Neuroception of Psychological Safety and Attitude Towards General AI in uHealth Context

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Line 43 - We face a pervasive interest towards general AI - how this is justified?

Lines 63-67 - polyvagal theory - here a very clear cut what is proven and what is not shall be introduced. Furthermore, vagus responses are related to very basic functions, if they really are so clearly associated with higher functions involved in AI, is not clear.

NPSS scale - a context is needed, it is proofed in mass surveys via electronic forms which are very far from classical validation approaches and have therefore a completely different context. Projecting mega-mass surveys onto an individual's assessment is very risky, shall be clearly confined

 

2. Materials and Methods

 

Insofar a mass-survey was used, some characteristic of the method is needed - what can be said about participants, time span after announcement, IP addresses etc. - in methodology part, and what it means

Is the NPSS a true ePROM? Self reporting not necessarily means that patients will correctly fill the survey by themselves.

line 120 -  a scenario of AI-driven approaches in uHealth was provided as the specific context of scale application, explaining the multidisciplinary nature of uHealth at the intersection of 
healthcare, technology and sustainability - where are those approaches defined, described?

Please state clearly what exactly was surveyed by the application of the method

Line 235 -  The especially large 
value of the regression coefficient of NPSS score for social engagement - what can be said about collinearity of measurements and dependent variable to be partially defined via independent?

Table 7a - if the test is not significant, what can be said as a conclusion?

Discussion

hypothesized connection between NPS and the perception of general AI - actually the connection has been found between two scales, there is no clear characteristic of NPS. Furthermore, a mass validation has been used (not characterized persons). This is reflected in the limitations, however, Conclusions have a very high degree of uncertainty. Response times etc were not analyzed.

 

 

 

 

 

Author Response

We thank you very much for your professional feedback, constructive comments and observations, which helped us improve the manuscript and better convey our message.

Comments and Suggestions for Authors

R1.1

Line 43 - We face a pervasive interest towards general AI - how this is justified?

Answer 1.1

The sentence was rephrased to soften the tone (lines 43 – 47):

"Recent years have also seen growing interest in general AI, with high expectations for its adoption in the health sector, particularly for mobile health (mHealth) and connected or ubiquitous health (uHealth), whose potential to transform health services is recognized [3–8], including the psychological and mental health support [7,9–11]."

 

R1.2

Lines 63-67 - polyvagal theory - here a very clear cut what is proven and what is not shall be introduced. Furthermore, vagus responses are related to very basic functions, if they really are so clearly associated with higher functions involved in AI, is not clear.

NPSS scale - a context is needed, it is proofed in mass surveys via electronic forms which are very far from classical validation approaches and have therefore a completely different context. Projecting mega-mass surveys onto an individual's assessment is very risky, shall be clearly confined

Answer 1.2

An additional paragraph has been added to provide the context of our investigation (lines 87 – 91):

"We hypothesized that individual differences in baseline neuroceptive subjective safety orientation would modulate responses to novel or ambiguous technological agents. That is to say, even if the safety of interaction with the AI itself cannot be measured, NPS scores would reflect individual attitudes towards AI and its perceived beneficial or harmful role in human society."

 

R1.3

Materials and Methods

Insofar a mass-survey was used, some characteristic of the method is needed - what can be said about participants, time span after announcement, IP addresses etc. - in methodology part, and what it means

Is the NPSS a true ePROM? Self reporting not necessarily means that patients will correctly fill the survey by themselves.

line 120 -  a scenario of AI-driven approaches in uHealth was provided as the specific context of scale application, explaining the multidisciplinary nature of uHealth at the intersection of healthcare, technology and sustainability - where are those approaches defined, described?

Please state clearly what exactly was surveyed by the application of the method

Answer 1.3

Additional information regarding the survey application has been provided in lines 111 – 112:

"No identifying data were collected, nor were any IP addresses tracked;"

A description of the NPSS-scenario has been provided in lines 128 – 137:

"For the application of NPSS, participants were asked to consider a standardized scenario involving the use of AI-based connected health systems (namely, uHealth) in a healthcare setting. This scenario described digital health services in which AI tools support care delivery by analyzing health data and providing recommendations, alerts, or risk assessments to aid clinical decision-making or self-management. The AI system was presented as a decision support system rather than a standalone system, operating within existing care workflows and not replacing healthcare professionals. Participants were asked to answer the NPSS questions based on their anticipated experience of using, or being supported by, such AI-based uHealth systems, rather than on interpersonal interaction with a human agent."

