NOTE-259: Wellbeing, Freedom and Social Justice: The Capability Approach Re-Examined

Contribution

Before this book, the capability approach was usually described as “an open, flexible and multi-purpose framework” (p. 25). The book replaces that description with an anatomy that says what is compulsory, what is compulsory but has open content, and what is optional. From that anatomy it draws a criterion for when a theory counts as a capability theory, which Robeyns presents as a defence against “inflation” (pp. 77–78, 211–212). It also gives a principled way to sort critiques: they are aimed either at the approach (A-module) or at particular choices (B- and C-modules). The book develops Robeyns’s 2016 “cartwheel” paper (Robeyns 2016b, p. 36) and is written as an introductory “helicopter view” and FAQ (pp. 4, 19).

Key insight

There is one capability approach and many capability theories (p. 30). The approach says only that interpersonal comparisons of advantage should be made in the space of what people are able to do and be (capabilities) and what they actually do and are (functionings). It adds a small set of structural claims (the A-module). Almost everything people argue about is a choice a particular theory makes, and each such choice needs its own justification: which dimensions, whose agency, which account of social structure, whether to count functionings or capabilities, which distributive rule, which addressee. Those choices are the B- and C-modules. “Many of the intellectual and ideological battles actually take place in arguing about the B-modules and the C-modules” (p. 215).

Assumptions

These are the premises of the reconstruction itself.

Method. The account is inductive: a generalisation from how the literature uses the approach. It is also critical: parts of that literature that “do not survive careful analysis” are rejected (p. 21 n. 1). The account is offered as a revisable proposal (“If someone has convincing arguments why one element or module should be deleted, modified, or added, then that should be done”, p. 75).
Scope. Human capabilities only. Non-human animals, and the “capabilities of firms”, are set aside (p. 9 n. 4).
“Advantage” is the neutral technical term for “those aspects of that person’s interests that matter”. Within the approach it has two specifications: achieved wellbeing and wellbeing freedom (pp. 25–26). Sen’s four-fold version adds agency achievement and agency freedom (Sen 1985c, per Robeyns). Robeyns treats it as optional, because whether agency freedom is always an end “is disputed” (p. 26 n. 10).
Minimal metaphysics. Functionings are “constitutive of human life”. Robeyns concedes this is a metaphysical claim but says it is a “very ‘minimal’” one (p. 39 and n. 20).
Typology of research. Research is conceptual, descriptive, explanatory, evaluative or prescriptive (pp. 27–28). The approach is used mostly for evaluative and prescriptive work, least for explanatory work.

Key results

The definition (ch. 2). The approach is “a conceptual framework, which is in most cases used as a normative framework for the evaluation and assessment of individual wellbeing and that of institutions” (p. 26). It “purports that freedom to achieve wellbeing is a matter of what people are able to do and to be” (p. 24). Robeyns’s definition is broader than Sen’s (Sen 2009a, 16: “…ability to do the different things a person has reason to value doing or being”, quoted p. 7). She treats Sen’s “having reason to value” clause as “a special case” of her general definition (p. 8 n. 1).

What is necessary: the A-module (pp. 38–59).

	proposition	what Robeyns says it commits a theory to
A1	functionings and capabilities are the core concepts	capabilities are “real freedoms or real opportunities” to achieve functionings; functionings are the realised beings and doings (pp. 38–39)
A2	they are value-neutral categories	”bad” functionings exist (violence, stress, the capability to rape, pp. 41–42). Defining functionings as what one “has reason to value” is “a mistake” that merges two separate normative moves (pp. 43–44)
A3	conversion factors	personal, social and environmental factors govern how resources become functionings (the bicycle example, pp. 45–47)
A4	means–ends distinction	the ultimate ends in interpersonal comparison are valuable capabilities; resources are means (pp. 47–51)
A5	functionings and/or capabilities form the evaluative space (the whole of it or part)	this is the first normative claim. Alkire’s formulation should be extended to “the weakening of those functionings that have a negative value” (pp. 51–52)
A6	there may be other dimensions of ultimate value	the approach is “incomplete as an account of the good”; examples are Sen’s process aspect of freedom, procedural fairness and basic liberties (pp. 53–55)
A7	value pluralism	two kinds. One is principle pluralism (Sen 1985c, 176; Sen 1982 “Rights and Agency”). The other is multidimensionality: a one-functioning capability theory (happiness only) would be “logically possible” but “makes no sense” (pp. 55–57)
A8	each person as an end	Nussbaum’s principle, which is ethical or normative individualism. It is “desirable” (needed to treat people as moral equals) and “unavoidable” (functionings belong to embodied, physically separate persons) (pp. 57–59)

