On the very idea of normative foundations in critical social theory

History of the problem

In traditional Marxist accounts, criticism of capitalist societies was justified by the position of the laboring class. They (we) create the wealth of the society we live in, but because the workers are exploited, they do not benefit from the production of wealth. The working class will eventually overcome the inadequacies of modern capitalism by fulfilling the ideals that capitalism claims to uphold (Postone 2003, 64–83). This justified Marxist theories of revolution “scientifically,” and also politically: the Marxists were on the right side of history. Something like this argument can be found in most of the theorists of the Second International, as well as in Lukacs’ analysis of class consciousness (Jay 2019, 137-8).

According to one common narrative, Friedrich Pollock was able to convince Adorno and Horkheimer that they were living in a new kind of capitalism, for which the state bore as much responsibility as the economy. The economic contradictions that allowed the proletariat to act as the subject of history no longer existed. So, Adorno and Horkheimer rejected class analysis and the progressive philosophy of history. Their Dialectic of Enlightenment described a society that had never been without reification, and which therefore could not generate “the standards for its own criticism” (Chari 2015, 73). That is, Dialectic of Enlightenment suggested that criticism of our society could not be rationally justified by appeal to any standards or criteria (e.g., Cooke 2006, 39–41; Apel 1996, 294). Lukacs responded to this by suggesting that its authors resided in the Grand Hotel Abyss (Jay 2019, 137). Adorno in turn suggested that Lukacs’ tedious prose style was a perfect representation of “that simple repetition or duplication of a dogmatically ossified doctrine which is also quite appropriate to the content of his thought” (Adorno 2017, 211). It seemed that critical social theory was trapped between nihilism and dogmatism.

Jürgen Habermas was the most notable author of this narrative. According to him, Adorno’s Negative Dialectics suggests that “even the faculty of knowledge is not elevated above the frail and damaged state of the subject,” which should lead us to ask “how critical thought can be justified,” and to demand “that the reasons for the right of criticism be made explicit” (Habermas 1983, 106). ² Horkheimer and Adorno should have left “at least one rational criterion intact for their explanation of the corruption of all rational criteria.” That would have allowed them to “set the normative foundations of critical social theory.” Instead, they failed (Habermas 1987a, 127-9). Critical theorists had to take up this challenge: to describe and justify “a standpoint from which society and its institutional practices can be meaningfully theoretically criticized,” as Axel Honneth put it (50). There is room for a lot of argument here, though. Why should the absence of normative grounds be a problem for criticism? Why should we want normative grounds at all? ³

Why would we want normative foundations?

The history in the previous section can be reconstructed as a continuing argument about the status of normative foundations. There are good reasons for rejecting appeals to nature or to God as foundations: they both fall victim to what Karl-Otto Apel calls the “Baron von Münchhausen trilemma” (Apel 1993). Any appeal to an ultimate, metaphysical foundation must be guilty of regress (Hosea calls on the Lord, but who does the Lord call on?); circularity (The Daodejing calls on nature, but we want to know why it calls on nature); or dogmatism, the claim that the “ultimate ground of all things is, quite simply, intuitively obvious,” even though most supposed ultimate foundations are not intuitively obvious to most people (Apel 1993, 172). This trilemma suggests that ultimate, metaphysical foundations are not possible. ⁴

In addition to the problem of possibility, metaphysical foundations do not mesh well with the political and ethical commitments of contemporary critical social theorists. If we assume that “knowledge and validity are always mediated linguistically and that language is conditioned by history and context,” any appeal to a supposedly non-mediated, transhistorical, metaphysical foundation will inevitably appear authoritarian (Cooke 2006, 4). As well as being impossible, metaphysical foundations seem undesirable.

On the other hand, theories without metaphysical foundations often seem circular, incomplete, or relativistic. Honneth, for instance, was concerned about the “functionalist reductionism” of the first generation of Frankfurt school theorists (1995, 70), and this worry is still prominent. Consider a classic functionalist argument: capitalism undermines itself in its very workings, because capitalism is meant to lead to greater productive capacity, but the relations of production do not allow capitalism to fully unleash the forces of production. There is no need to appeal to any rational, normative ground. The problem with capitalism is that it does not work properly. It is dysfunctional.

But such an argument cannot stand on its own. As Rahel Jaeggi has shown, functionalist arguments are necessarily incomplete, because what it means to function is a normative question: “we have to ask: dysfunctional with respect to what?” (Fraser and Jaeggi, 117). Purely functionalist cannot answer this question, and so they cannot convince anyone that there are problems with a system. Some normative standard of what it means to function properly is also required for the argument to work. The claim that capitalism is dysfunctional because it does not unleash the forces of production tacitly assumes the normative claim that unleashing the forces of production would be a good thing and that claim must be rationally justified if it is to give us our normative foundation.

This is not quite the same as saying that a theory of society must have a metaphysical foundation. The functionalist argument can be completed by adding to it a simple statement of interest or desire. I might coherently oppose capitalism, for instance, by noting that it tends to destroy the labor power that is the condition of its possibility; that I am that labor power; and that I do not want to be destroyed. Or, I might argue in a functionalist way that capitalism is premised on the notion of freedom, but that it tends to make people less free, and add that I want to be free.

