This excellent volume contains 14 chapters exploring the idea of fragmentation: the division of a belief state into parts (“fragments”) that can represent the world in distinct, jointly incoherent ways. For instance, I might know that sea cucumbers are a type of animal related to starfish when I am asked in a biological context, but when I am at a restaurant and see them on the menu, I think that sea cucumbers are a vegetable. I have two ways of thinking of sea cucumbers, two fragments that are both sets of beliefs about the world but are in some sense separate from each other. Most of the contributions concentrate on whether fragmentation is a good model of belief and how a fragmented state of mind can be rationally evaluated, though the final section contains a paper by Gertler (chap. 13) applying a case of fragmentation and subsequent belief change to a question about agency.While fragmentation theories share the commitment to multiple (potentially) incoherent belief states, this volume reveals a deep divide between two families of views. The first, dispositionalism, holds that what it is to have a fragment is just to be disposed to exhibit a pattern of actions1 that is best explained by more than one set of beliefs (given one’s background beliefs and desires). On this view, fragments are by definition coherent within themselves, and also by definition at odds with one another. For Elga and Rayo (chap. 1) and Greco (chap. 2), at odds means picking out a different set of possible worlds whereas for Yalcin (chap. 6), it means partitioning possible space differently. Representationalists, on the other hand, such as Bendaña and Mandelbaum (chap. 3) or Murez (chap. 7) hold that a fragment is a psychologically real entity. This means that in principle we can ask whether fragments are internally coherent without triviality.This distinction between representationalism and dispositionalism is not just important to broader questions about belief, but directly bears on fragmentation. This is most clear when we consider that many of the questions raised in one of the two frameworks in this volume are not even able to be formulated on the alternative framework. I’ll give two examples.Bendaña and Mandelbaum ask: when do new fragments arise? Their answer is the “Environmental Principle”: new environments open up new fragments, so that if in my Portuguese class, I saw a chart of types of pastry, I might encode these separately (and potentially incoherently) from my stored knowledge of pastries acquired in other contexts. But notice that on the dispositionalist view, fragments arise if and only if one’s behavior is at odds with other parts of behavior, given background beliefs and desires. We can of course ask when behavior comes to be at odds in this way, and perhaps the answer might appeal to contexts. But this is not the same question at all, since Bendaña and Mandelbaum treat the creation of a new fragment as a mental event happening at a particular time. The question of when this event occurs is one that has ordinary efficient causes, such as the prior mental awareness of a new environment. On the contrary, the dispositionalist can only ascribe fragments to swathes of behavior, so there may be no clear beginning to a fragment. Likewise, the question of new fragments leaves open the possibility that, at least in principle, beliefs might be incoherent without acquiring a new fragment and, conversely, that new fragments can arise even when nothing contradictory is represented. That is, it can be true that I have a new fragment relative to Portuguese class