Abstract:P300 as an effective biomarker to index attention and memory has been widely used for brain-computer interface, cognitive evaluation, and clinical diagnosis. To evoke clear P300, an oddball paradigm consisting of two types of stimuli, i.e., infrequent target stimuli and frequent standard stimuli, is usually used. However, to simply and quickly explore the P300-related process, previous studies predominately focused on the target condition but ignored the fusion of target and standard conditions, as well as the difference of brain networks between them. Therefore, in this study, we used the hidden Markov model to investigate the fused multi-conditional electroencephalogram dataset of P300, aiming to effectively identify the underlying brain networks and explore the difference between conditions. Specifically, the inferred networks, including their transition sequences and spatial distributions, were scrutinized first. Then, we found that the difference between target and standard conditions was mainly concentrated in two phases. One was the stimulation phase that mainly related to the cortical activities of the postcentral gyrus and superior parietal lobule, and the other corresponded to the response phase that involved the activities of superior and medial frontal gyri. This might be attributed to distinct cognitive functions, as the stimulation phase is associated with visual information integration whereas the response phase involves stimulus discrimination and behavior control. Taken together, the current work explored dynamic networks underlying the P300-related process and provided a complementary understanding of distinct P300 conditions, which may contribute to the design of P300-related brain-machine systems.

Dynamic Networks of P300-related Process