Mark Schroeder’s latest book delves deeper into the topic of normativity and reasons, while moving his focus from ethics to epistemology. His central aims are, first, to argue that theorizing in normative epistemology profits from comparison with other normative domains (his “Core Hypothesis” [9]); and second, to defend a picture of epistemic normativity that puts reasons first: they can be used to explain and analyze all other epistemic normative phenomena.Part 1 of the book provides a compelling account of normative reasons as competitors (which compete in determining, for instance, what one ought to do or believe) that are act-oriented rather than outcome-oriented, and can be acted on (my reasons to φ can be the reasons for which I φ). Schroeder assumes that there are both objective reasons, which bear on the correctness of belief, and subjective reasons, which determine its rationality or justification.Part 2 aims to solve the problem of unjustified belief for Reasons First epistemology. On this view, normative standings such as justification/rationality and knowledge bottom out in epistemic reasons. Yet it seems that only justified belief or knowledge can provide a subject S with reasons, so that we cannot take reasons as fundamental. So, apparently, perceptual experience—given that it itself is neither knowledge nor justified—cannot provide us with reasons or evidence. But this cannot be right, since perceptual experience is undoubtedly a privileged source of evidence concerning our surroundings. According to Schroeder, to allow for perceptual justification, we need a world-implicating conception of perceptual evidence, as endorsed by disjunctivism, which takes evidence to entail truths about the external world. At the same time, and contrary to disjunctivism, we must conceive of such evidence as nonfactive—it does not have to be true (or consist in a relation to a truth) and so is available not only in the good case of veridical perception but also in illusion or hallucination. Whether S’s belief is rational cannot hinge on minimal differences, as implied by disjunctivism. Schroeder illustrates this with a pair of cases C1 and C2 that are identical except that in C1, S undergoes a veridical perception, and in C2, she undergoes an indistinguishable illusion. (Say, in C1, S is looking at a red ball, but in C2, she is facing a white ball that appears red due to red lighting.) But, importantly, this illusion is a one-time occurrence—S has an otherwise flawless perceptual track record in C1 and C2. In both cases, S’s belief is equally rational, or so Schroeder argues.Schroeder thus rejects disjunctivism. Instead, he endorses the apparent factive attitude view: basic perceptual reasons are—nonfactive—subjective reasons, such as the proposition that I see that the ball is red. But since they entail worldly facts (such as: the ball is red), they are nonetheless world-implicating. For me to possess the reason, it has to appear to me that I see that the ball is red.In part 3, Schroeder addresses the problem of sufficiency—the worry that the balance of epistemic reasons by itself cannot properly determine whether a belief is justified or knowledge. For it is