GABA-ergic inhibition in human MT predicts visuo-spatial intelligence mediated by reverberation with frontal cortex

Article activity feed

1. 

   Version published to 10.7554/elife.97545.1 on eLife

   May 17, 2024
2. 

   Version published to 10.7554/elife.97545 on eLife

   May 17, 2024
3. 

   eLife

   May 16, 2024

   Author response:

   Thanks for the eLife assessment

   “This study employed a comprehensive approach to examining how the MT+ region integrates into a complex cognition system in mediating human visuo-spatial intelligence. While the findings are useful, the experimental evidence is incomplete and the study design, hypothesis, analyses, writing, and presentation need to be improved.” We plan to revise the manuscript according to the comments of Public Reviews.

   We are grateful for the excellent and very helpful comments, and now we address provisional author responses.

   Reviewer #1 (Public Review):

   Summary:

   The study of human intelligence has been the focus of cognitive neuroscience research, and finding some objective behavioral or neural indicators of intelligence has been an ongoing problem for scientists for many years. Melnick et al, …

   Author response:

   Thanks for the eLife assessment

   “This study employed a comprehensive approach to examining how the MT+ region integrates into a complex cognition system in mediating human visuo-spatial intelligence. While the findings are useful, the experimental evidence is incomplete and the study design, hypothesis, analyses, writing, and presentation need to be improved.” We plan to revise the manuscript according to the comments of Public Reviews.

   We are grateful for the excellent and very helpful comments, and now we address provisional author responses.

   Reviewer #1 (Public Review):

   Summary:

   The study of human intelligence has been the focus of cognitive neuroscience research, and finding some objective behavioral or neural indicators of intelligence has been an ongoing problem for scientists for many years. Melnick et al, 2013 found for the first time that the phenomenon of spatial suppression in motion perception predicts an individual's IQ score. This is because IQ is likely associated with the ability to suppress irrelevant information. In this study, a high-resolution MRS approach was used to test this theory. In this paper, the phenomenon of spatial suppression in motion perception was found to be correlated with the visuo-spatial subtest of gF, while both variables were also correlated with the GABA concentration of MT+ in the human brain. In addition, there was no significant relationship with the excitatory transmitter Glu. At the same time, SI was also associated with MT+ and several frontal cortex FCs.

   Strengths:

   (1) 7T high-resolution MRS is used.

   (2) This study combines the behavioral tests, MRS, and fMRI.

   Weaknesses:

   (1) In the intro, it seems to me that the multiple-demand (MD) regions are the key in this study. However, I didn't see any results associated with the MD regions. Did I miss something??

   Thank reviewer for pointing this out. After careful consideration, we agree with your point of view. According to the results of Melnick 2013, the motion surround suppression (SI) and the time thresholds of small and large gratings representing hMT+ functionality are correlated with Verbal Comprehension, Perceptual Reasoning, Working Memory, and Processing Speed Indicators, with correlation coefficients of 0.69, 0.47, 0.49, and 0.50, respectively. This suggests that hMT+ does have the potential to become the core of MD system. However, due to our results only delving into “the GABA-ergic inhibition in human MT predicts visuo-spatial intelligence mediated by reverberation with frontal cortex”, it is not yet sufficient to prove that hMT+is the core node of the MD system, we will adjust the explanatory logic of the article, that is, emphasizing the de-redundancy of hMT+ in visual-spatial intelligence and the improvement of information processing efficiency, while weakening the significance of hMT+ in MD systems.

   (2) How was the sample size determined? Is it sufficient??

   Thank reviewer for pointing this out. We use G*power to determine our sample size. In the study by Melnick (2013), they reported a medium effect between SI and Perception Reasoning sub-ability (r=0.47). Here we use this r value as the correlation coefficient (ρ H1), setting the power at the commonly used threshold of 0.8 and the alpha error probability at 0.05. The required sample size is calculated to be 26. This ensures that our study has adequate power to yield valid statistical results. Furthermore, compared to earlier within-subject studies like Schallmo et al.'s 2018 research, which used 22 datasets to examine GABA levels in MT+ and the early visual cortex (EVC), our study includes a more extensive dataset.

   (3) In Schallmo elife 2018, there was no correlation between GABA concentration and SI. How can we justify the different results different here?

