Next Article in Journal
Continuity and Quality in Pre-Service Teacher Preparation Across Modalities: Core Principles in a Crisis Leadership Framework
Previous Article in Journal
AI in the Classroom: GPT Usage and Learner Typologies in Programming
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Education Sciences
Volume 15
Issue 10
10.3390/educsci15101354
education-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Re-Viewing the Same Artwork with Emotional Reappraisal: An Undergraduate Classroom Study in Time-Based Media Art Education

by
Haocheng Feng
1,
Tzu-Yang Wang
1,*,
Takaya Yuizono
1 and
Shan Huang
2
1
Graduate School of Advanced Science and Technology, Japan Advanced Institute of Science and Technology, 1 -1 Asahidai, Nomi 923-1211, Ishikawa, Japan
2
College of Fine Art and Design, Shenyang Normal University, 253 Huanghe N Ave, Huanggu District, Shenyang 110034, China
*
Author to whom correspondence should be addressed.
Educ. Sci. 2025, 15(10), 1354; https://doi.org/10.3390/educsci15101354
Submission received: 2 September 2025 / Revised: 8 October 2025 / Accepted: 9 October 2025 / Published: 12 October 2025
(This article belongs to the Section Education and Psychology)
Browse Figure
Versions Notes

Abstract

Learning and understanding of art are increasingly understood as dynamic processes in which emotion and cognition unfold over time. However, classroom-based evidence on how structured temporal intervals and guided prompts reshape students’ emotional experience remains limited. This study addresses these gaps by quantitatively examining changes in emotion over time in a higher education institution. Employing a comparative experimental design, third-year undergraduate art students participated in two structured courses, where emotional responses were captured using an emotion recognition approach (facial expression and self-reported text) during two sessions: initial impression and delayed impression (three days later). The findings reveal a high consistency in dominant facial expressions and substantial agreement in self-reported emotions across both settings. However, the delayed impression elicited greater emotional diversity and intensity, reflecting deeper cognitive engagement and emotional processing over time. These results reveal a longitudinal trajectory of emotion influenced by guided reflective re-view over time. Emotional dynamics extend medium theory by embedding temporal and affective dimensions into TBMA course settings. This study proposes an ethically grounded and technically feasible framework for emotion recognition that supports reflective learning rather than mere measurement. Together, these contributions redefine TBMA education as a temporal and emotional ecosystem and provide an empirical foundation for future research on how emotion fosters understanding, interest, and appreciation in higher media art education.

1. Introduction

In media art education, learning is not only directed toward the acquisition of conceptual knowledge but also toward the cultivation of perceptual acuity, imagination, and reflective judgment (Liao, 2016; Peppler, 2010). Within this process, learners undergo an emotionally infused aesthetic experience: an initial rapid appraisal generates a preliminary emotional impression, which is subsequently reshaped through reappraisal as additional cues, prior knowledge, goals, and social context intervene (Keltner & Haidt, 2003; Menninghaus et al., 2020). This dynamic process illustrates that learners’ emotions when viewing the same artwork are not static, but rather reappraisal over time often follows discernible patterns when structured by temporal stimuli. With the ongoing development of digital technologies and media forms, aesthetic experience has increasingly been understood as a dynamic process involving multidimensional interactions between perception, cognition, and emotion (Quaranta, 2013), particularly within the field of time-based media art (TBMA). As articulated by the Guggenheim Museum, TBMA works, such as video-, film-, slide-, audio-, or computer-based pieces, are inherently subject to continual change due to technological obsolescence and the allographic nature of their display. This characteristic underscores the temporal and processual qualities of TBMA as an art form (Laurenson, 2013). In this sense, TBMA aligns with the temporality of aesthetic experience, whereby guided reflective re-viewing can distribute learners’ emotions across successive moments of self-assessment, framing a discernible emotional trajectory.
In this context, the central pedagogical challenge in media art education should not be confined to technological implementation alone but should instead address how structured curricular time can be employed to reshape emotional experience, thereby deepening artistic understanding. At the same time, recent advances in emotion recognition have made it possible to assess learning processes rapidly and with minimal disruption to teaching (Poria et al., 2017). Over the past two decades, emotion recognition technology (ERT) has advanced from early unimodal approaches, such as facial or vocal analysis (Pantic & Rothkrantz, 2002), to sophisticated multimodal integration systems and has been gradually introduced into educational contexts (Gupta et al., 2023). ERT captures changes in students’ emotions within learning environments, enabling teachers to adjust instructional strategies and promote more personalized and inclusive educational experiences (Yuvaraj et al., 2025). However, current applications remain largely confined to conventional classroom or virtual learning environments, with limited adoption in art education, particularly in TBMA instruction. Moreover, there is no systematic methodological framework that integrates time structures, media interaction, and technological intervention to empirically examine the emotional trajectories generated through guided reflection in TBMA classroom settings.
Therefore, the present study seeks to propose a comprehensive research framework by integrating medium theory, models of aesthetic–emotional cognition, and emotion recognition technologies, with the aim of investigating the dynamic evolution of emotional cognition in TBMA education. Specifically, this study examines how two structured course sessions, spaced three days apart, can be used to describe, analyze, and support temporal patterns of emotional change within aesthetic experience. This study was conducted to answer the following research question: How does guided reflective re-viewing of the same TBMA work after a three-day interval change students’ emotional impressions?

2. Literature Review

2.1. Aesthetic Model and Medium Theory in TBMA Education

Aesthetic experience has shifted from being framed as moments of epiphanic insight to being understood as a dynamic process driven by the coupling of emotion, perception, and cognition (Silvia, 2005; Walton, 1993). Contemporary aesthetic models indicate that initial viewing of an artwork elicits a rapid primary appraisal, often manifesting as a dominant initial impression orientation, and that as time passes and contextual cues intervene, reappraisal gradually reshapes the emotional response (Leder & Nadal, 2014; Pelowski & Akiba, 2011; Pelowski et al., 2017). Complex emotions are therefore not instantaneous; they unfold through temporally spaced repeated experiences and reflective processes. Emotion during aesthetic experience could be viewed as a layered process influenced by retention and reactivation. Short-term memory preservation allows the dominant emotional trace formed during the initial viewing to persist, guiding subsequent appraisals (Conway, 2009; Semeniuta et al., 2016). At the same time, episodic recall and contextual reactivation modulate subtle affective variations, giving rise to non-dominant emotions that coexist with, and gradually refine, the dominant one (Talarico & Rubin, 2003). This interplay between emotional stability and differentiation provides a cognitive basis for understanding how guided re-viewing transforms affective trajectories over time.
In parallel, medium theory has developed from a narrow, material-centered definition to a complex system encompassing temporal, experiential, and technological dimensions (Elleström, 2010). The medium is no longer merely a conduit of transmission but a key locus of meaning-making (Potts, 2004). Recent theories of the medium underscore variability, modularity, and interactivity, emphasizing how the medium, perception, and technology co-shape experience over time (Chierico, 2016; Rancière, 2011). Within the integrated framework of contemporary models of aesthetic experience and recent media theory, TBMA is, fittingly, an art form centered on temporal structuring and interactive participation; it therefore requires interdisciplinary collaboration and stable strategies for re-presentation in exhibition and instructional settings (Rinehart & Ippolito, 2022; Vassos et al., 2016). TBMA works typically depending on specific playback devices and environmental conditions to function as intended. Technological and environmental obsolescence often renders works unplayable, creating substantial preservation challenges (Phillips, 2012). Institutions thus face critical decisions about retaining original equipment, migrating content to new formats, or reconstructing outdated environments. Each approach affects authenticity and integrity, and researchers and conservators must confront these dilemmas and identify appropriate strategies to re-present TBMA artworks.

2.2. Emotion Recognition as a Methodological Instrument

Over the past two decades, affect recognition has progressed from early unimodal, lab-constrained approaches to multimodal, low-intrusion measurement frameworks that are deployable in authentic educational settings (Pantic & Rothkrantz, 2002). Unimodal methods based on facial action units, acoustic prosody, or posture can infer emotional states reliably under controlled conditions, but their ecological validity is limited (García-Hernández et al., 2024). In response, multimodal integration (face, voice, physiological signals, and text) improves robustness and temporal sensitivity in classrooms, enabling the capture of emotional temporal trajectories across learning episodes (Alves-Noreña et al., 2025; Franěk et al., 2022). In education, these methods have been used to characterize the onset, transitions, and dwell time of emotions, and to relate these dynamics to learning strategies and achievement (Kołakowska et al., 2014). Accordingly, emotion recognition technologies could serve as instrumentation in TBMA education research, providing auditable quantitative indicators of longitudinal, cross-time-point emotional change, thereby supporting tests of time-structured pedagogy and learning mechanisms.

