A Synergistic Workspace for Human Consciousness Revealed by Integrated Information Decomposition

Reviewer #2 (Public Review):

The authors analysed functional MRI recordings of brain activity at rest, using state-of-the-art methods that reveal the diverse ways in which the information can be integrated in the brain. In this way, they found brain areas that act as (synergistic) gateways for the 'global workspace', where conscious access to information or cognition would occur, and brain areas that serve as (redundant) broadcasters from the global workspace to the rest of the brain. The results are compelling and consisting with the already assumed role of several networks and areas within the Global Neuronal Workspace framework. Thus, in a way, this work comes to stress the role of synergy and redundancy as complementary information processing modes, which fulfill different roles in the big context of information integration.
In addition, to prove that the identified high-order interactions are relevant to the phenomenon of consciousness, the same analysis was performed in subjects under anesthesia or with disorders of consciousness (DOC), showing that indeed the loss of consciousness is associated with a deficient integration of information within the gateway regions.

However, there is still a standing issue that could be the basis for an improved analysis: the concepts of gateways and broadcasters allude to a directionality in the information flow. In fact, Figure 1 depicts Stage (i) and Stage (iii) as one-way processes. However, the identification of gateway and broadcaster regions relies on matrices that are symmetrical, i.e. they are not directed. Would it be possible to assess the gateway or broadcaster nature of a region taking into account the directionality of the information flow? In other words, if region X is a gateway, I would expect a synergistic relationship between the past of X,Y and present of Y (Y not being a gateway) towards the present of X; but not necessarily the other way around (i.e. the present of Y being less dependent on the past/present of X). A similar reasoning can be made for broadcasters.

Although regional differences in haemodynamics complicate attempts to map directed information flow from fMRI recordings, perhaps the IID framework could be leveraged to extract directed data (i.e., there are many atoms that are explicitly directed). As an avenue for future research, it would be interesting to discuss the feasibility or limitations of such analysis.

Also, there is something confusing in Figure 4B-C and its description. Awake should be similar to recovery (they are both awake, aren't they? Not much info is given, anyway); thus it seems counterintuitive that anesthesia minus awake looks so different than anesthesia minus recovery. The first is mostly blue-ish and the second is mostly red. Is it possible that Figure 4C is actually recovery minus anesthesia? That would make much more sense, also for Figure 4D. Please correct me if I am wrong.

Reviewer #3 (Public Review):

The work proposes a model of neural information processing based on a 'synergistic global workspace,' which processes information in three principal steps: a gatekeeping step (information gathering), an information integration step, and finally, a broadcasting step. They provided an interpretation of the reduced human consciousness states in terms of the proposed model of brain information processing, which could be helpful to be implemented in other states of consciousness. The manuscript is well-organized, and the results are important and could be interesting for a broad range of literature, suggesting interesting new ideas for the field to explore.

Comments on revised version:

The authors have addressed all my comments made in the previous revision.

Author Response

The following is the authors’ response to the original reviews.

eLife assessment

This article presents important results describing how the gathering, integration, and broadcasting of information in the brain changes when consciousness is lost either through anesthesia or injury. They provide convincing evidence to support their conclusions, although the paper relies on a single analysis tool (partial information decomposition) and could benefit from a clearer explication of its conceptual basis, methodology, and results. The work will be of interest to both neuroscientists and clinicians interested in fundamental and clinical aspects of consciousness.

Public Reviews:

Reviewer #1 (Public Review):

Summary:

In this paper, Luppi et al., apply the recently developed integrated information decomposition to the question how the architecture of information processing changes when consciousness is lost. They explore fMRI data from two different populations: healthy volunteers undergoing reversible anesthesia, as well as from patients who have long-term disorders of consciousness. They show that, in both populations, synergistic integration of information is disrupted in common ways. These results are interpreted in the context of the SAPHIRE model (recently proposed by this same group), that describes information processing in the brain as being composed of several distinct steps: 1) gatekeeping (where gateway regions introduce sensory information to the global synergistic workspace where 2) it is integrated or "processed" before 3) by broadcast back to to the brain.

I think that this paper is an excellent addition to the literature on information theory in neuroscience, and consciousness science specifically. The writing is clear, the figures are informative, and the authors do a good job of engaging with existing literature. While I do have some questions about the interpretations of the various information-theoretic measures, all in all, I think this is a significant piece of science that I am glad to see added to the literature.