   Thank reviewer for pointing this out. There are several differences between us:

   a. While the earlier study by Schallmo et al. (2018) employed 3T MRS, we utilize 7T MRS, enhancing our ability to detect and measure GABA with greater accuracy.

   b. Schallmo elife 2018 choose to use the bilateral hMT+ as the MRS measurement region while we use the left hMT+. The reason why we focus on left hMT+ are describe in reviewer 1. (6). Briefly, use of left MT/V5 as a target was motivated by studies demonstrating that left MT/V5 TMS is more effective at causing perceptual effects (Tadin et al., 2011).

   c. The resolution of MRS sequence in Schallmo elife 2018 is 3 cm isotropic voxel, while we apply 2 cm isotropic voxel. This helps us more precisely locate hMT+ and exclude more white matter signal.

   (4) Basically this study contains the data of SI, BDT, GABA in MT+ and V1, Glu in MT+ and V1-all 6 measurements. There should be 6x5/2 = 15 pairwise correlations. However, not all of these results are included in Figure 1 and supplementary 1-3. I understand that it is not necessary to include all figures. But I suggest reporting all values in one Table.

   We thank the reviewer for the good suggestion, we are planning to make a correlation matrix to reporting all values.

   (5) In Melnick (2013), the IQ scores were measured by the full set of WAIS-III, including all subtests. However, this study only used the visual spatial domain of gF. I wonder why only the visuo-spatial subtest was used not the full WAIS-III?

   We thank the reviewer for pointing this out. The decision was informed by Melnick’s findings which indicated high correlations between Surround suppression (SI) and the Verbal Comprehension, Perceptual Reasoning, Working Memory, and Processing Speed Indexes, with correlation coefficients of 0.69, 0.47, 0.49, and 0.50, respectively. It is well-established that the hMT+ region of the brain is a sensory cortex involved in visual perception processing (3D perception). Furthermore, motion surround suppression (SI), a specific function of hMT+, aligns closely with this region's activities. Given this context, the Perception Reasoning sub-ability was deemed to have the clearest mechanism for further exploration. Consequently, we selected the most representative subtest of Perception Reasoning—the Block Design Test—which primarily assesses 3D visual intelligence.

   (6) In the functional connectivity part, there is no explanation as to why only the left MT+ was set to the seed region. What is the problem with the right MT+?

   We thank the reviewer for pointing this out. The main reason is that our MRS ROI is the left hMT+, we would like to make different models’ ROI consistent to each other. Use of left MT/V5 as a target was motivated by studies demonstrating that left MT/V5 TMS is more effective at causing perceptual effects (Tadin et al., 2011). In addition, we will check the results of our localizer to confirm whether similar findings are consistently replicated.

   (7) In Melnick (2013), the authors also reported the correlation between IQ and absolute duration thresholds of small and large stimuli. Please include these analyses as well.

   We thank the reviewer for the good advice. Containing such result do help researchers compare the result between Melnick and us. We are planning to make such picture in the revised version.

   Reviewer #2 (Public Review):

   Summary:

   Recent studies have identified specific regions within the occipito-temporal cortex as part of a broader fronto-parietal, domain-general, or "multiple-demand" (MD) network that mediates fluid intelligence (gF). According to the abstract, the authors aim to explore the mechanistic roles of these occipito-temporal regions by examining GABA/glutamate concentrations. However, the introduction presents a different rationale: investigating whether area MT+ specifically, could be a core component of the MD network.

   Strengths:

   The authors provide evidence that GABA concentrations in MT+ and its functional connectivity with frontal areas significantly correlate with visuo-spatial intelligence performance. Additionally, serial mediation analysis suggests that inhibitory mechanisms in MT+ contribute to individual differences in a specific subtest of the Wechsler Adult Intelligence Scale, which assesses visuo-spatial aspects of gF.

   Weaknesses:

   (1) While the findings are compelling and the analyses robust, the study's rationale and interpretations need strengthening. For instance, Assem et al. (2020) have previously defined the core and extended MD networks, identifying the occipito-temporal regions as TE1m and TE1p, which are located more rostrally than MT+. Area MT+ might overlap with brain regions identified previously in Fedorenko et al., 2013, however the authors attribute these activations to attentional enhancement of visual representations in the more difficult conditions of their tasks. For the aforementioned reasons, It is unclear why the authors chose MT+ as their focus. A stronger rationale for this selection is necessary and how it fits with the core/extended MD networks.

