Undergraduates' field of study is intricately linked to inequality in the US, where women have surpassed men in most indexes of academic achievement but continue to be less likely than men to complete STEM degrees. This gendered variation in major selection has substantial implications for stratification: college major choice is closely related to labor market outcomes and advancement to future degrees. Building on recent theoretical developments in social aesthetics and field theory, the project argues that academic interests are developed in concert with encounters in the environment, and that position in academic fields at the start of university, gendered distributions of interest patterns, and peer influence play a critical role in gender differentiation in college major choice. The project uses a unique longitudinal data combining complex administrative databases from an elite American university, merging admissions, housing, course, financial aid, and alumni data. Multiple correspondence analysis shows that students' interests are organized in academic fields characterized by divisions between knowledge domains: science interests oppose social sciences, economics interests oppose humanities, and life sciences are differentiated from hard sciences. Knowledge domains share features of retention and attraction, and movement between disciplines that are close together in students' interest spaces are more common. Using clustering methods, the project shows that there are important distinctions in how students are interested in disciplines: some students are particularly devoted to knowledge domains, while other students are generalists. These finding have important implications for women and men, who have different interest patterns. There is durability in gender differences in high school interests reinforced by both retention and attraction to disciplines once at school. The last chapter of the dissertation explores the role that peer influence plays in these outcomes. Because students' interests are organized in academic fields, peer influence on academic major choice is better understood as a field effect. Utilizing the fact that roommate assignment is random at this university, the project shows that choosing a major is associated with roommate's interests coming into college, and this association depends on students' own initial interests when applying to university. Generalist science students are more likely to complete science degrees when they have science or engineering roommates compared to those who have humanities roommates, while devoted science students are less mutable. Because women are less likely to have roommates who are in sciences and engineering, gender segregation of roommates contributes to gender difference in STEM outcomes. By reframing choice as a question of social aesthetics, the project makes important contributions to understanding choice, inequality and peer influence.