1 INTRODUCTION

Most of us would sincerely and justifiably express views that portray ourselves as unprejudiced agents, consciously committed to egalitarianism in all its forms. Yet, we are often surprised to discover that we behave in ways that betray such egalitarian, self-reported beliefs. We may thus very well find that it takes us longer to e.g., categorize women as surgeons or Mediterranean people as hard-working, even if we disavow sexism and racism. The mismatch between our explicit egalitarian beliefs and the stereotypically biased, often discriminatory, prejudicial behavior we often display makes it plausible to think that there must be mental states, other than our explicit beliefs, which are at least partially responsible for such a behavior. The expressions “implicit biases,” “implicit attitudes,” “implicit prejudices” or simply “prejudices,” here taken as synonyms, are used to refer to such posits. Implicit biases are automatically activated by certain categorical cues in relevant contexts and permeate our perception, actions, and decisions. They are seldom the object of awareness,¹ which makes them especially resilient to change—resilient, but not unchangeable.

Some authors characterize implicit biases as traits, i.e., broad-track dispositions that manifest themselves not just in behavior, but also in e.g., attention patterns and emotions. Broad-track dispositions are thus like character traits (Machery, 2016). A construal of implicit biases in terms of dispositions is also at the heart of e.g., Céline Leboeuf's (2020) account, who regards them as perceptual habits (see below). More often, however, implicit biases are taken to be occurrent mental states with a content that reflects stereotypical social features and is evaluatively charged. The central cases in the literature are those in which the content of such mental states is at odds with our explicit (self-reported) attitudes. It is, in fact, the low correlation between implicit and self-reported attitudes that has driven most of the research on implicit biases in social psychology and the reason I take the implicitly biased egalitarian as the central case in this paper.²

This lack of consensus about the ontology of implicit biases extends to the question of their representational structure when they are considered to be occurrent mental states. On the one hand, according to what is sometimes called “the associationist view,” implicit biases are sui generis states, which, unlike explicit attitudes, have an associative structure (see e.g., Brownstein & Madva, 2012; Gendler, 2008a, 2008b; Madva, 2016; Toribio, 2018). On the other hand, on the so-call “propositionalist view,” implicit biases are propositional states, i.e., either beliefs (De Houwer, 2014; Egan, 2011; Hughes & Barnes-Holmes, 2011; Mandelbaum, 2013, 2016; Mitchell & De Houwer, 2009) or states that fall short of being beliefs, but are, nevertheless, propositionally structured (Levy, 2015). There are also hybrid models such as Sullivan-Bissett's (2019) unconscious imagination model and models that fall outside the propositional/associationist dichotomy, such as Nanay's (2021) mental imagery account.

Regardless of their ontology, implicit biases are taken to have an affective component, with some associationist accounts such as Gendler's (2008a, 2008b) or Madva and Brownstein (2018) taking the affective element to be an essential part of their content. And, again, no matter how they are conceptualized, there is no question that implicit biases trigger an (often) discriminatory behavior toward members of the applicable social category—a discriminatory behavior that is directed toward these members just in virtue of their membership to that category. The focus of this paper is the nature of this discriminatory behavior. In particular, I aim to defend the view that the discriminatory behavior triggered by implicit biases is best understood as a type of habitual action—as a harmful, yet deeply entrenched, passively acquired, and socially relevant type of habit.³

As I have mentioned, the appeal to habits in the explanation of both implicit and explicit biases is not new. Helen Ngo (2016), following Merleau-Ponty's Phenomenology of Perception, develops an account of racism as “deeply embedded in our bodily habits of movement, gesture, perception, and orientation” (Ngo, 2016, p. 847). Also following the steps of Merleau-Ponty (1945/2002), Céline Leboeuf (2020) argues for a view of implicit biases as perceptual, embodied, social habits. Nathifa Green (2020) relies on the notion of habit to give an account of stereotype threat, i.e., the feeling of being at risk of underperforming in tasks which are traditionally associated with stereotypes of the social group to which a person belongs—be it gender, race, sexual orientation, nationality, or profession. Green argues that the anxiety and alienation triggered by stereotype threat is a form habit disruption. Stereotype threat consists in the feeling of being knocked out of a person's skills and habits normal “flow.”⁴

I acknowledge the importance and depth of all these analyses, most of which are located within the phenomenological tradition or within the so-called embodied and embedded approach to cognition. I offer an argument for the thesis that the prejudicial behavior is best conceptualized as a type of passively acquired habit that completements, rather than compete, with similar views in the phenomenological tradition. This central claim is, to be clear, not a thesis about implicit biases themselves. It is a thesis about the behavior triggered by such implicit biases—understood as either mental or physical action. My argument here is an inference to the best explanation, premised on the view of habitual action that psychology and neuroscience have to offer. I elaborate on some of the marks of habitual action and suggest a more precise characterization of the central notions involved in such an account.

