Metaethical constructivism is the view that insofar as there arenormative truths, they are not fixed by normative facts that areindependent of what rational agents would agree to under somespecified conditions of choice. The appeal of this view lies in thepromise to explain how normative truths are objective and independentof our actual judgments, while also binding and authoritative forus.
Constructivism takes different forms. Some versions seek to explainall normative reasons or normative truths, while others are limited tomoral reasons, moral obligations, or moral truths. Their relation tothe traditional classifications of metaethics is problematic,particularly with respect to the realism–antirealism debate.These disputes reflect deeper divergences concerning the verydefinition of metaethics, the relation between normative andmetaethical claims, and the methods thought to be distinctive ofmetaethical inquiry. In what follows, varieties of metaethicalconstructivism will be considered by focusing on the specificquestions that constructivist theories are designed to address.
Moral constructivism has also be advanced as a normative theory ofjustification: it holds that the moral principles we ought to acceptare those that agents would agree to, or endorse, under conditions ofhypothetical or idealized rational deliberation. Constructivists,however, differ in how they conceive of such deliberation. Some followadaptations of Kant’s categorical imperative, while others drawon contractualist approaches, or adopt the method of reflectiveequilibrium.
Section 1 sets out the core claims and underlying motivations ofconstructivism. Sections 2 and 3 examine two main varieties ofmetaethical constructivism while section 4 turns to hybrid theoriesthat integrate central constructivist commitments withnon-constructivist elements. Section 5 discusses the position ofconstructivism within the metaethical debate. Section 6 offers acritical assessment of its constitutivist strategies, and section 7provides concluding remarks.
The term ‘constructivism’ entered debates in moral theorywith John Rawls’ seminal Dewey Lectures “KantianConstructivism in Moral Theory” (Rawls 1980), wherein Rawlsoffered a reinterpretation of the philosopher Immanuel Kant’sethics and of its relevance for political debates.
According to Rawls, these debates fail to effectively address thepolitical problem of ethical disagreements because they adoptmetaphysical standards of objectivity, which appeal to the independentreality and truth of values. In his view, such standards areinadequate to address disagreement in a political debate in which allthe parties in the dispute claim to be defending the only true view,because they lead to a stalemate in the discussion, with each partyaccusing its opponent of being ‘blind’ to the moraltruth.
Rawls is especially concerned with coordination problems that arise inpluralistic contexts, wherein citizens hold different and to someextent incommensurable moral views. The need for objectivity,according to Rawls, ispractical: it arises in contexts inwhich people disagree about what to value and need to reach anagreement about what to do. He attributes to Kant the idea that weought to approach objectivity as a practical problem and that we canfruitfully address moral disputes by reasoning about them (Rawls 1971:34, 39–40, 49–52). Rawls thus turns to Kant in order toargue for a conception of objectivity that is not metaphysical, thatis, a conception of objectivity that avoids claims to universal andfundamental moral truths that are independent of our fully rationaljudgments. On this conception, nobody is assumed to have a privilegedaccess to moral truth, but all have equal standing in reasoning aboutwhat to do. To this extent, Kant’s theory is regarded asproviding a metaethical alternative both to realism and skepticismabout the existence and nature of moral truths.
Rawls’ account of Kantian constructivism in moral theory (1980)generated a large literature, and produced several varieties ofconstructivism. Some of these views depart from Rawls’ ownconception of constructivism.
Kantian constructivism is defended in a variety of ways, but itsdistinguishing feature is that it understands the nature of moral andnormative truths based on considerations about practical reason andits relation to agency, although some focus on rational agency as suchwhile others take into account also embodiment and socialembeddedness. On this view, reasons for being moral do not spring fromour interests or desires; instead, they are rooted in our nature asrational agents. Insofar as moral obligations are justified in termsof rational requirements, they are universally and necessarily bindingfor all rational beings. Because of its claim about the universalauthority of reason and obligations, Kantian Constructivism isregarded as the most ambitious form of metaethical constructivism. Inthis section, we will consider three main varieties, starting with theconstructivist interpretation of Kant’s ethics.
John Rawls first advanced a constructivist interpretation ofKant’s account of practical reason and moral obligation (1980,2000). On his reading, the novelty of Kant’s view commits him toconstructivism, best appreciated against rival accounts of obligation(Rawls 1980, 1989, 2000). According to Kant, all previous ethicaltheories fail to account for the objectivity and authority of moralobligation because they fail to explain how reason plays a role in ourlives because they misunderstand its practical function andmischaracterize its relation with the ends of choice (Kant G 4:441–444; C2 5: 35–41, 153, 157). Kant’s argumentsspecifically address sentimentalism and ‘dogmaticrationalism’. Sentimentalism, championed by Francis Hutcheson,David Hume, and Adam Smith, holds that ethical judgments stem fromsentiments and regards reason as incapable of moving us to action onits own. According to the sentimentalist, the role of reason is solelyinstrumental. That is, reason merely finds the means to satisfy anagent’s ends, and it is not capable of indicating which ends areworth pursuing. This claim exposes sentimentalism as a“heteronomous” doctrine, which fails to establish theobjectivity of moral obligations. This is because sentimentalismtreats moral obligations as conditional upon our interests, and thusas having limited authority.
Kant raises an analogous objection against dogmatic rationalism,championed by Christian Wolff and Gottfried Wilhelm Leibniz, whichasserts that there are moral truths accessible by rational insight(Kant G 4: 443; Rawls 2000: 50, 228). This view represents anintuitionist form of moral realism according to which reasonrecognizes objective values or moral ends that exist prior to andindependently of our reasoning and of the kinds of agents that we are.The function of reason is to identify these moral truths withoutitself participating in their formation or determining their content.The question arises how such pre-fixed moral ends are authoritativeover us and action guiding. The answer must be that moral truths guideus only on the condition that we have a corresponding desire to beguided by what is rational (Rawls 1980: 343–46; Rawls 1989:510–13). By narrowing the function of reason in this way,Kant’s point is that dogmatic rationalism fails to secure theconclusion that moral obligations have unconditional authority over us(Kant G, 4: 441). This is because, for dogmatic rationalism, moraltruths guide us only on the condition that we have a correspondingdesire to be guided by what is rational (Rawls 1980: 343–46;Rawls 1989: 510–13).
Kant’s general diagnosis is that all such doctrines fail tocapture the practical function of reason because they areheteronomous. They deny the authority and efficacy of reason, eitherholding that reason can only recognize objective ends that existindependently of its operations, or claiming that reason can bindagents only with the help of inclination or interest. For Kantheteronomy is a form of moral skepticism, understood as skepticismabout the power of reason to establish moral truths and theirauthority. On this reading, then, constructivism is part ofKant’s overall argument for grounding ethics in reason, againstthe skeptical view that there are no normative truths (Korsgaard1996a; Stern 2013; Wallace 2012). Skepticism is avoided only if reasonis accounted as autonomous, and its authority does not derive fromanything outside it. Reason is autonomous if its authority rests onits proper activity, rather than being derived from elements of theworld outside of reason. Thus, the norm governing the activity ofreason must be internal to reason, rather than dependent on any givenvalue, interest, or desire. That is to say that reason is a“self-legislative activity” (Kant G 4: §2), and itslegislative activity is governed by a norm, which Kant calls the‘Categorical Imperative.’
The Categorical Imperative expresses the autonomy of reason and is itsgoverning principle. It is not a mere decision-procedure to determinewhat to do, but the ‘constitutive norm’ of reason, thatis, the basic standard of rationality in thinking and acting (Rawls1989: 498–506; Rawls 2000: 166, 240–244; Korsgaard 1996a:36–37; O’Neill 1989b: 18–19, 59n, 128, 180; Reath2006: 221–222; Reath & Timmermann 2010; Engstrom 2009:Chapter 5; Turan 2019). The Categorical Imperative comes in differentformulations that Kant regards as equivalent (G 4: 421, 429, 431,433); ultimately, it is the requirement that in deliberating, we testour motives by considering whether the principle they express can beadopted as a universal law, a principle that applies to and binds allagents endowed with rational capacities. To this extent, Kant iscommitted to the “constitutivist view” that the source ofthe categorical force of moral obligations lies in the constitutivefeatures of rational agency (Rawls 2000: 263–265; O’Neill1989a, Korsgaard 1996a: 236ff). We will return to this point insection 6.3.
Scholars are divided about the constructivist interpretation ofKant’s theory of practical reason, even though nobody deniesthat, for Kant, the laws of the mind are laws of reason (compare Guyer2013 and Engstrom 2013). The most general source of reservations isthat constructivism builds upon the critique of realism, butKant’s claims about objective moral knowledge seem bestvindicated by moral realism (Schmidt and Santos 2017). Some doubt thatmoral realism is Kant’s own intended target, hence suggestingthat Kant’s constructivism does not build upon a critique ofmoral realism (Stern 2012a: 7–68, 2012b; Schmidt & Santos2017). A related contested issue is whether Kantian constructivismradically departs from perfectionism or entails it (Pollock 2017,forthcoming; cf. Zylberman forthcoming; Reath forthcoming). It shouldbe noted, however, that in this dispute, constructivism is generallytaken to be a form of antirealism (Ameriks 2003: 268, 274; Wood 2008:108, 337, 374–375; Thorpe 2019; Vesper 2020; Rodriguez 2022),while Rawls introduced Kant’s constructivism as a distinctalternative to both realism and antirealism, where the latter includessubjectivism and relativism (Rawls 1980; O’Neill 1989a: 1;Engstrom 2013: 138ff; Bernstein 2025).
