableness below.

A key desideratum here is that our theories be, at least in principle, acceptable to individuals holding reasonable but conflicting moral views within pluralistic societies. What counts as a correct normative or evaluative principle in a society with little reasonable disagreement need not always count as correct in a highly pluralistic society, where the range of reasonable disagreement is greater. In light of this, some

theorists avoid using the notion of truth in political theory altogether, and prefer to replace it with other, less loaded notions, like reasonableness or reasonable acceptability (Rawls 1996; cf. J. Cohen 2009).

But what might justify the shift from the “epistemic” to the “constitutive” status of disagreement in the political domain? At least two answers are possible. First, the shift might be morally justified. On this view, a commitment to respect for persons places a burden of “reasonable acceptability” on the principles put forward by the political theorist, insofar as these principles may be permissibly enforced against individuals, for instance through state action. The idea is that political theory focuses on  the  development  of  “enforceable  rules”,  and  these  rules  are  normatively appropriate – meet the relevant criterion of correctness – only if they are acceptable to reasonable individuals to whom they apply.

Second, reasonable disagreement may be taken to constrain correctness or warranted assertability in political theory for pragmatic reasons. After all, most of the principles put forward by political theorists are meant to help regulate social life in complex and highly pluralistic societies. A normative or evaluative theory that appeals only to people of a particular moral persuasion would be of little use in this respect; it could not give rise to a durable and stable social order except through autocratic imposition. And so again, acceptability to individuals with competing reasonable views may be deemed a criterion of correctness of the theory.

The inspiration for many of the foregoing reflections can be found in John Rawls’s second  major  book,  Political  Liberalism  (1993/1996).  We  now  offer  a  brief discussion of Rawls’s treatment of the relevance of pluralism to political theorizing.

6.2 The distinction between “political” and “comprehensive” theories

One of the key innovations of Political Liberalism, compared to A Theory of Justice, is Rawls’s insistence on his theory being “political not metaphysical” (Rawls 1985). This  means  two  things.  First,  the  theory  does  not  adjudicate  all  aspects  of interpersonal conduct (as a “comprehensive” moral theory might do) but concerns only the “public” realm (most notably, the basic structure of society). Second, the theory is grounded not in any “comprehensive” moral theory (“doctrine”), on which there is likely to be deep disagreement in any pluralistic liberal society, but in ideals that are drawn from the public culture of such a society. Thus a “political” theory is typically (i) restricted to a smaller domain of issues and (ii) less morally and metaphysically “loaded” (thereby “thinner”) than a “comprehensive” theory. In particular, it refrains from taking a stand on issues that are too controversial.

For example, a theory of justice that adjudicates all aspects of personal, and not just political and social, life would be “comprehensive”, as would be a theory based on the metaphysically loaded premise that human beings are morally equal because they are created “in the image of God”. Both theories would be the object of reasonable disagreement within a pluralistic society. By contrast, a theory of justice that focuses on the basic structure of society and is based on a commitment to “citizens’ freedom and equality” would count as “political”, since its domain does not “over-reach” and its premises are entrenched in the public culture of liberal democracies and arguably

shared by individuals holding competing yet reasonable comprehensive doctrines. But what does the notion of “reasonableness” stand for?

6.3 The notion of “reasonable disagreement”

The notion of “reasonableness” is hard to pin down in general, and this is true in Rawls’s case as well (for discussion, see Gaus 1996, 131–132; Gaus 1999). In particular, the notion may be interpreted in epistemological and/or moral ways. Under an epistemological interpretation, something is reasonable if it is consistent with a proper use of reason in light of the evidence available. Rawls believes that, because of what he calls the “burdens of judgment”, we should not expect agreement between the views of different people, even when they are all developed by consistently applying the powers of reason (Rawls 1996, Lecture 2, sec. 2). Under a moral interpretation, reasonableness refers to a view’s compatibility with certain fundamental normative requirements, such as respect for citizens as free and equal, or a commitment to mutual justification (Rawls 1996, Lecture 2, sec. 3). This would make views that reject such a commitment unreasonable.

Both interpretations of “reasonableness” have virtues and vices, especially in the context of what we earlier called the “moral” justification for deeming acceptability in the context of reasonable disagreement an assertability condition for theories within political theory. As far as the epistemological interpretation of reasonableness is concerned, on the positive side, it offers a compelling rationale for treating only some contested  views  as  “worthy  of  respect”,  namely  those  with  a  given  epistemic pedigree.  On  the  negative  side,  this  interpretation  potentially  allows  morally repugnant views to count as reasonable, assuming they involve no breach of reason, purely epistemically understood. If a necessary condition for the overall acceptability of a normative or evaluative theory is that it is acceptable to all individuals with reasonable  views,  an  acceptable  theory  must  then  potentially  appeal  even  to individuals with morally repugnant views, and this may be too much to ask.

