An Eye-Tracking Study of Pain Perception Toward Faces with Visible Differences

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The authors situate their work within stigma and social perception research, arguing that visible facial differences (VFDs) draw disproportionate visual attention and may disrupt the processing of socially relevant cues. They cite the need to extend prior eye-tracking work by examining how VFDs modulate perception of pain expressions—an emotion category that relies heavily on internal facial features. The study uses a within-subjects eye-tracking paradigm with 44 university students. Participants viewed standardised facial stimuli (pain vs. neutral) either with or without an added VFD. Gaze behaviour was quantified using dwell times on predefined areas of interest (internal vs. peripheral features), and participants provided pain-intensity ratings for each face. The presence of a VFD reliably redirected gaze away from internal features and towards peripheral regions, replicating known attentional capture effects. Contrary to preregistered predictions, VFD faces were rated as experiencing more pain, independent of expression. No correlations emerged between gaze patterns and pain judgments, suggesting dissociation between visual attention and evaluative outcomes.

 

Introduction

 

I feel the introduction fails to set the context for the paper as a whole. What are the theoretical underpinnings that the authors predict underly the effects of VFD in terms of attentional and eyetracking patterns, and in terms of pain perception? How do these effects relate to broader effects of social prejudice and ostracisation that individuals with VFD face? As pain perception is often used in empathy research, this may also be worth exploring: do VFD faces elicit more or less empathy? Or conversely, are the authors looking to test whether the ability of viewers to discriminate facial emotion is impaired in VFD faces? This all remains unclear from the introduction and calls for significant editing.

 

L34-35: Review English. What is the “expected norm”?

 

L37-38: Very vague, be more specific than “social environment”. What factors specifically are being overlooked?

 

L38: “recent work” Was this a review, a meta-analysis, or a article? What was the context for moving towards a more macro view of VFD?

 

L48: “tainted” – subjective + often pejorative term, please change to language more suitable for research study

 

L54: Again, article could do with a brief review by a native English speaker. “more quickly” not “quicker”.

 

L54: “the area”: what area of interest? How was the area defined?

 

L54: If citing “evidence”, references should be in the form of an article, not review.

 

L61: Is this an “assumption” or a theory/explanation in the literature? Not clear. If there is an underlying assumption behind attentional capture in the context of VFD, this should be laid out in more detail.

 

L64 + 69: Standardise the terminology here please. Is “over-gazing” or “over-attention” the more accepted term in the literature?

 

L71: Are these findings “seminal”? This wasn’t made clear in the previous paragraph.

 

L73: See my original comment with regards to the introduction. What “affects” are you referring to? Pain v. pleasure? Aversion v. attraction?

 

L84: Why is pain the focal point of the study and one of the main measures by which response to VFD faces are studied? This should have been the main focus of the preceding paragraphs, leading to a clear rationale for selecting this method.

 

L98: “hypothesised” not “assumed”. These hypotheses need to be more specific (ie. What direction of effect were you expecting – higher pain evaluation for VFD or for control faces? Less gaze time to internal AOIs v peripheral AOIs? Positive correlation between gaze time to internal AOIs and pain evaluation?

 

Methods + Results

 

I see from the stimuli the authors provided that these VFDs were in the form of facial scars. This should be explicitly stated and justified in the context of the study aims.

 

L126: “VFD were added” – were these digitally added or done in makeup? Was the location of the VFD counterbalanced, given the methodology behind the division of AOIs into 7 different face quadrants (3 internal, 4 peripheral)? Were an equal number of VFDs in internal as peripheral AOIs? I see there was a variance in location from the experimental images provided but it should be made clear whether this was effectively controlled for.

 

If your stated goal was to test whether peripheral VFDs bias attention away from internal areas of the face, then vary peripheral quadrants only. Bear in mind that this may undermine your ecological validity as surely there is an equal chance of injuring your internal as external face regions?

 

L160: Would it not make more sense to run a combined analysis of both peripheral and internal AOIs with location as an included factor? Or were both dependent variables (dwell time – Peripheral, dwell time – Internal) included in the same ANOVA? It is not clear from your description here, especially as:

 

L178: You mention an interaction between the two DVs here. How is this possible if two separate repeated-measures ANOVAs were conducted?

 

L182: Table 1 is very confusing for the reader both in terms of its content and formatting. You have external columns for “Expression type, Neutral, Painful + All” which don’t correspond to the actual content of the table. You also have internal columns for “AoI + Face type”, but “Pain rating” isn’t an AoI! I would not include this table in the article at all as it doesn’t provide much useful information beyond the later graphs. But even as an annex, this needs to be better edited and formatted – it is a real mess of information.

 

L186: From Figure 2 it looks like your ppts spent longer looking at internal AOIs in faces without VD than with VD but you say: “As expected, participants looked longer at internal features when the face had VFD than when it did not have VFD”. What is going on here?

 

Also from your introduction, the rationale if not the stated hypothesis was that individuals spend less time looking at the internal areas of the face as the VFD captures attention so how could this result possibly be expected?

 

Figure 2. You found a main effect of VFD vs control faces for gaze time to the internal AOI, independent of expression. So show it! Have a graph with VFD vs control across all faces and make it clear this is a significant difference in the graph (* asterisk). Then show a graph with the other main effect between painful vs normal expressions for gaze time to the internal AOI, independent of face type. Again make clear where the significant difference lies.

 

L202: What are the two dependent variables? See comment for L160.

 

Figure 3. From the figure and the mean error bars included, this does not look like a significant difference between pain rating for the VFD vs normal faces. Again, a graph comparing VFD vs normal across all expressions would help to show the reader at a glance what significant effects your study found. Include brackets and asterisks to show statistical significance between bars.

 

Table 2. Once again, this formatting is very poor and unclear as to what the data represents. The “Pain rating” label is very misleading as this should be placed above the second two columns. The labels for “Gaze duration” in the rows is very difficult to parse at a glance. This may essentially be left out as the correlations are essentially randomly distributed around 0. Instead you can add to the sentence “None of the expected correlations were evidenced” …as per our previous hypothesis of a significant positive correlation between pain rating and gaze time to the internal features of the face (or something to that effect).

 

Discussion

 

The key issue here is that it’s not clear what you set out to investigate in the first place: 1) If individuals attend less to internal facial features in VFD faces and consequently are less able to discriminate between pained and non-pained faces? 2) If individuals are less empathetic towards faces with VFDs than non-VFD faces? 3) More empathetic?

 

All of these rationales for conducting the study should be made clear in the introduction as mentioned previously. If it is the first rationale, why use pain as an independent variable in the first place if there is an obvious confound with facial scarring inducing pain? Why not use a face expression categorisation or face memory task, where you would also expect performance to decrease given the attentional capture of the VFD?

 

L214-215: From L186 “As expected, participants looked longer at internal features when the face had VFD than when it did not have VFD”.

Now you state: “participants spent less time on internal facial features but looked for longer duration at peripheral areas when the face had VFD than did not have VFD”.

 

Which one is it?

 

L215-217: “Moreover, the results revealed a moderating effect of facial expression, as the visible difference (in the peripheral area) was gazed for shorter duration when the face displayed pain.” But there was no significant interaction here, so this may simply be speculation?

 

L218: What do you mean “overestimated”? Overestimated the pain ratings compared to what? Non-VFD faces? It may have been helpful to initially validate the pain ratings for set of VFD and non-VFD faces separately with a small number of participants. It may simply be the case as you later state that by comparing the two, the VFD faces receive a more “sympathetic” pain rating than non-VFD, or that non-VFD faces receive a lower pain rating than they would otherwise. In this case, you can’t state that there is an “overestimation” or “underestimation” of pain ratings, only that one is rated more highly than the other.

 

L221: The authors cite several studies in social psychology describing patterns of behaviour and social response to VFDs but which may not be possible to replicate or even investigate using methods in visual perception and emotion research. This is fine for the initial “macro” view in the Introduction but not for analysing the findings in the Discussion. Please include more relevant studies in eyetracking and visual perception.

 

L257: This is one of the few references the authors make directly to empathy for pain. Both potential readers and this reviewer could be forgiven for thinking that this would be the whole rationale behind the study as a whole, given the importance of pain perception in research into empathy and other aspects of theory of mind. I would urge the authors to review this evidence more carefully in the context of their study, and whether the focus of the study is on empathy to pain in VFDs or on impaired discrimination of facial information due to attentional capture as mentioned previously. Make this focus clearer to the reader!

 

L271: The authors state: the (VFD) “stimuli were mainly scars”. I could only see scars among the experimental stimuli provided. What other VFD exemplars were used? If there were different types of VFD this should be made clear in the Methods earlier.

Comments on the Quality of English Language

Could do with revision by native English speaker

Author Response

[Comment_1] I feel the introduction fails to set the context for the paper as a whole. What are the theoretical underpinnings that the authors predict underly the effects of VFD in terms of attentional and eyetracking patterns, and in terms of pain perception? How do these effects relate to broader effects of social prejudice and ostracisation that individuals with VFD face? As pain perception is often used in empathy research, this may also be worth exploring: do VFD faces elicit more or less empathy? Or conversely, are the authors looking to test whether the ability of viewers to discriminate facial emotion is impaired in VFD faces? This all remains unclear from the introduction and calls for significant editing.

