The Epistemic Role of Consciousness (Smithies 2019) develops a theory of epistemic justification that gives a fundamental epistemic role to phenomenal consciousness. The book works from the bottom up to build this epistemological theory from foundational assumptions in the philosophy of mind. This précis reverses the order of presentation: it works from the top down to put the epistemological motivations for the theory in the foreground.

The book aims to advance the debate between internalism and externalism in epistemology. I generally avoid this terminology, since there are too many ways of drawing the divide, but mine is an internalist theory of epistemic justification in every sense of the term. I argue for phenomenal accessibilism: a version of evidentialism that combines a phenomenal conception of evidence with a higher-order accessibility constraint on the evidential support relation. Here I present the theory as a 12-step program for epistemological addicts.

Step 1 is evidentialism: the thesis that you have epistemic justification to believe whatever your total evidence supports (6.1). This is an extremely flexible framework for building a theory of epistemic justification, since it leaves open the nature of evidence and the evidential support. It is not indispensable to the theory, but it provides the most powerful framework for developing it.

Step 2 is probabilism: the thesis that the evidential support relation is a kind of probability (7.2.1). It is not just a mathematical convenience to model degrees of evidential support in a way that respects formal principles of logic and probability theory. It reflects an epistemological insight that your total evidence supports hypotheses it entails or makes probable. Probabilism raises familiar concerns about epistemic idealization, but there are resources for addressing them (Smithies 2015). This probabilistic version of evidentialism is common ground between many on all sides of the internalism/externalism debate.

Step 3 is a theory of higher-order probabilities (7.1). What are the evidential probabilities of evidential probabilities themselves? This is not guaranteed by the axioms of the probability calculus alone, but I argue on epistemological grounds for an accessibility constraint on higher-order probabilities: necessarily, if it's evidentially probable that p to degree n, then it's evidentially certain that it's evidentially probable that p to degree n. This underpins the JJ Principle: necessarily, you have justification to believe that p if and only if you have justification to believe that you have justification to believe that p.

A second argument is that justification for belief is constrained by the aim of knowledge: you have justification to believe only what you have justification to believe you're in a position to know. A third argument is that justification is stable under justified reflection: you cannot have justification to believe something that you have justification to abandon upon justified reflection (8.1-8.2, 9.5).

Step 5 is to defend accessibilism against objections (5.2, 8.4-8.6, 10.1-10.4, 11.2-11.5). Accessibilism requires that evidence and evidential support relations are self-evident in the sense that they always make themselves evidentially certain (7.2). Defending accessibilism requires answering the objection that you can have misleading higher-order evidence about what your evidence supports and what your evidence is. I respond by arguing that we can defend accessibilism about higher-order probabilities by using the very same resources we need to defend probabilism. The key point is that the evidential support relation is not constrained by the limitations of your doxastic capacity to respond to your evidence.

Step 6 is to motivate this idealized conception of the evidential support relation (10.6, 11.6). Why should we care about an epistemic ideal of respecting the evidence that extends so far beyond our limited doxastic capacities? I argue that this normative ideal is implicit in our ordinary understanding of good reasoning. Enkratic agents who violate logical omniscience are thereby committed to fallacious reasoning. Similarly, enkratic agents who violate evidential omniscience are thereby committed to disrespecting their own evidence. Since ideal rationality requires respecting logic and evidence, while also requiring enkrasia, it thereby requires omniscience about logic and evidence.

Step 7 is to motivate a constraint on evidence (7.2.2). If accessibilism is true, then your evidence must be self-evident in the sense that it always evidentially certain which facts are included in your evidence. What must your evidence be like in order to make itself evident in this way? If your evidence consists of contingent facts, then it's hard to see how your evidence can be self-evident, since contingent facts cannot always make themselves evidentially certain. Or can they?

Step 8 makes an exception for introspection (5.1). A simple theory of introspection says that some facts about your mental states are self-evident in the sense that the fact that you're in some mental state M makes it evidentially certain that you're in M. Once again, the argument appeals to epistemic akrasia (5.3). If M makes it probable that p, but it's uncertain whether you're in M, then you might have justification for believing that p while disbelieving or doubting that you have justification for believing that p. You cannot have justification for epistemic akrasia. So, if M makes it probable that p, then it cannot be uncertain whether you're in M.