As far as the moral interpretation of reasonableness is concerned, on the positive side, it allows us to “filter out” morally repugnant views. But, on the negative side, it does so at the cost of referring to substantive values that themselves require justification. The suspicion is that, on a moralized interpretation, the “reasonable” simply corresponds to the moral commitments that the liberal theorist considers non- negotiable: it is ad hoc (see Mouffe 2005). Again, we do not take a stand on this, but note that these difficulties threaten the plausibility of defining the overall acceptability of a normative or evaluative theory simply in terms of its acceptability to all individuals with reasonable views; identifying those reasonable views may itself rely on moral premises.

6.4 The quest for an “overlapping consensus”

Finally, turning to the “pragmatic” justification for taking disagreement to have a “constitutive” rather than purely “epistemic” status in political theory, we conclude with some remarks on Rawls’s notion of an “overlapping consensus”. An overlapping consensus on a normative or evaluative theory in a social or political domain occurs when this theory is endorsed from the perspective of different reasonable comprehensive doctrines held in society. A normative theory that cannot be endorsed

from different such perspectives could not hope to gain sufficient support and to offer a stable basis for social organization in pluralistic liberal democracies (Rawls 1996, Lecture 4). The notion of an overlapping consensus operationalizes the idea that the criterion of correctness of a “political” theory is its acceptability to individuals who hold competing but reasonable comprehensive moral theories.

7. Concluding remarks

We have reviewed the methodology of analytic political theory from what we hope is a somewhat novel and helpful angle. By drawing on ideas from the philosophy of science, we have attempted to highlight the ways in which theorizing in political theory relates to theorizing in other areas of philosophy and positive science. We have also reviewed some recent debates and controversies within the political-theory literature,  which  has  only  recently  given  greater  attention  to  methodological questions. Our hope is that this article will prove to be a clarifying contribution to the growing methodological debate in political theory.

References

Arrow, Kenneth Joseph. 1951. Social Choice and Individual Values. New York: John

Wiley & Sons.

Baehr, Amy R. 2013. “Liberal Feminism.” In The Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta, Winter 2013. http://plato.stanford.edu/archives/win2013/entries/feminism-liberal/.

Baker, Alan. 2013. “Simplicity.” In The Stanford Encyclopedia of Philosophy, edited

by                Edward                N.                Zalta,                Fall                2013. http://plato.stanford.edu/archives/fall2013/entries/simplicity/.

Blackorby, Charles, Walter Bossert, and David J. Donaldson. 2005. Population Issues in  Social  Choice  Theory,  Welfare  Economics,  and  Ethics.  New  York:

Cambridge University Press.

Bratman,  Michael.  1999.  Faces  of  Intention:  Selected  Essays  on  Intention  and

Agency. Cambridge: Cambridge University Press.

Brownlee,  Kimberley,  and  Zofia  Stemplowska.  forthcoming.  “Trapped  in  an

Experience  Machine  with  a  Famous  Violinist:  Thought  Experiments  in

Normative Theory.” N/A

Carter, Ian. 2012. “Positive and Negative Liberty.” In The Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta, Spring 2012. http://plato.stanford.edu/archives/spr2012/entries/liberty-positive-negative/.

Chakravartty, Anjan. 2013. “Scientific Realism.” In The Stanford Encyclopedia of

Philosophy,      edited      by      Edward      N.      Zalta,      Summer      2013. http://plato.stanford.edu/archives/sum2013/entries/scientific-realism/.

Christiano,   Thomas.   2008.   “Democracy.”   In   The   Stanford   Encyclopedia   of

Philosophy,       edited       by       Edward       N.       Zalta,       Fall       2008.

http://plato.stanford.edu/archives/fall2008/entries/democracy/.

Cohen, G. A. 1988. History, Labour, and Freedom: Themes from Marx. Oxford:

Clarendon Press.

———.  1995.  Self-Ownership,  Freedom,  and  Equality.  Cambridge:  Cambridge

University Press.

———. 2008. Rescuing Justice and Equality. Cambridge, MA: Harvard University

Press.

———. 2009. Why Not Socialism? Princeton, N.J.: Princeton University Press.

Cohen, Joshua. 2009. “Truth and Public Reason.” Philosophy & Public Affairs 37 (1):

2–42.

Daniels, Norman. 2013. “Reflective Equilibrium.” In The Stanford Encyclopedia of

Philosophy,      edited      by      Edward      N.      Zalta,      Winter      2013. http://plato.stanford.edu/archives/win2013/entries/reflective-equilibrium/.

Dennett, Daniel C. 2013. Intuition Pumps And Other Tools for Thinking. New York: W. W. Norton.

Dworkin, Ronald. 1975. “The Original Position.” In Reading Rawls: Critical Studies on Rawls’s A Theory of Justice, edited by Norman Daniels, 16–52. Stanford,

CA: Stanford University Press.