2.3. Curriculum and Context in TBMA Education: U.S., Japan, and China

As indicated in comparative overviews (Table 1), universities in the United States have established time-based media as a distinct academic category, supported by curricula and preservation strategies, whereas universities in Japan and China more often subsume time-based practices under video art, media art, or digital media, emphasizing the moving image, installation, interactivity, and technological integration.
While the U.S. has recognized time-based media as a formal category in academic and museum contexts, Japan and China address similar practices through different cultural and institutional lenses. The emphasis in Japan and China frequently falls on technical proficiency, digital aesthetics, and media experimentation rather than on the ontological significance of time in art (Spielmann, 2012). Therefore, regional TBMA curriculum development could integrate the conceptual rigor of a time-based framework with locally specific forms of media arts education while employing emotion-recognition technologies to render the emotional dimension of TBMA learning quantifiable and visible.

3. Materials and Methods

This study used a within-participant comparative design in a tiered classroom at Shenyang Normal University. Students attended two course sessions in the same week, scheduled three days apart. In both sessions, they viewed the same time-based media art (TBMA) video, but the temporal structure differed (Table 2). At present, video documentation affords the most faithful representation of the essential features of TBMA works. In the first session, the video was screened continuously, and the students completed an immediate written reflection. In the second session, the same work was presented as a guided reflective re-viewing that used curated still frames of fixed duration with brief prompts and short pauses to support recall and appraisal.
Initial impression (II)—During the initial session, the participants directly viewed videos related to TBMA. Facial expressions of the participants were recorded using their private smart mobile devices (smartphones or tablets), while self-reported textual questionnaires were simultaneously administered.
Delayed impression (DI)—As this study adopted an empirical research methodology and anticipated future extensions over a broader temporal scope, the delayed session was determined based on the actual academic calendar of Shenyang Normal University, where courses are held twice a week with a three-day interval. Accordingly, the delayed session was scheduled to take place on the third day following the conclusion of II. During the delayed session, the participants recalled the video content by viewing still frames extracted from the same footage. Their facial expressions were recorded using their private smart mobile devices, and they subsequently completed self-report textual questionnaires.
Given that the interval between the two viewing sessions in this study was limited to three days, the influence of forgetting was expected to be minimal. Consequently, the dominant emotional orientation formed during the initial impression was likely to persist, although its intensity could slightly decrease. In contrast, non-dominant emotions were expected to undergo more subtle modulation, becoming richer and more nuanced through guided reflective re-viewing. Based on this reasoning, two hypotheses were formulated to examine the consistency of and variation in emotions between the initial and delayed impressions:
Hypothesis 1.
The dominant emotions in the II and DI conditions exhibit a high degree of consistency.
Hypothesis 2.
The DI condition leads to a greater variety of emotional categories and more profound non-dominant emotions compared with II.

3.1. Participant Demographics and Ethical Considerations

The participants in this study comprised 24 third-year undergraduate students majoring in painting at Shenyang Normal University (Table 3). They were randomly assigned into four groups of six students each (Mage = 21.1, SDage = 0.68). All participants were from the same class. Prior to their enrollment at Shenyang Normal University, they had undergone systematic training in drawing, color theory, sketching, and art design.
According to the latest publicly available data released by the Chinese Ministry of Education in 2023, the proportion of female students in university art programs is relatively high. Therefore, the gender distribution of the sample in this study reflects the natural distribution, without any artificial balancing of gender proportions (http://en.moe.gov.cn/).
Given that the experiment involved the recording of the participants’ facial expressions, careful ethical considerations were implemented to protect the participants’ privacy and rights. All participants signed a written informed consent form prior to participation. The participants were allowed to use their own smartphones or personal devices to ensure comfort and control during data collection. The recording duration was designed to be brief yet effective to minimize discomfort and intrusion. The participants were informed that the devices used had no unauthorized image enhancement or surveillance functions. They were also provided with opportunities to give feedback and express any concerns after the experiment. Data collection was conducted on-site, and all 24 participants provided the corresponding data. All participants had normal or corrected-to-normal vision.

3.2. Experimental Design and Procedure

This study involved four groups of participants, each of whom took part in two on-site experimental sessions conducted three days apart. Each group viewed a different video recording of a TBMA project sourced from YouTube. These videos varied in duration, stylistic characteristics, levels of abstraction, and visual complexity (Table 4).
The videos were projected onto a 195 cm × 195 cm screen, maintaining the original resolution and aspect ratio as downloaded from YouTube. The setting of the tiered classroom and large screen ensured the controllability and effectiveness of the experiment to the greatest extent possible (Ravaja et al., 2004; Wang et al., 2009). During the video viewing, each participant’s facial expressions were recorded using their private smart mobile devices, and after watching the videos, they were asked to write down their thoughts and emotional responses regarding the content. During the recording of the participants’ facial expressions, on-site researchers ensured the independence of each participant by excluding other participants from the footage.
Similarly, the questionnaires were completed without any mutual influence between the participants. Both experimental sessions were conducted in the same tiered classroom (see Figure 1).

3.2.1. Detailed Description of the Experimental Procedure (For Initial Impression)

The design of the experiment was based on the 45 min course time allocation at Shenyang Normal University.
Stage 1: The participants were introduced to the experimental procedures, including a five-minute informed consent briefing, followed by a five-minute explanation of the experimental process and assistance in setting up the facial expression recording devices.
Stage 2: Each group of six participants watched a different TBMA video while sitting in a tiered classroom facing the projection screen. Each participant’s private smart mobile device was mounted on a tripod to record their facial expressions during the viewing.
Stage 3: To ensure that the participants retained a clearer memory for completing the questionnaire, only a 30 s break was given after the video presentation (Semeniuta et al., 2016).
Stage 4: Based on emotion measurement under laboratory conditions, a questionnaire was designed (Wallbott & Scherer, 1989). The participants completed a self-report questionnaire on their emotional responses, writing a 300-word reflection on their experience while watching the video. The prompt for the self-report was as follows: “Please describe your feelings after watching this video (describe the emotions you felt during it in as much detail as possible; in about 300 words).”
After the experiment, an open-ended group discussion was conducted in the tiered classroom to assess whether the participants experienced any discomfort or concerns. The participants were assured that they could discuss any personal concerns privately, ensuring their privacy and well-being.

3.2.2. Detailed Description of the Experimental Procedure (For Delayed Impression)

The second session followed a similar four-stage structure, with notable modifications in time allocation and procedural details.
In Stage 1, the informed consent explanation was omitted, as the participants had already provided consent in the first session. Instead, they only received a briefing on the specific procedures for the DI.
In Stage 2, the participants did not rewatch the full videos but instead viewed still frames extracted from the same TBMA videos at one-minute intervals. Each still frame was presented for 17 s, in accordance with the optimal viewing duration suggested by Brieber et al., allowing the participants to recall their previous viewing experience and associated emotions (Brieber et al., 2020).
In Stage 3, to stimulate the participants’ recall and facilitate better cognitive processing while also preserving their short-term memory, the break duration was extended to two minutes (Cowan & AuBuchon, 2008).
In Stage 4, the participants were asked to complete the self-report emotional questionnaire, attempting to recall their initial impressions from the first session while writing a 300-word reflection on their experience. The prompt for the self-report was as follows: “Please describe your feelings after watching content related to this video? (Describe the emotions you felt during it in as much detail as possible; in about 300 words.”
At the conclusion of the experiment, a detailed explanation of the artistic intent behind the TBMA works was provided, addressing the participants’ questions and concerns regarding the artworks.

