Fitting Things Together defends the distinctive normativity of structural rationality, which requires one’s mental states to fit together correctly. More specifically, Worsnip argues that structural rationality is “[a] genuine, [b] autonomous, [c] unified, and [d] normatively significant” (x). (a) and (b) are part of his case for rationality dualism, “the view that structural and substantive rationality are two distinct but equally genuine kinds of rationality, neither … reducible to the other” (4). His account of (c) establishes a link between structural rationality and the metaphysics of attitudes: structurally irrational combinations of attitudes are ones that agents are disposed, owing to the nature of the attitudes at issue, to revise under “full transparency” (133). His take on (d) meshes with this proposal: structural rationality matters because it is fitting to structure deliberation in ways that treat incoherent combinations as “off-limits” (256).The book has three parts. The first supports rationality dualism. The second develops Worsnip’s conception of the unity and normativity of structural rationality. The third explores the implications for topics in epistemology and metaethics. I’ll focus on the first two parts, but I’ll mention aspects of the third in passing. Research on many of the topics Worsnip discusses will benefit from attention to his views. But the traffic goes both ways: some predictions of his outlook for these topics may provide reasons to rethink it.Let’s examine Worsnip’s case for bifurcating rationality. He isolates structural rationality by noting, as is standard, that requirements of coherence seem to explain the irrationality in some examples. The examples Worsnip highlights are mainstream: means-end incoherence, inconsistency, akrasia, and cyclical preferences. These examples highlight the negative, peremptory side of structural rationality, predisposing readers to understand it as the absence of incoherence, as his “wide-scope in spirit” view holds (chap. 6).From such examples, Worsnip infers some hallmarks of structural rationality (7):∙ It is circumstance-independent and evidence-independent.∙ It is ‘formal’: structural irrationality violates a pattern that abstracts from content.∙ Structural rationality judgments “can be made in abstraction from disagreements about what is worth doing.”He notes that these hallmarks make structural rationality narrow (17–19). He narrows it further when separating requirements of structural rationality from norms of correct reasoning (180) and apparent reasons (35).Although Worsnip is mostly making explicit a conception T. M. Scanlon (1998) and John Broome (2013) fostered, it is worth considering whether one can broaden the diet of examples without changing the subject. As he partially acknowledges, other rationality evaluations have his hallmarks but suggest a notion of “fitting-togetherness” that is not just the absence of synchronic incoherence (18–19). Evaluations of reasoning qua reasoning are good examples. Michael Smith has also long recommended nonarbitrariness and systematicity as demands of structural rationality, which he invokes in a Kantian analysis of reasons. While Smith’s program may fail, it draws attention to examples that could be treated as paradigmatic, and it revives rationalist themes that merit attention. While I don’t think we are rationally required to believe the principle of sufficient reason, there may be related structural demands. Consider the puzzlingness of believing p, believing there is objective reason to believe p, believing one doesn’t understand why p, yet lacking any inclination to understand why. Such cases suggest an aspirational side to structural rationality. While it is only a rational imperfection to believe without understanding, this suggests an imperfect duty to seek understanding if one finds it lacking.A different demand that deserves more attention is a principle of sufficient apparent reason: rationality demands φ-ing only if it appears to you that there are objective reasons to φ. Worsnip thinks this demand is either evidence-dependent or reduces to anti-akrasia (chap. 2). But while structural rationality is independent of a posteriori evidential relations, it may not be independent of the a priori demand to respect appearances. Following Kant, one might think this demand reflects a structural requirement to harmonize one’s receptive and spontaneous mental capacities. This explains why it is hard to imagine subjects whose attitudes float free from the way the world appears to them: it is better to suppose that the world looks different, say, if one has schizophrenia.Appearances aren’t well treated by the categories Worsnip uses to frame substantive rationality in chapter 2. He grounds substantive rationality in ‘evidence-relative reasons’, which he contrasts with ‘fact-relative’ and ‘belief-relative’ reasons. Appearances are not fact-relative or belief-relative reasons. They aren’t evidence-relative reasons either. Those are more demanding: as Worsnip’s discussion of higher-order evidence in chapter 3 suggests, one can have conclusive evidence for p even though it doesn’t appear from one’s overall perspective that p (and wouldn’t on reflection). Holding that structural rationality is coherence with one’s overall perspective, including appearances, seems more fitting.There are other cases where Worsnip’s divisions feel ill-fitting. In chapter 6, Worsnip maintains that structural rationality is synchronic and