This response, may however, be open to accusations of relativism. Raymond Geuss suggests that this was a major reason for Habermas’ turn away from the “contextualism” of Adorno’s work: “sometime in the mid 1960s, [Habermas] seems to have been frightened by the specter of relativism, and retreated into a kind of transcendentalism” (Geuss 1981, 64). Desires of the kind I have described above—my wish not to be destroyed, my wish to enjoy freedom—are simply facts about me, right now. In that case, my criticism is nothing more than a restatement of my own personal desires. Against my critique, you are always free to respond: “Well, I want to be destroyed,” and then cast your vote for the capitalist parties. ⁵

Michel Foucault is often accused of this kind of relativism. Honneth argues that Foucault’s “normative criteria remain on the whole so obscure … that the normative direction of his critique of power can often only be deduced from his political-journalistic statements” (Honneth 2007, 40). Foucault is said to have turned his back on the “modern, liberal normative framework” without offering any alternative that could explain why we should support the “insurrection of subjugated knowledges,” to use Foucault’s terms (Fraser 1981, 273; Foucault 1994, 20). Foucault clearly saw himself as a partisan, but he also sometimes suggested that “the pertinent opposition is not between the legitimate and illegitimate,” which would require normative explanation, “but between struggle and submission,” which can be seen as merely descriptive (Foucault 1994, 30). The difficulty is that if his theory is “totalized,” that is, if all aspects of modern life are created by power, then he does not offer us any reason to take the side of the subjugated (Bernstein 1994, 220). The subjugated are products of power—but then, so is Rupert Murdoch. Foucault’s theory, then, is not genuinely critical, because it does not offer us any way of explaining why we should want to struggle in the interests of the subjugated. ⁶

So, critical social theorists find themselves caught between two bad options. On the one side, there are theories of metaphysical foundations, which would have us believe that we can justify our actions based on God or nature. Logical considerations show that such foundations are impossible (because of the Munchhausen trilemma) and political considerations show that they are undesirable (because they are authoritarian). On the other side, functionalist arguments and relativist arguments give us no reason to act or think as those arguments tell us we should act or think. You say the proletariat is the historically necessary grave-digger of the bourgeoisie—but what reason can you give me for why I should pick up the shovel? You say there is an insurrection of the subjugated against modern forms of power/knowledge—but can you give me any reason to care for anyone but myself?

Confronted by this problem, many critical social theorists have tried to “maintain an idea of context-transcending, ethical validity without violating their own authoritarian impulses.” They seek foundations that are normative, but not metaphysical. In a catch phrase, they must pursue a “dialectics of immanence and transcendence” (Cooke 2006, 4, 37).

Varieties of normative foundations

Critical social theorists have pursued this dialectic in a bewildering variety of ways and have produced a wealth of categorizations. Rahel Jaeggi suggests that there are three possible approaches: “freestanding morality or normativity for which you don’t need a notion of progress or social change,” such as Kantianism; “Foucauldianism, which… tends not to be able to rebut the relativism with which it is constantly confronted”; or “some kind of immanent critique” (Allen, Jaeggi, and von Redecker 2016, 228–229). Maeve Cooke, on the other hand, sees four ways that critics can offer normative foundations. Conventionalism takes norms from the Sittlichkeit of its own society; radical contextualism seeks “normative ideas implicit but not fully realized” in its society; context-transcending theories appeal “to normative ideas that are at once immanent to the sociocultural context in question and transcend it”; and authoritarian positions appeal to “a transcendent, final authority” (2006, 14–15).

Both Cooke and Jaeggi then immediately contract their lists. For Jaeggi, Foucauldianism reduces to a freestanding morality, so the only genuine options are freestanding morality or immanent critique (Jaeggi’s approach). For Cooke, the authoritarian position is obviously verboten, and the conventionalist position is so relativistic as to be unusable. The competition is between radically contextual (Foucauldian) and context-transcending theories (Cooke’s approach). Only the latter can guarantee criteria for judging ethical progress. That is the kind of normative foundations we must seek (Cooke 2006, 22). ⁷

The most popular way to avoid getting impaled on the horns of relativism or of metaphysics has been to find norms that have developed historically, but are nonetheless said to be universally binding. Thus, Honneth once claimed that critical theory should uncover “in the social reality of a given society those normative ideals that offer a reference point for a justified critique because they represent the embodiment of social reason.” This move is possible if the theorist works with “a concept of the directed development of human rationality,” (Honneth 2009, 50–51). To put it even more strongly, critical social theory requires a concept of progress, which can best be understood as a learning process (Owen 2002, 22, 186).

Habermas is the origin of this approach. He argues that “the modern understanding of the world is indeed based on general structures of rationality,” and is therefore universal, even though we arrived at it through “learning processes” (Habermas 1984, 66–67). He rejects the claim that all reason is subject to ideology critique, because, he argues, communicative action is necessarily normative, unlike instrumental action (Habermas 1979, 119). Given those conditions, his own “theory of communicative action can ascertain for itself the rational content of anthropologically deep-seated structures… found in the intuitive knowledge of competent members of modern societies” (Habermas 1987b, 383). That is, any proposal can be normatively validated if it would have been the conclusion reached by people in the ideal speech situation, free of coercion: “only the rules and communicative presuppositions that make it possible to distinguish an accord or agreement among free and equals from a contingent or forced consensus have legitimating force today” (cf Benhabib 1986, 346; Habermas 1979, 188). We can understand this, and so we are able to justify our criticisms on the basis of modern rationality and the ideal speech situation.