According to realist interpreters, Kant’s defense of theautonomy of reason takes place within a realist foundationalistproject grounded on the absolute value of humanity (Wood 1999: 157,114; Rauscher 2002; Langton 2007; Johnson 2007; Hills 2008; Krasnoff1999; Kain 2006a, Kain 2006b; Irwin 2009; Galvin 2011; Lyons 2020;Scholten 2020). First, critics point out that Kant’s defenseprovides a transcendental argument, an argument that highlights theconditions under which it is possible for something to be the case.The value of humanity is the condition of the possibility of allvaluing. The issue revolves around the nature of transcendentalarguments, and whether they commit us to moral realism (Stern 2000;Larmore 2008: 121; Watkins & Fitzpatrick 2002; Fitzpatrick 2005;Tiffany 2006; Brum, Stern, & Werner 2017), something thatconstructivists deny. Realist critics also dispute the force and thescope of the argument for the autonomy of practical reason, a debatethat partly stems from differing assessments of the constitutiviststrategies, which will be discussed in sections 6.2 and 6.3.
Second, realist critics hold that Kant’s claim in theCritique of Practical Reason that we are immediatelyconscious of the moral law as a “fact of reason”, (C25:46–48) represents a realist foundation of morality (Ameriks2003: 263–282; Kleingeld 2010: 55–72). By contrast,constructivist interpreters tend to downplay the role of the fact ofreason in Kant’s general argument for the objectivity of moralobligations (O’Neill 2002: 81–97; Łuków 1993:204–221). Rawls takes the fact of reason to show that Kantdevelops “not only a constructivist conception of practicalreason, but a coherentist account of its authentication” (Rawls1999: 524; Rawls 2000: 268–273). In his view, the fact of reasonindicates that the deliverances of practical reason cohere with ourmoral experience. This congruence is an integral part of Kant’svindication of ethical objectivity, but it is no commitment torealism. Rather, it simply confirms that there is no discrepancybetween the requirements of practical reason (which are expressed bythe Categorical Imperative) and our ordinary experience of morality(Rawls 1980: 340; Rawls 1989: 523–524; Rawls 2000:253–272, 268, 273; Kant C2, 5: 15). For Rawls, the realistnotion of objectivity is “unnecessary for objectivity”(1980: 570). However, his argument in support of a coherentistconception of practical reason is found too weak to captureKant’s view of moral obligations as objective rationalrequirements (Larmore 2008: 83–84; Stern 2012a: 7–40). Asan alternative to these readings, some interpreters argue that Kant isconstructivist about the authority of moral obligations and practicallaws for finite agents, but not about the contents of such laws, whichapply to all rational agents as such (Engstrom 2009, 2013; Sensen2013). For others, Kant is a constructivist about a limited set ofgeneral substantive moral principles – and as a constititutivistabout moral authority (Reath 2022). According to Schafer, Kant’sconstitutivism is more appropriately interpreted as‘reason-first’ rather than ‘agent-first’(Schafer 2019, 179). On this reading, the constitutive basis ofnormativity is located not in the rational agent, but in rationalactivity as such. This thesis is broadly shared within Kantianscholarship (cf. Reath forthcoming). What remains contentious,however, is Schafer’s proposed shift in focus from‘autonomy’ to ‘understanding’ in explicatingrational activity (Schafer 2023, cf. O’Neill 1989, 2015; Sensenforthcoming).
Constructivist interpreters have largely deemphasized the metaphysicalimplications of Kant’s ethics in order to highlight hisrecognition of the limits of finite agency and to foreground therational basis of morality as a fundamentally cooperative enterprise.Kant embarks on the project of vindicating reasoning, starting fromvery modest considerations about rational agency. Such starting pointsare not sufficient “to sustain or revive classical philosophicalambitions to build vast metaphysical structures on reason alone”(O’Neill 2015: 3; C1: A xiii; cf. Schneewind 1970). Theconstructivist interpretation of Kant’s ethics highlights thecooperative nature of the standards of reasoning:
“Kant’s repeated use of metaphors ofconstructionandcollaboration in his discussion of reasoning make itnatural to speak of his approach and method asconstructivist, and of his aim as theconstructionof reason’s authority, and thereby of a basis for offeringothers reasons for truth claims and moral claims, reasons forfavouring some rather than other practical and political aims.”(O’Neill 2015: 4)
Ultimately, the standards of reasoning are justified by a practicalcooperative function: “it must exhibit patterns that otherscould discern, and thus it must be law-like” (O’Neill2015: 4).
The antimetaphysical orientation of constructivism is central toleading defenses of metaethical constructivism. Christine Korsgaardcharacterizes Kantian constructivism as a form of “proceduralrealism” – the view that “there are answers to moralquestionsbecause there are correct procedures for arrivingat them”; and she contrasts procedural realism with“substantive realism” – the view that
there are correct procedures for answering moral questionsbecause there are moral truths or facts, which existindependently of those procedures, and which those procedures track.(Korsgaard 1996a: 36–37, see also Korsgaard 1983: 183)
Substantive realism holds that moral judgments have objective criteriaof correctness only if they represent facts about the world.Constructivism, by contrast, claims that such criteria arise fromobjective standards of reasoning about practical matters. Thus,reasons against deception and manipulation are generated throughpractical reasoning rather than discovered by empirical investigation,grasped by the intellect, or revealed by some god. What makes thisview Kantian is that practical reasoning has a single criterion– the Categorical Imperative – through which moralobligations, as requirements of reason, are objectively grounded.
Korsgaard’s defense of constructivism parallels Kant’scase for the autonomy of practical reason, as reconstructed by Rawls.She argues that substantive realism cannot answer the skeptic whodenies that there are reasons to be moral, since realism merelypresupposes objective standards without justifying them. Hence itfails to explain the authority of moral obligations (Korsgaard 1996a;Korsgaard 2008: 30–31, 55–57, 67–68; Stern 2012a;Brady 2002). Realists claim that reasons must be anchored in normativefacts, yet such facts cannot explain how they bind or motivaterational agents. If it is a normative fact that deception is wrong,how does that fact compel us to refrain from deceiving? This is notjust a psychological question about motivation, but a normativequestion about authority.
For Korsgaard, humans are reflective agents, able to step back fromtheir thoughts and desires and question their legitimacy (1996a:10–11, 17, 93). Reflection suspends the authority of givenattitudes and allows agents to ask what there is reason to do orbelieve. Because they are reflective, humans form ideals of thepersons they aspire to be and can guide themselves accordingly. Theircapacity for self-governance lies in endorsing universal standards,and the proper form of such governance is self-legislation (Korsgaard1996a: 36, 91, 231–232; Korsgaard 2008: 3).
Rational agents are guided by universal principles they havelegislated. This does not mean, however, that individuals arbitrarilydetermine the moral law; otherwise, evil people would not be bound byit (Korsgaard 1996a: 234–235; O’Neill 2003c; Reath 2006:92–170, 112–113; Korsgaard 2008: 207–229). Rather,the moral law obliges only insofar as it is self-legislated: agentscan act autonomously on moral requirements only if they legislatethem. Universal principles ensure that action expresses integrityrather than unreflective preferences or desires, since they areconstitutive of rational agency. By contrast, agents who actmindlessly or compulsively lack this integrity. Still, rational agentsmay permissibly act on desires that endure reflective scrutiny.
A canonical objection to grounding morality solely in rationality isthat it neglects the special bonds with loved ones, and thus fails tocapture integrity and morality (Williams 1981: chs. 1–2). Toaddress this, Korsgaard introduces “practical identities,”roles that serve as sources of special obligations (Korsgaard 1996a:101, §3.3.1; 2009: 20). For example, Camille values herself underthe descriptions of artist, French citizen, and Auguste’s lover.These identities guide her choices, sustain her integrity, andgenerate special obligations to pupils, compatriots, and friends. Yetsuch roles bind us only when reflectively endorsed (Korsgaard 1996a:§3.3.1; 2009: 22). Reflective endorsement requires testingloyalties against the principle of universality, which commits us tomorality. Thus, to value ourselves under particular descriptions, wemust also value humanity in ourselves and others (Korsgaard 2008:Lecture 6, 25–26).
Korsgaard (1996a) offers a transcendental argument that what we oughtto do is justified by the norms constitutive of rational agency.Valuing humanity – understood as the capacity for rationality– is the condition of valuing anything at all (1996a:121–123; 1998: 60–62). In deliberation we confer value onour ends, thereby attributing value to ourselves. Hence, the value ofobjects depends on the rational capacity of evaluators, and‘humanity’ names an unconditional value. We must valuehumanity in ourselves and others on pain of incoherence. Practicalidentities alone cannot sustain integrity when they conflict with thisrequirement: a Mafioso, for example, cannot justify his conduct onuniversal principles and thus fails as a rational agent. Integrity,necessary for agency, requires commitment to morality grounded inreason.
