Perceptual Uncertainty and Its Monitoring in Visual Snow Syndrome: A Multimodal Framework
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe paper proposes that heightened perceptual uncertainty and impaired metacognition underlie the visual disturbances in Visual Snow Syndrome (VSS), using a Bayesian framework to model disrupted sensory processing. However, the reasoning is inconsistent and oversimplified, conflating distinct levels of cognitive and perceptual mechanisms without providing sufficient empirical support or clarity.
1. The authors claim, “We suggest that heightened perceptual uncertainty in patients with VSS reflects in higher prediction errors.” This statement is conceptually incorrect. Heightened perceptual uncertainty (i.e., increased variance in sensory inputs) should lead to smaller precision-weighted prediction errors, as the system downweights unreliable sensory information. The authors may need to revisit their understanding about how Bayesian models compute and propagate prediction errors.
2. The authors state, “patients with VSS process visual sensory information [and] may face a discrepancy between actual and expected sensory data, that healthy subjects may not,” yet their figure appears to contradict this. In the figure, healthy controls show a clear discrepancy between priors and likelihoods, while for VSS patients, these distributions seem to align closely, suggesting little to no discrepancy. This contradiction undermines the internal consistency of the proposal.
3. Even if the proposal was internally consistent, it is unclear whether the Bayesian framework can help in any meaningful way. Bayesian theories have been widely applied to various mental disorders, including autism and psychosis, but with mixed empirical support and significant overgeneralizations. For instance, Chrysaitis, N. A., & Seriès, P. (2023). 10 years of Bayesian theories of autism: a comprehensive review. Neuroscience & Biobehavioral Reviews, 145, 105022 comprehensively reviewed Bayesian theories of autism and found the empirical results to be inconsistent and inconclusive. Similarly, broader critiques (e.g., Karvelis, P., Paulus, M. P., & Diaconescu, A. O. (2023). Individual differences in computational psychiatry: A review of current challenges. Neuroscience & Biobehavioral Reviews, 148, 105137) highlight how the Bayesian accounts end up being too general, which does not agree with empirical observations. These challenges should temper the enthusiasm for applying the framework to VSS, especially in a handwavy and uncritical way.
4. The authors introduce metacognition as a central component of their model, but the explanation feels implausible. Metacognition typically refers to high-level cognitive processes like self-monitoring and self-regulation, whereas the low-level perceptual disturbances in VSS seem far removed from these higher-order functions. There are multiple intermediary layers between low-level perception and metacognition. I don’t think it is accurate to refer to the mechanisms at the level of perceptual recalibration and gain control as metacognition.
Author Response
The paper proposes that heightened perceptual uncertainty and impaired metacognition underlie the visual disturbances in Visual Snow Syndrome (VSS), using a Bayesian framework to model disrupted sensory processing. However, the reasoning is inconsistent and oversimplified, conflating distinct levels of cognitive and perceptual mechanisms without providing sufficient empirical support or clarity.
- The authors claim, “We suggest that heightened perceptual uncertainty in patients with VSS reflects in higher prediction errors.” This statement is conceptually incorrect. Heightened perceptual uncertainty (i.e., increased variance in sensory inputs) should lead to smaller precision-weighted prediction errors, as the system downweights unreliable sensory information. The authors may need to revisit their understanding about how Bayesian models compute and propagate prediction errors.
Thank you so much for your comment that helped us refine and hopefully strengthen our argument regarding the interplay of perceptual uncertainty and metacognition in patients with VSS
Indeed, you are right in saying, that using the Bayesian rationale, sensory uncertainty as increased variance of likelihood should result in minimization of prediction error. We believe that in healthy participants this can be assured by effective filtering of noise based on perceptual priors. However, there are conditions when sensory uncertainty may become salient- even in healthy participants- necessitating the updating of prior perceptual beliefs, via heightened prediction errors. In VSS cortical hyperexcitability may suggest that this might translate in an increase in sensory gain. We tried to strengthen our argument and account for your criticism in our new version and hope we manage to convey our argument (see 2.1 Predictive Coding framework)
- The authors state, “patients with VSS process visual sensory information [and] may face a discrepancy between actual and expected sensory data, that healthy subjects may not,” yet their figure appears to contradict this. In the figure, healthy controls show a clear discrepancy between priors and likelihoods, while for VSS patients, these distributions seem to align closely, suggesting little to no discrepancy. This contradiction undermines the internal consistency of the proposal.
Thank you again for highlighting this. We believe, that we weren’t concise enough in our argument. We hope that we were able to strengthen our argument in the revised manuscript and refer to page 3,2.1 Predictive Coding.