What is compulsory with open content: the B-modules (pp. 59–67). B1 purpose (including scope and audience). B2 selection of dimensions: the method depends on purpose and on practical constraints, and it should never be “technocratic”. B3 an account of human diversity. B4 an account of agency, working definition Sen 1999a, 19; Nussbaum “explicitly refrains” from agency, but her “practical reason” capability supplies one (p. 64 n. 37). B5 an account of structural constraints. B6 functionings, capabilities or both. B7 meta-theoretical commitments.

What is optional or contingent: the C-modules (pp. 67–73). C1 additional ontological and explanatory theories. For example, whether the gendered division of labour is judged unjust depends on the explanatory theory of it (pp. 68–69). C2 weighting, which is “not required for each capability theory”, unlike selection (p. 69). C3 empirical methods. C4 additional normative principles, such as non-domination, non-discrimination, empowerment and political liberalism.

Hybrid theories (pp. 75–77). A theory that uses functionings and capabilities but violates an A-proposition is “hybrid”; Robeyns says the label is not a value judgement. The candidate cases are two:

Sen’s recent work on justice (Sen 2006, 2009c). This holds only if Baujard & Gilardone’s reading is right that Sen leaves the metric of justice to public reasoning and so declines A5.
Master-value theories, where capabilities only matter through one further value (Schokkaert’s life-satisfaction weights, Claassen’s “navigational agency”). These are hybrid if they breach A7, which Robeyns leaves “not sufficiently analysed” (pp. 71–72).

Sen versus Nussbaum (pp. 77–80, and passim). The modular view makes Nussbaum’s work “a capability theory … rather than another version of the approach”. Robeyns endorses Qizilbash (2013, 38): Nussbaum’s approach “emerges as one particular application or development of Sen’s original formulation”. She says there is “only one capability approach which is a generalisation of the work by Sen together with further developments by many others” (pp. 79–80). She names specific points where Nussbaum wrongly promotes her own choices to the core:

political liberalism, which belongs in C4 (p. 78 n. 40);
the priority of capabilities over functionings (p. 112 n. 8);
the government as the agent of change (§4.5);
the claim that there are only two modes of use (p. 35);
“the CA is a type of human rights approach” (p. 168 n. 31).

Sen, conversely, “would have benefited from a more systematic description” of the approach’s anatomy (p. 79).

What Sen specifically holds, per Robeyns (ch. 3, and pp. 94, 99–103, 155–156, 172–174). Everything here is Robeyns’s reading of Sen, not read in Sen himself:

