Mori Taiga, Den Yasuharu, Jokinen Kristiina
Artificial Intelligence Research Center, National Institute of Advanced Industrial Science and Technology, Tokyo, Japan.
Graduate School of Science and Engineering, Chiba University, Chiba, Japan.
PLoS One. 2025 May 22;20(5):e0323448. doi: 10.1371/journal.pone.0323448. eCollection 2025.
Head nods are a commonly observed gesture in daily conversations and has attracted the interest of many researchers in human, social, and computer sciences since early studies. However, there has been little research focusing on the structure of nod movement, which sometimes involve repetition of upward and downward movements. In this study, we shed light on this structure of nods which has been previously overlooked. Prior to the analysis, we propose systematic conceptualization of nod movement. First, we define a cycle as a consecutive upward and downward movement as the basic unit of analysis. We next define the number of cycles within a nod as its length and the relative position of a cycle within a nod as its position. We then define the magnitude as the difference between the lowest and highest points of the head within a cycle. In the analysis, we demonstrate that this magnitude varies depending on length and position, thereby providing evidence that nods exhibit a structured pattern. Specifically, three structural patterns were observed: (1) the magnitude of the first cycles increases with length (anticipatory rising), (2) the magnitude decreases proportionally with position (declination), and (3) the magnitude of the final cycles is noticeably smaller than predicted from the preceding trend (final lowering). Finally, we discuss the similarities between the discovered structure and the phonological structure of utterances, suggesting that these structures may represent a universal characteristic of human repetitive actions.
点头是日常对话中常见的一种手势，自早期研究以来，就吸引了许多人文、社会和计算机科学领域研究人员的关注。然而，很少有研究聚焦于点头动作的结构，点头动作有时会包含上下动作的重复。在本研究中，我们揭示了这种此前被忽视的点头结构。在进行分析之前，我们对点头动作提出了系统的概念化。首先，我们将一个连续的上下动作定义为一个循环，作为基本分析单位。接下来，我们将一次点头中的循环次数定义为其长度，将一个循环在一次点头中的相对位置定义为其位置。然后，我们将幅度定义为一个循环内头部最低点和最高点之间的差值。在分析中，我们证明了这个幅度会根据长度和位置而变化，从而提供了点头呈现出结构化模式的证据。具体而言，观察到了三种结构模式：（1）第一个循环的幅度随长度增加（预期上升），（2）幅度随位置成比例减小（倾斜），（3）最后一个循环的幅度明显小于根据先前趋势预测的值（最终下降）。最后，我们讨论了所发现的结构与话语语音结构之间的相似性，表明这些结构可能代表了人类重复动作的一种普遍特征。
