Author response:
The following is the authors’ response to the original reviews.
eLife Assessment
This Review Article provides a thorough overview of whole-brain activity changes induced by brain stimulation and summarizes the current state of the field. However, it lacks integration across spatial and mechanistic scales, which limits the reader's ability to understand how the different findings relate to one another. In addition, several key concepts are not explained in sufficient depth for non-expert readers. The manuscript would benefit from the development of a cohesive conceptual framework to more clearly synthesize the existing literature.
Thank you for the positive assessment. We fully agree, and as suggested we have added a new conclusion paragraph that outlines a synthesis of the paper and suggests a conceptual framework :
“In this paper, we have reviewed aspects of neuronal responsiveness, from the microscale level of neurons and circuits, the mesoscale level of single brain areas, and the macroscale level of the whole brain. At the microscale, it is apparent that the circuit operating in an asynchronous mode displays the highest responsiveness, as seen in brain slices (D’Andola et al., 2018). The underlying mechanism is that the high levels of synaptic « noise » in asynchronous states set neurons in a high responsive mode, as seen in models of single neurons (Ho & Destexhe, 2000). This higher responsiveness is confirmed at mesoscale, and can be seen for example with Utah-array recordings comparing wake and anesthesia (Dwarakanath et al., 2025). Similarly, propagating waves occur in the asynchronous state in awake monkey (Muller et al., 2014), and sensory inputs evoke more propagating patterns (and higher PCI) in wakefulness with asynchronous states compared to slow-wave states of anesthesia in mice (Montagni et al., 2024). At the whole-brain scale, experiments also find that evoked responses are more complex and propagating compared to slow-wave states (Massimini et al., 2005), a situation which models can reproduce (Goldman et al., 2023; Sacha et al., 2025). Other measures, such as fluidity (Breyton et al., 2024) and reversibility (Camassa et al 2024) also point to the same conclusion. Collectively, these results show that asynchronous and irregular activity states set neurons in a high responsive mode, which in turn impacts network behavior and favors the propagation of activity as mesoscale propagating waves, or macroscale activity patterns that propagate across brain regions. It is therefore not surprising that the best correlate of conscious states is the asynchronous activity (Koch et al., 2016).”
Public Reviews:
Reviewer #1 (Public review):
Summary:
This paper is a comprehensive review of perturbation studies and the state-dependence of the brain's response to perturbation at the circuit, mesoscale, and macroscale levels.
Strengths:
The strengths of the paper are the thorough description of many perturbation studies at different levels of organization, and the integration of both experimental and modeling studies. The review clearly communicates the need to consider (1) brain or local-population state, and (2) multiple levels of organization, in order to understand perturbation responses. Another major strength is the ability for the reader to reproduce figures using the EBRAINS platform.
Weaknesses:
Two major points of improvement should be resolved with the review, in order to make it useful for a broad audience.
The first is that the review does not include a significant integration across scales, and as a result, reads like three separate (though comprehensive) reviews. Currently, the only integration across the scales is in the brief conclusion paragraph. I would recommend adding an additional section, in which the overarching picture is discussed. (i.e. a unifying view of state dependence, and what is learned by considering across scales). This need not be too long, but it should be longer than a single conclusion paragraph.
Thank you for the positive assessment. We fully agree with the excellent suggestion of adding a concluding paragraph where the conceptual framework and overarching picture are presented. Please see the new conclusion paragraph that we added to the paper (copied above in the reply to Editors).
The second major weakness is that there is a lack of clarity on many points throughout, which is needed for the reader to fully understand the results described.
See our answer to the specific comments below for the list of unclear points.
Reviewer #2 (Public review):
Summary:
In this review article, the authors discuss the whole-brain activity changes induced by brain stimulation. They review the literature on how these activity changes depend on the cognitive state of the brain and divide the results by the scale of the change being induced, from microscale changes across small groups of neurons, up to macroscale changes across the entire brain. Finally, they describe attempts to model these changes using computational models.
Strengths:
The review provides an overview of the results within this subfield of neuroscience, and the authors are able to discuss a lot of prior results. The framing of the changes in neuronal activity in terms of computational changes is also a helpful approach.
Weaknesses:
However, the authors are not able to contextualize these results within a single framework, i.e. explaining from first principles how different aspects of stimulus-induced changes interact to generate functional changes in the brain, and how different changes - at distinct spatiotemporal scales - combine to form larger effects. This is a significant weakness in generating a review of the literature, since the authors do not provide a cohesive conceptual framework on which to frame the results. Similarly, the authors do not explain how their different computational models fit together, and how one can get a singular computational understanding of the distinct mechanisms of brain activity changes due to stimulation under different brain states, by combining the results derived from each separate model.
Thank you for the positive assessment. This is an excellent suggestion, actually also requested by Reviewer 1. We have now added a new conclusion paragraph where the conceptual framework is explained (copied above in the reply to Editors).
Major Comments:
(1) The authors have written this review as if it were intended for an audience who is already familiar with the topics. For example, they introduce concepts like complexity, spiral vs planar waves, without much explanation.
