Critical influence of implicit motion features on hand-drawn lines’ emotional expression

Xinya Liu, Ruimin Lyu, Huan Liu, Guoying Yang, Yuefeng Ze

Keywords: hand-drawn lines, emotion, motion futures, empirical aesthetics, embodied simulation, affective computing, ordered logistic regression, random forest

Abstract: As a basic visual element, the way lines express and evoke human emotional responses is an important topic of interest in fields such as cognitive psychology and human-computer interaction. In the past, many studies have attempted to identify the relationship between the static features of lines and emotions but have overlooked the motion features of hand-drawn lines. However, some recent studies have shown that the motion behaviour implied by hand-drawn lines has a significant impact on viewers’ aesthetic-emotional judgments. In this study, we sought to explore the motion features that have a key impact when people actively express emotions through drawing lines. In the experiment, subjects actively expressed specified emotions by drawing lines by hand, and their pen movement data were fully recorded on a digital screen. Several static and motion features were estimated using the experimental data. Moreover, we quantified their correlations with emotional pleasure and arousal through multiple ordered logistic regression analysis. The resulting static and motion features that significantly affect the emotional expression of lines were identified. In addition, a variety of motion features were crucial in the emotional expression of hand-drawn lines. This could help guide artists and designers in their creative work and has the potential to be used in the development of emotional computing applications for line, picture, interactive art, and graphic design.

Track: Graphics

Summary Of Changes:

(1) Be clearer that the authors study emotional expression via lines：In this revision, we have placed greater emphasis on studying emotional expression via lines. Specifically, in the abstract and introduction, we use language such as "explore the motion features that have a key impact when people express emotions through drawing lines" to be clear that our focus is on studying emotional expression via lines. Furthermore, in the last paragraph of the introduction, we have added the following statement: "It is important to emphasise that the research in this paper focuses on the motion influencing factors on the expression of emotion through the line from the creator's point of view, and does not yet address the motion influencing factors that affect the viewer's perception." This is intended to clarify our research goals for the reader at an early stage.
(2) More details about how features are computed: In the supplementary materials, we provide new tables 5, 6, 7, and 8 to supplement details about the features. Table 5 describes all the raw data used for feature computation, table 6 describes the static features we extracted, and tables 7 and 8 describe the motion features we extracted. In tables 6, 7, and 8, we describe where each feature is extracted from, and in conjunction with table 5, readers can understand how these features were obtained. Additionally, we also note in tables 6, 7, and 8 which features come from previous research.
(3) Better definitions and introductions of terms: We have emphasized adjusting the logic of the introduction and modified some expressions to make the introduction smoother to read. We have also made some adjustments to certain terms, such as changing "the 1st order derivative of x(t)" to "velocity in the x direction", in the hope of improving terminology. As our features are computed using multiple calculations, their term descriptions can be somewhat obscure. We hope that the feature description tables we have provided will enhance understanding of these feature terms.
(4) Fix typos throughout the text: As non-native English speakers, we have done our best to correct grammar and spelling errors throughout the text, and to optimize the wording to make the article easier to read.
(5) Check axes and colors of all figures, especially Figures 3, 5, and 7: We have adjusted the brightness of the images for Figures 3 and 7 to ensure that the colors are distinguishable when printed in black and white. For Figure 5, we have corrected the order of the clustering levels for the arousal dimension.