Some critics argue that Korsgaard’s argument tacitly rests on arealist premise regarding the value of humanity, and thus it is not acomplete alternative to moral realism (Watkins & Fitzpatrick 2002;Fitzpatrick 2005; Ridge 2005; Stieb 2006; Kain 2006a, Kain 2006b;Papish 2011; O’Shea 2015; Schafer 2015a), nor does it offer adistinctive reply to skeptical challenges to ethical objectivity.Others find Korsgaard’s argument vulnerable to the objection of‘regress on conditions’, which shows that it is notlogically necessary that the condition of a thing’s value bevaluable itself (Rabinowicz & Rønnow- Rasmussen 2000;Kerstein 2001; Ridge 2005; Coleman 2006). If so, humanity may be thecondition of the possibility of value and yet lack value itself.
Korsgaard replies that “there need be no such regress if thereare principles that are constitutive of the very rational activitiesthat we trying to perform” (2008: 5, 1996b: 164–67, 2009).Indeed, some of these objections overlook the distinction betweensubstantive realism and constitutivism (or structural realism, Nozick1981:545–551); nevertheless, the question of the realistimplications of constitutivist strategies resurfaces.
For Korsgaard, the principles of practical rationality areconstitutive of rational agency, in the sense that they are standardsarising from and justified by the nature of the object in question(i.e., the rational agent). More specifically, unless the objectconforms to the standard, it ceases to be the kind of object that itis (‘constitution requirement’). If the functionof the house is to serve as a habitable shelter, then it must conformto the standard of being a habitable shelter. A thing that does notserve this purpose is not a house. Secondly, some objects makethemselves into the kind of objects that they are by conforming totheir constitutive standards (‘self-constitutionrequirement’). Rational agents make things happen, that is, theyare efficacious. They exert their efficacy in a specific way, i.e.,autonomously. Autonomous agents must conform to the categoricalimperative, and in order to be efficacious, they must conform to thehypothetical imperative. It is by conforming to these principles,Korsgaard argues, that one makes oneself into an agent.
Her more recent works point toward a new direction in the debate aboutconstructivism, which combines Aristotelian and Kantian features(Korsgaard 2008, 2009). However, in her view, Kantian constitutivismuniquely captures the requirement of self-constitution, a featureabsent from the Aristotelian conception of virtue as excellence ofcharacter (Korsgaard 2019). The Kantian idea of self-constitution isthus indispensable for any adequate account of rational – andthereby moral – agency.
Korsgaard’s constitutivist strategy of grounding moralnormativity in the constitutive features of agency raises worries thatwill be discussed in sections 6.2–6.3.
Onora O’Neill defines metaethical constructivism as a thirdalternative between realism and subjectivism (O’Neill 1989b:279). She departs from the versions of Kantian constructivismdiscussed above because she makes no appeal to transcendentalarguments and rejects the idealized conceptions of rational agencythat are at play in other versions of Kantian constructivism. In herview, “finite rational beings” should not be construed as“beings whose rationality is finite”, but as “finitebeings who are rational”. Finitude does not entail anylimitation on strategic reasoning, even though there is no way toestablish that finite rational agents have access to all sorts ofreasoning that infinite or disembodied rational agents may have. Thekey thesis is that instrumental principles are not the only principlesof rationality, and more importantly, they never operate in isolation(O’Neill 1989b: 74).
According to O’Neill, this more austere constructivism is closerto Kant’s own theory than other varieties (O’Neill 1989a;O’Neill 2015). In her view, Kant’s constructivism ismotivated by the vivid awareness of interdependency, finitude, andmutual vulnerability. Humans are prone to mistakenly rely on claimsthat are not warranted, and thus they need to check and criticize theunjustified and arbitrary assumptions they make in reasoning(O’Neill 1989b). O’Neill shares Kant’s claimthat
reasoning is fundamentally practical: it aims to providestandards ornorms that thought, action andcommunication can (but often fail) to meet. (O’Neill 2015:2)
Second, she shares Kant’s view that reasoning is necessarybecause humans are finite and interdependent beings. Third, she agreesthat the principles of reason are not given by intuition orintrospection but must be discerned through the exercise of rationalpowers. Finally, she accepts Kant’s claim that norms ofreasoning must be universal if they are to fulfil their function.
The difference lies in how universality is justified. ForO’Neill, universality is required because reasoning must beshared among a plurality of finite, interdependent agents whose actionand communication are not antecedently coordinated (by instinct,providence, or law). The requirement of universality – orfollowability – thus sets minimal constraints on what counts asreasoning (O’Neill 2015: 3). From this it follows that noplurality of agents can coherently adopt principles that destroy orundermine the agency of any of its members (O’Neill 1989a: 10;see also 1989b, 2015). Hence practical reasoning justifies theprohibition of harming, coercing, and deceiving others.
This is not to suggest that the requirement of followability canresolve all moral questions or fully determine duties. O’Neillholds that many issues demand further substantive argument. Still, heraccount promises to vindicate reason’s capacity – andauthority – to distinguish genuine justification from mererationalization. While ‘contractualism’ is often linked tometaethical constructivism (Rawls 1971; Scanlon 1998; Hill 1989, 2001;Milo 1995), and many regard construction itself as a hypotheticalprocedure akin to contract (see e.g., Darwall, Gibbard, & Railton1997: 13; Hill 2001, 2002; Walden 2012; cf. O’Neill 2003a,b;Street 2010: 365), offers a sharp account of the difference betweenKantian constructivism and contractualism in terms of their scope,audience, and ambitions. Kantian constructivism holds that practicalreasoning yields unconditionally authoritative moral claims, whereascontractualism does not.
O’Neill agrees with Kant that only reason itself can verify thecredentials of its own claims. The process of identifying theprinciples of reason is avowedly circular, yet this circularity is notproblematic, since the process of verification is reflexive: itinvolves reason critiquing the claims of reason itself. Morespecifically, the critique of reason uncovers a basic principle ofreasoning: we should rely only on principles that other rationalagents can share. The authority of reason is thus conferred throughpublic communication among free, rational agents, and lies in the factthat the principles governing our thought are neither self-serving norself-defeating. We learn what these principles require by submittingour claims to free and critical debate, the practice that constitutes“the public use of reason” (O’Neill 1989b:70–71, 206).
Since the critique of reason is a continuous, progressive, andreflexive process, reason has a history that coincides with thedevelopment of practices of tolerance and mutual recognition(O’Neill 1999: 174, 2002). Such practices establish theauthority of reason. This ‘developmental’ view ofpractical reason and its autonomy accounts for change and progress,and provides a significant explanatory advantage over rival views thatneglect the historical and dialectical dimension of truth and reason(O’Neill 1989b: 70–71; Arruda 2016). This requires theconstructivist to provide an account of truth and objective knowledgeas altering in time. O’Neill’s defense of the virtuouscircularity of constructivism identifies a solution to a problem thataffects constructivism in general.
Extending O’Neill’s approach, recent forms of Kantianconstructivism emphasize embodiment and social embeddedness ratherthan ‘rational agency as such’. Some point to theconstitutive role of moral sensibility and practical attitudes in theaccount of rational agency (Bagnoli 2013, 2017, forthcoming; Muddforthcoming). While this line of inquiry sharpens the distinctionbetween constructivist and contractualist reasoning (O’Neill2003a, O’Neill 2003b), others advances a naturalistic and socialaccount, according to which agency is an emergent property –like those found in biology or economics – arising from thecomplex interactions among agents, and the normativity of a specificcontractualist procedure is sustained (Walden 2012, 2018a, 2018b; cf.Smith 2020, 2023).
Finally, Dover and Gingerich (2024) defend a form of constructivisminspired by Simone de Beauvoir, which bears striking affinities withKantian constructivism, but grounds the ‘existential imperativeto will freedom’ in contingent features of our embodiedsubjectivity.
Humean constructivism builds on the Kantian insight that normativetruths are not simply “out there”, but denies that theyare deliverances of practical reasoning (Bagnoli 2002: 131; Street2008a, 2010, 2012; Velleman 2009; Lenman 2010, 2012). The case forHumean constructivism rests on the alleged inadequacy of competingviews: “it is what we are forced to by the untenability ofrealism plus the failure of Kantian versions of metaethicalconstructivism” (Street 2010: 371). Humean constructivistsabandon the claim that moral obligations are requirements of practicalreason. Thus, Humeans maintain that an ideally coherent Caligula whovalues maximizing suffering is conceivable. Such a person would havereasons for torturing people for fun, because the value of humanity isnot a constitutive norm of reasoning (Street 2010: 371; Street 2009,2012).
In contrast to Kantian constructivism, Humean constructivists abandonthe claim that moral obligations are requirements of practicalreason.While Korsgaard holds that proper reflection leads everyone tothe same substantive moral principles prescribing particular actions,Humeans deny that by reflection we can identify any specific moralcontent. The constitutive norms of practical reason may favormorality, but do not require it (Street 2012; Velleman 2009:150–154; Lenman 2010: 192). Thus, Humeans maintain that anideally coherent Caligula whose aim is to maximize people’ssuffering is conceivable. Such a person would have reasons for makingpeople suffer, which is just to say that the value of humanity is nota constitutive norm of reasoning (Street 2010: 371; Street 2009, 2012;Morton 2018; Gill 2025).