not concerned per se with the form of one’s processing. These claims oppose Kolodny’s idea that rational requirements concern the structure of one’s reasoning. Worsnip instead suggests a trinitarian view with a distinctive category of norms of reasoning (180n31). This feels artificial: it is more natural to start by treating fallacious reasoning as a paradigm of deficient structural rationality, since it displays the hallmarks.Let’s turn to Worsnip’s case against rationality monism. Worsnip argues first against substantivist views that reduce structural rationality to substantive rationality or eliminate it (chap. 3), and then against structuralist views that reduce substantive rationality to structural rationality (chap. 4). Focusing on Benjamin Kiesewetter and Errol Lord, Worsnip assumes substantivists need a Guarantee Hypothesis, on which structural irrationality guarantees the presence of some substantively irrational attitude(s). He then makes two moves. First, he argues that even if the Guarantee Hypothesis is true, there are decisive objections to substantivism. Second, he gives counterexamples to the Guarantee Hypothesis and argues that substantivists cannot explain them away.Worsnip’s case against substantivist eliminativism appeals to two intuitions. One is the ‘counting intuition’: incoherent subjects with substantively irrational attitudes make two kinds of mistakes. Another is the intuition that coherent subjects are more rational in one respect than incoherent subjects. Worsnip’s case against Lord’s and Kiesewetter’s diagnoses of these intuitions may succeed. But he neglects the diagnosis of Niko Kolodny’s (2005) ‘Transparency Account’. He considers a later statement of Kolodny’s view, taking Kolodny to agree that there is one genuine structural requirement—anti-akrasia. But this ignores a key theme in Kolodny (2005: 509): “The normative ‘pressure’ that we feel, when rational requirements apply to us, derives from … the reasons that, as it seems to us, we have.” Kolodny stresses that “this account does not appeal to … an additional normative concept beyond that of a reason” (558).Friends of this approach can agree that there is a distinctive nonnormative property of coherence but hold that it has no essential significance. Preface cases recommend this view: here it seems more rational to become inconsistent by humbly allowing that one surely made some mistake, but couldn’t find it. This point reveals a problem for Worsnip. His best counterexamples to the Guarantee Hypothesis—preface cases—undermine the intuitions he wields against opponents. Preface cases make it plausible that incoherence is insufficient for irrationality: what could be a clearer example of incoherence, after all, than having beliefs that straightforwardly entail a contradiction? Such cases recommend the following re-diagnosis: incoherence ensures a mistake of rationality only when, by being incoherent, one thereby invites substantive irrationality.Let’s consider structuralism. Worsnip starts with arguments against structuralist eliminativism, then attacks Humean and Kantian reductive views. He suggests that Broome’s eliminativism assumes an overly narrow conception of rationality that ignores the rational significance of evidence-relative reasons (95–99). But defenders of a similar view could respond that rationality supervenes on apparent reasons, which aren’t relative to evidence in the natural sense but rather appearances, where the requirement to respect appearances reflects a structural demand to harmonize two mental capacities.Worsnip’s case against reductionism doesn’t do it full justice. He assumes the best views are “counterfactual and idealizing,” invoking “the attitudes that an idealized counterpart of us would have, where the relevant idealization … mak[es] the counterpart structurally rational” (99). Some of Worsnip’s objections are compelling: counterfactual views suffer from indeterminacy, and Humean views fail to explain moral reasons. He doesn’t, however, give a decisive case against Kantianism.Worsnip believes that Kantians in the end must appeal to substantive rationality. But Kant’s view plausibly rests on a regress argument that makes a point about the structure of practical reasoning. It is only intelligible to pursue the means to one’s ends if it makes sense to have them in the first place. How, then, does practical reasoning get started? Kantians deny that it can intelligibly begin with externally dictated ends: this is arbitrary and heteronomous. One can terminate the regress of practical reasoning autonomously only if there are ends that one can necessarily will autonomously. There will be, if some ends are constitutive of practical reason. Here Kantians use an analogue of the cogito argument: some ends cannot intelligibly be questioned if one is to have any end at all. Chief among them is reason’s capacity to set laws for itself: willing presupposes the authority of this capacity.Discussing Markovits’s relative of this argument, Worsnip reasonably suggests that the claim that humanity is a special end is “a substantive judgment” (113), but no such judgment features in the version above. He may insist I appealed to a substantive judgment in holding that it is arbitrary and heteronomous to submit to externally dictated ends, but this is less obvious. Anti-arbitrariness and anti-heteronomy can be framed as structural requirements. Kantians could then propose that it is substantively rational to φ iff there is a well-founded