Honneth identified a problem with Habermas’ position: although he theorized the problems of modernity, Habermas did not explain how actual people could experience those problems as injustice. ⁸ Honneth offered a theory to fill this gap: “moral injustice is at hand,” he argued, whenever “human subjects are denied the recognition that they feel they deserve.” That is, Honneth could provide an even more fundamental normative foundation than Habermas. Communication relies not merely on the linguistic structures of Habermas’ theory, but also on “social recognition” (2007, 71). When we feel ourselves to be the victims of injustice, we are feeling a lack of recognition from others. Our very acts of cognition, Honneth would later argue, rely on a pre-existing recognitive structure, which can be identified through the discipline of child psychology (56, 46). This kind of psychological argument suggests that recognition is an anthropological constant, but Honneth does not see recognition as unchanging. His theory of normative foundations includes an account of historical change: different kinds of recognition have developed over time, in a progressive manner. We can know that our modern orders of recognition—love, legal equality, and social esteem—are superior to those that preceded us, because they make possible “individual self-realization” (Fraser and Honneth 2003, 177–183). ⁹ Habermas has been seeking ever-stronger normative foundations in social and biological science, and ever-stronger proof that modernity offers universally binding normative foundations; until recently, the same was true of Honneth. ¹⁰

Others, however, have started to offer less definitive theories of normative foundations. Nancy Fraser found a normative foundation in parity of participation. This requires that material resources be distributed in such a way that individuals can participate equally in social interactions and that all are accorded equal respect and social esteem (Fraser and Honneth 2003, 36). More recently, Fraser has argued that social criticism must be based on a “triple-movement” in which marketization, social protection, and emancipation are all politically ambivalent. There are conditions in which any of them can be progressive; there are also conditions in which they can be oppressive. Marketization’s worst effects can be exacerbated by emancipation; social protection without emancipation is a recipe for domination (Fraser 2013a). Insofar as parity of participation can be understood as a form of emancipation, the foundation on offer here is less certain than Honneth’s or Habermas’. ¹¹

Rahel Jaeggi has taken Habermas’ and Honneth’s emphasis on learning processes, but reduced their content to a bare minimum. She argues that we can distinguish and judge between forms of life based on the purely formal criterion of whether that form of life enables learning processes or “a process of accumulating experiences.” This relies on the idea that “history is a crisis-driven dynamic of problem-solving,” so we can judge a form of life based on its ability to solve the problems that face it without shutting off possible future experiences (Allen, Jaeggi, and von Rodecker 2016, 240-4). Although Jaeggi continues to use the normatively laden language of modern rationality—openness, growth, pluralism, and differentiation—she offers extremely thin normative foundations, which she has suggested can be reduced to “freestanding determinate negation” (Allen, Jaeggi, and von Rodecker 2016, 241, 246; Jaeggi 2018, 10.3–10.4).

By contrast, Amy Allen has very thick normative principles, but is far more skeptical that those principles are foundations that can be rationally justified. Allen argues that one “can be a contextualist about normativity and still believe in certain normative principles… that are non-relativistic and that may even be universalistic in scope.” That is, a theorist can be restrained when it comes to judging between normative principles or between forms of life, while still strongly defending a specific set of normative principles—Allen names freedom from domination, responding to suffering, and an understanding of freedom as autonomy (Allen, Jaeggi, and von Rodecker 2016, 243–246). But Allen rejects any story of progress through learning processes (Allen 2016, 229). So, compared to Habermas’ or Honneth’s stronger projects, Fraser’s theory of normative foundations leaves more room for ambivalence; Jaeggi’s aims for pure formalism; and Allen’s gives up on the project of rationally distinguishing between historical stages or societies.

Criticisms: Incoherent, -centric, and totally depraved

If theories of normative foundations have grown weaker over time, this is in part because the later thinkers are very much aware of the many criticisms of such theories. There are at least three families of criticism: some point to internal failures on the part of the theories; some suggest that their universalism is unwarranted; some question the basic assumption of theories of normative foundation. ¹²

Hilary Putnam offered an early version of the internal failure criticism. If Habermas is arguing with someone who does not claim that their ethical beliefs are rational, then his theory offers no universal or certain foundations at all. A quasi-Nietzschean, for instance, would reject Habermas’ theory because “democratically structured societies themselves produce ‘false consciousness’,” and “it is only in a certain kind of inegalitarian society that the truth about human existence can shine through” (Putnam 1987, 56). In other words, Putnam can imagine a hypothetical participant, even in the ideal speech situation, who would reject Habermas’ conclusions. ¹³

Raymond Geuss offers a similar argument against theories of normative foundations. ¹⁴ He suggests that projects that try to “‘abstract’ the normative in [their] society” lead to a lop-sided theoretical structure combining “high morality,” which lays down rules that are literally impossible to follow entirely (“One must never lie”), and “really existing morality,” which is how we bend and avoid those impossible rules (“One must never lie… except to protect a lady’s honor”). Theorists then tend to claim that “our [high] morality would be appropriate for any society.” Geuss, to the contrary, is convinced by Marx’s argument that high morality is a contextually bound “compensatory fantasy.” Theories of normative foundations fail, then, because they assume that there is a realm of the purely normative, untouched by values and hypothetical imperatives. But there is not (Geuss 2016, 110–112). ¹⁵

Christoph Henning has argued from a Marxist perspective that theories like Habermas’ or Honneth’s mistake the object of their own critique. Just as Geuss questions whether rational normativity can be analytically disentangled from historical experiences, Henning criticizes German sociology’s tendency to divide facts—or economic structures that simply function without normative input—from non-economic, social, normative frames (Henning 2015, 2.4.1). This project is deeply rooted in the sociological tradition and is taken over in distinctions such as Habermas’s between the system and the lifeworld. Henning blames this false division for a fruitless oscillation between functionalist theories and rational, normative foundation theories, neither of which can be adequate to the proper object of social critique, which is bourgeois society.

