Electroencephalographic responses to unexpected musical events allow researchers to test listeners’ internal models of syntax. One major challenge is dissociating cognitive syntactic violations—based on the abstract identity of a particular musical structure—from unexpected acoustic features. Despite careful controls in past studies, recent work by Bigand, Delbe, Poulin-Carronnat, Leman, and Tillmann (2014) has argued that ERP findings attributed to cognitive surprisal cannot be unequivocally separated from sensory surprisal. Here we report a novel EEG paradigm that uses three auditory short-term memory models and one cognitive model to predict surprisal as indexed by several ERP components (ERAN, N5, P600, and P3a), directly comparing sensory and cognitive contributions. Our paradigm parameterizes a large set of stimuli rather than using categorically “high” and “low” surprisal conditions, addressing issues with past work in which participants may learn where to expect violations and may be biased by local context. The cognitive model (Harrison & Pearce, 2018) predicted higher P3a amplitudes, as did Leman’s (2000) model, indicating both sensory and cognitive contributions to expectation violation. However, no model predicted ERAN, N5, or P600 amplitudes, raising questions about whether traditional interpretations of these ERP components generalize to broader collections of stimuli or rather are limited to less naturalistic stimuli.

中文翻译：

---

使用流行音乐和弦序列重新评估与语法相关的 ERP 组件

对意外音乐事件的脑电图反应使研究人员能够测试听众的内部句法模型。一个主要挑战是将基于特定音乐结构的抽象身份的认知句法违规与意想不到的声学特征分开。尽管在过去的研究中进行了仔细的控制，但 Bigand、Delbe、Poulin-Carronnat、Leman 和 Tillmann（2014）最近的工作认为，归因于认知惊奇的 ERP 发现不能明确地与感觉惊奇分开。在这里，我们报告了一种新的 EEG 范式，它使用三个听觉短期记忆模型和一个认知模型来预测由几个 ERP 组件（ERAN、N5、P600 和 P3a）索引的惊喜，直接比较感官和认知贡献。我们的范式将一大组刺激参数化，而不是使用绝对“高”和“低”的意外条件，解决过去工作的问题，在这些问题中，参与者可能会知道在哪里会出现违规行为，并且可能会受到当地环境的影响。认知模型 (Harrison & Pearce, 2018) 预测更高的 P3a 振幅，Leman (2000) 模型也是如此，表明感官和认知对违反期望的贡献。然而，没有模型预测 ERAN、N5 或 P600 幅度，这引发了对这些 ERP 组件的传统解释是否可以推广到更广泛的刺激集合，或者更确切地说仅限于不太自然的刺激的问题。认知模型 (Harrison & Pearce, 2018) 预测更高的 P3a 振幅，Leman (2000) 模型也是如此，表明感官和认知对违反期望的贡献。然而，没有模型预测 ERAN、N5 或 P600 幅度，这引发了对这些 ERP 组件的传统解释是否可以推广到更广泛的刺激集合，或者更确切地说仅限于不太自然的刺激的问题。认知模型 (Harrison & Pearce, 2018) 预测更高的 P3a 振幅，Leman (2000) 模型也是如此，表明感官和认知对违反期望的贡献。然而，没有模型预测 ERAN、N5 或 P600 幅度，这引发了对这些 ERP 组件的传统解释是否可以推广到更广泛的刺激集合，或者更确切地说仅限于不太自然的刺激的问题。