Previous research suggests that the influence of facial trustworthiness is irresistible in the process of making judgments even when relevant behavioral information is available. This leaves the possibility that people's evaluations of others, especially implicit evaluations, might be determined by facial appearances alone, despite the lack of evidence showing the accuracy of inferences made from faces. In this paper, however, we identified properties of behavioral information—extremity and reliability—that are key to rapidly, effectively, and durably changing implicit evaluations based on facial trustworthiness. Across three pre-registered studies, we showed that implicit evaluations based on facial trustworthiness were updated when, and only when, the propositional information was extreme (Study 1). However, extreme information alone was not enough. Extreme information needed to be perceived as reliable (believable) in order to change face-based implicit evaluations (Study 2). Moreover, updating face-based implicit evaluations with one piece of reliable and extreme information showed stability over a three-day interim (Study 3). These results suggest that impressions based on facial trustworthiness are not insurmountable. After a brief, single-session learning session with a specific type of new evidence, initial implicit evaluations were reversed and persisted across three days.

以往的研究表明，即使有相关的行为信息，面部可信度的影响在做出判断的过程中也是不可抗拒的。这使得人们对他人的评价，尤其是隐性评价，可能仅由面部外观决定，尽管缺乏证据表明面部推断的准确性。然而，在本文中，我们确定了行为信息的属性——极端性和可靠性——对于基于面部可信度快速、有效和持久地改变隐性评估至关重要。在三项预先注册的研究中，我们表明基于面部可信度的内隐评估在且仅当命题信息极端时才会更新（研究 1）。然而，仅靠极端信息是不够的。极端信息需要被视为可靠（可信）才能改变基于面部的隐式评估（研究 2）。此外，使用一条可靠且极端的信息更新基于面部的隐式评估表明在三天的过渡期内具有稳定性（研究 3）。这些结果表明，基于面部可信度的印象并非不可克服。在对特定类型的新证据进行了一次简短的单次学习会议后，最初的内隐评估被逆转并持续了三天。用一条可靠和极端的信息更新基于面部的隐式评估显示了三天的稳定性（研究 3）。这些结果表明，基于面部可信度的印象并非不可克服。在对特定类型的新证据进行了一次简短的单次学习会议后，最初的内隐评估被逆转并持续了三天。用一条可靠和极端的信息更新基于面部的隐式评估显示了三天的稳定性（研究 3）。这些结果表明，基于面部可信度的印象并非不可克服。在针对特定类型的新证据进行了一次简短的单次学习会议后，最初的内隐评估被逆转并持续了三天。