Capability as a set. In Sen’s original terminology a person has one “capability”, a set of functioning vectors (Sen 1992a, 40). The plural “capabilities” as separate options is the later, Nussbaum-influenced usage, which Sen later uses “(perhaps reluctantly)” (pp. 91–92).
“Basic capability”. This was Sen’s 1979 first answer to “equality of what?”. Later “basic capability” came to mean a threshold for poverty (“the ability to satisfy certain elementary and crucially important functionings up to certain levels”, Sen 1992a, 45 n. 19) (p. 94).
Freedom. Sen first called capabilities “positive freedoms”, citing Berlin approvingly. Robeyns argues this sits badly with Berlin’s notion of self-mastery. Sen later redefined positive freedom as “the person’s ability to do the things in question taking everything into account” (Sen 2002a, 586). Sen also described capabilities as opportunities (Dewey Lectures 1985c, 201; Sen 1992a, 31). Robeyns’s own conclusion is that capabilities are Pettit-style option freedoms, broader than Taylor’s opportunity concept (pp. 99–105).
Opportunity versus process freedom. Capabilities capture the opportunity aspect of freedom but not its process aspect, so a rights theory needs process freedoms added through A6 (pp. 53, 164–166).
Principle pluralism was “all along” in Sen, and later readers lost it (pp. 55–56).
No fixed list. Sen is “(a) against one list that would apply to all capability theories and applications, and (b) … in favour of seriously considering procedural methods” (pp. 173–174). The reason given is his weight on agency, public reasoning and social choice (p. 173).
Distributive rule. Sen is “uncommitted to one single distributive rule”. “Equality of What?” was “merely asking” which equalisandum to use if one wants equality of something (pp. 155–156).
Justice. Sen is “averse to building a well-defined theory of justice” and criticises “transcendental” theories (pp. 152, 156–157, 173).
Individualism. Sen’s reply to Deneulin & Stewart was “I fear I do not see at all the basis of their diagnosis” (Sen 2002b, 80), and Robeyns sides with Sen (p. 184 n. 6). Sen’s “socially dependent individual capability” is distinguished from collective capability (p. 116).
Happiness is one important functioning, not the metric (Sen 2008, 26, pp. 135–136). Adaptive preferences are a reason to reject mental metrics (Sen 1985c, 21, p. 138).
Sen versus Robeyns on groups. Sen has “great faith in people’s abilities to be rational and to resist social and moral pressure stemming from groups” (Sen 1999b, 2009b), unlike Alkire, Nussbaum and Robeyns herself (p. 187).

The list debate (§4.3, pp. 171–174). The demand that Sen supply a list “misfires” when aimed at the approach, because no single list can serve all purposes. For particular theories Robeyns distinguishes a weak and a strong critique (from Robeyns 2005a). The strong critique is Nussbaum’s: there must be one well-defined list, or Sen’s approach has no bite for justice. Robeyns sides with Sen that selection belongs at the level of individual theories and must be sensitive to purpose. Dimensions can be selected on philosophical grounds (Nussbaum, Claassen), by procedure (Byskov), by bias-minimising criteria for empirical work (Robeyns 2003), or from the UDHR (Vizard, Burchardt). Robeyns’s verdict is that the debate “seems to be settled now” (p. 30 n. 15).

Individualism (§2.6.8, §4.6). Ethical or normative individualism holds that only individual persons are units of ultimate moral concern. Methodological individualism comprises semantic, ontological and explanatory individualism: “all social phenomena are to be explained wholly and exclusively in terms of individuals and their properties” (Bhargava 1992, 19, p. 184). The approach is committed to the first by A8. It is not committed to the second: B5 and C1 make room for, and usually supply, social ontology. As evidence she cites Sen’s “cooperative conflict” analysis and Drèze & Sen’s “social opportunity”. She reads the groups critique in two forms. The strong form, that the approach cannot attend to groups, is “obviously false”. The weak form, that the literature does not attend to groups enough, she partly concedes, recommending more dialogue with sociology, anthropology and history (pp. 186–188). She suggests replacing the term “normative individualism” with “the principle of each person as an end” (p. 59).

Verdicts on the other critiques (ch. 4).