Finally, Frieder Vogelmann (2017) argues that critical theories of normative foundations are not properly speaking critical at all. They perform critical work by measuring an object using a yardstick (i.e., a normative foundation). However, the construction of the yardstick cannot be critical in the same sense: Habermas, for instance, can do critique by saying that a social form is undermining the possibility of communicative action. But when he explains what communicative action is and what it requires, he is not doing critique—he is doing traditional theory. This kind of critical theory is, at best, half critical, and half traditional. These theories “must either acknowledge their theoretical activity as uncritical,” or accept that other forms of critical theory are possible and valuable. This, Vogelmann archly suggests, “seems to be an unlikely concession” (106).

Other criticisms of theories of normative foundations focus on their supposed universalism. For instance, Nancy Fraser claimed that Habermas’ theory is “in important respects androcentric and ideological” (Fraser 2013b, 32). The categories of gender play no role in Habermas’ theory, and yet, Fraser argues, “feminine and masculine gender identity run like pink and blue threads” through all of the institutions of modern life, and “male dominance is intrinsic rather than accidental to classical capitalism.” Therefore, any critical social theory needs “gender-sensitive categories.” Habermas lacked them, so he “naturalized androcentric features of the social order” (Fraser 2013b, 38, 45, 6). How universal could a normative foundation be, if it ignored the experience of women?

Amy Allen argues that Habermas and Honneth base their normative foundations on a backward-looking history of progress, which suggests that humanity’s learning process has led up to us, that is, modern humans, and our form of life (Allen 2016, 12-3). The political problem here is quite clear: progress, and so the supposedly universal, rational foundations, are bound up with “colonialism, neocolonialism, and the civilizing mission” (Allen, Jaeggi and von Rodecker 2016, 232). But Habermas and Honneth are blind to their Eurocentrism, which leads to an epistemological problem: how do we know that European modernity is a universal rationality? Can we trust that in setting up a universal theory, we are not claiming transhistorical validity for our own contextual standards? (Allen, Jaeggi, and von Rodecker 2016, 231). We cannot, so we must accept that “the normative principles and ideals on which we rely… may rest on a contingent foundation” (34).

Charles Mills has taken up Allen’s argument in the context of race. Mills accuses contemporary Frankfurt School critical theories of liberal eliminativism: Habermas and Honneth at least implicitly claim that “races do not exist, only individuals.” So, they cannot explain (or even witness) “the difference race makes to the interaction of individuals with society.” Theories that do not take nonWhite experiences or knowledges into account cannot live up to their own universalist claims. Mills quite mildly suggests that the Western philosophical tradition’s “deprecation of nonwhite reason will make it difficult to come to an agreement on norms of discursive justification that is satisfactory” to all (Mills 2017, 236, 242).

A third, more radical criticism of theories of normative foundations has emerged more recently. Nancy Fraser has recently argued that second wave feminism contributed to the “new spirit of capitalism.” Boltanski and Chiapello showed how capitalism reappropriates the anti-capitalist critique of the New Left, so that “the higher, moral significance” of the critique was attributed to capitalism, rather than its critics (Boltanski and Chiapello 2005; Fraser 2013b, 219). Fraser suggested that the same thing had happened to second wave feminist arguments against the “family wage.” By giving a moral sanction to the increased exploitation implicit in shattered glass ceilings, leanings-in, and flexibility, second wave feminist arguments helped to “intensify capitalism’s valorization of waged labor” (Fraser 2013b, 221). This kind of argument strongly suggests that any emancipatory, normative claim can turn oppressive given even slight changes in historical context, and so hints at the possibility that there cannot be any solid normative foundation during the era of neoliberal capitalism. Fraser, however, rejects the idea her claims about second wave feminisms can be generalized. All normativity, she concluded, “has its own emancipatory surplus,” as Honneth had long argued (Fraser and Jaeggi 2018, 151).

But others have made the more radical claim. Rocío Zambrana argues that Honneth’s theory of recognition “involves a notion of normativity whereby norms have critical potential if they remain insulated from capitalist resignification.” In other words, the pure norm of freedom can be distorted by its social context, but freedom itself retains its normative surplus and can be rationally foundational. Zambrana argues to the contrary that “normative integrity cannot be maintained” during neoliberalism (Zambrana 2013, 105). Today, recognition, freedom, or Fraser’s participatory parity are all fallible, and none can work as foundations. Zambrana concludes that “the authority of normative language requires political rather than philosophical justification” (Zambrana 2013, 115). ¹⁶

Finally, Amy Allen has criticized Habermas and Honneth for presuming “that the normative can be purified of power relations” (Allen 2016, 206). As Adorno and Foucault have shown, “all our thought forms, including our conceptions of reason, are… socially and historically conditioned” (Allen 2016, 220). There cannot be a normative stance free from the oppressions that fill our societies and histories. This radical criticism suggests that theories of normative foundations are fundamentally flawed. To use theological language, individuals under neoliberal capitalism are totally depraved. Although liberal norms may once have been available as a critical tool, they are vitiated by neoliberal societies (Chari 2015, 44–45). To ignore this is a mis-reading of history that diverts “attention from real, existing catastrophic tendencies” (Abromeit 2018, 22; cf Freyenhagen 2017, 463).

