Defending Instrumental Rationality against Critical Theorists

Central to much critical theory is the critique of instrumental rationality (roughly, the ability to pick good means to ends). This critique is overstated, I suggest. Critical theorists often depict instrumental rationality too narrowly, and many criticize the wrong target, for example, attacking capitalist instrumental rationality when the fundamental problem is capitalism, not instrumental rationality. Nonetheless, critical theorists’ critique requires certain changes to orthodox accounts of instrumental rationality. I offer a more palatable definition, highlight instrumental rationality’s essential contestability, and show that it can actually help us pick ends. Everyone needs instrumental rationality, especially Habermasian critical theorists. And far from instrumental rationality being amoral, I argue that because instrumental rationality almost always involves multiple ends, one end may prohibit immoral means, acting as a side-constraint. Ultimately, the substance of critical theorists’ critiques remains highly important but should not be framed in opposition to instrumental rationality.

Such views have been extremely influential-and not only among political theorists and political philosophers (e.g., Allen 2015;Benhabib 1986;Fraser 1985;Marasco 2015). Critical theory is one of the nine major theoretical perspectives covered in The Oxford Handbook of International Relations (Reus-Smit and Snidal 2008) and underpins many empirical studies in that field (see Risse 2015). Work on the public sphere by Jürgen Habermas, a key critical theorist, underlies theoretical work on civil society (e.g., Cohen and Arato 1992) and empirical work by political scientists (e.g., Wampler and Avritzer 2004). Critical theory is particularly important in deliberative democracy, both in theory (e.g., Dryzek 1990;Hammond 2019;Rostbøll 2008) and underlying much empirical analysis (e.g., Steiner et al. 2004). So, critical theory is not just abstract: it is also widely applied in empirical research in politics. The same holds in other disciplines (Blau 2011, 40;. Critical theory is highly and increasingly diverse (Rush 2004, 7), but the critique of instrumental rationality is "probably its most notable contribution" (Alford 2017, 425). Instrumental rationality-roughly, the ability to choose good means to ends-is typically seen by critical theorists as amoral, narrow, or having damaging effects. Such concerns go back to predecessors of critical theory, including Kant, Hegel, Marx, Nietzsche, Weber, and Lukács (Schecter 2010). And similar criticisms are made by many outside critical theory (e.g., in an economics textbook by Varoufakis 1998, 92-93, 100-101, 114, 264-65, 356-58). Much of my response applies to such thinkers too. Of course, not all critical theorists discuss instrumental rationality, and those who do so sometimes offer different analyses (Smulewicz-Zucker 2017). But the critique is central for many critical theorists and informs many empirical studies as exemplified above.
My key argument is that this critique of instrumental rationality is overstated. Critical theorists can and should attack particular conceptions and applications of instrumental rationality. But instrumental rationality itself, properly understood, is a sensible idea. And it is one that 958492P RQXXX10.1177/1065912920958492Political Research QuarterlyBlau research-article2020 1 King's College London, UK critical theorists actually need, for example, for deliberative democracy.
But defending instrumental rationality requires some tweaks to standard accounts-my second aim. I apply some ideas not previously prominent in the literature, such as John Rawls's surprising depiction of instrumental rationality as an essentially contested concept, and some ideas not previously related to instrumental rationality at all, such as Joseph Raz's tripartite account of ends.
This paper thus strengthens instrumental rationality against criticisms. But the paper also strengthens critical theory. I do not oppose critical theory; indeed, I have offered some contributions myself (Blau, 2019). But the critique of instrumental rationality needs rewriting. Critical theorists can still attack narrow or immoral ends, inadequate questioning of those ends, and so on. But these are not problems with instrumental rationality itself.
This sounds like a fundamental challenge to a central concern of a prominent theoretical perspective. But instrumental rationality should never have been a central concern in critical theory. Indeed, critical theorists themselves need instrumental rationality. Reframing the critique thus strengthens, not weakens, critical theory: focusing on what really matters makes the objections harder to ignore.
My paper builds on work by James Johnson and Joseph Heath. Each advocates closer links between Habermasian critical theory and rational choice theory, especially game theory. Johnson (1991Johnson ( , 1993 shows how Habermasian critical theory and rational choice theory illuminate each other, while Heath argues that rational choice theory strengthens Habermas's accounts of communicative action (Heath 2001, chapters 2-4) and social order (Heath 1996, 51-59;2001, chapters 6-7).
Nonetheless, my focus is different. I largely sidestep rational choice and game theory. Instrumental rationality should be part of any theory of action or morality, not just rational choice. Many scholars have an almost instinctive distrust of rational choice and game theory. Instrumental rationality should be less controversial: we need it to cross the road to buy a present for a friend to show our affection for them, say.
Moreover, Johnson and Heath primarily address Habermas. (Indeed Johnson, writing in the early 1990s, cannot address Habermas's change of direction in 1993, as we will see). Habermas is not exactly positive about instrumental rationality, but he is less negative than many critical theorists. This paper thus covers Habermas and other critical theorists too.
I should add three important caveats. First, I am not defending instrumental rationality as the only idea of rationality, or even the main one; nor do prominent exponents of instrumental rationality. Any complete theory of rationality will probably include instrumental rationality (Nozick 1993, 133, 176;Pauer-Studer 2007, 76, 79), but instrumental rationality is only one part of this broader value-system (Rawls 1971, 412-24;1996, 50-51). Instrumental rationality does not exhaust rationality: at "the core of our rational capacity" is "the ability to engage in self-critical reflection" about our "standard of deliberation" (Gauthier 1986, 183), and our ends ultimately matter more than our means (Rescher 1988, vii, 1-2, 6-8). Instrumental rationality can even be irrational: seeking means to ends which are based on palpably irrational beliefs is hardly the epitome of rationality (Elster 1983, 15-26).
