Personality, Mental Health, and AI Research Lab
-
We develop multimodal AI, including transformer-based language models, for personality and mental health assessment, using behavioral signals such as language, speech, movement, and facial expressions.
-
We study how personality relates to health outcomes over time, and how personality and mental health change, including in response to therapy.
-
We examine how people see their own personality and mental health, how others perceive them, and how these perspectives diverge.
Advancing psychological science in three areas
Artificial Intelligence
Harnessing machine learning to improve psychological assessment
We develop models that detect nuanced behavioral signals associated with personality traits and clinical symptoms. Applied in clinical and other settings, these tools can assess personality and mental health more efficiently -- a direct, behavior-based complement to self-report.
​
There is still much to learn about disorders such as depression, anxiety, and PTSD. By improving how we measure them, AI-based assessment can deepen our understanding of these conditions and help advance treatment.
Personality and Mental Health
Using science to understand personality and health
We use the five-factor model (FFM) as our framework for understanding personality. The FFM has a strong empirical foundation, with roots in the natural language people use to describe one another. Its traits are closely tied to mental health and predict important life outcomes, including physical health. One goal of the lab is to improve how we assess personality, so we can uncover the causal pathways linking it to health.
Commitment to Diversity and Equity
Developing psychological science in the context of group differences
Both psychological assessment and AI have a history of bias and inequity. A primary goal of the Personality, Mental Health, and AI lab is to build new behavioral assessments using data from people of diverse backgrounds, so we can understand group differences and reduce bias from the start.