3. Even if the proposal was internally consistent, it is unclear whether the Bayesian framework can help in any meaningful way. Bayesian theories have been widely applied to various mental disorders, including autism and psychosis, but with mixed empirical support and significant overgeneralizations. For instance, Chrysaitis, N. A., & Seriès, P. (2023). 10 years of Bayesian theories of autism: a comprehensive review. Neuroscience & Biobehavioral Reviews, 145, 105022 comprehensively reviewed Bayesian theories of autism and found the empirical results to be inconsistent and inconclusive. Similarly, broader critiques (e.g., Karvelis, P., Paulus, M. P., & Diaconescu, A. O. (2023). Individual differences in computational psychiatry: A review of current challenges. Neuroscience & Biobehavioral Reviews, 148, 105137) highlight how the Bayesian accounts end up being too general, which does not agree with empirical observations. These challenges should temper the enthusiasm for applying the framework to VSS, especially in a handwavy and uncritical way.
Thank you for sharing so openly your point of view and highlighting pitfalls in using a Bayesian framework, which we partially share. We believe, that thinking of mechanistic frameworks in VSS may be useful in order to study the complex interrelations between sensory and cognitive factors at play. However, we agree that their applicability needs to be tested and empirically validated.
4. The authors introduce metacognition as a central component of their model, but the explanation feels implausible. Metacognition typically refers to high-level cognitive processes like self-monitoring and self-regulation, whereas the low-level perceptual disturbances in VSS seem far removed from these higher-order functions. There are multiple intermediary layers between low-level perception and metacognition. I don’t think it is accurate to refer to the mechanisms at the level of perceptual recalibration and gain control as metacognition.
We revised the manuscript and hope that we were able to convey more clearly why we believe that accounting for metacognition might be worthwhile thinking about in VSS (see 1.3 and 3)
This hypothesis not only stems from neurofunctional studies, that were able to show weakened activity in the pulvinar, a structure known to gate sensory information and generate perceptual confidence, but also from clinical observation as patients seem to be unable to manage and monitor effectively their sensory perception. Still, we thank the reviewer for his/her comments and believe that this needs to be accounted for empirically.
Reviewer 2 Report
Comments and Suggestions for AuthorsFor the authors
The article presents an interesting work, it introduces the concept of perceptual uncertainty and its impaired monitoring as potential mechanisms in VSS, offering a novel perspective on the condition's visual impairments. Before considered for acceptance, the authors are encouraged to address the following issues.
1. The authors' experimental theory on VSS is predicated on one of its pathogenic mechanisms, 'disruption of higher cortical visual information processing,' yet other potential mechanisms such as 'disruption of thalamocortical rhythm,'(ref.1) 'metabolic abnormalities,'(ref.2) and 'overexcitation of the visual cortex' (ref.3) and 'visual evoked potentials' (ref.4) are not sufficiently addressed. I suugest the author should note these oversights and clarified within the manuscript; and it is crucial for the authors to clarify whether their model is applicable to patients presenting with these alternative mechanisms. If not, the manuscript should describe inclusion or exclusion criteria of such patients to ensure the study's accuracy.
2. In the study, the author utilizes sensory stimuli as the foundational data for analysis. please elucidate 'specialised neural areas' and detail them within a table. Furthermore, it could be useful to conduct comparative analyses across distinct neural regions, such as Region 1 and Region 2, to ascertain their respective responses to the sensory stimuli.
3. In FIGURE A, the specific locations of the Posterior, Likelihood, and Prior within the flowchart should be clearly indicated. Additionally, a detailed explanation is required for the 'interaction' described in the legend: ' Our model of the world produces a prediction (prior) which is updated by sensory data (likelihood)'.
4. The author defines received stimuli as 'signals that exceed a threshold.' Is there clinical evidence to support the existence of a threshold for visual stimuli, such as a snowflake? Assuming that such neural damage persists, as in the theory of 'hyperexcited cortex,' how should the model be adjusted?
References:
1. Lauschke JL, Plant GT, Fraser CL. Visual snow: A thalamocortical dysrhythmia of the visual pathway? J Clin Neurosci. 2016 Jun;28:123-7. doi: 10.1016/j.jocn.2015.12.001. Epub 2016 Jan 11. PMID: 26791474.
2. Puledda F, Schankin C, Digre K, Goadsby PJ. Visual snow syndrome: what we know so far. Curr Opin Neurol. 2018 Feb;31(1):52-58. doi: 10.1097/WCO.0000000000000523. PMID: 29140814.
3. Chen WT, Lin YY, Fuh JL, Hämäläinen MS, Ko YC, Wang SJ. Sustained visual cortex hyperexcitability in migraine with persistent visual aura. Brain. 2011 Aug;134(Pt 8):2387-95. doi: 10.1093/brain/awr157. Epub 2011 Jul 4. PMID: 21729907.
4. Luna S, Lai D, Harris A. Antagonistic Relationship Between VEP Potentiation and Gamma Power in Visual Snow Syndrome. Headache. 2018 Jan;58(1):138-144. doi: 10.1111/head.13231. Epub 2017 Nov 29. PMID: 29193050.
Reviewer.
Author Response
The article presents an interesting work, it introduces the concept of perceptual uncertainty and its impaired monitoring as potential mechanisms in VSS, offering a novel perspective on the condition's visual impairments. Before considered for acceptance, the authors are encouraged to address the following issues.
