Often the source of information is as important as the information itself. The present study examined the impact of contextual threat and safety signals (source information) on memory encoding and recognition of faces (item information). In two experimental sessions, 30 participants viewed neutral face pictures. In the encoding session, 60 faces were presented with contextual background colors (blue or green, 30 pictures each) which were verbally instructed to signal either threat-of-shock or safety. In the recognition session, the 60 old faces intermixed with 30 new faces were shown while a combined old/new recognition and source memory task was performed. During the encoding session, face processing varied as a function of contextual source information. Confirming successful threat-of-shock manipulation, threatening compared to safe face–context compounds revealed differential neural processing (early parieto-occipital and late fronto-central negativity) as well as pronounced threat ratings. During the recognition session, participants had serious problems identifying old from new faces with poor source memory. Intriguingly, however, brain activity differentiated previously seen faces from newly presented pictures (old/new ERP effect). Moreover, old faces presented within a threat context were associated with distributed late negativities compared to old safe faces. Thus, threat effects not only emerged during face encoding (incidental learning) but also during face recognition, although no valid judgements could be made regarding the threatening or safe sources. These findings support the notion that contextual source information critically modulates person perception and recognition as a form of an expectation based remembering in the absence of conscious recognition.

通常，信息来源与信息本身一样重要。本研究考察了情境威胁和安全信号（源信息）对记忆编码和面部识别（项目信息）的影响。在两次实验中，30 名参与者观看了中性的面部图片。在编码会话中，60 张人脸被呈现上下文背景颜色（蓝色或绿色，每张 30 张图片），这些颜色被口头指示以表示电击威胁或安全。在识别过程中，展示了 60 张旧脸与 30 张新脸混合在一起，同时执行了新旧识别和源记忆任务。在编码会话期间，面部处理随上下文源信息而变化。确认电击威胁操作成功，与安全的面部上下文复合物相比，威胁性显示出不同的神经处理（早期顶枕叶和晚期额中央负性）以及明显的威胁等级。在识别过程中，参与者在识别具有较差源记忆的新面孔时遇到了严重的问题。然而，有趣的是，大脑活动将先前看到的面孔与新呈现的图片区分开来（旧/新 ERP 效应）。此外，与旧的安全面孔相比，在威胁上下文中呈现的旧面孔与分布式后期消极性相关。因此，威胁效应不仅在人脸编码（附带学习）期间出现，也在人脸识别期间出现，尽管无法对威胁或安全来源做出有效判断。