   We really appreciate reviewer’s opinions. The reason why we focus on hMT+ is following: According to the results of Melnick 2013, the motion surround suppression (SI) and the time thresholds of small and large gratings representing hMT+ functionality are correlated with Verbal Comprehension, Perceptual Reasoning, Working Memory, and Processing Speed Indicators, with high correlation coefficients of 0.69, 0.47, 0.49, and 0.50, respectively. In addition, Fedorenko et al. 2013, the averaged MD activity region appears to overlap with hMT+. Based on these findings, we assume that hMT+ does have the potential to become the core of MD system.

   (2) Moreover, although the study links MT+ inhibitory mechanisms to a visuo-spatial component of gF, this evidence alone may not suffice to position MT+ as a new core of the MD network. The MD network's definition typically encompasses a range of cognitive domains, including working memory, mathematics, language, and relational reasoning. Therefore, the claim that MT+ represents a new core of MD needs to be supported by more comprehensive evidence.

   Thank reviewer for pointing this out. After careful consideration, we agree with your point of view. Due to our results only delving into visuo-spatial intelligence, it is not yet sufficient to prove that hMT is the core node of the MD system. We will adjust the explanatory logic of the article, that is, emphasizing the de-redundancy of hMT+in visual-spatial intelligence and the improvement of information processing efficiency, while weakening the significance of hMT+ in MD systems.

   Reviewer #3 (Public Review):

   Summary:

   This manuscript aims to understand the role of GABA-ergic inhibition in the human MT+ region in predicting visuo-spatial intelligence through a combination of behavioral measures, fMRI (for functional connectivity measurement), and MRS (for GABA/glutamate concentration measurement). While this is a commendable goal, it becomes apparent that the authors lack fundamental understanding of vision, intelligence, or the relevant literature. As a result, the execution of the research is less coherent, dampening the enthusiasm of the review.

   Strengths:

   (1) Comprehensive Approach: The study adopts a multi-level approach, i.e., neurochemical analysis of GABA levels, functional connectivity, and behavioral measures to provide a holistic understanding of the relationship between GABA-ergic inhibition and visuo-spatial intelligence.

   (2) Sophisticated Techniques: The use of ultra-high field magnetic resonance spectroscopy (MRS) technology for measuring GABA and glutamate concentrations in the MT+ region is a recent development.

   Weaknesses:

   Study Design and Hypothesis

   (1) The central hypothesis of the manuscript posits that "3D visuo-spatial intelligence (the performance of BDT) might be predicted by the inhibitory and/or excitation mechanisms in MT+ and the integrative functions connecting MT+ with the frontal cortex." However, several issues arise:

   (1.1) The Suppression Index depicted in Figure 1a, labeled as the "behavior circle," appears irrelevant to the central hypothesis.

   We thank the reviewer for pointing this out. In our study, the inhibitory mechanisms in hMT+ are conceptualized through two models: the neurotransmitter model and the behavior model. The Suppression Index is essential for elucidating the local inhibitory mechanisms within behavior model. However, we acknowledge that our initial presentation in the introduction may not have clearly articulated our hypothesis, potentially leading to misunderstandings. We plan to revise the introduction to better clarify these connections and ensure the relevance of the Suppression Index is comprehensively understood.

   (1.2) The construct of 3D visuo-spatial intelligence, operationalized as the performance in the Block Design task, is inconsistently treated as another behavioral task throughout the manuscript, leading to confusion.

   We thank the reviewer for pointing this out. We acknowledge that our manuscript may have inconsistently presented this construct across different sections, causing confusion. To address this, we plan to ensure a consistent description of 3D visuo-spatial intelligence in both the introduction and the discussion sections. But we would like to maintain 'Block Design task score' within the results section to help readers clarify which subtest we use.

   (1.3) The schematics in Figure 1a and Figure 6 appear too high-level to be falsifiable. It is suggested that the authors formulate specific and testable hypotheses and preregister them before data collection.

