Impact of Architecture Façade Design on Neurophysiological Stress Using Functional Near-Infrared Spectroscopy and Heart Rate Variability

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The article addresses an important and timely topic concerning the neurophysiological consequences of architectural façade design. The authors adopt an ambitious multi-modal approach combining ViStA computational analysis, fNIRS, HRV, and subjective ratings. While the study’s aim is novel and the theoretical framing is compelling, the manuscript exhibits substantial methodological and interpretative limitations that limit the strength of its conclusions. A major limitation concerns the extremely small effective sample size for fNIRS (n = 12), resulting from a 28% data loss due to optode–scalp coupling issues. Given the known challenges of occipital fNIRS recording, this level of attrition is not surprising, but it severely undermines statistical power. The manuscript repeatedly interprets null results cautiously, yet the extent of underpowering raises doubts as to whether the experimental design was adequate to test the stated hypotheses. The authors should explicitly quantify expected effect sizes and justify whether the recorded sample enables valid inferences. As it stands, the null fNIRS findings are uninterpretable rather than informative. The HRV analysis suffers from similar constraints: HRV was extracted from a 10 Hz fNIRS PPG signal rather than ECG, which is widely considered insufficient for accurate time-domain and unusable for frequency-domain metrics. The tiny marginal R² (0.0006) indicates that, despite statistical significance, the façade effect is negligible in practical terms. It is therefore unwarranted to discuss “autonomic modulation” without acknowledging that the analytical method lacks clinical-grade precision. The ecological validity of the projection setup is commendable, but other potential confounds are not adequately controlled. Luminance, contrast, and colorimetric properties of the projected stimuli are not photometrically calibrated, despite their centrality to visual stress research. Moreover, exposure durations (20 seconds for most images) are too short to produce reliable haemodynamic or autonomic responses. The authors acknowledge this partially, yet the discussion often drifts toward interpretative claims that exceed what the data support. The theoretical argument for a cortical–autonomic cascade is intriguing but underdeveloped empirically. The authors invoke allostatic load theory, but the physiological data offered here do not meaningfully test that framework. A more restrained interpretation is warranted: the study demonstrates a strong relationship between ViStA metrics and subjective ratings but offers no credible evidence of cortical activation differences and only minimal evidence of autonomic modulation. The manuscript should more clearly distinguish what is actually demonstrated from what remains speculative. Several methodological advances described—use of generative AI for stimuli, large-scale projection, integration of multiple modalities—are promising. However, their novelty does not compensate for foundational issues such as insufficient statistical power, inadequate instrumentation for HRV, absence of short-separation channels, and lack of real cortical coverage beyond V1. The authors’ acknowledgment of DEI-related signal loss is important, yet they do not discuss concrete steps to mitigate these biases in future work. The writing is generally clear and well structured, although occasionally overly descriptive and repetitive, particularly in the Introduction and Method sections. The Discussion is lengthy but could benefit from sharper differentiation between empirical findings and theoretical extrapolation.

Before publication, the authors should substantially revise the interpretation of physiological findings, strengthen methodological justification, reduce speculative language, and clarify the study’s contribution as primarily exploratory and methodological rather than confirmatory.

Author Response

Reviewer 1

Comment 1: Claims and Interpretation Too Strong

Reviewer Comment

“the manuscript exhibits substantial methodological and interpretative limitations that limit the strength of its conclusions… the discussion often drifts toward interpretative claims that exceed what the data support… the manuscript should more clearly distinguish what is actually demonstrated from what remains speculative.”

Response
We agree with the reviewer that clearer boundaries were required between empirically supported findings and theoretical interpretation. The Discussion has been revised to explicitly distinguish results directly supported by the data from speculative or conceptual framing. A new paragraph has been added clarifying what the present study can and cannot conclude, and interpretative language has been consistently tempered throughout to avoid overstatement.

Comment 2: Underpowering of fNIRS

Reviewer Comment

“A major limitation concerns the extremely small effective sample size for fNIRS (n = 12)… it severely undermines statistical power… null fNIRS findings are uninterpretable rather than informative.”

Response
We agree that the effective fNIRS sample size (n = 12) is severely underpowered for detecting the small-to-moderate effects expected in visual stress research. We have added a dedicated section (Section 5.3: Sample Size, Expected Effect Sizes, and Statistical Power) explicitly quantifying expected effect sizes (d ≈ 0.3–0.5), demonstrating that the study was only powered to detect large effects (d ≥ 0.8), and clarifying that null fNIRS findings should be interpreted as reflecting insufficient power rather than evidence of no cortical effect. Interpretation of the fNIRS results has been revised accordingly.

Comment 3: Need to Quantify Expected Effect Sizes

Reviewer Comment

“The authors should explicitly quantify expected effect sizes and justify whether the recorded sample enables valid inferences.”

Response
This issue has been addressed in the newly added Section 5.3, where expected effect sizes are derived from prior literature and formal power considerations are presented. The manuscript now clearly states that the sample size limits inferential strength and that findings should be interpreted as exploratory.