- The authors' experimental theory on VSS is predicated on one of its pathogenic mechanisms, 'disruption of higher cortical visual information processing,' yet other potential mechanisms such as 'disruption of thalamocortical rhythm,'(ref.1) 'metabolic abnormalities,'(ref.2) and 'overexcitation of the visual cortex' (ref.3) and 'visual evoked potentials' (ref.4) are not sufficiently addressed. I suugest the author should note these oversights and clarified within the manuscript; and it is crucial for the authors to clarify whether their model is applicable to patients presenting with these alternative mechanisms. If not, the manuscript should describe inclusion or exclusion criteria of such patients to ensure the study's accuracy.
Thank you for this helpful comment. We expanded and revised the manuscript and tried to account for the neural underpinnings in VSS more extensively in order to explain conceptually why perceptual uncertainty and the impaired metacognition might explain the chronic misperceptions in these patients (see Section 1 Neural Underpinnings of Perceptual Uncertainty in VSS).
- In the study, the author utilizes sensory stimuli as the foundational data for analysis. please elucidate 'specialised neural areas' and detail them within a table. Furthermore, it could be useful to conduct comparative analyses across distinct neural regions, such as Region 1 and Region 2, to ascertain their respective responses to the sensory stimuli.
Sure. We tried to explicitly evoke the key areas in the revised manuscript under Section 1 Neural Underpinnings of Perceptual Uncertainty in VSS and created a table (Table A)enumerating the locations, the neural mechanism at stake with its functional consequence.
3. In FIGURE A, the specific locations of the Posterior, Likelihood, and Prior within the flowchart should be clearly indicated. Additionally, a detailed explanation is required for the 'interaction' described in the legend: ' Our model of the world produces a prediction (prior) which is updated by sensory data (likelihood)’.
Thank you fort your suggestion. We think we mistakingly put the legend of figure B in figure A. We hope to explain sufficiently in out updated manuscript text (Section 2) and in the now updated Figure B and legend.
4. The author defines received stimuli as 'signals that exceed a threshold.' Is there clinical evidence to support the existence of a threshold for visual stimuli, such as a snowflake? Assuming that such neural damage persists, as in the theory of 'hyperexcited cortex,' how should the model be adjusted?
Thank you for this very interesting comment and question. We believe that research by Emre Eren, providing evidence for a lower visual contrast sensitivity at 15Hz in patients with VSS (see reference below), can be read through the lens of a different threshold in detecting visual information.
We tried to account for this (1) threshold in signal detection (2) model adjustment in the situation of hyperexcitabiliy in section 2 of our revised manuscript, when introducing to the Signal Detection Theory. We therefore hope to answer to your questions in the revised manuscript.
Eren, O.E., Straube, A., Schöberl, F. et al. Age- and frequency-dependent changes in dynamic contrast perception in visual snow syndrome. J Headache Pain 22, 148 (2021). https://doi.org/10.1186/s10194-021-01355-y
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThank you for revising the manuscript.
Author Response
Thank you for revising the manuscript.
Thank you for your comments and questions that challenged our hypothesis and motivated us to expand on and strengthen our argument.
Reviewer 2 Report
Comments and Suggestions for AuthorsFor the authors,
The author has provided satisfactory explanations for the questions I raised. However, there are still some issues that need to be addressed:- I am unable to see the revision marks on the manuscript. Please highlight the changes in red (or another color) to make them visible.
- The revised figure (Figure B and legend) should retain the revision marks.
- The information in the revised table (Table A) is not visible; I can only see black. Please ensure that the table is properly displayed.
Author Response
The author has provided satisfactory explanations for the questions I raised. However, there are still some issues that need to be addressed:
Thank you for your patience considering our revision and again for your very helpful comments and questions in the first round that helped us improve and refine our argument.
- I am unable to see the revision marks on the manuscript. Please highlight the changes in red (or another color) to make them visible.
Yes, we understand your wish to easily find the changes. However, as we have been asked by the editors to expand the manuscript to minimum 2500 words, the initial structure changed importantly. Thus, to highlight with color would eventually highlight most parts of the manuscript. However, we numbered the lines in the manuscript, that you can access it easily. Basically the whole first section of our hypothesis "Neural Underpinnings of Perceptual Uncertainty" (line 35-92) has been designed to 1) make our argument more concise and espectially to 2) account for your questions.
2. The revised figure (Figure B and legend) should retain the revision marks.
We tried to highlight with color the changes made in the legend. Again, we hope that we were able to elaborate on the interaction between likelihood and prior beliefs in the section 2 of our manuscript "Formal Frameworks Explaining the Mechanisms" (line 95-116).
3. The information in the revised table (Table A) is not visible; I can only see black. Please ensure that the table is properly displayed.
We are very sorry for this and it is slightly embarrassing as we should have controlled for that. It ist due to an exportation problem (Pages to Microsoft Word).
I kindly ask the author to clearly mark all revisions on the manuscript so that I can review them according to the author's response.