 

Response: We thank the reviewer for this important comment, which helped us clarify the theoretical positioning of the paper. We have revised the Introduction to better clarify the theoretical context of the study. The revised version now more explicitly situates our hypotheses within prior eye-tracking research on visual attention to faces with VFD and discusses how altered gaze patterns may relate to social evaluation and stigmatizing reactions.

Regarding pain perception, we clarify that the present study examines observers’ visual exploration of faces and their pain intensity ratings, rather than empathic responses or emotion recognition accuracy.

Although empathy is a highly relevant framework in pain perception research, it was not directly measured in the present study; we therefore address its potential role as an interpretative perspective in the Discussion rather than as a tested construct. Overall, these changes aim to clarify the scope of the study and its contribution to understanding social perception of faces with VFD.

 

Modification:

• Introduction: see the modified version of the Introduction (modifications are marked in red).
• Discussion: We added discussion points: “As such, focusing on the VFD might bias empathy for pain, i.e., the ability to share and understand the pain experience of someone else (Goubert et al., 2005). Importantly, however, empathy was not directly assessed in the present study, as participants were not instructed to engage in perspective-taking or affect sharing. Any role of empathy should therefore be interpreted as an indirect, downstream mechanism rather than a primary explanatory factor. One possibility is that reduced attention to pain-diagnostic facial features in faces with VFD limits access to individuating emotional information, which may constrain cognitive empathy (perspective-taking). Prior eye-tracking work suggests that attention to internal facial features, particularly the eyes, supports accurate emotion decoding and perspective-taking (Cowan et al., 2014). However, the present results indicate that atypical gaze allocation did not translate into lower pain ratings. Instead, participants systematically rated pain higher in faces with VFD, suggesting that judgments relied less on perceptual input and more on top-down expectations. In the context of disability, visible differences are often associated with stereotypes of vulnerability and suffering, which can shape empathic responses independently of actual emotional expressions (Nario-Redmond et al., 2019). Recent work shows that such stereotypes may lead to biased or paternalistic forms of empathy, characterized by lower affective sharing (Granjon et al., 2023). Notably, Granjon and colleagues (2023) demonstrated that early affective empathy, as measured via an electro-encephalography (EEG) paradigm, and which can also be interpreted as reflecting attentional biases, was altered when participants viewed faces with disabilities, whereas cognitive empathy remained unaffected. This raises the intriguing possibility that the attentional biases captured via eye-tracking in our study could reflect a similar early, automatic component of empathy, linking perceptual-attentional processes with affective responses to VFD. Future research combining eye-tracking with EEG markers of early and late empathic components would help to disentangle these processes and clarify how attention and empathic processing interact in the social perception of VFD.” (l.287-313)

 

 

[Comment_2] L34-35: Review English. What is the “expected norm”?

 

Response: As requested we carefully reviewed English.

 

Modification:

• Introduction: We clarified: “These facial differences diverge from both beauty standards and social expectations of facial features. Due to what could be perceived as diverging from the norm, VFD is still frequently negatively viewed by others, potentially leading to social difficulties (e.g., see Sarwer et al., 2022).” (l.34-36).

 

 

[Comment_3] L37-38: Very vague, be more specific than “social environment”. What factors specifically are being overlooked?

 

Response: We clarified this point.

 

Modification:

• Introduction: We changed for “public reactions to VFD” (l.39).

 

 

[Comment_4] L38: “recent work” Was this a review, a meta-analysis, or a article? What was the context for moving towards a more macro view of VFD?

 

Response: Actually, this is a narrative review that highlights that the experience of social difficulties constitutes a challenge common to individuals with a VFD, regardless of the specific condition. These difficulties may arise in interactions with others, whether through spontaneous reactions to the perception of facial difference, during social interactions themselves, or in the management of disclosure and explanation of the VFD. This is not to deny the existence of other difficulties, such as functional impairments that may be associated with certain VFDs (e.g., those resulting from cancer), but rather to emphasize that managing a VFD also involves dealing with the reactions of the social environment (i.e., people in first encounters, but also people met occasionally, on a regular basis, etc.) to a situation that does not directly affect it. In this respect, the authors demonstrated that sociocultural factors play a significant role in the lived experience of facial difference and should not be overlooked.

To illustrate the relevance of adopting a more macro-level perspective, the authors draw a parallel with models of disability. Taking into account the impact of barriers arising from the social environment (as emphasized in the social model of disability) goes beyond the idea that the challenges associated with VFD are limited to the existence of an essentialized specific pathology (as in the biomedical model of disability). In other words, the difficulties associated with having a VFD are not necessarily inherent to the condition itself, but may also stem from others’ reactions to that condition.

 

Modification:

• Introduction: We added more details: “So as to better integrate the importance of the social challenges in the lived experience of people with visible differences, a recent narrative review called for moving beyond an individual lens towards a socio-structural lens (Harcourt et al., 2025).” And “In line with this recommendation, this study draws on a social stigma framework (Bos et al., 2013; Goffman, 1963) to investigate public reactions to VFD from the perspective of people who do not have a VFD.” (l.39-l.44).

 

 

[Comment_5] L48: “tainted” – subjective + often pejorative term, please change to language more suitable for research study

 

Response: As requested, we changed for more suitable language.

 

Modification:

• Introduction: we changed for “marked” (l.50).

 

 

[Comment_6] L54: Again, article could do with a brief review by a native English speaker. “more quickly” not “quicker”.

 

Response: As requested, and as mentioned above, we carefully reviewed English of the entire manuscript by a native English speaker.

 

Modification:

• Introduction: we changed for “more quickly” (l.56).
• Throughout the document: modifications for improvement of English language.

 

 

[Comment_7] L54: “the area”: what area of interest? How was the area defined?

 

Response: Asaad et al. (2020) conducted a systematic review of empirical studies using eye-tracking to investigate gaze behavior toward various visible facial differences (VFDs). In this literature, the term “area” refers to predefined areas of interest. However, to avoid ambiguity, we decided to remove the term “area” from the manuscript and clarify the wording accordingly.

 

Modification:

• Introduction: we clarified: “Evidence consistently shows that, compared with equivalent faces without VFD, observers direct their gaze more quickly and for longer durations toward VFD.” (l.54-57).

[Comment_8] L54: If citing “evidence”, references should be in the form of an article, not review.

 

Response: We respectfully note that Asaad et al. (2020) is a systematic review synthesizing converging empirical evidence on gaze behavior toward visible facial differences. To clarify this point and avoid any ambiguity, we revised the wording to explicitly refer to a systematic review of eye-tracking studies.

 

Modification:

• Introduction: we clarified: “for a systematic review of eye-tracking studies, see Asaad et al., 2020)” (l.56-57).

 

 

[Comment_9] L61: Is this an “assumption” or a theory/explanation in the literature? Not clear. If there is an underlying assumption behind attentional capture in the context of VFD, this should be laid out in more detail.

 

Response: We clarified this point and added a definition of attentional capture.

 

Modification:

• Introduction: We added details and clarification “The behavioral gaze pattern observed in response to VFD can be explained by attentional capture, a phenomenon in which objects receive prioritized attentional processing even when they are not task-relevant (see Theeuwes, 2025). Accordingly, VFD would attract observers’ gaze because they are perceptually distinctive, novel, unexpected, and unique facial features (Boutsen et al., 2018). As a result, this over-attention to the VFD would leave little room for the processing of other individual information, thereby biasing evaluations and behavior.” (l.63-69).

 

 

[Comment_10] L64 + 69: Standardise the terminology here please. Is “over-gazing” or “over-attention” the more accepted term in the literature?

 

Response: We decided to use different terminologies because we differentiate between studies on attention and studies on gaze behavior. Attention is inferred from gaze behavior (e.g., fixations are an indicator of attention), but attention goes beyond gaze behavior.

 

Modification:

• Introduction: we added details “As such, this research demonstrated the negative impact of over-attending, and more generally over-gazing, at VFD on the processing of relevant individual information provided in parallel.” (l.72-75)

 

 

[Comment_11] L71: Are these findings “seminal”? This wasn’t made clear in the previous paragraph.

 

Response: After proofreading, we decided to change for “previous”.

 

Modification:

• Introduction: we changed for “Adding to these previous findings, the purpose of the present study was to investigate the impact of VFD on the visual processing of other task-relevant directly available facial information.” (l.76-78)

 

 

[Comment_12] L73: See my original comment with regards to the introduction. What “affects” are you referring to? Pain v. pleasure? Aversion v. attraction?

 

Response: We agree that the term “affects” was too broad. We therefore revised the text to specify that we refer to affective information conveyed by emotional expressions, in order to clarify the type of affective cues considered.

 

Modification:

• Introduction: we changed for “Among individual cues, affective information, such as emotional expressions, are particularly relevant because they provide critical information about a person’s intentions, feelings, and social states (Brosch et al., 2013; Hareli & Hess, 2012; Zadra & Clore, 2011).” (l.78-81)

 

 

[Comment_13] L84: Why is pain the focal point of the study and one of the main measures by which response to VFD faces are studied? This should have been the main focus of the preceding paragraphs, leading to a clear rationale for selecting this method.