unclear by how much S’s evidence that p has to outweigh her evidence that not-p so as to render her belief that p justified or knowledge. It might appear that solving this problem forces us to abandon Reasons First epistemology since it presupposes the concept of rationality: what is needed is enough evidence to make belief rational. The related problem of near ties is that evidence that p that is at least as good as evidence that not-p does not always rationalize belief that p; in cases of (near) ties between evidence pro and con, withholding belief is rational instead. By contrast, in the practical case, reasons to φ that are at least as strong as reasons not to φ insure that φ-ing is rational. How might this difference be explained? A further puzzle discussed by Schroeder is raised by cases that have motivated philosophers to defend pragmatic encroachment. For instance, in Jason Stanley’s (2005) famous bank cases, in the high-stakes scenario, the believer plausibly needs more evidence in order to form a justified belief or to know than in the corresponding low-stakes scenario.Schroeder uses these and other puzzles to argue that there must be nonevidential epistemic reasons against believing that enter into the balance of reasons; the overall balance then determines which attitude is epistemically justified, thereby preserving Reasons First epistemology. Nonevidential reasons against belief include, among others, facts about the future availability of evidence or about the costs of error—that is, the costs of having or acting on false belief. Such reasons are ubiquitous and so can explain why in cases of near ties, the overall balance of reasons, which includes nonevidential reasons against belief in addition to the (nearly) balanced evidential reasons, does not rationalize belief that p. Regarding the problem of sufficiency: taking on board nonevidential reasons, the balance of epistemic reasons can all by itself determine the epistemic standing of belief.True to his core hypothesis, Schroeder uses comparisons with reasons for other attitudes, such as admiration, to argue that some nonevidential reasons against belief are right-kind reasons, which affect a belief’s epistemic standing. Right-kind reasons are considerations to which belief must be sensitive in order to successfully play its role in our cognitive economies—the role of providing us with considerations to rely on in reasoning by default, or as a matter of policy. Beliefs do so by simplifying reasoning and decision-making, as compared to credences, since they allow us to ignore remote error possibilities. According to Schroeder’s Pragmatic Intellectualism, facts about the (practical or moral) costs of error are then right-kind reasons against belief because they are considerations to whose truth belief must be sensitive to play its cognitive role well. The costs of error explain why otherwise sufficient evidence fails to justify belief in high-stakes scenarios: they are so high that they outweigh the subject’s evidence that p.In part 4, Schroeder details his Kantian account of knowledge. He analyzes knowledge as believing well, where the belief is a response to reasons that are jointly