Interindividual Brain and Behavior Differences in Adaptation to Unexpected Uncertainty
Round 1
Reviewer 1 Report
Interindividual brain and behavior differences in adaptation to unexpected uncertainty
Soussi, C., Berthoz, S., Chirokoff, V., and Chanraud, S.
Summary
Soussi and colleagues report on a sample of 28 individuals who participated in a modified associative learning paradigm while undergoing fMRI scanning. The paradigm was separated into learning and cheat phases, and resting-state scans were acquired pre-task, between phases, and post-task. From the task phases, the authors derived measures of exploitation and cost of uncertainty for each individual. The authors also computed change in resting-state functional connectivity between two of the resting-state scans for each individual to identify flexible connections. A median split on the cost of uncertainty variable divided the sample into groups who exhibited higher and lower cost of uncertainty. Groups displayed differential changes in resting-state functional connectivity. In the full sample, the exploitation index was also associated with specific connectivity changes, as were various behavioral measures.
Overall, the manuscript has promise, introducing a novel way to look at explore/exploit behaviors and tolerance of uncertainty. However, the manuscript is lacking in some key areas—namely motivation, sufficient discussion of past work, prediction, and coherent presentations of analyses/results—which limit its readability and impact.
General Comments
-The introduction would benefit greatly from more detail and organization. As the authors are presenting a novel explore/exploit paradigm, it would be really helpful for them to discuss the tasks that have been used in previous findings, not only to contextualize the findings they present but also to discuss why they decided to deviate from what’s been done. It would also be useful for them to discuss their predictions. Specifically, it’s not at all clear why they chose to look at change in resting-state functional connectivity (between 2/3 of the scans) instead of task-based connectivity.
-My main concern has to do with the validity of the exploitation and uncertainty indices. As they didn’t relate to any behavioral variables, how are the authors sure that they are measuring what they intend to measure without any evidence from previous findings?
-The methods section is also lacking in detail and coherence. I’ve pin-pointed specific places below.
-The procedure figure is helpful. Perhaps the authors could either add to this figure or add another figure that illustrated their analyses of interest?
-The connectivity figure is not conducive to understanding the results. Please consider an alternative way to depict these findings.
-The use of the word ‘networks’ is misleading as only regional connectivity is explored.
Specific Comments
Abstract
-The abstract would benefit from some organization overall: 1) it would help to say that resting-state was collected before and after task fMRI to understand what “task-related changes in intrinsic brain network connectivity” refers. 2) The beginning mentions that brain network relationships to behavior are overlooked, but the results mention describe connectivity between regions, please fix. 3) The beginning of the last line is a bit misleading as no hypotheses were introduced. Consider editing to “These results demonstrate that the exploration/exploitation trade-off involves…”
Introduction
-P2 L75-80: I don’t follow the reasoning here. It would be helpful to unpack this and provide references.
-P2 L81-82: Please describe the tasks used here.
-P2. L90-92: “To do so… those two kinds of behavior.” This sentence is somewhat vague. What do you mean by internal and external stimuli and why do they have to be considered continuously?
-P3. L111-113: “Thus.. architecture.” What limits? Why is rest more useful than task in the present study?
-Please include some hypotheses. What changes in rFC do you expect at the network level? And how do you predict these will relate to behavior?
Methods
-18-60 is a wide range, with participants likely already presenting with some age effects in rFC and explore/exploit behavior. Can you provide some justification for this range?
-Eyes closed resting state: how did you ensure that participants were not sleeping?
-Acquisition: please include task fMRI acquisition parameters. Multi-echo acquisition mentioned, but only one TE provided. Was it multi-echo or single-echo? Processing suggests single-echo.
-Perhaps it’s just me, but I’m having trouble following the calculation of scores. Why were only the first and last 60 trials from each phase used? When calculating accuracy in the CHEAT phase, can you confirm that this was independent of the feedback provided? More broadly, this seems like an indirect way of measuring exploration/exploitation—is there any precedent for this task in the literature?
-Please include descriptive statistics for the number of trials in learning phase for participants
-Is there any way to sanity check/validate your exploitation/uncertainty indices of interest? Do you have any additional behavioral data from outside of the scanner on individual differences in uncertainty and/or exploration/exploitation?
-Please specify which resting-state scan was used as the ‘before.’ Pre-task? Or pre-cheat? If the latter, please explain why you wouldn’t use the pre-task if you were interested in comparing flexibility from ‘intrinsic functional architecture.’
