Built environment and urban vibrancy

Within the field of environmental psychology, Brunswik’s probabilistic lens model (Brunswik, 1956) theorized that the environment contains an abundance of cues. As shown in Figure [X], Brunswik likened the process of perception to a lens through which stimuli are perceived and come into focus. The lens model further indicates that observers make personality judgments about objects based on environmental cues. Similarly, Gibson (1979) contended that rather than perceiving individual features, or “cues,” humans organize environmental features into recognizable patterns, or “affordances,” based on an arrangement of factors that inform perception. These contributions inspired many later empirical studies in environmental psychology (Winkel, 2009), providing a solid general theoretical framework to explain how humans acquire perceptual information and process this information into images. However, there is room for improvement in understanding the detail of the built environment as perceived and the specifics of those patterns.

One widely used framework is Campbell et al. ’s (1976) model of neighborhood satisfaction, which proposes that life satisfaction results from objective stimuli, cognitive responses, and domain-specific satisfaction. Campbell et al. (1976) also argued that personal characteristics are important determinants of perception, domain satisfaction, and life satisfaction. Their model conceptualizes how objective environmental attributes affect life satisfaction in terms of two mediating processes: perceived environmental attributes and satisfaction with one’s residential environment. Although several previous studies have measured both the objective and perceived elements of built environment (and sometimes residential satisfaction), few have applied a sound theoretical model to empirical research. For example, although Hur et al. (2010) explored the influence of building density and vegetation rate on the perceptions and evaluations of

neighborhood satisfaction, they did not explicitly link their model to any theoretical framework. Most environment-related variables are measured by human auditing and hence pay little attention to the street view (see Campbell, 1976; Diener et al., 1985; Amérigo & Aragones, 1997, Kahn & Juster, 2002; Lee et al., 2010; Hur, 2010; Florida, 2011; Kim et al., 2014; Cao & Wang, 2016; Hadavi, 2017; Zhu & Fan, 2018).

Challenging the conventional planning paradigm, Jacobs (1961) argued that mixed land uses, small block sizes, varied building sizes, and a high population density are essential in creating lively places to live. Over time, these ideas became the foundation for various place-making theories. Whyte (1980) took Jacobs’ ideas a step further by using time-lapse photography to record and analyze people’s social lives in New York’s public spaces; he found that crowds drew people to public spaces and that the presence of water, trees, and food also positively influenced an area’s vitality. Lynch (1984) proposed a normative theory of high-quality urban form that encompassed vitality, sense, fit, access, control, efficiency, and justice. In a deeper analysis of the relationship between urban form and human activity, Gehl (1987) identified three subtypes of activity (necessary, optional, and social), suggesting that the optional and social activities were more strongly influenced by the built environment than necessary activities, apparently because the latter were considered compulsory. In a subsequent review of the literature, Montgomery (1998) listed 12 physical conditions that were necessary for place-making. Recently, Mehta (2007) adopted Whyte’s approach to record Boston’s street life and found that people seemed equally concerned with the social, physical, and land use aspects of their streets. Although most studies are conducted using very thorough ideas, because of the limitations of most data collection techniques, researchers usually have preferred to approach this topic qualitatively, mainly by using case studies. More importantly, the theories proposed by the literature are relatively fragmented and cannot be

merged in an intuitive way. Carmona (2003) attempted to summarize all the literature in one framework of 10 major factors, but most of the concepts he proposed cannot be easily quantified. A lack of a holistic general theory that could guide the setting of the variable has indeed limited the quantitative empirical study of this topic until the recent major breakthrough by Ewing and Clemente(2013).

Ewing and Clemente (2013) constructed a systematic framework to quantify the urban form measures for urban design and place-making–related concepts, including enclosure, human scale, transparency, tidiness and imageability, to represent street quality. However, they still used high-cost and labor-expensive human auditing methods to collect the data. In a recent review of the possible GIS-based measures of streetscape when it comes to livability, Harvey and Aultman- Hall (2016) noted the potential of applying computer vision techniques in a streetscape analysis. The development of these techniques can also enhance data collection when exploring perceived vibrancy. Crowdsourcing and online surveys now facilitate neighborhood cognition audits that involve groups of remote volunteers, providing much larger samples than traditional survey methods. In MIT’s Place Pulse project (Dubey et al., 2016; Salesses et al., 2013), more than one million votes were collected for about a hundred thousand places worldwide. This approach has great potential for studies of urban environment perception. Based on these data, Naik et al. (2014) developed a scene-understanding algorithm to predict the perceived safety of a streetscape. Additionally, Harvey et al. (2015) used these data to identify a set of “streetscape skeleton” design variables and examined these variables’ relationship to perceived safety scores created by Place Pulse.