“Not everything called a capability is one” (§4.2): upheld. Holland’s “meta-capability” of ecological functioning is a precondition, not a capability.
Needs (§4.4): partly upheld. The approach “does not have the resources” for the needs/wants distinction (non-contingent needs), and capability scholars “neglect to take the philosophy of needs seriously” (pp. 178–179).
Government as sole addressee (§4.5): rejected, against Nussbaum; the agent of change is a B/C choice.
Power (§4.7.1): the A-module “do[es] not necessarily have significant implications for the role of power”. Robeyns’s own view is that a rich account of power is needed (pp. 192–193).
Political economy (Pogge 2002, Jaggar) (§4.7.2): “not … a valid critique of the capability approach as a general framework”, only of research priorities (p. 194).
Liberalism (§4.8): Sen’s and Nussbaum’s theories are liberal in the philosophical sense, and neither is neoliberal. The approach itself is not necessarily liberal, and non-liberal capability theories are possible. Violating A8 makes a theory hybrid, not capabilitarian. Robeyns withdraws her earlier view that the approach is liberal as “premature” (p. 195 n. 13).
Human development (§4.9): human development is not the same as the capability approach. It is a capability theory with thick C1/C4 choices, better suited to challenging the Washington consensus (pp. 197–202).
Welfare economics (§4.10): a capabilitarian welfare economics is possible but harder. Sugden’s operationality objection is answered by existing applications and by incomplete or dominance rankings, but there may be a trade-off between informational richness and complete formal orderings (pp. 207–208).

Other ch. 3 claims.

Robustness, the probability that a capability can actually be exercised, should qualify capabilities (pp. 97–98).
The choice between functionings and capabilities depends on paternalism, responsibility-sensitivity, institutional reciprocity and interdependence (the vaccination case). The claim that a capability theory must focus on capabilities is “mistaken” (pp. 107–112).
Collective capabilities are “a subset of personal capabilities” that need group action (p. 117).
The approach contains several accounts of wellbeing: desire-fulfilment in first-person and institutional uses, objective-list in poverty measurement. “The often-heard view that the account of wellbeing given by the capability approach is an objective list theory doesn’t seem to be true” (p. 126).
Happiness is one functioning. Robeyns gives three objections to it as the metric: ontological (Nozick’s experience machine), adaptation and social comparison, and group comparisons (pp. 126–137).
Adaptive preferences are real but easy to over-diagnose (Khader’s “occupational hazards”). They threaten participatory selection of dimensions and capability-only metrics (pp. 137–142).
The approach is “not very suitable for ethical issues that only concern questions about the right”; abortion is the example (pp. 144–145).
A capability theory of justice must add several things: a justificatory basis, outcome versus opportunity, the selection of capabilities, a comparison with other metrics, a distributive rule, a line between individual and collective responsibility, and duty-bearers (pp. 153–157).
Capabilities are insufficient for a theory of human rights, because process freedoms must be added (pp. 164–168).

Ch. 5 (pp. 211–216). Many debates are “settled”, and theory-pluralism should be embraced. Robeyns proposes six directions:

push the approach to its limits in fields where it is immature, including book-length capability theories of justice and of welfare economics;
use it in explanatory and behavioural models;
use it as an interdisciplinary lingua franca;
build policy tools beyond GDP alternatives;
identify capability theories that are logically possible but empirically implausible, partly to avoid “hijacking” by powerful actors;
guide practice, as in social work.

She warns: “There is no point in pretending that the capability approach can do more than it is able to do” (p. 216).