The section of limitations has been carefully revised to correctly reflect the shortcomings of this survey-based investigation.

 

R1.4

Line 235 -  The especially large value of the regression coefficient of NPSS score for social engagement - what can be said about collinearity of measurements and dependent variable to be partially defined via independent?

Answer 1.4

The sensitivity analysis has been extended to address all issues.

A paragraph has been added to the section of methods/ data analysis (lines 178 – 182):

"Sensitivity/robustness analysis was conducted considering two approaches: (a) separate univariate multiple linear regressions for the two dependent variables in SEM models with multicollinearity testing based on the variance inflation factor (VIF); (b) alternative SEM models with gender as a grouping factor (namely, rather than introducing gender as a covariate in the SEM model)."

The section of results has a separate sub-subsection presenting the results of univariate multiple linear regression (MLR; lines 300 – 309):

"3.4.1. Univariate multiple linear regression analysis

Univariate multiple linear regression confirmed the regression results for AIAS4, namely the strength and significance of NPSS scores for social engagement and compassion, and significant contributions of covariates such as demographic cohort, level of education and gender. In addition, the regression results for the outcome AI as a threat confirmed the nonsignificant results of the covariance-based SEM analysis.

The VIF values ranged from 1.08 to 2.06, meaning there was no collinearity of the independent variables in the regression models.

The R code for this analysis and the results in full are provided as supplementary material."

Supplementary material with the R code and the results of univariate MLR has been provided.

 

R1.5

Table 7a - if the test is not significant, what can be said as a conclusion?

Answer 1.5

Comments on the sensitivity analysis, namely the results of gender-split SEM, are provided in sub-subsection 3.4.2. Referring to the results in Table 7a, the narrative text in lines 314 – 318 read:

"Although some of the fit indices are slightly better in this approach (such as CFI, RMSEA, SRMR; Table 7a), the chi-square test of the fitting and the Vuong test showed that introducing the gender in the model (namely, the previous model with gender included in the regression equations) would be a significantly better fit for the actual data (Likelihood ratio LR=66.417, p=0.008)."

 

R1.6

Discussion

hypothesized connection between NPS and the perception of general AI - actually the connection has been found between two scales, there is no clear characteristic of NPS. Furthermore, a mass validation has been used (not characterized persons). This is reflected in the limitations, however, Conclusions have a very high degree of uncertainty.

Answer 1.6

The whole section of discussions has been carefully revised.

The conclusions have been completely rephrased (lines 480 – 485):

"In the context of ubiquitous healthcare solutions, neuroception-related dimensions appear to shape general attitudes towards AI, whereas explicit perception of AI as a threat may be driven by distinct and more reflective processes. The significant covariance between these measures indicates conceptual relatedness without redundancy. Gender, demographic generation, and level of education are covariates which significantly impact the general attitudes, rather than explicit perception of threat."

 

R1.7

Response times etc were not analyzed.

Answer 1.7

Response times have been mentioned as a limitation (453 – 455):

"Moreover, the temporal distribution of responses was highly skewed (nearly half were received in the first four days following the public announcement of the questionnaire)."

We appreciated that further analysis of the responses' temporal distribution would not bring meaningful insight into our investigation, given the convenience sample and the limited sample size.

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript explores a novel and timely research question by linking the Neuroception of Psychological Safety (NPS) with attitudes toward General AI in the context of uHealth. The attempt to apply a physiological state of safety to explain technology acceptance offers a fresh perspective compared to traditional cognitive models. However, there are significant concerns regarding the theoretical foundation, instrument validity in this specific context, sampling bias, and the interpretation of statistical results. These issues need to be addressed to ensure the academic rigor of the study.

