Dartmouth College, Hanover, NH.
Bucknell University, Lewisburg, PA.
J Cogn Neurosci. 2022 Jul 1;34(8):1326-1339. doi: 10.1162/jocn_a_01858.
Notes in a musical scale convey different levels of stability or incompleteness, forming what is known as a tonal hierarchy. Levels of stability conveyed by these scale degrees are partly responsible for generating expectations as a melody proceeds, for emotions deriving from fulfillment (or not) of those expectations, and for judgments of overall melodic well-formedness. These functions can be extracted even during imagined music. We investigated whether patterns of neural activity in fMRI could be used to identify heard and imagined notes, and if patterns associated with heard notes could identify notes that were merely imagined. We presented trained musicians with the beginning of a scale (key and timbre were varied). The next note in the scale was either heard or imagined. A probe tone task assessed sensitivity to the tonal hierarchy, and state and trait measures of imagery were included as predictors. Multivoxel classification yielded above-chance results in primary auditory cortex (Heschl's gyrus) for heard scale-degree decoding. Imagined scale-degree decoding was successful in multiple cortical regions spanning bilateral superior temporal, inferior parietal, precentral, and inferior frontal areas. The right superior temporal gyrus yielded successful cross-decoding of heard-to-imagined scale-degree, indicating a shared pathway between tonal-hierarchy perception and imagery. Decoding in right and left superior temporal gyrus and right inferior frontal gyrus was more successful in people with more differentiated tonal hierarchies and in left inferior frontal gyrus among people with higher self-reported auditory imagery vividness, providing a link between behavioral traits and success of neural decoding. These results point to the neural specificity of imagined auditory experiences-even of such functional knowledge-but also document informative individual differences in the precision of that neural response.
音阶中的音符传达出不同程度的稳定性或不完整性，形成了所谓的音高层次。这些音阶传达的稳定程度部分负责产生旋律进行时的期望，以及这些期望的实现（或未实现）所产生的情感，并负责判断旋律整体的良好形成。即使在想象的音乐中，这些功能也可以被提取出来。我们研究了功能磁共振成像（fMRI）中的神经活动模式是否可以用于识别听到和想象的音符，以及与听到的音符相关的模式是否可以识别仅仅被想象的音符。我们向受过训练的音乐家展示了音阶的开头（音调和音色都有所变化）。音阶中的下一个音符要么是听到的，要么是想象的。探测音任务评估了对音高层次的敏感性，以及意象的状态和特质测量作为预测因子。多体素分类在初级听觉皮层（Heschl gyrus）中产生了高于随机的结果，用于解码听到的音阶程度。想象中的音阶程度解码在跨越双侧颞上、下顶叶、中央前回和额下回的多个皮质区域中是成功的。右侧颞上回成功地对听到的想象的音阶程度进行了交叉解码，表明音高层次感知和意象之间存在共享通路。右和左颞上回以及右额下回的解码在具有更分化的音高层次的人中和在具有更高自我报告的听觉意象生动性的人中的左额下回更成功，为行为特征与神经解码的成功之间提供了联系。这些结果表明想象中的听觉体验具有神经特异性，即使是这种功能知识，也记录了神经反应的精确性存在个体差异。
