Evidence level: C
选钵要靠「跟随签名」做个体匹配,但签名在哪个调制频率以上才重测得稳,目前没人逐频点标定过——这条信度-频率曲线缺着,低频段的选钵决策就没有地基。
正文
用听觉稳态反应(ASSR)给颂钵选型做个体校准,是本系列近几篇反复推的一个方案:给受试者扫一组幅度调制声,记录每个调制频率上的脑电跟随强度(相位锁定与功率调制),得到一条「个体签名」,再拿它去匹配钵的拍频特性。前几篇已经裁决了标定窗下限(PMID 21909024 显示跨会话幅度与潜伏期变异不可忽略)和「签名 vs 体质量表」两条路线的分歧。但这套方案还压着一块没搬开的石头:信度本身是随频率变化的,0.5-20Hz 这段扫频里,并不是每个频点的读数都值得拿去做决策。
现有文献已经给出三条线索。第一,调制频率不同,重测信度明显不同:MEG 实验里 40Hz 调制的相位锁定指标(ITPC)跨会话 ICC 高达 0.86-0.96,而 5Hz 调制条件的信度整体低于 40Hz 条件,功率类指标 ICC 只有 0.61-0.82(PMID 26216274)。第二,指标类型也在分叉:同一批数据里,事件相关谱扰动(ERSP)的重测信度好到优秀,逐试次相位相干(ITC)却只有差到中等——同一频率、同一批人,换个读数口径信度结论就翻转(PMID 35030446)。第三,跨设备、跨时间的稳定性虽有正面报告——EEG 与 MEG 测 40Hz ASSR 信度相当(PMID 28161286),多中心方案在 2025 年补了标准化流程(PMID 40480970),健康人与临床样本的 ASSR 时频指标重测结果同样呈「功率稳、相位分化」格局(PMID 31228795),儿童多频 ASSR 也有独立信度数据(PMID 38050889)——但这些正面证据几乎全部落在 20Hz 以上的伽马段,0.5-20Hz 逐频点的重测信度曲线,文献里没有现成答案。
为什么这条曲线对选钵特别要命?因为钵的声学特征恰恰压在低频段:两只钵的拍频(两个相邻分音的频率差)通常落在零点几赫兹到十几赫兹之间,正好扫进 0.5-20Hz 这个信度未标定区。如果逐频点重测下来,G 系数要到某个频率(比如 10Hz 或 15Hz)以上才普遍过 0.6 的可用线,那么大量钵的拍频特征就落在「测不准」的区间——用它们做个体匹配,等于拿一个重测都复现不了的读数去当匹配依据。反过来,如果低频段信度意外地好,这套方案才立得住。
所以本篇提出的裁决设计很直接:0.5-20Hz 按 0.5Hz 步长扫频,同一批受试者隔天重测,每个频点单独算 G 系数(或 ICC),画出一条「信度-频率曲线」,读出 G≥0.6 的下限频率;再把四只钵的实际基频与拍频逐个映射到这条曲线上,标出哪些落在可用区、哪些落在不可用区。这个设计同时回答两件事:签名方案的地基在哪,以及现有钵库有多大比例被排除在决策之外。此外,读数口径要按 PMID 35030446 的教训双报(功率与相位分开),不能只报一个合成指标。
阈值为什么定 G≥0.6?G 系数来自概化理论,它把「受试者间真实差异」和「测量误差」分开算:低于 0.6 意味着误差方差的占比大到足以让两个真实不同的人测出接近的读数——这时拿签名做匹配,分组错误率会盖过匹配收益。0.6 不是金标准而是工程下限,类似结论在临床生物标志物信度评估里被广泛采用。对选钵这个具体决策来说,下限频率每低 1Hz,可用的钵就多一批;每高 1Hz,就有更多钵的拍频被排除在决策之外——所以这条曲线的形状直接决定整套方案的覆盖面,而不只是精度问题。
还有一个必须提前想清楚的后手:如果低频段信度确实过不了线,替代路线不是放弃,而是换读数对象。可选项至少有三个:一是放弃拍频、改用钵的主基频(通常在 200-600Hz,虽然不直接对应调制频率,但物理量本身重测稳定);二是把神经签名的提取端上移到 20Hz 以上已知高信度区,让钵的特征先做频率变换再进签名空间;三是承认低频段只能做群体层结论,个体匹配只在相位锁定达标的那几个频点上做。三条路线哪条可行,同样要等那条信度-频率曲线画出来才能裁决。
证据与局限
已验证: 40Hz ASSR 的相位锁定与功率指标跨会话重测信度良好到优秀(ICC 0.61-0.96,MEG 与 EEG 双系统一致,PMID 26216274 / 28161286);ERSP 读数在三次重测中稳定(PMID 35030446)。
部分验证: 多中心标准化方案与儿童多频 ASSR 信度(PMID 40480970 / 38050889);跨会话幅度与潜伏期变异的量级(PMID 21909024)。
待验证: 0.5-20Hz 逐频点的重测信度曲线目前完全空白;G≥0.6 的下限频率无人测过;钵的实际拍频在该曲线上的落点分布无任何实测数据;低频拍频签名能否支撑选钵决策属研究假说。
相关阅读
L1 网络状态:本主题(0.5-20Hz 逐频点重测信度标定)尚未进入 L1 对应网络;L1 相邻边仅两条 A 级边——「assr-minimum-calibration-window-bowl-selecti」(PMID 21909024,脑电)与「assr-signature-vs-dosha-bowl-selection」(PMID 29269119,次声)。本篇证据等级按规则封顶 C。
证据等级: C
参考文献: PMID 26216274, PMID 28161286, PMID 35030446, PMID 31228795, PMID 40480970, PMID 38050889, PMID 21909024, PMID 29269119, PMID 40702066
证据等级:A=有PubMed直接支撑,B=科学类比/间接证据,C=传统文化共识/初步探索
免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。
20 Hz Reliability Floor for ASSR Bowls: Which Trace in the 0.5–20 Hz Sweep Must Be Followed for Re-testing
Evidence level: C
ASSR Bowl Selection Decision Reliability Lower Bound at 20 Hz
