Berryessa et al. (2022) consider how prior experience as a criminal prosecutor may influence judicial behaviour, but their concerns about prior experience apply much more broadly in the case of American judges. In the United States, unlike many other countries, lawyers with experience as advocates comprise the great majority of persons selected to be judges (Volcansek, 2010; Wilets et al., 2022), and most American lawyers focus their practice on particular areas and particular client types, such as representing employers or employees in employment disputes. Specialized experience of any kind, whether as a prosecutor or corporate attorney—and both these types of lawyers have been over-represented among recent judicial nominees to the federal bench in the United States (Shepherd, 2021)—may lead to preconceptions, preferences and blind spots that influence how judges discharge their duties.

Berryessa et al. rightly question greater diversification of the bench as a solution because balancing biases across judges does nothing to protect the parties who must appear before the pro-prosecution, pro-business or pro-plaintiff judge. Furthermore, given the many prior advocacy roles judges bring to the bench, it is unrealistic to expect sufficient experiential diversification to achieve balance in the aggregate. Berryessa et al. also rightly question the power of short-term training to produce long-term debiasing effects (see, e.g. Bezrukova et al., 2016; Greenwald et al., 2022). We should not despair, however, for American legal systems employ several measures that constrain judges' idiosyncratic beliefs and personal values to produce merits-based decision-making. In particular, the use of adversarial presentation of arguments and evidence, paired with structured decision-making taking place inside a strong accountability matrix, should promote impartial decision-making.

One of the most effective methods to reduce confirmation bias and biased assimilation of evidence involves considering opposing viewpoints and alternative possibilities (e.g. Lilienfeld et al., 2009; van Brussel et al., 2020). The adversarial presentation of evidence and arguments by parties with conflicting incentives, playing on a field levelled by procedures that reduce the significance of resource disparities, ensure that judges leaning to one side or the other are exposed to competing arguments and factual narratives. This process promotes individuation and perspective-taking that should disrupt preconceptions or stereotypes judges bring to a case (Devine et al., 2012).

Judges do not have free reign to admit evidence and make decisions as they see fit but rather must act according to extensive procedural and substantive laws that govern how judges should handle and decide cases, along with rules directing judges to recuse themselves from cases in which their personal interests, as opposed to the law, might determine the outcome. Many laws employ bright-line rules or multi-factor tests that constrain judicial discretion and focus judges on the legally relevant considerations (e.g. Girvan, 2016).

Judges who place their personal interests above the law can expect to have their actions challenged, for American judges act within an extensive accountability matrix. The most obvious check on judicial power is the hierarchical structure of American court systems, which ensures that all cases can be reviewed by at least one panel of disinterested judges sitting on a court of appeals. This appellate review is designed to detect both factual and legal errors after the conclusion of proceedings in trial courts (most cases cannot be appealed from the trial court until a final judgement by that court has been issued); thus, judges in the trial courts have a strong incentive to decide cases by applying governing law to a reasonable assessment of the facts to avoid having to expend more effort later on the same case. (Because the norm is for cases overturned on appeal to be remanded to the original judge (Heytens, 2014), judges understand the importance of issuing a defensible decision in the first instance.) Reversals of decisions also affect judicial reputations, which may be of particular concern to elected state court judges and federal judges who aspire to appointment to higher courts (Epstein et al., 2013; Paynter & Kearney, 2010).

Another important part of this accountability matrix is the requirement that judges issue reasoned explanations for their actions (Oldfather, 2008). Typically these explanations take the form of written opinions following a bench trial or disposing of motions submitted to the court, but during trial judges usually offer oral explanations for their rulings on evidentiary objections, with the explanations captured by the transcript for the record on appeal. Knowing before issuing a decision that the decision will be reviewed by others for compliance with the law serves as an effective way of inducing judges to issue merits-based decisions. Decision-makers who know before deciding that they will be held accountable for following prescribed procedures in an unbiased way are more likely to reach unbiased decisions than unaccountable decision-makers (Lerner & Tetlock, 1999). The requirement of written decisions by appellate courts also facilitates learning: by providing clear feedback on how a case should have been handled or decided, lower courts should be less likely to repeat those mistakes in subsequent cases.

Yet, another important part of the accountability matrix is the use of multi-member courts and the random assignment of judicial panels to cases in the courts of appeal. Because no particular viewpoint or perspective dominates on courts of review (though certainly some appellate courts lean left or right in their collective decisions over time) and because the composition of the review panel will be randomly determined, judges whose decisions are subject to review can only be sure that their factual findings will be evaluated for accuracy and their legal rulings will be evaluated for legal fidelity. In other words, a trial court judge will be uncertain about what biases the reviewing judges will hold other than a bias in favour of following the law.

In sum, although Berryessa et al. offer good reasons to worry about former prosecutors (and any other former advocates) serving as judges, American legal institutions create conditions that should foster merits-based decision-making. Injustices no doubt occur—sometimes because of ineffective assistance of counsel, sometimes because judges fail to apply a critical eye to expert testimony and other evidence, and sometimes because judges fail to act impartially—but judges with different backgrounds often treat similar cases similarly, suggesting that existing constraints on judges work well to reduce the influence of judges' personal biases (see Harris & Sen, 2019; Mitchell, 2019). Nonetheless, further research is necessary to ensure that former prosecutors serving as judges are not uniquely immune from the constraints that American legal institutions impose on judges.