1. While the authors utilize Polyvagal Theory as the central theoretical framework, it is crucial to acknowledge that this theory is not universally accepted in the scientific community, a limitation the authors briefly mention but do not sufficiently integrate into their analysis. The interpretation of the SEM results appears overly deterministic regarding the biological pathways suggested by the theory. The authors should tone down the definitive causal claims derived from this theory. The discussion should more balanced, explicitly recognizing the theoretical controversy earlier in the paper, not just in the limitations section.
2. The Neuroception of Psychological Safety Scale (NPSS) was originally developed to measure safety in interpersonal or social contexts. The current study applies this scale to Human-AI interaction by simply providing a scenario/context. It is not immediately clear that the construct of "social engagement" or "compassion" as measured by NPSS is valid for interaction with a non-human agent (AI) without more rigorous validation. The authors need to provide a stronger justification or evidence for the construct validity of NPSS in a non-human, uHealth context. A discussion on how "social engagement" translates to "AI usage" beyond a metaphorical level is needed.
3. The study relies on a convenience sample (N=201) that raises concerns about representativeness and reliability. The sample is heavily skewed towards females (63.7%) and younger generations (Gen Z) who hold disproportionately high education levels (e.g., Gen Z with Master's degrees seems unusually high or requires explanation regarding the coding "4 = master degree" for this age group). In addition, Table 3 reports an Intraclass Correlation Coefficient (ICC) of 0.428 for the total NPSS, which is below the authors' own stated threshold of 0.5 for "moderate to good consistency". The low ICC suggests potential issues with the consistency of the instrument in this specific sample. The authors must address how this low reliability impacts the robustness of the SEM results. The limitations of the non-probability sampling and the specific imbalances (education/age) should be discussed more critically.
4. The results show a significant negative covariance (-2.152) between AIAS4 (Attitude/Benefit) and AI Perceived Threat. The authors interpret this as a "dual perception" where benefits and threats are not mutually exclusive. However, given the measurement structure, this might simply reflect scale inverse relationships or ambivalence. Thus, a more in-depth analysis is required. Is this relationship an artifact of the scale design, or does it represent a psychological "attitudinal ambivalence"?
5. The standard deviation for "AI Perceived Threat" (3.17) is notably larger than for AIAS4 (2.59), indicating high polarization. Furthermore, the correlation structure among the three NPSS factors needs clearer explanation in the context of the model. The authors need to elaborate on why the variance for the Threat variable is so high.
6. In the Discussion, the interpretation of gender differences relies heavily on biological essentialism and gender stereotypes. Specifically, the statement that "women, with more developed frontal lobes... are more sociable but also more likely to perceive the cognitive level... as less satisfying" lacks sufficient scientific backing in the context of AI interaction and appears to be an over-generalization. Interpretations of gender differences should be grounded in recent, empirical evidence regarding technology acceptance and gender, rather than speculative biological differences. I strongly recommend revising this section to avoid reinforcing gender stereotypes without robust evidence.

 

Author Response

We thank you very much for your professional feedback, constructive comments and observations, which helped us improve the manuscript and better convey our message.

Comments and Suggestions for Authors

R2.0

While the authors utilize Polyvagal Theory as the central theoretical framework, it is crucial to acknowledge that this theory is not universally accepted in the scientific community, a limitation the authors briefly mention but do not sufficiently integrate into their analysis. The interpretation of the SEM results appears overly deterministic regarding the biological pathways suggested by the theory. The authors should tone down the definitive causal claims derived from this theory. The discussion should more balanced, explicitly recognizing the theoretical controversy earlier in the paper, not just in the limitations section.

Answer 2.0

We have carefully revised the whole section of discussions. The conclusions were completely rephrased.

The theoretical controversy over the polyvagal theory has been explicitly stated for the first time in the section of methods (lines 124 – 127) and three additional refences were added:

"Although the polyvagal framework remains debated on mechanistic grounds [30–33], the NPSS has been psychometrically validated as a subjective measure of perceived safety, the level at which it is used in the present study."

 

R2.1

The Neuroception of Psychological Safety Scale (NPSS) was originally developed to measure safety in interpersonal or social contexts. The current study applies this scale to Human-AI interaction by simply providing a scenario/context. It is not immediately clear that the construct of "social engagement" or "compassion" as measured by NPSS is valid for interaction with a non-human agent (AI) without more rigorous validation. The authors need to provide a stronger justification or evidence for the construct validity of NPSS in a non-human, uHealth context. A discussion on how "social engagement" translates to "AI usage" beyond a metaphorical level is needed.