If the thesis that I defend here is plausible, and we accept that an important part of our life as egalitarian agents is nevertheless a life dotted with prejudicial habits—automatic actions of which we seem to be unaware—can we hold ourselves responsible for such prejudicial actions? Understanding prejudicial behavior as a passively acquired type of habit offers a diagnosis of the array of intuitions that inform extant approaches to responsibility for this type of actions. It also helps develop an ability-based account of the responsibility we have for prejudicial action that justifies a blaming response to the prejudiced agent while acknowledging that she is not blameworthy for her actions. This ability-based account of responsibility is neutral regarding any of the characterizations of implicit biases portrayed above.

The paper is organized as follows: In Section 2, I outline the main features traditionally associated with the notion of habit and highlight some of the neurobiological and neuropsychological processes that underwrite habitual action so as to pin down the type of habit that I consider to be relevant for characterizing prejudicial behavior. In Section 3, based on the previous considerations, I argue for the claim that prejudicial behavior is a passively acquired type of habitual action. In Section 4, I address the issue of the responsibility we have for our habits. As habits are ultimately susceptible of being controlled, the agent is responsible, and ought to be blamed, I argue, for failing to develop the ability to spot the kind of situations that require the exercise of her intellectual, moral, social, and prudential obligations. Being thus responsible, however, is consistent with the agent's not being blameworthy. For the automaticity of the blamed agent's implicitly biased behavior makes it unintentional relative to intellectual, moral, social, and prudential values that she already cares about.⁵

中文翻译：

---

对隐性偏见行为的责任：基于习惯的方法

1 简介

我们中的大多数人会真诚而合理地表达将自己描绘成不带偏见的代理人的观点，有意识地致力于各种形式的平等主义。然而，我们常常惊讶地发现，我们的行为方式背叛了这种平等主义、自我报告的信念。因此，我们很可能会发现，即使我们否认性别歧视和种族主义，我们也需要更长的时间才能将女性归类为外科医生或地中海人是勤奋的人。我们明确的平等主义信念与我们经常表现出的刻板偏见、通常是歧视性的偏见行为之间的不匹配，使得认为除了我们明确的信念之外必须存在至少部分对这种行为负责的心理状态是合理的。“隐含偏见”、“隐含态度”、“隐含偏见”或简称“偏见”等表达方式 ”这里作为同义词，用来指代这样的立场。隐性偏见由相关上下文中的某些分类线索自动激活，并渗透到我们的感知、行动和决策中。他们很少是意识的对象，¹这使他们对变化特别有弹性——有弹性，但并非一成不变。

一些作者将隐性偏见描述为特征，即不仅表现在行为中，而且表现在注意力模式和情绪中的广泛倾向。因此，广泛的倾向就像性格特征（Machery，2016）。对性格方面的隐性偏见的解释也是例如 Céline Leboeuf ( 2020 ) 的核心) 帐户，谁将它们视为感知习惯（见下文）。然而，更常见的情况是，隐性偏见被认为是发生的心理状态，其内容反映了刻板的社会特征并受到评价。文献中的中心案例是这些精神状态的内容与我们明确的（自我报告的）态度不一致的案例。事实上，正是内隐态度和自我报告态度之间的低相关性推动了大多数关于社会心理学内隐偏见的研究，也是我将内隐偏见的平等主义作为本文中心案例的原因。²

