META-TOO: Student Projects
Master’s theses
Personalized and Gender-Swapped Embodiment: A Virtual Reality Intervention on Gender Bias
Author: Kátia dos Santos Estevães
Granting University: Lusófona University (Erasmus Mundus)
Abstract: Social virtual reality platforms are enabling new forms of interaction but also perpetuating old challenges, such as harassment toward women that reflects real-world gender bias. We investigated whether embodying a female avatar experiencing sexual harassment in a metaverse simulation of a job interview leads to a reduction in implicit gender bias. Building on the Golden Rule Embodiment Paradigm, 37 male participants were assigned to either an experimental condition embodying a female victim of harassment or a control condition embodying a female without experiencing harassment. We implemented personalized avatars to enhance identification, and a Lexical Decision Task (LDT) was used to measure implicit bias. Contrary to our hypothesis, quantitative results showed no statistically significant reductions in implicit gender bias, as well as embodiment did not correlate with bias change. However, qualitative data indicated greater emotional engagement, perspective-taking, and gender-based violence awareness in the experimental group. These findings demonstrate the complexity of changing implicit gender bias and suggest that emotionally intense VR scenarios may enhance awareness and reflection, though they may not be sufficient to produce implicit learning. We discuss implications for VR interventions and future research directions.
The Role of Sentiment Analysis in Identifying Changes in Attitudes Toward Gender-Based Harassment in Virtual Reality
Author: Abigya Abayneh Melese
Granting University: Lusófona University (Erasmus Mundus)
Abstract: This dissertation investigates how virtual reality (VR) can influence participants’ attitudes toward gender-based harassment (GBH), with a particular focus on the role of sentiment analysis in identifying emotional and attitudinal shifts. By enabling users to embody diverse perspectives, VR facilitates changes in social cognition, empathy, and implicit bias, offering a novel medium for addressing persistent societal issues such as GBH (Slater et al., 2009; Banakou et al., 2016). While previous studies have demonstrated VR’s potential in fostering perspective-taking, this work explores how computational tools especially sentiment analysis can support the detection and interpretation of attitudinal changes. Sentiment analysis is considered alongside other evaluative techniques to understand participants’ emotional responses and potential shifts in viewpoint following immersive experiences. Drawing from existing literature and practical implementations, the dissertation highlights the added value of integrating sentiment analysis into multi-method approaches for assessing the effectiveness of VR-based interventions. The findings contribute to ongoing efforts to develop scalable, data-informed methods for promoting prosocial attitudes and reducing tolerance toward GBH.
Bachelor of Science theses
Development of an Immersive Virtual Environment to Support Research on Gender-Based Harassment in the Metaverse
Author: Mary Diamandi
Granting university: National and Kapodistrian University of Athens
Abstract: As immersive virtual reality (VR) technologies become increasingly integrated into work, education, and social spaces, new concerns have emerged around user safety and misconduct, particularly gender-based harassment. This thesis presents my contribution to the design and implementation of a cross-national VR study developed as part of the META-TOO project. The project investigates how virtual embodiment, where participants experience a scenario from the perspective of another avatar, can influence male participants’ attitudes, awareness, and behaviors related to harassment. My work focused specifically on the technical development of the second part of the study, which included creating the “cafeteria scene in the virtual environment (VE) used for the study, animating non-player characters, integrating voiceovers and facial expressions, and ensuring that the scene functioned correctly in VR. I was also involved in the practical execution of the study sessions that took place in Greece, assisting with technical setup, procedure, and participant support. The study is based on the Golden Rule Embodiment Paradigm (GREP), which encourages behavioral change through embodied perspective-taking. Across two sessions, participants were exposed to realistic workplace scenarios in the metaverse. In the first session, they observed a job interview (either neutral or harassing) and then re-experienced the same event from the perspective of a female interviewee. In the second session, they witnessed a cafeteria harassment scenario, where their verbal and non-verbal reactions were observed to assess bystander behavior. The experiment was conducted in both Greece and Spain, with a total of 40 male participants. Personalized avatars were created for each participant to enhance the sense of embodiment. Data collection included behavioral observations, semi-structured interviews, and a mix of implicit and explicit psychological measures, including the Sexual Harassment Attitude Scale (SHAS), the Ambivalent Sexism Inventory (ASI), and a Lexical Decision Task (LDT). All VR content was developed in Unity and deployed on Meta Quest headsets. This thesis contributes to research on immersive behavioral change and digital ethics by documenting the technical implementation of a virtual reality scenario designed to explore whether perspective-taking can help reduce gender bias and encourage more active bystander responses.