The myth of the Given

In their well-known debate, Nancy Fraser accused Axel Honneth of using prepolitical suffering as a reference point for critical theory that was “simultaneously empirical and primordial.” This, she argued, was “an instance of ‘the myth of the given’” (Fraser and Honneth 2003, 204). Honneth dismissed the myth of the given as a “blunt, all-purpose weapon” (245-6). Fraser did not press the point at the time. I will do so here. The analogy between the problem of rational, normative foundations and the problem of the given suggests that critical social theorists are trying to do something impossible. This, in turn, could explain the gradual weakening of the normative foundations on offer, from Habermas’ and Honneth’s borderline transcendental claims, to the less certain foundations offered by Fraser, Jaeggi, and Allen. These later theorists are aware of the difficulties involved and have hinted at critical social theories that do not focus on normative foundations.

The idea of the given is of great importance for the philosophical attempt to explain how we come to have knowledge. In an argument dating back to Aristotle, (Post. An. I. 2–3), philosophers claim that we have most of our knowledge through demonstration: for instance, I can prove to you that, if x + y = 0, then x must be -y, unless x and y both = 0. But in order to demonstrate that, I have to call upon other mathematical facts that do not need to be demonstrated or cannot be demonstrated—axioms and so on. If those undemonstrated facts did not exist, we would find ourselves in an infinite regress of demonstration, and we would never be able to prove anything. Something similar is true for empirical trains of thought. If I believe that my water bottle contains hemlock, and I believe that I will soon be forced to read about Aristotle, I may then hold a further belief that I would like to take a hefty sip from my water bottle. But are those original beliefs based on anything certain? If not, there is a regress, as I try to justify those original beliefs. ¹⁷ If they are based on something certain, the regress comes to an end at some foundational, empirical fact. Whether the knowledge in question is empirical, mathematical, or logical, we can only avoid a regress if there is some foundation for our knowledge that does not require further demonstration or proof. Or, so foundationalist theories suggest. ¹⁸

This foundation is sometimes called what is given in experience. Evan Fales defends a theory of the given that makes three claims: there are “experiences to whose content or character an individual has direct—that is, noninferential—cognitive access”; some of the content and some of the truth value of some of these experiences “can be determined immediately and with certainty”; and these experiences “play a foundational role with respect to the acquisition of knowledge and justified beliefs” (Fales 1996, 6). In short, we can be said to know something when that thing is a given experience, or when we validly infer on the basis of given experiences.

Criticisms of the given as a foundation for knowledge are almost as aged as Aristotle’s argument; the Münchausen trilemma described above draws on Hellenistic Greek skeptical arguments. More recently, John Dewey offered an argument against the given, in defense of his pragmatic view of knowledge acquisition. He argued that it was fallacious to take something indisputably important for scientific inquiry, that is, the need to start somewhere, and to convert that temporary need into a substantial “independent structure,” which is said to be really existing. The given, he argues, is only a postulate that makes scientific inquiry possible (Dewey 1938, 149). So, for instance, the assumption of a deterministic universe makes it much easier to do experimental science, but it would be fallacious to then claim that the universe was, in reality, entirely deterministic. ¹⁹

Wilfrid Sellars’ argument against the given is still the most well-known, despite its infamous inscrutability. He was worried that philosophers were being pushed to accept either foundationalism or coherentism, that is, the idea that our beliefs can mutually justify each other without any foundation. He described the first as an elephant resting on a tortoise and the second as “a great Hegelian serpent of knowledge with its tail in its mouth” (Sellars 1963, 170 §38). He hoped to avoid this dilemma entirely, but his description of the myth of the given focused on the problem of foundationalism. ²⁰ Sellars structured his argument around two claims. For the foundationalist, there is something that can “be non-inferentially known to be the case, but presupposes no other knowledge” and is “the ultimate court of appeals for all factual claims.” Knowledge about this thing “is noninferential, yet it is, after all, knowledge. It is ultimate, yet is has authority” (Sellars 1963, 164 §32, italics Sellars’ own). This latter idea is what causes so many problems: it is not clear how something can be noninferential, but also knowledge. How can something justify claims to knowledge, but not be justifiable itself? ²¹

Philosophers have tried to clarify Sellars’ claims ever since he made them. Bonjour described the dilemma in terms of whether the given foundation of empirical knowledge is cognitive in nature or non-cognitive. If cognitive, he argues, the given will be able to justify further knowledge, but will also stand in need of justification, and so not be given (since cognition requires justification); if non-cognitive, the given will not need to be justified, and so will be a genuine given, but it will not be able to justify any further knowledge—because only a cognition can justify another cognition. In this case, the given will not perform the function it is meant to perform, that is, putting a stop to the regress of justification (Bonjour 1978, 11). Maher similarly suggests that a given can either be conceptual, or non-conceptual. If it is conceptual, then it cannot be independent of other concepts, because having one concept is just to have all of them (cf Sellars 1963, 148 §19). But if the given is non-conceptual and is therefore independent of other concepts, it would not be “epistemically efficacious” (Maher 2012, 17), because a concept can only be justified by a concept.