pattern of reasoning that begins with the ends that are constitutive of practical reason and concludes with φ-ing. One might doubt this view ensures the substantive rationality of morality. But it shouldn’t therefore be excluded. Worsnip’s evidence-relative view doesn’t do so either, if developed plausibly.There are epistemic analogues of this story. Kantians in epistemology defend a priori canons of theoretical reasoning via transcendental arguments that mirror the practical regress argument. Worsnip doesn’t consider such views. The coherentist views he considers are better cast in this spirit (116–18). Rather than holding that ‘a belief of yours is supported by evidence iff it coheres with your other beliefs’, one could say that a belief is supported by E iff it makes sense of E, where there are a priori principles of sense-making grounded in the constitution of theoretical reason. Such views needn’t appeal to ‘substantive judgments about support relations’: it is incoherent to think in ways that defy the constitution of theoretical reason. While the approach requires a disputable view about the nature of theoretical reason, it is not a substantive view in the relevant sense: Worsnip’s account of structural rationality rests on a similar view.Let’s consider Worsnip’s positive views. His account of the unity of structural rationality seeks to explain why it is puzzling (129). Focusing on incoherence, he notes that it is very puzzling: incoherent tangles of attitudes seem only attributable given failures of transparency, and are liable to unravel once spotted. This suggests a connection between coherence and the metaphysics of attitudes: “a set of attitudinal mental states is jointly incoherent iff it is (partially) constitutive of (at least some of) the states in the set that any agent who holds this set of states has a disposition, when conditions of full transparency are met, to revise at least one of those states” (133). Worsnip notes that this proposal gives some support to constitutivism and the idea that norms of rationality are preconditions of interpretation (150). But these ideas allegedly only work for structural rationality (156).The basic shape of this view is compelling, but a less restrictive version explains more. More cases are relevantly puzzling. Consider reasoning that radically flouts inductive and abductive canons, gruesome conceptual schemes, and paranoid beliefs. While such errors are conceivable, it seems inconceivable to engage in theoretical reasoning without commitment to the ideals that explain them. It is very puzzling if such shortcomings are noted but don’t incline one to return to the drawing board.1 The underlying explanation is plausibly constitutivist: besides being constrained by constitutive prohibitions against incoherent combinations, our reasoning is regulated by constitutive ideals. These are not compliance constitutive (149)—reasoners needn’t be disposed to meet them—but they are commitment constitutive: we must be guided by them to count as having rational capacities.This story helps with the normativity of rationality. Worsnip thinks structural rationality is normative because there are reasons to structure deliberation in ways that treat incoherent combinations as off-limits. But why structure deliberation this way? Worsnip offers a “rather uninformative” answer: “because it’s fitting to structure deliberation in ways that respect coherence constraints” (265). He offers a little more by noting that his view of incoherence “drive[s] home just how little sense these combinations of attitudes make” (266). But similar facts make it unfitting to commit to counterinductive and counterabductive canons of ampliative inference, invalid deductive patterns, and gruesome conceptualization. Inference and conceptualization are constitutively regulated by ideals. If one asks why they matter, the answer is that we’re committed to them in virtue of being rational beings.This is not Worsnip’s proposal: he thinks it is a substantive fact that it is fitting to treat structural irrationality as off-limits. But rational beings cannot but see such combinations as off-limits. If this weren’t true, I doubt it would be clear that it is necessarily fitting to avoid them, any more than it is to shun odd socks. Indeed, when inconsistency is something we can invite—for example, in preface cases—it is accepted as an outcome of our crooked timber. To insist that it still is unfitting to tolerate inconsistency here, as Worsnip does (270–71), is to worship the hobgoblin of consistency. While Worsnip admits that it is permissible to be incoherent here (303), this sits unevenly with the claim that it is unfitting: better to say that it is fitting to tolerate some inconsistency, given our inescapably crooked frame.Further explanation of this fittingness is needed. The Kantian option is to say that global consistency and fit with appearances are constitutive ideals of reasoning, associated with imperfect duties of rationality. There is no conflict of requirements, just two ideals. A wider constitutivist approach, which includes not just compliance-constitutive requirements but commitment-constitutive ideals, allows for this. This approach takes the question of why rationality matters seriously, without supposing that an answer must bottom out in the promotion of value. To decline the question with an appeal to fittingness seems no less ‘tyrannical’ than the instrumentalist outlook Worsnip rightly rejects in the coda.

