Author response:
The following is the authors’ response to the original reviews.
Public Reviews:
Reviewer #1 (Public review):
Summary:
This is a study utilizing several types of analyses (computational modeling, neuronal cultures, rodent epilepsy model, and human intracranial multi-scale recordings) to address a highly relevant conceptual question: Are fast ripples (FRs) distinct pathological entities or largely emergent products of stochastic spike clustering? The results can potentially reshape current approaches to incorporating fast ripples into the epilepsy surgery evaluation.
Strengths:
The conceptualization of fast ripples as potentially arising by chance is highly novel and builds effectively on questions raised in prior studies that have never been satisfactorily resolved.
The integration across biological scales and models is a major strength. The state dependency analysis provides additional, strong support. The methodology and statistical approaches used are thoughtfully presented and rigorously applied.
In particular, this paper provides a strong response to the findings from Gliske et al, Nat Commun 2018. This study utilized long-term data analysis to uncover low rates of FRs detected from most recording sites, suggesting spurious detections, although FRs were concentrated within seizure onset areas.
We fully agree with this comparison. Although we had already cited this paper, we now further emphasize this observation in the Discussion:
“Furthermore, the variability of FRs across time (Gliske et al., 2018) indicates that longer nocturnal recordings in humans are necessary. It also suggests that changes in excitability across time could explain this change in FR incidence.”
Weaknesses:
The authors clearly aimed to use a statistical rather than a mechanism-based approach in this work. However, the paper's framing of true fast ripples as oscillatory events with stochastic fast ripples considered as confounders does not take prior investigations into biological mechanisms, particularly prior studies that point to an important role for stochastic fast ripples in some contexts. Incorporating recognition of these mechanisms would strengthen the manuscript and provide a more complete and nuanced characterization.
Some examples from the literature:
Eissa et al, eNeuro 2016, a paper that closely parallels this manuscript but took a mechanistic rather than statistical approach, showed that fast ripples can arise from population paroxysmal depolarizations - a key feature of epileptiform discharges - as temporally clustered, jittered population firing, with FRs appearing in LFP or EEG due to summated postsynaptic potentials (which are slower than action potentials and can generate signals in the high gamma range).
Foffani et al., 2007, Neuron, and Ibarz et al., 2010, J Neurosci, argue that FRs are pseudo-oscillations created by jittered neuronal populations in the setting of altered spike timing.
Smith et al., 2020, Sci Rep, contrasts FR characteristics in different regimes, i.e., intact inhibition early in a seizure vs. implied collapse of inhibition after recruitment. Schlingloff et al., 2025, J Neurosci, reported analogous findings in an animal model.
We agree with the reviewer that even stochastic events may be of biological importance and an increase in stochastic events will occur when there is an increase in synchronisation and excitability, two properties of pathological cortex. We also don’t disagree that FRs can occur as distinct entities, although our work indicates that most are due to chance.
To address this point, we have clarified our claims in the Abstract:
“This work does not rule out FRs as potential indicators of epileptogenic tissue, but it does challenge prevailing assumptions about their generation and specificity. Their higher prevalence in epileptogenic tissue is likely primarily due to increased excitation and/or neural synchronization, rather than peculiar abnormalities in network behavior.”
In addition, we expand on these points at various junctures in the Discussion. In particular, we reiterate our assertion that FRs may still be a useful biomarker, but that their interpretation should be moderated to reflect the fact that they often occur by chance:
“Importantly, we do not question the potential of FRs to delineate the seizure-onset zone. Instead, our results suggest that the observed increase in FRs within the epileptogenic zone is an emergent phenomenon – arising due to changes in secondary network properties such as excitability and synchronization, not as a direct result of some pathology that is specific to epilepsy. In addition, we show that long-durations FRs are more likely to be distinct oscillations than stochastic events; and so FR duration is a key parameter that should be considered in future studies.”
The computational model and subtraction approach provide a strong case for the random emergence of clustered activity in the high gamma band, given its assumptions. However, any such modeling effort needs to account for inhibitory activity, including impaired inhibitory function that is expected in epileptic brain regions, which has a strong modulating effect on excitatory firing and is thought to play a significant role in FR generation.