Claims

id	claim	strength	support
C1	Every capability theory must satisfy A1–A8; the B-modules are compulsory with open content; the C-modules are contingent	moderate, as a definitional proposal	Argued proposition by proposition in §§2.6–2.8, with examples. It is stipulative and inductive, and explicitly revisable (p. 75). No test against a corpus of self-described capability work is reported.
C2	Functionings and capabilities must be defined value-neutrally, and defining them as what one “has reason to value” is a mistake	moderate	Informal argument (pp. 41–45): bad functionings, contested cases (care work, sex work, vaccination), and two separate normative moves merged into one. This departs from Sen’s own definitional formula, which Robeyns treats as a special case (p. 8 n. 1).
C3	The capability approach is not a theory of justice, of wellbeing or of human rights; each of these is one purpose (B1) a capability theory can have	strong, within the framework	Follows from the approach/theory distinction (§2.3) and is applied consistently in §§3.7, 3.13 and 3.14
C4	Nussbaum’s work is one capability theory, not a co-equal version of the approach	moderate	Informal argument from the modular view (p. 79), plus appeal to Qizilbash 2013. It presupposes the modular view (C1). Nussbaum (2011) holds otherwise, as Robeyns notes.
C5	The “list” critique misfires against the approach; for particular theories, selection must fit the theory’s purpose, and Sen favours procedural selection	moderate	Informal argument (§4.3). Sen’s position is reported from Sen 1993, 2004a and 2004, which were not read here.
C6	The approach is ethically individualist but neither methodologically nor ontologically individualist	strong as a conceptual distinction; moderate as a claim about the literature	Conceptual analysis (§4.6.1, following Bhargava). The evidence about the literature is selective citations (Sen’s cooperative conflicts, Drèze & Sen 2002, group-inequality studies).
C7	Most standing critiques (list, individualism, power, political economy, liberalism, government focus) are invalid against the approach and at most valid against particular theories	moderate	The same move is repeated across §§4.3–4.8. It is partly true by construction: by Robeyns’s own admission the A-module is thin enough that little can be valid against it (pp. 192, 210).
C8	Sen’s recent justice work may be a “hybrid” capability theory that does not commit to A5	weak	This is reported as Baujard & Gilardone’s (2017) interpretation, with a conditional “if that interpretation is correct”. Robeyns disputes it for Sen’s earlier work (p. 76 n. 39).
C9	The capability approach lacks the needs/wants (non-contingent needs) distinction that needs theories supply	moderate	Informal argument (pp. 178–179), drawing on Reader & Brock. It is one of the few critiques Robeyns concedes against the approach itself.
C10	Capabilities are best understood as option freedoms in Pettit’s sense, not Berlinian positive freedom	moderate	Conceptual analysis (§3.3), comparing Berlin, Taylor and Pettit
C11	The capability approach entails no single account of wellbeing; the “objective list” classification is wrong	moderate	Informal argument (§3.7.3). It depends on counting first-person uses of capability language as capability theories.
C12	Capability-based measurement makes a substantive empirical difference from income metrics	moderate, second-hand	Cited studies: Ruggeri Laderchi 1997 (income “insignificant as a determinant of the shortfall in health, schooling and child nutrition”, p. 8); Sen 1985a and Drèze & Sen 2013 (Table 1.1: India ranks 1st of 16 on GDP per capita but 9th–15th on the listed functionings, p. 13); Van Zanden et al. 2014. None was re-analysed here.

Concepts

Capability approach vs capability theory — the general, open, underspecified framework, as opposed to a specific use of it for a purpose. “Capability theory” is shorthand for “capability account, analysis, application or theory” (pp. 29, 37).
Advantage — “those aspects of that person’s interests that matter”; used to stay agnostic about what is being compared (p. 25).
Functioning / capability — realised beings and doings / the real opportunity to achieve them. Robeyns uses one capability per functioning (the later plural usage), not Sen’s early usage in which a person has one capability set (pp. 91–92).
Conversion factors — personal, social and environmental determinants of how much functioning a resource yields (pp. 45–47).
Structural constraints — institutions, laws and norms that shape conversion factors and can also shape capability sets directly, without passing through resources (pp. 65–66, 81).
Hybrid theory — uses functionings and capabilities but violates some A-proposition (pp. 75–77).
Inflation — calling capabilitarian something that fails the A-module, e.g. opportunity sets over commodity bundles (pp. 77–78).
General vs specific capabilities — “being able to wear appropriate clothes” versus “being able to wear a sari” (pp. 96–97). Robeyns prefers “general” to Williams’s “basic”.
Basic capabilities — four senses: Sen 1980 (“first rough attempt”); Sen later (a poverty threshold, the dominant usage); analytical philosophy (morally urgent entitlements); Nussbaum (innate equipment) (pp. 94–96).
Robustness — the probability that a capability can be realised if chosen (pp. 97–98).
Ethical / normative individualism — only individual persons are units of ultimate moral concern. Robeyns identifies it with Nussbaum’s “principle of each person as an end” (p. 57).
Ontological / explanatory / methodological individualism — respectively: only individuals and their properties exist; all social phenomena are explicable in terms of individuals; and a doctrine combining the semantic, ontological and explanatory versions (pp. 184–185).
Narrow vs broad use — the narrow use takes interpersonal comparison as its purpose and has no C1 or C4; the broad use adds explanatory theories and normative principles (pp. 84–86).
Collective capability — a personal capability whose realisation requires group action, such as winning suffrage; a subset of personal capabilities (pp. 115–117).