Step 9 is to answer a question about the scope of introspection (5.4-5.5). Which facts about your mental states are self-evident? There is no good reason to suppose that all mental states are self-evident, since the argument for the simple theory applies only to mental states that provide evidence. Indeed, the universal claim is manifestly false, since subdoxastic mental representations don't make themselves evident. I argue that your current mental states are self-evident just when they are individuated either by their phenomenal character or by their phenomenal dispositions. These phenomenally individuated mental states are good candidates for self-evidence.

Step 10 gives a top-down argument for the phenomenal conception of evidence, which says that your current evidence is exhausted by the phenomenally individuated facts about your current mental states (6.4). The phenomenal conception of evidence is motivated by an accessibility constraint, which says your evidence is always self-evident, together with the claim that only phenomenally individuated facts about your current mental states are self-evident (7.2).

Step 11 is to give a bottom-up argument for this phenomenal conception of evidence by appealing to intuitive counterexamples to reliabilism, including clairvoyance and the new evil demon (3.1-3.4, 6.3). I try to improve these standard arguments and respond to objections, but I also give them a distinctive spin by exploring parallels with blindsight to motivate the epistemic significance of phenomenal consciousness.

Step 12 is to show how these intuitions about cases can be explained and vindicated by locating them within a more general theory of epistemic justification (7.3-7.4). They are not just gut reactions. A purely intuition-driven approach to epistemology is dialectically ineffective and theoretically unsatisfying. Phenomenal accessibilism is explicitly designed to explain why these intuitions are well motivated and to provide them with theoretical support.

Internalist theories of epistemic justification tend to be strong on intuitive conviction but weak on theoretical principle. One exception is phenomenal conservatism, which promises to explain all epistemic justification in terms of a single epistemic principle that you should always believe what seems on balance to be true (12.1-12.2). But while this theory gives a plausible epistemology of perception, it severely distorts the epistemology of memory, introspection, inference, and higher-order evidence (12.3-12.4).

My hope is that phenomenal accessibilism will be taken seriously as an alternative to phenomenal conservatism by anyone who is interested in assessing the prospects for an internalist theory of epistemic justification that is not only intuitively compelling, but also theoretically principled.

意识的认知作用(Smithies 2019) 发展了一种认知论证理论，赋予现象意识以基本的认知作用。这本书自下而上地从心灵哲学的基本假设中构建这个认识论理论。这个概要颠倒了呈现的顺序：它从上到下工作，将理论的认识论动机置于前台。

本书旨在推进认识论中内在主义和外在主义之间的争论。我通常避免使用这个术语，因为划分界限的方法太多了，但我的是一个内在主义的认知论证理论，从这个术语的每一个意义上讲。我主张现象可访问性：一种证据主义版本，它将证据的现象概念与对证据支持关系的高阶可访问性约束相结合。在这里，我将理论作为认识论成瘾者的 12 步程序介绍。

第 1 步是证据主义：你有认知理由相信你的全部证据支持的任何东西的论点（6.1）。这是用于构建认知论证理论的极其灵活的框架，因为它保留了证据的性质和证据支持。它对于理论来说并非不可或缺，但它为发展理论提供了最强大的框架。

第二步是概率论：证据支持关系是一种概率的命题（7.2.1）。以尊重逻辑和概率论的形式原则的方式对证据支持程度进行建模不仅仅是数学上的便利。它反映了一种认识论的洞察力，即您的全部证据支持它所包含或可能的假设。概率论引发了对认知理想化的常见担忧，但有解决这些问题的资源（Smithies 2015）。这种证据主义的概率版本是内在主义/外在主义辩论各方之间的共同点。

步骤 3 是高阶概率理论(7.1)。证据概率本身的证据概率是多少？这并不能仅由概率微积分的公理来保证，但我在认识论的基础上争论对高阶概率的可访问性约束：必然地，如果 p 显然有可能达到 n 度，那么显然可以肯定的是，这显然有可能p到n级。这支持了 JJ 原则：必然地，当且仅当您有理由相信您有理由相信p时，您才有理由相信p。

第二个论点是，信念的辩护受到知识目标的限制：你有理由只相信你有理由相信你有能力知道的东西。第三个论点是，在合理的反思下，辩护是稳定的：你不能有理由相信你有理由在合理的反思中放弃的东西（8.1-8.2，9.5）。