Comment 4: HRV Validity Concerns

Reviewer Comment

“HRV was extracted from a 10 Hz fNIRS PPG signal rather than ECG… widely considered insufficient… it is unwarranted to discuss ‘autonomic modulation’ without acknowledging that the analytical method lacks clinical-grade precision.”

Response
We agree and have revised the manuscript to remove strong claims of autonomic modulation. HRV findings are now explicitly framed as exploratory. A new subsection in the Methods (Section 3.6.2) clarifies the limitations of deriving HRV from fNIRS-PPG signals, and an expanded limitations section (Section 5.4.3) discusses the lack of clinical-grade precision and the necessity of ECG-based measurement for robust autonomic inference.

Comment 5: Stimulus Calibration Missing

Reviewer Comment

“luminance, contrast, and colorimetric properties of the projected stimuli are not photometrically calibrated, despite their centrality to visual stress research.”

Response
We agree that photometric calibration is important for visual stress research. The Methods (Section 2.4.2) now explicitly state that the stimuli were not photometrically calibrated. While viewing geometry and presentation conditions were held constant across stimuli, absolute luminance, contrast, and colorimetric values were not measured. The Limitations section (Section 5.3) has been expanded to bound interpretation accordingly and to identify full photometric calibration as a requirement for future studies.

Comment 6: Exposure Duration Too Short

Reviewer Comment

“exposure durations (20 seconds for most images) are too short to produce reliable haemodynamic or autonomic responses.”

Response
We respectfully disagree that 20-second exposure windows are inherently insufficient for detecting haemodynamic responses. fNIRS reliably captures stimulus-evoked visual cortical responses on this timescale, including in prior visual discomfort research. The present design was intended to probe initial stimulus-locked responses rather than sustained effects. We nevertheless acknowledge that brief exposures reduce sensitivity to prolonged haemodynamic dynamics and preclude reliable autonomic inference. This distinction is now clarified in the Methods and Limitations (Section 5.3), and longer exposure blocks are identified as a requirement for future work examining sustained cortical–autonomic coupling.

Comment 7: Cortical–Autonomic Cascade Not Tested

Reviewer Comment

“the theoretical argument for a cortical–autonomic cascade is intriguing but underdeveloped empirically… the physiological data offered here do not meaningfully test that framework.”

Response
We agree. The Discussion has been revised to explicitly separate the cortical–autonomic framework as conceptual motivation rather than an empirically tested pathway. The manuscript now clarifies that the study examines co-occurring cortical and autonomic responses rather than causal propagation, and outlines the experimental requirements needed to directly test such a cascade in future work.

Comment 8: Novel Methods Do Not Offset Foundational Limits

Reviewer Comment

“novelty does not compensate for foundational issues such as insufficient statistical power, inadequate instrumentation for HRV, absence of short-separation channels, and lack of real cortical coverage beyond V1.”

Response
We agree. The Limitations section (Section 5.3) now explicitly acknowledges insufficient power, HRV instrumentation limitations, absence of short-separation channels, and restricted cortical coverage. The manuscript has been revised to clarify that this study is a pilot investigation focused on feasibility, variance structure, and methodological integration rather than definitive physiological inference. Concrete methodological improvements are outlined for future work.

Comment 9: DEI-Related Signal Loss

Reviewer Comment

“authors’ acknowledgment of DEI-related signal loss is important, yet they do not discuss concrete steps to mitigate these biases in future work.”

Response
We agree that acknowledgment alone is insufficient. Sections 2.1 and 5.3 now outline concrete mitigation strategies, including hair-sensitive and brush-style optodes, flexible cap designs, standardized hair preparation protocols, inclusion of short-separation channels, expanded pilot testing with diverse participants, and exploration of complementary modalities less sensitive to hair and skin characteristics. We also clarify how DEI-related signal loss constrains generalisability and reflects a broader methodological challenge in the fNIRS field.

Comment 10: Writing Quality – Repetition

Reviewer Comment

“writing is generally clear… although occasionally overly descriptive and repetitive.”

Response
We have streamlined the Introduction and Methods to reduce repetition and over-description, consolidating duplicated explanations and tightening structure while preserving essential methodological detail.

Comment 11: Discussion Too Long / Drifts

Reviewer Comment

“The Discussion is lengthy but could benefit from sharper differentiation between empirical findings and theoretical extrapolation.”

Response
The Discussion has been restructured to reduce redundancy and sharpen the distinction between empirical findings, interpretation, limitations, and future directions.

Comment 12: Clarify Contribution

Reviewer Comment

“authors should… clarify the study’s contribution as primarily exploratory and methodological rather than confirmatory.”

Response
We have reframed the manuscript to explicitly position the study as exploratory and methodological, establishing feasibility and integration rather than confirmatory neurophysiological claims.