One specific question I have is that I am still a little unsure about what "synergy" really is in this context. From the methods, it is defined as that part of the joint mutual information that is greater than the maximum marginal mutual information. While this is a perfectly fine mathematical measure, it is not clear to me what that means for a squishy organ like the brain. What should these results mean to a neuro-biologist or clinician?

Right now the discussion is very high level, equating synergy to "information processing" or "integrated information", but it might be helpful for readers not steeped in multivariate information theory to have some kind of toy model that gets worked out in detail. On page 15, the logical XOR is presented in the context of the single-target PID, but 1) the XOR is discrete, while the data analyzed here are continuous BOLD signals w/ Gaussian assumptions and 2) the XOR gate is a single-target system, while the power of the Phi-ID approach is the multi-target generality. Is there a Gaussian analog of the single-target XOR gate that could be presented? Or some multi-target, Gaussian toy model with enough synergy to be interesting? I think this would go a long way to making this work more accessible to the kind of interdisciplinary readership that this kind of article with inevitably attract.

We appreciate this observation. We now clarify that:

“redundancy between two units occurs when their future spontaneous evolution is predicted equally well by the past of either unit. Synergy instead occurs when considering the two units together increases the mutual information between the units’ past and their future – suggesting that the future of each is shaped by its interactions with the other. At the microscale (e.g., for spiking neurons) this phenomenon has been suggested as reflecting “information modification” 36,40,47. Synergy can also be viewed as reflecting the joint contribution of parts of the system to the whole, that is not driven by common input48.”

In the Methods, we have also added the following example to provide additional intuition about synergy in the case of continuous rather than discrete variables:

“As another example for the case of Gaussian variables (as employed here), consider a 2-node coupled autoregressive process with two parameters: a noise correlation c and a coupling parameter a. As c increases, the system is flooded by “common noise”, making the system increasingly redundant because the common noise “swamps” the signal of each node. As a increases, each node has a stronger influence both on the other and on the system as a whole, and we expect synergy to increase. Therefore, synergy reflects the joint contribution of parts of the system to the whole that is not driven by common noise. This has been demonstrated through computational modelling (Mediano et al 2019 Entropy).”

See below for the relevant parts of Figures 1 and 2 from Mediano et al (2019 Entropy), where Psi refers to the total synergy in the system.

Author response image 1.

Strengths

The authors have a very strong collection of datasets with which to explore their topic of interest. By comparing fMRI scans from patients with disorders of consciousness, healthy resting state, and various stages of propofol anesthesia, the authors have a very robust sample of the various ways consciousness can be perturbed, or lost. Consequently, it is difficult to imagine that the observed effects are merely a quirk of some biophysical effect of propofol specifically, or a particular consequence of long-term brain injury, but do in fact reflect some global property related to consciousness. The data and analyses themselves are well-described, have been previously validated, and are generally strong. I have no reason to doubt the technical validity of the presented results.

The discussion and interpretation of these results is also very nice, bringing together ideas from the two leading neurocognitive theories of consciousness (Global Workspace and Integrated Information Theory) in a way that feels natural. The SAPHIRE model seems plausible and amenable to future research. The authors discuss this in the paper, but I think that future work on less radical interventions (e.g. movie watching, cognitive tasks, etc) could be very helpful in refining the SAPHIRE approach.

Finally, the analogy between the PID terms and the information provided by each eye redundantly, uniquely, and synergistically is superb. I will definitely be referencing this intuition pump in future discussions of multivariate information sharing.

We are very grateful for these positive comments, and for the feedback on our eye metaphor.

Weaknesses

I have some concerns about the way "information processing" is used in this study. The data analyzed, fMRI BOLD data is extremely coarse, both in spatial and temporal terms. I am not sure I am convinced that this is the natural scale at which to talk about information "processing" or "integration" in the brain. In contrast to measures like sample entropy or Lempel-Ziv complexity (which just describe the statistics of BOLD activity), synergy and Phi are presented here as quasi-causal measures: as if they "cause" or "represent" phenomenological consciousness. While the theoretical arguments linking integration to consciousness are compelling, is this is right data set to explore them in? For example, the work by Newman, Beggs, and Sherril (nee Faber), synergy is associated with "computation" performed in individual neurons: the information about the future state of a target neuron that is only accessible when knowing both inputs (analogous to the synergy in computing the sum of two dice). Whether one thinks that this is a good approach neural computation or not, it fits within the commonly accepted causal model of neural spiking activity: neurons receive inputs from multiple upstream neurons, integrate those inputs and change their firing behavior accordingly.