Answer 2.1

An additional paragraph has been added to provide the context of our investigation (lines 87 – 91):

"We hypothesized that individual differences in baseline neuroceptive subjective safety orientation would modulate responses to novel or ambiguous technological agents. That is to say, even if the safety of interaction with the AI itself cannot be measured, NPS scores would reflect individual attitudes towards AI and its perceived beneficial or harmful role in human society."

A more informative description of the NPSS-scenario has been provided in lines 128 – 137.

We have carefully revised the whole section of discussions.

 

R2.2

The study relies on a convenience sample (N=201) that raises concerns about representativeness and reliability. The sample is heavily skewed towards females (63.7%) and younger generations (Gen Z) who hold disproportionately high education levels (e.g., Gen Z with Master's degrees seems unusually high or requires explanation regarding the coding "4 = master degree" for this age group). In addition, Table 3 reports an Intraclass Correlation Coefficient (ICC) of 0.428 for the total NPSS, which is below the authors' own stated threshold of 0.5 for "moderate to good consistency". The low ICC suggests potential issues with the consistency of the instrument in this specific sample. The authors must address how this low reliability impacts the robustness of the SEM results. The limitations of the non-probability sampling and the specific imbalances (education/age) should be discussed more critically.

Answer 2.2

The low reliability of the combined NPSS was addressed in lines 206 – 208. The whole paragraph reads:

"The reliability and reproducibility of the scales' measurements are presented in Table 3. The combined three-factor NPSS resulted in a rather low ICC in this dataset, therefore the three factors were considered separately in the multivariate SEM analysis, and no subsequent analysis was conducted on the combined NPSS, apart from descriptive statistics."

The section of limitations has been carefully revised.

 

R2.3

The results show a significant negative covariance (-2.152) between AIAS4 (Attitude/Benefit) and AI Perceived Threat. The authors interpret this as a "dual perception" where benefits and threats are not mutually exclusive. However, given the measurement structure, this might simply reflect scale inverse relationships or ambivalence. Thus, a more in-depth analysis is required. Is this relationship an artifact of the scale design, or does it represent a psychological "attitudinal ambivalence"?

Answer 2.3

We have carefully revised the whole section of discussions; the significant negative covariance (-2.152) between AIAS4 (attitude/benefit) and AI perceived threat is addressed multiple times in slightly different contexts.

The narrative text in the section of results which refers to the covariance (lines 269 – 273) has been revised and now reads:

"This significant relationship would suggest a partial interdependence between base-line safety orientation and explicit threat judgements. It should also be noted that, as a single item construct, perceived threat has a higher variance than the four-item AIAS4 (as presented in Tables 4 and 5)."

 

R2.4

The standard deviation for "AI Perceived Threat" (3.17) is notably larger than for AIAS4 (2.59), indicating high polarization. Furthermore, the correlation structure among the three NPSS factors needs clearer explanation in the context of the model. The authors need to elaborate on why the variance for the Threat variable is so high.

Answer 2.4

The comment on standard deviation is included in lines 269 – 273 (previous answer). As a single item construct, the "AI threat" is expected to have a higher variance than the 4-item AIAS-4. On the other hand, the variance is indeed high, therefore we additionally noted it in lines 287 – 288 of the results. Apart from these notifications, we appreciated that, for our investigation, further elaboration would be too speculative. However, the significant relationship between the AIAS-4 scores and "AI threat" is addressed multiple times in the present version of discussions.

 

R2.5

In the Discussion, the interpretation of gender differences relies heavily on biological essentialism and gender stereotypes. Specifically, the statement that "women, with more developed frontal lobes... are more sociable but also more likely to perceive the cognitive level... as less satisfying" lacks sufficient scientific backing in the context of AI interaction and appears to be an over-generalization. Interpretations of gender differences should be grounded in recent, empirical evidence regarding technology acceptance and gender, rather than speculative biological differences. I strongly recommend revising this section to avoid reinforcing gender stereotypes without robust evidence.

Answer 2.5

We dropped those statements. The only reference to biological differences is in lines 426  - 428:

"Differences related to education and demographic cohorts can be readily explained by simple exposure, but the gender difference may also have a biological component [24,43]."

