Dissecting Dynamic Inter-Network Relationships During Attention

Post by D. Chloe Chung

What's the science?

When we switch our minds from being restful to being attentive, different neural networks in our brain act in different ways. The dorsal attention network (DAN) and the salience network (SN) are known to be activated during attention-requiring tasks, while another network called the default mode network (DMN) behaves in the opposite direction - its activity is suppressed. Although previous studies using functional magnetic resonance imaging (fMRI) have reported this negative correlation between DMN and DAN/SN network activity, there is a lack of our knowledge on how this “anti-correlation” is relevant to human behaviors. This week in Nature Communications, Kucyi and colleagues present clear evidence that each network has a distinct response profile to attention-requiring tasks and dynamic relationships among these networks are closely related to the efficient switch between attention and rest.

How did they do it?

While fMRI offers important information on the activity of brain regions based on changes in the blood oxygen level, the authors chose to use intracranial electroencephalography (iEEG) that places electrodes directly on the surface of the brains of patients undergoing surgery. This way, the authors were able to record the electrical activity of the brain regions that constitute attention networks – DAN, SN, or DMN – with much higher temporal and anatomical resolution. The authors obtained iEEG data from more than 3500 sites across 31 human participants who were performing the attention-evaluating test, which shows images that gradually and continuously change every 800 milliseconds. In this test, participants were asked to specifically respond when images of city scenes presented to them changed to different city scenes, but not to respond when they changed to mountain scenes. When the participants successfully responded to image changes, their responses were categorized as “correct”, but otherwise, the responses were considered to be “incorrect”. During the iEEG recording, the authors measured the electrical activity within the high-frequency broadband that ranges from 70 to 170 Hz, as this activity range has been shown to well-represent the negative correlation between attention networks.

What did they find?

From the iEEG recording, the authors first detected an increase in high-frequency broadband activity from the brain regions that constitute the DAN/SN and a decrease in high-frequency broadband activity from the regions constituting the DMN. All of these changes occurred several hundred milliseconds after the image change during the attention-evaluating test and returned to baseline after 1 to 2 seconds. These changes in activity confirmed that the DAN/SN are activated while the DMN is deactivated when human participants paid attention to external stimuli. When they took a closer look at the precise timing of when the activity peaked in each network, the authors found that the DAN was the first to peak in its activity, followed by the SN, and then the DMN. This observation of a unique timing profile for each network suggests that there is a clear temporal lag in electrical activity across attention networks during attention-requiring tasks. By calculating the correlation between this temporal lag in the activity of attention networks and the accuracy of attention-task performance, the authors showed that the lagged anti-correlation between DAN and DMN was especially important for performance on the attention task. In addition to these findings, the authors found that when human subjects failed to correctly perform the attention-requiring tasks, activities of the DAN/ SN were noticeably elevated while the activity of DMN was not sufficiently suppressed. This interesting finding further supports the significance of anti-correlation among attention networks in accurate attentional performance.

What’s the impact?

This study is the first to show the significance of delays in the timing of activity among the DAN, SN and DMN networks in attention. Powered by a large group of human participants who had electrodes directly implanted within the brain areas that represent these different attention networks, this study provides findings that will be valuable in critically interpreting neuroimaging studies that investigate the brain states between rest and active tasks. What we learned from this study will also serve as a foundation for subsequent research on how these dynamic inter-network relationships can change as we age or develop neurological disorders.

Kucyi et al. Electrophysiological dynamics of antagonistic brain networks reflect attentional fluctuations. Nature Communications (2020). Access the original scientific publication here.