Connections

Sen, “Equality of What?” (Tanner Lecture, 1979). Robeyns dates the approach from this lecture (p. 7), and several of her statements about it check against the text:
- Confirmed.
  - The lecture does not speak of freedoms and uses “ability” and “power” (p. 99 n. 3). The basic-capability examples are “the ability to move about”, “the ability to meet one’s nutritional requirements” and “the power to participate in the social life of the community” (lecture p. 218).
  - The conversion argument is present: “the conversion of goods to capabilities varies from person to person substantially” (p. 219).
  - Weighting is left culture-dependent (“any application of it must be rather culture-dependent, especially in the weighting of different capabilities”, p. 219). This supports Robeyns’s reading that Sen never meant a fixed list.
  - Sen calls basic capability equality only “a partial guide to the part of moral goodness that is associated with the idea of equality” and lets the index be used in leximin or utilitarian-like ways (p. 220). This fits A6/A7 and her reading that no single distributive rule is fixed.
- Flags: places where the 1979 text does not bear Robeyns out.
  1. Egalitarianism. p. 94 n. 2 says that in the lecture Sen “argued that almost any theory of distributive justice is egalitarian, in the sense that they all advocate equality of something”. The lecture makes no such general argument. It examines three conceptions of equality (utilitarian, total-utility and Rawlsian) and calls the equal outcome under utilitarianism “egalitarianism by serendipity” (p. 202). The “every theory is egalitarian in some space” thesis is, to my knowledge, the argument of Sen’s Inequality Reexamined (1992), ch. 1, which was not read here, so this is unverified.
  2. Module A. p. 76 n. 39 lists Sen 1980 among works that “could be seen as an endorsement of all propositions of module A” for justice. The lecture has no concept of functionings: the word does not occur. Its basic capabilities are introduced as “the interpretation of needs” with a “notion of urgency” (p. 218), a value-laden framing of the kind Robeyns herself calls a mistake under A2 (pp. 43–44). So A1 (in its functionings half) and A2 are not endorsed in the 1979 text. A3, A4, A6 and A7 are.
  3. “Basic capability” later became “capability”. p. 94 says the 1979 “basic capability” “would later become ‘capability’”. The 1979 notion is explicitly the needs and urgency notion, so it anticipates Sen’s later threshold sense of “basic capabilities” at least as much as the general “capability”. The continuity runs both ways.
  4. “Merely asking”. pp. 155–156 say Sen was “merely asking” what the equalisandum should be. The lecture does more: it promises “propaganda” for basic capability equality (opening paragraph) and claims “certain clear advantages over other types of equality” (p. 220). Robeyns is right that Sen fixes no single distributive rule. “Merely asking” understates his advocacy.
SEP, “The Capability Approach” (Robeyns & Byskov). This is the same author’s shorter synthesis. It is the place to go for Pogge’s resourcist critique, which this book does not summarise (see corrections).
Nussbaum, “Capabilities as Fundamental Entitlements” (2003). This is the “strong critique” of §4.3 (cited as Nussbaum 2003a). Robeyns’s answer is that the critique conflates approach and theory. Robeyns also borrows from that paper Nussbaum’s rape example for A2 (pp. 41–42).
Wells, IEP “Sen’s Capability Approach”. A Sen-centred survey whose criticisms section (illiberalism, under-theorisation, individualism) overlaps §§4.6–4.8 here. Robeyns answers the same critiques by relocating them to B/C choices.
Valentini, “A Paradigm Shift in Theorizing about Justice?” (2011), and Zwolinski’s review of The Idea of Justice. Both concern Sen’s anti-transcendental, comparative view of justice. Robeyns reports that view (pp. 152, 157, 173) and floats the reading that it makes Sen’s justice work a public-reasoning “hybrid” (§2.10), but does not assess it.
Sen, Nobel lecture “The Possibility of Social Choice” (1998), and “The Impossibility of a Paretian Liberal” (1970). Robeyns ties Sen’s refusal of a list to his social-choice background (p. 173) and his welfare-economic “non-welfarism” to widening the informational basis (pp. 204–206). She does not discuss either paper.
LIT-202 (Pettigrew, Consent and the Formation of Preferences). §3.9 is the capability literature’s side of the adaptive-preferences debate that LIT-202 recasts in terms of formation mechanisms and “free exercise of agency”. Robeyns reports Elster’s process account, Khader’s perfectionist three-condition definition (Khader 2011, 51), Nussbaum’s list-relative account, and Burchardt’s cohort evidence. Her claim that adaptive preferences threaten participatory selection of dimensions (B2) and capability-only metrics (B6) is an application LIT-202 does not make.
LIT-130 (Personal Identity and Ethics). A8’s defence quotes Nussbaum’s “each person has just one life to live” and grounds ethical individualism in bodily separateness (pp. 57–58). This is the separateness-of-persons theme LIT-130 surveys (Rawls, Parfit, Nozick). Robeyns does not engage the personal-identity literature.
LIT-179 (Algorithmic Nudging). §3.4 argues that systematic irrationality (Ariely, Kahneman) can justify some paternalism, i.e. promoting functionings rather than capabilities, within a capability theory (pp. 108–109). This is a B-module choice relevant to how LIT-179’s oversight argument would be framed capabilitarianly. The connection is thematic only.
Anthology of the SOTA. No ANTH- document is relevant.