———. 1986. Law’s Empire. Cambridge, MA: Harvard University Press.

Elster, Jakob. 2011. “How Outlandish Can Imaginary Cases Be?” Journal of Applied

Philosophy 28 (3): 241–258.

Fabre, Cécile. 2012. Cosmopolitan War. Oxford: Oxford University Press.

Farrelly, Colin. 2007. “Justice in Ideal Theory: A Refutation.” Political Studies 55

(4): 844–864.

Gallie, W. B. 1955. “Essentially Contested Concepts.” Proceedings of the Aristotelian

Society 56 (1): 167–198.

Galston,  William  A.  2010.  “Realism  in  Political  Theory.”  European  Journal  of

Political Theory 9 (4): 385–411.

Gärdenfors, Peter. 2000. Conceptual Spaces: The Geometry of Thought. Cambridge,

MA: MIT Press.

Gaus,  Gerald  F.  1996.  Justificatory  Liberalism  :  An  Essay  on  Epistemology  and

Political Theory. Oxford: Oxford University Press.

———.  1999.  “Reasonable  Pluralism  and  the  Domain  of  the  Political:  How  the

Weaknesses of John Rawls’s Political Liberalism Can Be Overcome by a

Justificatory Liberalism.” Inquiry 42 (2): 259–284.

Gerring, John. 1999. “What Makes a Concept Good? A Criterian Framework for

Understanding Concept Formation in the Social Sciences.” Polity 31 (3): 357–

393.

Gilabert, Pablo, and Holly Lawford-Smith. 2012. “Political Feasibility: A Conceptual

Exploration.” Political Studies 60 (4): 809–825.

Gilbert, Margaret. 1989. On Social Facts. Princeton, NJ: Princeton University Press.

Goodin, Robert E. 2009. The Oxford Handbook of Political Science. Oxford: Oxford

University Press.

Harsanyi, John C. 1955. “Cardinal Welfare, Individualistic Ethics, and Interpersonal

Comparisons of Utility.” Journal of Political Economy 63 (4): 309–321.

James, Aaron. 2005. “Constructing Justice for Existing Practice: Rawls and the Status

Quo.” Philosophy & Public Affairs 33 (3): 281–316.

Joyce, Richard. 2009. “Moral Anti-Realism.” In The Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta, Summer 2009. http://plato.stanford.edu/archives/sum2009/entries/moral-anti-realism/.

Larmore,   Charles.   2013.   “What   Is   Political   Philosophy?”   Journal  of  Moral

Philosophy 10 (3): 276–306.

Leopold, David, and Marc Stears. 2008. Political Theory: Methods and Approaches.

Oxford: Oxford University Press.

Linz,  Juan  J.,  and  Alfred  Stepan.  1996.  Problems of Democratic Transition and

Consolidation:   Southern   Europe,   South   America,   and   Post-Communist

Europe. Baltimore: Johns Hopkins University Press.

List,  Christian.  2011.  “The  Logical  Space  of  Democracy.”  Philosophy & Public

Affairs 39 (3): 262–297.

———. 2013. “Social Choice Theory.” In The Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta, Winter 2013. http://plato.stanford.edu/archives/win2013/entries/social-choice/.

List, Christian, and Philip Pettit. 2011. Group Agency: The Possibility, Design, and

Status of Corporate Agents. Oxford: Oxford University Press.

Lovett, Frank. 2013. “Republicanism.” In The Stanford Encyclopedia of Philosophy,

edited           by           Edward           N.           Zalta,           Spring           2013. http://plato.stanford.edu/archives/spr2013/entries/republicanism/.

MacCallum, Gerald C. 1967. “Negative and Positive Freedom.” The Philosophical

Review 76 (3): 312–334.

Margolis, Eric, and Stephen Laurence. 2012. “Concepts.” In The Stanford Encyclopedia of Philosophy, edited by Edward N. Zalta, Fall 2012. http://plato.stanford.edu/archives/fall2012/entries/concepts/.

McDermott,  Daniel.  2008.  “Analytic  Political  Philosophy.”  In  Political  Theory:

Methods and Approaches, edited by David Leopold and Marc Stears, 11–28. Oxford: Oxford University Press.

Miller, David. 1992. “Distributive Justice: What the People Think.” Ethics 102 (3):

555–593.

———.  2002.  “Two  Ways  to  Think  About  Justice.”  Politics,  Philosophy  & Economics 1 (1): 5–28.

Mills, Charles W. 2005. “‘Ideal Theory’ as Ideology.” Hypatia 20 (3): 165–183. Mouffe, Chantal. 2005. “The Limits of John Rawls’s Pluralism.” Politics, Philosophy

& Economics 4 (2): 221–231.

