Review Reports - Nonlinear Perceptual Thresholds and Trade-Offs of Visual Environment in Historic Districts: Evidence from Street View Images in Shanghai

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This research investigates how the built environment shapes human perception in historic districts within modern megacities, using six case studies in central Shanghai, by using street-view imagery (to quantify visual elements like greenery and buildings), Points-of-Interest data (to map urban functions), and online surveys (to capture subjective perceptions like beauty, liveliness, and depression). The main innovation lies in its methodology and findings. It moves beyond simple correlations to reveal threshold effects and perceptual trade-offs. For instance, a "green view ratio" significantly boosts a sense of beauty only after surpassing 30-40%, while commercial density increases vibrancy but can trigger feelings of depression when too concentrated. The study concludes that a district's appeal doesn't come from maximizing any single element but from achieving a dynamic balance among three key factors: natural elements (for aesthetics and comfort), functional facilities (for vitality), and order-related features (for a sense of safety and cleanliness).

The research is naturally limited by its data. The static, daytime, and car-perspective street-view images cannot capture the dynamic experience of a district at different times, in various weather, or from a pedestrian's eye level. Furthermore, the online participant pool might not fully represent the diverse perceptions of all social and cultural groups. Future work could thus benefit from incorporating dynamic visual data, multi-sensory measurements, and a more diverse set of evaluators to deepen our understanding of these lived environments.

1. Reframe the work around its core discovery: non-linear thresholds and perceptual trade-offs. The title and abstract should highlight this, not just the assessment methodology.

2. Strengthen the introduction by explicitly framing the study around the central tension in historic districts: heritage aesthetics vs. metropolitan functionality. Position your "perceptual balance" as a new lens to resolve this tension.

3. Add brief, critical justifications in the methodology:XGBoost: Explain why it was chosen over other models for this specific task.Perception Dimensions: Reference their established use in literature (e.g., Place Pulse) to bolster validity.SHAP/PDPs: Frame them as essential tools for interpretability, directly addressing a known gap in the field.

4. Elevate the discussion by synthesizing findings into a concise conceptual model. Visually illustrate how the balance between Naturalness, Functionality, and Order drives positive or negative perception. Connect your "thresholds" directly to established urban design concepts.

5. Reframe limitations as a direct research agenda. Propose specific next steps: multi-temporal street-view analysis, pedestrian-perspective VR studies, and cross-cultural comparisons to address the current constraints.

Author Response

Reviewer #1:

This study is naturally constrained by its data. Static, daytime, vehicle-based street-view images cannot capture the dynamic experiences of the same areas under different times, weather conditions, or from a pedestrian perspective. In addition, online participants may not fully represent the perceptual diversity of all social and cultural groups. Therefore, future research could incorporate dynamic visual data, multi-sensory measurement methods, and a more diverse pool of evaluators to deepen our understanding of these lived environments.

Point 1:

Reframe the work around its core discovery: non-linear thresholds and perceptual trade-offs. The title and abstract should highlight this, not just the assessment methodology.

Response 1: Thank you for this constructive suggestion. We have revised both the title and abstract to more clearly foreground the study’s core discovery—namely, the non-linear thresholds and perceptual trade-offs that structure perceptual responses in historic districts. At the same time, we have preserved and clarified the role of our street-view–based machine-learning assessment, as it is an essential component of the study’s analytical contribution. The revised versions now present a more balanced framing that highlights both the methodological innovation and the theoretical insights derived from the analysis.

Point 2:

The format of citations in this paper does not comply with the journal's requirements. The author should carefully review and ensure consistency in citation style throughout the manuscript. Additionally, the formatting of formulas and special symbols is inconsistent; also, please standardize the decimal points in the figures to either two or three decimal places.

Response 2: Addressed. In accordance with your comment, we have significantly strengthened the introduction. The revised text now explicitly highlights that historic districts in modern metropolitan contexts simultaneously embody the aesthetic values of cultural heritage and accommodate the functional demands of contemporary urban life—an inherent and central spatial tension.

Building on this, we position perceptual balance as the core analytical lens of the study, explaining how individuals cognitively and emotionally negotiate between heritage aesthetics and metropolitan functionality.

Point 3:

Add brief, critical justifications in the methodology:XGBoost: Explain why it was chosen over other models for this specific task.Perception Dimensions: Reference their established use in literature (e.g., Place Pulse) to bolster validity.SHAP/PDPs: Frame them as essential tools for interpretability, directly addressing a known gap in the field.

Response 3: Thank you for the suggestion. We have added methodological clarifications: highlighting XGBoost’s suitability for nonlinear relationships, referencing Place Pulse to support the perception dimensions, and identifying SHAP and PDP as key tools for improving model interpretability.