subjectively and objectively sufficient to outweigh all existing reasons against the belief, and thus sufficient to make it rational and correct. He motivates the account by analogy with right-reasons accounts of the moral worth of actions and by focusing on how reasons generally give rise not only to standards of correctness for a certain response, but—as something for which we can act—also to standards of acting well.I now turn to a critical comment on Schroeder’s argument against disjunctivism sketched above. He presents perceptual reasons in a propositional idiom—the idea being that we can attribute reasons by way of associated propositions without being committed to any particular ontology of reasons (41). For instance, the phrase “that S sees that the ball is red” can be used equally to characterize S’s mental state of seeing that the ball is red or the consideration that S sees that the ball is red, as her epistemic reason. My worry is that using a propositional idiom encourages a misconception of disjunctivism—namely, that it is committed to a hard line between veridical perception, on the one hand, and illusion and hallucination, on the other. Put differently, the propositional idiom encourages the thought that the only version of metaphysical disjunctivism available to the epistemological disjunctivist is a V v IH view. This misconception undergirds Schroeder’s argument against disjunctivism. Let me elaborate.We can identify the epistemological disjunctivist with Schroeder’s disjunctivist opponent in the book, who endorses the view that perception provides factive, world-implicating reasons. By contrast, metaphysical disjunctivism concerns the nature of perceptual experience. Crudely put, it states that veridical perception is a mental state of a fundamentally different nature than hallucination. Regarding the further perceptual state of illusion, it comes in two varieties, V v IH and VI v H (Byrne and Logue 2009: xi). V v IH says that veridical perception as the good case is fundamentally distinct from the bad case of illusion or hallucination; VI v H holds that the good case includes veridical perception and illusion, which share the same nature and differ fundamentally from the bad case of hallucination.If we characterize the reasons provided by perceptual experience propositionally, the truth of V v IH may seem obvious. Take my visual experience as of a red ball. An associated proposition is that the ball is red. For my experience to be veridical is for this proposition to be true, as in perception. In illusion and hallucination, the associated proposition turns out false—in illusion, for instance, the white ball looks red even though it is not; in hallucination, there is not even a ball that is visually presented. This suggests a metaphysical classification of illusion with hallucination as the bad case.If, however, we think of perceptual experience as fundamentally nonpropositional and instead directed at objects, it is natural to say—with VI v H—that perception and illusion both successfully acquaint us with objects. Accordingly, both my veridical and my illusory experience acquaint me with the ball in front of me. Illusion merely involves