 

Response: Beyond being a crucial social signal, pain represents a particularly relevant dimension in the context of VFD. Pain is a common experience among individuals with VFD, especially when the facial difference is associated with conditions requiring surgery or results from traumatic injury (Sarwer et al., 2022).Importantly, pain is not only a physical experience but also a socially meaningful cue that shapes observers’ perceptions, emotional responses, and behavioral intentions. We made modifications to put these points forward.

 

Modification:

• Introduction: we changed how we introduced pain “Pain serves as the focal point of this study for several important reasons. As a ubiquitous and inherently aversive experience that engages both sensory-discriminative and affective-motivational components of the pain matrix (Loeser & Melzack, 1999; Tracey & Mantyh, 2007), faces expressing pain reliably capture attention (Vervoort et al., 2013). Observers decode other’s pain by attending to specific facial cues, such as eye narrowing, brow lowering, nose wrinkling, and upper lip raising (Blais et al., 2019). Beyond its role as a core social signal, pain is particularly relevant in the context of VFD. Individuals with VFD, especially when resulting from traumatic injury, frequently experience pain (Sarwer et al., 2022), making pain expressions both ecologically valid and socially meaningful. Thus, pain is not only a physical experience but also a cue that shapes observers’ perceptions, emotional responses, and behavioral intentions (Hadjistavropoulos et al., 2011). Together, these features make pain an especially informative emotional signal for examining how facial differences, such as facial scars, influence social perception and attentional processes.” (l.88-101)

 

 

[Comment_14] L98: “hypothesised” not “assumed”. These hypotheses need to be more specific (ie. What direction of effect were you expecting – higher pain evaluation for VFD or for control faces? Less gaze time to internal AOIs v peripheral AOIs? Positive correlation between gaze time to internal AOIs and pain evaluation?

 

Response: We rectified and clarified our hypotheses.

 

Modification:

• Introduction: we changed for “We hypothesized that: (i) the visual processing of the face would be influenced by the presence of a peripherally located VFD, resulting in increased gaze toward the VFD and reduced gaze toward internal features; (ii) the evaluation of the pain experience would be biased by the presence of a VFD, such that faces with a VFD would be evaluated as experiencing less pain; and (iii) visual processing and pain evaluation would be correlated, with gaze toward the internal features of the face being positively associated with pain assessment, whereas gaze toward the VFD on peripheral features would be negatively associated with pain assessment.” (l.104-112)

 

 

[Comment_15] I see from the stimuli the authors provided that these VFDs were in the form of facial scars. This should be explicitly stated and justified in the context of the study aims.

 

Response: As advised, we added this clarification.

 

Modification:

• Introduction: we changed for: “Together, these features make pain an especially informative emotional signal for examining how facial differences, such as facial scars, influence social perception and attentional processes. Thus, the overarching aim of this study is to investigate how the presence of a VFD (i.e., a facial scar) moderates the processing of pain, both in terms of visual processing, as measured by gaze duration, and self-reported pain evaluation.” (l.99-104)
• Methods: we changed for: “Since our study specifically aimed to investigate the impact of peripherally located VFD while controlling for potential effects of individual faces, scar-like VFD were digitally added on the peripheral areas of the face by a graphic designer” (l.138-140)

 

 

[Comment_16] L126: “VFD were added” – were these digitally added or done in makeup? Was the location of the VFD counterbalanced, given the methodology behind the division of AOIs into 7 different face quadrants (3 internal, 4 peripheral)? Were an equal number of VFDs in internal as peripheral AOIs? I see there was a variance in location from the experimental images provided but it should be made clear whether this was effectively controlled for.

If your stated goal was to test whether peripheral VFDs bias attention away from internal areas of the face, then vary peripheral quadrants only. Bear in mind that this may undermine your ecological validity as surely there is an equal chance of injuring your internal as external face regions?

Response: VFD were digitally added. We were specifically interested in peripherally located VFD and we varied the location of the VFD (forehead, cheeks, chin).

We agree that focusing on peripheral VFD misses the investigation of what happens for central VFD. Our goal was specifically to test whether peripherally located VFDs bias visual attention away from individuating internal facial features. For this reason, we deliberately restricted the manipulation to peripheral facial regions in order to maximize experimental control and to allow a clear separation between areas of interest in the eye-tracking analyses. We acknowledge that this design choice comes at the expense of ecological validity, as facial differences may also occur on internal facial features. However, varying both peripheral and internal locations within the same design would have made it difficult to disentangle gaze directed at the VFD from gaze directed at internal facial features, thereby complicating the interpretation of attentional biases. Future studies could address this with complementary methodologies.

 

Modification:

• Methods: we changed for: “Since our study specifically aimed to investigate the impact of peripherally located VFD while controlling for potential effects of individual faces, scar-like VFD were digitally added on the peripheral areas of the face by a graphic designer” (l.138-140)
• Throughout the document: we specified that VFD were peripherally located in our study.
• Discussion: we added a limitation point: “Finally, we focused on peripherally located VFDs; however, future research would benefit from investigating the impact of VFDs located on internal facial features (e.g., cleft lip) as well. If VFDs systematically bias the visual processing of individuating information regardless of their location, eye-tracking alone may fail to distinguish visual processing directed toward the VFD and that directed toward the internal facial features. Complementary methodological approaches would therefore be needed and could also help address other limitations of the present research.” (l.333-340).

 

 

[Comment_17] L160: Would it not make more sense to run a combined analysis of both peripheral and internal AOIs with location as an included factor? Or were both dependent variables (dwell time – Peripheral, dwell time – Internal) included in the same ANOVA? It is not clear from your description here, especially as:

L178: You mention an interaction between the two DVs here. How is this possible if two separate repeated-measures ANOVAs were conducted?

 

Response: We decided to conduct separate ANOVAs for several reasons. Including area of interest as a factor would have considerably increased the complexity of the model by introducing higher-level interactions, which we deemed unwarranted given our design. Moreover, a direct comparison between the two types of areas was not straightforward because the internal and peripheral AOIs differ in size (internal < peripheral). Accounting for this difference would have required data transformation (e.g., computing proportions), which we chose to avoid to maintain interpretability.

We also apologize for the confusion regarding the reported interaction. The interaction we referred to was between our independent variables (i.e., face type × expression type), not between the two dependent variables.

 

Modification:

• Results: we rectified by “There was also a significant interaction effect between the two independent variables.” (l.194) and “There was no significant interaction effect between the two independent variables.” (l.226).

 

 

[Comment_18] L182: Table 1 is very confusing for the reader both in terms of its content and formatting. You have external columns for “Expression type, Neutral, Painful + All” which don’t correspond to the actual content of the table. You also have internal columns for “AoI + Face type”, but “Pain rating” isn’t an AoI! I would not include this table in the article at all as it doesn’t provide much useful information beyond the later graphs. But even as an annex, this needs to be better edited and formatted – it is a real mess of information.

 

Response: As requested, we revised Table 1 to make it clearer and less confusing. Specifically, we retained only descriptive statistics and removed results from the repeated-measures ANOVA and post-hoc comparisons, which are now fully reported in the main text.

 

Modification:

• Results: We removed results of repeated measures ANOVA and post-hoc comparisons from table andincorporated them into the main text: e.g. “A significant main effect of face type was observed on gaze duration directed towards the peripheral features of the face (F(1, 43) = 48.50, p <.001, η²_G= 0.09; see Table 1 and Figure 2). As expected, participants looked longer at peripheral features when the face had VFD than did not have VFD. There was also a significant main effect of expression type (F(1, 43) = 27.74, p <.001, η²_G = 0.03), showing that participants looked longer at peripheral features for neutral conditions than for pain ones. There was also a significant interaction effect between the two independent variables (F(1, 43) = 8.73, p = .01, η²_G = 0.01). Post-hoc analyses revealed, as expected, that participants gazed at peripheral features of the face for significantly longer duration when the face had VFD than when it did not have VFD, the effect being larger for faces displaying neutral expression (t(43) = 7.41, p_Bonf < .001, d_Cohen = 0.82) than pain expression (t(43) = 4.06, p_Bonf < .001, d_Cohen = 0.45).” (l.189-200).

 

 

[Comment_19] L186: From Figure 2 it looks like your ppts spent longer looking at internal AOIs in faces without VD than with VD but you say: “As expected, participants looked longer at internal features when the face had VFD than when it did not have VFD”. What is going on here?

Also from your introduction, the rationale if not the stated hypothesis was that individuals spend less time looking at the internal areas of the face as the VFD captures attention so how could this result possibly be expected?

 

Response: We apologize for this mistake.

 

Modification:

• Results: we rectified “As expected, participants looked longer at internal features when the face did not have VFD than when it did have VFD” (l.191-192).

[Comment_20] Figure 2. You found a main effect of VFD vs control faces for gaze time to the internal AOI, independent of expression. So show it! Have a graph with VFD vs control across all faces and make it clear this is a significant difference in the graph (* asterisk). Then show a graph with the other main effect between painful vs normal expressions for gaze time to the internal AOI, independent of face type. Again make clear where the significant difference lies.

 

Response: We respectfully disagree that adding separate two-bar plots or significance markers directly onto the graphs would improve the interpretability of the results. In a factorial design including both main effects and interactions, visual markers such as asterisks or brackets can be ambiguous and may suggest simple effects that were not tested.