So, Sellars seems to have left us with a depressing situation: knowledge requires some given foundation if we are not to find ourselves in a regress, but there is nothing that can function as such a foundation. His solution to this was to reframe the idea of justification, so that it was neither an elephant resting on a tortoise (but what does the tortoise rest on?) nor an ouroborus (but where does it begin?). Instead, he appealed to the idea of the “space of reasons.” When we talk about knowing something, we should really think of ourselves as putting that thing “in the logical space of reasons, of justifying and being able to justify what one says” (Sellars 1963, 169 §36). That allows us to see that rationality is “a self-correcting enterprise which can put any claim in jeopardy, though not all at once” (Sellars 1963, 170 §38). ²² Only that which is within the space of reasons can justify our claims to knowledge; anything outside of the space of reasons cannot play any justifying role. We no longer need to keep moving back and back and back to find some ultimate foundation. Instead, the space of reasons as a whole is the justification of last resort—and to claim that the space of reasons is not justifying is to leap into brute irrationalism.

To see how this is meant to work, we can introduce another distinction, between the natural and the normative. One way out of Sellars’ argument against the given is to say that something in the world can affect us in a non-cognitive manner and that that effect can justify our cognitions. If a right-wing zealot has sabotaged my chair by jamming large knives through the back of it, they will tear my flesh whether I know they are there or not. My torn flesh is non-cognitive, but my belief I should get out of this chair is both cognitive and well-justified. But Sellars, and thinkers after him, reject the claim that “thinking about things can be assimilated to being merely affected by things” (Maher 2012, 1). Only if the pain I feel is put within the space of reasons am I acting when I leap out of the chair; otherwise, I am only re-acting, much as a house plant turns towards light without deciding, rationally, to do so. Therefore, thinking rationally is a normative activity, not a natural, law-bound one. Sellars’ argument against the given goes along with the claim that natural relations “are different in kind from the normative relations” of human rationality (McDowell 1994, xv). Natural relations are not inside the space of reasons; rational actions are. Natural events do not give us a reason to act, but they can become a reason to act, if they are taken to be such—and therefore placed in the space of reasons. The same person will react differently to the same temperature in a sauna during the day and in their bedroom at night.

The vast industry of interpretation about this argument is beyond the scope of this paper. I do not want to defend the argument’s validity with regard to epistemology. ²³ I only want to set it up here as one half of an analogy, in order to criticize normative foundations in critical social theories. For that, this general outline will suffice: an epistemological given cannot fulfill the function it is designed to fulfill, that is, justifying our inferential knowledge without needing to be justified in turn.

The argument against normative foundations

So far, I have outlined the origins and development of theories of normative foundations and given a brief overview of the argument against the given in epistemological terms. The two seem to me to fit together in an analogy. Foundationalists in epistemology seek some grounds for human knowledge that can justify our claims to knowledge and avoid a regress of justification. Those grounds must need no further justification. The argument against the given claims that this is not possible, because the feature of the given that allows it to justify our knowledge (being within the space of reasons) also requires that the given be justified in turn. That is, only the normative can justify the normative; only the rational can justify the rational (cf Bonjour 1978, 12).

In exactly the same way, theories of normative foundations seek to avoid a regress of rational justification. Such a regress threatens to leave us with relativism, decisionism, or dogmatism: why should we fight climate change? Because it has reduced biodiversity. But why should I care about biodiversity? Because biodiversity is important for human flourishing. But why should I care about human flourishing? According to theories of normative foundations, this is the point at which we need a backstop. Here, we must posit some normative foundation that can avoid the regress and the relativism, and allow us to rationally justify our critique of capitalism or modern society. In order to avoid the regress, however, that foundation itself must not need any justification.

My suspicion is that just as the epistemological given can only justify our knowledge if it, in turn, requires justification, so a normative foundation can only justify our critique if it, too, requires justification. The easiest way to see why is to discuss it in the terms Maeve Cooke used in Cooke (2000). There, she suggested that we try to find normative foundations that are both contextual and objective. She argued that although objectivism “stresses the need for evaluative standards that are non-arbitrary and universal,” it could also incorporate contextualism. This is what she sought in her own theory: “whereas ‘objectivism’ without ‘contextualism’ is blind, ‘contextualism’ without ‘objectivism’ is impoverished” (Cooke 2000, 199).

Cooke later changed her terminology and claimed that a sufficient theory must be both contextual and context-transcending, rather than objective (Cooke 2006). This makes sense: something can certainly be contextual and context-transcending. For instance, I am intrigued and impressed by the art of the Aksum kingdom, despite being in a very, very different context from that of its artists. But, on the other hand, nothing can be contextual—that is, tied to its immediate historical and social situation—and also objective, that is, of identical standing for all historical and social situations. Disappointingly few people are intrigued by Aksumite art: its value is context-transcending, but not objective. Cooke was correct, then, to alter her earlier formulation. But the problem persists despite the change in terminology. If something is universal enough to be a rational normative foundation, it cannot be contextual. But something that is context-transcending need not be universal enough to be a certain, normative foundation.

Another way to describe “contextual” here might be “socially mediated”: objects and ideas are not the products of an individual’s mind or hard work, but rather gain their standing and content from the character of the society in which they are created. A sword in the middle ages was an object of brute power that worked to keep peasants almost ontologically distinct from lords; the very same material object today is a museum piece. Even though museums might still reproduce class hierarchies, the meaning of the object and its function are now very different. The question of normative foundations could then be posed in this way: can we find some foundation that is both socially mediated, but plays the same role in all societies? And this seems no more plausible than something that is both contextual and universally objective. If the proposed foundation is socially mediated, it will not be universal enough to function as a foundation; if it is universal enough to function as a foundation, it will be too abstract to justify anything in any particular society.