   We thank the reviewer for pointing this out. We are planning to revise the Figure 1a and make it less abstract and more logical. For Figure 6, the schematic represents our theoretical framework of how hMT+ works in the 3D viso-spatial intelligence, we believe the elements within this framework are grounded in related theories and supported by evidence discussed in our results and discussions section, making them specific and testable.

   (2) Central to the hypothesis and design of the manuscript is a misinterpretation of a prior study by Melnick et al. (2013). While the original study identified a strong correlation between WAIS (IQ) and the Suppression Index (SI), the current manuscript erroneously asserts a specific relationship between the block design test (from WAIS) and SI. It should be noted that in the original paper, WAIS comprises Similarities, Vocabulary, Block design, and Matrix reasoning tests in Study 1, while the complete WAIS is used in Study 2. Did the authors conduct other WAIS subtests other than the block design task?

   Thanks for pointing this out. Reviewer #1 also asked this question, we copy the answers in here “The decision was informed by Melnick’s findings which indicated high correlations between Surround suppression (SI) and the Verbal Comprehension, Perceptual Reasoning, Working Memory, and Processing Speed Indexes, with correlation coefficients of 0.69, 0.47, 0.49, and 0.50, respectively. It is well-established that the hMT+ region of the brain is a sensory cortex involved in visual perception processing (3D perception). Furthermore, motion surround suppression (SI), a specific function of hMT+, aligns closely with this region's activities. Given this context, the Perception Reasoning sub-ability was deemed to have the clearest mechanism for further exploration. Consequently, we selected the most representative subtest of Perception Reasoning—the Block Design Test—which primarily assesses 3D visual intelligence.”

   (3) Additionally, there are numerous misleading references and unsubstantiated claims throughout the manuscript. As an example of misleading reference, "the human MT ... a key region in the multiple representations of sensory flows (including optic, tactile, and auditory flows) (Bedny et al., 2010; Ricciardi et al., 2007); this ideally suits it to be a new MD core." The two references in this sentence are claims about plasticity in the congenitally blind with sensory deprivation from birth, which is not really relevant to the proposal that hMT+ is a new MD core in healthy volunteers.

   Thanks for pointing this out. We have carefully read the corresponding references and considered the corresponding theories and agree with these comments. Due to our results only delving into “the GABA-ergic inhibition in human MT predicts visuo-spatial intelligence mediated by reverberation with frontal cortex”, it is not yet sufficient to prove that hMT+ is the core node of the MD system, we will adjust the explanatory logic of the article, that is, emphasizing the de redundancy of hMT+in visual-spatial intelligence and the improvement of information processing efficiency, while weakening the significance of hMT+ in MD systems. In addition, regarding the potential central role of hMT+ in the MD system, we agree with your view that research on hMT+ as a multisensory integration hub mainly focuses on developmental processes. Meanwhile, in adults, the MST region of hMT+ is considered a multisensory integration area for visual and vestibular inputs, which potentially supports the role of hMT+ in multitasking multisensory systems (Gu et al., J. Neurosci, 26(1), 73–85, 2006; Fetsch et al., Nat. Neurosci, 15, 146–154, 2012.). Further research could explore how other intelligence sub-ability such as working memory and language comprehension are facilitated by hMT+'s features.

   Another example of unsubstantiated claim: the rationale for selecting V1 as the control region is based on the assertion that "it mediates the 2D rather than 3D visual domain (Born & Bradley, 2005)". That's not the point made in the Born & Bradley (2005) paper on MT. It's crucial to note that V1 is where the initial binocular convergence occurs in cortex, i.e., inputs from both the right and left eyes to generate a perception of depth.

   Thank you for pointing this out. We acknowledge the inappropriate citation of "Born & Bradley, 2005," which focuses solely on the structure and function of the visual area MT. However, we believe that choosing hMT+ as the domain for 3D visual analysis and V1 as the control region is justified. Cumming and DeAngelis (Annu Rev Neurosci, 24:203–238.2001) state that binocular disparity provides the visual system with information about the three-dimensional layout of the environment, and the link between perception and neuronal activity is stronger in the extrastriate cortex (especially MT) than in the primary visual cortex(V1). This supports our choice and emphasizes the relevance of MT+ in our study. We will revise our reference in the revised version.