without anything being contradictory. But of course, for the dispositionalist, this is conceptually impossible. Thus the question raised (and answered) by Bendaña and Mandelbaum means something quite different to the dispositionalist. Or rather, the real question they are asking, what psychological process makes me break off my representations into new fragments, cannot be asked at all—instead, new fragments are born (and expire) as a matter of conceptual necessity.From the dispositionalist side, Egan asks: “What constraints shall we impose on the proliferation of belief states in order to avoid collapse into triviality and maintain the status of doxastic states as explanatory of behavior rather than as a mere summary or redescription of it?” (123). This question, on the other hand, is ill posed for the representationalist. For her, there is a matter of fact about how many fragments you have and one that is not decided from the point of view of the external modeler. And more importantly, there is no problem of triviality—a person has the number of fragments that she has, and whatever number that is cannot be too many since it reflects her representational capacity.These two questions are instances of a deeper issue dividing this volume. In a sense, all of the fragmentation debate is motivated by incoherence. But there are two very different projects around understanding how we are less than fully logically consistent. The first is the project of understanding how, when, and why we strive to be coherent. (By “striving” I do not necessarily mean first-personal effort but rather the deployment of psychological processing). On the other hand, we might also ask whether and how it is conceptually possible to be incoherent. These questions are not just different but, I would suggest, require different standpoints: the former a view of incoherence between real representations, and the latter a view of incoherence as a function of a whole person. This divide is not a problem with the volume, but it does lend it a somewhat disjointed feeling as authors on either side use similar language and examples to speak at distinct levels.The middle section of the book is devoted to connecting work on fragmentation to the topic of mental files. A mental file is a way of understanding what is going on in Frege cases where, for instance, I inadvertently think of the very same person under two different descriptions (Havel the politician/Havel the playwright). There is obviously a connection between these and fragmentation cases—in fact, it is sometimes hard to distinguish between them, though a difference is that fragmentation emphasizes a difference in propositions that are represented between fragments, whereas mental files emphasize a difference in the mode of presentation of a referent. This connection, while intuitive, has not received much attention, and this kind of work, connecting similar topics across different literatures, is often neglected. Recananti (chap. 9) in particular puts forward an illuminating connection between the two.One doubt I began to have about fragmentation in general started with a point made by Egan. He describes a case where someone can answer the question “Was the youngest von Trapp child’s name ‘Gretl’?” but not “What was the name of the youngest von Trapp child?” (111). After this, he observes, “Here is a puzzling feature of this sort of case: the information required to answer the two questions is exactly the same” (112). This seems false. I need more information to answer the open question about the name. The information cannot be the