Recently, Professor Qiuhai Yue from the School of Psychology, Shenzhen University, together with collaborators, published a research article in Nature Communications (2023 Impact Factor = 14.7), a prestigious journal indexed in the Nature Index. The paper, titled “Ultrafast fMRI Reveals Serial Queuing of Information Processing During Multitasking in the Human Brain,” employed ultrafast, high-field functional magnetic resonance imaging (fMRI) to reveal that the human frontoparietal multiple-demand network and the motor cortex jointly constitute the serial processing bottleneck that limits multitasking in the human brain.Professor Yue is the first author of the paper, with Shenzhen University as the first-affiliated institution. He and Professor René Marois from Vanderbilt University are co-corresponding authors.
Research Introduction
The human brain contains over 87 billion neurons, each forming thousands of connections, giving rise to its remarkable parallel processing capacity. Nevertheless, influential cognitive models propose the existence of a central bottleneck—distinct from sensory encoding and motor response stages—that constrains our ability to perform two cognitively demanding tasks simultaneously, forcing information to be processed sequentially.
Due to technological limitations, previous research has not been able to clearly delineate the neural basis of this processing bottleneck—namely, when and where in the brain information processing shifts from a parallel to a serial queuing mode. The main challenge lies in the lack of neuroimaging techniques that combine both high spatial precision and high temporal resolution.
To address this gap, the present study employed ultrafast (199 ms) high-field (7T) functional magnetic resonance imaging (fMRI) together with multivariate pattern analysis (MVPA). This approach enabled the researchers to trace the flow of task-specific neural activity throughout the entire processing stream under dual-task overlap conditions—from stimulus perception to action execution.
Figure 1. Experimental design and behavioral results
The study employed the classic psychological refractory period (PRP) paradigm, in which participants performed two tasks: making an eye movement in response to different auditory stimuli and pressing a key with their fingers in response to different visual colors (Fig. 1a–c). Behaviorally, the researchers observed the classic PRP effect—under dual-task overlap (interference) conditions, responses to the second task were significantly delayed. This delay reflects the sequential processing mode of the brain when confronted with overlapping task demands (Fig. 1d, e).
By analyzing brain activation data from the single-task conditions, the researchers identified brain regions and networks associated with different stages of information processing. These included modality-specific sensory regions (auditory and visual areas), motor regions (eye-movement and finger-movement areas), as well as a multimodal central processing region. Notably, this central processing region overlapped substantially with the well-established multiple-demand network, which is known to support performance across a broad range of cognitively demanding tasks (Fig. 2).
Figure 2. Task-specific regions of interest (ROIs) at different processing stages
Figure 3. Neural decoding time series in the multiple-demand network (MDN) under single-task and dual-task conditions
Using multivariate pattern analysis (MVPA), the researchers tracked task-specific neural activity across different brain regions. They found that in primary sensory areas, task-related neural activity was not affected by task overlap. In other words, during the sensory processing stage, stimulus information was processed in a parallel manner regardless of dual-task interference.
In contrast, within the multiple-demand network, neural activity associated with the first stimulus in the dual-task condition was largely unaffected by the presence of a second task. However, the neural activity associated with the second stimulus was significantly delayed (Fig. 3a). Importantly, this delay in neural activity predicted the behavioral psychological refractory period effect—namely, the reaction time delay (Fig. 3b). These results indicate that, at the central processing stage, information processing shifts to a serial queuing mode.
Similarly, in motor response regions, the researchers observed a delay in neural activity for the second task, paralleling the pattern observed in the central processing stage.
Figure 4. Reaction time (RT)-based analysis under the single-task condition
At the same time, the study revealed that, contrary to conventional understanding, the primary motor cortex is not merely involved in the simple execution of movements. Instead, it also participates in response selection processing during the central processing stage. In specific tasks, the primary motor cortex and the multiple-demand network jointly constitute the central bottleneck of information processing in the human brain.Analyses conducted under the single-task condition further supported this conclusion. Both the examination of neural activity based on reaction time differences (Fig. 4) and the results of Granger causality analysis indicated that the primary motor cortex plays a role beyond action execution, contributing directly to the central bottleneck of human information processing.
In conclusion, the results clearly demonstrate that task information within the frontoparietal multiple-demand network is processed in a serial queuing mode. Moreover, the study found that this network, in combination with modality-specific motor regions, jointly defines the central bottleneck function at the stage of response selection. These findings provide direct neural evidence for serial queuing in information processing and precisely localize the neural substrates that support the central bottleneck.In addition, the study highlights that combining ultrafast imaging techniques with the high spatial resolution of fMRI offers a powerful new approach for understanding higher-level cognitive processes from a neurobiological perspective.
This research was supported by the Scientific Research Start-up Fund for Newly Recruited High-level and Urgently Needed Talents of Shenzhen.