error with respect to how the object is presented—in the example, the white ball is misleadingly presented as red. This metaphysical picture of perceptual experience can be combined with the epistemological claim that the reasons provided by perception are the very objects with which we are acquainted (Brewer 2018). For the epistemological disjunctivist, this opens up the possibility that the subject’s epistemic rationality in the illusory and veridical cases may be on a par.Reconsider Schroeder’s argument that disjunctivism implausibly entails that a minimal one-time difference between good case C1 and bad case C2 makes for a major difference with respect to the subject’s rationality. According to VI v H, the minimally different case in which the subject undergoes an illusion instead of an indistinguishable veridical experience (64) is a good case. Perceptual illusion provides the same objectual reason as veridical perception. In the subject’s misleading circumstances and given her objectual reason, it is rational, at least in the sense of perfectly understandable, that she forms a false perceptual belief; her belief is justified, but not knowledge. (True, the disjunctivist is still committed to a major rational difference between veridical perception and hallucination. But it is much less plausible anyway that we can find a minimal one-time difference between veridical perception and hallucination, since much more has to go wrong for a subject to hallucinate.)Such a view would have to be elaborated in detail to judge its merits. Yet its in-principle availability suffices to show that disjunctivism is not automatically committed to implausible rational differences between minimally different good and bad cases. This, in turn, undermines one main motivation for adopting Schroeder’s apparent factive attitudes view. I suspect that his use of the propositional idiom partly explains why he fails to address the VI v H option that is available to his opponents.This criticism notwithstanding, Reasons First is a rich and rewarding read that paints a sweeping and convincing picture of epistemology as part of a broader normative landscape grounded in reasons. Especially intriguing is Schroeder’s account of how the balance of epistemic reasons, across a range of prima facie problematic cases, determines whether a belief is justified, and of how this balance of reasons can further be used to analyze the concept of knowledge. I expect Reasons First to greatly advance the debate about epistemic reasons in the coming years.I thank Giulia Martina and Simon Wimmer for helpful comments on an earlier version of this review.

中文翻译：

---

首先是原因

马克·施罗德（Mark Schroeder）的最新著作更深入地探讨了规范性和原因的主题，同时将他的关注点从伦理学转向了认识论。他的中心目标首先是论证规范认识论的理论化受益于与其他规范领域的比较（他的“核心假设”[9]）；其次，捍卫将理由放在首位的认知规范性图景：它们可以用来解释和分析所有其他认知规范现象。本书的第一部分对作为竞争者的规范理由进行了令人信服的说明（它们在确定、竞争方面进行竞争）例如，一个人应该做什么或相信什么），这些都是以行动为导向的，而不是以结果为导向的，并且可以采取行动（我对 φ 的理由可以是我 φ 的理由）。施罗德假设既有影响信念正确性的客观原因，也有决定其合理性或正当性的主观原因。第二部分旨在解决理性第一认识论的不合理信念问题。根据这种观点，正当性/理性和知识等规范地位在认知原因中触底。然而，似乎只有合理的信念或知识才能为主体S提供理由，因此我们不能将理由视为根本。因此，显然，感知经验——鉴于它本身既不是知识，也不是合理的——不能为我们提供理由或证据。但这不可能是正确的，因为感知经验无疑是有关我们周围环境的证据的重要来源。根据施罗德的观点，为了实现知觉辩护，我们需要一种涉及世界的知觉证据概念，正如析取主义所认可的那样，它需要证据来蕴含关于外部世界的真理。