Second, my account of instrumental rationality is broader than that used by most economists and rational choice theorists, who often specify fixed ends, egoism, complete preferences, and so on. Economists have long known that such assumptions are false (e.g., Sen 1977) but usually need them for modeling, explaining, or predicting (Varoufakis 1998, 54). That said, rational choice theorists increasingly amend such assumptions (Fearon and Wendt 2002, 55-56), a point occasionally recognized by critical theorists (White 2004, 313).
Third, when I discuss "critical theorists," I primarily mean first-generation and second-generation Frankfurt-School critics of instrumental rationality (especially Max Horkheimer, Theodor Adorno, and Jürgen Habermas), and, importantly, the huge number of "applied" scholars they have influenced, conducting empirical analysis of deliberative democracy, civil society, and so on. I largely sidestep more recent Frankfurt-School theorists. Writers like Rainer Först hardly address instrumental rationality. Axel Honneth, indeed, sees recognition as more fundamental than rationality (Honneth 2007), and as noted below, criticizes some caricatures of instrumental rationality. But such writers have fewer "applied" followers studying politics. This paper runs as follows. After briefly summarizing and starting to challenge critical theorists' account of instrumental rationality, I criticize five common fallacies in this critique. Nonetheless, the critique requires changes to orthodox depictions of instrumental rationality. I thus define instrumental rationality more broadly than is standard. I then use Rawls's insight about instrumental rationality's essential contestability, to deflect allegations that instrumental rationality necessarily implies efficiency, speed, or cost, say. I also discuss "second-order" instrumental rationality: the ability to choose good means to choosing means or ends, which Habermasian critical theorists need. I then supplement a point made by others-that every end is a means to other ends-to reject the criticaltheory view that instrumental rationality is non-cognitive, that is, says nothing about our ends. I also challenge critical-theory objections about instrumental rationality's amorality, by showing that because instrumental rationality almost always involves multiple ends, one end may act as a side-constraint, precluding immoral means. Overall, the more nuanced critique of instrumental rationality recommended here, although apparently undercutting a core commitment for many critical theorists, actually strengthens critical theory.

Critical-Theory Critiques of Instrumental Rationality: Overview
Frankfurt-School critical theory originated in the Institute for Social Research in Frankfurt in the 1920s. In 1937, Horkheimer (1972 published his famous essay contrasting traditional and critical theory, but as regards instrumental rationality, the key early texts were published in 1947: Horkheimer's (2004) Eclipse of Reason, and Horkheimer and Adorno's (1972) Dialectic of Enlightenment. Their excessive pessimism (see Schecter 2010, 89, 223-24) led Habermas, from the 1970s onward, to offer more constructive accounts involving different ideas of rationality, primarily communicative and discursive rationality (especially Habermas 1979Habermas , 1984Habermas , 1990. Habermas (1987b, 333) even criticizes the way that Horkheimer and Adorno "demonized" instrumental rationality.
Simplifying considerably, there are typically two stages to these criticisms of instrumental rationality, including those criticisms by adherents of Horkheimer, Adorno, and Habermas. Stage 1 makes conceptual and/or empirical claims: critical theorists define instrumental rationality, note its presuppositions, and/or describe its effects. Stage 2 negatively evaluates what stage 1 depicted and, except in Horkheimer and Adorno's pessimistic accounts, adds prescriptions, for example, restricting instrumental rationality or recommending alternatives. I will suggest that stage 1 depictions of instrumental rationality are too narrow: instrumental rationality in general is conflated with particular conceptions or applications of it. This encourages stage 2 criticisms that attack instrumental rationality in general rather than particular conceptions or applications.
Before elaborating, I should address one potential objection: that critical theorists are not actually attacking instrumental rationality in general, only some applications. As Fred Rush notes, we all use instrumental rationality, for example, deciding how to get to the cinema to see a movie that one wants to compare with another movie. So, "there does not seem to be anything particularly suspicious about 'instrumental reason' understood in this manner, although of course one may doubt its application in specific contexts." Critical theorists "do not begrudge rational agents their means and, so, are not 'against' instrumental reason in these broad terms" (Rush 2013, 192).
Unfortunately-and this matters greatly for my paper-there is little textual evidence for this position (one exception being Honneth and Joas, discussed shortly). To the contrary: there is ample evidence that many critical theorists, although doubtless not all, are attacking instrumental rationality in general. For example, Horkheimer, arguably the most influential of early critical theorists (Rush 2004, 7), clearly criticizes instrumental rationality in general: "the optimum adaptation of means to ends." And note how he immediately depicts this as "a pragmatic instrument oriented to expediency, cold and sober" (Horkheimer 1941, 368, emphasis added). Yet, instrumental rationality can also be warm and merry, used for planning parties, finding friends, deciding which charities to support, showing appreciation of kind acts by loved ones, and so on. Portraying instrumental rationality as "cold and sober" lets Horkheimer (1941, 368) link instrumental rationalitywhich he defines broadly, I reiterate-to "the modern technique of war," "the conduct of business," and "the dictators of today." But instrumental rationality can equally be used by pacifists, socialists, and liberals.
Elsewhere, Horkheimer (2004, 5) depicts instrumental rationality as devoted exclusively to the relation of . . . an object or concept to a purpose, not to the object or concept itself. There is no reasonable aim as such, and to discuss the superiority of one aim over another in terms of reason becomes meaningless. This is clearly instrumental rationality in general. Yet, he links this to self-preservation and self-interest (Horkheimer 2004, 5). That claim is self-defeating: if instrumental rationality has "no reasonable aim as such," then it need not involve self-interest and self-preservation. Habermas (1979, 41, 117;1984, 285-86;1996, 25, 27, 90, 161, 337), too, repeatedly claims that instrumental rationality and its social variant strategic rationality are egocentric. Habermas never justifies this assertion, nor can he (Blau 2019;Johnson 1991, 189-91; see also Honneth and Joas's critique of Habermas, discussed in Deranty 2009, 173-77).