In contrast, here, we are looking at BOLD data, which is a proxy measure for gross-scale regional neural activity, which itself is a coarse-graining of millions of individual neurons to a uni-dimensional spectrum that runs from "inactive to active." It feels as though a lot of inferences are being made from very coarse data.

We appreciate the opportunity to clarify this point. It is not our intention to claim that Phi-R and synergy, as measured at the level of regional BOLD signals, represent a direct cause of consciousness, or are identical to it. Rather, our work is intended to use these measures similarly to the use of sample entropy and LZC for BOLD signals: as theoretically grounded macroscale indicators, whose empirical relationship to consciousness may reveal the relevant underlying phenomena. In other words, while our results do show that BOLD-derived Phi-R tracks the loss and recovery of consciousness, we do not claim that they are the cause of it: only that an empirical relationship exists, which is in line with what we might expect on theoretical grounds. We have now clarified this in the Limitations section of our revised manuscript, as well as revising our language accordingly in the rest of the manuscript.

We also clarify that the meaning of “information processing” that we adopt pertains to “intrinsic” information that is present in the system’s spontaneous dynamics, rather than extrinsic information about a task:

“Information decomposition can be applied to neural data from different scales, from electrophysiology to functional MRI, with or without reference to behaviour 34. When behavioural data are taken into account, information decomposition can shed light on the processing of “extrinsic” information, understood as the translation of sensory signals into behavioural choices across neurons or regions 41,43,45,47. However, information decomposition can also be applied to investigate the “intrinsic” information that is present in the brain’s spontaneous dynamics in the absence of any tasks, in the same vein as resting-state “functional connectivity” and methods from statistical causal inference such as Granger causality 49. In this context, information processing should be understood in terms of the dynamics of information: where and how information is stored, transferred, and modified 34.”

References:

(1) Newman, E. L., Varley, T. F., Parakkattu, V. K., Sherrill, S. P. & Beggs, J. M. Revealing the Dynamics of Neural Information Processing with Multivariate Information Decomposition. Entropy 24, 930 (2022).

Reviewer #2 (Public Review):

The authors analysed functional MRI recordings of brain activity at rest, using state-of-the-art methods that reveal the diverse ways in which the information can be integrated in the brain. In this way, they found brain areas that act as (synergistic) gateways for the 'global workspace', where conscious access to information or cognition would occur, and brain areas that serve as (redundant) broadcasters from the global workspace to the rest of the brain. The results are compelling and consisting with the already assumed role of several networks and areas within the Global Neuronal Workspace framework. Thus, in a way, this work comes to stress the role of synergy and redundancy as complementary information processing modes, which fulfill different roles in the big context of information integration.

In addition, to prove that the identified high-order interactions are relevant to the phenomenon of consciousness, the same analysis was performed in subjects under anesthesia or with disorders of consciousness (DOC), showing that indeed the loss of consciousness is associated with a deficient integration of information within the gateway regions.

However, there is something confusing in the redundancy and synergy matrices shown in Figure 2. These are pair-wise matrices, where the PID was applied to identify high-order interactions between pairs of brain regions. I understand that synergy and redundancy are assessed in the way the brain areas integrate information in time, but it is still a little contradictory to speak about high-order in pairs of areas. When talking about a "synergistic core", one expects that all or most of the areas belonging to that core are simultaneously involved in some (synergistic) information processing, and I do not see this being assessed with the currently presented methodology. Similarly, if redundancy is assessed only in pairs of areas, it may be due to simple correlations between them, so it is not a high-order interaction. Perhaps it is a matter of language, or about the expectations that the word 'synergy' evokes, so a clarification about this issue is needed. Moreover, as the rest of the work is based on these 'pair-wise' redundancy and synergy matrices, it becomes a significative issue.