   Results & Discussion

   (1) The missing correlation between SI and BDT is crucial to the rest of the analysis. The authors should discuss whether they replicated the pattern of results from Melnick et al. (2013) despite using only one WAIS subtest.

   We thank for reviewer’s suggestion. Now the correlation result is placed in the supplemental material, we will put it back to the main text.

   (2) ROIs: can the authors clarify if the results are based on bilateral MT+/V1 or just those in the left hemisphere? Can the authors plot the MRS scan area in V1? I would be surprised if it's precise to V1 and doesn't spread to V2/3 (which is fine to report as early visual cortex).

   We thank for reviewer’s suggestion. We plan to draw the V1 ROI MRS scanning area and use the visual template to check if the scanning area contains V2/3. If it does, we will refer to it as the early visual cortex rather than specifically V1 in our reporting.

   (3) Did the authors examine V1 FC with either the frontal regions and/or whole brain, as a control analysis? If not, can the author justify why V1 serves as the control region only in the MRS but not in FC (Figure 4) or the mediation analysis (Figure 5)? That seems a little odd given that control analyses are needed to establish the specificity of the claim to MT+

   We thank for reviewer’s suggestion. We plan to do the V1 FC-behavior connection as control analysis. For mediation analysis, since V1 GABA/Glu has no correlation with BDT score, it is not sufficient to apply mediation analysis.

   (4) It is not clear how to interpret the similarity or difference between panels a and b in Figure 4.

   We thank reviewer for pointing this out. We plan to further interpret the difference between a and b in the revised version. Panels a represents BDT score correlated hMT+-region FC, which is obviously involved in frontal cortex. While panels b represents SI correlated hMT+-region FC, which shows relatively less regions. The overlap region is what we are interested in and explain how local inhibitory mechanisms works in the 3D viso-spatial intelligence. In addition, we would like to revise Figure 4 and point out the overlap region.

   (5) SI is not relevant to the authors‘ priori hypothesis, but is included in several mediation analyses. Can the authors do model comparisons between the ones in Figure 5c, d, and Figure S6? In other words, is SI necessary in the mediation model? There seem discrepancies between the necessity of SI in Figures 5c/S6 vs. Figure 5d.

   We thank the reviewer for highlighting this point. The relationship between the Suppression Index (SI) and our a priori hypotheses is elaborated in the response to reviewer 3, section (1). SI plays a crucial role in explicating how local inhibitory mechanisms function within the context of the 3D visuo-spatial task. Additionally, Figure 5c illustrates the interaction between the frontal cortex and hMT+, showing how the effects from the frontal cortex (BA46) on the Block Design Task are fully mediated by SI. This further underscores the significance of SI in our model.

   (6) The sudden appearance of "efficient information" in Figure 6, referring to the neural efficiency hypothesis, raises concerns. Efficient visual information processing occurs throughout the visual cortex, starting from V1. Thus, it appears somewhat selective to apply the neural efficiency hypothesis to MT+ in this context.

   We thank the reviewer for highlighting this point. There is no doubt that V1 involved in efficient visual information processing. However, in our result, the V1 GABA has no significant correlation between BDT score, suggesting that the V1 efficient processing might not sufficiently account for the individual differences in 3D viso-spatial intelligence. Additionally, we will clarify our use of the neural efficiency hypothesis by incorporating it into the introduction of our paper to better frame our argument.

   Transparency Issues:

   (1) Don't think it's acceptable to make the claim that "All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary information". It is the results or visualizations of data analysis, rather than the raw data themselves, that are presented in the paper/supp info.

   We thank reviewer for pointing this out. We realized that such expression will lead to confusion. We will delete this expression.

   (2) No GitHub link has been provided in the manuscript to access the source data, which limits the reproducibility and transparency of the study.

   We thank reviewer for pointing this out. We will attach the GitHub link in the revised version.

   Minor:

   "Locates" should be replaced with "located" throughout the paper. For example: "To investigate this issue, this study selects the human MT complex (hMT+), a region located at the occipito-temporal border, which represents multiple sensory flows, as the target brain area."

   We thank reviewer for pointing this out. We will revise it.