In short, many critical theorists do seem to conflate instrumental rationality in general with particular notions or applications of it, without explicitly distinguishing the two scopes or stating that they are only criticizing the latter-let alone noting that the former is perfectly defensible, at least sometimes. Rush's defense of critical theorists does not seem to apply as widely as he suggests.

Stage 1 of the Critique of Instrumental Rationality
I now offer more detail on the two stages of critical theorists' attack on instrumental rationality. In stage 1 (conceptual and/or empirical), instrumental rationality is defined, its presuppositions are noted, and/or its effects are described. Here, instrumental rationality is typically depicted too narrowly, even where critical theorists start with a broader notion, as just discussed. Thus, we are told that-in terms of definitions, presuppositions, or effects-instrumental rationality involves dominating nature (Horkheimer 2004, 64), the predominance of science and technology (Marcuse 1991, 157-63), abstract reason and mind/body dualism (Gardiner 2000, 9), bureaucracy (Dryzek 1990, 4-5), treating other people merely as a means (Alker 1990, 174), and conservatism (Dryzek 2006, 113). These linkages may overlap, as when Michael Ott relates instrumental rationality to atomism, bourgeois egoism, neoliberalism, capitalism, class domination, colonialism, imperialism, and a subject-object dualism where subjects try to control and manipulate an objectified world (Ott 2006, 121-23), or when Patrick Murray (2018, 769) states that instrumental rationality "presupposes a world stripped of social forms and purposes, devoid of moral, aesthetic, gendered, social, legal, and political features-a world without a way of life." This very last claim-instrumental rationality presupposing a world with no way of life-is theoretically impossible. Instrumental rationality is about picking means to ends. Without a way of life, without ends, we cannot pick means to them; we cannot be instrumentally rational. This is important. What instrumental rationality actually presupposes is a value-system, otherwise one has no ends to pick means to. Instrumental rationality will be dangerous in a bad value-system, and desirable in a good one.
The substance of critical theorists' claims thus matters hugely, and many adherents of instrumental rationality overlook these problems. But I think that only two of the above links are potentially inherent to instrumental rationality: the assumption of fixed, pre-given ends (e.g., Habermas 1984, 170-72;Horkheimer and Adorno 1972, 36-42), and amorality, such that instrumental rationality can be applied to murderous ends (Forester 1985, 49;Weizenbaum 1976, 252, 256). I address these challenges toward the end of the paper.
Those two possibilities aside, none of the above claims necessarily follows from seeking good means to ends. Consider again Horkheimer's (1941, 370;2004, 3; see too 69) claim that in instrumental rationality, the ends "serve the subject's interest in relation to self-preservation." This is not a necessary relationship. As Horkheimer knew, kamikaze pilots can pick good means to their (literal) end. Similarly, we can pick means to environmental flourishing, rather than just dominating nature; we can pick means to feminist ends, rather than patriarchal ones; we can pick means to socialist ends, not just neoliberal ones.
Such overstated criticisms thus involve a "type/token fallacy," conflating instrumental rationality in general with particular examples of it, or what I call "rottencherry-picking": instead of defending something by cherry-picking good examples of it, one attacks something by only picking bad examples of it. In short, most depictions of instrumental rationality by critical theorists only fit particular notions or applications of it.

Stage 2 of the Critique of Instrumental Rationality
I now turn from narrow depictions of instrumental rationality at stage 1 to stage 2: negative evaluations of instrumental rationality and, usually, prescriptions about removing or supplementing it. A common fallacy here is what I call "the out-of-tune violin fallacy." If we hear someone playing a violin out of tune, we do not criticize the violin, because we know it can be played well. Nearly all of the critical theorists mentioned in the previous section move from the type/token fallacy just mentioned to the out-of-tune violin fallacy, criticizing instrumental rationality in general due to particular examples of it.
Another stage 2 fallacy is what I call "the nothingbut-carrots fallacy." If someone dies after eating nothing but carrots, it is fallacious to criticize carrots. We can legitimately criticize someone who tries to live by instrumental rationality alone, but we should not criticize instrumental rationality-because instrumental rationality, within a broader framework of morality and rationality, is desirable.
Yet, critical theorists often attack a world reduced to instrumental rationality as if instrumental rationality is the problem. Thus, Joseph Weizenbaum (1976, 252), drawing on Horkheimer, describes it as "entirely reasonable" to use B52s and napalm to get what you want "if 'reason' means instrumental reason." But this is only right if reason means instrumental reason alone. Incorporating other kinds of reason may preclude napalm and B52s. Instrumental rationality should be praised when it makes useful contributions to broader accounts of reason.
Likewise, as noted above, Murray (2018, 769) claims that instrumental rationality "presupposes a world stripped of social forms and purposes, devoid of moral, aesthetic, gendered, social, legal, and political features-a world without a way of life." Murray's criticism is fallacious in several respects-I mention another shortly-but here the problem is that Murray is attacking a purely instrumental world. A world composed solely of blue things would be unpleasantly monotonous, but we would not criticize the color blue.
Consider too Horkheimer's (2004, 69) criticism of the "complete transformation of the world into a world of means rather than of ends." Habermas (1987a, 112; see also 113-14) later censured Adorno and Horkheimer for the way they "oversimplify [their] image of modernity so astoundingly." But even if Adorno and Horkheimer's gloomy assertion were empirically correct, the normative evaluation is flawed. The problem is not instrumental rationality but the failure to supplement instrumental rationality with more substantive kinds of rationality.