Humean constructivism also rejects the Kantian claim that there areuniversal rational norms that bind all rational agents. For instance,Street argues that “the substantive content of a givenagent’s reasons is a function of his or her particular,contingently given, evaluative starting points” (Street 2010;see also Lenman 2010: 180–181). Consequently, “truth andfalsity in the normative domain must always be relativized to aparticular practical point of view” (Street 2008a: 224).Agreement among various practical standpoints is possible but it isnot guaranteed by facts about the nature of reason or the principlesof reason that are authoritative for all rational agents. Humeans holdthat there is nothing alarming about the sort of relativism that theirposition implies (Street 2008a: 245). Even though moral norms are notnecessary requirements of reason, there is a sense in which they arenot contingent because they play a large role in our lives.Furthermore, by analogy with attitudes such as love, which is bothcontingent and compelling, Street argues that the fact that moralcommitments are contingent does not weaken their normative force. Thelatter claim has been disputed on the ground that the driving force oflove should be distinguished from its normative authority (Bratman2012).
Critics argue that Street’s Humean constructivism fails tocapture the diachronic complexity of moral reasons (Jaeger 2015;Guardo 2019), to account for moral improvement and prudentiallearning, and to explain the social dimension and key functions ofmoral evaluation (LeBar 2023). In cases of conflicting judgments, italso appears implausibly committed to denying that there is any reasonto choose among the available options (Levy 2019). More broadly,Street’s account of practical reason underestimates thepervasive role of reasoning and its central function in fosteringagreement and coordination within social life (Bagnoli 2019); this isa weakness in contrast to many other varieties of constructivism thathave emphasized the coordinating function of morality (Gauthier 1986;Copp 2005; Brennan, Goodwin and Southwood 2013; O’Neill 1989,2015).
These objections suggest that Street’s individualist,means–ends conception of coherence cannot secure diachronic orinterpersonal justification, thereby opening the way for alternativeconstructivist strategies that foreground the social and temporaldimensions of normativity.
To address this problem, one might expect Humean constructivists toappeal to artificial virtues or reduce them to social conventions,thereby aligning with conventionalism – the view that moralclaims rest on social agreement and are grounded in the actualpractices of groups within particular traditions, a view oftenillustrated through the metaphor of construction (Wong 2008).
If Street’s difficulties are to be overcome, however, Humeanconstructivism must address, above all, the challenge of securingobjectivity. While constructivism might recognize objectivemind-independent normative truths, these are not mind-independent inthe sense of being available to every agent regardless of whether suchreasons can be derived, logically or instrumentally, from thatagent’s evaluative perspective combined with the non-normativefacts. In contrast to Kantian constitutivism, which defendsobjectivity in terms of the activity of the will, Street insteadexploits the notion ofvaluing as a philosophical termencompassing a wide range of attitudes and experiential states (2016a:174, 177ff).
Street (2016b) argues that the objectivity of ethics can bevindicated, without metaphysical or epistemological mystery, if thereis a universal problem faced by every agent for which moralityprovides the universal and best (or only) solution. The particularsolution she explores is an ethics of compassion, which dispenses withmoral concepts such as right, wrong, and obligation (2016b: 166), incontrast to the forms of moral constructivism discussed in sections4.1–4.2. Drawing on the Buddhist tradition, she conceives theethical standpoint as “the standpoint of pure awareness,”which addresses the problem of attachment and loss: “when oneoccupies this standpoint, one is identifying with a universal point ofview on the world that transcends any particular, finite point ofview; in doing so, one is identifying with a point of view that is notitself vulnerable to loss” (Street 2016a: 186). This standpointis ethical because it is “that of pure subject – themaximally thin point of view of the ‘one who isaware’” (Street 2016a: 187, cf. Street 2016b, Street2017).
This move addresses some of the issues noted above, but it facesdifficulties in its impersonal characterization of the ethicalstandpoint and its tension with the first-person perspective, on painof alienation (cf. Williams 1981, 1985; Nagel 1986; Wallace 2013). Asimilar issue arises for Salazar (2021), whose constructivist theoryalso draws on Buddhist ethics. She recognizes strong affinities withKantian constructivism but proposes a dual source of normativity. Thisduality, she argues, rectifies some of the difficulties ofautonomy-based morality by redirecting attention on the suffering ofothers, thereby challenging the individualistic conception of the selfas separate and impermeable to others (2021: 175–176).
Other strategies to enhance Humean constructivism leads to hybridtheories because they incorporate elements external to this tradition;these will be examined in section 4.3. In section 5, we will considerhow these varieties of constructivism relate to antirealism andexpressivism, and in sections 6.2–6.3 we will discuss thefeatures of Humean constitutivist strategies.
The debate over the viability of Kantian and Humean constructivism hasgiven rise to a range of hybrid theories, so called because theyincorporate non-constructivist elements drawn from diversephilosophical traditions, particularly in their accounts of moralnormativity and moral rationality, thereby shifting the place ofconstructivism in the traditional metaethical map. In this section, weexamine forms of constructivism limited to moral principles of rightand wrong but combined with non-constructivist accounts of reasons(4.1), or grounded in a realist metaphysics (4.2). Section 4.3considers objectivist developments of Humean constructivism addressingthe quest for objectivity articulated in section 3. Section 4.3. turnsto Aristotelian constructivism, which extends beyond moralnormativity, but rejects Kantian and Humean formalist accounts, andaligns with a modest variety of normative realism. Section 4.5discusses Hegelian constructivism, which sharpens some key elements ofO’Neill’s developmental conception of practical reason,but points to actual ontological structures underlying it.
Thomas Scanlon defends a restricted constructivist account ofjustification for a specific class of moral judgments of right andwrong (1998: 11–12, ch. 4, §7.2; 2008; 2014: 94–98).Unlike Korsgaard and O’Neill, he rejects Kantian constructivismas a general metaethical view about all normative truths. Moralquestions, he argues, cannot be settled by appealing to the barestructure of rationality; they require substantive argument (2003b:14–15; 2014: 90–104) – a point many Kantians wouldconcede. His aim is to explain the truth of claims about right andwrong in terms of their being entailed from the standpoint of acontractual situation.
The contractualist test holds that an act is wrong if it would beprohibited by any set of principles that no one could reasonablyreject as a basis for informed, unforced, general agreement. This testof rejectability specifies the content of moral principles andexplains why it is rational to adopt them. The correctness ofprinciples depends on a hypothetical agreement among individuals,defined by their motivations and reasoning. There are no correct moralprinciples independently of such agreement: rightness is constitutedby what reasonable agents, under specified conditions, would notreject (1998: 380 n.48). The test allows for disagreement – overstandards for assessing conduct or the reasons supporting them –without entailing relativism, since not all answers are equallyvalid.
Scanlon objects that constructivist accounts of general normativityfall prey to bootstrapping: they attempt to justify the normativity ofreasons by procedures that already presuppose it (2014: 96–104;Wallace 2012). Unlike Korsgaard or Street, he denies that truths aboutreasons are constructed. We reach conclusions about reasons throughnormative reasoning (e.g., reflective equilibrium), but such reasoningrequires substantive judgments as inputs; hence, not all facts aboutreasons can be constructed (2014: 102–103). To this extent, hedefends ‘primitivism’ about reasons (1998: 17, 21; 2013):reasons are mind-dependent features of the world, existing in virtueof considerations standing in a “favoring relation” tomental states such as intentions. While this is a form ofnon-naturalism, it differs sharply from Moorean or robust realism,which grounds reasons in mind-independent facts (2014: 14; cf. Enoch2011a: 112–113; Smith 2023). For Scanlon, truths about reasonsdo not require metaphysical guarantees beyond what sound normativereasoning secures. There is no “external” question abouttheir existence, any more than there is about mathematical facts(2014: 16–26). In this respect, his view bears affinities withCarnap’s (1956).
Society-based constructivism, developed by David Copp, holds thatthere are true moral standards produced by a decision procedure thattakes into account the needs and values of a society, together withfacts about its circumstances (Copp 1995, 2007). Moral truth thusdepends on what it would be rational for societies to choose.
Copp’s view shares several features with Kantian constructivism.First, it maintains that societies require their members to endorse asuitable moral code in order to sustain cooperation, therebyconceiving of morality as a cooperative enterprise and grounding theneed for objective moral standards in practical considerations.Second, it explains moral truth in procedural terms, denying theexistence of moral facts independent of such procedures (Rawls 1980:307). Third, it insists that any adequate metaethics must account forthe normativity and practical relevance of moral claims (Copp 2007:4–7), holding that moral obligations bind us regardless of ourmotivational states.
Unlike Kantian constructivism, however, Copp’s account does notcommit to a specific conception of autonomy. Finally, he defendssociety-based constructivism as both realist and naturalistic: realistin treating moral propositions as truth-evaluable and some moralproperties as instantiated, and naturalistic in claiming that theseproperties are natural ones (Copp 1995; 2013).
Recent developments in Humean constructivism have produced hybridtheories that borrow from different traditions, and whose metaethicalstanding diverges significantly, partly in response to theshortcomings highlighted in section 3 regarding the quest for moralobjectivity.
First, James Lenman (2012, 2024) develops a hybrid account thatcombines constructivism, treated as a first-order account of thejustification of moral reasons, with an expressivist semantics ofethical judgment. Although Humean in orientation, his view convergeswith Kantian constructivism in affirming the unifying role of reasonin agency and in offering a normative procedure embodying ideals ofequality and freedom – what he calls “a form ofexpressivism articulated in reason.” Unlike Street (2008),Lenman rejects immoralism, insisting that cases such as torture forfun must be ruled out not by brute preference but by co-deliberationwithin a diachronic community of selves. This requires a modestprinciple of agency-unification: to deliberate normatively at all,agents must integrate their commitments across time, though withoutinvoking any essentialist conception of the self. On this view,failures such as Caligula’s reflect political collapse ratherthan reflective defect.
