Introduction

Korean pop music, synonymously called K-Pop, has risen in popularity among the West in the past decade.

Even though groups consist of several members, going as high as 24, fans will often pick out their favorite member, a “Bias”. Selecting a bias within a group has become a cultural phenomenon on its own, shaping sub-fandoms within the larger construct of the group fandom. Visual representations play a central role in shaping fan preferences, particularly in the selection and formation of favorite group members. The fascination for certain artists within a group can often be enhanced by subtle visual cues embedded in music videos, promotional materials, and live performances. 

The study, „Perception and Bias Formation in the Korean Music Industry,“ utilizes eye-tracking technology to investigate how visualelements influence fan preferences and contribute to the formation of favorite group members (“Bias”).

Research Questions

The following questions were researched during the study:

Does the frequency of an artist’s appearance in a music video correlate with increased viewer engagement and gaze fixation?

  • This question seeks to determine if there's a direct relationship between how often an artist is seen and the amount of attention they capture from the audience.

In what ways do the styling choices (costumes, makeup, and hairstyles) of K-Pop artists affect the viewer’s gaze distribution across different group members?

  • This question examines whether the distinct visual styles of artists within the same group lead to differential attention from viewers.

How does the background activity and visual complexity of a music video scene impact attention distribution among K-Pop artists?

  • This explores whether visually busy backgrounds distract from or enhance viewer focus on the main artists.

Do member roles matter when it comes to the gaze behavior?

  • This question aims to identify if the gaze behavior is impacted by the assigned roles of the members (e.g. Visual, Main Dancer, Leader etc.)

Methodology

This study relies on the usage of a remote eye tracking device and the functions of its recording and analytics softwares.

For our research, we use the Tobii Pro Nano for eye tracking and Tobii Pro Labs for the analysis and processing of the eye tracking data.

Additionally, we conducted our tests with the talking out loud method, allowing our test subjects to speak about their thoughts as they watch the provided content.

Test subjects

The study included a total of seven participants, consisting of five males and two females, with an average age of 24 years old. Among the participants, 57% regularly listen to K-Pop. However, these regular listeners do not identify with any particular fandom and have not developed a specific K-Pop bias. The remaining 43% of the participants do not listen to K-Pop at all.

Results

Key Findings

The study revealed several key findings. Firstly, participants demonstrated a strong preference for focusing on faces. Throughout the study, their visual attention was consistently drawn to the faces of the K-Pop group members. Secondly, the formation of biases, or the selection of favorite members, was significantly influenced by the faces, facial expressions, and overall aura of the individuals. This indicates that the visual and emotive aspects of the members played a crucial role in how participants made their choices.

In addition to facial features, the outfits worn by the K-Pop group members also significantly impacted bias formation. The style and appearance of the members' clothing contributed to how participants perceived and selected their favorites.  Interestingly, the participants reported that their bias formation was not influenced by the specific roles of the members within the group, such as being the visual or the leader. This suggests that visual appeal and personal expressions were more critical factors than the predefined roles within the group.

Lastly, the study found that participants had limited success in consistently recognizing the group members. While some members were more easily recognizable than others, overall recognition was only partially achieved, indicating challenges in differentiating and recalling individual members.

Detailled Findings

Bias Forming

Momo was the most frequently mentioned person in the ratings, with 25% of participants selecting her. She was closely followed by Dahyun, who garnered 18% of the mentions. Both Momo and Dahyun stood out to the participants due to their striking outfits in the videos „Set me Free“ and „Moonlight Sunrise.“ In contrast, Chaeyoung was not mentioned as a bias by any of the participants.

Jeongyeon was looked at the most during the discussions, with an average fixation duration of 29 seconds on her images. Among all the members, she was viewed the longest when participants were looking at the photos. During the discussion phase, she was particularly noticeable due to her easily recognizable short hair.

Despite being observed the longest, Jeongyeon was, on average, the least favored member among the participants.

Background

When the background is monotonous, participants' focus remains horizontally level, and they do not look around. An example of this is in choreography videos. In contrast, when B-roll footage or zoomed-out shots are used, there is a tendency for participants to look at the background more frequently. The farther the shot, the more likely the focus is on the background. Additionally, the more playful or intricate the background, the more positive the impression on the person, although this requires further verification.

Editing

In static videos, such as choreography videos, participants' gaze is more spread out. They do not focus on the center or the person but rather look around. Generally, rapid cuts made it more difficult for participants to distinguish between the members.

Outfits

Outfits are noticed either peripherally or in half-shots of the person. Participants took short glances at the outfits without long fixations. Certain elements of the outfits stood out more, such as bushy sleeves or jackets.  Jeongyeon outfits were looked at more frequently. Subjects explained this behaviour by claiming that her outfits stood out from the rest of the group as they were less styled or plain ugly in their opinions. 

Face

The focus on the face was primarily on the eyes, followed by the nose and then the mouth. Hair was also mentioned as a recognition feature since it frames the face. Special makeup, such as white eyeliner, helped individuals stand out from the crowd. If participants could clearly distinguish someone from the group, they were more likely to focus on or look at that person for a longer duration, although this needs further verification.

Dance

Three out of seven participants (43%) mentioned dance skills as a factor in evaluating their bias. Bias formation was not strongly influenced or was not influenced at all by center time or center position, as exemplified by Nayeon who is the center of the group.

Interpretation

The study concludes that highly cluttered backgrounds create a significant cognitive load for viewers. The more inviting or well-presented a person appears, the more likely it is that viewers will develop a bias towards them. For instance, if a person is frequently shown smiling, dressed naturally, and with a distinctive feature (such as a standout accessory), they are more likely to be noticed.

There is a delicate balance between negative and positive dissonance. While a gray outfit in a crowd of black outfits can positively impact a person by making them stand out, a reserved and unremarkable outfit can have a negative effect. Core moments, where a person is in the spotlight, can also positively influence bias formation, benefiting rappers or performers of a solo.

Additionally, natural close-up shots, such as ending fairies, also promote a positive bias, although they are not decisive on their own.

Conclusion

Overall, eye tracking provides many measurable values for researching bias formation, from which actions can be derived. However, qualitative insights offer deeper information and observations. For example, the quantitative results showed that Jeongyeon was the most viewed person during the discussions. In contrast, the qualitative results revealed that she was the least popular member in the study. Because we combined the thinking out loud method with eye tracking, there could be extra fixations that manipulated the eye-tracking data, which had to be considered during the analysis.

Report//Presentation