A third stage 2 fallacy is what I will call "the poisoned carrots fallacy." If someone dies from a poisoned carrot, we criticize the poison, not the carrot. Criticizing the carrot is similar to critical theorists attacking Nazis and others who seek instrumentally rational means to evil ends. The problem here is evil ends, not instrumental rationality. After all, we do not criticize those who seek instrumentally rational means to charitable or altruistic ends.
Thus, when Weizenbaum (1976, 256) depicts the Holocaust as "a textbook exercise in instrumental reason," the problem was not instrumental rationality but Nazi morality. The Nazis's opponents also used instrumental rationality, but for better ends. True, some Nazis failed to question their ends, asking only how to reach those ends. But many opponents of fascism simply asked how to reach their non-fascist ends. The morality of the ends matters most.
So, it cannot be right that "[c]apitalism radicalizes guilt because its morality is simply identical with the means/ends rationality of modern instrumental thinking" (Ross 2017, 362). Critical theorists cannot hold both that instrumental rationality is about finding means to ends, whatever those ends are, and that instrumental rationality is about finding means to capitalist ends. By contrast, Honneth (2007, 82, emphasis added) gets the emphasis right in describing "capitalist instrumentalization," not instrumentalization per se.
However, the most damaging fallacy at stage 2 of the critique of instrumental rationality is a version of "the false dilemma fallacy," which oversimplifies choices (e.g., "you are either with us or you are with the terrorists"). Here, the implication is that we must choose instrumental rationality or something else. Denouncing instrumental rationality thus prompts readers to reject it.
One response is that instrumental rationality might be less problematic, perhaps even desirable, if combined with other kinds of rationality. If asked to pick P or Q, we might pick both. I endorse this response later. Another response, consistent with this, is to reject only narrow forms of instrumental rationality, not instrumental rationality in general. When critical theorists start with instrumental rationality as picking means to ends, and then attack pathological forms of this-capitalist ends, patriarchy, and so on-they leave space for forms of instrumental rationality that do not involve such pathologies. If asked to pick P or Q, we might ask if there are alternatives. There definitely are alternatives to the caricatures of instrumental rationality presented by many critical theorists.
The false dilemma fallacy is very common. For example, Gardiner (2000, 7, 38, 85, 90, 96, 134, 148-51) several times contrasts instrumental or "purely" instrumental approaches with "non-instrumental" approaches, as if there is no middle ground (see likewise Benhabib 1986, 224, though not 137-38). The false dilemma fallacy is particularly common among Habermas's followers. Habermas (1984, 10-12, 86-88, 285-86) disparages instrumental rationality (and its social counterpart, strategic rationality) and defends communicative rationality (and its subset, discursive rationality, the rationality of discourse ethics). We might note in passing that Habermas has a decidedly imperfect understanding of instrumental rationality, rational choice theory, and game theory (Heath 1995, 225-28;2001, 4;Johnson 1991, 189-91, 199), and that Habermas's response to Johnson (1991) misses the point (Habermas 1996, 554). But for here, the problem is that Habermas's followers sometimes present this as a choice between instrumental and communicative rationality (e.g., Dryzek 1993, 214, though not Dryzek 1992Forester 1985, 49-51, though not Forester 1993. Instrumental rationality but only in restricted situations. John Dryzek (1990, 14;see also 9, 218;1994, 169), for example, sees instrumental rationality as "often unavoidable," but wants to keep it in "a subordinate domain." Yet all critical theorists need instrumental rationality, in their daily lives and for their political projects. We could not eat or travel without instrumental rationality. We could not even communicate: we constantly make choices about how to communicate our ideas, for example, whether a personal anecdote is a good way of making a general point. Habermas (1979, 68) implies the same: speech is "a medium in which . . . linguistic means . . . are employed instrumentally" (see too Johnson's 1993 critique of Habermas, which shows how game theory informs communicative agreement). Unfortunately, as discussed above, there is ample textual evidence that many critical theorists oppose instrumental rationality as a whole.
In 1993, Habermas (1993, 1-17) changed tack, implicitly accepting instrumental rationality in his account of pragmatic reason. Unfortunately, Habermasian critical theorists applying his ideas to deliberative democracy and suchlike mostly address his 1980s work on communicative action, which was more hostile to instrumental rationality (Blau, 2019).
Instrumental rationality is even needed in Habermasian discursive rationality, although to my knowledge critical theorists do not say so explicitly. As William Rehg (1994, 46-49) and Simone Chambers (1996, 90-91) note, discursive reason must consider knock-on consequences. In Rehg's example, capitalist exchange may seem legitimate to participants in a discourse until they consider the consequences. Consider Nozick's claim that it seems reasonable to freely pay money to watch a sports star, if that star ends up richer, fair enough-literally. But Habermas's (1990, 65) universalization principle, central to discourse ethics, asks if all affected by a norm accept its consequences and side-effects. Are Nozickean capitalist exchanges good means to a desirable kind of society? Might inequality have undesirable effects on health and education, say?
So, instrumental rationality plays some part in Habermasian communicative and discursive rationality (see too Blau, 2019). Instrumental rationality should not be restricted to certain domains. Rather, it should be restricted in terms of how much it controls, by placing it in a broader theory of rationality and morality, whether Habermasian or not.
Instrumental rationality is a bit like a carrot which contributes to our health, as long as we can eat more than just carrots, and as long as it is not poisoned. It is like a violin that can be played in tune or out of tune; we should criticize the latter, but might want the former. And we do not face a choice between instrumental rationality and other kinds of rationality.
In short, critical theorists make important criticisms of instrumental rationality, but these criticisms are mostly too strong. More nuanced criticisms help critical theorists see the value, not just the dangers, of instrumental rationality.

