State Key Laboratory of Media Convergence and Communication, Communication University of China, Beijing, China.
Neuroscience and Intelligent Media Institute, Communication University of China, Beijing, China.
Psych J. 2023 Jun;12(3):421-429. doi: 10.1002/pchj.645. Epub 2023 Apr 25.
Accurately predicting whether a short video will be liked by viewers is a topic of interest to media researchers. This study used an electroencephalogram (EEG) to record neural activity in 109 participants as they watched short videos (16 clips per person) to see which neural signals reflected viewers' preferences. The results showed that, compared with the short videos they disliked, individuals would experience positive emotions [indexed by a higher theta power, lower (beta - theta)/(beta + theta) score], more relaxed states (indexed by a lower beta power), lower levels of mental engagement and alertness [indexed by a lower beta/(alpha + theta) score], and devote more attention (indexed by lower alpha/theta) when watching short videos they liked. We further used artificial neural networks to classify the neural signals of different preferences induced by short videos. The classification accuracy was the highest when using data from bands over the whole brain, which was 75.78%. These results may indicate the potential of EEG measurement to evaluate the subjective preferences of individuals for short videos.
准确预测短视频是否会受到观众喜爱，是媒体研究人员感兴趣的话题。本研究使用脑电图（EEG）记录了 109 名参与者观看短视频时的神经活动（每人 16 个片段），以观察哪些神经信号反映了观众的偏好。结果表明，与不喜欢的短视频相比，个体在观看喜欢的短视频时会产生积极的情绪[由较高的θ功率、较低的（β-θ）/（β+θ）分数表示]，更放松的状态[由较低的β功率表示]，较低的精神投入和警觉性[由较低的β/（α+θ）分数表示]，并投入更多注意力[由较低的α/θ分数表示]。我们进一步使用人工神经网络对不同偏好的短视频引起的神经信号进行分类。当使用全脑频段的数据时，分类准确率最高，为 75.78%。这些结果可能表明 EEG 测量在评估个体对短视频的主观偏好方面具有潜力。