We appreciate the reviewer’s concerns, but we believe that the impact of inhibitory interneuron activity on excitatory firing rates and synchronisation is incorporated indirectly into our simulations, while keeping our model as parsimonious as possible by not directly incorporating interneuron activity into our simulations. We have addressed this point in the Methods section:
“Varying synchrony allowed us to test the impact, on the network, of inhibitory cells, which have been shown to favour synchrony (Bocchio et al., 2024; Cobb et al., 1995).”
The shuffling procedure aims to preserve the power spectrum but randomizes high frequency phase (>200 Hz). However, this procedure removes biologically meaningful spike timing correlations, as well as structured cross-frequency coupling. The subtraction method thus likely underestimates the incidence of structured "distinct" FRs, while perhaps overestimating "chance" FRs due to biologically infeasible activity, making the statement that most FRs are due to chance correlation too strong.
We appreciate this concern, which is especially important given that our results depend crucially on the validity of our shuffling procedure (as described in the Discussion). To address this issue, we have implemented an additional shuffling algorithm that preserves cross-frequency coupling (see last section of the Results, especially Supplementary Fig. 10f). This method showed no qualitative difference, compared with other alternative methods presented in Supplementary Fig. 10. These new results are described in the Methods section:
“Last, we also implemented a method based on wavelet-IAAFT with preservation of cross-frequency coupling, since fast ripples are typically locked to low-frequency phase (Sheybani et al., 2019). The code detects the highest phase-amplitude coupling (PAC) in the original signal between [300-6000 Hz] for amplitude and several low-frequency bands ranging from 2-20 Hz, bandwidth of 3 Hz. PAC is computed using the modulation index (Tort et al., 2008). Then, in the shuffled signal under construction and during convergence testing of PSD (see above), the PAC between high-frequency part of the signal (300-6000 Hz) and the identified low frequency for phase is normalized to that of the highest PAC identified earlier.”
The kainate findings underscore this point: the increase in the number of FR detections could be, as the authors state, an increase in chance clustering due to increased network excitability generally. However, the likelihood of a parallel increase in pathological FRs cannot be ruled out, given likely pro-epileptic alterations in spike timing and circuit function.
We appreciate the reviewer’s point but wish to re-emphasise our interpretation of these findings – that the observed increase in the incidence of FRs occurs as a result of increased network excitability/synchrony, secondary to the pathological mechanisms of epilepsy. We have updated the Discussion accordingly:
“Importantly, we do not question the potential of FRs to delineate the seizure-onset zone. Instead, our results suggest that the observed increase in FRs within the epileptogenic zone is an emergent phenomenon – arising due to changes in secondary network properties such as excitability and synchronization, not as a direct result of some pathology that is specific to epilepsy. In addition, we show that long-duration FRs are more likely to be distinct oscillations than stochastic events; and so FR duration is a key parameter that should be considered in future studies.”
To further emphasise this important point, we have also updated the Abstract:
“This work does not rule out FRs as potential indicators of epileptogenic tissue, but it does challenge prevailing assumptions about their generation and specificity. Their higher prevalence in epileptogenic tissue is likely primarily due to increased excitation and/or neural synchronization, rather than peculiar abnormalities in network behavior.”
Reviewer #2 (Public review):
Summary:
This paper asks an important question that has not been discussed much in the extensive literature on the High Frequency Oscillations (HFOs) that have been extensively studied in patients with epilepsy and experimental models of epilepsy. The question is whether the Fast Ripples (FRs), the HFOs in the 250-500 Hz frequency band, represent a pathological phenomenon or represent a physiological phenomenon that occurs in the healthy brain but happens to be more frequent in epileptic tissue. It is an important question that has not been systematically addressed until now. The authors conclude, from very extensive simulations, from extensive experimental animal studies (the systemic kianate model of epilepsy in rats), and from a modest amount of human data, that FRs occur in healthy brains as a result of the chance occurrence of bursts of action potentials, and that in epileptic tissue, their frequency of occurrence is approximately 30% higher than what is expected by chance. They conclude that FRs are not a separate phenomenon of epileptic tissue. This finding is reinforced by the recent findings of FRs in experimental models of Alzheimer's disease.
Strengths:
This is a valuable study because it asks an important and original question and because it evaluates it from several angles (simulation, tissue culture, experimental animals, and human patients). The simulations and the analyses of real data are performed very carefully and with original and solidly documented approaches, using extensive simulations and extensive data sets in the cultured cell data and in the in vivo experiments. The paper is clearly written and well-illustrated.