Thank you for this helpful comment. We agree that the Introduction should be more accessible to readers who are less familiar with these concepts, and we have therefore revised the text to provide a clearer definition of complexity and a more explicit explanation of propagating wave patterns.
Specifically, we now clarify that complexity can be understood as the richness of the set of accessible states of a system, which in our context can be related to the diversity of slowwave propagation modes and to the high-entropy, desynchronized activity of the awake brain. We also expanded the description of propagating slow waves to distinguish planar from spiral waves and to explain how their relative prevalence changes with anesthesia depth.
In the Introduction we replaced the sentence “New methods … at various scales” with “New methods for characterizing the complexity of network dynamics and their response patterns have emerged, particularly recently (Krohn et al., 2023; Wolf et al., 2018), and are presented here at various scales. Here, complexity is associated with the set of accessible states of a system (Parisi, 2006). In the present context, this notion can be linked to the diversity of slow-wave propagation modes and to the richness (i.e., the entropy) of perturbation-evoked responses in brain activity.”
While in Results (p. 13) the sentence “Spontaneous slow waves … administered (Huang et al., 2010).” has been expanded in “Spontaneous slow waves can also display propagating patterns, as shown in anesthetized mice (Huang et al., 2010; Mohajerani et al., 2010; Pazienti et al., 2022; Stroh et al., 2013). These patterns may take the form of planar waves, which travel across the cortex along a relatively regular front, or spiral waves, which rotate around a central core and therefore produce a more complex spatiotemporal organization. Under relatively deep anesthesia, spiral waves occur more frequently than planar waves, whereas the opposite imbalance is observed as anesthesia is lightened (Huang et al., 2010).”
(2) Regarding complexity, the authors present a quantification termed PCI. However, in the associated box, they state that PCI could be implemented in a number of different ways, using analogous metrics (which are, nonetheless, not identical). Yet the authors simply claim that all these metrics are sufficiently similar to be grouped together as "PCI". The authors do not provide much intuition about this, and they also don't present any other potential quantifications. This makes any interpretation of their results strongly dependent on your understanding of the concept of PCI. It would be helpful to present some other, analogous metric to demonstrate that the results that the authors are focusing on are not somehow tied to the specific computational structure of the PCI metric.
Thank you for pointing out to this inconsistency. We agree that the rationale for focusing on perturbational complexity was not sufficiently introduced in the original version of the manuscript.
Broadly speaking, complexity measures used in consciousness research can be divided into two major classes. The first includes observational measures, which are computed from spontaneous ongoing activity and quantify statistical dependencies within neural time series. The second includes perturbational measures, which quantify the deterministic causal interactions revealed by a controlled perturbation of the system and their spatiotemporal propagation across the network (see Sarasso et al., 2021).
The primary aim of the present Review was to discuss how complexity changes across spatial and temporal scales in response to perturbations. For this reason, we focused on perturbational complexity measures and, in particular, on the Perturbational Complexity Index (PCI), which remains one of the most widely adopted and validated approaches in this category.
As described in Box 2, different implementations of PCI have been proposed. The two most established versions are PCI based on Lempel–Ziv complexity (PCI^LZ) and PCI based on state transitions in principal component space (PCI^ST). Although these implementations differ algorithmically, they were developed to operationalize the same theoretical construct and have been shown to correlate strongly when applied to the same datasets (Comolatti et al., 2019). For this reason, throughout the Review we use the term “PCI” as an umbrella label encompassing these related perturbational complexity measures.
Importantly, all complexity measures discussed in the studies reviewed here belong to this broader class of perturbational approaches. While adaptations of the original algorithms are often required when dealing with different recording modalities and spatial scales, these modifications mainly concern preprocessing and signal representation rather than the underlying theoretical construct being quantified.
To clarify this point, we have revised the Introduction to explicitly motivate our focus on perturbational complexity, to distinguish perturbational from observational complexity measures, and to explain why different PCI implementations can be discussed within a common conceptual framework. We believe that these additions make the rationale of the Review substantially clearer and reduce the impression that the conclusions depend on a specific implementation of PCI.
(3) The authors divide the review into sections organized by the spatial extent of the effects that they are exploring (e.g. from microscale to macroscale). However, they don't bring together these insights into a cohesive structure - for example, by providing potential explanations of the macroscale effects by using the microscale changes.
We agree – and this is now the focus of the newly-added conceptual-framework conclusion paragraph.
(4) The authors completely ignore any aspect of cell-type specificity in their review, despite the known importance of specific cell types at the microcircuit scale. This makes it difficult to map their results onto the true biological system.
We agree that cell-type specificity could be made more explicit. The revised manuscript now clarifies that several models already include cell-type specificity. For example, the AdEx-based models distinguish excitatory regular-spiking or pyramidal populations with adaptation from inhibitory fast-spiking populations without adaptation. This differentiation is not made with other models like leaky or quadratic integrate-and-fire models. At the mesoscale, mean-field approaches can be derived for different structures, such as cortex, thalamus, hippocampus, striatum, or cerebellum, and can incorporate the experimentally experimentally observed firing properties of relevant cell classes.