For Lenman, constructivism addresses expressivism’s centralchallenge – explaining moral objectivity – but only bysubstantive normative means, so it cannot serve as a metaethics on itsown. Expressivism provides the metaethical underpinning, whileconstructivism supplies the normative procedure. Together theyvindicate objectivity without metaphysical costs and preserve thepractical significance of moral judgment without reductive naturalism(2012: 217–18). In this case, constructivism, ensures theadvantages of Blackburn’s quasi-realism (1993): normativediscipline transforms passions into judgments governed by stability,coherence, and commonality, which, if it succeeds, can be truth-aptand objective (Lenman 2012: 220; cf. Street 2011). Yet a fundamentaldifficulty remains: the nature of desire is either an evaluativenotion or a merely psychological one – an ambiguity that must beresolved if Humean constructivism is to advance (Scanlon 1998:7–8, ch. 1).
A second attempt to articulate the quest for objectivity in Humeanterms, proposed by Julia Driver (2017), extends the individualstandpoint through general rules grounded in the practical perspectiveof social creatures, thereby accommodating the possibility of moralfailure, self-correction, and practices of social alignment. Itpurports to improve over Street’s constructivism and Kantiantranscendental arguments by explaining the pressure toward generalityas a natural phenomenon rooted in human sociability. However, thestrategy presupposes an indeterminate account of “humannature”, which is problematic (Driver 2017: 176). Second, itsgeneralizing resources, grounded in sympathy, may fall short ofsecuring a stable relation between morality and the self (Driver 2017:180–81). Third, the key question is whether the Humean appeal togenerality can vindicate its normative significance, particularly incases of moral disagreement where the demand for correction requiresjustification (Korsgaard 2008: 278–79). The Humean approach isthus hard pressed to explain how social animals with capacities suchas sympathy become bound by norms of sympathy, or why they have reasonto endorse the general point of view.
Finally, Dale Dorsey (2018) proposes ‘perfectionist’constructivism inspired by Hume’s esthetics. This view holdsthat individual attitudes are constructed out of humanity’sshared evaluative nature. Like its Kantian counterpart, theperfectionist variety of Humean constructivism holds that some moralreasons apply universally but denies that obligations are groundedsolely in rationality (Dorsey 2018: 591). The immoralist’s errorlies not in incoherence but in being socially dysfunctional andself-defeating, undermining our shared self-understanding as sociallyproductive beings. A qualified agreement by those whose sentiments are“in a sound state” counts as “evidence that humannature issues a particular verdict in a given case” (Dorsey2018: 585). This formulation is problematic for three reasons: it iscircular in presupposing a common core that should itself beconstructed; it rests on the questionable assumption that actualpractices converge, ignoring both the difficulty of achievingconvergence and the possibility that it may result from irrelevantfactors; and it excludes the prospect of divergent yet equally validmoralities, which should at least remain conceptually open even ifultimately rejected on practical grounds. The perfectionist strategyaligns Humean constructivism more closely with moral realism bypositing a subject matter independent of the agent’sconstruction.
Despite these refinements aimed at capturing the social significanceof moral and practical reasoning, it remains doubtful whether thesehybrid views have substantially advanced beyond Street’sposition.
Aristotelian constructivism holds that our true normative judgmentsrepresent a normative reality, but this reality is not independent ofthe exercise of moral and practical judgment (LeBar 2008: 182;2013a,b). Like the Kantian varieties of constructivism, Aristotelianconstructivism appeals to constitutive features of practicalreason:
practical truth is constructed, not discovered, because it is activityin accordance with the norms of practical rationality, which arethemselves constitutive of agency. (LeBar 2008: 191)
Unlike Kantian models, however, it rejects formal and proceduralprinciples in favor of a substantive account of the good life, rootedin eudaimonism. Rational agents set the standard of practicalrationality through the exercise of intellectual and moral virtues(Aristotle,NE II.6; cf. Berryman 2019). The negative casefor Aristotelian constructivism targets the Kantian view of practicalreasoning as law-like, governed by universality. Critics argue thatKantians struggle to apply universal principles to concrete cases(Höffe 1993; LeBar 2013b; Millgram 2005: ch. 6). Aristoteliansclaim an advantage here: their account allows practical reasoning toadapt to particulars and offers a framework for norms of successfuljudgment, an area where Kantian ethics is seen as deficient (Millgram2005: ch. 6). Kantian ethics is often criticized for either rigorismor empty formalism, and though constructivists such as O’Neill(1975) and Herman (1993) attempt to address the role of circumstances,Aristotelians contend these efforts remain inadequate (LeBar2013b).
To identify the substantive standards of practical reasoning,Aristotelian constructivism starts with a study of the complexity ofour rational animal nature, which excludes that the principlesconstitutive of human rationality can be merely formal. In contrast toKantian self-legislation, Aristotelian constructivism emphasizes theinterplay between rational and animal nature, focusing especially ontraining and shaping the affective and sensitive aspects of ournature. In particular, it emphasizes the transformative effects ofreflection on passions and desires, and the possibility of developinga ‘second nature’, thanks to complex processes such ashabituation and education (LeBar 2008: 197). Practical rationalitydoes not merely direct affective responses toward adequate objects butalso structurally transforms our animal sensibility into character.Aristotelian constructivism rests on two central concepts: the goodlife (eudaimonia) and practical wisdom (phronesis),whose contents are progressively defined through their mutualinteraction (LeBar 2023: 479). While general standards exist, inparticular cases it isphronesis that determines them. Thereare no independent facts that establish what a good life is apart fromthe judgment of the wise. Such judgments are substantive rather thanformal, and particular rather than universal.
Hegel-inspired ethical constructivism affirms the primacy of socialprocesses over individual agency, though they diverge on how toconceive this priority, producing different positions toward realism.Some interpret Hegel as claiming that moral normativity is“invented” (Pippin 2008), while others insist that humannature is always already shaped by ethical order (Laitinen 2020;Thompson 2020). To stress antirelativist commitments, one strandretains a qualified normative realism, weakened or historicallymediated (Rockmore 2016; Werner 2017; Gladhill & Stein 2020;Ostritsch 2020; Laitinen 2016, 2020). Others treat the social natureof constructions as neutral but objectivist (Westphal2003, 2016,2017), while some, following Korsgaard, argue that constructivismdestabilizes the realism/skepticism dichotomy (Wretzler 2020).
In stark contrast to formalist and transcendent aspirations,Neo-Hegelians emphasize actuality: reason unfolds through concreteforms while leaving space for alternative or thwarted paths (Ng 2009;Moyland 2011; Yeomans 2012; Redding 2020). This yields a robustaccount of practical reason that affirms its historicity and embedsautonomy in complex social processes (Bildung). On this view,the constraints of practical cognition are socially grounded, arisingfrom the norms of a community of rational agents at a given historicalmoment. Rational agency is thus not reducible to self-relation butrequires recognition and engagement with others. Autonomy cannot besimply assumed but must be situated within a sociopolitical reality,prompting inquiry into its ontological constitution.Self-determination and choice become possible only within suchcontexts, and the authority of reasons varies across socialformations. Constructivism thus becomes a theory of social agency,foregrounding autonomy as lived within institutions.
The Hegelian notion of ‘sublation’ (Aufhebung)designates a process of preservation and transformation. It impliesthat social constructions, though indispensable for rational agency,are not ultimate normative standards but concrete frameworks withinwhich reasoning unfolds, shaping practical truth through structural,epistemic, and contextual functions. Ethical life thus preserves pastcommitments that continue to shape collective and individualidentities.
A strength of this model is its fallibilism: norms may obscureobjective contents, yet practices remain open to critique andcorrection. Here moral sensibility – especially the experienceof suffering – plays an epistemic role in detecting injusticeand empowering ethical and political contestation (Ostritsch 2020;Laitinen 2020). By focusing on concrete formations, this variety ofconstructivism highlights how agents discern the limits of theirhistorical realities, illuminating responsibility, social roles, andhistorically conditioned alienation missed by individualist accountsof reflective endorsement.
The view that ethical norms are social artifacts leans towardantirealism but struggles to vindicate their truth-claims. Purelysocial constructivism, lacking ontological grounding, faces twoobjections: relativism, since reasons bound to sociohistoricalcontexts risk losing critical force, and reification, since reasonsembedded in practices may appear immune to scrutiny (Thompson 2020).Hegelian approaches reply by positing a social ontology: human beingsare essentially social, and their relations constitute reality itself,historically reshaped in diverse forms (Redding 2020; Thompson 2020).On this account, objectivity is internal to concrete social formationsrather than the contingent product of convention. Its strength lies inavoiding relativism while grounding normativity socially; its weaknessis reliance on agents to apprehend reasons expressive of the systemictotality of their social world. Whether this confidence is justified,and how it avoids reification, remains an open question.
Constructivism has become a central position in both practicalphilosophy and metaethics, appealing for its attempt to secure moralobjectivity while avoiding the epistemological and ontological burdensof nonnaturalist realism (Darwall, Gibbard, & Railton 1992;Shafer-Landau 2003, ch. 4). Yet this aim is not unique toconstructivism: naturalist realism also promises objectivity withoutnonnaturalist commitments, and antirealist theories likewise defendobjectivity on non-ontological grounds (Hare 1952; Gibbard 1990;Wright 1992). What distinguishes constructivism is its potential toreconcile the key features of normative truths – theirobjectivity, their intelligibility, and their practical authority(Scanlon 2014: 91).