At the same time, to improve clarity for the reader, we added explicit notes below the figures indicating which effects were statistically significant. This allows the significant differences to be clearly identified without introducing potentially misleading graphical annotations or redundant plots.

The corresponding statistical results were already fully reported in the Results section (including test statistics and p-values), and these remain unchanged.

 

Modification:

• Results: We added a note: e.g. for figure 2 “Note. Both main effects and interaction effects are significant.” (l.214)

 

 

[Comment_21] L202: What are the two dependent variables? See comment for L160.

 

Response: Again, we apologize for this mistake. We meant “independent variables”.

 

Modification:

• Results: we rectified by “There was also a significant interaction effect between the two independent variables.” (l.194) and “There was no significant interaction effect between the two independent variables.” (l.226).

 

 

[Comment_21] Figure 3. From the figure and the mean error bars included, this does not look like a significant difference between pain rating for the VFD vs normal faces. Again, a graph comparing VFD vs normal across all expressions would help to show the reader at a glance what significant effects your study found. Include brackets and asterisks to show statistical significance between bars.

 

Response: As mentioned in our response to your Comment 20, we added explicit notes below the figures indicating which effects were statistically significant.

 

Modification:

• Results: We added a note: e.g. for figure 4: “Note. Only main effects are significant.” (l.229).

 

 

[Comment_22] Table 2. Once again, this formatting is very poor and unclear as to what the data represents. The “Pain rating” label is very misleading as this should be placed above the second two columns. The labels for “Gaze duration” in the rows is very difficult to parse at a glance. This may essentially be left out as the correlations are essentially randomly distributed around 0. Instead you can add to the sentence “None of the expected correlations were evidenced” …as per our previous hypothesis of a significant positive correlation between pain rating and gaze time to the internal features of the face (or something to that effect).

 

Response: As advised, we removed the table.

 

Modification:

• Results: We removed Table 2 and changed text for: “None of the expected correlations between visual processing and pain evaluation were found. There was neither a significant positive correlation between gaze toward the internal features of the face being and pain assessment, nor a significant negative correlation between gaze toward VFD on peripheral features and pain assessment.” (l.231-234).

 

 

[Comment_23] The key issue here is that it’s not clear what you set out to investigate in the first place: 1) If individuals attend less to internal facial features in VFD faces and consequently are less able to discriminate between pained and non-pained faces? 2) If individuals are less empathetic towards faces with VFDs than non-VFD faces? 3) More empathetic?

 

Response: In line with Comment 14, we have revised and clarified our hypotheses to better specify our initial research objectives. Our study was designed to investigate whether the presence of a VFD biases visual processing of facial information and, in turn, affects the evaluation of pain. We did not formulate any a priori hypotheses regarding empathy. However, following the reviewer’s suggestion, we now discuss empathy as a possible interpretative framework in the Discussion section, while acknowledging that it was not directly measured.

 

Modification:

• Introduction: we changed for “We hypothesized that: (i) the visual processing of the face would be influenced by the presence of a peripherally located VFD, resulting in increased gaze toward the VFD and reduced gaze toward internal features; (ii) the evaluation of the pain experience would be biased by the presence of a VFD, such that faces with a VFD would be evaluated as experiencing less pain; and (iii) visual processing and pain evaluation would be correlated, with gaze toward the internal features of the face being positively associated with pain assessment, whereas gaze toward the VFD on peripheral features would be negatively associated with pain assessment.” (l.104-112)
• Discussion: We added discussion points: “As such, focusing on the VFD might bias empathy for pain, i.e., the ability to share and understand the pain experience of someone else (Goubert et al., 2005). Importantly, however, empathy was not directly assessed in the present study, as participants were not instructed to engage in perspective-taking or affect sharing. Any role of empathy should therefore be interpreted as an indirect, downstream mechanism rather than a primary explanatory factor. One possibility is that reduced attention to pain-diagnostic facial features in faces with VFD limits access to individuating emotional information, which may constrain cognitive empathy (perspective-taking). Prior eye-tracking work suggests that attention to internal facial features, particularly the eyes, supports accurate emotion decoding and perspective-taking (Cowan et al., 2014). However, the present results indicate that atypical gaze allocation did not translate into lower pain ratings. Instead, participants systematically rated pain higher in faces with VFD, suggesting that judgments relied less on perceptual input and more on top-down expectations. In the context of disability, visible differences are often associated with stereotypes of vulnerability and suffering, which can shape empathic responses independently of actual emotional expressions (Nario-Redmond et al., 2019). Recent work shows that such stereotypes may lead to biased or paternalistic forms of empathy, characterized by lower affective sharing (Granjon et al., 2023). Notably, Granjon and colleagues (2023) demonstrated that early affective empathy, as measured via an electro-encephalography (EEG) paradigm, and which can also be interpreted as reflecting attentional biases, was altered when participants viewed faces with disabilities, whereas cognitive empathy remained unaffected. This raises the intriguing possibility that the attentional biases captured via eye-tracking in our study could reflect a similar early, automatic component of empathy, linking perceptual-attentional processes with affective responses to VFD. Future research combining eye-tracking with EEG markers of early and late empathic components would help to disentangle these processes and clarify how attention and empathic processing interact in the social perception of VFD.” (l.287-313)

 

 

[Comment_24] All of these rationales for conducting the study should be made clear in the introduction as mentioned previously. If it is the first rationale, why use pain as an independent variable in the first place if there is an obvious confound with facial scarring inducing pain? Why not use a face expression categorisation or face memory task, where you would also expect performance to decrease given the attentional capture of the VFD?

 

Response: We conducted this study to examine how the visual processing of a VFD biases the processing of other information conveyed by individuals with VFDs, and more specifically information related to their internal states. We were particularly interested in assessing whether such biases could be linked to eye-tracking measures. From this perspective, the proposed protocol appeared appropriate for addressing our research questions.

That said, in light of the present findings, it is clear that this paradigm has limitations and that other experimental approaches may further clarify the mechanisms underlying our results. For instance, one possibility is that what is biased is not so much the perceived intensity of pain, but rather the detection or activation of pain-related representations during the categorization of faces with VFDs. In this respect, the paradigms suggested by the reviewer (e.g., facial expression categorization or face memory tasks) constitute promising avenues for future research, which we intend to consider in subsequent studies.

 

 

Modification:

• Discussion: we added a call for other paradigms “Complementary methodological approaches would therefore be needed and could also help address other limitations of the present research.” (l.338-240).

 

 

[Comment_25] L214-215: From L186 “As expected, participants looked longer at internal features when the face had VFD than when it did not have VFD”. Now you state: “participants spent less time on internal facial features but looked for longer duration at peripheral areas when the face had VFD than did not have VFD”. Which one is it?

 

Response: Again, we apologize for this mistake.

 

Modification:

• Results: we rectified “As expected, participants looked longer at internal features when the face did not have VFD than when it did have VFD” (l.191-192).

 

 

[Comment_26] L215-217: “Moreover, the results revealed a moderating effect of facial expression, as the visible difference (in the peripheral area) was gazed for shorter duration when the face displayed pain.” But there was no significant interaction here, so this may simply be speculation?

 

Response: Actually, there is a significant interaction effect. As mentioned in the results section: “There was also a significant interaction effect between the two independent variables. Post-hoc analyses revealed, as expected, that participants gazed at peripheral features of the face for significantly longer duration when the face had VFD than when it did not have VFD, the effect being larger for faces displaying neutral expression than pain expression.” (F(1, 43) = 8.73, p = .01, η²_G = 0.01).

 

 

[Comment_27] L218: What do you mean “overestimated”? Overestimated the pain ratings compared to what? Non-VFD faces? It may have been helpful to initially validate the pain ratings for set of VFD and non-VFD faces separately with a small number of participants. It may simply be the case as you later state that by comparing the two, the VFD faces receive a more “sympathetic” pain rating than non-VFD, or that non-VFD faces receive a lower pain rating than they would otherwise. In this case, you can’t state that there is an “overestimation” or “underestimation” of pain ratings, only that one is rated more highly than the other.

 

Response: We agree with you. We revised the sentence as advised by Reviewer 2.

 

Modification:

• Discussion: We changed for “Contrary to our hypothesis, participants did not rate pain intensity as lower when presented with faces with VFD. We observed the opposite: the pain ratings were higher for faces with VFD compared to faces without VFD.” (l.242-245)

 

 

[Comment_28] L221: The authors cite several studies in social psychology describing patterns of behaviour and social response to VFDs but which may not be possible to replicate or even investigate using methods in visual perception and emotion research. This is fine for the initial “macro” view in the Introduction but not for analysing the findings in the Discussion. Please include more relevant studies in eyetracking and visual perception.

 

Response: In response, we revised the Discussion by building on the opening provided by empathy-related frameworks to integrate more relevant eye-tracking and visual perception studies.