3.3. Data Preprocessing and Statistical Analysis

3.3.1. Data Preprocessing of Facial Expressions

Before conducting the data analysis, the facial expression data collected under the initial impression (II) and delayed impression (DI) conditions across all four groups were systematically preprocessed. The preprocessing steps were as follows:
Part 1: The participants’ facial expression recordings were stored in folders categorized by group number and experimental condition (II vs. DI). Each participant’s video file was labeled for easy identification.
Part 2: One frame per second was extracted from each participant’s video using Python 3.10 (implemented in PyCharm 2023.3.3, an integrated development environment commonly used for machine learning applications) to capture close-up facial images. Each frame was timestamped and stored in sequential order. Given that macro-expressions typically last between 0.5 and 4 s and are readily observable, a sampling rate of 1 fps is generally sufficient for effective analysis, particularly when the original video is recorded at a standard frame rate (25–30 fps) (Ben et al., 2021). This approach was adopted in the present study to reduce the computational load while preserving essential emotional cues.
Part 3: A Convolutional Neural Network (CNN) emotion analysis model was applied to automatically classify the emotional states in each extracted image frame. The analysis was conducted in Python using PyCharm. The model had pre-trained CNN architecture based on PyTorch for facial expression recognition. This publicly available model was obtained from the GitHub website (model link, https://github.com/WuJie1010/Facial-Expression-Recognition.Pytorch, accessed on 2 August 2025) and used to ensure reproducibility and benchmark-level performance. Each extracted frame was categorized into one of seven emotional states: anger, disgust, fear, happiness, sadness, surprise, and neutral (He et al., 2016). For each frame, the model generated probability scores for all emotion categories, and the dominant emotion per second was determined automatically based on the highest probability score. The entire classification process was completed by the algorithm without any human judgment or manual labeling. Finally, the frequency (in seconds) of each detected dominant emotion was recorded for every participant to support a quantitative comparison between experimental conditions.
Part 4: Since the video durations varied across the groups and the recording times for the II and DI conditions differed, a percentage-based normalization was applied to the emotion frequency data. This ensured unbiased comparisons across groups, eliminating potential distortions due to time differences (Aksu et al., 2019).
Part 5: After normalization, Min–Max normalization was applied to compute the dominant emotions for each participant under the II and DI conditions (Rafique et al., 2022). The dominant emotion of each participant was recorded for further statistical analysis.

3.3.2. Data Preprocessing of Textual Emotions

The textual emotion data collected under the II and DI conditions across all four groups were systematically preprocessed. The preprocessing steps were as follows:
Part 1: The participants’ handwritten questionnaires were transcribed and stored in an electronic Word format. The files were categorized into folders by group number and experimental condition (II vs. DI), with each file labeled for identification purposes.
Part 2: Using PyCharm, emotion analysis was conducted based on a Chinese emotion lexicon, extracting emotion-related words from the participants’ responses. The emotion analysis model classified words into seven emotional categories: happy, good, anger, sad, fear, dislike, and surprise (Li et al., 2010). Additionally, the model identified the frequency of words associated with each emotion, as well as the stop words, total word count, and sentence count (Deng & Nan, 2022).
Part 3: Similar to the preprocessing of facial expression data, the dominant emotion in the textual responses was processed using Min–Max normalization. The dominant emotion for each participant was computed and recorded for the II and DI conditions.
These preprocessing steps ensured that the facial expression and textual emotion data were systematically analyzed while maintaining comparability across the experimental conditions.

3.3.3. Statistical Analysis of Facial Expressions

To evaluate the consistency of the dominant emotions identified under the II and DI conditions across a three-day interval, Cohen’s Kappa coefficient was applied as a statistical measure to evaluate the agreement between the II and DI dominant emotions (Höfel & Jacobsen, 2003). Since this study was based on quantitative analysis with a small sample size and slight variations between the groups, the normality of the data distribution was first assessed before analyzing the diversity and intensity of emotions under the II and DI conditions (Altman & Bland, 1995). After confirming that all data approximately followed a normal distribution, the homogeneity of variance across the four groups was further evaluated to ensure the validity of subsequent analyses. Given that the participants were divided into four groups, each viewing a different video, a one-way ANOVA was conducted to confirm the absence of significant baseline differences in emotional responses between the groups (Judd et al., 2017). After confirming that the results of the normality tests and the one-way ANOVA met the assumptions required for conducting a paired-sample t-test, to examine the emotional diversity, a paired-sample t-test was conducted to compare the range of emotions exhibited under the II and DI conditions. This test evaluates whether the mean difference between paired observations is statistically significant. Specifically, the analysis focused on whether the DI condition elicited a broader spectrum of emotional expressions compared with the II condition. Furthermore, to determine whether non-dominant facial emotions (NFEs) were more prominent and intense under the DI condition compared with the II condition, the total proportion of non-dominant emotions observed for all participants was calculated for both conditions. A paired-sample t-test was then conducted to assess any significant differences (Blair & Higgins, 1985).

3.3.4. Statistical Analysis of Textual Emotions

This study analyzed self-reported emotional text data by assessing the consistency, diversity, and intensity of emotional expressions under the II and DI conditions. Cohen’s Kappa coefficient was used to evaluate the consistency of the dominant emotions, ensuring the reliability of the categorized emotional data derived from the textual analysis. Similarly, after confirming that the results of the normality test and the one-way ANOVA satisfied the assumptions required for conducting a paired-sample t-test, emotional diversity was examined through a paired-sample t-test, which compared the range of emotions expressed in self-reported textual responses across the two conditions. The goal was to determine whether the DI condition elicited a broader spectrum of emotional expression than the II condition. Differences in the intensity of non-dominant emotions were assessed using a paired-sample t-test on three key indicators: total non-dominant emotion words (TNEWs), representing the overall number of emotion-related words associated with non-dominant emotions; non-dominant emotion density (NED), reflecting the proportion of non-dominant emotion words within the total number of emotion words; and non-dominant emotion intensity per sentence (NEIPS), defined as the average number of non-dominant emotional words per sentence (Yadollahi et al., 2017).

4. Results

4.1. Consistency of Dominant Emotions Between Initial and Delayed Impressions

Table 5 displays a crosstabulation of the dominant facial emotions identified under the initial impression (II) and delayed impression (DI) conditions. The analysis revealed complete consistency between the two conditions: all 15 participants who were classified as exhibiting a dominant “neutral” expression in the II condition were likewise identified as “neutral” in the DI condition, and all 9 participants whose dominant emotion was classified as “sad” in the II condition were also categorized as “sad” in the DI condition. No discrepancies in the emotion categories were observed between the two conditions. These results suggest a high degree of consistency in the expression of dominant facial emotions across the II and DI conditions.
The self-reported textual data presented in Table 6 further support Hypothesis 1, indicating that most participants demonstrated consistent self-reported emotions. Specifically, among the 24 participants, 19 were consistently classified as expressing the dominant emotion of “good” under both the II and DI conditions, while 3 participants were consistently categorized as expressing “dislike.”
However, two participants exhibited discrepancies: one participant’s dominant emotion shifted from “good” in the II condition to “dislike” in the DI condition, while the other showed the reverse pattern. Compared with facial expression data, the consistency of dominant textual emotions across the two conditions was relatively high, though not perfect. Overall, most participants (22 out of 24) demonstrated consistent dominant emotional expressions between the II and DI conditions.
The Cohen’s Kappa coefficients and associated significance levels for the consistency of dominant emotions identified from facial expressions and self-reported textual responses are presented in Table 7.
For the facial expression data, the Kappa coefficient was 1.000 (p < 0.001), indicating perfect agreement between the II and DI conditions (Sim & Wright, 2005). This result corroborates the descriptive findings in Table 5 and demonstrates the complete stability of the participants’ dominant emotional expressions across the II and DI conditions. In contrast, the Kappa coefficient for the self-reported textual responses was 0.700 (p = 0.001), indicating substantial agreement. Although this consistency remains statistically significant at the p < 0.05 level, the lower Kappa value and higher p-value, relative to the facial expression data, reflect greater variability in the participants’ self-reported emotional states. Nonetheless, a certain degree of consistency between these two forms of emotional expression is maintained within a bounded range.

4.2. Comparative Analysis of Initial and Delayed Impressions

4.2.1. Comparison Across Groups

This study first conducted a one-way ANOVA on the difference scores obtained by subtracting the II score from the DI score within each group to examine whether variability at the group level (Groups 1 to 4) had a significant impact on the six emotional indicators (Table 8). This step was taken to ensure that any differences identified in the subsequent paired-sample t-test would not be confounded by random group assignment.
The ANOVA results indicate no statistically significant differences between the four groups on any of the emotional items (all p > 0.05). The effect size values (r) ranged from 0.138 to 0.465, suggesting small-to-moderate magnitudes. Furthermore, all mean values for the difference scores were positive, suggesting that emotional responses under the delayed impression (DI) condition were consistently higher than those under the initial impression (II) condition across all groups. Given these results, the groups did not appear to introduce systematic variation in the emotional responses. As a result, the data were aggregated across all groups for the paired-sample t-test, as group-level random effects were unlikely to bias the within-subject comparisons between the II and DI conditions (Q. Liu & Wang, 2021). The mean DI minus II differences in the non-dominant emotion density (NED) were tightly clustered across the groups (0.02 to 0.03). The between-group effect was not only non-significant but also negligible in magnitude (r < 0.10), indicating minimal group influence. Interpreted as a proportion, a difference of 0.02 corresponds to roughly 2 additional non-dominant emotion words per 100 words of reflection, which is meaningful given the brevity of the self-report writing.