same since I could much more easily guess the right answer to the yes/no question. Likewise, I could answer the yes/no question by lacking knowledge of the name and just knowing facts such as “the name started with an A.” The question that includes the name contains more information, and I need less information to answer it.Egan is wrong about this case in an interesting way: he, like many of the other authors, seeks to explain the phenomenon of seemingly incoherent patterns of action and behavior through ascribing a more complex mental state to the agent than standard, unified belief. But this excludes, or rather internalizes, the role of the environment. When someone asks you “was the child named Gretl?” she is giving you information that is withheld in the other way of asking the question. But what becomes of this difference on the fragmentation picture?Elga and Rayo get nearest to this question. On their view, belief states are not relations between persons and propositions, but among persons, elicitation conditions, and propositions. Thus, relative to the elicitation condition of being asked this or that question, you are related to different sets of propositions. This does acknowledge the role of the environment, insofar as the elicitation conditions are environmental, and it is not committed to the idea that your beliefs are independent of the environment. But at the same time, the role of the environment is minimized in two ways. First, by including possible environmental conditions in the belief state, Elga and Rayo treat the variability between contexts as a feature of you rather than one shared by anyone in your circumstances. Second, by compiling all the elicitation conditions in a list, they elide differences in information between them: a person who has an easier time answering more informative questions like the yes/no question and a person who has an easier time answering when she is spinning in a circle are modeled as sharing the same kind of relativity to circumstance, whereas intuitively, a question that conveys information has a different kind of influence on belief than a physical motion that merely disposes you to some frame of mind.More generally, I worry that the category of fragmentation cases is heterogeneous between informationally relevant conditions and mere nonrational shifts, and treating them alike would miss that we expect many agents to respond to the von Trapp questions differently and indeed find such a pattern to reflect a rational uptake of implicitly conveyed information. Our relationship to our environments is an important part of the story of how we come to be incoherent, when incoherence is maintained, and how it is even possible, and so a major missing piece from the important work in this book is a focus beyond the agent’s mind.Finally, a striking feature of this book is its contemporary framing. Fragmentation, as introduced in this volume, is an idea originating from Cherniak, Lewis, and Stalnaker. In fact, almost every chapter in this book discusses a case from Lewis revolving around a mental map of Nassau Street. This unity has upsides, in allowing the book to be tightly focused and well-integrated compared to similar volumes. But it did leave me with questions about the broader history of this debate.Among surely many other places, historical antecedents can be found in the many descriptions of internal conflict in the Confessions (Augustine, 2008). In book 7, Augustine begins by describing his state of mind as he in some sense comes to believe that God is immaterial, but at the same time is unable to