Judging the mood of the crowd: Attention is focused on happy faces [Dataset]

Previous research on valence biases in face perception revealed inconsistent findings either proposing angry or happy faces to be detected more efficiently. We argue that the typical experimental task in this field, the face-in-the-crowd (FiC) paradigm, lacks ecological validity and leads to ambiguous results. In the present paper, we introduce a new task, the mood-of-the-crowd (MoC) paradigm that can complement existing FiC findings. In the new task, participants have to decide which expression is shown by most faces in a crowd. In Experiment 1, photographs were used as stimuli, whereas computer-generated faces were presented in Experiment 2. While in the first experiments either happy and neutral or angry and neutral emotions were shown within one crowd, in Experiment 3, crowds were composed of angry and happy faces. Gaze position was recorded to gain further insights on attentional processes. In the first experiment, happy moods were identified faster and with a higher accuracy. Although we could only find higher accuracy rates for happy moods in experiment 2, happy faces were fixated more frequently indicating faster processing of more information in the same amount of time. A higher proportion of fixations and first fixation on happy faces occurred in experiment 3. Moreover, target gender was found to be an important moderator. In female crowds, emotion was generally assessed more accurately. Additionally, while emotional faces were overall focused more often than neutral faces, this effect was more pronounced in female crowds. The same pattern emerged in the comparison of happy vs. angry faces: Female happy faces had the highest probability to attract attention. The close connection between femininity and emotionality or rather happiness is discussed as a possible reason for these findings.

过往有关面孔知觉效价偏差（valence biases）的研究结论存在分歧：部分研究提出愤怒面孔的探测效率更高，另有研究则认为快乐面孔更易被快速察觉。我们认为，该领域的经典实验任务——人群中的面孔（face-in-the-crowd, FiC）范式——缺乏生态效度（ecological validity），进而导致研究结果模棱两可。本文提出一种全新实验范式：人群情绪（mood-of-the-crowd, MoC）范式，可作为现有FiC研究发现的补充。在该新任务中，参与者需判断群体面孔中占多数的表情类型。

实验1采用实拍照片作为刺激（stimuli）材料，实验2则使用计算机生成面孔作为刺激。在实验1与实验2中，每组面孔群仅包含快乐与中性，或愤怒与中性两类情绪；而实验3的面孔群由愤怒与快乐面孔混合组成。本研究同步记录眼动注视（fixations）位置，以深入探析注意加工机制。

实验1结果显示，快乐情绪的识别速度更快、准确率更高。尽管实验2仅在快乐情绪的识别准确率上表现更优，但快乐面孔的注视频次更高，这表明在相同时长内可更快完成更多信息的加工。实验3中，对快乐面孔的注视占比更高，且首次注视快乐面孔的概率也更大。

此外，研究发现目标面孔的性别为重要调节变量（moderator）。在全女性面孔群中，情绪类型的评估准确率普遍更高。同时，尽管整体而言情绪面孔的注视频次高于中性面孔，但该效应在女性面孔群中更为显著。在快乐与愤怒面孔的对比中，同样呈现出这一模式：女性快乐面孔最易吸引注意。本文将女性气质与情绪性（更确切地说，与快乐情绪）之间的紧密关联，作为上述研究结果的潜在成因展开讨论。