Nozick, Robert. 1974. Anarchy, State, and Utopia. New York: Basic Books.

O’Neill,  Onora.  1996.  Towards  Justice  and  Virtue:  A  Constructive  Account  of

Practical Reasoning. Cambridge: Cambridge University Press.

Okasha,  Samir.  2002.  Philosophy of Science: A Very Short Introduction.  Oxford: Oxford University Press.

Oppenheim,  Felix  E.  1981.  Political  Concepts:  A  Reconstruction.  Oxford:  Basil

Blackwell.

Otsuka,   Michael.   2003.   Libertarianism   Without   Inequality.   Oxford:   Oxford

University Press.

Parfit, Derek. 1984. Reasons and Persons. Oxford: Clarendon Press.

Pettit,  Philip.  1993.  The  Common  Mind:  An  Essay  on  Psychology,  Society,  and

Politics. New York: Oxford University Press.

———.  1997.  Republicanism.  A  Theory  of  Freedom  and  Government.  Oxford:

Oxford University Press.

———. 2011. “The Instability of Freedom as Noninterference: The Case of Isaiah

Berlin.” Ethics 121 (4): 693–716.

Quine, W. V. 1975. “On Empirically Equivalent Systems of the World.” Erkenntnis 9

(3): 313–328.

Rawls, John. 1985. “Justice as Fairness: Political Not Metaphysical.” Philosophy and

Public Affairs 14 (3): 223–251.

———. 1996. Political Liberalism. New York: Columbia University Press.

———. 1999. A Theory of Justice. Oxford: Oxford University Press.

Robeyns, Ingrid. 2008. “Ideal Theory in Theory and Practice.” Social Theory and

Practice 34 (3): 341–362.

Roemer, John E. 1998. Equality of Opportunity. Cambridge, MA: Harvard University

Press.

Ronzoni,  Miriam.  2012.  “Life  Is  Not  a  Camping  Trip  –  on  the  Desirability  of

Cohenite Socialism.” Politics, Philosophy & Economics 11 (2): 171–185.

Sangiovanni, Andrea. 2008. “Justice and the Priority of Politics to Morality.” Journal of Political Philosophy 16 (2): 137–164.

Sayre-McCord,  Geoff.  2011.  “Moral  Realism.”  In  The Stanford Encyclopedia of

Philosophy,      edited      by      Edward      N.      Zalta,      Summer      2011.

http://plato.stanford.edu/archives/sum2011/entries/moral-realism/. Searle, John R. 1995. The Construction of Social Reality. London: Penguin.

Sen, Amartya K. 1970. Collective Choice and Social Welfare. San Francisco: Holden- Day.

———.  1980.  “Equality  of  What?”  In  The Tanner Lectures on Human Values, Volume I, edited by Sterling M. McMurrin, 195–220. Cambridge: Cambridge

University Press.

———. 1987. The Standard of Living. Edited by Geoffrey Hawthorn. Cambridge:

Cambridge University Press.

Simmons, A. John. 2010. “Ideal and Nonideal Theory.” Philosophy & Public Affairs

38 (1): 5–36.

Singer,  Peter.  1972.  “Famine,  Affluence,  and  Morality.”  Philosophy  and  Public

Affairs 1 (3): 229–243.

Southwood, Nicholas. 2013. “Democracy as a Modally Demanding Value.” Noûs

Online early: n/a–n/a.

Stemplowska, Zofia. 2008. “What’s Ideal About Ideal Theory?” Social Theory and

Practice 34 (3): 319–340.

Thomson, Judith Jarvis. 1985. “The Trolley Problem.” The Yale Law Journal 94 (6):

1395–1415.

Tollefsen, Deborah Perron. 2002. “Collective Intentionality and the Social Sciences.”

Philosophy of the Social Sciences 32 (1): 25–50.

Tuomela, Raimo. 2007. The Philosophy of Sociality: The Shared Point of View. New

York: Oxford University Press.

Valentini,  Laura.  2009.  “On  the  Apparent  Paradox  of  Ideal  Theory.”  Journal of

Political Philosophy 17 (3): 332–355.

———.  2012.  “Ideal  Vs.  Non-ideal  Theory:  A  Conceptual  Map.”  Philosophy

Compass 7 (9): 654–664.

Van Fraassen, Bas C. 1980. The Scientific Image. Oxford: Oxford University Press.

Waldron, Jeremy. 1999. Law and Disagreement. Oxford: Oxford University Press. Walzer, Michael. 1983. Spheres of Justice: a Defense of Pluralism and Equality. New

York: Basic Books.

———.  1987.  Interpretation  and  Social  Criticism.  Cambridge,  MA:  Harvard

University Press.

Williams, Bernard Arthur Owen. 2005. In the Beginning Was the Deed: Realism and

Moralism in Political Argument. Princeton, N.J.: Princeton University Press.