4.2.2. Diversity of Emotional Categories Between Initial and Delayed Impressions

To examine the diversity of emotional expressions, a paired-sample t-test was conducted to compare the number of emotional categories expressed under the II and DI conditions. The total number of emotion categories was seven. Table 9 shows that the mean number of FECs in the II condition was 4.42 (SD = 0.93), whereas in the DI condition, it increased to 6.21 (SD = 0.72). The paired-sample t-test revealed a statistically significant difference in the DI condition (t23 = −8.98, p < 0.001). The extremely low p-value indicates a highly significant difference, and the effect size (d = 1.834) suggests a strong impact of the delay on the diversity of facial emotional expressions (Fritz et al., 2012). Similarly, for the self-reported textual emotions, the mean number of TECs was 2.63 (SD = 0.97) in the II condition, which increased to 4.83 (SD = 0.91) in the DI condition. The results indicate a statistically significant difference (t23 = −10.60, p < 0.001), confirming that self-reported emotional diversity was significantly higher in the DI condition. Notably, the effect size (d = 2.163) was even larger than that observed for facial expressions, suggesting that the delay had a greater impact on the self-reported emotional diversity. These results are consistent with the initial hypothesis, namely, that the DI condition would elicit richer and more diverse emotional expressions compared with the II condition.

4.2.3. Comparison of Non-Dominant Emotional Expressions Between Initial and Delayed Impressions

In this study, the non-dominant emotions examined in the paired-sample t-test referred to the total sum of emotional expressions exhibited by each participant, excluding their dominant emotion. These indicators were compared between the II and DI conditions to determine whether the DI condition elicited more diverse and intense emotional expressions (B. Liu, 2022; S. M. Mohammad, 2021). The proportion of NFEs (Table 9) significantly increased from M = 0.20 (SD = 0.12) in the II condition to M = 0.41 (SD = 0.12) in the DI condition (t23 = −10.59, p < 0.001, D = 2.173). The negative t-value and large effect size (d = 2.173) indicate that over time, the students exhibited stronger expressions of non-dominant emotions. The extremely low p-value (p < 0.001) suggests a highly significant statistical difference.
To further assess the intensity and prominence of non-dominant emotions, a detailed examination was conducted on three variables: TNEWs, NED, and NEIPS (Table 9). The number of TNEWs significantly increased from M = 3.63 (SD = 2.86) in the II condition to M = 9.92 (SD = 4.80) in the DI condition (t23 = −12.19, p < 0.001, d = 2.489), nearly doubling in usage. This suggests that when given more time, the students expressed emotions more profoundly, indicating a cognitive reconstruction of emotional perception in time-based media art experiences. Similarly, NED increased significantly from M = 0.02 (SD < 0.01) to M = 0.04 (SD = 0.01) (t23 = −9.83, p < 0.001, d = 2.006). This result demonstrates that not only were more emotions expressed but they also occupied a larger proportion of the students’ descriptive language, reinforcing the idea that the DI condition facilitates deeper emotional reflection and expression rather than merely increasing the emotional word count. Additionally, NEIPS increased from M = 0.44 (SD = 0.31) to M = 1.01 (SD = 0.41) (t23 = −11.06, p < 0.001, d = 2.26), confirming a statistically significant increase in the non-dominant emotional word usage per sentence. These findings suggest that the participants’ emotional expressions extended across multiple sentences, resulting in more nuanced emotional narratives, as they integrated emotions more deeply into their descriptions through additional reflective time.
The paired-sample t-test results strongly support the hypothesis that the DI condition elicits broader and more intense emotional responses. Facial expression analysis revealed a significant difference in non-dominant emotions, confirming that a more diverse range of emotions emerged over time. In the DI condition, all textual emotional items (TNEWs, NED, and NEIPS) significantly increased, demonstrating that the participants used more emotional language, with higher density and intensity, when given additional time. Moreover, the large effect sizes (d > 0.8 for all variables) confirm that these differences are substantial and meaningful, rather than being merely statistically significant.

5. Discussion

This study investigated how guided reflective re-viewing within a time-structured design mediates emotional reappraisal in time-based media art (TBMA) by comparing two course sessions scheduled three days apart. Specifically, guided reflective re-viewing increased the breadth and intensity of non-dominant emotions, while the dominant emotion maintained a high degree of consistency.

5.1. The Roles of Negative and Positive Appraisals in Dominant and Non-Dominant Emotional Dynamics

The consistency analysis in this study revealed an intriguing discrepancy: while the dominant facial expressions were concentrated in “neutral” and “sad”, which are typically associated with negative affect, the dominant emotions in the self-reported texts were predominantly positive, with the participants describing their experiences as “good”. A possible explanation for this phenomenon is that the selected TBMA projects, characterized by abstraction, a slow pace, and melancholic aesthetic qualities, may have elicited low-arousal negative facial expressions (sad, neutral) during the guided-prompt durations. However, during the reflective evaluation (self-reported texts), the participants may have recognized cognitive or aesthetic appreciation, leading them to self-report more positive emotions. Alternatively, the participants may have suppressed or masked their negative emotional expressions in self-reports to conform to perceived social norms. This pattern is consistent with findings from previous studies on emotion regulation strategies. Gross emphasized the costs of emotional suppression, particularly its tendency to reduce external expression while increasing internal physiological stress (Gross & Levenson, 1997; John & Gross, 2004). Goldin et al. further distinguished between cognitive reappraisal and behavioral suppression (Goldin et al., 2008). In contrast to these studies, the present study focused on the more naturalistic aesthetic experiences induced by TBMA without explicit emotion regulation instructions. The findings suggest that even in the absence of strong external pressure to suppress or reappraise emotions, the participants spontaneously shifted from initially subdued or negative facial expressions to more positive retrospective self-evaluations.
Another important finding of this study is that while the participants’ dominant emotional expressions (sad, neutral, good, and dislike) remained stable between the initial and delayed impressions, non-dominant emotional features, such as the number of emotional categories, emotional density, and emotional intensity, significantly increased after three days during the delayed impressions. This finding aligns with the conclusions summarized in the review by Trupp et al. (Trupp et al., 2025), which indicate that reflective and educational accessory engagements serve as positive factors shaping outcomes rather than mere viewing alone (Bell & Robbins, 2007; Drake et al., 2024; Fancourt et al., 2020). However, this study further supplements existing research by suggesting that emotional changes during artistic engagement cannot be fully explained by positive factors alone. Specifically, deeper aesthetic engagement may depend on the dynamic interplay between positive and non-positive emotions (good, neutral, or sad). From the perspective of the experimental design of this study, sustained engagement and reflection facilitate emotional integration rather than the immediate enhancement of emotion. This process relies on the coexistence of positive and negative emotions elicited by the stimuli, fostering a more authentic sense of self-awareness and psychological resilience development. The differences in non-dominant emotions observed in this study make it plausible that the negative aesthetic emotions experienced during the initial impression (sad, neutral, dislike) acted as emotional catalysts, stimulating deeper emotional exploration during the reflective process. Moreover, these negative aesthetic emotions did not diminish over time; instead, they appeared to stabilize the participants’ fundamental aesthetic orientation toward the artworks. Given prior research indicating that negative aesthetic emotions, such as sadness, can enhance critical thinking and cognitive engagement, and that individuals tend to allocate more cognitive resources to processing negative stimuli (Schaefer et al., 2013; Silvia, 2009; Silvia & Brown, 2007), the present findings empirically challenge the traditional view that only positive emotions foster emotional broadening. Instead, the results suggest that negative aesthetic emotions can serve as potent drivers of emotional complexity and cognitive elaboration over time.
Moreover, rather than being driven solely by either positive or negative appraisals, the emotional outcomes were also strongly influenced by viewing duration optimization. Previous studies have indicated that overly short or excessively long guided-prompt durations do not necessarily enhance emotional engagement or comprehension, as they may lead to cognitive fatigue, overprocessing, or emotional detachment (Brieber et al., 2020; Codispoti et al., 2009). Therefore, this study adopted the 17 s optimal presentation duration proposed by Brieber et al. to display the TBMA works under the DI condition (Brieber et al., 2020). However, the participants in this study differed significantly from those of Brieber et al., as all participants had received systematic training in art-related knowledge within an art school laboratory setting. This distinction highlights a pedagogical implication in which educators can tailor the length and pacing of guided reflection according to learners’ prior experience, allowing sufficient cognitive space for emotional reappraisal without inducing fatigue. In this way, reflective viewing time functions as a flexible pedagogical variable that can be adjusted to balance emotional immersion and critical distance in art appreciation.
The experimental results of the present study supplement the work of Contreras-Medina et al. (González-Zamar & Abad-Segura, 2021), whose analysis emphasized the spatial and disciplinary structures of art education, mapping how visual culture operates as a cultural and epistemological system. In contrast, the current study advances this discussion by introducing temporal and affective dimensions through the lens of TBMA. By examining the emotional dynamics that unfold during repeated viewing, this research redefines the educational ecosystem as a temporal ecology, one in which learning unfolds through iterative emotional reappraisal rather than through the static transmission of knowledge. While the analysis by Contreras-Medina et al. centered on the circulation of knowledge across artistic disciplines (Douglas & Coessens, 2012; Igdalova & Chamberlain, 2023; Schlegel et al., 2015), the present study emphasizes that affective circulation is an equally essential component of the educational ecosystem. The observed stability of dominant emotions and the diversification of non-dominant emotions across viewing stages illustrate an adaptive balance between cognitive regulation and emotional openness, a micro-level mechanism that reflects the macro-level adaptability emphasized in the literature on visual art ecosystems. Accordingly, the integration of structured re-viewing within the TBMA pedagogy proposed here exemplifies how future models of art education may translate macro-ecological goals, such as adaptability, creativity, and emotional literacy, into micro-level pedagogical practices. In this sense, the present study not only aligns with the educational ecosystem framework but also deepens it by shifting the focus from knowledge structures in visual arts toward affectively mediated learning processes in TBMA.