Instrumental Rationality: Responding to Critical Theorists
These criticisms of instrumental rationality by many critical theorists are too strong but still have much power. Particularly important are two claims: instrumental rationality involves fixed ends, precluding the questioning of ends; and instrumental rationality is amoral, permitting immoral actions. Moreover, instrumental rationality's adherents do often depict it in egocentric, economistic ways. I thus offer an account of instrumental rationality to avert these objections. My account will, I hope, be broadly palatable both to traditional defenders of instrumental rationality-my position is broadly orthodox, despite these tweaks-and to critical theorists too.
I start by defining instrumental rationality as the ability to choose good means to ends. An agent (individual or collective) is instrumentally rational to the extent that it can choose optimal means to ends. Instrumental rationality is a matter of degree (Kolodny and Brunero 2016, section 1), although I usually talk dichotomously, for ease of expression.
An "agent" can be an individual or a group, for example, a family. "Ability" means the faculty or faculties by which agents reach decisions. These faculties need not include "reason": instinctively fleeing from a snake may be instrumentally rational (see, similarly, Rawls 1971, 423). Instrumental rationality and emotion are perfectly consistent (Gilboa 2010, 139-42). Emotions provide ends on which instrumentally rational individuals act, for example, love of children (Gilboa 2010, 139), and can be heuristic shortcuts to instrumentally rational actions (Nussbaum 2001, 106-109). Rational decisions can even be subconscious (Becker 1976, 7).
This fits the Habermasian (1984, 9, 11, 15, 17, 22, 115-16; 1996, 119-20) understanding of reason and rationality as the ability to give reasons or good reasons. This idea must include giving reasons counterfactually. When I make instrumentally rational decisions, for example crossing a road, I could have explained my choice, even when I actually do not. This counterfactual view of rationality must also apply in Habermasian communicative rationality. The validity-claims inherent in communicative rationality are usually tacit: we typically understand and agree with each other's propositions without interrogating them. But if we do question a proposition, reasons would have to be given for the communicative understanding and agreement to be rational (see, for example, Habermas 1979, 1-5;1996, 4).
So, instrumental rationality does not necessarily involve the cold, calculating, conscious, cost/benefit computations described by many critics. Nor have I assumed fixed or given ends, as I explain later, or selfinterest, as noted above. And permitting agents to be collective avoids implying anything individualistic, let alone that instrumental rationality opposes communicative or discursive decision-making.

The Essential Contestability of Instrumental Rationality
The next step in deflecting critical-theory critiques of instrumental rationality is more surprising: highlighting instrumental rationality's essential contestability. In an important and oft-overlooked discussion, Rawls (1971, 411-12) notes that "good" means can mean different things, for example, the most efficient means (such as speed or cost), the likeliest means for reaching a goal, and the means giving the greatest degree of a goal. The fastest means may not be the cheapest; the likeliest way of reaching a goal may not provide as much of the goal as something riskier. Reasonable people may thus disagree about which option is best. This is philosophically profound and extremely important. It shows the paucity of many economics textbooks' accounts of instrumental rationality, as criticized by Yanis Varoufakis in the academic stage of his career (see above). Even within economistic frameworks, deciding what is instrumentally rational may involve trade-offs and indifference curves.
Emphasizing essential contestability also refutes any implication by critical theorists that instrumental rationality requires the quickest or cheapest means, say. For example, Werner Bonefeld (2014, 206) states that instrumental rationality "is interested in essentially two things: 'how long did it take?' and 'how much did it cost?'" This is true for some instrumental rationalists but not all. Critical theorists themselves can use instrumental rationality to ask "how many people can we emancipate, and how should we do this?" for example. Quite simply, instrumental rationality itself is unable to indicate what a "good" means to an end is.
Zygmunt Bauman (1989, 18), influenced by Adorno, voices an important concern: "there is nothing in those rules [of instrumental rationality] which disqualifies the Holocaust-style methods of 'social-engineering.'" This commits the poisoned carrots fallacy-Bauman's fundamental concern was Nazi ends, not instrumental rationality. But the key issue here is that "the rules of instrumental rationality" say little to guide our choice of means either. If Nazis want to kill Jews, gays, and Communists, instrumental rationality itself does not say whether the best means are deaths which are fast or slow, cheap or expensive, painless or painful, or public or private. I return to this later, but even normal value-systems will see multiple candidates for "good" means to ends. It is these broader value-systems which clarify instrumental rationality by indicating what "good" means require. Critical theorists who think instrumental rationality requires cheap or quick means are usually assuming capitalist instrumental rationality, which is only one form of instrumental rationality.

Second-Order Instrumental Rationality
I now extend the basic idea of instrumental rationality, to deflect claims that instrumental rationality should be kept in a subordinate domain, for example, instrumental rationality as the rationality of bureaucracy, with communicative rationality as the rationality of democracy (Dryzek 1990, 4-5, 14-22; but see Blau 2011, 41-42, 50-54). My argument further accentuates instrumental rationality's importance for critical theorists.
I thus introduce what I call "second-order instrumental rationality": the ability to pick good means to picking means (or ends). If Aliya wants to get to London, secondorder instrumental rationality lets her choose a good way of picking the means-checking the Internet, asking friends, tossing a coin, and so on.