中文翻译：

---

 将事物组合在一起


《将事物整合在一起》捍卫了结构理性独特的规范性，这要求一个人的心理状态能够正确地整合在一起。更具体地说，沃斯尼普认为结构理性是“[a]真正的、[b]自主的、[c]统一的、[d]具有规范意义的”(x)。 (a) 和 (b) 是他的理性二元论案例的一部分，“结构理性和实体理性是两种不同但同样真实的理性，两者都不能还原为另一种”(4)。他对（c）的解释在结构理性和态度形而上学之间建立了联系：由于所讨论态度的性质，行为者倾向于在“完全透明”下修改态度的结构性非理性组合（133） 。他对（d）的看法与这一建议相吻合：结构合理性很重要，因为它适合以将不连贯的组合视为“禁区”的方式进行结构审议（256）。本书分为三个部分。第一个支持理性二元论。第二个发展了沃斯尼普的结构理性的统一性和规范性概念。第三部分探讨了认识论和元伦理学主题的含义。我将重点关注前两部分，但我会顺便提及第三部分的各个方面。对沃斯尼普讨论的许多主题的研究将受益于对他的观点的关注。但流量是双向的：他对这些话题的前景的一些预测可能会提供重新思考的理由。让我们来看看沃斯尼普的理性分歧的案例。他按照标准指出，连贯性的要求似乎可以解释某些例子中的非理性，从而分离出结构理性。 沃斯尼普强调的例子是主流的：手段与目的不一致、不一致、失控和周期性偏好。这些例子突出了结构理性消极的、专制的一面，使读者容易将其理解为缺乏不连贯性，正如他的“精神上的宽广范围”观点所主张的那样（第 6 章）。从这些例子中，沃斯尼普推断出结构理性的一些特征。理性 (7)：∙ 它是独立于环境和证据的。∙ 它是“形式的”：结构非理性违反了从内容中抽象出来的模式。∙ 结构理性判断“可以从关于什么值得做的分歧中抽象出来”他指出，这些特征使结构理性变得狭窄（17-19）。当将结构理性的要求与正确推理的规范（180）和明显的理由（35）分开时，他进一步缩小了范围。尽管沃斯尼普主要明确了 T. M. Scanlon (1998) 和 John Broome (2013) 提出的概念，但值得考虑是否可以在不改变主题的情况下扩大例子的范围。正如他部分承认的那样，其他理性评估也有他的标志，但提出了“契合性”的概念，而不仅仅是缺乏共时性的不连贯性（18-19）。对推理作为推理的评估就是很好的例子。迈克尔·史密斯也长期以来一直建议将非任意性和系统性作为结构理性的要求，他在康德式的理性分析中引用了这一点。虽然史密斯的计划可能会失败，但它引起了人们对可以被视为范例的例子的关注，并且它复兴了值得关注的理性主义主题。虽然我认为我们没有理性地要求相信充分理由原则，但可能存在相关的结构性要求。 想想相信 p 的令人费解的地方，相信有客观理由相信 p，相信一个人不理解为什么 p，但却缺乏理解为什么的倾向。这些案例表明了结构理性的理想一面。虽然在没有理解的情况下相信只是一种理性的缺陷，但这表明如果一个人发现缺乏理解，寻求理解的不完美义务。值得更多关注的一个不同要求是充分明显理由的原则：理性仅在出现时才要求 φ-ing向您表明 φ 是有客观原因的。沃斯尼普认为这种需求要么依赖证据，要么简化为反阿克拉西亚（第 2 章）。但是，虽然结构理性独立于后验证据关系，但它可能并不独立于尊重表象的先验要求。按照康德的观点，人们可能会认为这种需求反映了协调一个人的接受能力和自发心智能力的结构性要求。这解释了为什么很难想象主体的态度会不受世界对他们的呈现方式的影响：最好假设世界看起来有所不同，例如，如果一个人患有精神分裂症。沃斯尼普类别并没有很好地对待外表他在第二章中用来构建实质性合理性。他将实质性合理性建立在“证据相关的原因”之上，并与“事实相关”和“信念相关”的原因进行对比。外表不是与事实相关或与信仰相关的原因。它们也不是证据相关的原因。这些要求更高：正如沃斯尼普在第 3 章中对高阶证据的讨论所表明的那样，人们可以拥有 p 的确凿证据，即使从整体角度来看，p 似乎并不存在（反思起来也不会如此）。 认为结构理性是与一个人的整体观点（包括外表）一致，似乎更合适。