5.2. The Mere Exposure Effect and Conscious Reappraisal

Early studies on the Mere Exposure Effect (MEE) typically employed repeated presentations of simple visual stimuli, such as geometric figures, symbols, or faces, under tightly controlled laboratory conditions (Zajonc, 1968; Zajonc et al., 1974). These studies emphasized subliminal or unconscious exposure to index automatic affective preferences and showed that preference formation follows a non-linear function of exposure frequency that is consistent with associative learning and a curvilinear affective response, yet still operates through implicit associative mechanisms. Subsequent work by Monahan et al. extended this paradigm to examine affective diffusion (Monahan et al., 2000), demonstrating that repeated exposure can generate diffuse positive affect, even in the absence of awareness, without engaging conscious or reflective emotional processes. Zajonc further argued that MEE rests on noncognitive affective systems, reinforcing the view that the effect functions primarily at an unconscious level (Zajonc, 2001). Taken together, MEE research has conceptualized emotional change as a passive associative mechanism in which repetition increases perceptual fluency and thereby produces positive affect without conscious involvement.
By contrast, the present study relocates exposure within a conscious and reflective viewing context. The participants deliberately re-viewed the same TBMA video across a three-day interval. Rather than testing familiarity-based liking with simple stimuli, the design used complex content, together with temporal structuring and guided reflection, enabling emotional reappraisal through memory retention and cognitive engagement. Furthermore, this study departed from strictly controlled laboratory settings by embedding emotion measurement in a naturalistic educational classroom environment. Moreover, whereas MEE research highlights unconscious affective diffusion, the present study indicates a layered affective development in which dominant emotions remain stable while non-dominant emotions become richer and more differentiated through conscious reappraisal. This shift from unconscious priming to conscious affective reorganization illustrates how reflective re-viewing could transform mere exposure into a process of aesthetic and pedagogical emotional learning.

5.3. Emotion Recognition as a Pedagogical Tool in Art Education

In this study, emotion recognition technology not only provided a quantitative means of assessing aesthetic responses but also functioned as an instructional aid in art classrooms that was capable of sustaining attention and delivering timely feedback. Previous research has shown that emotion can modulate attention and learning (Kołakowska et al., 2014; Tonguç & Ozkara, 2020); therefore, visualizing affective states allows educators to adjust student engagement in real time. However, real-time modulation often requires higher hardware costs and more controlled classroom conditions (Nguyen et al., 2017). The auxiliary method proposed in this study offers a more practical approach for most art-teaching contexts while maintaining sufficient accuracy. Specifically, emotion measurements collected during the initial impression sessions can be aggregated to estimate the probability of core emotions over time and to flag small groups or the entire class when “attention decline” occurs (for instance, prolonged eye closure and absence of detectable facial expressions may indicate reduced attention). Educators can then incorporate a 60 s pause during the delayed-impression session for guided reflection, pose targeted prompts, or replay a short segment of the TBMA work. Prior studies on vision-based attention monitoring and classroom analytics have confirmed the feasibility of such micro-interventions.
Within the context of reflective art viewing, technology-assisted emotion tracking may serve as a metacognitive tool that supports reflective learning (Bartsch et al., 2008; S. Mohammad et al., 2016). Rather than restricting emotion recognition to evaluative assessment, its integration into instructional design enables learners to externalize and monitor their emotional states, thereby fostering higher-order reflection and self-regulation. In addition to the objective detection of emotions from facial expressions, the self-reported texts provided by students offered rich introspective supplements that can inform pedagogical interventions in art education. Self-narratives often reveal emotional fluctuations tied to specific moments or themes within the artwork. For example, a student might note, “When the frame lingered on the decaying egg, I felt both sadness and curiosity,” or “I hesitated here, feeling confused but also intrigued.” Such affective cues allow educators to design micro-discussions or prompts that help students articulate latent meanings, connect emotional reactions to the formal or conceptual aspects of artworks, and transition from a sensory response to critical interpretation.
Moreover, comparing students’ self-reported descriptions with the emotions detected from their facial expressions may encourage reflection on moments of discrepancy, such as asking, “Why did the facial expressions detect surprise when I actually felt calm in my heart?” Through this continuous comparative process, learners can develop broader emotional vocabulary, stronger self-evaluation skills, and a more intentional interpretive stance. Over time, this iterative cycle of emotional attention, comparison, and verbalization may cultivate emotional literacy, deeper engagement, and more nuanced aesthetic judgment, contributing to the sustained innovation and development of art education.

5.4. Limitations and Future Work

However, despite the valuable insights offered by this study into the relationship between time, emotion, and aesthetic experience, several limitations warrant further investigation. First, this study employed only two course settings (II and DI) and a relatively small sample size (N = 24), which constrains the generalizability of the findings. Additionally, the study examined only four TBMA-related video projects, which may not fully represent the diversity of styles, themes, and artistic intentions in time-based media. Finally, potential confounding variables such as individual differences in art familiarity, viewing motivation, and emotional expressivity may have influenced the observed results. Although the mixed-method design mitigated some biases by integrating both facial data and self-reported data, future research should adopt larger, more diverse participant groups and employ statistical controls, such as mixed linear modeling, to enhance robustness.
Future research will explore a broader range of TBMA works, including those with abstract, surreal, or controversial elements, to evaluate whether different artistic styles exhibit distinct emotional dynamics. Moreover, future studies will aim to extend the methodological framework developed in this research by further examining how the observed emotional changes, particularly the increased diversity and intensity of non-dominant emotions during delayed reflection, relate to students’ art understanding, interest, and liking. By combining quantitative emotion recognition with qualitative self-report analysis, subsequent experiments could track how different emotional patterns correspond to levels of aesthetic appreciation and conceptual comprehension in art education. For instance, a comparative design may explore whether heightened emotional variability predicts deeper understanding or sustained artistic interest over time. Furthermore, by expanding the scope of stimuli, refining guided-prompt durations, and incorporating additional cognitive and physiological measures, future studies can offer a more comprehensive understanding of the interplay between time, emotion, and aesthetic appreciation. Such investigations would ultimately clarify the pedagogical significance of emotion modulation in structured art-viewing experiences.

6. Conclusions

This study provides empirical evidence that a time-structured curricular design centered on guided reflective re-view with a three-day interval shapes students’ emotional experience of TBMA. Using a longitudinal comparative analysis that combines facial-expression recognition with text-based self-reports, this study shows that dominant emotions remain stable, while emotional diversity and intensity increase significantly during delayed reflection. These findings indicate that deeper cognitive engagement and emotional complexity develop dynamically through structured reflection.