 

Modification:

• Discussion: We added discussion points: “As such, focusing on the VFD might bias empathy for pain, i.e., the ability to share and understand the pain experience of someone else (Goubert et al., 2005). Importantly, however, empathy was not directly assessed in the present study, as participants were not instructed to engage in perspective-taking or affect sharing. Any role of empathy should therefore be interpreted as an indirect, downstream mechanism rather than a primary explanatory factor. One possibility is that reduced attention to pain-diagnostic facial features in faces with VFD limits access to individuating emotional information, which may constrain cognitive empathy (perspective-taking). Prior eye-tracking work suggests that attention to internal facial features, particularly the eyes, supports accurate emotion decoding and perspective-taking (Cowan et al., 2014). However, the present results indicate that atypical gaze allocation did not translate into lower pain ratings. Instead, participants systematically rated pain higher in faces with VFD, suggesting that judgments relied less on perceptual input and more on top-down expectations. In the context of disability, visible differences are often associated with stereotypes of vulnerability and suffering, which can shape empathic responses independently of actual emotional expressions (Nario-Redmond et al., 2019). Recent work shows that such stereotypes may lead to biased or paternalistic forms of empathy, characterized by lower affective sharing (Granjon et al., 2023). Notably, Granjon and colleagues (2023) demonstrated that early affective empathy, as measured via an electro-encephalography (EEG) paradigm, and which can also be interpreted as reflecting attentional biases, was altered when participants viewed faces with disabilities, whereas cognitive empathy remained unaffected. This raises the intriguing possibility that the attentional biases captured via eye-tracking in our study could reflect a similar early, automatic component of empathy, linking perceptual-attentional processes with affective responses to VFD. Future research combining eye-tracking with EEG markers of early and late empathic components would help to disentangle these processes and clarify how attention and empathic processing interact in the social perception of VFD.” (l.287-313)

 

 

[Comment_29] L257: This is one of the few references the authors make directly to empathy for pain. Both potential readers and this reviewer could be forgiven for thinking that this would be the whole rationale behind the study as a whole, given the importance of pain perception in research into empathy and other aspects of theory of mind. I would urge the authors to review this evidence more carefully in the context of their study, and whether the focus of the study is on empathy to pain in VFDs or on impaired discrimination of facial information due to attentional capture as mentioned previously. Make this focus clearer to the reader!

 

Response: We agree that pain perception is often central to empathy research and that this could create ambiguity regarding the primary focus of our study. To address this, we have clarified that the present work is not designed to test empathy for pain or theory-of-mind processes (see also our response to your comment 1). Instead, our primary focus is on visual attention to faces with VFD and on observers’ pain intensity ratings, without claims about empathic accuracy or emotion recognition. Empathy is therefore introduced and discussed only in the Discussion, as a broader interpretative framework to contextualize the findings, rather than as a core rationale of the study. 

 

Modification:

• Discussion: We added discussion points: “As such, focusing on the VFD might bias empathy for pain, i.e., the ability to share and understand the pain experience of someone else (Goubert et al., 2005). Importantly, however, empathy was not directly assessed in the present study, as participants were not instructed to engage in perspective-taking or affect sharing. Any role of empathy should therefore be interpreted as an indirect, downstream mechanism rather than a primary explanatory factor. One possibility is that reduced attention to pain-diagnostic facial features in faces with VFD limits access to individuating emotional information, which may constrain cognitive empathy (perspective-taking). Prior eye-tracking work suggests that attention to internal facial features, particularly the eyes, supports accurate emotion decoding and perspective-taking (Cowan et al., 2014). However, the present results indicate that atypical gaze allocation did not translate into lower pain ratings. Instead, participants systematically rated pain higher in faces with VFD, suggesting that judgments relied less on perceptual input and more on top-down expectations. In the context of disability, visible differences are often associated with stereotypes of vulnerability and suffering, which can shape empathic responses independently of actual emotional expressions (Nario-Redmond et al., 2019). Recent work shows that such stereotypes may lead to biased or paternalistic forms of empathy, characterized by lower affective sharing (Granjon et al., 2023). Notably, Granjon and colleagues (2023) demonstrated that early affective empathy, as measured via an electro-encephalography (EEG) paradigm, and which can also be interpreted as reflecting attentional biases, was altered when participants viewed faces with disabilities, whereas cognitive empathy remained unaffected. This raises the intriguing possibility that the attentional biases captured via eye-tracking in our study could reflect a similar early, automatic component of empathy, linking perceptual-attentional processes with affective responses to VFD. Future research combining eye-tracking with EEG markers of early and late empathic components would help to disentangle these processes and clarify how attention and empathic processing interact in the social perception of VFD.” (l.287-313)

 

 

[Comment_30] L271: The authors state: the (VFD) “stimuli were mainly scars”. I could only see scars among the experimental stimuli provided. What other VFD exemplars were used? If there were different types of VFD this should be made clear in the Methods earlier.

 

Response: We removed “mainly”.

 

Modification:

• Discussion: we changed for “Stimuli were scars that can be perceived as resulting from a painful injury.”(l.324)

 

Reviewer 2 Report

Comments and Suggestions for Authors

I am grateful to have had the opportunity to review this paper. The authors have investigated how gaze behavior may influence the perception of pain in individuals with visible facial differences (VFD). Their use of eye tracking for this research is appropriate and lends itself well to answer this research question.

However, I strongly believe the paper would benefit from major revisions, as the findings are not always clearly presented and the reader must at times make certain assumptions about the experiment design or statistical analyses. Also, the correlation analysis is mentioned but barely detailed, which leaves many questions unanswered as well as making it unclear why authors conducted these analyses. Eye tracking data is somewhat underutilized in the authors’ analysis plan and would benefit from more in-depth analyses.

I believe this paper would be a very welcome publication for this journal after revision, as it can have an impact on how we understand the perception of pain in individuals with VFD. It will be my pleasure to review this paper again once my comments have been addressed.

Here is a list of more precise comments:

Introduction

• I would either clarify the hypotheses for this study (if you expected a directional effect) or nuance the use of “as expected” later in the paper. With your hypotheses as they are, it is unclear whether you expected 1) a significant effect regardless of direction or 2) a significant effect in this specific direction.

Method

• (L. 106) You write seven participants were excluded due to “meeting exclusion criteria.” Is it possible to include more details as to what these criteria were?
• (L. 110-111) It is unclear to me why you based your power analysis on a within-between interaction while you only report within participant effects using repeated measures ANOVA. However, G*Power returns the same suggested sample size (n=36) when performing a power analysis for a repeated measure, within factor ANOVA. If this is a typo, this should be corrected; otherwise, would it be possible to give more details as to why a within-between interaction was used despite no between participant factors being reported in the paper?
• (L. 133-141) Would it be possible to explicitly state how you combined fixation duration across both AoI? I assume it’s a simple addition of dwell times across the selected features (eyes/nose/mouth for internal and cheeks/chin/forehead for external), but it would be good to have this clearly stated for readers less familiar with eye tracking paradigms or this type of analysis.
• Would it be possible to add a Figure showing an example pair of stimuli to visualize the differences between one identity with/without VFD? It would be informative to see how face stimuli were altered to show VFD.

Results

• I assume dwell times are reported in milliseconds, but as you state earlier that “Dwell time corresponds to the total time (in seconds) spent fixing a specific area of interest (AoI) over a trial.” (L. 134-136), this could cause confusion at first glance. Either add a precision to your figure caption in Figure 2 or change the unit you name at lines 134-136.
• Related to my previous point in the method section, you state that “as expected, participants looked longer at peripheral features when the face had VFD than did not have VFD” (L. 173-174). I feel we are missing information to state that this is “as expected”. In your introduction, you correctly state that the presence of VFD orients gaze towards the VFD; considering this, wouldn’t a VFD located closer to internal features (ex. a cleft lip) cause a longer dwell time on internal features? I assume, from what is stated here, that the VFD in your experimental stimuli were all located in external AoI, but the paper would benefit from a clear statement that this is or isn't the case.
• You state that “As expected, participants looked longer at internal features when the face had VFD than when it did not have VFD.” (L. 187-188) This is in direct contradiction to what you show in Figure 2 as well as Table 1, where reported dwell times are lower for internal features for faces with VFD compared to faces without VFD, as well as what you mention in your discussion. Please review either the text or the Figures to ensure that your results are properly reported.
• I would suggest to perhaps clarify what is written in column 1 of Table 2, or consider a more intuitive layout for the table. It is somewhat hard to read what exactly is being correlated here. Also, please consider adding p values to your correlation table.
• (L. 207) Is it possible to include more information on what the “expected correlations” were? You only vaguely mention correlation analyses in your methods and appear to gloss over them in your results section. What, exactly, did you expect to find in your correlation analyses? How did your results differ from what was expected?

Discussion

• You mention (L.217-219) that participants did not underestimate pain ratings when faces had VFD, but rather “overestimated” them. I would suggest more nuance in this statement; while true that pain ratings were higher when faces had VFD, do you have an objective pain rating linked to your face stimuli (ex. by the actor) to compare with your pain estimations? I believe this would be necessary to conclude there is an “overestimation”. Otherwise, I would recommend simply referring to “higher pain ratings in faces with VFD compared to faces without VFD.”
• While you did not find links between gaze duration and pain ratings, I wonder if other information located in the eye tracking data (ex. saccades, initial fixations, number of fixations, fixation location as a function of time) may yield further insights on the question. I believe extracting more information from the eye tracking data could help improve the reported correlation analyses by going beyond simply gaze duration, as duration does not give a complete picture of gaze behavior. If you have the time and someone with the technical ability to perform these analyses, I believe they could greatly improve the conclusions you can make regarding this.

