- Title
- Affective and cognitive variations account for sex differences in the perception of facial avatar realism and uncanniness
- Creator
- Bailey, Jacqueline Deanna
- Relation
- University of Newcastle Research Higher Degree Thesis
- Resource Type
- thesis
- Date
- 2021
- Description
- Research Doctorate - Doctor of Philosophy (PhD)
- Description
- Avatars (virtual humans) are used to engage with end-users across many areas, including entertainment media, digital games, serious games, and simulation training. But, regardless of the application, a common issue for avatars is how the end-users perceive them. Previous research has widely explored some factors relating to avatar characteristics, including age, ethnicity, and physical appearance. Of particular interest in this work is the role gender(sex) plays in end-user perceptions. In this research, the term gender(sex) is taken from existing literature and refers to the socially constructed perception of gender whereby gender identification, expression and performance may not align with the biological sex of a person. There is limited existing work that focuses solely on gender(sex)-based perceptions in human-avatar interaction (HAI). Therefore, this thesis intends to examine how the gender(sex) of end-users and the perceived gender(sex) of avatars affect the perception of avatars during these interactions. This research will contribute to a broader understanding of genders(sexes) influence in the design and perception of facial avatars for developers and end-users in HAI. This will be achieved by building on the existing literature and developing a conceptual framework from systematic and narrative reviews relevant to the thesis aims. Key themes from these reviews indicate that various cosmetic and functional attributes contribute to the observed variability in avatar perception. End-user perception may be influenced by the avatar's appearance and behaviour, including the avatars’ perceived gender(sex). These perceptions may also be affected by factors related to the development of the avatar. These factors may also impact the level of perceived realism and can lead to a sense of uncanniness. There are also many individual user characteristics that may affect human factors of HAI, such as the user's gender(sex). Two rounds of data collection were conducted using a between-subjects experimental design to investigate the potential impact of gender(sex) in HAI related to both the avatar and the user. A series of core lab-based experiments examined participant affective states and considered cognitive abilities related to spatial awareness, affective and cognitive empathetic abilities. These cognitive abilities were chosen as the literature suggests that these two abilities have known gender(sex)-based differences. In these experiments, participants (n= 92) viewed a set of emotionally expressive facial avatars while their affective state was recorded via facial electromyography (f-EMG) signals. Participants also rated each avatars’ level of humanness, eeriness, and attractiveness in a series of Godspeed surveys. To validate the outcomes, a further (n=2065) participants completed a secondary study survey that asked them to rank the avatars in terms of perceived realism and uncanniness levels. A key finding from the secondary study showed that in terms of realism perceptions for the mid-low realism avatars, the male avatars are ranked as more realistic than the female avatars. However, this trend was reversed for the real humans and high realism avatars, suggesting a relationship between gender(sex)-based perceptions of avatars and their visual realism. The key findings from this research demonstrate gender(sex) differences in perceptions of avatars. The analysis of the f-EMG responses indicates that the gender(sex) of both the avatar and the participant impacts the perception of facial avatars. An additional method and analysis using novel categorisation of f-EMG responses to account for individual variability was developed. This new method allows objective determination of an end-users perception of an avatar as either pleasant, neutral, or unpleasant. The analysis of these categorised responses suggests that while statistically significant results are present, the results may not be generalisable due to limitations discussed in this work. However, these results do warrant further investigation in future work. The research also identified that perceptions of avatar eeriness and attractiveness are associated with participant gender(sex)-differences, but humanness is not. Further, this research also considered whether spatial and empathetic abilities, which are known to vary by gender(sex), influence perceptions of facial avatars. From the research, spatial ability was not a reliable predictor of perceptions of facial avatars and does not account for gender(sex)-based differences in the perception of pleasantness, humanness, eeriness, and attractiveness in avatars. Similarly, empathetic abilities were not strong indicators of end-user perceptions in human avatar interactions. These results point to more socially constructed factors impacting human perceptions of facial avatars rather than inherent cognitive abilities that vary between genders(sexes). The research contributes to the broader understanding of the nature of gender(sex)-based differences in human perceptions of avatar faces. The conceptual model created forms a framework for identifying the factors that potentially influence these perceptions. The research confirms that end-users and avatars' gender(sex) will affect how facial avatars are perceived. In particular for pleasant, unpleasant, and neutral perceptions of avatars using a novel approach not present in the current literature to contribute to the field of human-avatar interactions. Additionally, this thesis examines spatial and empathetic abilities as potential sources of gender(sex) variations in the perceptions of avatars that are a unique contribution to the field of human-avatar interaction.
- Subject
- avatars; virtual humans; facial realism; gender-based differences
- Identifier
- http://hdl.handle.net/1959.13/1506220
- Identifier
- uon:55831
- Rights
- Copyright 2021 Jacqueline Deanna Bailey
- Language
- eng
- Full Text
- Hits: 131
- Visitors: 149
- Downloads: 28
| Thumbnail | File | Description | Size | Format | |||
|---|---|---|---|---|---|---|---|
| View Details Download | ATTACHMENT01 | Thesis | 8 MB | Adobe Acrobat PDF | View Details Download | ||
| View Details Download | ATTACHMENT02 | Abstract | 335 KB | Adobe Acrobat PDF | View Details Download |