Author Contributions

Conceptualization, H.F.; methodology, T.-Y.W.; software, H.F.; validation, T.-Y.W., T.Y. and S.H.; formal analysis, H.F.; investigation, H.F.; resources, T.Y. and S.H.; data curation, H.F. and S.H.; writing—original draft preparation, H.F.; writing—review and editing, T.-Y.W., T.Y.; supervision, T.-Y.W. and T.Y.; All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

This study involved human participants, and ethical review was performed by College of Fine and Design of Shenyang Normal University, No approval code was issued due to the regulations of the ethics committee, 2 November 2022.

Informed Consent Statement

Informed consent was obtained from all subjects involved in this study.

Data Availability Statement

The data are available from the corresponding authors upon reasonable request.

Acknowledgments

The first author would like to express sincere gratitude to Ziyang Lv (Shenyang Normal University), Kazunori Miyata (Japan Advanced Institute of Science and Technology) and Jing Wang (Dalian Polytechnic University) for their valuable support and guidance.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Aksu, G., Güzeller, C. O., & Eser, M. T. (2019). The effect of the normalization method used in different sample sizes on the success of artificial neural network model. International Journal of Assessment Tools in Education, 6(2), 170–192. [Google Scholar] [CrossRef]
  2. Altman, D. G., & Bland, J. M. (1995). Statistics notes: The normal distribution. BMJ, 310(6975), 298. [Google Scholar] [CrossRef]
  3. Alves-Noreña, A. C., Rodríguez-Conde, M.-J., Hernández-Ramos, J. P., & Castro-Salgado, J. W. (2025). Technology and emotions: AI-driven software prototyping for the analysis of emotional states and early detection of risky behaviors in university students. Education Sciences, 15(3), 350. [Google Scholar] [CrossRef]
  4. Bartsch, A., Vorderer, P., Mangold, R., & Viehoff, R. (2008). Appraisal of emotions in media use: Toward a process model of meta-emotion and emotion regulation. Media Psychology, 11(1), 7–27. [Google Scholar] [CrossRef]
  5. Bell, C. E., & Robbins, S. J. (2007). Effect of art production on negative mood: A randomized, controlled trial. Art Therapy, 24(2), 71–75. [Google Scholar] [CrossRef]
  6. Ben, X., Ren, Y., Zhang, J., Wang, S.-J., Kpalma, K., Meng, W., & Liu, Y.-J. (2021). Video-based facial micro-expression analysis: A survey of datasets, features and algorithms. IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(9), 5826–5846. [Google Scholar] [CrossRef]
  7. Blair, R. C., & Higgins, J. J. (1985). Comparison of the power of the paired samples t test to that of Wilcoxon’s signed-ranks test under various population shapes. Psychological Bulletin, 97(1), 119. [Google Scholar] [CrossRef]
  8. Brieber, D., Forster, M., & Leder, H. (2020). On the mutual relation between art experience and viewing time. Psychology of Aesthetics, Creativity, and the Arts, 14(2), 197. [Google Scholar] [CrossRef]
  9. Chierico, A. (2016). Medium specificity in post-media practice. V! RUS, 1(12), 1–11. [Google Scholar]
  10. Codispoti, M., Mazzetti, M., & Bradley, M. M. (2009). Unmasking emotion: Exposure duration and emotional engagement. Psychophysiology, 46(4), 731–738. [Google Scholar] [CrossRef] [PubMed]
  11. Conway, M. A. (2009). Episodic memories. Neuropsychologia, 47(11), 2305–2313. [Google Scholar] [CrossRef]
  12. Cowan, N., & AuBuchon, A. M. (2008). Short-term memory loss over time without retroactive stimulus interference. Psychonomic Bulletin & Review, 15(1), 230–235. [Google Scholar] [CrossRef]
  13. Deng, X., & Nan, P. (2022). cntext: A Python tool for text mining (Version 1.7. 9). Available online: https://github.com/hiDaDeng/cntext (accessed on 2 August 2025).
  14. Douglas, A., & Coessens, K. (2012). Experiential knowledge and improvisation: Variations on movement, motion, emotion. Art, Design & Communication in Higher Education, 10(2), 179–198. [Google Scholar] [CrossRef] [PubMed]
  15. Drake, J. E., Eizayaga, M., & Wawrzynski, S. (2024). Making versus viewing art: Effects on affect, enjoyment, and flow. Empirical Studies of the Arts, 42(1), 147–165. [Google Scholar] [CrossRef]
  16. Elleström, L. (2010). The modalities of media: A model for understanding intermedial relations. In Media borders, multimodality and intermediality (pp. 11–48). Springer. [Google Scholar]
  17. Fancourt, D., Garnett, C., & Müllensiefen, D. (2020). The relationship between demographics, behavioral and experiential engagement factors, and the use of artistic creative activities to regulate emotions. Psychology of Aesthetics, Creativity, and the Arts, 19(3), 439–449. [Google Scholar] [CrossRef]
  18. Franěk, M., Petružálek, J., & Šefara, D. (2022). Facial expressions and self-reported emotions when viewing nature images. International Journal of Environmental Research and Public Health, 19(17), 10588. [Google Scholar] [CrossRef]
  19. Fritz, C. O., Morris, P. E., & Richler, J. J. (2012). Effect size estimates: Current use, calculations, and interpretation. Journal of Experimental Psychology: General, 141(1), 2. [Google Scholar] [CrossRef] [PubMed]
  20. García-Hernández, R. A., Luna-García, H., Celaya-Padilla, J. M., García-Hernández, A., Reveles-Gómez, L. C., Flores-Chaires, L. A., Delgado-Contreras, J. R., Rondon, D., & Villalba-Condori, K. O. (2024). A systematic literature review of modalities, trends, and limitations in emotion recognition, affective computing, and sentiment analysis. Applied Sciences, 14(16), 7165. [Google Scholar] [CrossRef]
  21. Goldin, P. R., McRae, K., Ramel, W., & Gross, J. J. (2008). The neural bases of emotion regulation: Reappraisal and suppression of negative emotion. Biological Psychiatry, 63(6), 577–586. [Google Scholar] [CrossRef]
  22. González-Zamar, M.-D., & Abad-Segura, E. (2021). Visual arts in the university educational ecosystem: Analysis of schools of knowledge. Education Sciences, 11(4), 184. [Google Scholar] [CrossRef]
  23. Gross, J. J., & Levenson, R. W. (1997). Hiding feelings: The acute effects of inhibiting negative and positive emotion. Journal of Abnormal Psychology, 106(1), 95. [Google Scholar] [CrossRef]
  24. Gupta, S., Kumar, P., & Tekchandani, R. K. (2023). Facial emotion recognition based real-time learner engagement detection system in online learning context using deep learning models. Multimedia Tools and Applications, 82(8), 11365–11394. [Google Scholar] [CrossRef]
  25. He, K., Zhang, X., Ren, S., & Sun, J. (2016, June 27–30). Deep residual learning for image recognition. IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA. [Google Scholar]
  26. Höfel, L., & Jacobsen, T. (2003). Temporal stability and consistency of aesthetic judgments of beauty of formal graphic patterns. Perceptual and Motor Skills, 96(1), 30–32. [Google Scholar] [CrossRef]
  27. Igdalova, A., & Chamberlain, R. (2023). Slow looking at still art: The effect of manipulating audio context and image category on mood and engagement during an online slow looking exercise. Psychology of Aesthetics, Creativity, and the Arts, 19(3), 522–534. [Google Scholar] [CrossRef]
  28. John, O. P., & Gross, J. J. (2004). Healthy and unhealthy emotion regulation: Personality processes, individual differences, and life span development. Journal of Personality, 72(6), 1301–1334. [Google Scholar] [CrossRef] [PubMed]
  29. Judd, C. M., Westfall, J., & Kenny, D. A. (2017). Experiments with more than one random factor: Designs, analytic models, and statistical power. Annual Review of Psychology, 68(1), 601–625. [Google Scholar] [CrossRef]
  30. Keltner, D., & Haidt, J. (2003). Approaching awe, a moral, spiritual, and aesthetic emotion. Cognition and Emotion, 17(2), 297–314. [Google Scholar] [CrossRef] [PubMed]
  31. Kołakowska, A., Landowska, A., Szwoch, M., Szwoch, W., & Wrobel, M. R. (2014). Emotion recognition and its applications. In Human-computer systems interaction: Backgrounds and applications 3 (pp. 51–62). Springer. [Google Scholar]