同时，与析取主义相反，我们必须将此类证据视为非事实的——它不必是真实的（或存在于与真理的关系中），因此不仅在真实感知的良好情况下可用，而且在真实感知的情况下也可用。也处于幻觉或幻觉中。S 的信念是否理性不能取决于最小的差异，正如析取主义所暗示的那样。施罗德用一对相同的案例 C1 和 C2 来说明这一点，除了在 C1 中，S 经历了真实的感知，而在 C2 中，她经历了无法区分的幻觉。（比如说，在 C1 中，S 正在看着一个红球，但在 C2 中，她面对的是一个因红色照明而呈现红色的白球。）但是，重要的是，这种错觉是一次性发生的 - S 有一个其他的情况C1 和 C2 中完美的感知记录。在这两种情况下，S 的信念同样是理性的，至少施罗德是这么认为的。施罗德因此拒绝析取主义。相反，他赞同明显的事实态度观点：基本的感知原因是非事实的主观原因，例如我看到球是红色的命题。但由于它们涉及世俗事实（例如：球是红色的），因此它们仍然涉及世界。为了让我拥有理由，在第三部分中，施罗德解决了充分性问题——担心认知原因的平衡本身无法正确确定一个信念是否合理或知识是否合理。因为尚不清楚 S 的 p 证据必须在多大程度上胜过她的非 p 证据，才能使她对 p 的信念成为合理的或知识。解决这个问题似乎迫使我们放弃理性第一认识论，因为它预设了理性的概念：需要的是足够的证据来使信念变得理性。密切联系的相关问题是，证明 p 至少与证明非 p 的证据一样好，并不总是使相信 p 合理化；在赞成和反对证据之间存在（接近）联系的情况下，不相信是理性的。相比之下，在实际情况中，选择 Φ 的理由至少与不选择 Φ 的理由一样强，确保 Φ-ing 是合理的。如何解释这种差异？施罗德讨论的另一个难题是由激励哲学家为实用主义侵犯辩护的案例提出的。例如，在贾森·斯坦利（Jason Stanley，2005）著名的银行案例中，在高风险场景中，与相应的低风险场景相比，相信者似乎需要更多的证据才能形成合理的信念或了解情况。施罗德使用了这些和其他证据。令人困惑的是，必须有非证据的认知理由来反对相信，进入理由的平衡；然后，总体平衡决定哪种态度在认识论上是合理的，从而保留了理性第一的认识论。反对信仰的非证据理由包括，除其他外，有关未来证据可用性或错误成本的事实，即拥有错误信仰或根据错误信仰采取行动的成本。这些理由是普遍存在的，因此可以解释为什么在密切联系的情况下，理由的整体平衡，除了（几乎）平衡的证据理由之外，还包括反对信念的非证据理由，并没有使p的信念合理化。关于充分性问题：考虑到非证据原因，认识论理由的平衡本身就可以决定信仰的认识论地位。施罗德忠实于他的核心假设，使用与其他态度的理由（例如钦佩）进行比较来论证：一些反对信仰的非证据理由是正确的理由，它们影响信仰的认知地位。正确的理由是信念必须敏感的考虑因素，以便在我们的认知经济中成功发挥其作用——为我们提供默认推理或政策问题上可依赖的考虑因素。与信任相比，信念通过简化推理和决策来实现这一点，因为它们使我们能够忽略远程错误的可能性。根据施罗德的实用知性主义，关于错误的（实际或道德）成本的事实是反对信仰的正确理由，因为它们是对真理信仰必须敏感才能很好地发挥其认知作用的考虑因素。错误的成本解释了为什么其他足够的证据无法证明高风险场景的信念是合理的：它们是如此之高，以至于超过了主体的证据。在第四部分中，施罗德详细介绍了他对知识的康德式解释。他将知识分析为良好的信念，其中信念是对主观和客观原因的反应，这些原因在主观和客观上都足以压倒所有现有的反对信念的理由，因此足以使其变得合理和正确。他通过类比行动的道德价值的正确理由解释来激发这一解释，并通过关注原因通常如何不仅产生某种反应的正确性标准，而且——作为我们可以采取行动的东西——也产生标准现在我对施罗德反对上述分离主义的论点进行批评性评论。他用命题习语提出了感知原因——这个想法是我们可以通过关联命题的方式来归因原因，而不需要致力于任何特定的原因本体论（41）。例如，短语“S看到球是红色的”同样可以用来描述S看到球是红色的心理状态或认为S看到球是红色的，作为她的认知原因。我担心的是，使用命题习语会鼓励对析取主义的误解——也就是说，它致力于在真实知觉与幻觉和幻觉之间划清界限。换句话说，命题习语鼓励这样的想法：认识论析取主义者可用的形而上学析取主义的唯一版本是 V v IH 观点。这种误解强化了施罗德反对析取主义的论点。让我详细说明一下。我们可以将认识论的析取主义与施罗德在书中的析取主义对手等同起来，施罗德赞同知觉提供了事实性的、涉及世界的理由这一观点。相比之下，形而上学的分离主义关注的是感知体验的本质。粗略地说，它指出真实知觉是一种与幻觉本质上不同的心理状态。关于幻觉的进一步知觉状态，它有两种变体：V v IH 和 VI v H (Byrne and Logue 2009: xi)。V v IH 说，作为好情况的真实知觉与错觉或幻觉的坏情况根本不同；VI v H认为，好的情况包括真实的知觉和幻觉，它们与幻觉的坏情况具有相同的性质和根本的不同。如果我们用命题来描述知觉经验提供的理由，那么V v IH的真实性似乎是显而易见的。以我对红球的视觉体验为例。一个相关的命题是球是红色的。我的经验之所以真实，是因为这个命题是真实的，就像在知觉中一样。在错觉和幻觉中，相关的命题结果是错误的——例如，在错觉中，白球看起来是红色的，尽管它不是；在幻觉中，甚至没有视觉上呈现的球。这表明幻觉的形而上学分类，幻觉是最坏的情况。然而，如果我们认为知觉经验本质上是非命题的，而是针对对象的，那么很自然地可以说——用VI v H——知觉和幻觉都成功地认识了我们和物体。因此，我的真实体验和虚幻体验都让我熟悉了面前的球。错觉仅仅涉及物体呈现方式的错误——在这个例子中，白球被误导性地呈现为红色。这种感知体验的形而上学图景可以与认识论主张相结合，即感知提供的原因正是我们所熟悉的对象（Brewer 2018）。对于认识论析取主义者来说，这开启了一种可能性，即主体在虚幻案例和真实案例中的认知理性可能是同等的。重新考虑施罗德的论点，即析取主义令人难以置信地意味着好案例 C1 和坏案例 C2 之间的最小一次性差异使得就主体的合理性而言存在重大差异。根据VI v H，主体经历幻觉而不是难以区分的真实体验（64）的最小差异情况是一个很好的情况。知觉错觉提供了与真实知觉相同的客观原因。在主体的误导性环境中，并考虑到她的客观原因，她形成错误的感性信念是合理的，至少在完全可以理解的意义上是这样；她的信念是合理的，但知识却不是。（诚​​然，析取主义者仍然致力于真实知觉和幻觉之间的重大理性差异。但无论如何，我们能找到真实知觉和幻觉之间最小的一次性差异的可能性要小得多，因为要获得更多的结果，就必须出错。 ）这种观点必须经过详细阐述才能判断其优点。然而，它原则上的可用性足以表明，析取主义不会自动致力于在极小差异的好案例和坏案例之间令人难以置信的理性差异。这反过来又破坏了采用施罗德明显的事实态度观点的一个主要动机。我怀疑他对命题习语的使用部分解释了为什么他未能解决他的对手可以使用的 VI v H 选项。尽管有这样的批评，《理由第一》还是一本内容丰富且有益的读物，它描绘了一幅全面而令人信服的认识论图景：是基于原因的更广泛规范景观的一部分。特别有趣的是施罗德的解释，即在一系列表面上有问题的案例中，认识论理由的平衡如何决定一个信念是否合理，以及如何进一步使用这种理由的平衡来分析知识的概念。我预计 Reasons First 将在未来几年极大地推进有关认知原因的辩论。我感谢 Giulia Martina 和 Simon Wimmer 对本评论早期版本的有益评论。