德国 Brain Products 脑电放大器 actiCHamp

摘要 · 中文整理：引言：大语言模型（LLM）正在改变大多数行业，并将成为人类数字体验的基石。尽管将显式人类反馈纳入基于 LLM 的聊天机器人的训练与开发对当前的进展至关重要，但如何使其更好地与人类价值观对齐仍需更多研究。被动式脑机接口（pBCI）所实现的隐式人类反馈有望揭示用户在与聊天机器人交互时认知与情感状态的细微差别。本研究探讨利用 pBCI 解码对文本刺激反应的心理状态的可行性，为神经自适应聊天机器人奠定基础。方法：设计了两种以文本刺激诱发道德判断与错误加工的范式。在受试者完成阅读任务时使用 64 个凝胶电极记录脑电。采用窗口均值方法与线性判别分析（LDA）在离线状态下针对全成分与脑成分数据获得心理状态分类器，并对相应事件相关电位（ERP）进行目视检查。结果：道德显著性在单试次水平被成功解码，基于 600 ms 数据窗的平均校准准确率为 78%。后续分类器无法区分道德判断的一致（道德认同）与不一致（道德不认同）。对事实错误的错误加工解码平均校准准确率为 66%。所识别的 ERP 与其他研究发现部分一致。讨论：本研究证明了利用 pBCI 从读者脑数据中在单试次水平区分心理状态的可行性。仍需进一步工作以从离线转向在线研究，并了解在控制较少的语言任务和更真实的聊天机器人交互中能否获得可靠的 pBCI 分类器。本工作为理解和利用基于神经的隐式人类反馈进行 LLM 对齐迈出了初步步伐。

原文摘要：Introduction: Large-language models (LLMs) are transforming most industries today and are set to become a cornerstone of the human digital experience. While integrating explicit human feedback into the training and development of LLM-based chatbots has been integral to the progress we see nowadays, more work is needed to understand how to best align them with human values. Implicit human feedback enabled by passive brain-computer interfaces (pBCIs) could potentially help unlock the hidden nuance of users' cognitive and affective states during interaction with chatbots. This study proposes an investigation on the feasibility of using pBCIs to decode mental states in reaction to text stimuli, to lay the groundwork for neuroadaptive chatbots. Methods: Two paradigms were created to elicit moral judgment and error-processing with text stimuli. Electroencephalography (EEG) data was recorded with 64 gel electrodes while participants completed reading tasks. Mental state classifiers were obtained in an offline manner with a windowed-means approach and linear discriminant analysis (LDA) for full-component and brain-component data. The corresponding event-related potentials (ERPs) were visually inspected. Results: Moral salience was successfully decoded at a single-trial level, with an average calibration accuracy of 78% on the basis of a data window of 600 ms. Subsequent classifiers were not able to distinguish moral judgment congruence (i.e., moral agreement) and incongruence (i.e., moral disagreement). Error processing in reaction to factual inaccuracy was decoded with an average calibration accuracy of 66%. The identified ERPs for the investigated mental states partly aligned with other findings. Discussion: With this study, we demonstrate the feasibility of using pBCIs to distinguish mental states from readers' brain data at a single-trial level. More work is needed to transition from offline to online investigations and to understand if reliable pBCI classifiers can also be obtained in less controlled language tasks and more realistic chatbot interactions. Our work marks preliminary steps for understanding and making use of neural-based implicit human feedback for LLM alignment.

关键词：被动式脑机接口、被动式脑机接口、大语言模型、错误加工、道德判断、人工智能对齐

研究目标：探讨能否利用被动式脑机接口从脑电中单试次解码读者对文本刺激的道德显著性与错误加工等心理状态，为基于隐式神经反馈的神经自适应聊天机器人与 LLM 对齐提供基础。

主要结论：被动式脑机接口可在单试次水平解码道德显著性（约 78%）与错误加工（约 66%），但无法区分道德一致与不一致；证明了该方法的可行性，后续需向在线、真实交互场景推进。

