Embedded Files
Project 1
Project 1
Investigating the Role of Metaphors in Human-AI Interaction
Investigating the Role of Metaphors in Human-AI Interaction
Collaborator: Dr. Rui Mao, School of Computer Science, Nanyang Technological University
Metaphorical expressions are widely used in daily communication between humans to improve the understanding of both complex and abstract concepts. However, it is unclear whether the usage of metaphorical language is helpful for enhancing the conversational engagement between humans and AI. In this work, we leverage a state-of-the-art computational metaphor processing tool to gain insights from human and Chat- GPT conversations. Our quantitative analysis finds that although metaphors may enhance the quality of human-AI interactions, they do not directly lead to higher levels of conversational engagement, which is measured by the duration of the conversation. However, regression analysis shows a notable relationship between user and ChatGPT metaphor usage, suggesting that ChatGPT is adept at reflecting the linguistic style of users, especially in terms of metaphorical language. Additional topic modeling and concept mapping analyses further explored the patterns of metaphorical language across various engagement levels and topics in user and ChatGPT messages.
Maheshwary, P., Lui, Q., Mao, R. (2024, December). Investigating the Role of Metaphors in AI-Human Conversation and Engagement. In Proceedings of IEEE’s International Conference on Data Mining Workshops, Abu-Dhabi, UAE.
Project 2
Project 2
Investigating the Impact of Early Abacus Training on Later Math performance
Investigating the Impact of Early Abacus Training on Later Math performance
Advisor: Dr. Martha Alibali, Dept. of Psychology, University of Wisconsin-Madison
Advisor: Dr. Martha Alibali, Dept. of Psychology, University of Wisconsin-Madison
Maheshwary, P., Anthony, L., Alibali, M. (2024, April). The Impact of Early Abacus Training on Later Math performance. Poster presentation at the Midwestern Psychological Association, Chicago, IL.
Maheshwary, P., Anthony, L., Alibali, M. (2024, June). Does Childhood Experience With the Abacus Influence Mathematics Performance in Adulthood? Poster presentation at the Midwestern Psychological Association, Chicago, IL.
Past research suggests that children who are trained in using the abacus use a “mental abacus” when performing mental calculations. Specifically, children who have had abacus training are less accurate at performing mental calculations when they are asked to tap a finger on the table concurrently with performing mental calculations (Brooks et al., 2018). This interference from concurrent motor movements suggests that they use a “mental abacus” to perform mental calculations. In this study, we plan to explore the long-term impact of childhood abacus training on mental calculation. Specifically, we will test whether adults who received extensive training in the abacus as children will demonstrate evidence of using the mental abacus for calculation in adulthood. Participants will perform mental calculations under two conditions: when free to move their hands, and when asked to tap a finger on the table while performing calculations. If participants continue to use a mental abacus in adulthood, they should display degraded performance when asked to tap a finger on the table concurrently with mental calculations. Control participants who did not receive abacus training as children should not display this pattern. We will also examine whether early experience with the abacus is associated with enhancements in participants’ spatial ability and flexibility in representing quantities. If abacus training does have long-term effects, we expect abacus users to score higher than non-users on spatial reasoning and representational flexibility tasks.
This project has secured $7000 in funding in total through the Hilldale Undergraduate Research Fellowship and Summer Senior Honors Thesis Grant.
Project 3
Project 3
Designing an Effective Navigation System for People with Low-Vision
Designing an Effective Navigation System for People with Low-Vision
Advisor: Dr. Yuhang Zhao, Dept. of Computer Science, University of Wisconsin-Madison
This study evaluates a novel augmented reality (AR) navigation system designed to enhance indoor navigation for individuals with low vision using the Microsoft Hololens. By integrating augmented reality with strategically placed landmarks, the system aims to improve spatial awareness and independence in indoor environments. We conducted a mixed-methods evaluation involving 21 participants (16 with low vision and 5 sighted) to assess the usability and effectiveness of the system. Results from user testing with sighted participants indicate that while the AR system enhances the navigational experience in terms of engagement and confidence, it does not significantly improve the accuracy of mental maps or navigation efficiency compared to traditional methods. Feedback from sighted participants helped refine the system, emphasizing the importance of intuitive interfaces and reliable landmark augmentation. This study contributes to the understanding of AR applications in accessibility, suggesting further research to optimize these technologies for diverse user needs.