Thus far, constructivists have not developed a distinctive account ofthe meaning or logical behavior of moral and normative terms. In theabsence of a constructivist semantics, some suggest thatconstructivism is best seen as “a family of substantive moraltheories” (Darwall et al. 1992: 140; Hussain & Shah 2006,2013; Enoch 2009; Hussain 2012). On this view, constructivism is not ametaethical theory, since by treating normative concepts as solutionsto practical problems it remains within the normative domain (Ridge2012). This objection, however, assumes a sharp boundary betweennormative ethics and metaethics. Constructivists reject this, aiminginstead to clarify the reasons that make moral principles ones no onecould reasonably reject (Scanlon 1998: 246–247; 2003a:429–435). The dispute thus turns on what counts as metaethics.Some constructivists even regard the semantic focus of traditionalmetaethics as misguided, diverting attention from the real issue– the nature of normativity itself (Korsgaard 1996a; 2003: 122fn.49; Street 2008a: 239), and hence call for a break with the“platitudes” of metaethics (Korsgaard 2003: 105).
Still, some constructivists pursue also a semantic task from within abroader understanding of metaethics. Some argue that identifying theconstitutive norms of valuing provides the basis for an inferentialistsemantics, whereby normative terms are explained through the kinds ofinferences – such as those about means and ends – requiredto count as employing them at all (Street 2010: 239–242). Thiscommits constructivists to showing that their proposal improves onrival accounts (cf. Jaeger 2015). Others articulate a pragmatistvariety of constructivism about normative discourse (Richardson 1994,1995, 1998, 2008, 2013, 2018; Elgyin 1997; Misak 2000; Schwartz 2017).Müller (2020) defends a theory of reasoning as requiring reasonjudgments, where reason judgments are not representational yet stilltruth apt, analysing the nature of these judgments in functionalistterms as guiding reasoning in a certain way (cf. Frugé 2021;Silverstein 2022).
Constructivists must deny the correspondence theory of truth, the viewthat truth is correspondence to a fact. At the same time, they canallow that normative judgments are truth-apt. This view is hospitableto the constructivist claim that moral and normative truths may alterover time, as a result of ongoing rational deliberation and revision(Richardson 2013; LeBar 2013b; O’Neill 1992, 2015; Laitinen2016). Some attempts to deal with semantic issues bring to light aresemblance between constructivism and pragmatism, which rejects theidea that truth is correspondence to a mind-independent world,stressing instead that truth is tied to inquiry responsive to argumentand evidence: it is the state of belief that, after full inquiry,would not be improved or defeated, although we me not know when wehave reached such a belief. Human inquiry is fallible and provisional,but truth is tied to the ideal of inquiry’s completion (Misak2000, cf. Richardson 1998). Like pragmatism, constructivism appeals tothe practical point of view to account for truth, in contrast tostandard forms of realism about truth (Proulx 2016; Elgin 1997;Richardson 2013; Schwartz 2017). Some critics are skeptical about thepossibility of developing a constructivist account of truth (Hussain2012: 189ff; Dorsey 2012). Schwartz and Velasco (2019) develop asemantic account of metanormative constructivism, where moral claimsare true if endorsed by idealized agents through practical reasoning.Their framework clarifies how constructivism can treat moral judgmentsas truth-apt and normatively authoritative without appealing toindependent moral facts.
The varieties of constructivism occupy different positions in therealism–antirealism debate. Kantian constructivism has been castas a third option between realism and subjectivism (Rawls 1980), orbetween realism and relativism (O’Neill 1989a; Korsgaard 1996a:36). Some interpret constructivism as realist (Copp 2013; LeBar 2013,2023), others as antirealist (Lenman 2010, 2024), while still othersdefend hybrid views, such as “prioritism,” which combinesrealism about reasons with constructivism about moral judgments(Scanlon 1998, 2014, 2003b: 18). Street (2006, 2008a, 2009, 2012)places constructivism on the antirealist side, even embracing a formof relativism. This diversity suggests it is a mistake to assessconstructivism’s significance solely in terms of therealism–antirealism debate (Korsgaard 2008: 312, 325 n.49; Copp2013; Engstrom 2013: 138ff).
Part of the difficulty lies in divergent definitions of realism(Sayre-McCord 2015; Joyce 2015; Miller 2014; LeBar 2023; Smith 2018).Realists generally hold that (a) moral discourse is cognitivist, (b)there are moral properties such as rightness, (c) some moralproperties are instantiated, and (d) moral predicates expressproperties. Disagreement arises over whether realism requires thestronger claim that (e) moral properties are mind-independent. Furtherdivisions concern naturalism, which holds that (f) moral propertiesare metaphysically on a par with non-moral properties and (g) moralassertions express ordinary beliefs about their instantiation.Constructivists generally accept (nonreductive) naturalism but somereject (e): they hold that normative properties depend on humansensibility or rational agency rather than being mind-independent.Like realist naturalism and antirealism, constructivism is committedto metaphysical parsimony, distinguishing it from Moorean and robustmoral realism (Enoch 2011a).
Korsgaard highlights a shared assumption of both realism andantirealism, which constructivists reject: that the function ofconcepts in truth-apt judgments is representational, such thatnormative truths must refer to something “out there.”Constructivists instead claim that normative concepts serve apractical function: they articulate solutions to practical problemsrather than represent moral facts (Korsgaard 2008: 302ff). Thus,equity, for example, does not name a property but proposes a way ofdistributing goods. Constructivism therefore differs from substantiverealism, which grounds truth in a mind-independent normative reality,and from antirealism, which denies normative truth altogether:practical judgments can be true or false without representingindependent normative facts (Korsgaard 2003: 325 n.49).
This characterization, however, does not neatly captureconstructivism’s place. First, Korsgaard equates realism withmind-independence, but many define it more broadly to include viewswhere normative truths depend on mental states (Sayre-McCord 1988;Copp 2007: 7; LeBar 2023). Second, some antirealists allow that moraljudgments are truth-apt but treat truth as a deflationary semanticnotion (Ridge 2012; Lenman & Shemmer 2012b). Constructivism thusmarks a middle ground between forms of realism committed tomind-independent truths and forms of antirealism that deny normativetruth altogether. In this sense, it resembles “sophisticatedsubjectivism” (Wiggins 1993; McDowell 1985), though Kantianconstructivism grounds normative truth in rational agency rather thansensibility.
Some forms of Humean constructivism align closely with expressivism,which treats normative language as guiding action rather thanrepresenting facts (Chrisman 2010; Meyers 2012; Ridge 2012; Tiberius2012). Lenman (2012) argues that this link helps constructivismaddress the problem of identifying the mental states expressed bynormative judgments. Yet many constructivists deny that expressivismcan capture the authority of normativity, reducing moral discourse tomere “emotional expletives” (Korsgaard 2003: 105) or tonormative states (Street 2010: 239–242; Korsgaard 2003, 2008,2009; LeBar 2023). For Korsgaard in particular, the problem ofnormative authority must be addressed from the first-personperspective, something both realism and expressivism fail to achieve(1996a; 2008: 30–31, 55–57, 67–68, 234). Otherconstructivists doubt that expressivism can capture the socialdimension of moral evaluation (LeBar 2023: 472, 479), and endorse amodest variety of realism. Some hybrid theories endorse explicitlymoral (Copp 1995) or normative realism (Rockmore 2016; Werner 2017;Gladhill & Stein 2020; Ostritsch 2020; Laitinen 2016, 2020) oralign to it (Driver 2017; Dorsey 2018), while others combineconstructivism with expressivism (Lenman 2012), or reject therealism/antirealist dichotomy (Westphal2003, 2016, 2017, Wretzler2020).
Constructivism explains practical truths through the constitutiveactivity of reasoning captured by the notion of“construction.” It provides “a potentiallydevastating argument against realism” (Shafer-Landau 2003: 49),showing an explanatory power that realism lacks (Shafer-Landau 2003:51). But all these comparative advantages are nulled by anEuthyphro-style dilemma: if reasoning is unconstrained, its resultsare arbitrary; if constrained, the constraints are not constructed andentail realism (Shafer-Landau 2003: 42). Thus, constructivism appearseither arbitrary or collapses into realism.
This objection can be formulated and motivated in at least two ways,which will be examined in sections 6.1–6.2, while section 6.3will discuss the viability of constitutivism as an account of moralnormativity.
The first objection is that constructivism relies on an“unconstructed,” and thus unjustified, set of normativeconstraints. Kantian constructivism, for example, seems grounded inthe value of moral impartiality, which demands equal respect forpersons (Scanlon 1998: 22–33, 287–290; Rawls 1993:38–54). The worry is that constructivism risks vacuity, since“its test yields results only by presupposing moral views whichcan only be established independently of it” (Raz 2003: 358;Hare 1983; Brink 1987, 1989; Timmons 2003; Cohen 2003). Moreover, asTimmons argues, there is “no non-question-begging feature towhich the constructivist can help herself in breaking symmetry amongthe various competing sets of constructed principles” (2003:§3).
