My Research
I am generally interested in how social interactions, whether in person or mediated by technology, engage emotion, perception, and memory to shape impressions, decisions, and the connections that sustain relationships. My work combines experimental and computational methods to investigate these dynamics across diverse social contexts.
LLM-Based Scoring of Narrative Memory Reveals Affect's Selective Effects
This project introduces an LLM-based, scalable framework to score central and peripheral recall from two public naturalistic narrative datasets (Sherlock and Filmfest). We show that higher emotional arousal selectively improves memory for central information while reducing recall of peripheral details.
Working with Data Across Modalities
Pupil Data Processing: Preprocessed timestamped eye-tracking data with blink-aware interpolation and filtering to examine how arousal dynamics predict memory fidelity. | |
AI Narratives and Cultural Perception: Analyzed 640K Reddit posts discussing ChatGPT and DeepSeek using emotion labeling (RoBERTa), topic modeling (BERTopic), and clustering to reveal cross-cultural narratives of AI innovation. | |
Public Sentiment and the United Healthcare CEO Shooting: Applied machine learning models (TF-IDF + logistic regression) to 20K Reddit posts, finding increased negative sentiment toward health insurance after the incident. |
Collectively, these projects reflect my interest in combining computational modeling with psychological theory to understand how emotion and cognition manifest in both physcial and social environments.