Bowl selection relies on subject matching via “following signatures,” but the modulation frequency above which these signatures can be reliably retested remains unstated in current literature—no individual frequency points have been systematically calibrated. Consequently, this reliability-frequency curve is missing, leaving the low-frequency range of bowl selection decisions without a foundational basis.
Body
Using Auditory Steady-State Responses (ASSR) to calibrate singing bowl selection for individual profiles is the proposal repeatedly advanced in recent installments of this series: presenting amplitude-modulated sound to subjects, recording EEG following strength at each modulation frequency (phase-locking and power modulation), deriving an “individual signature,” and matching it against the bowl’s beat frequency characteristics. Previous articles have already adjudicated the lower bound of the calibration window (PMID 21909024 shows inter-session amplitude and latency variability cannot be ignored) and the divergence between two routes: “signature versus body mass scale.” This current proposal, however, rests on an unresolved stone: reliability itself varies with frequency; within the 0.5–20 Hz sweep, not every frequency point’s reading is worth using for decision-making.
Existing literature provides three clues. First, modulation frequency significantly affects test-retest reliability: in MEG experiments, the phase-locking index (ITPC) for 40 Hz modulation shows inter-session ICCs as high as 0.86–0.96, whereas the 5 Hz modulation condition exhibits overall lower reliability than the 40 Hz condition, with power-based metrics ICCs ranging only from 0.61 to 0.82 (PMID 26216274). Second, metric type also diverges: within the same dataset, event-related spectral perturbation (ERSP) demonstrates excellent test-retest reliability, while trial-by-trial phase coherence (ITC) ranges from poor to moderate—changing the reading metric flips the reliability conclusion for the same frequency and same subjects (PMID 35030446). Third, although cross-device and cross-time stability has been positively reported—EEG and MEG measuring 40 Hz ASSR show comparable reliability (PMID 28161286), multi-center protocols standardized procedures in 2025 (PMID 40480970), and both healthy and clinical samples exhibit ASSR time-frequency metrics following a “power-stable, phase-divergent” pattern (PMID 31228795), with children’s multi-frequency ASSR also having independent reliability data (PMID 38050889)—these positive findings almost exclusively fall within the gamma band above 20 Hz. No existing literature provides a ready-made answer for the test-retest reliability curve across individual frequency points from 0.5 to 20 Hz.