 

Comments on the Quality of English Language

The english language is mostly appropriate, but could benefit from some revision as in some cases, sentence structure could be improved for better readability.

Author Response

[General] I am grateful to have had the opportunity to review this paper. The authors have investigated how gaze behavior may influence the perception of pain in individuals with visible facial differences (VFD). Their use of eye tracking for this research is appropriate and lends itself well to answer this research question.

 

However, I strongly believe the paper would benefit from major revisions, as the findings are not always clearly presented and the reader must at times make certain assumptions about the experiment design or statistical analyses. Also, the correlation analysis is mentioned but barely detailed, which leaves many questions unanswered as well as making it unclear why authors conducted these analyses. Eye tracking data is somewhat underutilized in the authors’ analysis plan and would benefit from more in-depth analyses.

 

I believe this paper would be a very welcome publication for this journal after revision, as it can have an impact on how we understand the perception of pain in individuals with VFD. It will be my pleasure to review this paper again once my comments have been addressed.

 

Here is a list of more precise comments:

 

Response: We thank you for your constructive feedback. We hope that this new version meets your expectation.

 

 

[Comment_1] (L. 106) You write seven participants were excluded due to “meeting exclusion criteria.” Is it possible to include more details as to what these criteria were?

 

Response: These participants were excluded because they inferred that the pain was caused by the facial scar itself rather than by an external cause. We considered this a source of bias, as our objective was to examine whether the presence of a VFD biases the processing of pain information, rather than serving as a direct cue for pain evaluation.

 

Modification:

• Methods: we added more details: “seven for inferring pain as being due to the scar” (l.117).
• Methods: we specify that this information was collected during debriefing: “The demographic questionnaire was administered after the task, followed by a debrief which included questions about the stimuli.” (l.172-173).

 

 

[Comment_2] (L. 110-111) It is unclear to me why you based your power analysis on a within-between interaction while you only report within participant effects using repeated measures ANOVA. However, G*Power returns the same suggested sample size (n=36) when performing a power analysis for a repeated measure, within factor ANOVA. If this is a typo, this should be corrected; otherwise, would it be possible to give more details as to why a within-between interaction was used despite no between participant factors being reported in the paper?

 

Response: Regarding the statistical analyses computed, the analyses reported in the manuscript are indeed repeated-measures ANOVAs including only within-participant factors. The reference to a within-between interaction in the power analysis was not intended to reflect the statistical model ultimately reported, but rather the experimental structure of the design. Specifically, although both independent variables were analyzed as within-participant factors, the visible facial difference was manipulated in a between-stimulus manner, in the sense that a given participant never saw the same face with and without a VFD. Participants were allocated to 4 variations. These were between subjects: each variation presented different individual stimuli (i.e., faces). However, the modalities of our variables (with or without VDF, pain or no pain) were within subjects.

 

Modification:

• Methods: we changed for “An a priori power analysis was conducted via G*Power. Desired sample size was estimated N = 36 for a small effect (f = 0.25), alpha significance (α = .05) and a power = .95 for repeated measures ANOVA.” (l.120-122)

 

 

[Comment_3] (L. 133-141) Would it be possible to explicitly state how you combined fixation duration across both AoI? I assume it’s a simple addition of dwell times across the selected features (eyes/nose/mouth for internal and cheeks/chin/forehead for external), but it would be good to have this clearly stated for readers less familiar with eye tracking paradigms or this type of analysis.

 

Response: We agree with the Reviewer that these details were insufficiently specified in the original version of the manuscript. We have now explicitly clarified how fixation duration was combined across areas of interest.

 

Modification:

• Methods: We added details: “This study considered dwell times of two AoI: one composed of the sum of the dwell times on internal features of the face (i.e., eyes and eyebrows, nose, and mouth) and one composed of the sum of the dwell times on the peripheral features of the face (i.e., forehead, cheeks, and chin).” (l.151-154).

 

 

[Comment_4] I assume dwell times are reported in milliseconds, but as you state earlier that “Dwell time corresponds to the total time (in seconds) spent fixing a specific area of interest (AoI) over a trial.” (L. 134-136), this could cause confusion at first glance. Either add a precision to your figure caption in Figure 2 or change the unit you name at lines 134-136.

 

Response: We apologize for this mistake.

 

Modification:

• Methods: We rectified: “the total time (in milliseconds) spent fixing a specific area of interest (AoI)” (l.149).

 

 

[Comment_5] Related to my previous point in the method section, you state that “as expected, participants looked longer at peripheral features when the face had VFD than did not have VFD” (L. 173-174). I feel we are missing information to state that this is “as expected”. In your introduction, you correctly state that the presence of VFD orients gaze towards the VFD; considering this, wouldn’t a VFD located closer to internal features (ex. a cleft lip) cause a longer dwell time on internal features? I assume, from what is stated here, that the VFD in your experimental stimuli were all located in external AoI, but the paper would benefit from a clear statement that this is or isn't the case.

 

Response: We acknowledge that our hypotheses and rationale were not sufficiently explicit in the original version of the manuscript, and we have added clarifications accordingly. All VFDs used in our experimental stimuli were deliberately located on peripheral facial areas. Our objective was specifically to test whether peripherally located VFDs bias visual processing away from individuating internal facial features.

For this reason, we restricted the manipulation to peripheral facial regions in order to maintain experimental control and to allow a clear separation between areas of interest in the eye-tracking analyses. Introducing both peripheral and internal VFD locations within the same design would have made it difficult to disentangle gaze directed toward the VFD from gaze directed toward internal facial features, thereby complicating the interpretation of visual processing biases.

We agree that this design choice does not allow us to address how centrally located VFDs (e.g., cleft lip) may affect gaze patterns. This limitation is now acknowledged, and we note that future studies could investigate central VFDs using complementary methodological approaches.

 

 

Modification:

• Introduction: we changed for “We hypothesized that: (i) the visual processing of the face would be influenced by the presence of a peripherally located VFD, resulting in increased gaze toward the VFD and reduced gaze toward internal features; (ii) the evaluation of the pain experience would be biased by the presence of a VFD, such that faces with a VFD would be evaluated as experiencing less pain; and (iii) visual processing and pain evaluation would be correlated, with gaze toward the internal features of the face being positively associated with pain assessment, whereas gaze toward the VFD on peripheral features would be negatively associated with pain assessment.” (l.104-112)
• Throughout the document: we specified that VFD were peripherally located in our study.
• Discussion: we added a limitation point: “Finally, we focused on peripherally located VFDs; however, future research would benefit from investigating the impact of VFDs located on internal facial features (e.g., cleft lip) as well. If VFDs systematically bias the visual processing of individuating information regardless of their location, eye-tracking alone may fail to distinguish visual processing directed toward the VFD and that directed toward the internal facial features. Complementary methodological approaches would therefore be needed and could also help address other limitations of the present research.” (l.333-340).

 

 

[Comment_6] You state that “As expected, participants looked longer at internal features when the face had VFD than when it did not have VFD.” (L. 187-188) This is in direct contradiction to what you show in Figure 2 as well as Table 1, where reported dwell times are lower for internal features for faces with VFD compared to faces without VFD, as well as what you mention in your discussion. Please review either the text or the Figures to ensure that your results are properly reported.

 

Response: We apologize for this mistake.

 

Modification:

• Results: we rectified “As expected, participants looked longer at internal features when the face did not have VFD than when it did have VFD” (l.191-192).

 

 

[Comment_7] I would suggest to perhaps clarify what is written in column 1 of Table 2, or consider a more intuitive layout for the table. It is somewhat hard to read what exactly is being correlated here. Also, please consider adding p values to your correlation table.

 

Response: Since we found no correlations, and as advised by Reviewer 1, we decided to remove the table

 

Modification:

• Results: We removed Table 2 and changed text for: “None of the expected correlations between visual processing and pain evaluation were found. There was neither a significant positive correlation between gaze toward the internal features of the face being and pain assessment, nor a significant negative correlation between gaze toward VFD on peripheral features and pain assessment.” (l.231-234).

 

 

[Comment_8] (L. 207) Is it possible to include more information on what the “expected correlations” were? You only vaguely mention correlation analyses in your methods and appear to gloss over them in your results section. What, exactly, did you expect to find in your correlation analyses? How did your results differ from what was expected?

 

Response: Once again, we acknowledge that our hypotheses and rationale were not sufficiently explicit in the original version of the manuscript, and we have added clarifications accordingly.

 

Modification:

• Introduction: we changed for “We hypothesized that: (i) the visual processing of the face would be influenced by the presence of a peripherally located VFD, resulting in increased gaze toward the VFD and reduced gaze toward internal features; (ii) the evaluation of the pain experience would be biased by the presence of a VFD, such that faces with a VFD would be evaluated as experiencing less pain; and (iii) visual processing and pain evaluation would be correlated, with gaze toward the internal features of the face being positively associated with pain assessment, whereas gaze toward the VFD on peripheral features would be negatively associated with pain assessment.” (l.104-112)

[Comment_9] You mention (L.217-219) that participants did not underestimate pain ratings when faces had VFD, but rather “overestimated” them. I would suggest more nuance in this statement; while true that pain ratings were higher when faces had VFD, do you have an objective pain rating linked to your face stimuli (ex. by the actor) to compare with your pain estimations? I believe this would be necessary to conclude there is an “overestimation”. Otherwise, I would recommend simply referring to “higher pain ratings in faces with VFD compared to faces without VFD.”