  32. Laurenson, P. (2013). Emerging institutional models and notions of expertise for the conservation of time-based media works of art. Technè. La Science au Service de l’Histoire de l’Art et de la Préservation des Biens Culturels, 37, 36–42. [Google Scholar]
  33. Leder, H., & Nadal, M. (2014). Ten years of a model of aesthetic appreciation and aesthetic judgments: The aesthetic episode–Developments and challenges in empirical aesthetics. British Journal of Psychology, 105(4), 443–464. [Google Scholar] [CrossRef]
  34. Li, J., Xu, Y., Xiong, H., & Wang, Y. (2010, August 16–17). Chinese text emotion classification based on emotion dictionary. 2010 IEEE 2nd Symposium on Web Society, Beijing, China. [Google Scholar]
  35. Liao, C. (2016). From interdisciplinary to transdisciplinary: An arts-integrated approach to STEAM education. ART Education, 69(6), 44–49. [Google Scholar] [CrossRef]
  36. Liu, B. (2022). Sentiment analysis and opinion mining. Springer Nature. [Google Scholar]
  37. Liu, Q., & Wang, L. (2021). t-Test and ANOVA for data with ceiling and/or floor effects. Behavior Research Methods, 53(1), 264–277. [Google Scholar] [CrossRef]
  38. Menninghaus, W., Schindler, I., Wagner, V., Wassiliwizky, E., Hanich, J., Jacobsen, T., & Koelsch, S. (2020). Aesthetic emotions are a key factor in aesthetic evaluation: Reply to Skov and Nadal (2020). Psychological Review, 127(4), 650–654. [Google Scholar] [CrossRef]
  39. Mohammad, S., Shutova, E., & Turney, P. (2016, August 11–12). Metaphor as a medium for emotion: An empirical study. Fifth Joint Conference on Lexical and Computational Semantics, Berlin, Germany. [Google Scholar]
  40. Mohammad, S. M. (2021). Sentiment analysis: Automatically detecting valence, emotions, and other affectual states from text. In Emotion measurement (pp. 323–379). Elsevier. [Google Scholar]
  41. Monahan, J. L., Murphy, S. T., & Zajonc, R. B. (2000). Subliminal mere exposure: Specific, general, and diffuse effects. Psychological Science, 11(6), 462–466. [Google Scholar] [CrossRef]
  42. Nguyen, B. T., Trinh, M. H., Phan, T. V., & Nguyen, H. D. (2017, April 16–19). An efficient real-time emotion detection using camera and facial landmarks. 2017 Seventh International Conference on Information Science and Technology (ICIST), Da Nang, Vietnam. [Google Scholar]
  43. Pantic, M., & Rothkrantz, L. J. M. (2002). Automatic analysis of facial expressions: The state of the art. IEEE Transactions on Pattern Analysis and Machine Intelligence, 22(12), 1424–1445. [Google Scholar] [CrossRef]
  44. Pelowski, M., & Akiba, F. (2011). A model of art perception, evaluation and emotion in transformative aesthetic experience. New Ideas in Psychology, 29(2), 80–97. [Google Scholar] [CrossRef]
  45. Pelowski, M., Markey, P. S., Forster, M., Gerger, G., & Leder, H. (2017). Move me, astonish me… delight my eyes and brain: The Vienna Integrated Model of top-down and bottom-up processes in Art Perception (VIMAP) and corresponding affective, evaluative, and neurophysiological correlates. Physics of Life Reviews, 21, 80–125. [Google Scholar] [CrossRef]
  46. Peppler, K. A. (2010). Media arts: Arts education for a digital age. Teachers College Record, 112(8), 2118–2153. [Google Scholar] [CrossRef]
  47. Phillips, J. (2012). Shifting equipment significance in time-based media art. Electronic Media Review, 1, 139–154. [Google Scholar]
  48. Poria, S., Cambria, E., Bajpai, R., & Hussain, A. (2017). A review of affective computing: From unimodal analysis to multimodal fusion. Information Fusion, 37, 98–125. [Google Scholar] [CrossRef]
  49. Potts, A. (2004). Tactility: The interrogation of medium in art of the 1960s. Art History, 27(2), 282–304. [Google Scholar] [CrossRef]
  50. Quaranta, D. (2013). Beyond new media art. Link Editions Brescia. [Google Scholar]
  51. Rafique, H., Liu, M., Lin, Q., & Yang, T. (2022). Weakly-convex–concave min–max optimization: Provable algorithms and applications in machine learning. Optimization Methods and Software, 37(3), 1087–1121. [Google Scholar] [CrossRef]
  52. Rancière, J. (2011). What medium can mean. Parrhesia, 11, 35–43. [Google Scholar]
  53. Ravaja, N., Kallinen, K., Saari, T., & Keltikangas-Jarvinen, L. (2004). Suboptimal exposure to facial expressions when viewing video messages from a small screen: Effects on emotion, attention, and memory. Journal of Experimental Psychology: Applied, 10(2), 120. [Google Scholar] [CrossRef]
  54. Rinehart, R., & Ippolito, J. (2022). Re-collection: Art, new media, and social memory. MIT Press. [Google Scholar]
  55. Schaefer, S. M., Morozink Boylan, J., Van Reekum, C. M., Lapate, R. C., Norris, C. J., Ryff, C. D., & Davidson, R. J. (2013). Purpose in life predicts better emotional recovery from negative stimuli. PLoS ONE, 8(11), e80329. [Google Scholar] [CrossRef]
  56. Schlegel, A., Alexander, P., Fogelson, S. V., Li, X., Lu, Z., Kohler, P. J., Riley, E., Tse, P. U., & Meng, M. (2015). The artist emerges: Visual art learning alters neural structure and function. NeuroImage, 105, 440–451. [Google Scholar] [CrossRef]
  57. Semeniuta, S., Severyn, A., & Barth, E. (2016). Recurrent dropout without memory loss. arXiv, arXiv:1603.05118. [Google Scholar] [CrossRef]
  58. Silvia, P. J. (2005). Emotional responses to art: From collation and arousal to cognition and emotion. Review of General Psychology, 9(4), 342–357. [Google Scholar] [CrossRef]
  59. Silvia, P. J. (2009). Looking past pleasure: Anger, confusion, disgust, pride, surprise, and other unusual aesthetic emotions. Psychology of Aesthetics, Creativity, and the Arts, 3(1), 48. [Google Scholar] [CrossRef]
  60. Silvia, P. J., & Brown, E. M. (2007). Anger, disgust, and the negative aesthetic emotions: Expanding an appraisal model of aesthetic experience. Psychology of Aesthetics, Creativity, and the Arts, 1(2), 100. [Google Scholar] [CrossRef]
  61. Sim, J., & Wright, C. C. (2005). The kappa statistic in reliability studies: Use, interpretation, and sample size requirements. Physical Therapy, 85(3), 257–268. [Google Scholar] [CrossRef]
  62. Spielmann, Y. (2012). Hybrid culture: Japanese media arts in dialogue with the West. Mit Press. [Google Scholar]
  63. Talarico, J. M., & Rubin, D. C. (2003). Confidence, not consistency, characterizes flashbulb memories. Psychological Science, 14(5), 455–461. [Google Scholar] [CrossRef]
  64. Tonguç, G., & Ozkara, B. O. (2020). Automatic recognition of student emotions from facial expressions during a lecture. Computers & Education, 148, 103797. [Google Scholar] [CrossRef]
  65. Trupp, M. D., Howlin, C., Fekete, A., Kutsche, J., Fingerhut, J., & Pelowski, M. (2025). The impact of viewing art on well-being—A systematic review of the evidence base and suggested mechanisms. The Journal of Positive Psychology, 1–25. [Google Scholar] [CrossRef]
  66. Vassos, S., Malliaraki, E., Falco, F. d., Di Maggio, J., Massimetti, M., Nocentini, M. G., & Testa, A. (2016, November 15–18). Art-bots: Toward chat-based conversational experiences in museums. International Conference on Interactive Digital Storytelling, Los Angeles, CA, USA. [Google Scholar]
  67. Wallbott, H. G., & Scherer, K. R. (1989). Assessing emotion by questionnaire. In The measurement of emotions (pp. 55–82). Elsevier. [Google Scholar]
  68. Walton, K. L. (1993). How marvelous! Toward a theory of aesthetic value. The Journal of Aesthetics and Art Criticism, 51(3), 499–510. [Google Scholar] [CrossRef]
  69. Wang, M., Shen, R., Novak, D., & Pan, X. (2009). The impact of mobile learning on students’ learning behaviours and performance: Report from a large blended classroom. British Journal of Educational Technology, 40(4), 673–695. [Google Scholar] [CrossRef]
  70. Yadollahi, A., Shahraki, A. G., & Zaiane, O. R. (2017). Current state of text sentiment analysis from opinion to emotion mining. ACM Computing Surveys (CSUR), 50(2), 1–33. [Google Scholar] [CrossRef]
  71. Yuvaraj, R., Mittal, R., Prince, A. A., & Huang, J. S. (2025). Affective computing for learning in education: A systematic review and bibliometric analysis. Education Sciences, 15(1), 65. [Google Scholar] [CrossRef]