We are grateful for the opportunity to clarify this point. We should highlight that PhiID is in fact assessing four variables: the past of region X, the past of region B, the future of region X, and the future of region Y. Since X and Y each feature both in the past and in the future, we can re-conceptualise the PhiID outputs as reflecting the temporal evolution of how X and Y jointly convey information: the persistent redundancy that we consider corresponds to information that is always present in both X and Y; whereas the persistent synergy is information that X and Y always convey synergistically. In contrast, information transfer would correspond to the phenomenon whereby information was conveyed by one variable in the past, and by the other in the future (see Luppi et al., 2024 TICS; and Mediano et al., 2021 arXiv for more thorough discussions on this point). We have now added this clarification in our Introduction and Results, as well as adding the new Figure 2 to clarify the meaning of PhiID terms.

We would also like to clarify that all the edges that we identify as significantly changing are indeed simultaneously involved in the difference between consciousness and unconsciousness. This is because the Network-Based Statistic differs from other ways of identifying edges that are significantly different between two groups or conditions, because it does not consider edges in isolation, but only as part of a single connected component.

Reviewer #3 (Public Review):

The work proposes a model of neural information processing based on a 'synergistic global workspace,' which processes information in three principal steps: a gatekeeping step (information gathering), an information integration step, and finally, a broadcasting step. The authors determined the synergistic global workspace based on previous work and extended the role of its elements using 100 fMRI recordings of the resting state of healthy participants of the HCP. The authors then applied network analysis and two different measures of information integration to examine changes in reduced states of consciousness (such as anesthesia and after-coma disorders of consciousness). They provided an interpretation of the results in terms of the proposed model of brain information processing, which could be helpful to be implemented in other states of consciousness and related to perturbative approaches. Overall, I found the manuscript to be well-organized, and the results are interesting and could be informative for a broad range of literature, suggesting interesting new ideas for the field to explore. However, there are some points that the authors could clarify to strengthen the paper. Key points include:

(1) The work strongly relies on the identification of the regions belonging to the synergistic global workspace, which was primarily proposed and computed in a previous paper by the authors. It would be great if this computation could be included in a more explicit way in this manuscript to make it self-contained. Maybe include some table or figure being explicit in the Gradient of redundancy-to-synergy relative importance results and procedure.

We have now added the new Supplementary Figure 1 to clarify how the synergistic workspace is identified, as per Luppi et al (2022 Nature Neuroscience).

(2) It would be beneficial if the authors could provide further explanation regarding the differences in the procedure for selecting the workspace and its role within the proposed architecture. For instance, why does one case uses the strength of the nodes while the other case uses the participation coefficient? It would be interesting to explore what would happen if the workspace was defined directly using the participation coefficient instead of the strength. Additionally, what impact would it have on the procedure if a different selection of modules was used? For example, instead of using the RSN, other criteria, such as modularity algorithms, PCA, Hidden Markov Models, Variational Autoencoders, etc., could be considered. The main point of my question is that, probably, the RSN are quite redundant networks and other methods, as PCA generates independent networks. It would be helpful if the authors could offer some comments on their intuition regarding these points without necessarily requiring additional computations.

We appreciate the opportunity to clarify this point. Our rationale for the procedure used to identify the workspace is to find regions where synergy is especially prominent. This is due to the close mathematical relationship between synergistic information and integration of information (see also Luppi et al., 2024 TICS), which we view as the core function of the global workspace. This identification is based on the strength ranking, as per Luppi et al (2022 Nature Neuroscience), which demonstrated that regions where synergy predominates (i.e., our proposed workspace) are also involved with high-level cognitive functions and anatomically coincide with transmodal association cortices at the confluence of multiple information streams. This is what we should expect of a global workspace, which is why we use the strength of synergistic interactions to identify it, rather than the participation coefficient. Subsequently, to discern broadcasters from gateways within the synergistic workspace, we seek to encapsulate the meaning of a “broadcaster” in information terms. We argue that this corresponds with making the same information available to multiple modules. Sameness of information corresponds to redundancy, and multiplicity of modules can be reflected in the network-theoretic notion of participation coefficient. Thus, a broadcaster is a region in the synergistic workspace (i.e., a region with strong synergistic interactions) that in addition has a high participation coefficient for its redundant interactions.