第 5 步是针对异议捍卫无障碍主义（5.2、8.4-8.6、10.1-10.4、11.2-11.5）。无障碍主义要求证据和证据支持关系是不言而喻的，因为它们总是使自己在证据上是确定的（7.2）。捍卫可访问性需要回答这样的反对意见，即您可能拥有关于您的证据支持什么以及您的证据是什么的误导性高阶证据。我的回应是，我们可以通过使用捍卫概率论所需的相同资源来捍卫关于高阶概率的可访问性。关键是证据支持关系不受你对证据做出回应的信念能力的限制。

第 6 步是激发这种证据支持关系的理想化概念 (10.6, 11.6)。为什么我们应该关心尊重证据的认知理想，这些证据远远超出了我们有限的信念能力？我认为，这种规范理想隐含在我们对良好推理的普通理解中。违反逻辑全知的 Enkratic 代理人因此致力于谬误推理。同样，违反证据无所不知的狂热代理人因此致力于不尊重他们自己的证据。由于理想理性需要尊重逻辑和证据，同时也需要enkrasia，因此它需要对逻辑和证据的无所不知。

第 7 步是激发对证据的约束(7.2.2)。如果无障碍主义是正确的，那么你的证据必须是不言而喻的，因为它总是在证据上确定你的证据中包含哪些事实。你的证据必须是什么样的才能以这种方式证明自己？如果你的证据由偶然的事实组成，那么很难看出你的证据如何不证自明，因为偶然的事实并不总是使它们自己在证据上是确定的。或者他们可以吗？

第 8 步为内省(5.1)设置了一个例外。一个简单的内省理论说，关于你的精神状态的一些事实是不言而喻的，因为你处于某种精神状态 M 的事实使你在证据上确定你处于 M 状态。再一次，这个论点吸引了认知性失动作症 (5.3)。如果 M 使p成为可能，但不确定你是否在 M 中，那么你可能有理由相信p而不相信或怀疑你有理由相信p。你不能为认知失能症辩护。所以，如果 M 使p成为可能，那么就不能确定你是否在 M 中。

第 9 步是回答有关内省范围的问题 (5.4-5.5)。关于你的精神状态的哪些事实是不言而喻的？没有充分的理由假设所有心理状态都是不证自明的，因为简单理论的论证只适用于提供证据的心理状态。事实上，普遍的主张显然是错误的，因为亚信念的心理表征并不明显。我认为，当你当前的精神状态被它们的现象特征或它们的现象倾向所个体化时，它们就是不言而喻的。这些显着个体化的精神状态是自我证明的良好候选者。

第 10 步为证据的现象概念提供了自上而下的论证，它说你当前的证据已经被关于你当前精神状态的显着个体化的事实所耗尽 (6.4)。证据的现象概念是由可访问性约束激发的，它说你的证据总是不言而喻的，连同只有关于你当前精神状态的现象个体化的事实是不言而喻的主张 (7.2)。

第 11 步是通过诉诸可靠主义的直觉反例，包括千里眼和新的邪恶恶魔 (3.1-3.4, 6.3)，为这种现象性的证据概念提供自下而上的论证。我试图改进这些标准论点并回应反对意见，但我也通过探索与盲视的相似之处来激发现象意识的认知意义，从而赋予它们独特的旋转。

第 12 步是展示如何通过将这些关于案例的直觉定位在更一般的认知论证理论（7.3-7.4）中来解释和证明这些直觉是正确的。它们不仅仅是直觉反应。一种纯粹的直觉驱动的认识论方法在辩证上是无效的，在理论上也不能令人满意。现象可及性明确旨在解释为什么这些直觉有很好的动机，并为它们提供理论支持。

内在主义的认知论证理论往往在直觉信念上很强，但在理论原则上很弱。一个例外是现象保守主义，它承诺根据一个单一的认知原则来解释所有的认知论证，即你应该始终相信总的来说似乎是真的（12.1-12.2）。但是，虽然这个理论给出了一种似是而非的感知认识论，但它严重扭曲了记忆、内省、推理和高阶证据的认识论（12.3-12.4）。

我希望任何有兴趣评估不仅在直觉上令人信服而且在理论上有原则的认知辩护的内在主义理论的前景的人，都会认真对待现象可及性主义作为现象保守主义的替代方案。