Weaknesses:
I found only one serious weakness in this study, but it is one that is of importance. Although the original work on FRs was done in an experimental model of epilepsy, the field really became prominent when ripples and fast ripples were found first in microelectrode recordings of epileptic patients and then in the intracerebral EEG of such patients. Numerous studies have been performed since then, with a valuable meta-analysis including 700 patients (Wang Z, Guo J, van 't Klooster M, Hoogteijling S, Jacobs J, Zijlmans M. Prognostic Value of Complete Resection of the High-Frequency Oscillation Area in Intracranial EEG: A Systematic Review and Meta-Analysis. Neurology. 2024 May 14;102(9). Although the consensus at this point is that FRs are not the ideal and totally specific marker of epileptic tissue that many thought it could be, FRs are nevertheless much more frequent in epileptic tissue than in non-epileptic tissue and are a solid biomarker.
We agree with the reviewer, and do not intend to challenge the role of FRs as a marker of the seizure-onset zone, and potentially the epileptogenic zone. Instead, the aim of this study was to address the question of whether FRs are generated by intrinsic pathological mechanisms, or whether they arise due to the chance co-occurrence of action potentials that follow different dynamics in epileptogenic parenchyma. We have updated the Discussion accordingly:
“Importantly, we do not question the potential of FRs to delineate the seizure-onset zone. Instead, our results suggest that the observed increase in FRs within the epileptogenic zone is an emergent phenomenon – arising due to changes in secondary network properties such as excitability and synchronization, not as a direct result of some pathology that is specific to epilepsy. In addition, we show that long-durations FRs are more likely to be distinct oscillations than stochastic events; and so FR duration is a key parameter that should be considered in future studies.”
To further emphasise this important point, we have also updated the Abstract:
“This work does not rule out FRs as potential indicators of epileptogenic tissue, but it does challenge prevailing assumptions about their generation and specificity. Their higher prevalence in epileptogenic tissue is likely primarily due to increased excitation and/or neural synchronization, rather than peculiar abnormalities in network behavior.”
It is also well established that they are much more frequent in NREM sleep than in wakefulness, as reported in the original paper of Staba et al (Staba RJ, Wilson CL, Bragin A, Jhung D, Fried I, Engel J Jr. High-frequency oscillations recorded in human medial temporal lobe during sleep. Ann Neurol. 2004 Jul;56(1):108-15., not mentioned in this paper) and in the study of Bagshaw et al (2009). In this last paper, using SEEG in various brain regions, the average rate of FRs in NREM sleep is about 6 times that in wakefulness. In the paper by Staba, with microelectrodes in mesial temporal structures, it is about twice. As a separate issue, the paper of Fraucher et al (Frauscher B, von Ellenrieder N, Zelmann R, Rogers C, Nguyen DK, Kahane P, Dubeau F, Gotman J. High-Frequency Oscillations in the Normal Human Brain. Ann Neurol. 2018 Sep;84(3):374-385), which is not quoted, found that, in an extensive sample, non-epileptic human tissue sampled with SEEG generated extremely rare FRs (an average rate of 0.04/min/channel, i.e. 1 every 25 min).
The results above are mentioned because they do not fit with the data provided in the present study: FRs are much more frequent in NREM sleep than in wakefulness in human epileptic patients, and they are much more frequent (not 30% more, but many hundreds of percent more) in epileptic tissue than in non-epileptic human tissue. The fundamental phenomenon of interest is, I believe, the FRs in epileptic patients. The animal experiments, tissue studies, and simulations are models to study the human phenomenon. With respect to the modulation by sleep and the differentiation between epileptic and non-epileptic tissue, it seems that the systems studied in this paper are not good models of the human condition. The human results presented in the study only reflect wakefulness recordings, which is not the condition in which most HFO studies have been done and in which most HFOs occur. The authors refer to the study of long-term fluctuations in HFO rates by Gliske et al. (2018) to say that one has to be careful with the results regarding sleep, for example, Bagshaw et al (2009), but the clear predominance in of HFOs in NREM sleep has been observed by many studies. The cautions regarding fluctuations over extended periods also apply to the awake human data analyzed in this study. The study's conclusions regarding the generation of FRs are therefore questionably applicable to the human condition. I do not dispute their validity for the models and situations in which they were studied.