shake a way of thinking of God as occupying space: My heart vehemently protested against all the physical images in my mind, and by this single blow I attempted to expel from my mind’s eye the swarm of unpurified notions flying about there. Hardly had they been dispersed when in the flash of an eye (i Cor. 15:52) they had regrouped and were back again. They attacked my power of vision and clouded it. Although you were not in the shape of the human body, I nevertheless felt forced to imagine something physical occupying space diffused either in the world or even through infinite space outside the world. Admittedly I thought of this as incorruptible and inviolable and unchangeable, which I set above what is corruptible, violable, and changeable. (111)A few features of Augustine’s presentation of division might shed light on the issue of fragmentation. First, unlike the paradigm cases of fragmentation from this volume, Augustine’s divided state persists despite him noticing it. He may not have had the full understanding of the contradiction he notes in the light of hindsight, but the description implies an effort to dispel the material conception that must reflect some degree of awareness. This seems intuitively possible, though it might already strain some of the views discussed above on which we only reason from one fragment at a time (e.g., Yalcin’s). Second, the conflict is not merely between propositions, or even between fine- and coarse-grained propositions, but between visual and abstract modes of thinking. We could even imagine these differences explain the first feature, why the division can persist despite being noticed.Most interestingly, an affective element pervades Augustine’s description: the relationship between the two states of mind is described as a war. I wonder whether the volume might have benefitted from more engagement with history, especially when it comes to exploring the connection between incoherence and affect that Loeb (1998) traces through the Stoics, Hume, and Peirce. While in Augustine’s case, the feeling of strife and division might draw on awareness of the conflict, there are surely ways to think of the affective side of fragmentation even in cases of partial or minimal awareness. This dimension is significant because it relates to questions of how and when fragments are combined, to the particular recalcitrance of core beliefs as noted by Bendaña and Mandelbaum, and even to the ethically and politically charged issues raised in the section on implicit beliefs. Many of the contributors appeal to the idea that beliefs are a set of maps that we use to steer, but the link with affect suggests that divided states of mind have a motivational force of their own—in the sense that we’re bothered or made uneasy by the division, or even in the sense of each belief substate pushing us and pushing against the others.Overall, this book is a tightly connected collection of papers on a topic at the heart of the intersection between metaphysics of mind and epistemology (with many more compelling contributions than I have room to discuss here). It will be a core resource to anyone interested in diving into this debate.

中文翻译：

---

支离破碎的心灵