这种对隐性偏见本体论缺乏共识的问题延伸到了当它们被认为是发生的心理状态时的表征结构问题。一方面，根据有时被称为“联想主义观点”的观点，内隐偏见是自成一体的状态，与外显态度不同，隐含偏见具有关联结构（参见例如，Brownstein & Madva, 2012 ; Gendler, 2008a , 2008b ;玛德瓦，2016 年；托里比奥，2018 年）。另一方面，在所谓的“命题主义观点”中，隐含偏见是命题状态，即两种信念（De Houwer, 2014 ; Egan, 2011 ; Hughes & Barnes-Holmes,2011；曼德尔鲍姆，2013 年，2016 年；Mitchell & De Houwer, 2009 ) 或不属于信念但仍具有命题结构的状态 (Levy, 2015 )。还有混合模型，例如 Sullivan-Bissett ( 2019 ) 的无意识想象模型和不属于命题/联想二分法的模型，例如 Nanay ( 2021 ) 的心理意象帐户。

无论它们的本体论如何，隐含偏见都被认为具有情感成分，例如 Gendler's ( 2008a , 2008b ) 或 Madva and Brownstein ( 2018 )) 将情感元素作为其内容的重要组成部分。而且，再一次，无论它们是如何被概念化的，毫无疑问，隐性偏见会引发对适用社会类别成员的（通常）歧视性行为——这种歧视性行为仅仅因为这些成员的成员资格而针对该社会类别的成员。类别。本文的重点是这种歧视行为的性质。特别是，我的目标是捍卫这样一种观点，即由隐性偏见引发的歧视性行为最好被理解为一种习惯性行为——一种有害的、但根深蒂固的、被动获得的、与社会相关的习惯类型。³

正如我所提到的，在解释隐性和显性偏见时诉诸习惯并不是什么新鲜事。Helen Ngo ( 2016 ) 遵循 Merleau-Ponty 的感知现象学，将种族主义描述为“深深植根于我们的运动、手势、感知和方向的身体习惯”（Ngo, 2016 , p. 847）。同样遵循 Merleau-Ponty ( 1945/2002 ) 的步骤，Céline Leboeuf ( 2020 ) 主张将隐性偏见视为感知的、体现的社会习惯。纳蒂法·格林（2020 ）) 依赖于习惯的概念来说明刻板印象的威胁，即在传统上与一个人所属的社会群体的刻板印象相关的任务中表现不佳的感觉——无论是性别、种族、性方向、国籍或职业。格林认为，刻板印象威胁引发的焦虑和疏离是一种形式习惯的破坏。刻板印象威胁在于感觉一个人的技能和习惯被打乱了正常的“流动”。⁴

我承认所有这些分析的重要性和深度，其中大部分位于现象学传统或所谓的具身和嵌入认知方法中。我为这个论点提供了一个论据，即偏见行为最好被概念化为一种被动获得的习惯，它与现象学传统中的类似观点相辅相成，而不是与之竞争。需要明确的是，这一核心主张并不是关于隐性偏见本身的论文。这是一篇关于由这种隐含偏见引发的行为的论文——被理解为心理或身体行为。我的论点是对最佳解释的推论，其前提是心理学和神经科学必须提供的习惯性行为观点。

如果我在这里捍卫的论点是合理的，并且我们承认，作为平等主义代理人，我们生活的重要组成部分仍然是充满偏见的习惯——我们似乎没有意识到的自动行为——我们是否可以为这种偏见负责？行动？将偏见行为理解为一种被动获得的习惯类型，可以诊断出一系列直觉，这些直觉为现有的此类行为负责的方法提供了依据。它还有助于对我们对偏见行为的责任进行基于能力的描述，从而证明对有偏见的代理人的指责反应是正当的，同时承认她的行为不应该受到指责。这种基于能力的责任描述对于上述隐性偏见的任何特征都是中立的。

本文组织如下：在第 2 节中，我概述了传统上与习惯概念相关的主要特征，并强调了一些支持习惯行为的神经生物学和神经心理学过程，以便确定我认为的习惯类型。与表征偏见行为有关。在第 3 节中，基于前面的考虑，我主张偏见行为是一种被动获得的习惯性行为。在第 4 节中，我讨论了我们对自己的习惯负有责任的问题。由于习惯最终容易受到控制，代理人有责任，也应该受到指责，我认为，未能培养发现需要行使智力、道德、社会和谨慎义务的情况的能力. 然而，因此负责与代理人不应该受到指责是一致的。因为被指责的代理人隐含偏见行为的自动性使其与她已经关心的智力、道德、社会和审慎价值观无关。⁵