Second-order instrumental rationality helps Habermasian critical theorists who support deliberative democracy. Consider Innes and Booher's (1999) fascinating Habermasian study which shows how role-playing games foster empathy, helping participants with conflicting interests debate problems open-mindedly. This exemplifies second-order instrumental rationality, helping Habermasians implement deliberative democracy. Innes and Booher almost admit this. They condemn "the rational, instrumental model of seeking means to reach a given end": where "the ends are not clearly known at the outset," we need "nonlinear, holistic" methods of decisionmaking (Innes and Booher 1999, 15-16). But this implicitly accepts nonlinear, holistic methods as the best means for making decisions. Interestingly, in a passing comment, Habermas (1993, 3) accepts something similar: sometimes, "decision-making strategies themselves must be developed." Second-order instrumental rationality, while only a minor development of the standard idea, thus highlights instrumental rationality's place in a broader theory of rationality and morality-even for critical theorists.

Ends as Means: The Cognitive Value of Instrumental Rationality
I now offer a more important contribution, to tackle critical-theory objections that instrumental rationality says nothing about ends. Recall that many critical theorists, quite understandably, define instrumental rationality as picking means to pre-given, fixed ends (e.g., Habermas 1984, 170-72;Horkheimer and Adorno 1972, 36-42). This seems obviously right: after all, instrumental rationality is the ability to pick good means to ends, not the ability to pick good ends.
However, because everything has knock-on consequences (e.g., cookies may cause health problems), instrumental rationality can help us reflect on and maybe pick different ends, by showing how a means or an end might affect other things we value.
Writers from diverse viewpoints have noted this before, including Habermas, as I will shortly discuss.
But such comments are mostly quite brief, and critics of instrumental rationality typically overlook this, regularly lambasting instrumental rationality as solely about means, not ends (e.g., Forester 1993, 49;Horkheimer 2004, 3-4;Tribe 1973, 641, 648). They claim that "instrumental rationality cannot address the issue of desirable goals and norms" (Forester 1993, 49, emphasis added), that instrumental methods "simply cannot address the question of what one's ultimate ends and values ought to be" (Tribe 1973, 641, emphasis added). (Tribe's critique of instrumental rationality derives from Horkheimer: see Tribe 1974.) Dryzek articulates this critique most powerfully. Instrumental rationality is "non-cognitive," that is, values and preferences are "like emotions, beyond the reach of rational argument" (Dryzek 1992, 406), and are thus "given, formed independently of and before the choice at hand" (Dryzek 1996, 406). Yet political actors should be "subjecting their preferences to rational scrutiny and possible adjustment in the interests of determinate collective choices" (Dryzek 1992, 406). This is absolutely right: we need rational ways of discussing and picking ends. Dryzek (1990) was advocating deliberative democracy even before Habermas. Actual democracies, however, are systematically biased against open-minded deliberation, with citizens and politicians acting on existing preferences rather than deliberating and transforming their ideas: we need a more "reflexive" rationality, not instrumental rationality, to question ends rationally (Dryzek 2006, 84-85, 111-23).
I support these comments, especially Dryzek's perceptiveness about modern democracy's bias against openmindedness, a Habermasian insight expressed more incisively than Habermas himself achieves. Nonetheless, these insights do not damage instrumental rationality. I argue this first by applying Joseph Raz's distinction between different kinds of intrinsic value, which to my knowledge has not been used in discussions of rationality. I then highlight the ambiguity of the term "given" ends. After showing that other writers accept my position-including, sometimes, Habermas himself-I reiterate that critical theorists need instrumental rationality.
In The Morality of Freedom, Raz distinguishes between three kinds of intrinsic value: (1) something which is of ultimate value, (2) something which is good in itself but not of ultimate value, and (3) something which is a constituent element of one of the first two kinds of value (Raz 1986, 200-201).
Consider J. S. Mill's account of utility and liberty. Utility in the broad sense is of ultimate value. Liberty is good in itself: a freely chosen way of life is good even if utility does not result. Liberty is also a constituent element of utility: a life of genuine utility must be freely chosen, not coerced. Finally, liberty is also instrumental to utility: giving people freedom makes utility likelier.
Other things are good in themselves and constitutive goods, of course, and many things will be instrumental goods. "Ends" are often simply means to other ends. We have "long chains of instrumental desires," ultimately aimed at something valued in itself, but in practice often aimed simply at (intermediate) instrumental ends (Parfit 2011, 44). Getting to the other side was presumably not the chicken's ultimate goal.
Instrumental rationality's cognitivism may even inform deeply held values. Hobbes and Bentham want us to reflect on what really matters to us. Unlike homo economicus, writes David Schmidtz (1994, 251), "[w]e sometimes stop to wonder whether an end like maximizing profit is worth having." Instrumental rationality thus helps us "be rational in a more reflective sense, calling into question ends we happen to have, revising them when they seem unfit" (Schmidtz 1994, 227, emphasis removed). As Philip Pettit (2006, 142) notes, we can "interrogate" a goal "in the light of other goals that also appeal to us." Even those who think "justice should be done though the heavens fall" may worry if a just act would cause billions of deaths (Thomson 1990, 167).
This explains Horkheimer's error in discussing a judge in a traffic court, bound by instrumental rationality to consider only whether a driver's actions fit the law, respect life and property, and so on. The traffic judge "implicitly assumes that these values must be respected," complains Horkheimer (2004, 7). But the system would collapse if traffic judges started questioning core values, say. Higher level judges do sometimes ask such questions, though, and instrumental rationality can help here, by asking if a law has highly undesirable consequences. In both cases, instrumental rationality helps us question values.
The ambiguous term "given" helps explain why instrumental rationality is unfairly criticized as non-cognitive. "Given" could simply mean "assuming" (e.g., "given assumption P, we should do Q") or "fixed/unchangeable" (as in "condition X is a given: we can't change it"). Critics like Tribe and Forester imply the second sense, but only the first sense is required. As Amartya Sen (2002, 310) writes, depicting preferences as given "does not amount to taking individual preferences . . . to be unchanging or unalterable." So, Raz's typology shows that we can value something both instrumentally and non-instrumentally. Critical theorists tend to talk more dichotomously (an exception being Finlayson 2002, 17, on knowledge as both intrinsically and instrumentally valuable). Ends have knock-on consequences and may even clash. When they do clash, instrumental rationality does not tell us which to pick, without a higher end like utility or peace. But thinking through the consequences of what we value can help us choose. This is one reason why deontologists are "crazy" to ignore consequences (Rawls 1971, 30).