Reviewer 2 Report

Comments and Suggestions for Authors

Please address the following comments:

 

Please highlight what is genuinely new versus prior visual-discomfort work, and make the novelty explicit in the introduction of your work.

Do sample size and data loss limit power and generalizability? Please discuss.

If possible, report measured screen luminance and contrast at the eye, ambient illumination, and the computation of cycles/degree per image. Sharing the stimuli would aid replication.

Reproducibility of ViStA would benefit from releasing code and parameters. Please document tiling size/overlap, CSF weighting, any preprocessing.

HRV was derived from fNIRS. Discuss reliability of this choice and possible alternatives.

Clarify which time-domain indices were computed over the 5-min exposures.

fNIRS preprocessing is described, yet analysis uses 20 s windows from 5 min blocks and HbO only. Why not considering GLM-based modelling?

Analysing only the first 20 s of each 5-min exposure needs justification. Please discuss order fatigue effects, baseline stability across the session, and whether longer sustained windows change results for haemodynamics and HRV.

Mixed-effects analyses would benefit from reporting full formulas, random-effects structure, effect sizes with CIs, and multiple-comparison control.

For future developments, a next study with short-sep fNIRS, ECG, respiration, longer exposures, and broader cortical coverage could be suggested.

Comments on the Quality of English Language

Proofreading could be useful 

Author Response

Reviewer 2

Comment 1: Highlight Novelty

Reviewer Comment

“Please highlight what is genuinely new versus prior visual-discomfort work.”

Response
The Introduction has been revised to explicitly articulate novelty, including the integration of generative AI façade synthesis, ViStA computational analysis, and multimodal physiological measurement. Redundancy with prior publications has been reduced, with clear attribution to earlier methodological work.

Comment 2: Sample Size and Data Loss

Reviewer Comment

“Do sample size and data loss limit power and generalizability?”

Response
Yes. This is now explicitly discussed in the Methods (Section 2.1) and expanded in Section 5.3.1, detailing attrition, power limitations, and constraints on generalisability.

Comment 3: Stimulus Luminance and Spatial Metrics

Reviewer Comment

“If possible, report measured screen luminance and contrast…”

Response
As detailed above (Reviewer 1, Comment 5), photometric calibration was not performed. This limitation is now explicitly stated, and future calibration requirements are outlined.

Comment 4: Share Stimuli for Replication

Reviewer Comment

“Sharing the stimuli would aid replication.”

Response
A Data Availability statement has been added. All façade images analyzed are preserved and available from the authors upon reasonable request. Due to the generative nature of image production, exact prompt-level reproduction is not possible; however, the analyzed stimuli can be shared to support transparency and secondary analysis.

Comment 5: ViStA Reproducibility

Reviewer Comment

“Reproducibility of ViStA would benefit from releasing code and parameters.”

Response
Detailed methodological parameters are now clearly documented, and prior publications describing the ViStA framework are explicitly cited. While the ViStA code itself is under commercial confidence, foundational algorithms and parameters are fully described in the cited literature.

Comment 6: HRV Reliability

Reviewer Comment

“HRV was derived from fNIRS. Discuss reliability.”

Response
This has been addressed by reframing HRV results as exploratory, clarifying methodological limitations, and explicitly identifying ECG and respiration monitoring as requirements for future studies.

Comment 7: Clarify HRV Time-Domain Indices

Reviewer Comment

“Clarify which time-domain indices were computed.”

Response
Section 3.6.2 now specifies that HRV was quantified using a single time-domain metric (SDRR/SDNN) computed over full 5-minute exposures. Short-term and frequency-domain indices were not calculated due to signal limitations.

Comment 8: 20s Windows vs 5-min Blocks

Reviewer Comment

“Why analyze only the first 20 seconds?”

Response
The analytic rationale is now clarified in Section 3.6.1. The 10–20 s window captures the initial stimulus-locked haemodynamic response while minimizing baseline drift and attentional disengagement. Future work is identified as necessary for sustained-response modeling.

Comment 9: GLM-Based fNIRS Analysis

Reviewer Comment

“Why not GLM-based modelling?”

Response
Section 3.6.1 now explains that limited trial repetition, timing precision, and signal-to-noise ratio constrained reliable GLM estimation. GLM-based approaches are identified as appropriate for future studies with improved design.

Comment 10: Mixed-Effects Transparency

Reviewer Comment

“Mixed-effects analyses would benefit from reporting full formulas and CIs.”

Response
The statistical reporting has been expanded to clarify model structure, random effects, and reporting conventions.

Comment 11: Future Study Recommendations

Reviewer Comment

“A next study with short-sep fNIRS, ECG, respiration, longer exposures, and broader cortical coverage could be suggested.”

Response
A detailed future research roadmap has been added in the Discussion, outlining specific methodological enhancements including short-separation fNIRS, ECG and respiration monitoring, longer exposures, expanded cortical coverage, and larger samples.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The text has been improved, and I leave the final decision to the editor.

Reviewer 2 Report

Comments and Suggestions for Authors

thank you for addressing my comments