Point 4:

Elevate the discussion by synthesizing findings into a concise conceptual model. Visually illustrate how the balance between Naturalness, Functionality, and Order drives positive or negative perception. Connect your "thresholds" directly to established urban design concepts.

Response 4: Thank you for the valuable comment. In the revision, we have synthesized the key findings into a concise conceptual model that visually illustrates how the balance among Naturalness, Functionality, and Order drives positive and negative perceptions. The model incorporates the identified threshold ranges and presents them as integrated proportional relationships.
We also explicitly connect these thresholds to established urban design concepts to strengthen the theoretical relevance. The revised discussion and the new figure (Figure 10) now clearly demonstrate these linkages.

Point 5:

Reframe limitations as a direct research agenda. Propose specific next steps: multi-temporal street-view analysis, pedestrian-perspective VR studies, and cross-cultural comparisons to address the current constraints.

Response 5: Thank you for the constructive comment. We have revised the limitations section into a clear research agenda. The updated text now outlines (1) multi-temporal street-view analysis to overcome static imagery constraints, (2) pedestrian-perspective and VR-based experiments to bridge the gap between online and lived experiences, and (3) cross-cultural comparison frameworks to address sampling limitations. These additions directly respond to your suggestion and strengthen the forward-looking contribution of the study.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The study is an attempt to quantify the psychological experience of people immersed in a complex urban context. The objects of the study are historical districts (all - in Shanghai, China) with a typical clash of the attributes of habitual historical life and the demands of modern dynamic life. This is a difficult and urgent task at the same time. The research history is described in detailed manner and logically justified, the limitations of the data-driven approach are well-founded. Research provided in a multidimensional mental environment. The study included the decomposition of spherical panoramas (more than 3000 were made for this research) with automatic selection of thematic elements, and a survey of users (almost 1,000 participants), revealing the features of direct perception of the spatial context; method looks adequate and reasonable.

It is necessary to note the rich research data obtained in this study. Factors of user experience are identified, classified and explained convincingly.

Language is good, article is acceptably structured, the narrative is reasonable. Visual component (pictures) is very rich and compatible with text. References look adequate. Annotations and conclusions are informative. Thanks to huge volume of empirical data article, I guess, could triggered discussions.

No ethic issues, AI-intrusions etc. were detected.

Identifies issues/corrections:

2.4. (Chapter title) :357

Article could be published with tiny improvement (:357).

Author Response

Reviewer #2:

Language is good, article is acceptably structured, the narrative is reasonable. Visual component (pictures) is very rich and compatible with text. References look adequate. Annotations and conclusions are informative, but a few minor issues should be addressed before it can be published.

Major revision:

Point 1:

Identifies issues/corrections: 2.4. (Chapter title)

Response 1: Addressed, thank you for pointing this out. We have corrected the title of Section 2.4 accordingly. The updated version now reflects the intended content more accurately.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This paper is strong and well-put-together. It presents a careful study of how different environmental features influence people’s perceptions of historic districts in Shanghai, drawing on street-view imagery, POI data, and an online perception survey. In my view, the manuscript is close to publishable and mainly needs a few minor clarifications.

First, check 6/7 perceptual dimensions. This requires a decision and a brief explanation. A short description of the demographics frame should also be important. On the modelling side, I suggest adding a summary of model performance for each perceptual dimension (for example, R² and one error metric on a validation or test set), ideally collected in a small table. In the abstract, the final sentence (“This study offers quantitative evidence and theoretical insight to support the enhancement of landscape quality and the promotion of sustainable urban vitality.”) could be made more concrete by briefly indicating what kinds of planning or design actions you have in mind. In the conclusion, ending with three or four practical design or management principles for historic districts (for example, indicative ranges for greenery, commercial density or facility mix suggested by your findings) would underscore the architectural relevance of the work and give practitioners something very tangible to take away. Please review counter-bias measures, anonymity, and ethical research stances. Some SHAP and PDP plots are also very dense. Try to manage a better fit for that kind of complex visual information.

By fixing these elements, you should be able to properly address the research formalities and readability. The manuscript is methodologically strong, well aligned with the journal’s scope and of clear interest to researchers working on urban perception, historic districts and data-driven urban design. The main issues relate to clarification (sampling, aggregation, consistency of dimensions) and presentation rather than fundamental flaws.

Author Response

Reviewer #3:

This paper is strong and well-put-together. It presents a careful study of how different environmental features influence people’s perceptions of historic districts in Shanghai, drawing on street-view imagery, POI data, and an online perception survey. In my view, the manuscript is close to publishable and mainly needs a few minor clarifications.

