Nikhita Joshi
University of Waterloo
nvjoshi@uwaterloo.ca
Bio
Nikhita Joshi is a fifth-year PhD candidate at the University of Waterloo School of Computer Science. Her research area is broadly in human-computer interaction (HCI). Specifically, she is interested in creating and evaluating novel systems and interaction techniques for desktop, mobile, and mixed reality environments. Her dissertation is focused on leveraging user interface constraints to encourage positive user outcomes while reading and writing. Her work has been published at top-tier HCI venues, with 7 published at CHI and UIST and 4 winning best paper awards. She is the recipient of top scholarships at the federal, provincial, and university level, such as the NSERC Canadian Graduate Scholarship, Ontario Graduate Scholarship, and the Cheriton Graduate Scholarship. Recently, she was recognized as a Young Researcher by the Heidelberg Laureate Forum Foundation. She interned with Autodesk Research, Microsoft Research, and Meta Reality Labs Research, and even co-founded a microbrewery during her bachelor’s degree.
Areas of Research
- Human-Computer Interaction
Bounded Interfaces: Constrained User Interfaces for Positive User Outcomes while Reading and Writing
Constraints have the connotation of “limiting” or “restricting,” but prior work in psychology shows that they have many enabling properties, like improving creativity and productivity. In human-computer interaction (HCI), constraints have mostly been used for error-proofing, but I believe that constraints can be placed at the forefront of interaction design to encourage positive outcomes within software. I aim to create novel interaction techniques that intentionally constrain users in a positive way, a concept I formalize as “bounded interactions.” So far, I have created bounded interactions to support reading and writing. Scrolling is a fundamental user interaction that has become increasingly “mindless,” which fragments reading and lowers reading comprehension. I am currently designing and evaluating bounded interactions that encourage users to be more mindful of their scrolling and improve reading comprehension, such as timeouts between scroll events or adjusting the control-display gain. Highlighting helps readers remember information from documents, but it must be done selectively to be effective. I constrained the number of words that can be highlighted in a document reader. Results showed that when highlights were constrained to 150 words, participants did significantly better on a reading comprehension test than highlighting nothing or highlighting an unlimited number of words. Psychological ownership, feelings of something belonging to someone, is important to writers. Increased “collaborations” with generative AI assistants likely reduces psychological ownership, but writing longer prompts may improve it. I conducted two experiments to better understand these relationships and found that when participants wrote longer prompts, they felt more psychological ownership toward the generated output. I modified the prompt submission interface so users had to press and hold the prompt submission button for a few seconds to submit their prompt, which was effective at increasing prompt length.