在其他情况下，沃斯尼普的划分也让人感觉不合适。在第六章中，沃斯尼普认为结构理性是共时性的，其本身与一个人的加工形式无关。这些主张反对科洛德尼的观点，即理性要求涉及一个人的推理结构。相反，沃斯尼普提出了一种三位一体的观点，具有独特的推理规范类别（180n31）。这感觉很人为：更自然的做法是，首先将错误推理视为结构理性缺陷的范式，因为它显示出其特征。让我们转向沃斯尼普反对理性一元论的案例。沃斯尼普首先反对将结构理性还原为实质性理性或消除结构主义的观点（第3章），然后反对将实质理性还原为结构理性的结构主义观点（第4章）。沃斯尼普以本杰明·基瑟韦特和埃罗尔·洛德为重点，假设实质主义者需要一个保证假说，在这个假设上，结构性非理性保证了某些实质上非理性态度的存在。然后他做了两个动作。首先，他认为即使保证假说是正确的，也存在对实体主义的坚决反对。其次，他给出了保证假说的反例，并认为实体论无法解释它们。沃斯尼普反对实体主义取消主义的案例诉诸了两种直觉。一是“计数直觉”：语无伦次、态度实质上非理性的主体会犯两种错误。另一个是直觉，即连贯的主体在某一方面比不连贯的主体更理性。 沃斯尼普反对洛德和基塞韦特对这些直觉的诊断的案例可能会成功。但他忽视了 Niko Kolodny (2005)“透明度账户”的诊断。他考虑了科洛德尼后来的观点陈述，认为科洛德尼同意存在一个真正的结构性要求——反阿克拉西亚。但这忽略了科洛德尼（Kolodny，2005：509）的一个关键主题：“当理性要求适用于我们时，我们感受到的规范‘压力’来自……在我们看来，我们拥有的理由。”科洛德尼强调，“这种解释不诉诸……超出理由的额外规范概念”（558）。这种方法的支持者可以同意，一致性有一个独特的非规范属性，但认为它没有本质意义。前言案例推荐了这种观点：在这里，谦虚地承认自己肯定犯了一些错误，但找不到它，从而变得不一致，这似乎更理性。这一点暴露了沃斯尼普的一个问题。他对保证假说的最佳反例——序言案例——破坏了他针对对手的直觉。前言案例使得不连贯不足以构成非理性似乎是合理的：毕竟，还有什么比拥有直接导致矛盾的信念更明显的不连贯的例子呢？这些案例建议进行以下重新诊断：只有当由于不连贯而招致实质性的非理性时，不连贯才确保了理性的错误。让我们考虑一下结构主义。沃斯尼普首先反对结构主义取消主义，然后攻击休谟和康德的还原论观点。他认为布鲁姆的取消主义假设了一种过于狭隘的理性概念，忽视了证据相关原因的理性意义（95-99）。 但类似观点的捍卫者可能会回应说，理性源于明显的原因，这些原因与自然意义上的证据无关，而是与表象相关，其中尊重表象的要求反映了协调两种心理能力的结构性需求。 沃斯尼普反对还原论的案例并没有完全公正。他认为最好的观点是“反事实和理想化的”，援引“我们的理想化对手所具有的态度，其中相关的理想化……使对手具有结构理性”（99）。沃斯尼普的一些反对意见是令人信服的：反事实观点存在不确定性，休谟观点无法解释道德原因。然而，他并没有给出反对康德主义的决定性理由。沃斯尼普认为，康德主义者最终必须诉诸实质理性。但康德的观点似乎建立在回归论证之上，该论证提出了关于实践推理结构的观点。只有当首先拥有这些手段有意义时，追求达到目的的手段才是可以理解的。那么，实践推理如何开始呢？康德主义者否认它可以理解地以外部决定的目的开始：这是任意的和他律的。只有当存在人们必然能够自主地实现的目标时，人们才能自主地终止实践推理的倒退。如果某些目的是实践理性的组成部分，那么就会有。康德主义者在这里使用了我思论证的一个类比：如果一个人想要有任何目的，那么某些目的就无法被理解地质疑。其中最主要的是理性为自身制定法律的能力：意愿以这种能力的权威为前提。在讨论马可维茨与这一论点的相对论时，沃斯尼普合理地认为，人类是一种特殊目的的主张是“一种实质性判断”（113），但上述版本中没有这样的判断。他可能会坚持认为我诉诸了实质性判决，认为服从外部规定的目的是任意的和他律的，但这不太明显。反任意性和反他律性可以被框定为结构性要求。