School of Psychology, University of Sydney, NSW, 2006, Australia; ARC Centre of Excellence in Cognition and Its Disorders, NSW, 2109, Australia; Department of Cognitive Science, Macquarie University, NSW, 2109, Australia.
School of Psychology, University of Sydney, NSW, 2006, Australia; ARC Centre of Excellence in Cognition and Its Disorders, NSW, 2109, Australia; Department of Cognitive Science, Macquarie University, NSW, 2109, Australia.
Neuroimage. 2019 Aug 15;197:224-231. doi: 10.1016/j.neuroimage.2019.04.050. Epub 2019 Apr 19.
Rapid image presentations combined with time-resolved multivariate analysis methods of EEG or MEG (rapid-MVPA) offer unique potential in assessing the temporal limitations of the human visual system. Recent work has shown that multiple visual objects presented sequentially can be simultaneously decoded from M/EEG recordings. Interestingly, object representations reached higher stages of processing for slower image presentation rates compared to fast rates. This fast rate attenuation is probably caused by forward and backward masking from the other images in the stream. Two factors that are likely to influence masking during rapid streams are stimulus duration and stimulus onset asynchrony (SOA). Here, we disentangle these effects by studying the emerging neural representation of visual objects using rapid-MVPA while independently manipulating stimulus duration and SOA. Our results show that longer SOAs enhance the decodability of neural representations, regardless of stimulus presentation duration, suggesting that subsequent images act as effective backward masks. In contrast, image duration does not appear to have a graded influence on object representations. Interestingly, however, decodability was improved when there was a gap between subsequent images, indicating that an abrupt onset or offset of an image enhances its representation. Our study yields insight into the dynamics of object processing in rapid streams, paving the way for future work using this promising approach.
快速呈现图像并结合 EEG 或 MEG 的时分辨多元分析方法（快速-MVPA）在评估人类视觉系统的时间限制方面具有独特的潜力。最近的研究表明，可从 M/EEG 记录中同时解码顺序呈现的多个视觉对象。有趣的是，与快速呈现率相比，较慢的图像呈现率下的对象表示达到了更高的处理阶段。这种快速率衰减可能是由于流中的其他图像的前向和后向掩蔽引起的。在快速流中可能影响掩蔽的两个因素是刺激持续时间和刺激起始异步（SOA）。在这里，我们通过使用快速-MVPA 研究视觉对象的新兴神经表示来分离这些影响，同时独立地操纵刺激持续时间和 SOA。我们的结果表明，较长的 SOA 增强了神经表示的可解码性，而与刺激呈现持续时间无关，这表明后续图像充当有效的后向掩蔽。相比之下，图像持续时间似乎对对象表示没有分级影响。然而，有趣的是，当后续图像之间存在间隙时，可解码性得到了提高，这表明图像的突然出现或消失增强了其表示。我们的研究深入了解了快速流中对象处理的动态，为使用这种有前途的方法的未来工作铺平了道路。
