將注意力調控連結至聽覺注意力解碼：以固定時間結構之共置刺激為例

// Linking Attentional Modulation to Auditory Attention Decoding: Using Colocated Stimuli With a Fixed Temporal Structure

本研究評估現代聽覺注意解碼（AAD）方法對腦電圖（EEG）中注意神經指標的反映度｜三種解碼模型（前向、後向、卷積神經網路CNN）在4秒短時試驗中均優於機率，CNN模型表現最佳｜前向模型與經典注意事件相關電位（ERP）振幅關聯最強，強調結構化刺激有助於解碼結果連結注意力神經機制。

// This study tested how well modern auditory attention decoding (AAD) methods reflect established neural markers of attention in EEG signals during a two-stream listening task. Participants attended to one speech stream while EEG was recorded, and three decoding models (forward, backward, CNN) classified attention from short 4-second trials. All models performed above chance, with the CNN decoder achieving the highest accuracy, and decoding results correlated with classical attentional markers like ERP amplitudes. The forward model showed the strongest link to known attentional modulation, demonstrating how structured stimuli help connect decoding performance to neural mechanisms of attention.

Published 2026年7月15日