 

In addition, the language has been thoroughly revised to better convey the professional message of the manuscript.

Reviewer 3 Report

Comments and Suggestions for Authors

The authors present an important study on "Neuroception of Psychological Safety and Attitude Towards General AI in uHealth Context". The results are truly valuable for researchers in this field.

To improve the quality of the manuscript, the authors should explain in detail:

1. What was the motivation for the study, from the related knowledge gap to the description of specific situations in Timisoara, Romania?

2. The neuroception of psychological safety (NPS) used is a well-recognized and psychometrically sound tool for measuring individual sense of safety. Some authors present it as an ideal tool because it correlates with measures of well-being, trauma, and team safety. In this case, the results presented (based on social interaction and compassion) do NOT explore these correlations. Why?

3. Based on the results obtained by applying NPS, the authors should explain whether or not it is important to analyze risks and understand the detection of generated threats.

4. In Table 1, the education level 0 (no education) should be clarified. Are they illiterate or ignorant?

Author Response

We thank you very much for your professional feedback, constructive comments and observations, which helped us improve the manuscript and better convey our message.

Comments and Suggestions for Authors

R3.1

1. What was the motivation for the study, from the related knowledge gap to the description of specific situations in Timisoara, Romania?

Answer 3.1

A new, rephrased paragraph describes the starting hypothesis and motivation behind this investigation (lines 87 – 91):

"We hypothesized that individual differences in baseline neuroceptive subjective safety orientation would modulate responses to novel or ambiguous technological agents. That is to say, even if the safety of interaction with the AI itself cannot be measured, NPS scores would reflect individual attitudes towards AI and its perceived beneficial or harmful role in human society.

We opted not to describe situations specific to Timisoara in order to preserve the generalizability of our findings, given all the limitations related to convenience sampling and anonymous nature of the survey.

 

R3.2

2. The neuroception of psychological safety (NPS) used is a well-recognized and psychometrically sound tool for measuring individual sense of safety. Some authors present it as an ideal tool because it correlates with measures of well-being, trauma, and team safety. In this case, the results presented (based on social interaction and compassion) do NOT explore these correlations. Why?

Answer 3.2

Exploring the correlations with well-being, trauma, and team safety would require a different approach in sampling and data collection.

 

R3.3.

3. Based on the results obtained by applying NPS, the authors should explain whether or not it is important to analyze risks and understand the detection of generated threats.

Answer 3.3

We have added a new paragraph into the section of discussions (lines 439 – 445):

"We also acknowledge the relevance and importance of risk analysis related to the deployment of AI-based uHealth. However, our investigation did not aim at threat detection or risk analysis; our results are limited to user perception and acceptance. Technical risk detection and human perceived safety or security remain two distinct, complementary layers: acceptance and trust are influenced by factors not reducible to detected risks alone. The NPSS-informed findings highlight the need to address perceived safety alongside technical safety."

 

R3.4

4. In Table 1, the education level 0 (no education) should be clarified. Are they illiterate or ignorant?

Answer 3.4

Corrections were made in Tables 1 and 5.

Additional information was provided in the section of materials and methods, subsection "2.1. Tools for data collection: scales and variables" (last paragraph, lines 151 - 152), which now reads:

"The level of education was quantified on five levels (from 0 = basic formal education to 4 = master degree, namely at least five-year academic programs after high school)."

 

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have substantially improved the manuscript and addressed many of the concerns raised in the previous review. Nevertheless, several conceptual issues remain insufficiently resolved and warrant further clarification before publication.

First, regarding the use of the NPSS in a Human–AI context, the authors reframed their hypothesis by arguing that the scale does not assess the safety of AI interaction per se, but rather individuals’ baseline neuroceptive orientation toward safety, which in turn shapes attitudes toward AI. While this reframing represents a conceptual reconceptualization, it does not constitute evidence of construct validity in a non-human context. The claim that NPSS captures “attitudes toward AI” remains largely post hoc. Importantly, the mechanism by which biologically grounded constructs such as “social engagement” or “compassion,” originally developed for interpersonal contexts, translate into attitudes toward a non-biological agent remains speculative. This limitation should be stated more explicitly and treated as a substantive conceptual boundary of the study rather than a resolved issue.