   Use "hMT+" instead of "MT+" to be consistent with the term in the literature.

   We thank reviewer for pointing this out. We agree to use hMT+ in the literature.

   "Green circle" in Figure 1 should be corrected to match its actual color.

   We thank reviewer for pointing this out. We will revise it.

   The abbreviation for the Wechsler Adult Intelligence Scale should be "WAIS," not "WASI."

   We thank reviewer for pointing this out. We will revise it.

   Read the original source
4. 

   eLife

   May 16, 2024

   eLife assessment

   This study employed a comprehensive approach to examining how the MT+ region integrates into a complex cognition system in mediating human visuo-spatial intelligence. While the findings are useful, the experimental evidence is incomplete and the study design, hypothesis, analyses, writing, and presentation need to be improved. The work will be of interest to researchers in psychology, cognitive science, and neuroscience.

   Read the original source
5. 

   eLife

   May 16, 2024

   Reviewer #1 (Public Review):

   Summary:

   The study of human intelligence has been the focus of cognitive neuroscience research, and finding some objective behavioral or neural indicators of intelligence has been an ongoing problem for scientists for many years. Melnick et al, 2013 found for the first time that the phenomenon of spatial suppression in motion perception predicts an individual's IQ score. This is because IQ is likely associated with the ability to suppress irrelevant information. In this study, a high-resolution MRS approach was used to test this theory. In this paper, the phenomenon of spatial suppression in motion perception was found to be correlated with the visuo-spatial subtest of gF, while both variables were also correlated with the GABA concentration of MT+ in the human brain. In addition, there was no significant …

   Reviewer #1 (Public Review):

   Summary:

   The study of human intelligence has been the focus of cognitive neuroscience research, and finding some objective behavioral or neural indicators of intelligence has been an ongoing problem for scientists for many years. Melnick et al, 2013 found for the first time that the phenomenon of spatial suppression in motion perception predicts an individual's IQ score. This is because IQ is likely associated with the ability to suppress irrelevant information. In this study, a high-resolution MRS approach was used to test this theory. In this paper, the phenomenon of spatial suppression in motion perception was found to be correlated with the visuo-spatial subtest of gF, while both variables were also correlated with the GABA concentration of MT+ in the human brain. In addition, there was no significant relationship with the excitatory transmitter Glu. At the same time, SI was also associated with MT+ and several frontal cortex FCs.

   Strengths:

   (1) 7T high-resolution MRS is used.

   (2) This study combines the behavioral tests, MRS, and fMRI.

   Weaknesses:

   (1) In the intro, it seems to me that the multiple-demand (MD) regions are the key in this study. However, I didn't see any results associated with the MD regions. Did I miss something??

   (2) How was the sample size determined? Is it sufficient??

   (3) In Schallmo elife 2018, there was no correlation between GABA concentration and SI. How can we justify the different results different here?

   (4) Basically this study contains the data of SI, BDT, GABA in MT+ and V1, Glu in MT+ and V1-all 6 measurements. There should be 6x5/2 = 15 pairwise correlations. However, not all of these results are included in Figure 1 and supplementary 1-3. I understand that it is not necessary to include all figures. But I suggest reporting all values in one Table.

   (5) In Melnick (2013), the IQ scores were measured by the full set of WAIS-III, including all subtests. However, this study only used the visual spatial domain of gF. I wonder why only the visuo-spatial subtest was used not the full WAIS-III?

   (6) In the functional connectivity part, there is no explanation as to why only the left MT+ was set to the seed region. What is the problem with the right MT+?

   (7) In Melnick (2013), the authors also reported the correlation between IQ and absolute duration thresholds of small and large stimuli. Please include these analyses as well.

   Read the original source
6. 

   eLife

   May 16, 2024

   Reviewer #2 (Public Review):

   Summary:

   Recent studies have identified specific regions within the occipito-temporal cortex as part of a broader fronto-parietal, domain-general, or "multiple-demand" (MD) network that mediates fluid intelligence (gF). According to the abstract, the authors aim to explore the mechanistic roles of these occipito-temporal regions by examining GABA/glutamate concentrations. However, the introduction presents a different rationale: investigating whether area MT+ specifically, could be a core component of the MD network.