Pertaining specifically to the use of resting-state networks as modules, indeed our own (Luppi et al., 2022 Nature Neuroscience) and others’ research has shown that each RSN entertains primarily redundant interactions among its constituent regions. This is not surprising, since RSNs are functionally defined: their constituent elements need to process the same information (e.g., pertaining to a visual task in case of the visual network). We used the RSNs as our definition of modules, because they are widely understood to reflect the intrinsic organisation of brain activity into functional units; for example, Smith et al., (2009 PNAS) and Cole et al (2014 Neuron) both showed that RSNs reflect task-related co-activation of regions, whether directly quantified from fMRI in individuals performing multiple tasks, or inferred from meta-analysis of the neuroimaging literature. This is the aspect of a “module” that matters from the global workspace perspective: modules are units with distinct function, and RSNs capture this well. This is therefore why we use the RSNs as modules when defining the participation coefficient: they provide an a-priori division into units with functionally distinct roles.

Nonetheless, we also note that RSN organisation is robustly recovered using many different methods, including seed-based correlation from specific regions-of-interest, or Independent Components Analysis, or community detection on the network of inter-regional correlations - demonstrating that they are not merely a function of the specific method used to identify them. In fact, we show significant correlation between participation coefficient defined in terms of RSNs, and in terms of modules identified in a purely data-driven manner from Louvain consensus clustering (Figure S4).

(3) The authors acknowledged the potential relevance of perturbative approaches in terms of PCI and quantification of consciousness. It would be valuable if the authors could also discuss perturbative approaches in relation to inducing transitions between brain states. In other words, since the authors investigate disorders of consciousness where interventions could provide insights into treatment, as suggested by computational and experimental works, it would be interesting to explore the relationship between the synergistic workspace and its modifications from this perspective as well.

We thank the Reviewer for bringing this up: we now cite several studies that in recent years have applied perturbative approaches to induce transitions between states of consciousness.

“The PCI is used as a means of assessing the brain’s current state, but stimulation protocols can also be adopted to directly induce transitions between states of consciousness. In rodents, carbachol administration to frontal cortex awakens rats from sevoflurane anaesthesia120, and optogenetic stimulation was used to identify a role of central thalamus neurons in controlling transitions between states of responsiveness121,122. Additionally, several studies in non-human primates have now shown that electrical stimulation of the central thalamus can reliably induce awakening from anaesthesia, accompanied by the reversal of electrophysiological and fMRI markers of anaesthesia 123–128. Finally, in human patients suffering from disorders of consciousness, stimulation of intra-laminar central thalamic nuclei was reported to induce behavioural improvement 129, and ultrasonic stimulation 130,131 and deep-brain stimulation are among potential therapies being considered for DOC patients 132,133. It will be of considerable interest to determine whether our corrected measure of integrated information and topography of the synergistic workspace also restored by these causal interventions.”

Recommendations for the authors:

Reviewer #1 (Recommendations For The Authors):

I would appreciate it if the authors could revisit the figures and make sure that:

(1) All fonts are large enough to be readable for people with visual impairments (for ex. the ranges on the colorbars in Fig. 2 are unreadably small).

Thank you: we have increased font sizes.

(2) The colormaps are scaled to show meaningful differences (Fig. 2A)

We have changed the color scale in Figure 2A and 2B.

Also, the authors may want to revisit the references section: some of the papers that were pre-prints at one point have now been published and should be updated.

Thank you: we have updated our references.

Minor comments:

• In Eqs. 2 and 3, the unique information term uses the bar notation ( | ) that is typically indicative of "conditioned on." Perhaps the authors could use a slash notation (e.g. Unq(X ; Z / Y)) to avoid this ambiguity? My understanding of the Unique information is that it is not necessarily "conditioned on", so much as it is "in the context of".

Indeed, the “|” sign of “conditioning” could be misleading; however, the “/” sign could also be misleading, if interpreted as division. Therefore, we have opted for the “\” sign of “set difference”, in Eq 2 and 3, which is conceptually more appropriate in this context.

• The font on the figures is a little bit small - for readers with poor eyes, it might be helpful to increase the wording size.

We have increased font sizes in the figures where relevant.

• I don't quite understand what is happening in Fig. 2A - perhaps it is a colormap issue, but it seems as though it's just a bit white square? It looks like redundancy is broadly correlated with FC (just based on the look of the adjacency matrices), but I have no real sense of what the synergistic matrix looks like, other than "flat."

We have now changed the color scale in Figure 2.

Reviewer #2 (Recommendations For The Authors):

Besides the issues mentioned in the Public review, I have the following suggestions to improve the manuscript:

• At the end of the introduction, a few lines could be added explaining why the study of DOC patients and subjects under anesthesia will be informative in the context of this work.