Honneth has occasionally used the language of context-transcendence (2009, 44) and has also offered a slightly different way to get at the point. He once suggested that a criterion for measuring the “pathologies of a society” would “consist in culturally independent conditions that allow a society’s members to experience undistorted self-realization” (2007, 35). Honneth argued that communitarian arguments, like Charles Taylor’s, cannot do this, because they are relativistic (cf Honneth 2009, 50). Therefore, only a “weak, formal anthropology” could be an appropriate criterion (42). ²⁴ Here, again, we see the strangeness of the claim in the language itself: if human actions are socially (or culturally) mediated, then “culturally independent conditions” can have no motivational or normative force on actually existing, socially and culturally embedded human beings. We want what we want because of our culture and societies. So, a “condition” can either be culturally independent (and thus lack foundational, normative grip), or it can be culturally dependent, and then not universal. The claim that a condition is culturally dependent is made by communitarians like Charles Taylor, and Honneth rejects it. But the communitarian position has the benefit of being conceptually possible, no matter how troubling some counter-examples might be (e.g., what do communitarians say about culturally dependent norms in North Korea?).

Finally, Sellars’ “space of reasons” argument gives us one other way of considering the problem. Using this language, we can describe the search for normative foundations as one which seeks for some foundation that is both within the space of reasons (and hence able to rationally justify our critique), but also outside the space of reasons (and so not in need of rational justification itself). But this is to claim that the normative foundation will be both rational and non-rational; or normative and non-normative. We would be asked to use the non-rational element of the normative foundation to justify our rationality and the non-normative element of the normative foundation to justify our normative claims. But this is not possible. Only a reason can justify a reason; only a normative claim can justify another normative claim. It seems that no matter what language we use to describe the “dialectic” of immanence and transcendence in theories of normative foundations, we will not be able to find any candidate that is sufficiently contextual/immanent/socially mediated/normative, and context-transcending/objective/transcendent/immediate/natural.

Given this, why would anyone defend a theory of normative foundations? To use another mid-century modern philosophy analogy: Elizabeth Anscombe (1958) argued that moral philosophers of that period were caught in a trap. They wanted to uphold the traditional role of moral theory, without accepting the traditional source of moral theory (that is, God). Such philosophers used “the emphatic ought” as if it still had the force of a divine law, but rejected the idea of a “divine legislator” (11). This led them to try to replace God as the final justification for morality. Some found their justification in the norms of society; some in the laws of nature; some in a social contract; and some in human virtues. The first, Anscombe argued, leads to relativism; the second to something like social Darwinism; the third is conceptually incoherent; and the fourth was coherent but certainly did not offer the same certainty as the divine legislator. Anscombe’s paper was an early intimation of the later 20th century’s virtue ethics. It might also be a prescient critique of theories that want to have normative foundations without God or nature. On this second analogy, theories of rational normative foundations might be a kind of nostalgia for lost certainty. ²⁵

My argument here fits with previous criticisms of theories of normative foundations. As with Geuss and Putnam, it makes the point that existing theories of normative foundations cannot actually offer a certain ground for critique. As with Fraser, Mills, and Allen, my argument suggests that contextual theories of normative foundations do not and cannot have universal force, and those that have universal force cannot do justice to human histories and societies. And, like criticism based on total depravity, my argument suggests that theories of normative foundations will necessarily fail, today. I do not want to deny any of these other arguments against theories of normative foundations. They are convincing on political, moral, and empirical grounds.

Nonetheless, the argument I have laid out goes slightly further. The arguments that actually existing theories of normative foundations have failed suggest that the problem might be the type of foundation being sought. For instance, the arguments against normative foundations under neoliberalism sometimes suggest that Habermas was right to seek normative foundations under welfare-state capitalism or would have been right under classical liberalism.

But, as Bonjour (1978, 12) said of the argument against the epistemological given, there is a logical problem here, not merely an empirical one. A critical social theorist should not try to point to something that intuitively fits the criteria of a normative foundation and take their work to be completed. The question raised by the argument against the given, and, mutatis mutandis, my argument against normative foundations, is not what can function as a foundation—this was Cooke’s question—but whether such foundations are even possible. And the answer is that they are not possible in any current society.

Similarly, I am not suggesting that there was some “variety” of capitalism that was well-served by theories of normative foundations, or that, if we somehow manage to reconstruct welfare-state capitalism, we should return to such theories. ²⁶ My argument, rather, is that we cannot construct universal, certain normative foundations in any society that could be the object of a critical social theory—that is, in any society that produces abstract domination through social mediation (cf Postone 2003). To syllogize, norms are socially mediated or contextual; in societies of abstract domination, social mediation is tied up with domination; therefore, in societies of abstract domination, all norms are maculate.

Theorists sought normative foundations in part because the Dialectic of Enlightenment and the works of Foucault had supposedly left us in a theoretical dead-end, according to which there was no ontological possibility of a better world. That seems to me to be a very wrong reading of Dialectic of Enlightenment, Foucault, and the society we live in. There could be some form of society that is not based on abstract domination, and such a society could have adequate normative foundations. But such a society would not be an object for critical social theory: the normative foundations would be transparent, and properly functioning, rather than offering a standpoint for critique. ²⁷ Instead of reading Habermas, theorists in such a society could play around with algorithms in the morning and go mushrooming in the afternoon. But that charming daydream throws up one last question: what might critical social theorists do in our society, if not seek normative foundations for their critique?

