Projects
Exploring the Frontiers of Psychological Inquiry: My Research Portfolio
Welcome to a curated overview of my scholarly engagements, where rigorous research intersects with innovative application. This page is a testament to the ongoing quest for empirical understanding and the application of psychological principles through my current and previous research initiatives.
Current Projects
Insight Solutions
In the vanguard of interdisciplinary innovation, Insight Solutions epitomizes the confluence of machine learning and psychological assessment. Our current project is an ambitious endeavor to construct a robust machine-learning framework capable of elucidating cognitive processing styles and distortions through the analytical lens of written text. This pioneering work seeks to lay the groundwork for advancements in diagnostic methodologies and tailored therapeutic interventions. (Published paper coming soon)
The Relationship between Language and Emotion
This project represents a concerted effort to dissect the complex correlations between linguistic expression and the emotional dysregulation symptomatic of anxiety, PTSD, and depression. Through meticulous examination of linguistic narratives, our research aims to illuminate the underlying mechanisms that govern the manifestation of these psychopathologies, potentially informing new avenues for clinical intervention and psychotherapeutic methodologies. (Published paper coming soon)
Improving Data Accuracy with AI-Powered Question Rephrasing
​
The Challenge: The Hidden Biases in Our Questions
How we ask a question can dramatically change the answer we receive. In survey research, subtle biases in wording can lead to inaccurate data, flawed conclusions, and poor decisions. Biases like acquiescence (the tendency to agree with statements), social desirability (the tendency to give answers that make us look good), and the availability heuristic (overweighting recent information) are persistent threats to data validity. For decades, the only solution was a manual, time-consuming process of expert review and question revision. This research project asked: can we use Artificial Intelligence to automate this process and improve data quality at scale?
Our Solution: An AI for Neutral, Objective Questions
This study, which forms the scientific foundation for the technology at CogBias AI, investigated the effectiveness of an AI-powered tool designed to automatically detect and rephrase biased survey questions. We conducted a between-subjects experiment where one group of participants received original, biased survey questions, while another group received questions that had been rewritten by our AI to be more neutral and objective. We then compared the responses to measure the AI's impact.
Key Findings: A Significant Impact on Data Quality and Predictive Power
The results demonstrated that the AI intervention was highly effective at mitigating cognitive biases and improving the quality of the data collected.
•Significant Reduction in Bias: The AI-rephrased questions produced statistically significant changes in 33% of the targeted questions, proving that the intervention successfully altered how people responded.
•Dramatic Reversal of Acquiescence Bias: The tool was particularly effective against acquiescence bias. For one question about study preferences, the biased format suggested a majority of students (56%) preferred studying in a group. The AI-rephrased question revealed the opposite: a large majority (85%) actually preferred to study alone. This finding highlights how easily biased questions can lead to fundamentally incorrect conclusions.
•Improved Predictive Validity: Most importantly, the unbiased questions were a significantly better predictor of actual behavior. When asked to allocate resources for their study time, participants who received the unbiased questions showed a much stronger alignment between their survey answers and their behavioral choices. This indicates that the AI-rephrased questions captured a person’s true intentions more accurately.
Implications: Building More Trustworthy AI and Research
This research provides strong evidence that AI can serve as a powerful co-pilot in the research and data collection process. By automating the difficult task of bias reduction, tools like the one developed for this study—and now at the core of CogBias AI—have the potential to significantly improve the validity and reliability of data across the social sciences, market research, and business intelligence. This work represents a critical step forward in building more responsible, accurate, and trustworthy AI systems.
The Influence of PTSD Symptom Severity on Emotion-Laden Word use and Cognitive Processing Styles
​
This study investigates how PTSD symptom severity influences emotional language use and cognitive processing styles. By examining the use of emotion-laden words (e.g., "funeral") and emotion-label words (e.g., "happy") when describing traumatic events, we found that participants with higher PTSD severity used more positive emotion-laden words but often negated them, indicating complex emotional modulation and a defensive stance. In contrast, those with lower PTSD severity used these words more directly, reflecting healthier emotional processing. Additionally, PTSD was linked to maladaptive cognitive styles like overaccommodation (overgeneralizing) and assimilation (fitting new experiences into rigid beliefs), suggesting that rigid cognitive styles exacerbate emotional difficulties. Encouraging accommodative processing, where beliefs adapt based on new information, may help mitigate the negative effects of PTSD, improving emotional resilience. Future work will focus on clinical populations such as veterans, exploring the stronger associations observed between positive emotion-laden word use and PTSD severity. Ultimately, this research highlights the intricate connections between language, emotion, and cognitive processing in trauma-affected individuals, providing insights that can inform more targeted therapeutic interventions.
Past Projects
Mood Congruent Visual Perception
My prior research explored the psychophysical domain of affective cognition, specifically how emotional states can filter and influence the perceptual processing of visual stimuli. This project contributed empirical evidence to the discourse on mood-congruent perception, advancing the understanding of affective influences on sensory integration and cognitive interpretation.