   Strengths:

   The authors provide evidence that GABA concentrations in MT+ and its functional connectivity with frontal areas significantly correlate with visuo-spatial intelligence performance. Additionally, serial mediation analysis suggests that inhibitory mechanisms in MT+ contribute to …

   Reviewer #2 (Public Review):

   Summary:

   Recent studies have identified specific regions within the occipito-temporal cortex as part of a broader fronto-parietal, domain-general, or "multiple-demand" (MD) network that mediates fluid intelligence (gF). According to the abstract, the authors aim to explore the mechanistic roles of these occipito-temporal regions by examining GABA/glutamate concentrations. However, the introduction presents a different rationale: investigating whether area MT+ specifically, could be a core component of the MD network.

   Strengths:

   The authors provide evidence that GABA concentrations in MT+ and its functional connectivity with frontal areas significantly correlate with visuo-spatial intelligence performance. Additionally, serial mediation analysis suggests that inhibitory mechanisms in MT+ contribute to individual differences in a specific subtest of the Wechsler Adult Intelligence Scale, which assesses visuo-spatial aspects of gF.

   Weaknesses:

   While the findings are compelling and the analyses robust, the study's rationale and interpretations need strengthening. For instance, Assem et al. (2020) have previously defined the core and extended MD networks, identifying the occipito-temporal regions as TE1m and TE1p, which are located more rostrally than MT+. Area MT+ might overlap with brain regions identified previously in Fedorenko et al., 2013, however the authors attribute these activations to attentional enhancement of visual representations in the more difficult conditions of their tasks. For the aforementioned reasons, It is unclear why the authors chose MT+ as their focus. A stronger rationale for this selection is necessary and how it fits with the core/extended MD networks.

   Moreover, although the study links MT+ inhibitory mechanisms to a visuo-spatial component of gF, this evidence alone may not suffice to position MT+ as a new core of the MD network. The MD network's definition typically encompasses a range of cognitive domains, including working memory, mathematics, language, and relational reasoning. Therefore, the claim that MT+ represents a new core of MD needs to be supported by more comprehensive evidence.

   Read the original source
7. 

   eLife

   May 16, 2024

   Reviewer #3 (Public Review):

   Summary:

   This manuscript aims to understand the role of GABA-ergic inhibition in the human MT+ region in predicting visuo-spatial intelligence through a combination of behavioral measures, fMRI (for functional connectivity measurement), and MRS (for GABA/glutamate concentration measurement). While this is a commendable goal, it becomes apparent that the authors lack fundamental understanding of vision, intelligence, or the relevant literature. As a result, the execution of the research is less coherent, dampening the enthusiasm of the review.

   Strengths:

   (1) Comprehensive Approach: The study adopts a multi-level approach, i.e., neurochemical analysis of GABA levels, functional connectivity, and behavioral measures to provide a holistic understanding of the relationship between GABA-ergic inhibition and …

   Reviewer #3 (Public Review):

   Summary:

   This manuscript aims to understand the role of GABA-ergic inhibition in the human MT+ region in predicting visuo-spatial intelligence through a combination of behavioral measures, fMRI (for functional connectivity measurement), and MRS (for GABA/glutamate concentration measurement). While this is a commendable goal, it becomes apparent that the authors lack fundamental understanding of vision, intelligence, or the relevant literature. As a result, the execution of the research is less coherent, dampening the enthusiasm of the review.

   Strengths:

   (1) Comprehensive Approach: The study adopts a multi-level approach, i.e., neurochemical analysis of GABA levels, functional connectivity, and behavioral measures to provide a holistic understanding of the relationship between GABA-ergic inhibition and visuo-spatial intelligence.

   (2) Sophisticated Techniques: The use of ultra-high field magnetic resonance spectroscopy (MRS) technology for measuring GABA and glutamate concentrations in the MT+ region is a recent development.