Second, concerns regarding sample composition remain only partially addressed. In particular, the unusually high proportion of Gen Z participants holding Master’s degrees is not clearly explained. It remains unclear whether this reflects the sampling frame, coding decisions, or recruitment procedures. Moreover, the implications of this demographic skew for the interpretation and generalizability of the findings are not fully integrated into the theoretical discussion and remain confined largely to the limitations section.

Finally, with respect to gender differences, the removal of overtly essentialist claims (e.g., references to frontal lobe development and sociability) represents a clear improvement. However, the remaining assertion that gender differences “may also have a biological component” is insufficiently substantiated in the context of AI acceptance and use. Without direct empirical evidence or engagement with contemporary gender and technology research, this statement remains speculative and risks reintroducing a form of biological reductionism.

Author Response

We thank you very much for your professional feedback, constructive comments and advice, which have enabled us to improve the manuscript and its message.

Comments and Suggestions for Authors

R2.1

First, regarding the use of the NPSS in a Human–AI context, the authors reframed their hypothesis by arguing that the scale does not assess the safety of AI interaction per se, but rather individuals’ baseline neuroceptive orientation toward safety, which in turn shapes attitudes toward AI. While this reframing represents a conceptual reconceptualization, it does not constitute evidence of construct validity in a non-human context. The claim that NPSS captures “attitudes toward AI” remains largely post hoc. Importantly, the mechanism by which biologically grounded constructs such as “social engagement” or “compassion,” originally developed for interpersonal contexts, translate into attitudes toward a non-biological agent remains speculative. This limitation should be stated more explicitly and treated as a substantive conceptual boundary of the study rather than a resolved issue.

Answer 2.1

We framed this limitation as a conceptual boundary of the study and explicitly stated that the observed associations should be interpreted as exploratory and contextual (lines 470 to 479):

"More specifically, while the NPSS scale was designed to capture the biologically grounded dimensions of perceived safety in interpersonal contexts, this study does not establish a mechanism by which NPSS dimensions such as social engagement or compassion translate into attitudes towards a non-human agent. The observed associations should therefore not be interpreted as evidence that these concepts operate towards AI in an interpersonal or social sense, but rather as an indication that individual differences in baseline safety orientation can modulate evaluative responses to AI-based health technologies. Consequently, the application of the NPSS scale in this context is exploratory and limited, and does not constitute validation of the scale for human-AI interaction as such."

 

R2.2

Second, concerns regarding sample composition remain only partially addressed. In particular, the unusually high proportion of Gen Z participants holding Master’s degrees is not clearly explained. It remains unclear whether this reflects the sampling frame, coding decisions, or recruitment procedures. Moreover, the implications of this demographic skew for the interpretation and generalizability of the findings are not fully integrated into the theoretical discussion and remain confined largely to the limitations section.

Answer 2.2

Implications of the demographic skew were added into the discussions (lines 433 to 444):

"In this context, the observed effect of education level should be interpreted in light of the sample's generational composition. The relatively high proportion of young par-ticipants (namely, Generations Y and Z) with advanced university degrees likely reflects educational pathways specific to these cohorts and their greater familiarity with digital technologies, rather than formal education acting as an isolated determinant of attitudes towards AI. Consequently, the association between education and AI acceptance ob-served in this study could reflect, at least in part, unevenly distributed differences in functional digital health skills and digital literacy across generations. This demographic configuration suggests that the role of education in shaping attitudes towards AI-based health interventions depends on broader socio-technical contexts and warrants caution when extrapolating the magnitude of this effect to populations where formal education is less closely linked to digital skills or familiarity with AI."

Additional information was added into the subsection of limitations (line 458):

"Primarily due to the recruitment procedure, the sample was unbalanced with respect to gender and age relative to reported level of education."

 

R2.3

Finally, with respect to gender differences, the removal of overtly essentialist claims (e.g., references to frontal lobe development and sociability) represents a clear improvement. However, the remaining assertion that gender differences “may also have a biological component” is insufficiently substantiated in the context of AI acceptance and use. Without direct empirical evidence or engagement with contemporary gender and technology research, this statement remains speculative and risks reintroducing a form of biological reductionism.

Answer 2.3

The assertion regarding the biological component of gender differences has been removed.