**1. Introduction**

Urbanization can affect the hydrologic functions of urban watersheds and precipitation patterns [1–5]. The consequential increased use of impervious surfaces results in substantial increments of stormwater runoff volume and peak flow [6]. Thus, the transition from the conventional approach into a more sustainable stormwater managemen<sup>t</sup> paradigm which includes green stormwater infrastructure (GSI), is indispensable to reducing substantial environmental, economic, and social damage [7–9]. Hence, there is also a need to understand the hindrances and limitations in GSI implementation.

GSI offers a promising solution to stormwater managemen<sup>t</sup> by mimicking natural hydrological processes to reduce localized flooding events and water quality improvement through decentralized natural or engineered processes to treat stormwater runoff at its source [10]. In the US (United States), awareness of GSI has slowly increased over the past two decades. Its historical progress in stormwater managemen<sup>t</sup> and background knowledge is documented in several in-depth publications [11–14]. Research teams across nations have developed various GSI practices and in addition, retrofits and hybrid measures on different spatial scales (such as watershed scale and site scale, etc.) with diverse primary purposes have been developed [15–20]. The details on these practices are well documented in the literature [21–28].

Numerous studies have evaluated the performance of GSI, particularly in economic and technical aspects [14,29–32]. GSI provides extra benefits to the community, such as raising property values, enriching life quality, and providing adaptable climate resilience [33–35]. Urban stormwater managemen<sup>t</sup> has advanced gradually over the last two decades, thus various terminologies are used to define new principles and practices, where the concepts behind them often overlap [14,36]. Using these different terms may reduce effective communication in certain circumstances, such as when documenting all the alternative stormwater practices used in the US to assess their performance in general [36]. To avoid confusion, the term GSI was used throughout this work in referring to all types of multi-purpose structural stormwater managemen<sup>t</sup> practices that involve natural processes for runoff volume and water quality control.

Despite the progress, there are limited study efforts on non-technical factors, such as public perceptions and knowledge, that could explain the slow advancement in the wide adaptation of GSI to the desired level for stormwater managemen<sup>t</sup> and sustainability capacity building [37]. The contradiction between the low implementation rate of GSI in major regions of the US and the actual demand to address climate change impacts suggests that certain factors are hindering the relevant decision-making processes [38,39]. Furthermore, a study discovered the mismatch in the percentage of their survey participants that expressed an intention to support GSI and the number of those who actually adopted GSI [40]. This result is in agreemen<sup>t</sup> with the findings in an exhaustive review [41]. Irrational decision-making behaviors in energy-related decisions have been interpreted through the cognitive bias perspective [42,43], where cognitive biases can be defined as a belief that hampers one's ability to make rational decisions given the facts and evidence [44]. It has been supported by various studies that cognitive biases are influential in decision making and planning [44]. Yet, little attention has been given to the potential influence of cognitive biases in GSI implementation, despite numerous studies on perceptions of various GSI stakeholder groups [45–47]. This study aims to bridge this knowledge gap.

Historically, quantitative decision support tools have been developed with the main aim to maximize GSI performance to control runoff and water pollution and to be costeffective [48–52]. On the other hand, despite the extensive attempts made to expand the assessment work to include the social aspect of decision support [17,48,53–64], they lack a deeper understanding of the public perceptions and associated cognitive bias perspective to resolve the implementation dilemma from a bottom-up approach [65] as examined in other environmental issues [43,44]. This shortcoming can affect the expected outcomes envisioned by major decision-makers [42,66]. This study focuses on the barriers that could be linked to biased perceptions due to social factors in GSI development and implementation.

This work was conducted to examine the relevant social factors through the lens of cognitive biases, which may lead to implementation barriers during GSI adoption processes. The scope of social factors can vary significantly as they are commonly assessed in combination with factors from other dimensions, such as socio-ecological, social-cultural, socio-economic, and socio-technical factors [10,67–70]. We use a concept adapted from Gifford and Nilsson [71] to define social factors as the internal differences among people and the contextual factors that define them in this study. This study aims to understand the potential connections of cognitive biases with these barriers, and to recommend an approach to analyze and address the associated problems. Studies have been conducted to analyze cognitive biases with agent-based modeling (ABM) in various contexts [72–74]. However, no study has done a similar analysis in the context of GSI implementation. ABM is a methodology that can incorporate the autonomy, heterogeneity, and adaptability of individuals in a social system to study the resulting global patterns through a bottom-up approach [75,76]. It is also an approach that can carry exploratory simulations for a deeper understanding of the underlying adaptive behaviors and interactions that could lead to the emergence of phenomena that was previously overlooked [40]. However, the models developed solely based on social and physical science are usually fragmented in their fields, rely on qualitative analysis, or are difficult to incorporate into quantitative models [77]. This work was conducted to answer the following questions:


To address these research questions, we reviewed the literature on GSI implementation barriers that arise from social aspects and on the connections between cognitive bias with these barriers. Subsequently, we reviewed the literature to show and assess the applicability of ABM in addressing the issue of social factors' hindrances to GSI adoption and implementation.
