Toward Understanding Real-World Social Impacts of Avatars
Many digital narratives feature avatars onto which we project our agency, aspirations, and biases – consciously and unconsciously. This paper presents two projects towards understanding why we construct the avatars that we do and how these avatars impact us. The upshot is that electronic literature authors should take constructing avatars in digital systems seriously since they can potentially reinforce real-world stereotypes.
The first project consists of a system called AIRvatar (named for the Advanced Identity Representation), which is an avatar constructor for collecting analytical data such as mouse-click events and the amount of time spent in the different parts of the menu.
With AIRvatar, we found that social phenomena such as gender-related stereotyping could be observed through choices made by players (Lim, 2015). For example, female players appeared to conform more toward stereotypical notions of masculinity and femininity. Many gave male avatars significantly more strength and endurance points than female avatars, and significantly more intelligence and wisdom points for female avatars than male avatars. This effect appears related to the idea of “cross-stereotyping,” a type of “identity tourism” (Nakamura, 2008) in which players attribute a more limited range of behaviors to other genders than they do to their own. The fact is that avatars constructed by users introduce new audience-driven types of stereotyping. Electronic literature authors must ask whether, if such stereotypes are commonplace, we want to subvert, challenge, or change them in the systems we create.
The second project studies how avatar construction impacts user performance, identity development, and emotional engagement (Kao and Harrell, 2015). Experiments were conducted in our game called Mazzy (1892 online participants total in the studies discussed here). We contrasted outcomes during which users either deployed a minimal avatar (black dot), an abstract avatar (geometrical shape), or a likeness avatar (that looks like the user). We also investigated the impacts of user face photos, famous figures, and user-selected role models.
Minimal and shape avatar users were more engaged, had significantly higher enjoyment, and less difficulty. Likeness avatar users had significantly higher affect towards their avatars, yet reported significantly higher difficulty. Results suggest that Black or African American participants have lower affect towards the game than White participants in the user face photo condition. Yet, women using famous figures performed better than when using shape avatars and low performing users with role-model avatars did better than low performing users with shape avatars.
Although game-oriented, our results are more broadly informative for electronic literature. The fact that the replayability and emotional engagement are impacted by the types of avatar used in light of the demographics of those users is important. We have shown that such systems impact how users see themselves, perform, and feel about themselves. As such, authors have a great responsibility to their users. We hope that the results and discussion here will help inform electronic literature authors who are concerned about their impacts on diverse audiences.
(Source: ELO 2015 Conference Catalog)