Why is this curve critical for bowl selection? Because the acoustic characteristics of bowls precisely occupy the low-frequency band: the beat frequency (the difference between two adjacent partials) of two bowls typically falls between a fraction of a hertz and several tens of hertz, squarely within the 0.5–20 Hz reliability-unspecified zone. If test-retest G coefficients only universally exceed 0.6 above a certain frequency (e.g., 10 Hz or 15 Hz), then the beat frequency characteristics of many bowls fall into the “unmeasurable” interval—using them for individual matching would equate to relying on readings that cannot be reproduced across repeated tests. Conversely, the scheme only holds if low-frequency reliability unexpectedly proves adequate.
Thus, this article proposes a direct adjudication design: sweep frequencies from 0.5 to 20 Hz in 0.5 Hz steps, retest the same subjects on consecutive days, calculate G coefficients (or ICCs) separately for each frequency point, and plot a “reliability-frequency curve,” reading out the lower limit frequency where G ≥ 0.6; then map the actual fundamental and beat frequencies of four bowls onto this curve, marking which fall into usable zones and which into unusable zones. This design simultaneously answers two questions: where the foundation of the signature scheme lies, and what proportion of existing bowl libraries are excluded from decision-making. Furthermore, reading metrics must be reported dually per the lesson of PMID 35030446 (separating power and phase), not as a single composite index.
Why set the threshold at G ≥ 0.6? The G coefficient derives from generalized theory, separating “true inter-subject differences” from “measurement error”: below 0.6, the proportion of error variance becomes large enough that two genuinely different individuals could yield nearly identical readings—using signatures for matching would then result in grouping error rates exceeding matching benefits. 0.6 is not a gold standard but an engineering lower bound; similar conclusions are widely adopted in clinical biomarker reliability assessments. For the specific decision of bowl selection, lowering the threshold frequency by 1 Hz adds more usable bowls; raising it by 1 Hz excludes more bowls’ beat frequencies from decision-making—thus, the curve’s shape directly determines the scheme’s coverage, not merely its precision.
One more contingency must be considered in advance: if low-frequency reliability indeed fails to meet the threshold, the alternative route is not abandonment but changing the reading target. At least three options exist: (1) abandon beat frequency and instead use the bowl’s primary fundamental frequency (typically 200–600 Hz, though not directly corresponding to modulation frequency, the physical quantity itself shows stable test-retest); (2) shift the neural signature extraction endpoint upward to the known high-reliability region above 20 Hz, having bowl characteristics undergo frequency transformation before entering the signature space; (3) acknowledge that low-frequency bands can only yield group-level conclusions, with individual matching restricted to those frequency points where phase-locking meets criteria. Which of these three routes is feasible must await the completion of that reliability-frequency curve before adjudication.
Evidence and Limitations
Verified: Phase-locking and power metrics of 40 Hz ASSR demonstrated good-to-excellent cross-session reliability (ICC 0.61–0.96) across both MEG and EEG systems (PMID 26216274 / 28161286); ERSP readings remained stable across three retest sessions (PMID 35030446).
Partially Verified: Cross-session reliability of multi-center standardized protocols and pediatric multi-frequency ASSR (PMID 40480970 / 38050889); magnitude of cross-session variation in amplitude and latency (PMID 21909024).
Pending Verification: Cross-session reliability curves for individual frequency points (0.5–20 Hz) remain entirely uncharted; no measurements have established the lower frequency threshold of G ≥ 0.6; the actual beat frequency of the bowl lacks empirical data on its distribution along this curve; whether low-frequency beat signatures can support bowl selection decisions remains a research hypothesis.
Related Reading
- ASSR bowl calibration: bifurcation decision based on peak location vs. peak height dual indicators
- Stochastic bowl selection vs. uniform standard bowl: spectral contingency decision
- Three-factor constitution tags vs. dual-model separation of imbalance states
- Research pool: bio-frequency code
L1 network status: This topic (0.5–20 Hz point-by-point retest reliability calibration) has not yet entered the L1 corresponding network; the L1 adjacent edge contains only two A-level edges—「assr-minimum-calibration-window-bowl-selecti」(PMID 21909024, EEG)and「assr-signature-vs-dosha-bowl-selection」(PMID 29269119, infrasound). Per the rules, this topic’s evidence level is capped at C.
Evidence level: C
References: PMID 26216274, PMID 28161286, PMID 35030446, PMID 31228795, PMID 40480970, PMID 38050889, PMID 21909024, PMID 29269119, PMID 40702066
Evidence level: A = direct PubMed support; B = scientific analogy/indirect evidence; C = traditional consensus/preliminary exploration.
Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.