 

Response: We agree with you. We revised the sentence as advised.

 

Modification:

• Discussion: We changed for “Contrary to our hypothesis, participants did not rate pain intensity as lower when presented with faces with VFD. We observed the opposite: the pain ratings were higher for faces with VFD compared to faces without VFD.” (l.242-245)

 

[Comment_10] While you did not find links between gaze duration and pain ratings, I wonder if other information located in the eye tracking data (ex. saccades, initial fixations, number of fixations, fixation location as a function of time) may yield further insights on the question. I believe extracting more information from the eye tracking data could help improve the reported correlation analyses by going beyond simply gaze duration, as duration does not give a complete picture of gaze behavior. If you have the time and someone with the technical ability to perform these analyses, I believe they could greatly improve the conclusions you can make regarding this.

 

Response: As this was a pre-registered study, we focused on the analyses specified in our pre-registration. That said, we agree that examining additional eye-tracking measures - such as saccades, initial fixations, or fixation locations over time - could reveal patterns not captured by gaze duration alone. Exploring these metrics in future work may provide a deeper understanding of gaze behavior and its relation to pain perception.

 

 

[Comment_10] The english language is mostly appropriate, but could benefit from some revision as in some cases, sentence structure could be improved for better readability.

 

Response: As requested, we carefully reviewed English of the entire manuscript by a native English speaker.

 

Modification:

• Throughout the document: modifications for improvement of English language.

 

Reviewer 3 Report

Comments and Suggestions for Authors

Thank you for the opportunity to review the article " Emotion Misread: An Eye-Tracking Study of Pain Perception Toward Faces with Visible Differences.” The authors present a single eye-tracking study examining the impact of expression (pain vs no pain) and visual face difference (VFD vs. no VFD) on gaze behavior and pain intensity judgments. I found the manuscript well-written and the research well-designed. The high-quality writing, balance of judgment and behavioral measures, and the use of transparent science practices are all strengths of the current work. My overall impression is that this work offers value in replicating past research and in disseminating null and unexpected effects. Despite strengths I often felt the research team overstepped the data and overreached in their claims about contribution. Below I’ll describe this major concern (which think can be addressed with conscientious writing) and several minor questions/requests for the authors’ consideration. I hope my feedback is helpful as the authors pursue publication of this important work.

Major

A strength of this paper is the straightforward presentation of null and counter-hypothesized effects in the results section. However, in the introduction and discussion I often felt tension between these findings and the authors’ claims of contribution. At times the authors’ claims and interpretations felt too strong but at other times I think they were unfounded. I’ll provide a couple examples below. Although I think this is a major issue but also one that can be addressed via writing revisions (rather than requiring new data or analyses).

“Emotion Misread: An Eye-Tracking Study of Pain Perception Toward Faces with Visible Difference” The title is misleading. There isn’t evidence that the pain of people with VFD is being misread in this paper. Given evidence that people systematically underestimate others’ pain intensity/experience – judgments of people with VFD could be more accurate estimates of their pain experience relative to the people pictured without VFD. The use of terms like “accurately” “misread” “over/under estimated” should be used with care and relative comparisons made clear.

“These findings contribute to elucidating the mechanisms by which people with VFD are perceived and how their emotions are interpreted.” “this study provides novel insights into the mechanisms underlying public stigmatization of VFD and underscores the importance of considering perceptual and cognitive factors when addressing social biases” Statements like this were made throughout the paper. These claims require evidence of process which the current study does not provide (gaze was not correlated with emotion ratings – which doesn’t allow for strong claims that attention does or does not influence judgments of people with VFD).

“Thus, it seems that the perception of pain others have of people with VFD is influenced by stereotypical categorical information:…”  Overreach – no data collected on stereotyping or categorization of faces.

“This research shows that observers are biased in their processing of the affective re-actions experienced by people with VFD, both in visual processing and in self-reported measures.” To me this claim would be supported by the interaction between face condition (with vs without VFD) and expression condition (pain vs no pain) but no interaction was observed for internal gaze nor pain judgments and the effect of gaze duration was actually attenuated for pain expressions in peripheral gaze.

“By spending less time fixating on these diagnostic features, observers may fail to accurately perceive the pain experience of people with VFD.” No evidence from current work to support this claim.

Throughout the manuscript the authors talk about accurately reading pain expression or recognizing pain expression – the current methods isn’t well-suited for these questions or claims. A measure of pain intensity doesn’t tell us much about the recognition of the pain expression nor the accurate interpretation of the expression (without a gold standard comparison … like self-reported pain of the target).

Minor

Additional background on gaze and pain processing would enhance the introduction -- there are some jumps in logic (e.g., that more gaze to internal features promotes pain recognition or pain intensity judgments)

“Participants gazed at peripheral features of the face for significantly longer duration when the face had VFD than when it did not have VFD, the effect being larger for faces displaying neutral expression than pain expression” I found this super interesting – I would love to see more attention to this finding in the discussion. For example, might emotion or at least pain decrease attention to the VFD presented in the peripheral of the face? I know this isn’t the focus of the work, but it is an intriguing finding.

It would be useful to know pairwise comparisons for significant moderation effects (e.g., is the difference in attention to peripheral features of face those with and without VFD eliminated in the pain condition?)

Can the authors include more justification for their approach in analyzing peripheral and internal separately rather than operationalizing as a relative measure. I ask because in discussion of the work the authors often describe peripheral and internal as dependent – like a tradeoff between the two.

More detail on stimuli creation (especially with respect to VFD) would be helpful

I noticed a few typos and errors in reporting “As expected, participants looked longer at internal features when the face had VFD than when it did not have VFD.” “There was no significant interaction effect between the two dependent variables.”

Author Response

[General] Thank you for the opportunity to review the article " Emotion Misread: An Eye-Tracking Study of Pain Perception Toward Faces with Visible Differences.” The authors present a single eye-tracking study examining the impact of expression (pain vs no pain) and visual face difference (VFD vs. no VFD) on gaze behavior and pain intensity judgments. I found the manuscript well-written and the research well-designed. The high-quality writing, balance of judgment and behavioral measures, and the use of transparent science practices are all strengths of the current work. My overall impression is that this work offers value in replicating past research and in disseminating null and unexpected effects. Despite strengths I often felt the research team overstepped the data and overreached in their claims about contribution. Below I’ll describe this major concern (which think can be addressed with conscientious writing) and several minor questions/requests for the authors’ consideration. I hope my feedback is helpful as the authors pursue publication of this important work.

A strength of this paper is the straightforward presentation of null and counter-hypothesized effects in the results section. However, in the introduction and discussion I often felt tension between these findings and the authors’ claims of contribution. At times the authors’ claims and interpretations felt too strong but at other times I think they were unfounded. I’ll provide a couple examples below. Although I think this is a major issue but also one that can be addressed via writing revisions (rather than requiring new data or analyses).

 

Response: We thank you for your constructive feedback. We hope that this new version meets with your satisfaction.

 

 

[Comment_1] “Emotion Misread: An Eye-Tracking Study of Pain Perception Toward Faces with Visible Difference” The title is misleading. There isn’t evidence that the pain of people with VFD is being misread in this paper. Given evidence that people systematically underestimate others’ pain intensity/experience – judgments of people with VFD could be more accurate estimates of their pain experience relative to the people pictured without VFD. The use of terms like “accurately” “misread” “over/under estimated” should be used with care and relative comparisons made clear.

 

Response: We agree with this point. In the revised version, we have avoided using terms such as “misread,” “accurate,” or “overestimation” when referring to our data.

 

Modification:

• Title: We changed for “An Eye-Tracking Study of Pain Perception Toward Faces with Visible Differences”
• Discussion: We changed for “Contrary to our hypothesis, participants did not rate pain intensity as lower when presented with faces with VFD. We observed the opposite: the pain ratings were higher for faces with VFD compared to faces without VFD.” (l.242-245)

 

 

[Comment_2] “These findings contribute to elucidating the mechanisms by which people with VFD are perceived and how their emotions are interpreted.” “this study provides novel insights into the mechanisms underlying public stigmatization of VFD and underscores the importance of considering perceptual and cognitive factors when addressing social biases” Statements like this were made throughout the paper. These claims require evidence of process which the current study does not provide (gaze was not correlated with emotion ratings – which doesn’t allow for strong claims that attention does or does not influence judgments of people with VFD).

 

Response: We have revised the manuscript to remove claims about mechanisms that were not clearly evidenced. 

 

Modification:

• Abstract: We changed for: “These findings provide insights into how people with VFD are perceived and how their emotions are interpreted.” (l.27-28)
• Conclusion: We changed for: “This research shows that the presence of VFD influences observers’ visual facial processing and their pain ratings. The findings are consistent with the idea that the processing of faces with VFD may shape how observers interpret facial information. By documenting these patterns, the present study contributes to identifying factors that may influence evaluations of faces with VFD. These results point to directions for future research aimed at better understanding and, ultimately, reducing misperceptions in social interactions involving individuals with facial differences.” (l.342-348).