Bearing on the record

No THEORY document is supported or contradicted. The record has no THEORY on welfare metrics or justice (the scoping report found nothing on Sen, capabilities or Rawls). If the Sen batch produces one, this book supports its framing: the approach is only an evaluative-space claim plus A2–A8, and the list, individualism and liberalism disputes are disputes between capability theories. It would also require the THEORY to state the needs/wants gap (C9) and the thinness admitted in C7.
For filing.
- Tags. ethics is first: the book is about what should count in evaluation. society-and-governance covers justice, human rights, human development and policy. social-science covers welfare economics, poverty measurement and empirical applications. agency is not proposed: it is one B-module among seven, treated briefly (pp. 63–64, 104–105).
- Status. Active.
- Citing Sen through this book. Attribute to Robeyns, not to Sen, the definition (“value-neutral” functionings), the A/B/C modules, “hybrid”, and the Nussbaum-as-a-theory verdict. Sen’s own definition includes the “reason to value” clause, which Robeyns rejects as a general requirement.
No ML-practice content. The book carries no instruction for machine-learning practice. A reader might ask whether “capabilities” in ML evaluation borrows anything; Robeyns expressly excludes non-person senses of “capabilities” (p. 9 n. 4), so nothing here licenses that link.

Note of 2026-10-16: THEORY-025 now cites LIT-287 as a source.

