Happiness in America: A Multimodal Assessment

This thesis examines the concept of happiness in America through a multidisciplinary lens, integrating insights from psychology, sociology, political science, and computational analysis. It explores the rising trends of depression and dissatisfaction in modern American society, arguing that understanding happiness requires a rethinking of its conceptualization and history. The study employs multimodal machine learning methods to analyze three distinct corpora: Google Ngrams data, Reddit posts from depressed and non- depressed users, and photographs from the Library of Congress. Key findings include increasing associations between aging and mental health issues over the 20th century, a strong focus on self and on feelings in depressed individuals' language, and complex relationships between political affiliations and mental states. The research suggests that the current operationalization of happiness in psychology and popular culture may be contributing to societal issues, and proposes a return to more holistic, eudaimonic conceptions of well-being.