Kant is keenly aware of the air of paradox surrounding the claim thatthe moral concepts, such as good and evil, are not determined prior toengaging in practical reasoning, but only as a result of engaging inpractical reasoning. This is known as the ‘paradox of themethod’ (Kant C2, 5: 62 ff.). As Rawls explains it,
The difficulty is that Kant appears to know in advance of criticalreflection how a constructivist doctrine might look, but this makes itimpossible to undertake such reflection in good faith. In reply tothis quite proper worry, we interpret constructivism as a view abouthow the structure and content of the soundest moral doctrine wouldlook once it is laid out after due critical reflection. (Rawls 2000:274)
Practical reasoning, then, is not a matter of discovering valuesindependent of its verdicts. For Kantian constructivists, it yields aform of self-conscious practical knowledge, intelligible only throughaccounts that treat cognition as self-conscious activity (Rawls 2000:148, 218; Engstrom 2013; Bagnoli 2013b; Jezzi 2016 [Other InternetResources]). Its novelty lies in explaining objective moral knowledgewithout presupposing an independent moral reality, but insteadgrounding it in reason’s own activity.
Debate persists about the role of intuitions in justification. ForScanlon (1998), appeal to intuitions is unavoidable: justificationproceeds by testing the fit between theoretical assumptions andintuitive moral judgments, which serve as initial data withoutconstituting independent truths. Others, such as O’Neill(1989b), deny intuitions any justificatory role. Still, mostconstructivists agree that metaethics must remain broadly congruentwith common understandings of rationality and morality (Smith 2013).In this respect, constructivism distinguishes itself from projectivistor error theories (Blackburn 1993; Mackie 1977), which regardevaluative discourse as systematically mistaken (Bagnoli 2002; Street2011; Lillehammer 2011).
A second formulation of the objection that constructivism avoidsarbitrariness by tacitly relying on unconstructed elements targets theconstitutivist strategy. How are constitutive norms justified? Theobjection holds that they are either arbitrary or realist, dependingon normative features of reality that are unconstructed (Fitzpatrick2013; Stern 2013; Baiasu 2016; Bratu & Dittmeyer 2016; cf. Bachman2018).
In providing an account of rational justification, both realists andconstructivists rely on unconstructed elements. Normative ornon-naturalist realists take normativity as primitive: certain factssimply are reasons, and “there won’t be any illuminatingexplanation” of what makes this true (Shafer-Landau 2003:47–48). Reductive naturalists, by contrast, identify normativefacts with natural facts, accessible through empirical methods –for instance, facts about agents’ responses under idealizedconditions (Smith 2012, 2013).
Constructivism, while compatible with non-reductive naturalism(Korsgaard 1996; Walden 2012), does not reduce moral properties tonatural ones but seeks to explain the standards of practical reason(O’Neill 1989b; Korsgaard 2003, 2009; Street 2010; Velleman2009). Unlike (naturalist and normative) realism, however,constructivism anexternal foundation of normative standards.Internality is construed in two ways. According to Korsgaard,
“Constitutive standards are opposed to external standards, whichmention desiderata for an object that are not essential to its beingthe kind of thing that it is” (Korsgaard 2008: 8).
A more modest interpretation insists on the role of internal standardsin practical reasoning, which is “avowedly circular”:
If the standards of practical reasoning are fundamental to all humanreasoning, then any vindication of these standards is either circular(since it uses those very standards) or a failure (since it is not avindication in terms of the standards that are said to befundamental). (O’Neill 1989b: 29)
The latter interpretation stands in contrast to the traditional“first principles philosophy”, which derives moral truthsfrom principles that have the status of axioms (cf. Schneewind 1970).The shared constructivist conviction is that appeal to standardsconstitutive of agency explains with ease why such standards arenormative (Korsgaard 2008: 8). They are normative because in the veryactivity of reasoning we are committing ourselves to being guided bythem. In forming our intentions and beliefs, we are answerable tocriteria of correctness that are internal to and constitutive of thevery exercise of rationality (Korsgaard 2008: 13–15,110–126, 207–229).
The place of constitutivism – and correspondingly ofconstitutivist constructivism – within the traditionalmetaethical landscape remains contested. Some regard constitutivism asbroadly realist, on the grounds that moral judgments express beliefsthat can be true, though not in the robust realist sense (Smith 2013,2018, 2020, 2023). Others interpret it as compatible with expressivism(Silverstein 2016; Ridge 2018). Varieties of Humean and Aristotelianconstitutivism or constructivism are often mind-dependent (Street2010, cf. 2016b; LeBar 2023), while others are mind-independent,appealing either to natural objects (Smith 2023) or to practical laws(O’Neill 1989; LeBar 2023: 471). Some versions are idealized(Smith 2013, 2020, 2023; Street 2016a), while others are non-idealized(O’Neill 1987, 1989; Walden 2018a, 2018b; Bagnoli forthcoming).Appeals to idealized rationality are often charged with circularity,since they offer no independent, non–question-begging reason forthose very standards. For some, idealization is a normative matter;for others, it concerns facts about the psychology of idealizedagents, which are themselves conceived as natural objects (Smith2023). Further disagreement arises over the definition of idealizedagency: for some, it includes features such as self-control andresistance to intra- and interpersonal conflict (Smith 2023: 101; cf.Lindeman 2019, Coleman 2023), while for others it involves thedissolution of the self (Salazar 2021; cf. Street 2016a). All forms ofconstitutivism are at least compatible with non-reductive naturalism,though some take a reductive naturalist form, and others may evencommit to essentialism. Some constitutivist accounts support moralobjectivity – either in a robust form that coheres withfirst-person deliberation (O’Neill 1989; Korsgaard 2009;O’Hagan 2014) or in an absolutist form (Smith 2023) –while others are relativistic (Street 2008, 2010). Certain versionsadopt a moral rationalist framework, according to which moralrequirements entail corresponding reasons for action, are knowablea priori, and are grounded in practical reason. For some,however, this capacity is reduced to facts about the psychology ofideal agents (Smith 2023: 103–104). In the next section, we willlimit the discussion to the constitutivist strategies relevant toconstructivism because constitutivism can be defended as a theoryindependent of (meta-ethical and moral) constructivism, as the generalview that the view that we can justify fundamental normative claims byshowing that agents become committed to these claims merely in virtueof acting (see Katsafanas 2019; cf. Ferrero 2021).
The constructivist use of constitutivist strategies in defending moralnormativity offers at least three advantages. First, it sidesteps theinternalist/externalist distinction by directly linking rationalitywith agency. Second, it resolves the problem of authority bydispensing with the need for external facts to legitimate claims ofnormative force. Third, it vindicatesa priori moral andpractical knowledge – providing rational justification for moralverdicts – while remaining consistent with a minimalist orconstructivist metaphysics of the moral domain (Walden 2018;Katsafanas 2018; Schroeter, Jones, & Schroeter 2018: 8; Ferrero2021).
An outstanding question is whether the constitutivist strategysuffices to account for moral obligation. Constructivists are devidedabout this claim. Korsgaard (2009) holds that the principles governingaction are “constitutive standards” of agency, that is,standards arising from the nature of agency itself, which explain thenormativity of moral obligations. The marks of agency are autonomy andefficacy, and the categorical and hypothetical imperatives are normsgrounded on these two properties. These norms are“constitutive” of rational agency in that (i) theydetermine what a rational agent is (the “constitutionrequirement”), and (ii) they empower an agent to realize heridentity as the particular person she is, by guiding her conductaccording to those standards (the “self-constitutionrequirement”). According to Korsgaard (2019), Kantianconstructivism is uniquely positioned to vindicate the self-constitution requirement while Humean and Aristotelian theories failto meet it. Yet the Kantian claims about the rational authority ofmorality are found puzzling (Brink 1992).
Humean Constitutivism is more modest (Street 2006), in that isacknowledges only the principle of logical consistency. Thedifferences among these two constitutivist strategies can beillustrated by comparing their respective diagnoses of the immoralist(Street 2010: 371; cf. Street 2016a). Humean constructivists admitsideally coherent eccentrics, like Caligula who “aims solely tomaximize the suffering of others”; for such agents, there arereasons for torturing others (Street 2009). Kantian constructivistsdeny that there might be reasons for maximizing the suffering ofothers but provide different arguments in support of this claim. Someargue that Caligula’s incoherence can be shown by spelling outthe norms that are constitutive of valuing. Such constitutive normsentail valuing humanity, and this proves that Caligula is mistaken byhis own lights, even though he may never fully realize this, due topoor reflection, ignorance of the non-normative facts or some otherlimitation (Korsgaard 1996a: 121–123; cf. Street 2016a). Adifferent Kantian argument establishes that an internally coherentCaligula is conceivable, that is, he can be thought withoutcontradiction, but his case is blocked by general facts about moralsensibility and interdependency, and coordination requirements(Engstrom 2009: 243). Ultimately, there are no reasons to inflictmaximal suffering. Crucially, unlike in realism, this conclusion doesnot depend on the existence of moral truths that precede or standindependent of practical reasoning (Street 2016a).