Limitations

The core is thin by design, and that thinness is what makes most critiques miss (C7). A reader should weigh “this critique hits only particular theories” against the fact that nearly everything contentious has been moved into optional modules. Robeyns concedes the core “do[es] not necessarily have significant implications for the role of power” (p. 192).
Evidence. The account is a synthesis argued informally. There is no systematic survey of how capability work actually specifies its modules, and the empirical claims (ch. 1, §3.9, §4.10) are second-hand.
Attributions to Sen and Nussbaum are Robeyns’s readings. She was Sen’s doctoral student (pp. 3–4) and takes sides in ch. 4 (p. 19). At least one attribution to Sen 1979 overstates what that lecture argues, and one of the texts she says endorses the whole A-module does not endorse A1 or A2 (see Connections).
Absent topics. Pogge’s resourcist critique is not treated. There is no treatment of Anderson’s democratic-equality critique of luck egalitarianism beyond citation. Responsibility is flagged as “a remarkable absence” in the literature (p. 156) but not developed.
Internal inconsistencies. Three places mislabel modules (pp. 142, 201, 202; see corrections), and A6 is sometimes still called “intrinsic” value despite p. 54.
Open status of master values. Whether master-value theories breach A7, and so count as hybrid, is left open (pp. 71–72, 77).
Scope. Human capabilities only (p. 9 n. 4).

Open questions

Master values and A7. Does a master-value theory violate A7? Robeyns suggests the answer may depend on whether capabilities constitute the master value or contribute to it (p. 72). An account of when aggregation by a master value preserves value pluralism would settle it.
Sen and A5. Is Sen’s post-2006 justice work hybrid, i.e. not committed to A5? A close reading of The Idea of Justice against Baujard & Gilardone would settle it. The Sen batch covers this only second-hand (Valentini, Zwolinski).
Needs. Can the needs/wants distinction be imported as a C4 principle without turning a capability theory into a needs theory (§4.4)?
Implausible theories. Which logically possible capability theories are empirically implausible (ch. 5, fifth direction)? No criterion is given.
“Reason to value”. Robeyns notes that “little work has been done so far to flesh out this embryonic idea” (p. 154) and doubts whether the approach “has a solid unified rationale” for a full account of justice.

Corrections to the seeded skim

none (there was no seed or dossier)
The scoping report says ch. 2 gives the modular core as “A1–A6”. The A-module has eight elements, A1–A8 (table 2.2 p. 38, table 2.5 p. 74): A6 is “other dimensions of ultimate value”, A7 “value pluralism”, A8 “valuing each person as an end”.
The scoping report says Pogge 2002 (the resourcist critique) is “summarised in SEP and Robeyns ch. 4”. The book does not summarise or answer Pogge’s resourcist argument: that capability metrics treat natural endowments as injustice and cannot be justified over Rawlsian primary goods. Pogge 2002 is cited twice. The first is as one item in a list of works comparing capability and Rawlsian justice (p. 150 n. 21). The second is in §4.7.2 (p. 193), for a different critique: that the approach, Sen’s work especially, “overemphasises the role of national and local governments” and neglects the injustice of the global economic order. Robeyns answers that second critique (“not … a valid critique of the capability approach as a general framework”, p. 194). For the resourcist critique the record must rely on the SEP entry or on Pogge himself.
The brief expects a treatment of “too much / too little theorising”. The book has no section framed that way. The nearest material is: the underspecification complaints of Roemer, Mehta and Sugden (pp. 10, 207–208); the list debate (§4.3); the requirements for a capability theory of justice (§3.13.2); and Sen’s anti-”transcendental” stance (pp. 152, 157, 173). Robeyns answers each through the approach/theory distinction, not as a debate about how much to theorise.
There are three internal slips in the module labels, which a filer quoting labels should not copy:
- p. 142 calls A6 “other dimensions of intrinsic values” and A7 “normative individualism”. In ch. 2, normative individualism is A8 and A7 is value pluralism, and p. 54 itself recommends dropping “intrinsic”.
- p. 201 pairs “a strong notion of agency (B3)” with “an elaborate account of social structures (B4)”. Agency is B4, and structural constraints are B5.
- p. 202 has “human agency (module B3)”; B3 is human diversity.

nucleation

Explorer