By comparing functional brain scans from transient anaesthetic-induced unconsciousness and from the persistent unconsciousness of DOC patients, which arises from brain injury, we can search for common brain changes associated with loss of consciousness – thereby disambiguating what is specific to loss of consciousness.

• On page and in general the first part of Results, it is not evident that you are working with functional connectivity. Many times the word 'connection' is used and sometimes I was wondering whether they were structural or functional. Please clarify. Also, the meaning of 'synergistic connection' or 'redundant connection' could be explained in lay terms.

Thank you for bringing this up. We have now replaced the word “connection” with “interaction” to disambiguate this issue, further adding “functional” where appropriate. We have also provided, in the Introduction, an intuitive explanation of what synergy and redundancy mean int he context of spontaneous fMRI signals.

• Figure 2 needs a lot of improvement. The matrix of synergistic interactions looks completely yellow-ish with some vague areas of white. So everything is above 2. What does it mean?? Pretty uninformative. The matrix of redundant connections looks a lot of black, with some red here and there. So everything is below 0.6. Also, what are the meaning and units of the colorbars?.

We agree: we have increased font sizes, added labels, and changed the color scale in Figure 2. We hope that the new version of Figure 2 will be clearer.

• Caption of Figure 2 mentions "... brain regions identified as belonging to the synergistic global workspace". I didn't get it clear how do you define these areas. Are they just the sum of gateways and broadcasters, or is there another criterion?

Regions belonging to the synergistic workspace are indeed the set comprising gateways and broadcasters; they are the regions that are synergy-dominated, as defined in Luppi et al., 2022 Nature Neuroscience. We have now clarified this in the figure caption.

• In the first lines of page 7, it is said that data from DOC and anesthesia was parcellated in 400 + 54 regions. However, it was said in a manner that made me think it was a different parcellation than the other data. Please make it clear that the parcellation is the same (if it is).

We have now clarified that the 400 cortical regions are from the Schaefer atlas, and 54 subcortical regions from the Tian atlas, as for the other analysis. The only other parcellation that we use is the Schaefer-232, for the robustness analysis. This is also reported in the Methods.

• Figure 3: the labels in the colorbars cannot be read, please make them bigger. Also, the colorbars and colorscales should be centered in white, to make it clear that red is positive and blue is negative. O at least maintain consistency across the panels (I can't tell because of the small numbers).

Thank you: we have increased font sizes, added labels, indicated that white refers to zero (so that red is always an increase, and blue is always a decrease), and changed the color scale in Figure 2.

• The legend of Figure 4 is written in a different style, interpreting the figure rather than describing it. Please describe the figure in the caption, in order to let the read know what they are looking at.

We have endeavoured to rewrite the legend of Figure 4 in a style that is more consistent with the other figures.

• In several parts the 'whole-minus-sum' phi measure is mentioned and it is said that it did not decrease during loss of consciousness. However, I did not see any figure about that nor any conspicuous reference to that in Results text. Where is it?

We apologise for the confusion: this is Figure S3A, in the Supplementary. We have now clarified this in the text.

Reviewer #3 (Recommendations For The Authors):

(1) In the same direction, regarding Fig. 2, in my opinion, it does not effectively aid in understanding the selection of regions as more synergistic or redundant. In panels A) and B), the color scales could be improved to better distinguish regions in the matrices (panel A) is saturated at the upper limit, while panel B) is saturated at the lower limit). Additionally, I suggest indicating in the panels what is being measured with the color scales.

Thank you: we have increased font sizes, added labels, and changed the color scale in Figure 2.

(2) When investigating the synergistic core of human consciousness and interpreting the results of changes in information integration measures in terms of the proposed framework, did the authors consider the synergistic workspace computed in HCP data? If the answer is positive, it would be helpful for the authors to be more explicit about it and elaborate on any differences that may be found, as well as the potential impact on interpretation.

This is correct: the synergistic workspace, including gateways and broadcasters, are identified from the Human Connectome Project dataset. We now clarify this in the manuscript.

Minors:

(1) I would suggest improving the readability of figures 2 and 3, considering font size (letters and numbers) and color bars (numbers and indicate what is measured with this scale). In Figure 1, the caption defines steps instead stages that are indicated in the figure.

Thank you: we have increased font sizes, added labels, and replaced steps with “stages” in Figure 1.