   Weaknesses:

   Study Design and Hypothesis
   (1) The central hypothesis of the manuscript posits that "3D visuo-spatial intelligence (the performance of BDT) might be predicted by the inhibitory and/or excitation mechanisms in MT+ and the integrative functions connecting MT+ with the frontal cortex." However, several issues arise:
   1.1 The Suppression Index depicted in Figure 1a, labeled as the "behavior circle," appears irrelevant to the central hypothesis.
   1.2 The construct of 3D visuo-spatial intelligence, operationalized as the performance in the Block Design task, is inconsistently treated as another behavioral task throughout the manuscript, leading to confusion.
   1.3 The schematics in Figure 1a and Figure 6 appear too high-level to be falsifiable. It is suggested that the authors formulate specific and testable hypotheses and preregister them before data collection.

   (2) Central to the hypothesis and design of the manuscript is a misinterpretation of a prior study by Melnick et al. (2013). While the original study identified a strong correlation between WAIS (IQ) and the Suppression Index (SI), the current manuscript erroneously asserts a specific relationship between the block design test (from WAIS) and SI. It should be noted that in the original paper, WAIS comprises Similarities, Vocabulary, Block design, and Matrix reasoning tests in Study 1, while the complete WAIS is used in Study 2. Did the authors conduct other WAIS subtests other than the block design task?

   (3) Additionally, there are numerous misleading references and unsubstantiated claims throughout the manuscript. As an example of misleading reference, "the human MT ... a key region in the multiple representations of sensory flows (including optic, tactile, and auditory flows) (Bedny et al., 2010; Ricciardi et al., 2007); this ideally suits it to be a new MD core." The two references in this sentence are claims about plasticity in the congenitally blind with sensory deprivation from birth, which is not really relevant to the proposal that hMT+ is a new MD core in healthy volunteers.
   Another example of unsubstantiated claim: the rationale for selecting V1 as the control region is based on the assertion that "it mediates the 2D rather than 3D visual domain (Born & Bradley, 2005)". That's not the point made in the Born & Bradley (2005) paper on MT. It's crucial to note that V1 is where the initial binocular convergence occurs in cortex, i.e., inputs from both the right and left eyes to generate a perception of depth.

   Results & Discussion
   (1) The missing correlation between SI and BDT is crucial to the rest of the analysis. The authors should discuss whether they replicated the pattern of results from Melnick et al. (2013) despite using only one WAIS subtest.

   (2) ROIs: can the authors clarify if the results are based on bilateral MT+/V1 or just those in the left hemisphere? Can the authors plot the MRS scan area in V1? I would be surprised if it's precise to V1 and doesn't spread to V2/3 (which is fine to report as early visual cortex).

   (3) Did the authors examine V1 FC with either the frontal regions and/or whole brain, as a control analysis? If not, can the author justify why V1 serves as the control region only in the MRS but not in FC (Figure 4) or the mediation analysis (Figure 5)? That seems a little odd given that control analyses are needed to establish the specificity of the claim to MT+.

   (4) It is not clear how to interpret the similarity or difference between panels a and b in Figure 4.

   (5) SI is not relevant to the authors' priori hypothesis, but is included in several mediation analyses. Can the authors do model comparisons between the ones in Figure 5c, d, and Figure S6? In other words, is SI necessary in the mediation model? There seem discrepancies between the necessity of SI in Figures 5c/S6 vs. Figure 5d.

   (6) The sudden appearance of "efficient information" in Figure 6, referring to the neural efficiency hypothesis, raises concerns. Efficient visual information processing occurs throughout the visual cortex, starting from V1. Thus, it appears somewhat selective to apply the neural efficiency hypothesis to MT+ in this context.

   Transparency Issues:
   (1) Don't think it's acceptable to make the claim that "All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary information". It is the results or visualizations of data analysis, rather than the raw data themselves, that are presented in the paper/supp info.

   (2) No GitHub link has been provided in the manuscript to access the source data, which limits the reproducibility and transparency of the study.

   Minor:
   "Locates" should be replaced with "located" throughout the paper. For example: "To investigate this issue, this study selects the human MT complex (hMT+), a region located at the occipito-temporal border, which represents multiple sensory flows, as the target brain area."

   Use "hMT+" instead of "MT+" to be consistent with the term in the literature.

   "Green circle" in Figure 1 should be corrected to match its actual color.

   The abbreviation for the Wechsler Adult Intelligence Scale should be "WAIS," not "WASI."

   Read the original source
8. 

   Version published to 10.1101/2024.03.20.585863v1 on bioRxiv

   Mar 22, 2024