-Based on the FC methods, ‘networks’ is not the right word to use throughout the manuscript. Please edit.
Results
-I find the rFC results hard to follow. Please make the sentences shorter and edit for clarity.
-Please remove mention of ‘non-significant but non-negligible’ correlations. Given the small sample size, it is misleading to report these if not significant.
Figures
-Figure 2: It’s not clear what the circle and triangle represent.
-Figure 3: This view of the brain images makes it very hard to see the connections as many are in frontal cortex. Please consider showcasing these in a different way (e.g, midsagittal and lateral views may be more appropriate?)
Discussion
-P9 L391: “Overall and as expected” seems out of place since no hypotheses were mentioned.
-P10 L403-405: “At the brain level… task consolidation.” It isn’t contentious that rFC is altered by task. Consider editing this statement to be more specific to what you did.
-P12 L522-525: “Herein… explicit choice.” This information would be useful in the introduction when talking about previous task-related findings.
Only surface edits needed
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
1. In the introduction section, the authors mention that there is only one resting-state functional connectivity study to investigate the exploration/exploitation trade-off. I think it is necessary for the authors to describe that study in detail and, wherever possible, to explain the relationship between that study and the current study.
2. The authors mentioned that their second aim was to explore psychological factors that may relate to the interindividual variability in balancing explorative and exploitative behaviors as well as changes in rFC. It is necessary for the authors to explain why these psychological factors were examined and to elaborate further on them.
3. The amounts of subjects need to be reported with details of the basic steps taken to determine the amounts of subjects in the Methods section, whether any subjects were rejected during MRI pre-processing due to excessive head movement, the criteria for rejection, etc. Importantly, Due to the between-subjects design used in this study, I doubt that the number of subjects in this study was sufficient, and I ask the authors to provide statistical effect sizes.
4. Explorative and exploitative behaviors were inferred using accuracy during the CHEAT phase and focused only on trials involving stimuli (geometrical figure) for which the accuracy reached at least 80% of correct responses at the end of LEARNING (last 30% of trials). Is there an appropriate scientific basis or reference? Correspondingly, the exploration index and cost of uncertainty mentioned later
5. All results were corrected for multiple comparison using a False Discovery Rate (FDR) correction with a α level of 0.05. What was the p-value before uncorrected and what was the threshold?
6. The presentation of statistical indicators needs to be harmonized, e.g., p-values with 2 or 3 decimal places.
7. The authors provide an overview of functional brain regions relevant to exploration and exploitation in the introduction section, but there are differences between these brain regions and what the authors have found, why there are such differences, and what such differences indicate, which the authors need to discuss more fully in the discussion section.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
The authors have addressed some of my comments, but this has been done somewhat piecemeal. The introduction has more information, but the motivation for the study and the chosen methods are not coherent. I also still struggle with the readability of the methods (particularly the calculation of dependent variables), and the brain-behavior correlations. The text can be streamlined to be more clear.
Introduction
-I appreciate that the authors have added a description of other tasks, but it’s not well-integrated in the text. Why the choice to deviate from other tasks? What measure can be extracted from the new task that is important for the present study?
-It strikes me as a bit strange to hypothesize about locus coeruleus connectivity with the reported acquisition protocol. Is there even adequate SNR to detect it?
Methods
-18-60 is a wide range, with participants likely already presenting with some age effects in rFC and explore/exploit behavior. Can you provide some justification for this range?
-Eyes closed resting state: how did you ensure that participants were not sleeping?
-p7,l279: I think you mean “The functional images were warped...”
-P8 fmri analyses/figure: please align the text with the figure. Was change in FC statically tested (as indicated in the text) or was it a simple subtraction (as indicated in the figure)? Did the regression with exploitation index on change in FC include group as a predictor or was the regression carried out in groups separately? Please clarify in text.
Results
-Figure 3 is now very confusing. What are the different colors? How is the difference between end of learning and end of cheat significant? The bars are overlapping. Please indicate what the bars represent. Please label y-axis.
-Figure 4: Were there any significant differences here? Please label y-axis.
-Organization of FC results is much better. Thank you!
-Please remove mention of ‘non-significant but non-negligible’ correlations. Given the small sample size, it is misleading to report these if not significant.
NA
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 2 Report
The authors have answered my concerns very well
Author Response
We thank the reviewer for his work on our manuscript.
