Modeling Human Judgments of Lifestyle, Health, and Wellness

MIT 9.660 Computational Cognitive Science

September 2025 - December 2025

Cambridge, Massachusetts

Bayesian Inference, Probabilistic Programming, WebPPL, Survey Analytics, Cognitive Modelling

Can human perceptions of durative measures for lifestyle, health, and wellness be modeled probabilistically? This project extended prior literature in cognitive modeling of human predictive judgments by focusing on how humans perceive durative measures for topics like Activities of Daily Living (ADLs), injury, and overall lifespan. The results from both a real-world participant survey, and a probabilistic programming model of the survey scenario, helped identify which types of cognitive judgments are based on informed priors and which are built on more error-prone heuristics. These insights can help us determine how and when to have technology message us about our health, which is particularly relevant to eldercare support.
and senior living.

This was an individual final project where I was responsible for the background researhc, survey procedure design, probabilistic programming for cognitive modelling, and presentation of the final results.

The results demonstrate that bayesian models are very useful in differentiating between types of questions that humans are confident in answering based on informed priors, and types of questions they struggle to answer because they either have no common knowledge about the topic area or it is difficult to recognize the underlying distribution of the observed data. For this reason, these types of exercises in comparing Bayesian prediction functions to human cognition could actually serve as a proxy to determine how well humans understand different topics in lifestyle, health, and wellness. This information will be incredibly useful to the field of Human Activity Recognition for eldercare communities and for seniors living at home by helping determine how technology should monitor and send alerts to an individual's care team. The project itself recieved high marks from instructor evaluation.

My central graduate research topic is Human Activity Recognition for eldercare support, and this project was very complementary to this. It allowed me to think more about the practical applications of my technology and how it will be used by real people or informed by their cognitive judgements. It was also a good exercise in applying bayesian inference and probabilistic programming methods I have been developing in my research on a smaller scale project.