We looked at this in more detail. Our simulations were intended to test how the incidence of FRs can vary with different parameters of network activity (neuronal count, firing rate, synchronization). Indeed, since their incidence is known to vary across regions and within regions and across states, we wanted to test how FRs are controlled by different factors. As such, we do not wish to draw firm conclusions about the observed sleep-wake changes in FR incidence in rodents, and how it relates to humans – evidence shows that pathological FRs in rodents do not display state-specific preferential occurrence (Ewell et al., 2019). We have added new text to the Abstract and Discussion to emphasize this.
Abstract:
“Our simulations showed that chance aggregation can generate fast-ripples and that their incidence changes depending on brain state, an observation that we confirmed in our rodent data.”
We acknowledge that previous publications have reported higher rates during sleep, although with shorter recordings than in our rodent recordings (Staba, 2004: one night; Bagshaw, 2009: 10 min; Frauscher, 2018: 20 min – only sleep recordings). We have rewritten the part of the Discussion on the effect of the sleep-wake cycle on FRs incidence:
“In our rodent data, we were initially surprised to find a higher rate of FRs during wakefulness, which contrasts with previous reports in humans (Bagshaw et al., 2009; Staba et al., 2004). However, previous studies only indicate that physiological vs pathological FRs are more easily distinguished during NREM sleep (von Ellenrieder et al., 2016) and that their incidence varies during sleep (Von Ellenrieder et al., 2017), but in hours-long recordings, no differences in incidence have been reported in the mesial temporal lobe (Dümpelmann et al., 2015). Furthermore, the variability of FRs across time (Gliske et al., 2018) indicates that longer nocturnal recordings in humans are necessary. It also suggests that changes in excitability across time could explain this change in FR incidence. Last, but not least, another report did not find a state-dependent expression of FRs in the kainate rat model of temporal lobe epilepsy (Ewell et al., 2019), thus indicating that the variability of FRs across sleep and wake is still an open question, at least in rodents. Hence, the main conclusion on the effect of sleep-wake transitions is that these transitions impact the likelihood of stochastic events, more than dictating the direction (increases vs decreases) of change. It also highlights that the specificity of FRs to epileptogenic parenchyma could vary across the sleep-wake cycle, which would be crucial in epileptology (Dimakopoulos et al., 2024; Roehri et al., 2018; Sheybani et al., 2019, 2018; Zijlmans et al., 2012, 2009). Hence, FRs reflect and are highly susceptible to changes in network excitability.”
Reviewer #3 (Public review):
Summary:
An outstanding question in the field of high-frequency oscillations (HFOs) in the context of epilepsy is how these oscillations emerge, considering that they occur at such high frequencies, i.e., 250Hz, well above the firing ability of single neurons. One hypothesis that has been suggested in the past is that neurons that fire in an out-of-phase fashion, or rather at random intervals, may contribute to a spectrum of HFOs ranging from 250-500Hz that are observed in epilepsy. However, how possible it is that random action potentials could aggregate to the extent that they could give rise to HFOs in the so-called fast ripple (FRs) frequency range (>200 according to the authors) remains unclear. To test this hypothesis, they used computational modeling to randomly insert action potentials in a signal, and they found that this approach is sufficient to generate FRs. Some of the predictors of whether FRs could occur were neuronal count, firing rate, and synchronization. Besides computational modeling, they used different model systems to test whether that would be possible to be observed in neuronal cultures, in epileptic rats (intrahippocampal kainic acid model), and human data. Neuronal cultures treated with picrotoxin did not show evidence that FRs could be generated beyond chance aggregation of action potentials. They then asked whether synchronization and firing rate could play a role in the emergence of FRs. They found that changes in neural firing and synchronization, such as those occurring during differences phase of the sleep-wake cycle, could affect the number of FRs occurring by chance aggregation, with more FRs seen during periods of wakefulness, a result that they replicated in human data.
The authors largely achieve their proposed aims of demonstrating that random neuronal firing can, in principle, generate FRs. Results from this study could influence current thinking around mechanisms generating FRs in epilepsy. The use of different computational approaches and model systems could offer new analytical methodologies for the study of FRs in the context of brain disease.
Strengths:
(1) The authors used a multi-level approach combining computational modeling with experimental datasets, including neuronal cultures, a rat model of temporal lobe epilepsy, and human data.
(2) Identification of key parameters such as neuronal count, firing rate, synchronization, and brain state in observed incidence of FRs generated through random aggregation of neural firing.
(3) Cross-species validation increases the likelihood of generalizability of the findings.
Weaknesses:
(1) Some of the simulated FRs appear short in duration and may not meet standard detection and definition criteria, potentially influencing validity.
We thank the reviewer for raising this important concern. To address this issue, we quantified and compared the duration of FRs in original and shuffled rodent data. Consistent with the reviewer’s suspicions, we found that FRs in shuffled signals are shorter than FRs in original signals. This is important because it shows that: (i) a longer duration should be considered a core feature of genuine FRs; and (ii) depending on the basal duration of FRs, the shuffling procedure will lead to different ratios of genuine to stochastic FRs. We have updated the Results accordingly:
“These findings demonstrate the challenge of identifying distinct FRs within a composite population of distinct and stochastic events. One parameter that could help disentangle these events is their duration. Indeed, one might expect stochastic events to be more likely to be short-lived, since the probability of consecutive APs continuing to co-occur across neurons decreases over time. Hence, we next compared the distribution of FR durations between original and shuffled rodent data and found that FRs in shuffled data are shorter than those in original data (Supplementary Fig. 9). This makes duration a key feature that could help identify distinctly generated FRs.”
And Discussion accordingly:
“In addition, we show that long durations FRs are more likely to be distinct oscillations than stochastic events; and so FR duration is a key parameter that should be considered in future studies.”
(2) The neuronal culture approach does not directly test random insertion of action potentials, limiting interpretation.
Neither the neuronal culture approach, the rat data or the human data directly test random insertion of action potentials. The insertion of random action potentials is only performed in the simulated data to test if FRs can arise from the chance insertion of action potentials. Once this was confirmed in the simulations, we then used the shuffling procedure in biological data to test if FRs are more frequent than expected by chance.
(3) Sleep is treated as a homogeneous state in the rat dataset, without accounting for stage-specific differences in synchronization, which may affect the results and interpretation.
We agree with the reviewer, but our primary aim was to answer the question of whether FRs can arise by chance. Although it was interesting to see that, in our longitudinal rodent data, the incidence of FRs varies across the sleep-wake cycle, any sleep-stage-specific changes are beyond the scope of this work.
(4) The analyses conducted in human data lack direct comparison with sleep data.
We agree that it would have been useful to investigate variations in the incidence of FRs across the sleep-wake cycle in human microelectrode recordings. Unfortunately, however, such sleep recordings were not available. Hence, while we cannot compare variations in FR incidence across brain states between humans and animal models, our conclusions that FRs arise mostly by the chance co-occurrence of action potentials still holds.
Recommendations for the authors:
Reviewer #1 (Recommendations for the authors):
(1) Please indicate where corrections for multiple comparisons were used.
P-values corrected for multiple comparisons are indicated by the accompanying phrase: “adjusted p-value”. We had previously omitted to mention this once in the Results, which we have now corrected.
(2) Delta amplitude is likely sufficient for detecting sleep-wake transitions, but the beta/delta ratio is better supported in the literature. Do the results change if beta activity is incorporated?
We have now computed the beta (15-40 Hz) to delta (0.5-4 Hz) ratio and find that this is closely correlated with delta across time. We have updated the Results accordingly:
“Importantly, these findings were robust to the specific method used to detect FRs (Supplementary Fig. 4c) (Padmasola et al., 2024; Sheybani et al., 2019, 2018). Also our use of delta power to identify periods of presumed wakefulness and sleep was highly (negatively) correlated with an alternative method of using the beta-to-delta ratio across time (another marker of increased vigilance; (Fraigne et al., 2023), see Supplementary Fig. 4d).”
Methods:
“We further verified that delta power across time displayed similar fluctuations to beta (15-40 Hz) to delta power ratio, another marker of vigilance (Fraigne et al., 2023).”
And we updated Supplementary Fig. 4d
“(d) Beta to delta power ratio across time is superimposed over delta power across time. There is a strong (inverse) correlation between the two time-series (inset), which is confirmed by the correlation coefficient across animals (right).”
(3) Figure 2's axis labeling with the 3D plots is hard to read.
We have enlarged the font size.
(4) The scaling of the histogram in Figure 3 is unclear.
This was on omission. The scale has now been added to Figure 3.
(5) There is a risk of overfitting in the regression model. Was cross-validation used?
We have now repeated this analysis with cross-validation, without any qualitative impact on the results (e.g. the model still performs well above chance). We have updated the Methods:
“To further confirm the performance of GBT, we used a cross-validation procedure where the GBT is trained on 80% of data and then tested on the 20% remaining. The procedure is repeated 1000 times and the r2 is saved at each round. We repeated the analysis with randomization of the outputs across 1000 rounds and saved this null distribution r2. We then compared the performance against original data.”
Legend of Fig. 3:
“(e) Performance of the GBT classifier using cross-validation (training: 80% of data; test: 20% remaining) using original (orange) and shuffled (blue) data. The difference is significant (paired t-test, p<0.0001).”
And Results:
“Furthermore, using a cross-validation approach with 80% of the data as training set and the remaining 20% as the test set, we obtained a significantly higher explained variance than when outputs were shuffled across the 125,000 solution points (paired t-test, p<0.0001, Fig. 3e), […]”
Reviewer #2 (Recommendations for the authors):
Maybe I missed it, but I did not find the length of human data analyzed or how the sections were selected.
Apologies for this omission. The methods have been updated accordingly:
“Microwire signals were selected based on high signal-to-noise ratio, as reflected by the detection of ≥ 1 single unit. Duration of recordings was of (median, interquartile range) 10 min and 17 s [3-13 min] and number of electrodes per patient was 4.5 [2.75-8].”
The authors use the term "virtual simulation", which I find odd. I think the simulation is very real in the sense that it simulates reality, and I do not understand how a simulation can be virtual.
We have updated the manuscript accordingly.
Reviewer #3 (Recommendations for the authors):
Major Comments:
(1) In Figure 1, the authors suggest that random insertion of action potentials in a signal is sufficient to yield FRs. However, the observed FRs shown in panel 1b (also in supplemental Figure 5) seem pretty short in duration and may not meet the mentioned criteria in methods that require at least 4 cycles and ".whose amplitude is 3 times that of the surrounding baseline..". Moreover, in panel 1b, it seems that the FR shows a candle-like appearance, which has often been associated with filtering of sharp transients. How did the authors validate that the detected FRs were "real" FRs?
Given the very large amount of data, it was not possible to visually verify all FRs. However, FRs were detected with published methods (Roehri et al., 2016; Roehri et al., 2017; and Sheybani et al., 2018 for confirmation of 24-hour variability in rodents) that have subsequently been used in several publications.
Regarding the candle-like appearance of the spectrogram, the Delphos algorithm precisely looks for isolated “islands” of increased power (see Roehri et al., 2018, Ann Neurol), thus excluding any candle-like appearance. Similarly, the detector in Sheybani et al. (2018) J Neurosci first detects candidate FRs but then excludes those that are associated with a peak in lower frequencies, thus also limiting the risk of detecting candle-like events.
Regarding duration, we have compared the duration of FRs in original and shuffled rodent data and found that FRs in original signals are indeed longer. This makes duration a key feature to identify distinct FRs. We have updated the Results accordingly:
“These findings demonstrate the challenge of identifying distinct FRs within a composite population of distinct and stochastic events. One parameter that could help disentangle these events is their duration. Indeed, one might expect stochastic events to be more likely to be short-lived, since the probability of consecutive APs continuing to co-occur across neurons decreases over time. Hence, we next compared the distribution of FR durations between original and shuffled rodent data and found that FRs in shuffled data are shorter than those in original data (Supplementary Fig. 9). This makes duration a key feature that could help identify distinctly generated FRs.”
(2) In the context of neuronal cultures, it is unclear how it could be deducted that the result relates to chance incidence of action potentials considering that no random action potentials were inserted, but only random shuffling of the high frequency component of the signal was attempted "Hence, neural networks with limited complexity (Kim et al., 2020; Saglam-Metiner et al., 2024; Sanchez-Vives and McCormick, 2000; Timofeev and Chauvette) fail to generate FRs beyond that expected from the chance coincidence of APs, even after increasing network excitability."
FRs arise from series of action potentials occurring at a delay corresponding to their oscillatory frequency (250-500 Hz). Simulations demonstrated that FRs can occur by chance. When the EEG is shuffled, the only FRs that remain are those occurring by chance, because those occurring as individual entities have been broken up. Hence, if the original EEG displays more FRs than the shuffled EEG, then it means that these additional FRs were generated as individual entities. We have improved the Results section to clarify this:
“We hypothesized that if FRs arise purely from chance firing, then temporally shuffling these recordings while conserving their spectral properties (Supplementary Fig. 3) would disrupt any oscillatory structure, leaving only FRs that occur due to chance.] Any additional FRs in the original data, compared to the number of FRs in the shuffled EEG, should thus be assumed to be individual entities.”