创新点：首次将被动式脑机接口与 LLM 聊天机器人对齐问题结合，设计文本诱发道德判断与错误加工两种范式，并比较全成分与脑成分数据的解码效果。

后续建议：将离线分类器迁移到在线实时神经自适应聊天机器人系统；扩大样本量并纳入更自然的自由阅读与对话场景；结合眼动数据进行多模态心理状态解码；探索深度学习模型以提升道德一致性判别能力；研究跨会话、跨被试的分类器泛化与校准迁移

APA 引文：Gherman, D. E., & Zander, T. O. (2025). Towards neuroadaptive chatbots: a feasibility study. Frontiers in Neuroergonomics, 6, 1589734. https://doi.org/10.3389/fnrgo.2025.1589734

仪器引用：A total of 64 active actiCAP slim gel electrodes (Brain Products GmbH, Gilching, Germany) were used according to the extended 10–20 international system (Klem et al., 1999). The signal was sampled at 500 Hz with an actiCHamp amplifier.

摘要 · 中文整理：本文提出一种基于脑电响应的脑机接口，用于检测行走过程中障碍物出现时的停止意图。在跑步机上对 5 名受试者进行测试，采用两个串联卷积神经网络将每分钟误报由 31.8 降至 3.9，无误报正确检测指标由 34.9% 提高到 60.3%。在下肢外骨骼闭环实时测试中，3 名健康受试者获得每分钟 3.8 次误报与 49.3% 的准确率。迁移学习技术缩短了校准时间；2 名不完全脊髓损伤患者获得 37.9% 的准确率与每分钟 7.7 次误报。

原文摘要：This work presents a brain-machine interface (BMI) that detects the intention to stop when an obstacle appears during walking, based on EEG responses. The interface was tested on a treadmill with five subjects; the use of two consecutive convolutional neural networks reduced false positives from 31.8 to 3.9 per minute and improved the metric of correct detections without false alarms from 34.9% to 60.3%. In real-time closed-loop tests with a lower-limb exoskeleton, three healthy subjects obtained 3.8 false positives per minute and 49.3% accuracy. Transfer-learning techniques allowed reducing the calibration time; two patients with incomplete spinal cord injury obtained 37.9% accuracy with 7.7 false positives per minute.

关键词：脑机接口、停止意图、外骨骼、脑电图、迁移学习、闭环

研究目标：解决外骨骼辅助行走中如何通过脑电实时、低误报地检测用户因障碍物出现而产生的停止意图，并缩短个体校准时间的问题。

主要结论：两级卷积网络结合迁移学习与微调可构建有效的实时停止意图脑机接口，显著降低误报，在健康受试者与脊髓损伤患者中均取得可接受的准确率，同时减少校准时间。

创新点：提出串联双卷积网络架构抑制误报；引入迁移学习与个体微调缩短校准；在外骨骼闭环环境及脊髓损伤患者中完成实时验证。

后续建议：扩大健康受试者与患者样本量以提升模型泛化能力；引入注意力机制或时空图网络进一步提升检测准确率；结合肌电或惯性信号做多模态融合以降低误报；研究长期使用中的模型自适应与在线增量学习；优化电极数量以实现更轻量化的日常可穿戴系统

APA 引文：Quiles, V., Ferrero, L., Iáñez, E., Ortiz, M., Gil-Agudo, Á., & Azorín, J. M. (2023). Brain-machine interface based on transfer-learning for detecting the appearance of obstacles during exoskeleton-assisted walking. Frontiers in Neuroscience, 17, 1154480. https://doi.org/10.3389/fnins.2023.1154480

仪器引用：The EEG signal was recorded at 500Hz and amplified using the actiCHamp (Brain Products GmbH, Germany). ... a hardware notch filter at 50Hz ... hardware also filtered the signal over 0.1Hz to mitigate the DC component.