Many writers make similar points. Nozick (1993, 139), who initially states that "instrumental rationality gives us no way to evaluate the rationality of [one's] goals," adds "except as instrumentally effective in achieving further goals taken as given." That, of course, is often the case. Herbert Simon (1983, 7-8, 11) described reason as "a gun for hire" which "cannot tell us where to go; at best it can tell us how to get there," but adds that because these ends may have further consequences, instrumental rationality may help us see whether these intermediate ends are suitable means to our broader objectives-again, a frequent occurrence. Talcott Parsons (1949, 58, 229-30) expands on this: any such "chain" may intersect with other chains, leading to a "complicated web" of "means-end relationships." Even Horkheimer spots this. Above, I quoted his view that instrumental rationality makes it "meaningless" to assess "the superiority of one aim over another in terms of reason." The quotation continues: "such a discussion is possible [in instrumental rationality] only if both aims serve a third and higher one, that is, if they are means, not ends" (Horkheimer 2004, 5). But most ends are means in this respect! This is extremely common and thus extremely important.
Interestingly, some critical theorists' understanding of this aspect of instrumental rationality may have been hampered by a misunderstanding/mistranslation of Max Weber. For Weber (1968, 26), purposive rationality (zweckrationalität) includes "the relations of the end to the secondary consequences." Weber's position is worth clarifying, because Roth and Wittich's edition unhelpfully translates zweckrational not as "purposive" or "goal" rational but "instrumentally" rational. Instrumental and purposive rationality are usually equated for Weber, both by adherents of instrumental rationality (e.g., Elster 2000, 31) and critical theorists (e.g., Benhabib 1986, 183;Murray 2018, 769;Smulewicz-Zucker 2017, 187). This fosters anachronistic readings of Weber (e.g., Elster 2000, 31, 35-36). But the full quotation, which is worth perusing carefully, states that actions are purposively rational when the end, the means, and the secondary results are all rationally taken into account and weighed. This involves rational consideration of alternative means to the end, of the relations of the end to the secondary consequences, and finally of the relative importance of different possible ends. (Weber 1968, 26) In other words, Weber's purposive rationality has three components: (1) what we now call instrumental rationality; (2) consideration of knock-on consequences, which I argued above can be included in instrumental rationality; and (3) consideration of the relative importance of different ends-which is not part of instrumental rationality as most people conceive it. In short, Weber's purposive rationality involves more than instrumental rationality.
I am not saying that considering consequences necessarily entails instrumental rationality, of course. A school student does not use instrumental rationality when asking if mitosis involves cells splitting into two or three, say. But instrumental rationality is entailed when considering consequences is part of considering whether an action is a good means to an end.
Instrumental rationality thus has some cognitive content: ends, values, and preferences are rationally assessable in terms of whether they are good means to other ends, to some extent. My overall position is hardly original, but my use of Raz, my discussion of the ambiguity of "given" ends, and my exposure of a common misreading of Weber support an important point not always recognized by adherents of instrumental rationality-and very rarely by critical theorists.

Ends as Goals and Criteria
Finally, I challenge the crucial argument that instrumental rationality is amoral, in that any means will do-whether those means are moral or immoral-if they facilitate our ends (e.g., Forester 1985, 49;Weizenbaum 1976, 252, 256). My defense partly involves reiterating prior arguments, but I also show that instrumental rationality itself can help deflect some immoral actions.
(Again, I am not questioning the substance of critical theorists' objections. Ruthlessly capitalistic ends, for example, regularly lead to abhorrent actions. But as the poisoned carrots fallacy suggests, the fundamental problem here is ruthless capitalism, not instrumental rationality).
Although instrumental rationality is mostly depicted as the ability to choose good means to an end, we always pursue multiple ends, because any end, even an ultimate end, has knock-on consequences. This is one reason why I earlier defined instrumental rationality in terms of good means to ends, plural.
This opens the door to instrumental rationality potentially precluding immoral actions. Consider this example, heavily adapted from Habermas (1993, 2). Jenny normally cycles to work. One morning her bicycle breaks. She has five options: (1) fix the bicycle, (2) walk to work, (3) stay home, (4) use her old and highly polluting car, or (5) steal a bicycle.
Assume that option 1 does not work and option 2 would take too long. Option 3, staying at home, highlights the previous section's point that instrumental rationality need not assume fixed ends: here, the end of getting to work is (temporarily) dropped. Some people would preclude option 4: environmental protection disqualifies some means and ends. Likewise, most of us would never consider option 5. Even if stealing a bicycle flitted across our minds, we would probably reject it instinctively: it is, almost literally, "not an option." Environmental protection and respecting property are thus ends which act like side-constraints on some people's choices.
Remember that we have multiple ends (see Brennan 2007, 118-19;Sen 1987, 81). I will thus distinguish between two kinds of end: "goals" and "criteria." Jenny's goal here is getting to work; the criteria are time, cost, respecting the environment, respecting property, and so on. Earlier, I noted the ambiguity of "good means" (Rawls 1971, 411-12): even narrow, amoral instrumental rationalists will almost always have more than one criterion. Most people also have moral values, and some are side-constraints on our means. As Sen (1993, 501) notes, our "values and scruples" can affect our choices. Sen finds this inconsistent with rational choice theory, but Daniel Hausman (2007, 56-67) rightly disagrees. Either way, it is consistent with instrumental rationality.