By itself, consistency is too thin to account for the kind ofconstraints that rational agents need to impose on the dynamic processof goal management (Gibbard 1999; Smith 1999). To account for thesenormative constraints, some add a broad principle of coherence, whichdemands a unified account of the agent’s goals, and thus goesbeyond strict logical consistency (Shemmer 2012; Lenman 2012). Othersargue for additional but related principles, which regulate attentionand disregard (James 2007, 2012), benevolence and non-interference(Smith 2013: 322, 328). For Kantians, however, that the normativity ofinstrumental principles of rationality rests on the normativity ofnon-instrumental principles (O’Neill 1989b: 73–74;Korsgaard 1997, 2008: esp. 67–69), expressing respect for othersas having equal standing (Bagnoli 2013b).
Disagreement persists over the scope and role of constitutivism inKantian ethics (Bagnoli and Bacin forthcoming). Some view it as aresponse to skepticism about practical reason (O’Neill 1989;Korsgaard 1983, 2019; Engstrom 2009; Bagnoli 2013a), while othersargue it misfits Kant’s project (Gobsch 2019; Saemi 2016;Tenenbaum 2016). Critics contend that the categorical imperative, as aconstitutive standard, is too thin to yield determinate moralobligations (Nozick 1981: 545–551; Cohen 1996; Bratman 1998;Gibbard 1999: 149, 152–153; Fitzpatrick 2005; Scanlon 2007). Infact, most Kantians deny that moral obligations can be derived fromuniversal features of bare rationality alone, and also deny thatappeal to constitutive norms of rationality is sufficient to provide acomplete system of moral duties (O’Neill 1989; Sensen 2013,2017; Timmons 2017; Bagnoli 2021).
Constitutivism faces serious challenges that threaten to undermine theconstructivist project in all its forms. The first concerns the statusof constitutive standards. These standards are meant to be bothnormative and, at least in part, descriptive of the very activity theygovern (Korsgaard 2008: 9). Critics argue, however, that it is unclearwhether – and how – such standards can be violated (Cohen1996: 177). For an agent to count as subject to norms, she must alsobe capable of violating them. Suppose, for instance, that there is anorm prohibiting deception: to say that a rational agent is guided bythis norm is to say that she can fail to comply with it. Yet if thenorm is constitutive of reasoning itself, how could it be broken inthe very act of reasoning? If constitutive norms are in principleinviolable, then constitutivism seems to entail that both immoralismand irrationality are impossible (Lavin 2004; cf. Cokelet 2008).
A related objection concerns the very possibility of bad action(Katsafanas 2018). Korsgaard responds that in order to count as actingat all, we must at least be attempting to follow the principles ofpractical reason, though we may fail to do so adequately or fully(Korsgaard 2009: 45–49, 159–176; Pauer-Studer 2018). Forexample, one may adopt the moral end of benevolence yet pursue itwithout grace or understanding, thereby provoking resentment ratherthan eliciting gratitude. Thus, a mistaken judgment about whatself-defense requires mischaracterizes the very nature ofself-defense. The issue is whether such defective approximationsconstitute transgressions of normative requirements, rather than merefailures of what is expected of agents. Addressing this requires amore radical strategy: a constitutivist metaphysics of capacities thatexplains error as an imperfection of action (Fix 2020) and chartsdifferent modes of alienation from constitutive norms (Tenenbaum2019). Herman suggests that in defective instances of action,constitutive norms are not misapplied but misrepresented (Herman 2007:171–172, 245–246).
A second objection is that the constitutivist strategy depends on theclaim that agency is inescapable, whereas critics argue that agency isno less optional than any other activity (Enoch 2006; Tiffany 2012;Leffler 2016, 2019, 2024a,b). Constitutivists cannot simply appeal tothe constitutive role of agency to explain why its rules bind, forinescapability by itself provides no normative force (Kolodny 2005).On this view, the authority of rational requirements is illusory: theymerely track antecedent reasons rather than generate new ones.Constitutivists reply that the objector overlooks the inescapabilityof agency itself. While it is possible to disengage from anyparticular activity, some form of agency always remains operative(Ferrero 2010a; Velleman 2009: 138–141; James 2012). In thissense, agency is not optional. Enoch counters that even if some formof activity is always present, this does not entail that the normsconstituting such activity yield normative reasons for any particularagent (Enoch 2011b). The constitutivist rejoinder is that the veryquestion of whether there are reasons to be agents can only arisewithin agency itself (Velleman 2009: 204–206). One does not needa reason to be an agent rather than not, and thus Enoch’schallenge is not a genuine or live question (Rosati 2016: 201 n.71;Silverstein 2015: 1136–1138). For Ferrero (2019) the appeal toinescapability works best as a defensive move but fails to sustainrobust moral normativity.
These replies are effective at least against some versions of theshmagency objection (Arruda 2017; Paakkunainen 2018). Whether agentshave conclusive reasons to be agents, however, might depend on theparticular version of constitutivism. For most constitutivists, thisinvolves grounding authoritative norms in the teleological structureof agency (Korsgaard 2009; Engstrom 2009). For others, theconstitutivist project can be salvaged only if it is supplemented witha reductive metanormative account of reasons for action, which linksreasons to sound or successful practical reasoning (Silverstein2016).
A third objection is that treating moral obligations as rationalrequirements does not resolve the question of their normativeauthority (Broome 2005; Kolodny 2005), unless such requirements areshown to be either intrinsically normative or valuable independentlyof their contingent benefits. Meeting this challenge thus depends onvindicating the intrinsically normative character of rationalityitself (Williams 1981, 1985, 1994). This debate is rooted in afundamental disagreement about the scope and powers of practicalreason.
Finally, other worries arise from Bratman (2012)’s charge thatconstructivism overconstrains deliberation, producing tension betweenreflective and prereflective judgments, especially when loyalties orattachments conflict with impartial duties. This is a problem evenwhen for the modest Humean constitutivism that recognizes only aformal requirement of consistency. If the input judgments includeattitudes such as love and caring, which are not necessarilyresponsive to intersubjective pressures, agents may end up endorsingreflective judgments that are not aligned with their unreflectivejudgments. This may be taken as a variety of alienation (Tenenbaum2019, Mudd forthcoming, Sensen forthcoming).
This debate indicates the importance of the temporal dimensions ofrational agency (James 2012; Ferrero 2009, 2010b; Smith 2013: 315ff).Recent varieties of constructivism take into account the predicamentsof contingency (Street 2012), the possibility normative revisions(Baldwin 2013, Richardson 2018), moral progress (Arruda 2016), andconcrete forms of normativity (Gledhill and Stein 2020, cf. Bagnoli2022, §4). It is noteworthy that the growing attention to thetemporal dimension of rational agency has been paralleled by anincreased focus on the social dimension of moral evaluation. EarlyHumean and Kantian accounts – especially Street (2006, 2008,2010) and Korsgaard (1996) – have often been criticized foroffering an overly self-centered conception of morality that fails totake its interpersonal character seriously (Cohen 1996; Williams 1996;Meyr 2019). More recent approaches address this shortcoming byemphasizing the social dimension of moral life (Walden 2012, 2018a;cf. Smith 2020, 2023). The foregoing discussion highlights reflectiveequilibrium as a method of moral justification aimed at achievingcoherence among our considered moral judgments, moral principles, andbackground theories. The method is iterative: when conflicts arise,either principles or judgments are revised until a state of mutualsupport is achieved. Although originally developed by Rawls, thismethodological tool has been adopted by constitutivists as diverse asSmith (2010: 134, 136–137, 2013, 2020, 2023) and LeBar (2023),as well as by others (Baldwin 2013; Werner 2017; Walden 2018).
In response to the debates surveyed in sections 5–6, recentconstructivist theories have increasingly embraced hybrid elements,e.g., treating constitutivism not as a comprehensive theory but as apartial account of moral normativity.
New varieties of metaethical constructivism are emerging that draw onthe insights of philosophers such as Spinoza (Zuck 2015), Adam Smith(Stueber 2016), Nietzsche (Katsafanas 2013; Silk 2015), Murdoch(Mylonaki 2018, Mylonaki forthcoming), or Habermas (Rees 2020). Notall explicitly situate themselves within metaethics, independently ofrealism and antirealism, yet they converge in treating the notion ofconstruction as a distinctive explanatory tool for accounting for theobjectivity and normativity of ethical truths.
Over the past three decades, constructivism has established itself asa distinctive and influential position within metaethics. Like otherleading theories, it faces serious objections; yet its capacity toreframe questions about normativity, objectivity, and practical reasonmarks it as a significant and enduring contribution to contemporarymetaethical inquiry.
How to cite this entry. Preview the PDF version of this entry at theFriends of the SEP Society. Look up topics and thinkers related to this entry at the Internet Philosophy Ontology Project (InPhO). Enhanced bibliography for this entryatPhilPapers, with links to its database.
anti-realism |cognitivism vs. non-cognitivism, moral |constructivism: in political philosophy |metaethics |moral anti-realism |moral realism |practical reason |Rawls, John |realism |reflective equilibrium
I would like to thank Caroline Arruda, Christine M. Korsgaard, LucaFerrero, Jeremy Fix, Jacob Librizzi, Elijah Millgram, ThomasPendlebury, Tim Scanlon, Karl Schafer, Michael Smith, Oliver Sensen,Ariel Zylberman, and especially David Copp and Connie Rosati, theEditors and Referees of theStanford Encyclopedia ofPhilosophy, for their comments on earlier drafts.
View this site from another server:
The Stanford Encyclopedia of Philosophy iscopyright © 2025 byThe Metaphysics Research Lab, Department of Philosophy, Stanford University
Library of Congress Catalog Data: ISSN 1095-5054