贝里萨等人。（2022）考虑了作为刑事检察官的先前经验如何影响司法行为，但他们对先前经验的担忧更广泛地适用于美国法官的案例。在美国，与许多其他国家不同，具有辩护律师经验的律师占被选为法官的绝大多数人（Volcansek，  2010；Wilets 等，  2022 ）），大多数美国律师的执业重点是特定领域和特定客户类型，例如在雇佣纠纷中代表雇主或雇员。任何类型的专业经验，无论是作为检察官还是公司律师——这两种类型的律师在美国联邦法官最近的司法提名人中所占比例过高（Shepherd，2021）——可能会导致偏见、偏好和 偏见。影响法官履行职责的盲点。

贝里萨等人。正确地质疑法官席更加多元化作为解决方案，因为平衡法官之间的偏见无助于保护必须出庭支持起诉、支持企业或支持原告的法官的各方。此外，考虑到法官之前担任过许多辩护角色，期望充分的经验多样化以实现总体平衡是不现实的。贝里萨等人。也正确地质疑短期训练产生长期去偏差效应的力量（参见 Bezrukova 等人，  2016 年；Greenwald 等人，  2022 年））。然而，我们不应该绝望，因为美国法律体系采取了多种措施来限制法官的特殊信仰和个人价值观，以做出基于案情的决策。特别是，使用对抗性的论点和证据陈述，再加上在强大的问责矩阵内进行的结构化决策，应该可以促进公正的决策。

减少确认偏差和有偏见的证据同化的最有效方法之一是考虑相反的观点和替代可能性（例如 Lilienfeld 等人，  2009 年；van Brussel 等人，  2020 年）。动机相互冲突的各方以对抗性方式提出证据和论点，在一个通过减少资源差异重要性的程序所营造的公平环境中进行竞争，确保倾向于某一方的法官能够接触到相互竞争的论点和事实叙述。这一过程促进了个性化和观点采择，从而打破了法官对案件的先入之见或刻板印象（Devine 等，  2012）。

法官没有接受证据和做出他们认为合适的决定的自由支配权，而是必须根据指导法官如何处理和裁决案件的广泛程序法和实体法行事，以及指导法官回避其案件的规则。个人利益（而不是法律）可能决定结果。许多法律采用明线规则或多因素测试来限制司法自由裁量权，并使法官重点关注法律相关的考虑因素（例如 Girvan，  2016）。

将个人利益凌驾于法律之上的法官的行为可能会受到挑战，因为美国法官在广泛的问责矩阵内行事。对司法权最明显的制约是美国法院系统的等级结构，它确保所有案件都可以由上诉法院中至少一个由公正的法官组成的小组进行审查。这种上诉审查的目的是在初审法院诉讼结束后发现事实和法律错误（大多数案件在初审法院做出最终判决之前不能上诉）；因此，初审法院的法官有强烈的动机通过适用适用法律对事实进行合理评估来判决案件，以避免以后在同一案件上花费更多精力。 2014），法官了解在一审中做出可辩护的判决的重要性。）判决的撤销也会影响司法声誉，这可能是民选州法院法官和渴望任命为高等法院法官的联邦法官特别关注的问题（Epstein 等）等人，  2013 年；Paynter & Kearney，  2010 年）。

该问责矩阵的另一个重要部分是要求法官对其行为做出合理解释（Oldfather，  2008）。通常，这些解释在法庭审判或处理提交给法院的动议后采取书面意见的形式，但在审判期间，法官通常对其对证据反对意见的裁决进行口头解释，并将解释记录在上诉记录中。在做出决定之前知道该决定将由其他人审查其是否符合法律，是促使法官做出基于案情的决定的有效方式。决策者在决定之前就知道自己将以公正的方式遵循规定的程序，比不负责任的决策者更有可能做出公正的决定（Lerner & Tetlock，1999  ））。上诉法院书面判决的要求也有助于学习：通过提供关于案件应如何处理或判决的明确反馈，下级法院应该不太可能在后续案件中重复这些错误。

然而，问责矩阵的另一个重要部分是使用多人法院以及对上诉法院案件随机分配司法小组。由于没有特定的观点或观点在复审法院中占据主导地位（尽管随着时间的推移，一些上诉法院在集体裁决中肯定会左倾或右倾），并且由于复审小组的组成将是随机确定的，因此其裁决受到复审的法官只能确保他们的事实调查结果将得到准确性评估，他们的法律裁决将受到法律忠实度评估。换句话说，初审法院法官将不确定复审法官除了支持遵守法律的偏见之外还会持有哪些偏见。

总之，尽管 Berryessa 等人。尽管美国法律机构有充分理由担心前检察官（以及任何其他前辩护人）担任法官，但它们创造了促进基于案情决策的条件。毫无疑问，不公正现象时有发生——有时是因为律师的协助不力，有时是因为法官未能对专家证词和其他证据采取批判性的眼光，有时是因为法官未能公正行事——但不同背景的法官往往会以相似的方式对待类似的案件，这表明对法官的现有限制可以很好地减少法官个人偏见的影响（参见 Harris & Sen，  2019；Mitchell，  2019 ））。尽管如此，还需要进一步研究，以确保担任法官的前检察官并非唯一能够免受美国法律机构对法官施加的限制。