Examining the Impact of Urban Design on Environmental Preference Or: How I Learned to Stop Worrying and Love the City

As urban areas continue to rapidly increase in population, understanding how people experience, evaluate, and remember urban environments becomes increasingly important, particularly in urban design. Chapter 1 asks if prospect (an open view s), and refuge (a safe place to hide), predict preference for street-level urban scenes. Refuge consistently predicted higher preference, supporting its relevance in urban environments. While both prospect and refuge were initially associated with preference, only refuge generalized to a new image set using proxy measures derived using both visual (i.e., low- and high-level visual features) and experiential features (i.e., street-level urban design quality ratings), suggesting that prospect may function differently in various urban contexts. Chapter 2 examines whether urban design quality, imageability (the extent to which a scene is distinct, recognizable and memorable), can be explained by objective measures of memorability. In Study 1, imageability ratings from Talen et al. (2022) were based on the perception of "character," and neither human nor model-based memorability predicted this set of imageability ratings. In Study 2, when imageability was based on subjective participant selection of the most memorable images, both human nor model-based memorability were able to positively predict imageability. These results demonstrate that how imageability is measured can meaningfully affect its relationship to memorability, underscoring the importance of more consistent and theoretically grounded approaches to studying imageability in urban design. Chapter 3 extended the investigation of prospect, refuge, and preference by testing whether these relationships remain stable across neighborhood-level SES (i.e., median household income) in ~190,000 street-level images. In a rated subset, refuge again predicted preference, while prospect was negatively associated. A transfer learning model was trained on both image sets to scale predictions citywide. This chapter found that refuge's effect 1 remained stable, while prospect's negative association with preference became less pronounced in higher-income neighborhoods, suggesting its role may be context-dependent and shaped by socioeconomic variation. This dissertation bridges the environmental psychology, urban design, and memorability literatures to broaden our understanding of pedestrian experiences of urban environments.