Department of Marketing Management, Rotterdam School of Management, Erasmus University Rotterdam, the Netherlands.
Department of Marketing Management, Rotterdam School of Management, Erasmus University Rotterdam, the Netherlands.
Neuroimage. 2020 Aug 1;216:116618. doi: 10.1016/j.neuroimage.2020.116618. Epub 2020 Feb 7.
This study explored the feasibility of using shared neural patterns from brief affective episodes (viewing affective pictures) to decode extended, dynamic affective sequences in a naturalistic experience (watching movie-trailers). Twenty-eight participants viewed pictures from the International Affective Picture System (IAPS) and, in a separate session, watched various movie-trailers. We first located voxels at bilateral occipital cortex (LOC) responsive to affective picture categories by GLM analysis, then performed between-subject hyperalignment on the LOC voxels based on their responses during movie-trailer watching. After hyperalignment, we trained between-subject machine learning classifiers on the affective pictures, and used the classifiers to decode affective states of an out-of-sample participant both during picture viewing and during movie-trailer watching. Within participants, neural classifiers identified valence and arousal categories of pictures, and tracked self-reported valence and arousal during video watching. In aggregate, neural classifiers produced valence and arousal time series that tracked the dynamic ratings of the movie-trailers obtained from a separate sample. Our findings provide further support for the possibility of using pre-trained neural representations to decode dynamic affective responses during a naturalistic experience.
本研究探索了利用短暂情感片段（观看情感图片）中的共享神经模式来解码自然体验中扩展的动态情感序列（观看电影预告片）的可行性。二十八名参与者观看了国际情感图片系统（IAPS）中的图片，并在另一个单独的会议中观看了各种电影预告片。我们首先通过 GLM 分析定位了双侧枕叶皮层（LOC）对情感图片类别的反应体素，然后根据参与者在观看电影预告片时的反应对 LOC 体素进行了主体间超对齐。超对齐后，我们在情感图片上训练了主体间机器学习分类器，并使用分类器在参与者观看图片和观看电影预告片时对一个样本外参与者的情感状态进行解码。在参与者内部，神经分类器识别了图片的效价和唤醒类别，并跟踪了视频观看期间的自我报告效价和唤醒。总的来说，神经分类器生成的效价和唤醒时间序列跟踪了从另一个样本中获得的电影预告片的动态评分。我们的研究结果进一步支持了使用预训练的神经表示来解码自然体验中动态情感反应的可能性。