Summary

I began by describing the historical development of theories of normative foundations. Traditional Marxist social theories relied on the idea that the standpoint of labor or the proletariat justified their social criticism, but the Dialectic of Enlightenment argued that we could not take that standpoint. In response, Habermas suggested that a critical social theory needed new foundations in order to rationally justify its critique. Without them, we would have no way to rationally convince others that we are right, no way to judge progress, and no way to coherently criticize our society at all. Critical social theories needed rational normative foundations to avoid relativism, without falling into the problems surrounding metaphysical foundations. This leads to the dialectic of immanence and transcendence, or of contextualism and context-transcendence: a theory’s foundations must be contextual in order to avoid authoritarianism, but context-transcending in order to avoid relativism. The most common way of threading this needle is to claim that some historically contextual norm is also universally rational, and so context-transcendent. This approach has been taken by both Habermas and Honneth. Recently, other theorists have offered weaker and weaker possible foundations.

Theories of normative foundations have been criticized for failing to deliver on the promise of coherent criteria for critique; as being far less universal than they claim; and for ignoring the conditions of neoliberalism, in which all norms come to be corrupted. Without disputing those criticisms, I have gone slightly further. My argument is that critical social theories, which take abstractly dominating societies as their object, cannot offer rational, universal, binding normative foundations. I argued that the dialectic of immanence and transcendence must fail. Just as the argument against the given shows that nothing can justify rational inferences without itself needing to be rationally justified, so, my argument suggests, nothing can be both contextual/historical/socially mediated (and therefore offer a non-authoritarian foundation), and context-transcending/universal/immediate (and therefore function as a foundation for critique). The prophet Hosea could offer metaphysical normative foundations, because he was not concerned about authoritarianism. A theorist who lives in a non-dominating society might offer non-oppressive normative foundations, if their society has them. But critical social theorists work under the conditions of abstract domination and the rejection of authoritarianism. We cannot offer rational normative foundations. What can be done, instead?

What could we do instead?

The critics I have discussed here offer many suggestions for moving past theories of normative foundations. Zambrana (2018) suggests Fraser’s triple movement as a starting place: it requires us to consider the interactions of various norms at specific historical moments, rather than simply choosing one. Although Zambrana offers this as a contrast with Jaeggi’s approach, it also fits well with Jaeggi’s own suggestion that the best we can do is “freestanding determinate negation” (Allen, Jaeggi, and von Rodecker 2016, 246). We cannot simply ask whether an idea or action is justified by a normative foundation; nor whether a specific norm is the ultimate foundation. Rather, any norm, or any event, requires a more fine-grained engagement with the many norms we actually use. Critics of normative foundations often do just this. When Allen argues that Honneth’s normative foundations are vitiated by their use of the concept of progress, she is just doing what Honneth should have done to begin with: ask himself how the norm in question fails to live up to itself.

So, rational, universally binding normative force is not available to us; our normative authority must come from something less secure and less certain. There is not yet a perfect language for describing what this could mean. Zambrana distinguishes foundationalist, rational, philosophical justification from political justification (2013, 115); Freyenhagen, similarly, suggests that a critical theory should focus on the interests of the suffering, rather than rationally binding norms (2017). And Chari has tried to bring the (admittedly problematic) resources of radical democratic theory to bear on a reconstruction of critical theory (2015). I take each of these points to make a similar claim: the justification for critique of the way things are is simply that we do not want them to be this way. Of course, this throws up a host of issues. Who gets included in the “we?” Any critical theory has to avoid the problems of the -centrisms discussed above. How does the group come together to decide what they (we?) want? Any discussion must beware the problems of abstract domination inherent in all of our interactions. How can the group be even moderately sure that it wants the things that it says it wants? Desires and needs are just as socially mediated as any normative foundation. And, of course, is this not just decisionism? (cf Freyenhagen 2019).

These are genuine problems that I do not want to dismiss. Nor would I want to try the reader’s patience by discussing them here. Imagine, however, that critical theorists did focus on explaining what anyone could actually, really, want, rather than focusing on what we can rationally justify wanting. ²⁸ The task of explaining and thinking through what we want might take us back to Adorno and his near-definition of negative dialectics: “reciprocal critique of the general and the particular, the identifying acts which judge whether the concept does justice to what it is dealing with, and whether the particular also fulfills its own concept, are the medium of the thinking of the non-identity of the particular and concept” (Adorno 2003, 149; translation, Adorno 2001). Here, neither the concept nor the particular are foundational. Instead, each is measured against the other, constantly. The concept of freedom is no more incorrigible than actually existing freedom. The fact that we apparently want freedom, the concept of freedom, freedom as it actually exists—none of these can offer a firm, rationally justifying foundation for our interests, desires, or demands, but all of them will need to be taken into account if we are to clarify what we want, let alone if we are to get it: do we want the freedom we have, or the ideal concept of freedom? What would it mean to want them? What would we get from having them? What do we get from having what we do have? As many of the scholars I have discussed here have noted, Adorno might be useful again, more useful, perhaps, than recent theorists of normative foundations. But these are speculations for now. In this paper, I have no certain conclusions to offer.