摘要 · 中文整理：针对应用于多种临床人群的即兴音乐治疗，其潜在神经动力学的真实场景研究极为缺乏。本单案例研究探讨了一次音乐治疗过程中痴呆患者（PwD）与音乐治疗师之间音乐特征及脑内与脑间耦合的会话内差异。记录了音乐治疗师与一名痴呆患者（男性，31 岁）的双人脑电。从音视频数据中提取音符密度、脉冲清晰度与同步性。三名音乐治疗师在两段鼓即兴中识别出兴趣时刻与非兴趣时刻（MOI/MONI）。比较了 MOI 与 MONI 之间反映时滞神经同步的整合耦合指数及音乐特征。MOI 期间 2 Hz 活动的脑间耦合增强，并显示治疗师神经活动的领先性。患者在 MOI 期间来自额区与中央区的脑内耦合更强，而治疗师的脑内耦合在 MONI 期间更强。音乐特征差异表明二者在 MOI 期间的音乐行为更为相似。神经同步与音乐特征的会话内差异凸显了音乐治疗的动态性。研究结果有助于更好地理解真实音乐治疗过程中特定时刻的大脑社会与情感加工及（互动性）音乐行为，可能为此类时刻在关系治疗过程中的作用提供启示。

原文摘要：Real-life research into the underlying neural dynamics of improvisational music therapy, used with various clinical populations, is largely lacking. This single case study explored within-session differences in musical features and in within- and between-brain coupling between a Person with Dementia (PwD) and a music therapist during a music therapy session. Dual-EEG from a music therapist and a PwD (male, 31 years) was recorded. Note density, pulse clarity and synchronicity were extracted from audio-visual data. Three music therapists identified moments of interest and no interest (MOI/MONI) in two drum improvisations. The Integrative Coupling Index, reflecting time-lagged neural synchronization, and musical features were compared between the MOI and MONI. Between-brain coupling of 2 Hz activity was increased during the MOI, showing anteriority of the therapist's neural activity. Within-brain coupling for the PwD was stronger from frontal and central areas during the MOI, but within-brain coupling for the therapist was stronger during MONI. Differences in musical features indicated that both acted musically more similar to one another during the MOI. Within-session differences in neural synchronization and musical features highlight the dynamic nature of music therapy. The findings contribute to a better understanding of social and affective processes in the brain and (interactive) musical behaviors during specific moments in a real-life music therapy session. This may provide insights into the role of such moments for relational-therapeutic processes.

关键词：音乐治疗、即兴、超扫描、脑电图、音乐信息检索

研究目标：探究真实即兴音乐治疗场景中痴呆患者与治疗师在治疗性兴趣时刻与非兴趣时刻之间脑内、脑间神经同步及音乐互动特征的差异。

主要结论：兴趣时刻期间 2 Hz 脑间耦合增强且治疗师活动领先，患者额中央区脑内耦合更强，音乐行为更为相似，提示神经同步动态支持音乐治疗中的关系过程。

创新点：在真实治疗场景中结合 MOVE 无线 actiCHamp 双人脑电超扫描、音乐信息检索特征与治疗师标注的兴趣时刻，采用整合耦合指数刻画时滞神经同步。

后续建议：扩展至多名痴呆患者的群体研究以验证单案例结论；增加治疗次数进行纵向追踪以考察疗效相关的神经同步变化；引入自动化兴趣时刻识别以减少人工标注主观性；结合生理信号（心率、皮电）进行多模态同步分析；采用源空间分析减少容积传导对脑间耦合估计的影响

APA 引文：Maidhof, C., Müller, V., Lartillot, O., Agres, K., Bloska, J., Asano, R., Odell-Miller, H., & Fachner, J. (2023). Intra- and inter-brain coupling and activity dynamics during improvisational music therapy with a person with dementia: an explorative EEG-hyperscanning single case study. Frontiers in Psychology, 14, 1155732. https://doi.org/10.3389/fpsyg.2023.1155732

仪器引用：The EEG was recorded with 500 Hz from 32 Ag/AgCl active electrodes (ActiCap, Brain Products GmbH, Germany) for each participant, placed according to the extended 10–20 system ... referenced to Cz of participant 1. To enable free movements of the participants, we used the MOVE interface in combination with an ActiCHamp amplifier (Brain Products GmbH, Gilching, Germany). This system enables wireless EEG data transmission by connecting electrode cables to transmitters that were attached to the shoulders/back of the participants.