This matters. Whether we talk of multiple ends, or present some ends as criteria/values, it shows that instrumental rationality need not be amoral. Instrumental rationality will thus only pick immoral ends if our ends permit that, as with unconstrained capitalists. Such actions are precluded where instrumental rationality is part of a more moral system.
Rawls's insight about instrumental rationality's essential contestability also applies here. Without criteria, we could almost never decide which means are better. Criteria are only irrelevant in incredibly rare situationsperhaps nonexistent in practice-when there is only one goal, which can either be attained or not, with only one means of reaching it, and no knock-on consequences. Everywhere else, criteria apply.
These criteria need not be economistic ones like speed or cost. They could equally involve respecting property or the environment. Alas, adherents and critics of instrumental rationality often see criteria too narrowly. Tribe's (1973, 618, 628-29) influential critique, for example, slips from instrumental rationality as choosing "efficacious" means, to criticizing cost-benefit analysis. Tribe is quite right that we should care about how we reach our ends (Tribe 1973, 629, 631-34), but criticizing instrumental rationality commits the poisoned carrots fallacy: the fundamental problem is placing instrumental rationality in a narrow cost-benefit framework. Sen (2002, 314;see also 12, 159, 162, 181, 189-92), addressing social choice theory, finds "nothing peculiar" about considering the goodness of processes and outcomes together, as with politicians seeking to be reelected fairly. Exactly: fairness is no less a potential criterion for instrumental rationality than speed or cost. As noted above, instrumental rationality is an essentially contestable concept, and different value-systems will impose different criteria for what makes something a "good" means.
In short, whatever the textbook simplifications, our choices are almost always governed by multiple criteria: to relate means solely to one end is almost inconceivable in reality. Instrumental rationality requires "all things considered" judgments (Varoufakis 1998, 50, emphasis removed): we choose means which best achieve our ends when our ends are considered together (Gert 1998, 773). Some rational choice theorists may see this as involving whatever it takes to reach an end, but most of us should see it more broadly. This crucial reformulation of instrumental rationality is implicit in many accounts, including economists who discuss trade-offs between two criteria (e.g., speed and cost).

Conclusion
Critical theory has been hugely influential, for theorists and empiricists, offering profound insights and prescriptions. But it is weakened by significant overstatements about instrumental rationality. Instrumental rationality is commonly defined too narrowly. Its alleged presuppositions and effects are frequently exaggerated. The ensuing criticisms are too often fallacious, leading most critical theorists to oppose instrumental rationality despite its importance to them.
Rather than rejecting instrumental rationality, critical theorists should censure those who claim to uphold instrumental rationality but who actually support bad means to their ends, or ends which are means to bad ends. This line of criticism is much more incisive. "You claim to be instrumentally rational," critical theorists can say. "But you are picking bad means to your ends, or your ends are means to things you dislike. So much for you being instrumentally rational!" Rather than attacking a caricature of instrumental rationality, critical theorists should attack faulty applications of the idea. They can make the same substantive points-more rigorously, albeit less rhetorically powerfully. This is significant. Underlying these caricatures and criticisms of instrumental rationality are extremely important challenges to the narrow and skewed morality underpinning much of modernity. The fundamental issue is not rationality but morality. This reframing means critical theorists should censure not instrumental rationality itself but impoverished value-systems. Critical theorists can make much the same points, but far more strongly, because adherents of instrumental rationality cannot reject the critique as a caricature. Nor can they reject critical theorists' arguments as self-defeating because critical theorists need instrumental rationality-especially Habermasians, in discursive rationality and for implementing deliberative democracy.
Nonetheless, standard critical-theory objections to instrumental rationality do necessitate some revisions to orthodox understandings of instrumental rationality. Instrumental rationality's adherents should be more explicit that instrumental rationality can apply to groups, need not assume self-interest, is consistent with emotion, may not involve fixed ends, and can help us question our ends. Instrumental rationality's essential contestability is also a deep and potentially troubling insight. While deflecting some critical-theory objections, it raises problems for instrumental rationalists: "good" or "best" means to ends are not necessarily obvious or uncontestable, and the particular means often prioritized by economists can be questioned.
I will conclude with two methodological points. First, future critical-theory studies of instrumental rationality may require more empirical analysis. Empirical analysis adds little if instrumental rationality is bad by definition. But as instrumental rationality is not bad by definition, its consequences depend on the broader framework in which it operates, or how well it is applied. Empirical analysis casts light on when it is beneficial or dangerous (see too Blau 2011, 52-54). Combining normative and empirical analysis always mattered in critical theory, of course, but Adorno and Horkheimer's empirical claims were cherry-picked to support their case; this empirical analysis was inadequate. Much recent critical-theory-inspired empirical work on deliberative democracy is evenhanded (e.g., Steiner et al. 2004), and we need similar research on instrumental rationality more specifically.
Second, what are the implications of the fallacies discussed in the third and fourth sections of this paper? Such fallacious arguments are widespread-and are found far beyond critical theory. Countless people disparage or spurn poststructuralism because part of it is nonsense, for example, or reject quantitative social science because of some mindless number-crunching. Conflating a thing with some examples of the thing helps "motivated reasoning," where we unconsciously accept inadequate evidence that supports a conclusion we want to reach.
Critical theorists who attack instrumental rationality thus exemplify fallacious arguments that are actually extremely common. And these fallacies undermine critical theorists' emancipatory goals. Of course, such intellectual constraints are nowhere near as important as the structural constraints that critical theorists primarily oppose. But they are constraints nonetheless. They stop us from freely and open-mindedly considering the strengths and weaknesses of things that may actually help us-like instrumental rationality.