然后，康德主义者可以提出，如果存在一种有充分根据的推理模式，则 φ 是实质理性的，该模式始于构成实践理性的目的，并以 φ-ing 结束。人们可能会怀疑这种观点确保了道德的实质合理性。但不应该因此被排除在外。沃斯尼普的证据相对论观点如果发展得合理的话，也不会这样做。这个故事有认知上的类似物。认识论中的康德主义者通过反映实践回归论证的先验论证来捍卫理论推理的先验准则。沃斯尼普不考虑这样的观点。他认为连贯论观点更好地体现了这种精神（116-18）。人们不应该认为“你的一个信念得到证据支持，只要它与你的其他信念一致”，而是可以说，一个信念得到 E 的支持，只要它使 E 有意义，其中存在基于意义建构的先验原则。在理论理性的构成中。这些观点不需要诉诸“关于支持关系的实质性判断”：以违背理论理性构成的方式思考是不连贯的。 虽然这种方法需要对理论理性的本质有一个有争议的观点，但它并不是相关意义上的实质性观点：沃斯尼普对结构理性的解释也建立在类似的观点之上。让我们考虑一下沃斯尼普的积极观点。他对结构理性统一性的解释试图解释为什么它令人费解（129）。在谈到语无伦次时，他指出这是非常令人费解的：态度语无伦次的混乱似乎只能归因于透明度的失败，而且一旦被发现就很容易被瓦解。这表明连贯性和态度形而上学之间的联系：“一组态度心理状态是共同不连贯的，当且仅当它（部分）构成集合中的（至少一些）状态，任何持有这组状态的主体当完全透明的条件得到满足时，有权至少修订其中一项规定”（133）。沃斯尼普指出，这一提议为构成主义和理性规范是解释的前提条件的观点提供了一些支持（150）。但据称这些想法仅适用于结构合理性（156）。这种观点的基本形式令人信服，但限制较少的版本解释了更多。更多的案例也令人费解。考虑一下从根本上藐视归纳和溯因准则、可怕的概念方案和偏执信念的推理。虽然这些错误是可以想象的，但在不致力于解释这些错误的理想的情况下进行理论推理似乎是不可想象的。如果这些缺点被注意到但又不促使人们重新开始，那将是非常令人费解的。1 潜在的解释似乎是构成主义的：除了受到针对不连贯组合的构成性禁令的约束之外，我们的推理还受到构成性理想的调节。这些不是顺从的本构性（149）——理性者不需要倾向于满足它们——但它们是承诺的本构性：我们必须受它们的指导才能算作具有理性能力。这个故事有助于理性的规范性。沃斯尼普认为结构理性是规范性的，因为有理由以将不连贯的组合视为禁区的方式来构建审议。但为什么要以这种方式进行审议呢？沃斯尼普提供了一个“相当无信息”的答案：“因为它适合以尊重连贯性约束的方式构建审议”（265）。他还提供了更多内容，指出他对不连贯性的看法“让人明白这些态度的组合是多么没有意义”（266）。但类似的事实使得我们不适合遵守反归纳法和反溯因法的放大推理、无效的演绎模式和可怕的概念化规范。推理和概念化在本质上受到理想的调节。如果有人问它们为什么重要，答案是我们作为理性存在而致力于它们。这不是沃斯尼普的建议：他认为将结构性非理性视为禁区是合适的，这是一个实质性的事实。但理性的人不能不认为这种组合是禁区。如果这不是真的，我怀疑避免它们是否一定是合适的，就像避开奇怪的袜子一样。事实上，当我们可以邀请不一致时——例如，在序言的情况下——它被认为是我们歪曲木材的结果。 坚持认为这里仍然不适合容忍不一致，就像沃斯尼普所做的那样（270-71），这是对一致性的崇拜。虽然沃斯尼普承认这里不连贯是允许的（303），但这与它不合适的说法不一致：更好的说法是，考虑到我们不可避免的弯曲框架，容忍一些不一致是合适的。对这种合适性的进一步解释是需要。康德的选择是说，整体一致性和与表象的契合是推理的本构理想，与不完美的理性义务相关。没有要求的冲突，只是两个理想。更广泛的本构主义方法，不仅包括合规本构要求，还包括承诺本构理想，可以实现这一点。这种方法提出了为什么理性如此重要的问题，而不假设答案必须在价值提升中触底。以诉诸合适性来拒绝这个问题，似乎与沃斯尼普在尾声中正确拒绝的工具主义观点一样“专横”。