Major revision:

Point 1:

Check 6/7 perceptual dimensions. This requires a decision and a brief explanation. A short description of the demographics frame should also be important.

Response 1: Addressed, We appreciate the reviewer’s insightful comment regarding the choice of perceptual dimensions and the need for demographic clarification. The Method section has been updated to include these clarifications. In urban perception research, commonly examined dimensions include safety, liveliness, beauty, wealthiness, depression, and boredom. Given that the historic districts examined in this study are high-density, daily-use environments in Shanghai, we included cleanliness as an additional perceptual indicator. Cleanliness and perceived order have been repeatedly demonstrated in environmental psychology and urban governance research (e.g., Lansing & Marans, 1969) to influence comfort, safety, and place image. Therefore, incorporating the “clean” dimension allows us to more accurately capture the experiential structure of local urban settings.

Regarding demographics, the participants in our sample are primarily university students and faculty members in Shanghai, most of whom have long-term living or study experience in the city. This group tends to have higher educational backgrounds and stronger environmental sensitivity, making them well suited to evaluate visual perception and everyday environmental qualities in historic districts.

Point 2:

On the modelling side, I suggest adding a summary of model performance for each perceptual dimension (for example, R² and one error metric on a validation or test set), ideally collected in a small table.

Response 2: We appreciate the reviewer’s constructive suggestion. Following your recommendation, we have added a performance summary for each perceptual dimension and presented the results in concise tables. These tables are now provided in Appendix B for clear reference.

Point 3:

In the abstract, the final sentence (“This study offers quantitative evidence and theoretical insight to support the enhancement of landscape quality and the promotion of sustainable urban vitality.”) could be made more concrete by briefly indicating what kinds of planning or design actions you have in mind.

Response 3: Addressed. Thank you very much for this constructive suggestion. In response, we have revised the final sentence of the abstract to make the practical implications more concrete. The revised version briefly outlines several planning and design actions derived from our findings, including enhancing visible greenery (e.g., adding street trees, introducing pocket greenery, and improving façade greening), improving street openness (e.g., optimizing view corridors, reducing visual obstructions, and adopting appropriate setback adjustments), guiding an appropriate mix and spatial distribution of commercial and service facilities, and strengthening the perceptibility of cultural landscape features (e.g., façade rehabilitation, streetscape coherence, and improved signage systems).

Point 4:

In the conclusion, ending with three or four practical design or management principles for historic districts (for example, indicative ranges for greenery, commercial density or facility mix suggested by your findings) would underscore the architectural relevance of the work and give practitioners something very tangible to take away.

Response 4: Addressed. Thank you for this valuable suggestion. In accordance with your recommendation, we have revised the Conclusion to include three concise, practice-oriented principles derived from our empirical findings. These additions specify indicative threshold ranges for visible greenery, facility mix, and the exposure of institutional and traffic-related elements, thereby translating the quantitative results into concrete design and management guidance for historic districts.

Point 5:

Please review counter-bias measures, anonymity, and ethical research stances.

Response 5: Addressed. The study implemented several measures to ensure ethical compliance, data quality, and participant protection. The questionnaire was administered through a custom-developed online system, which randomly selected street-view images from a predefined image pool and generated a unique anonymous ID for each participant. Each image set and its corresponding ID were locked prior to submission, ensuring that images were not reused within the same batch and reducing potential rating biases. The online survey did not collect any personal or identifiable information, and all participation remained fully anonymous. Standardized instructions and consistent visual stimuli were provided, and the image order was randomized to minimize ordering or framing effects.

The sample primarily consisted of university students and faculty members living or working in Shanghai; however, due to constraints related to research resources, available time, and concerns regarding the quality of online responses, certain sampling biases may still be present. Future studies will refine recruitment strategies and expand the sample to include more diverse age groups, occupations, and life backgrounds to enhance representativeness and generalizability. The study followed voluntary participation and informed consent principles, involved minimal risk, and complied with the Declaration of Helsinki. The research protocol was reviewed and approved by the Institutional Ethics Committee of the University of Shanghai for Science and Technology (Approval Code: USST20250519; Approval Date: 19 May 2025).

Point 6:

Some SHAP and PDP plots are also very dense. Try to manage a better fit for that kind of complex visual information.

Response 6: Addressed. Thank you for pointing this out. In response, we have refined the SHAP and PDP figures by adjusting layout density, enlarging key visual elements, and improving spacing and readability. These revisions provide a clearer fit for the complexity of the visual information and enhance interpretability for readers.