This project secured $1500 funding in total through the Trewatha Senior Honors Thesis Award by the Honors Program at College of Letters & Sciences at UW-Madison.
Chen, R., Jiang, J., Maheshwary, P., Cochran, B, Zhao, Y. (2025). Accepted. VisiMark: Characterizing and Augmenting Landmarks for People with Low Vision in Augmented Reality to Support Indoor Navigation. [Conference Paper] In Proceedings of the 2025 SIGCHI Conference on Human Factors in Computing Systems, ACM, New York, NY, USA.
Maheshwary, P., Kaku, M., Pan, S. (2024, April). Does the Pre-testing Effect Expand or Contract Over Time? Poster presentation at the Midwestern Psychological Association, Chicago, IL.
Project 4
Project 4
The Dynamics of the Pretesting Effect on Long-term Retention: Evidence from a Retention-Interval Study
Advisor/Collaborator: Dr. Steven Pan, Dept. of Psychology, National University of Singapore
Understanding the mechanics of memory retention is pivotal for optimizing educational strategies. The pretesting effect—a pedagogical approach where learners attempt to answer questions about content before engaging with the material—is suggested to facilitate better long-term retention. However, the temporal durability of this effect remains insufficiently examined. Our study aims to dissect the pretesting effect over an extended time frame, specifically assessing its persistence or attenuation over intervals of up to 28 days.
Methods
Methods
We employed a within-subjects design, enlisting 62 participants via an online platform, with 58 completing the study. The methodology entailed five testing sessions post-learning, with retention intervals at immediate, 1 day, 7 days, 14 days, and 28 days. Participants learned 50 trivia facts, with half of the items learned through a pretesting mechanism—participants attempted to recall or guess information prior to exposure to the correct content—and the remaining through a straightforward study. Each session thereafter assessed recall for 10 items, evenly split between pretested and studied.
Results
Results
Unexpectedly, the advantage of pretesting was ephemeral, prominently present only in the immediate and following day's sessions. Interestingly, at longer intervals—7 to 28 days—the studied items yielded higher recall rates. This decline in the pretesting benefit suggests a nuanced relationship between the method of learning and memory consolidation over time, potentially implicating the nature of the encoding and retrieval processes involved.
The initially observed pretesting benefit likely stems from enhanced cognitive engagement or the principles of encoding specificity, where the context of initial guessing matches that of subsequent testing. However, this benefit appears to decrease with time, possibly due to varying strengths in memory traces between pretested and studied information. These findings propose significant implications for educational frameworks and the field of cognitive psychology, proposing that pretesting, while beneficial in the short term, may not provide a stable scaffold for long-term knowledge retention.
Project 5
Project 5
A Novel Measure of Symbolic Presentational Flexibility in Mathematics for Children
A Novel Measure of Symbolic Presentational Flexibility in Mathematics for Children
Advisor: Dr. Martha Alibali, Dept. of Psychology, University of Wisconsin-Madison
The above developed measure was adapted in:
Anthony L., Becerra C., Stevens I., Stephens A., Maheshwary P., Donovan A., Matthews P., & Alibali M. (2025). Submitted. A Substitutive Conception of Equivalence: Relations with Representational Flexibility and Problem Solving. [Conference Paper] Society for Research in Child Development (SRCD), Minneapolis, MN, USA.
Project 6
Project 6
The Role of Anthropomorphism in Children's Learning about Viruses from Storybooks
The Role of Anthropomorphism in Children's Learning about Viruses from Storybooks
Advisor/Supervisor: Dr. David Menendez, Dept. of Psychology, University of California Santa-Cruz
Page updated
Report abuse