 

 

[Comment_3] “Thus, it seems that the perception of pain others have of people with VFD is influenced by stereotypical categorical information:…”  Overreach – no data collected on stereotyping or categorization of faces.

 

Response: We agree that our study did not include direct measures of stereotyping or categorical processing. In response, we have revised the manuscript to remove statements implying that our data demonstrate the influence of such processes. We reformulated with more caution.

 

Modification:

• Discussion: We changed for: “In line with models of person perception (Fiske & Neuberg, 1990), it is possible that judgments of pain in people with VFD may have been influenced by stereotypical categorical information rather than solely by individuating information available from the face.” (l.274-277).

 

 

[Comment_4] “This research shows that observers are biased in their processing of the affective re-actions experienced by people with VFD, both in visual processing and in self-reported measures.” To me this claim would be supported by the interaction between face condition (with vs without VFD) and expression condition (pain vs no pain) but no interaction was observed for internal gaze nor pain judgments and the effect of gaze duration was actually attenuated for pain expressions in peripheral gaze.

 

Response:  We agree with the reviewer that the original wording overstated our findings. We have revised this sentence to remove claims about interaction and focused on the main effect of the presence of facial difference.

 

Modification:

• Conclusion: We changed for: “This research shows that the presence of VFD influences observers’ visual facial processing and their pain ratings. The findings are consistent with the idea that the processing of faces with VFD may shape how observers interpret facial information. By documenting these patterns, the present study contributes to identifying factors that may influence evaluations of faces with VFD. These results point to directions for future research aimed at better understanding and, ultimately, reducing misperceptions in social interactions involving individuals with facial differences.” (l.342-348).

 

 

[Comment_5] “By spending less time fixating on these diagnostic features, observers may fail to accurately perceive the pain experience of people with VFD.” No evidence from current work to support this claim.

 

Response: Although the original wording was intentionally cautious, we agree that it could be interpreted as implying a consequence for pain perception that is not directly supported by our data. We have therefore revised this sentence to describe the main effect of VFD on visual attention to pain-informative facial features, without inferring consequences for pain perception.

 

Modification:

• Discussion: We changed for: “By spending less time fixating on these diagnostic features, observers allocated less visual attention to pain-informative facial features in faces with VFD.” (l.254-256).

 

 

[Comment_6] Throughout the manuscript the authors talk about accurately reading pain expression or recognizing pain expression – the current methods isn’t well-suited for these questions or claims. A measure of pain intensity doesn’t tell us much about the recognition of the pain expression nor the accurate interpretation of the expression (without a gold standard comparison … like self-reported pain of the target).

 

Response: We agree that our measure of pain intensity reflects observers’ ratings and does not provide a gold standard for assessing accurate recognition or interpretation of pain expressions. In the revised manuscript, we now clarify that pain intensity was measured independently of the targets’ actual pain experience, which could be more accurately assessed, for example, via self-reported pain.

 

Modification:

• Discussion: We changed for: “In our study, we measured observers’ ratings of pain intensity independently of the targets’ actual pain experience, which would be a more accurate indicator and could, for example, be assessed via self-reported pain. Nonetheless, it is possible that observers do not rely on any specific area of the face when evaluating pain intensity, but focusing on specific areas might be related to more accurate pain perception.” (l.281-286).

 

 

[Comment_7] Additional background on gaze and pain processing would enhance the introduction -- there are some jumps in logic (e.g., that more gaze to internal features promotes pain recognition or pain intensity judgments)

 

Response: We thank the reviewer for this comment and agree that the initial version of the Introduction implied a link between gaze allocation to internal facial features and pain evaluation that was not sufficiently supported. We acknowledge this as a limitation of our reasoning. In the revised manuscript, we no longer assume such a relationship and explicitly address this issue in the Discussion. We now clarify that we did not find evidence for a relationship between gaze processing of facial features and pain ratings, regardless of facial expression, in line with previous work (e.g., Stopyn et al., 2021). We further emphasize that pain intensity ratings in the present study reflect observers’ evaluations rather than accurate recognition of the targets’ pain experience.

 

Modification:

• Discussion: we changed for “We didn’t find evidence for any relationship between gaze processing of facial features and pain rating, whether the face was expressing pain or not. Previous research also failed to demonstrate relationships between gaze duration and pain ratings for stimuli without VFD (Stopyn et al., 2021). In our study, we measured observers’ ratings of pain intensity independently of the targets’ actual pain experience, which would be a more accurate indicator and could, for example, be assessed via self-reported pain. Nonetheless, it is possible that observers do not rely on any specific area of the face when evaluating pain intensity, but focusing on specific areas might be related to more accurate pain perception.” (l.278-286)

 

 

[Comment_8] “Participants gazed at peripheral features of the face for significantly longer duration when the face had VFD than when it did not have VFD, the effect being larger for faces displaying neutral expression than pain expression” I found this super interesting – I would love to see more attention to this finding in the discussion. For example, might emotion or at least pain decrease attention to the VFD presented in the peripheral of the face? I know this isn’t the focus of the work, but it is an intriguing finding.

 

Response: We have expanded the Discussion to acknowledge that interaction effects suggest that participants are less diverted by peripheral VFD when faces display pain, indicating that pain information subtly influences visual attention to faces with VFD. While this was not the primary focus of the study, we agree that it provides an intriguing insight into how emotional expressions can modulate attention to visible differences.

 

Modification:

• Discussion: We added: “Interaction effects indicated that attention to VFD was reduced when faces expressed pain, suggesting that pain-related information modulates visual processing of faces with VFD.” (l.260-262).

 

 

[Comment_9] It would be useful to know pairwise comparisons for significant moderation effects (e.g., is the difference in attention to peripheral features of face those with and without VFD eliminated in the pain condition?)

 

Response: We present hereafter all significant pairwise comparisons.

There was a significant difference between:

• Neutral faces with VFD compared to painful faces with VFD (t(43) = 5.87, p_Bonf < .001, d_Cohen= 0.55),
• Neutral faces with VFD compared to neutral faces without VFD (t(43) = 7.41, p_Bonf < .001, d_Cohen= 0.82),
• Neutral faces with VFD compared to painful faces without VFD (t(43) = 8.73, p_Bonf < .001, d_Cohen= 1.00),
• Painful faces with VFD compared to painful faces without VFD (t(43) = 4.06, p_Bonf < .001, d_Cohen= 0.45),

There was no significant difference between:

• Painful faces with VFD compared to neutral faces without VFD (t(43) = 2.43, p_Bonf = .28),
• Neutral faces without VFD compared to painful faces without VFD (t(43) = 1.86, p_Bonf = .17).

Please note that we only included pairwise comparisons that were relevant to our hypotheses.

 

 

[Comment_10] Can the authors include more justification for their approach in analyzing peripheral and internal separately rather than operationalizing as a relative measure. I ask because in discussion of the work the authors often describe peripheral and internal as dependent – like a tradeoff between the two.

 

Response: We decided to conduct separate ANOVAs for several reasons. Including area of interest as a factor would have considerably increased the complexity of the model by introducing higher-level interactions, which we deemed unwarranted given our design. Moreover, a direct comparison between the two types of areas was not straightforward because the internal and peripheral AOIs differ in size (internal < peripheral). Accounting for this difference would have required data transformation (e.g., computing proportions), which we chose to avoid to maintain interpretability.

 

 

[Comment_11] More detail on stimuli creation (especially with respect to VFD) would be helpful

 

Response: Since our study specifically aimed to investigate the impact of peripherally located VFD while controlling for potential effects of individual faces, VFD were digitally added to peripheral areas of the faces by a graphic designer.

 

Modification:

• Methods: we gave more information “Since our study specifically aimed to investigate the impact of peripherally located VFD while controlling for potential effects of individual faces, scar-like VFD were digitally added on the peripheral areas of the face by a graphic designer” (l.138-140).

 

 

[Comment_12] I noticed a few typos and errors in reporting “As expected, participants looked longer at internal features when the face had VFD than when it did not have VFD.” “There was no significant interaction effect between the two dependent variables.”

 

Response: We apologize for these mistakes.

 

Modification:

• Results: we rectified by “There was also a significant interaction effect between the two independent variables.” (l.194) and “There was no significant interaction effect between the two independent ” (l.226).
• Results: we rectified “As expected, participants looked longer at internal features when the face did not have VFD than when it did have VFD” (l.191-192).

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

I thank the authors for answering all my comments. I believe the paper is now suitable for publication as is.

Reviewer 3 Report

Comments and Suggestions for Authors

Thank you for the thoughtful revisions and collegial handling of my comments - so appreciated!

Most of my concerns were addressed in revision although I still think greater detail on stimulus manipulation would improve the manuscript. How was VFD communicated to the graphic designer. How were VFDs added to the stimuli (what software; what approach)? What types of VFD are included? How many of each type or variation are included? Did the size, area, contrast of the VFD on the stimuli vary? In face perception and eye tracking paradigms this level of detail is common and helpful for interpreting the findings.