这本优秀的著作包含 14 章，探讨了碎片化的概念：将信念状态划分为多个部分（“碎片”），这些部分可以以不同的、共同不连贯的方式代表世界。例如，当我被问到生物学背景时，我可能知道海参是一种与海星有关的动物，但当我在餐厅看到菜单上的海参时，我认为海参是一种蔬菜。我对海参有两种思考方式，这两个片段都是关于世界的两套信念，但在某种意义上是彼此分离的。大多数贡献都集中在碎片化是否是一个好的信念模型以及如何合理地评估碎片化的心态，尽管最后一节包含格特勒的一篇论文（第 1 章）。13) 将分裂和随后的信念改变案例应用于关于代理的问题。虽然分裂理论共同致力于多个（潜在的）不连贯的信念状态，但本卷揭示了两个观点之间的深刻分歧。第一种是倾向主义，认为拥有一个片段只是倾向于展示一种行为模式 1，这种模式最好用不止一组信念（给定一个人的背景信念和愿望）来解释。按照这种观点，片断在定义上是连贯的，并且在定义上也是相互矛盾的。对于 Elga 和 Rayo（第 1 章）和 Greco（第 2 章）而言，不一致意味着选择一组不同的可能世界，而对于 Yalcin（第 6 章）而言，这意味着以不同方式划分可能的空间。另一方面，表象主义者，Bendaña 和 Mandelbaum（第 3 章）或 Murez（第 7 章）等人认为片段是心理上真实的实体。这意味着原则上我们可以问碎片是否在内部连贯而不琐碎。表征主义和倾向主义之间的这种区别不仅对更广泛的信仰问题很重要，而且直接与碎片化有关。当我们考虑到在本卷的两个框架之一中提出的许多问题甚至无法在替代框架上提出时，这一点就最为清楚了。我举两个例子。Bendaña 和 Mandelbaum 问：新的片段什么时候出现？他们的答案是“环境原则”：新的环境开辟了新的片段，所以如果在我的葡萄牙语课上，我看到了一张糕点类型的图表，我可能会根据我在其他情况下获得的糕点知识对这些进行单独编码（并且可能不连贯）。但请注意，根据倾向主义者的观点，当且仅当一个人的行为与行为的其他部分不一致时，才会出现碎片，给定背景信念和愿望。我们当然可以询问行为何时以这种方式出现矛盾，也许答案可能会吸引上下文。但这根本不是同一个问题，因为本达尼亚和曼德尔鲍姆将新片段的创建视为在特定时间发生的心理事件。这个事件何时发生的问题是一个具有普通有效原因的问题，例如对新环境的先前心理意识。相反，倾向主义者只能将片段归因于行为的片段，因此片段可能没有明确的开始。同样，新片段的问题留下了这样一种可能性，即至少在原则上，如果没有获得新的片段，信念可能会不连贯，相反，即使没有任何矛盾的表现，新的片段也会出现。也就是说，我确实有一个与葡萄牙语类相关的新片段，没有任何矛盾。但是，当然，对于倾向主义者来说，这在概念上是不可能的。因此，Bendaña 和 Mandelbaum 提出（并回答）的问题与倾向主义者的意思完全不同。或者更确切地说，他们问的真正问题，是什么心理过程让我把我的表征分解成新的片段，根本不能问——相反，新片段的诞生（和过期）是概念上的必然性问题。从倾向主义者的角度来看, 伊根问：“我们应该对信念状态的扩散施加什么限制，以避免陷入琐碎，并保持信念状态作为行为解释的地位，而不是仅仅作为行为的总结或重新描述？” (123)。另一方面，这个问题不适用于表征主义者。对她来说，从外部建模者的角度来看，你有多少碎片和一个没有决定的碎片是一个事实。更重要的是，不存在琐碎的问题——一个人有多少个片段，多少个都不能太多，因为它反映了她的表现能力。这两个问题是本书划分的更深层次问题的实例. 从某种意义上说，所有关于碎片化的争论都是由不连贯引起的。但是有两个非常不同的项目围绕着理解我们如何在逻辑上不完全一致。第一个是理解我们如何、何时以及为什么努力保持连贯性的项目。（我所说的“努力”不一定是指第一人称的努力，而是指心理处理的部署）。另一方面，我们也可能会问，在概念上不连贯是否可能以及如何可能。这些问题不仅不同，而且我建议，需要不同的立场：前者是真实表征之间不连贯的观点，后者是将不连贯视为一个完整的人的功能的观点。这种鸿沟不是本书的问题，但它确实给人一种有些脱节的感觉，因为双方的作者都使用相似的语言和例子来表达不同的层次。本书的中间部分致力于将碎片化工作与心理文件主题联系起来。心理档案是理解弗雷格案例中正在发生的事情的一种方式，例如，我无意中想到了两种不同描述下的同一个人（政治家哈维尔/剧作家哈维尔）。这些与碎片案例之间显然存在联系——事实上，有时很难区分它们，尽管不同之处在于碎片强调碎片之间表示的命题的差异，而心理文件强调的是表达方式的差异。引用的呈现。这种联系虽然直观，但并未受到太多关注，而且这种将不同文献中的相似主题联系起来的工作常常被忽视。Recananti（章节。9) 特别提出了两者之间的启发性联系。我开始对一般碎片化产生怀疑是从 Egan 提出的一个观点开始的。他描述了一个案例，有人可以回答这个问题“最小的 von Trapp 孩子的名字是‘Gretl’吗？” 但不是“冯特拉普最小的孩子叫什么名字？” （111）。此后，他观察到，“这种情况有一个令人费解的特点：回答这两个问题所需的信息完全相同”（112）。这似乎是错误的。我需要更多信息来回答有关名称的开放性问题。信息不可能相同，因为我可以更容易地猜出是/否问题的正确答案。同样，我可以通过缺乏对名称的了解而只知道诸如“名称以 A 开头”之类的事实来回答是/否问题。” 包含名称的问题包含更多信息，我需要更少的信息来回答它。伊根以一种有趣的方式错误地解释了这个案例：他和许多其他作者一样，试图解释看似不连贯的模式现象通过将比标准、统一的信念更复杂的心理状态归因于代理人的行动和行为。但这排除了，或者更确切地说，内化了环境的作用。当有人问你“这个孩子叫 Gretl 吗？” 她正在向您提供以另一种提问方式隐瞒的信息。但是碎片图上的这种差异会变成什么？Elga 和 Rayo 最接近这个问题。在他们看来，信念状态不是人和命题之间的关系，而是人、启发条件和命题之间的关系。因此，相对于被问到这个或那个问题的启发条件，你与不同的命题集有关。这确实承认了环境的作用，只要启发条件是环境性的，并且它不致力于您的信念独立于环境的想法。但与此同时，环境的作用在两个方面被最小化了。首先，通过在信念状态中包含可能的环境条件，Elga 和 Rayo 将情境之间的可变性视为你的一个特征，而不是你所处环境中任何人共有的特征。其次，通过将所有的启发条件编译成一个列表，他们消除了它们之间的信息差异：我们与环境的关系是我们如何变得不连贯、何时保持不连贯以及它如何成为可能的故事的重要组成部分，因此本书重要工作中的一个主要缺失部分是超越代理人的想法。最后，这本书的一个显着特点是它的现代框架。本卷中介绍的碎片化是 Cherniak、Lewis 和 Stalnaker 提出的一个想法。事实上，本书几乎每一章都围绕拿骚街的心理地图讨论了刘易斯的一个案例。这种统一有好处，因为与类似的卷相比，这本书可以更加集中和良好地整合。但这确实让我对这场辩论的更广泛历史产生了疑问。在许多其他地方，历史前因可以在《忏悔录》中对内部冲突的许多描述中找到（奥古斯丁，2008 年）。在第 7 本书中，奥古斯丁首先描述了他在某种意义上开始相信上帝是非物质的，但同时又无法动摇一种认为上帝占据空间的思维方式：我的心强烈抗议所有我脑海中的物理图像，并通过这一击，我试图从我的心灵之眼中驱逐出在那儿飞来飞去的未净化的观念。他们刚一散去，就在眨眼之间（哥林多前书 15:52），他们重新集结并又回来了。他们攻击了我的视力并使它蒙上了阴影。虽然你不是人形，尽管如此，我还是不得不想象一些物理占据空间的东西散布在世界上，甚至散布在世界之外的无限空间中。诚然，我认为这是不朽的、不可侵犯的、不可改变的，我把它置于可腐败、可侵犯和可改变的东西之上。(111) 奥古斯丁对分裂的描述中的一些特征可能会阐明碎片化问题。首先，与本卷中的分裂范例不同，奥古斯丁的分裂状态持续存在，尽管他注意到了。事后看来，他可能没有完全理解他所指出的矛盾，但这种描述暗示了一种努力，要消除必须反映某种程度意识的物质概念。这在直觉上似乎是可能的，尽管它可能已经使上面讨论的一些观点变得紧张，我们一次只能从一个片段中推理出这些观点（例如，Yalcin 的）。其次，冲突不仅仅是命题之间，甚至是细粒度命题和粗粒度命题之间的冲突，而是视觉思维模式和抽象思维模式之间的冲突。我们甚至可以想象这些差异解释了第一个特征，即为什么这种分裂会在被注意到的情况下持续存在。最有趣的是，奥古斯丁的描述中充斥着一种情感因素：两种心理状态之间的关系被描述为一场战争。我想知道这本书是否可以从更多地参与历史中获益，尤其是在探索不连贯与影响之间的联系时，勒布 (1998) 通过斯多葛学派、休谟和皮尔士追溯。在奥古斯丁的例子中，冲突和分裂的感觉可能会引起对冲突的意识，即使在部分或最低意识的情况下，也肯定有办法考虑分裂的情感方面。这个维度很重要，因为它涉及片段如何以及何时组合的问题，涉及本达尼亚和曼德尔鲍姆指出的核心信念的特别顽固，甚至涉及隐含信念部分提出的伦理和政治问题。许多贡献者诉诸这样的观点，即信念是我们用来操纵的一组地图，但与情感的联系表明，分裂的心态有其自身的动力——在我们被打扰或被制造的意义上对分裂感到不安，或者甚至在每个信念亚状态推动我们和反对其他人的意义上感到不安。总的来说，这本书是关于心灵形而上学和认识论交叉核心主题的紧密联系的论文集（比我在这里讨论的空间有更多引人注目的贡献）。对于任何有兴趣参与这场辩论的人来说，这将是一个核心资源。