  72. Zajonc, R. B. (1968). Attitudinal effects of mere exposure. Journal of Personality and Social Psychology, 9(2p2), 1. [Google Scholar] [CrossRef]
  73. Zajonc, R. B. (2001). Mere exposure: A gateway to the subliminal. Current Directions in Psychological Science, 10(6), 224–228. [Google Scholar] [CrossRef]
  74. Zajonc, R. B., Markus, H., & Wilson, W. R. (1974). Exposure effects and associative learning. Journal of Experimental Social Psychology, 10(3), 248–263. [Google Scholar] [CrossRef]
Education 15 01354 g001
Figure 1. Plan view of the classroom setup used in both sessions.
Figure 1. Plan view of the classroom setup used in both sessions.
Education 15 01354 g001
Table 1. Representative institutions and course offerings in time-based media art across the U.S., Japan, and China.
Table 1. Representative institutions and course offerings in time-based media art across the U.S., Japan, and China.
University and InstitutionProgram Overview and Curriculum Focus
New York UniversityProgram: Conservation of Time-Based Media Art
Survey of time-based media technologies and care; advanced training
Interdisciplinary: CS, engineering, and film and video preservation
Method: Collaboration with specialists and professional networks
University of PennsylvaniaProgram: Time-Based and Interactive Media
Focus: Moving image, digital technologies, and interactivity
Outcome: Engage with emerging tools/methods and integrate into art/design practice
Sam Fox School of
Design and Visual Arts		Program: Time-Based Media Art
Focus: Time as the primary expressive medium
Courses: Expanded cinema, sound environments, and media performance
Emphasis: digital, electronic, acoustic, cinematic, and performative approaches
Maricopa Community CollegesProgram: Time-Based Media
Focus: Production techniques, compositing, motion graphics, and interactive design
Competencies: Media and photographic technologies, production equipment, and industry practices
University of Tennessee, KnoxvilleProgram: Time-Based art
Focus: Animation, video, film, performance, installation, sound art, and interactive time-based media
Kyoto Seika UniversityPrograms: Video and Media Art
Focus: Artistic animation, short films, media arts, music videos, commercials, video installations, and projection mapping
Emphasis: Creative expression and communication
Tama Art UniversityPrograms: Art and Media
Focus: Innovative expression in digital art and creative practice, video and animation production, programming, and 3D modeling
Emphasis: Digital fabrication
Communication University of ChinaProgram: Digital Media Art
Focus: Ontological theory and industrial applications, intelligent media, interactive art, motion graphics, virtual societies, and digital entertainment
Emphasis: A global perspective and a deep grounding in Chinese cultural heritage
Beijing Institute of Graphic CommunicationProgram: Digital Media Art
Focus: Appreciation of digital media art, appreciation of video and animation art, and introduction to film aesthetics
Table 2. Procedures used for initial impression and delayed impression sessions.
Table 2. Procedures used for initial impression and delayed impression sessions.
StageInitial ImpressionDelayed Impression
Stage 110 min—Explained informed consent and experimental details; calibrated and tested recording equipment.5 min—Explained experimental details; calibrated and tested recording equipment.
Stage 2Four groups viewed different TBMA videos projected on screen; facial expressions were recorded.Four groups viewed still frames extracted from previously shown videos; facial expressions were recorded.
Stage 330 s break.2 min break.
Stage 420 min—Completed self-report questionnaire.25 min—Completed self-report questionnaire.
After the experiment5 min—Checked for any adverse reactions; gathered brief feedback.10 min—Checked for any adverse reactions; explained creators’ intent; collected participant feedback.
Table 3. Demographic information of participants.
Table 3. Demographic information of participants.
II\DINPercent (%)Cumulative Percent (%)
Gender
Male520.820.8
Female1979.2100
Duration of art learning
3–5 years1458.358.3
5–10 years937.595.8
More than 10 years14.2100
Table 4. Information and links for videos viewed by each group.
Table 4. Information and links for videos viewed by each group.
GroupVideo Information
G 1Bill Viola, Ascension
G 2Bill Viola, The Raft
G 3Steve Reich, Music for Pieces of Wood- Visualization
G 4Steve Reich, Pendulum Music 1968
Table 5. Crosstabulation of facial impressions between initial and delayed impressions.
Table 5. Crosstabulation of facial impressions between initial and delayed impressions.
II\DINeutralSadTotal
Neutral15015
Sad099
Total15924
Note: II—initial impression; DI—delayed impression.
Table 6. Crosstabulation of self-reported texts between initial and delayed impressions.
Table 6. Crosstabulation of self-reported texts between initial and delayed impressions.
II\DIDislikeGoodTotal
Dislike314
Good11920
Total42024
Note: II—initial impression; DI—delayed impression.
Table 7. Consistency of dominant emotions between initial and delayed impressions.
Table 7. Consistency of dominant emotions between initial and delayed impressions.
ItemKappapSig.
Facial Expression1.000<0.001***
Self-report Text0.7000.001**
Note: *** p < 0.001, ** p < 0.01.
Table 8. One-way ANOVA results for differences in initial and delayed impressions across groups.
Table 8. One-way ANOVA results for differences in initial and delayed impressions across groups.
Item Mean (DI Minus II)
Group1Group2Group3Group4fpr
FECs1.831.332.002.00(3, 10) = 1.020.4220.286
TECs2.332.172.002.33(3, 11) = 0.100.9590.138
NFEs0.230.210.170.20(3, 10) = 0.410.7530.259
TNEWs6.508.004.835.83(3, 11) = 1.540.2590.465
NED0.020.030.030.02(3, 10) = 0.710.568<0.1
NEIPS0.590.710.570.42(3, 11) = 1.690.2270.409
Note: II—initial impression; DI—delayed impression; FECs—facial emotion categories; TECs—textual emotion categories; NFEs—non-dominant facial emotions; TNEWs—total non-dominant emotion words; NED—non-dominant emotion density; NEIPS—non-dominant emotion intensity per sentence. r = √η2.
Table 9. Comparison of paired-sample t-test between initial and delayed impressions.
Table 9. Comparison of paired-sample t-test between initial and delayed impressions.
ItemIIDI
MeanS.D.MeanS.D.t23pdSig.
FECs4.420.936.210.72−8.98<0.0011.834***
TECs2.630.974.830.91−10.60<0.0012.163***
NFEs0.200.120.410.12−10.59<0.0012.173***
TNEWs3.632.869.924.80−12.19<0.0012.489***
NED0.02<0.010.040.01−9.83<0.0012.006***
NEIPS0.440.311.010.41−11.06<0.0012.260***
Note: II—initial impression; DI—delayed impression; FECs—facial emotion categories; TECs—textual emotion categories; NFEs—non-dominant facial emotions; TNEWs—total non-dominant emotion words; NED—non-dominant emotion density; NEIPS—non-dominant emotion intensity per sentence. d—Cohen’s d, *** p < 0.001.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Feng, H.; Wang, T.-Y.; Yuizono, T.; Huang, S. Re-Viewing the Same Artwork with Emotional Reappraisal: An Undergraduate Classroom Study in Time-Based Media Art Education. Educ. Sci. 2025, 15, 1354. https://doi.org/10.3390/educsci15101354

AMA Style

Feng H, Wang T-Y, Yuizono T, Huang S. Re-Viewing the Same Artwork with Emotional Reappraisal: An Undergraduate Classroom Study in Time-Based Media Art Education. Education Sciences. 2025; 15(10):1354. https://doi.org/10.3390/educsci15101354

Chicago/Turabian Style

Feng, Haocheng, Tzu-Yang Wang, Takaya Yuizono, and Shan Huang. 2025. "Re-Viewing the Same Artwork with Emotional Reappraisal: An Undergraduate Classroom Study in Time-Based Media Art Education" Education Sciences 15, no. 10: 1354. https://doi.org/10.3390/educsci15101354

APA Style

Feng, H., Wang, T.-Y., Yuizono, T., & Huang, S. (2025). Re-Viewing the Same Artwork with Emotional Reappraisal: An Undergraduate Classroom Study in Time-Based Media Art Education. Education Sciences, 15(10), 1354. https://doi.org/10.3390/educsci15101354

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop