{"id":"e04d04e8-ef06-40d2-b84a-3a157a4cb148","arxiv_id":"2412.18445","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A VR-EEG study reports a parietal-positive, frontal-negative evoked potential around 600 ms that distinguishes strong vection from weak or no vection during forward acceleration.","lead":"Researchers recorded brain activity from 30 people in virtual reality and found a voltage peak about 600 milliseconds after a visual acceleration that was stronger when people felt they were moving. The result points toward an objective, brain-based measure of the illusion of self-motion, which could help VR systems detect and reduce motion sickness.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The P600 vection marker likely reflects trial-level pseudoreplication: SV and WV&NV trials are highly imbalanced across participants, and the bootstrap ignores subject clustering; the effect may not survive subject-level analysis.","rationale":"The paper presents a plausible and potentially useful exploration of EEG markers for vection, and it thoughtfully replicates earlier alpha-suppression findings and acceleration-direction effects. However, the headline contribution—a P600 evoked-potential marker of vection—is supported only by a trial-level bootstrap that treats single epochs as independent units, despite the data being hierarchical (30 participants, up to 78 trials each) and highly imbalanced across conditions. The reader's weakest_assumption correctly identified the pseudoreplication problem. I agree with the CONDITIONAL verdict because the concern is substantive and directly addressable: a subject-level or mixed-effects analysis could confirm or refute the marker's generality. The proposed concrete test—per-participant paired differences and leave-one-subject-out robustness—would settle whether the P600 survives when participant identity is respected. Until such an analysis is provided, the central claim should not be accepted as a general vection marker, but the work is valuable enough to warrant conditional acceptance pending this verification.","tokens_in":13893,"tokens_out":5968,"duration_ms":54551,"concrete_test":"For each participant, extract FA1 epochs labeled SV and WV&NV, compute the mean ERP amplitude at the parietal electrodes used in Fig. 8 within a 550–650 ms window, and average per participant per condition. Restrict to participants who have at least one epoch in both conditions, then perform a paired one-sample test (e.g., Wilcoxon signed-rank) on the per-participant SV minus WV&NV differences. If this subject-level test is not significant at p < 0.05 (or the effect size is negligible), the trial-level bootstrap result is an artifact of participant clustering. Additionally, run the original bootstrap analysis excluding each participant in turn; if significance at 600 ms disappears when any single participant is removed, the effect is not robust across subjects.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim (Section 4.3, Fig. 8) rests on a trial-level bootstrap over 87 SV and 92 WV&NV epochs from 30 participants. This treats each epoch as an independent observation, but epochs from the same participant are correlated due to individual anatomy, electrode placement, and baseline EEG. The imbalance is severe: Table 1 shows participants 9, 10, 11, 12, 16, 27, and 30 each contribute between 36 and 65 SV trials, while participants 15, 20, and 29 contribute zero SV trials. Consequently, the SV average waveform is dominated by a small subset of individuals. The bootstrap with replacement over epochs inflates the effective sample size and produces overconfident confidence intervals; the shaded 'statistically significant' periods in Fig. 8 may therefore reflect differences between these particular participants rather than an effect of vection per se. Without a subject-level analysis that accounts for within-subject correlation—such as a mixed-effects model, per-subject difference testing, or leave-one-subject-out validation—the claim that the P600 component is a general marker of vection is not yet established. This is a correctness risk, not merely a stylistic choice, because the pseudo-replication directly undermines the inferential validity of the primary novel result.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports a VR/EEG study with 30 participants who experienced sudden forward or backward visual acceleration in a starfield environment and rated their perceived vection on a four-point scale. The central claim is that an evoked potential with a positive deflection around 600 ms after stimulus onset in parietal regions and a simultaneous negative deflection in frontal regions distinguishes strong vection from weak/no vection (Abstract, Sec. 4.3, Fig. 8). The paper also reports replication of previously published EEG markers of acceleration, alpha suppression during vection, and correlations between vection reports and Simulator Sickness Questionnaire (SSQ) scores. Analyses are based on median waveforms with trial-level bootstrap confidence intervals and topographic maps.","tokens_in":14120,"tokens_out":3573,"duration_ms":35611,"significance":"If the P600 vection marker is statistically robust, the paper would provide a plausible objective and potentially real-time neural index of subjective vection in VR, addressing a recognized gap in vection research and offering practical value for adaptive VR systems. The authors are appropriately cautious about the need for subjective ground truth, and the paper includes a data-availability commitment (BIDS format, Sec. 3.6). The replication of alpha suppression and of acceleration-related EEG patterns is a useful confirmation of earlier findings. However, the primary novel claim—the P600 marker—rests on a trial-level analysis whose inferential validity is not established; the statistical issues are correctness risks that must be addressed before the central claim can be accepted.","major_comments":[{"comment":"The central P600 comparison is based on 87 strong-vection (SV) trials versus 92 weak/no-vection (WV&NV) trials pooled across 30 participants, analyzed with a trial-level bootstrap that resamples individual epochs with replacement. This treats epochs from the same participant as independent, which is implausible given within-subject correlations in EEG. Table 1 shows severe imbalance: participants 9, 10, 11, 12, 16, 27, and 30 each contribute between 36 and 65 SV trials, while participants 15, 20, and 29 contribute zero SV trials; hence the SV average waveform is dominated by a small subset of individuals. The manuscript reports no subject-level mixed-effects model, per-subject difference test, or leave-one-subject-out analysis. The shaded 'significant' periods in Fig. 8 may therefore reflect inter-individual differences rather than an effect of vection state. I request a subject-level analysis that accounts for within-subject correlation before the P600 marker is claimed as a general vection correlate.","section":"Sec. 4.3, Fig. 8"},{"comment":"Statistical significance is assessed by computing pointwise 95% bootstrap confidence intervals for each time sample and electrode (or averaged electrode groups) and declaring a shaded region significant where intervals do not overlap. With 128 Hz sampling over epochs of about 1.5 seconds and 14 electrodes, this entails hundreds or thousands of simultaneous comparisons with no multiple-comparison correction. The resulting 'statistically significant periods' are therefore likely to include numerous false positives. A cluster-based permutation test across time and electrodes, or an equivalent correction for multiple comparisons, is needed to support the existence of the P600 component and the acceleration markers.","section":"Secs. 4.2 and 4.3"},{"comment":"The binary categorization used for the main vection analysis is post hoc: the NV and WV categories are merged, MV is excluded, and the SV category is compared against this constructed class. This grouping is decided after inspecting the data (Sec. 4.1), which risks bias if the merging or exclusion criterion is correlated with the EEG signals under study. The manuscript should either justify the grouping a priori from the rating scale semantics or demonstrate robustness to alternative groupings (for example, NV-only versus SV, or an ordinal analysis including MV). This issue is load-bearing because the 'weak/no vection' class is the reference category for the central P600 comparison.","section":"Sec. 4.1"},{"comment":"The reported Pearson correlations between SV counts and SSQ total and subscale scores (r = 0.55, 0.45, 0.51, 0.62) are computed on only 30 subjects and four outcome measures without correction for multiple testing. Moreover, the per-subject SV count is derived from an unbalanced number of trials per participant, which can inflate or distort the correlation. These correlations are secondary to the main claim, but as reported they overstate the strength of evidence for a vection-sickness association; a mixed-effects or rank-based analysis with adjusted inference would be more appropriate.","section":"Sec. 4.3, SSQ correlations"}],"minor_comments":[{"comment":"There are several typos in this section: 'steated' should be 'seated', 'minimze' should be 'minimize', and 'bloc' should be 'block'.","section":"Sec. 3.5"},{"comment":"The artifact rejection threshold is given as '125mV', which is almost certainly a typo for microvolts (µV); also, the filter passband (0.3–10 Hz) is described without specifying filter order for the PSD plot, which would hamper exact reproduction.","section":"Sec. 3.6"},{"comment":"The figure caption says 'during FA1' while the text reports the graph is based on 92 WV&NV and 87 SV trials; it is unclear whether only forward-acceleration trials are used or whether both FA1 and BA1 are pooled. Please clarify the trial selection and specify which electrodes are averaged for the parietal and frontal regions.","section":"Sec. 4.3, Fig. 8"},{"comment":"Table 1 lacks row labels or subject identifiers, making it impossible to map the text's references to 'participants 9, 10, 11...' onto the table. Adding subject IDs would improve transparency.","section":"Table 1"},{"comment":"Reference [6] lists the author as 'Cz.'; this should be corrected to Helmholtz, H. von, or the original source should be cited properly.","section":"References"},{"comment":"The similarity between the P600 pattern and the P3 oddball response in Nolan et al. [26] is stated qualitatively; the claim would be more convincing with a quantitative comparison or by explicitly noting the limits of visual resemblance.","section":"Sec. 5"}],"recommendation":"major_revision","confidential_remarks":"I see no evidence of circularity in the analysis: no fitted quantity reduces to an input parameter, and the acceleration and alpha findings are consistent with prior work. The main issue is the pseudoreplication in the novel P600 result, which is fixable in principle by a subject-level analysis. Given that the paper is otherwise methodical and includes a data-sharing commitment, I believe the central claim can be rehabilitated, but it requires a substantive reanalysis rather than copy-editing."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The headline: this paper reports a genuinely new evoked-potential correlate of vection—a parietal-positive/frontal-negative deflection around 600 ms after forward acceleration—and it also replicates the known alpha-suppression effect. If the P600 holds up, it's a useful objective marker for VR and cybersickness research. The experimental design is solid: a minimalist star-field, controlled acceleration events, and a four-point vection scale. The analysis, however, has a load-bearing flaw: the main comparison between strong vection (SV) and weak/no vection (WV&NV) is done at the trial level with bootstrapping, but the trials are heavily clustered by participant. Subject 10 contributes 65 SV trials, while subjects 15, 20, and 29 contribute none. Treating 87 trial epochs as independent observations inflates the effective sample size, and the 95% CIs in Fig. 8 likely reflect a few individuals rather than a general effect. The post hoc merging of NV and WV and the exclusion of MV also makes the grouping rule look data-dependent. And there's no correction for multiple comparisons across time points and electrodes, so the shaded 'significant' intervals are optimistic.\n\nThe paper isn't sloppy in other respects. It correctly notes the novelty claim, appropriately cites prior work on alpha suppression and acceleration markers, and the SSQ correlations are a nice addition. The similarity to the vestibular oddball P3 is a plausible hypothesis, not overclaimed.\n\nThe bottom line: the P600 is a promising exploratory finding, but the inferential statistics don't yet support it as a general marker. I'd send this to peer review, not desk-reject it, but I'd ask the authors to redo the vection analysis with subject as a random effect (or per-subject averages) and with cluster-based permutation or FDR correction. The data-sharing plan helps.\n\nThis paper is mainly useful to people working on objective vection measures, passive BCIs, and cybersickness prediction. It's a good discussion piece for a reading group on pseudoreplication. I wouldn't cite the P600 as established, but I'd note the alpha replication.\n\nRecommendation: send to peer review with major revisions.","headline":"A promising but statistically under-supported P600 vection marker; worth reviewing but needs subject-level analysis.","tokens_in":14697,"tokens_out":3028,"would_cite":false,"duration_ms":28690,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"EEG reveals a 600-millisecond brain response that tracks the illusion of self-motion in virtual reality.","keywords":["vection","EEG","evoked potentials","P600","virtual reality","self-motion perception","cybersickness","acceleration perception"],"falsifier":"Average each participant's vection and no-vection trials into one epoch per condition, so each of the 30 participants contributes exactly one trial to each group, and retest the 600 ms parieto-frontal difference; if it disappears or is carried by one or two participants, the claimed general vection marker is refuted. A matched no-vection control, using optic flow that participants do not interpret as self-motion, should also fail to produce the parietal P600.","tokens_in":13692,"feed_emoji":"🧠","tokens_out":8549,"duration_ms":72932,"temperature":0.7,"pith_summary":"This paper tries to establish that the subjective illusion of self-motion in virtual reality, known as vection, leaves a measurable, time-locked trace in the EEG. In an experiment with 30 participants watching a star field accelerate forward or backward while wearing a VR headset, the authors compared brain responses from trials where participants reported strong vection with trials where they reported weak or no vection. They found a positive deflection around 600 ms after acceleration onset over parietal electrodes and a simultaneous negative deflection frontally, a pattern they propose as an evoked potential of vection. If the marker holds, VR systems could detect vection online from EEG rather than relying on questionnaires, opening the way to real-time adaptation for comfort and reduced cybersickness.","feed_headline":"EEG wave at 600 ms marks the VR illusion of self-motion","feed_subtitle":"Parietal positivity at 600 ms separates strong from weak or no vection, enabling real-time EEG-based VR adaptation.","key_machinery":"The load-bearing object is the 600 ms evoked potential (EP), a stimulus-locked voltage deflection in the EEG after the onset of a sudden visual acceleration. To isolate it, the authors compare the median response to forward acceleration in trials rated SV against trials rated WV or NV, merge the weak and no-vection categories, and use the moderate category as a buffer; they assess significance with 10,000 bootstrap resamples and plot the spatial contrast as topographic maps. The EP is what carries the vection claim, while the FCz/Cz acceleration and direction markers and alpha suppression are secondary signatures used for replication and future fusion.","core_discovery":"The paper's central claim is that subjective vection has a distinct evoked potential: for forward acceleration, trials rated as strong vection show a positive deflection near 600 ms after stimulus onset over parietal electrodes and a simultaneous negative deflection frontally, while weak/no-vection trials show the reverse polarity. The authors support this with a non-parametric bootstrap comparison over 87 strong-vection and 92 weak/no-vection trials from 30 participants, plus topographic maps of the spatial contrast. They position this as the first evoked potential tied to subjective vection and interpret it as the brain resolving a visual-vestibular conflict, analogous to a P3-like oddball response to real-world self-motion. The paper also reports distinct EEG markers for acceleration presence and direction, alpha suppression during vection, and a correlation between strong vection and simulator sickness scores.","pith_inferences":["The trial-level bootstrap treats 179 epochs from 30 participants as exchangeable; a subject-level mixed model or leave-one-subject-out validation would test whether the P600 is a general marker or the product of a few vection-prone participants.","If the P600 survives subject-level validation, the same paradigm could be extended to backward acceleration, lateral motion, or walking-in-place illusions; the paper only demonstrates it for forward acceleration.","Vection questionnaires ask for conscious ratings after each trial, whereas an online EEG marker would resolve when in the trial vection starts, potentially sharpening onset-time measures that self-report is known to distort.","The comparison with vestibular oddball responses suggests the component may index surprise or sensory conflict rather than vection per se; a task that manipulates expectedness without changing visual self-motion could separate the two."],"forward_implications":["A real-time EEG classifier could flag a vection episode within roughly 600 ms of an acceleration event, rather than waiting for a post-hoc questionnaire.","VR systems that adapt content, such as field of view, speed, or visual gain, could use the marker to reduce sensory conflict before cybersickness builds.","Because strong vection correlates with SSQ nausea and disorientation scores, the marker may double as an early warning of impending simulator sickness.","The acceleration and direction markers on FCz and Cz give additional features that could make a vection detector robust across different motion profiles.","Alpha suppression during vection offers an independent, slower-frequency signature that could be fused with the P600 marker in future detectors."],"supporting_citations":[{"why":"Defines vection and frames the need for objective measures, providing the ground-truth rationale for pairing EEG with self-report.","marker":"[30]"},{"why":"Supplies the experimental protocol and previously reported acceleration EEG markers that this study replicates with a larger cohort.","marker":"[46]"},{"why":"Reviews EEG and fMRI studies of vection and identifies alpha-band activity as the prior neural correlate this paper replicates.","marker":"[2]"},{"why":"Shows the vestibular oddball P3 pattern in real self-motion that the paper matches against its 600 ms vection response.","marker":"[26]"},{"why":"Reports alpha-wave modulation during self-motion perception under visual-vestibular conflict, supporting the alpha suppression finding.","marker":"[12]"},{"why":"Describes the vection-inducing stimulus made of moving spheres and the visually induced motion sickness paradigm adapted for the virtual environment.","marker":"[17]"},{"why":"Provides the Simulator Sickness Questionnaire whose scores correlate with strong vection in the results.","marker":"[15]"}],"fun_headline_variants":["EEG P600 signal reveals VR vection in real time","Parietal EEG wave at 600 ms detects vection","Vection's brain wave: P600 at 600 ms","Real-time vection detection via EEG P600","P600 EEG spike marks VR self-motion illusion"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The result depends on treating the 87 strong-vection and 92 weak/no-vection EEG segments from 30 participants as approximately independent samples; if the 600 ms difference is produced by a few vection-prone individuals, the marker would not generalize.","fun_headline_variants_meta":{"raw":{"variants":["EEG P600 signal reveals VR vection in real time","Parietal EEG wave at 600 ms detects vection","Vection's brain wave: P600 at 600 ms","Real-time vection detection via EEG P600","P600 EEG spike marks VR self-motion illusion"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000249,"raw_usage":{"total_tokens":1572,"prompt_tokens":986,"completion_tokens":586,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":602,"completion_tokens_details":{"reasoning_tokens":505}},"tokens_in":602,"tokens_out":586,"duration_ms":5553,"temperature":1.0,"reasoning_tokens":505,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T04:42:23.041889+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Average each participant's vection and no-vection trials into one epoch per condition, so each of the 30 participants contributes exactly one trial to each group, and retest the 600 ms parieto-frontal difference; if it disappears or is carried by one or two participants, the claimed general vection marker is refuted. A matched no-vection control, using optic flow that participants do not interpret as self-motion, should also fail to produce the parietal P600.","supporting_citations":[{"cited_title":"Future challenges for vection research: Definitions, functional significance, measures, and neural bases","cited_arxiv_id":null,"evidence_quote":"Defines vection and frames the need for objective measures, providing the ground-truth rationale for pairing EEG with self-report."},{"cited_title":"EEG Markers of Acceleration Perception in Virtual Reality","cited_arxiv_id":null,"evidence_quote":"Supplies the experimental protocol and previously reported acceleration EEG markers that this study replicates with a larger cohort."},{"cited_title":"Neuropsychological Approaches to Visually-Induced Vection: An Overview and Evaluation of Neuroimaging and Neurophysio- logical Studies","cited_arxiv_id":null,"evidence_quote":"Reviews EEG and fMRI studies of vection and identifies alpha-band activity as the prior neural correlate this paper replicates."},{"cited_title":"Neural correlates of oddball detection in self- motion heading: A high-density event-related potential study of vestibular integration","cited_arxiv_id":null,"evidence_quote":"Shows the vestibular oddball P3 pattern in real self-motion that the paper matches against its 600 ms vection response."},{"cited_title":"Modulation of alpha waves in sensori- motor cortical networks during self-motion perception evoked by different visual-vestibular conflicts","cited_arxiv_id":null,"evidence_quote":"Reports alpha-wave modulation during self-motion perception under visual-vestibular conflict, supporting the alpha suppression finding."},{"cited_title":"The effect of visual motion stimulus characteristics on vection and visually induced motion sickness","cited_arxiv_id":null,"evidence_quote":"Describes the vection-inducing stimulus made of moving spheres and the visually induced motion sickness paradigm adapted for the virtual environment."},{"cited_title":"Simulator Sickness Question- naire: An Enhanced Method for Quantifying Simulator Sickness","cited_arxiv_id":null,"evidence_quote":"Provides the Simulator Sickness Questionnaire whose scores correlate with strong vection in the results."}],"review_version":1}