(5) The authors introduce several different computational models, such as the Hopf model, the AdEx model, and the MPR model. However, they do not provide the reader with a conceptual understanding of the structure of each of these models (except through potentially more complex terminology, e.g. the Hopf model is a "phenomenological StuartLandau nonlinear oscillator"). Additionally, though they present the results of each simulation, they don't provide the reader with intuition about how these models compare against each other, and how best to interpret results derived from each model.
Very good question, and the answer is not easy. If the goal is to capture large-scale phenomena with models as simple as possible, then Stuart-Landau, Hopf, or Jahnsen-Rit may be appropriate. This approach is rather top-down. But if the goal is to assess how microscopic changes (synaptic receptors for example) affect large-scale brain activity, then we need a bottom-up approach, where mean-field models are derived. We can better explain this.
We agree that while the technical definitions of the whole-brain models (Hopf, AdEx, MPR) were provided, a clear conceptual framework comparing their underlying structures, specific trade-offs, and interpretation guidelines was missing. We have substantially revised the "Macroscale" section on Page 22 to provide immediate intuition regarding what each model represents structurally (e.g., macroscopic phenomenology vs. microscopic biological realism). We emphasized the structural assumptions of each of them, as well as the explicit utility in interpreting brain responsiveness. This ensures readers understand exactly why a researcher would choose one model over another depending on the mechanistic question at hand.
(6) In several cases, the authors make statements that they appear to believe to be completely straightforward (and require no justification), but that do not appear so to the reader. For example, they mention: "In wakefulness and REM sleep, ..., the membrane potential is depolarized and close to the spike threshold, which explains why neurons respond more reliably and with less response variability compared with slow-wave sleep". However, this statement is not obvious to the reader and requires explanation (for example, in a system that is close to balance, bringing cells closer to the firing threshold can result in increased response jitter).
We agree that the original statement was an over-simplification of a complex situation. We have revised it to avoid suggesting that depolarization alone monotonically increases reliability. The relevant mechanism is the combination of depolarization, desynchronized high-conductance synaptic input, balanced fluctuations, and reduced tendency to enter long silent Down states. In this regime, weak inputs are more likely to be converted into spikes and propagate through the network. However, too high conductance, excessive noise, can shunt inputs, enhance jitter, or saturate the network. We are now more explanatory.
Recommendations for the authors:
Reviewer #1 (Recommendations for the authors):
As stated in the public review, there is a lack of clarity on many points throughout, which is needed for the reader to fully understand the results described.
Points needing clarification:
(1) sPCI (slice Perturbational complexity index) - is this different from other PCIs in box 2? Regardless, the metric and its interpretation should be briefly explained in the main text.
We use sPCI to refer to the PCI measure adapted for application to cortical brain slices (D’Andola et al., 2017; see Box 2). It relies on the same core algorithm as PCI, namely the Lempel–Ziv complexity of the spatiotemporal pattern of significant responses (Casali et al., 2013) but differs in the preprocessing steps required for slice recordings. We now added this clarification to the text.
(2) Page 9 "by decreasing fast inhibition but also enhancing it" - What does that mean? More info about the model is needed.
Thanks for raising this point, since this sentence was indeed confusing. We have revised it now.
(3) Page 9 "balance between segregation and integration, a crucial ingredient on which sPCI relies" - How is this balance seen in the figure? All I see is sPCI and blockage of GABA.
The comment is correct, and this mention of segregation and integration has now been eliminated.
(4) Figure 3D, Page 11 "two different desynchronized (AI) states in a network of AdEx neurons"- What are the two different states? Why is the response different?
The different AI states correspond to different synaptic strength parameters, we added this precision in the text.
(5) Figure 3B - "Bifurcation diagram showing the different activity regimes displayed by spiking neuron network." Which model? Multiple are cited. This is a general issue throughout where multiple models are mentioned in the text, and it's unclear which is shown in the figure.
We agree with the Reviewer's helpful remark. We have revised the manuscript to explicitly state the types of models depicted in the different panels of Figure 3. Corresponding details have also been incorporated into the relevant text in the Results section (previously pages 10–12)."
A few editorial issues:
(1) The text in many of the figure panels was too small to read. This is a significant issue that must be addressed.
We will fix this at the next round, can you please let us know which figures are not visible?
(2) I recommend reading through for writing flow. E.g. In the first paragraph of the introduction, there are two sentences that start with "importantly, ..." in a row.
Thanks for noting this – this is now fixed.
(3) Figure 1E - How does the color on the left relate to the right? What is the y-axis?
The colour code corresponds to the latency of activation (light blue, 0 ms; red, 300 ms). The Y-axes is the global mean field power (voltage). It has now been included in the figure caption.
(4) Figure 1C - What is the stimulus?
The triangle corresponds to the electrical stimulation of the homotopic area 18 of the contralateral hemisphere. This information is now included in the figure legend.
