Author ORCID Identifier

https://orcid.org/0009-0004-4118-7390

Date of Award

Fall 11-24-2025

Document Type

Thesis (Master's)

Department or Program

Computer Science

First Advisor

Lorie Loeb

Second Advisor

John P Bell

Third Advisor

Yu-Chun Huang

Abstract

Virtual reality (VR) narrative games must guide players through complex story worlds without undermining immersion, yet prior work rarely compares different forms of in-world guidance under tightly controlled conditions. Designers routinely mix diegetic text, character behavior, and environmental motion, but it remains unclear how these cue families differentially shape cognitive load, task performance, and narrative experience when they convey the same instruction.

This thesis investigates how three common diegetic cue types—diegetic text, non-verbal gestures from a non-player character, and environmental motion and highlights—affect players in a first-person VR retelling of the Snow White story. In a between-subjects laboratory study, each participant played the same linear narrative but experienced only one cue family. A pretest iteratively tuned the three versions of each cue to be comparable in perceived clarity, attention capture, and naturalness. During the main study, we logged task completion and failures, collected ratings and open-ended interpretations for each cue, and measured immersion, narrative engagement, self-reported performance, and subjective workload using the NASA Task Load Index (NASA-TLX).

Our results indicate that cue modality primarily influenced players' behavior and the precision with which they interpreted guidance, rather than overall post-session cognitive load or immersion. NPC gestures supported faster and more reliable playthroughs and were perceived as clear and narratively coherent. Environmental motion and highlights were effective for lightweight orienting (e.g., where to look or go), but often failed to communicate constraint-heavy intent, which led to more errors and less specific interpretations. Diegetic text produced the most accurate understanding of both goals and constraints and, in our measures, did not reduce immersion or increase overall NASA–TLX workload.

These findings suggest that diegetic guidance should not be treated as a single design intervention. Symbolic cues (text), social cues (NPC gestures), and environmental cues (motion/highlights) can produce meaningfully different outcomes even when message content is held constant. For practitioners, the results provide actionable guidance for selecting cue modalities: environmental signals are well-suited to low-cost spatial direction, whereas risk-sensitive or highly constrained actions may require more interpretable channels such as NPC gestures or brief in-world text. Although we did not evaluate multimodal combinations, the complementary strengths observed here motivate future work on staged or adaptive diegetic guidance that increases explicitness when subtler cues are missed.

Download

Share

COinS