(3) In the rat dataset, sleep was treated rather homogenously, without accounting for the sleep stage that is characterized by different synchronization and firing. An analysis of different sleep stages would be valuable.
Although we agree that it would be scientifically interesting, we believe that our claim – that the ratio of genuine to stochastic FRs changes across the sleep-wake cycle – would hold. Unfortunately, lack of EMG prevents us from performing reliable sleep scoring. However, we do now include an alternative method for differentiating sleep from wake using the beta-to-delta ratio, which was highly correlated with delta activity, supporting our previous approach. Please refer to Supplementary Fig. 4d for further information.
(4) The authors found that chance aggregation was highest during periods of wakefulness. Analyses of human data also confirmed that FRs could occur by chance aggregation during wakefulness. However, a comparison with sleep data would further strengthen this finding.
We fully agree, but unfortunately, we do not have sleep data using microwires. Although our central claim – that FRs can occur by chance clustering of action potentials – would hold, we agree that it would have been scientifically interesting to add sleep data.
(5) The statistics section would benefit from addressing how normality was determined and power analysis, as well as the inclusion of the exact sample size for all experiments.
With large sample sizes, ANOVA and linear mixed models are robust to non-normality. Given the large sample sizes of our data, we thus used ANOVA and linear mixed model. For tests with small sample sizes where normality was violated, we used non-parametric tests, indicated by their name, e.g., Wilcoxon test for Supplementary Fig. 3b.
(6) Greater discussion on the implications of this study for proposed in-phase or out-of-phase FR generation mechanisms is suggested.
We have added further discussion on this. In the aim to keep the Discussion short and impactful, we could not elaborate too much. We have synthetized other parts of the Discussion to keep it within the right length. Here is the additional part:
“It has been argued that the very high frequency that can be obtained during FRs are due to out-of-phase firing of excitatory neurons (Foffani et al., 2007; Ibarz et al., 2010), which is also consistent with our concept of stochastic firing. The conceptual difference is the degree to which there is any underlying organization of this firing. We argue that in the majority of cases there is no organization, although a substantial minority cannot be explained on a stochastic basis.”
(7) More explanation around why wakefulness may drive chance aggregation and the clinical relevance of it, as often presurgical epilepsy recordings are being evaluated during sleep.
We have profoundly rewritten the Discussion regarding the effect of the sleep-wake cycle on FRs incidence:
“In our rodent data, we were initially surprised to find a higher rate of FRs during wakefulness, which contrasts with previous reports in humans (Bagshaw et al., 2009; Staba et al., 2004). However, previous studies only indicate that physiological vs pathological FRs are more easily distinguished during NREM sleep (von Ellenrieder et al., 2016) and that their incidence varies during sleep (Von Ellenrieder et al., 2017), but in hours-long recordings, no differences in incidence have been reported in the mesial temporal lobe (Dümpelmann et al., 2015). Furthermore, the variability of FRs across time (Gliske et al., 2018) indicates that longer nocturnal recordings in humans are necessary. It also suggests that changes in excitability across time could explain this change in FR incidence. Last, but not least, another report did not find a state-dependent expression of FRs in the kainate rat model of temporal lobe epilepsy (Ewell et al., 2019), thus indicating that the variability of FRs across sleep and wake is still an open question, at least in rodents. Hence, the main conclusion on the effect of sleep-wake transitions is that these transitions impact the likelihood of stochastic events, more than dictating the direction (increases vs decreases) of change. It also highlights that the specificity of FRs to epileptogenic parenchyma could vary across the sleep-wake cycle, which would be crucial in epileptology (Dimakopoulos et al., 2024; Roehri et al., 2018; Sheybani et al., 2019, 2018; Zijlmans et al., 2012, 2009).”
Minor Comments:
(1) Abstract, please include the frequency range of fast ripples explored in this study.
The abstract has been updated accordingly.
(2) Abstract, consider including the exact epilepsy model system in rats instead of "a rodent model of hippocampal epilepsy".
The abstract has been updated accordingly.
(3) Line 87, while Ylinen uses the term "high frequency oscillations" to refer to ripples up to 200Hz, which are different from the ones discussed here, better to rephrase or use another reference.
The reference has been changed for Bragin et al. (1999), Epilepsia
