Hierarchical control in task switching: Electrophysiological evidence

Abstract

The task-switching paradigm is one of the leading research paradigms that is widely used to explore cognitive control. Previous studies have shown that switch costs are greater for high hierarchical tasks than for low hierarchical tasks, and a number of ERP studies on rule structure learning, rule switching, task complexity, and asymmetric task switching have coherently found that N2, P3, and late components are associated with the hierarchical control process. For example, Lu et al. (2017) designed three levels of tasks, but were not concerned with switch costs. Li et al. (2019) also designed three levels of tasks, but were concerned with asymmetric switch costs. The other two studies focused on stimulus or rule switching without concern for task switching. However, to date, no study has clearly addressed the ERP correlates of hierarchical effects in task-switching. A nested cue-task switching paradigm was used to investigate the brain responses associated with different hierarchical effects in task switching. Participants were asked to perform two hierarchical tasks. In the low hierarchical task, participants judged digits (1–9, except 5) as large/small or odd/even, respectively. In the high hierarchical task, participants identified the semantic features of the presented digits (e.g., whether the digit was an even number) before they performed the low hierarchical task (e.g., the large/small task). For example, participants first identified whether the current number was a large digit (i.e., greater than five) and then made an odd/even judgment on it. If the current number was not greater than five, then they did not respond (no-go trials). The proportion of no-go trials was 16%, and the no-go and subsequent go trials were excluded from dataanalysis. Thirty Chinese students (15 males) participated in the EEG experiment. They were asked to press the “F” key for odd or large numbers and the “J” key for even or small numbers. The links between the attributes of the cues and response keys were counterbalanced between participants. Behavioral results showed that the RT was longer for the high hierarchical trials than for the low hierarchical trials, indicating that the high hierarchical task was more complex than the low hierarchical task. Furthermore, there was a significant interaction between transition type and hierarchical level, with greater switch costs occurring in the high hierarchical task than in the low hierarchical task, indicating that switching in a low hierarchical task is easier than in a high hierarchical task. Cue-locked ERP results showed that the main effect of the hierarchical level was significant in P2, with higher P2 amplitudes for the high hierarchical trials than for the low hierarchical trials. A significant main effect of transition type was found in the CNV, with higher CNV amplitudes for the high hierarchical trials than for the low hierarchical trials, and there was a significant interaction between transition type and hierarchical level. Further analysis of this interaction revealed that task-switching trials elicited larger CNV amplitudes than task-repeating trials in the high hierarchical task, but not in the low hierarchical task. The target-locked ERP results showed that the main effect of transition type was significant for N2, P3, and SP. The difference in N2 and SP amplitudes between high hierarchical task switching and task repetition was significantly greater than between low hierarchical task switching and task repetition. The purpose of the present study was to explore the ERP correlates of hierarchical effects in task-switching. The behavioral results replicated previous findings. Cue-locked ERP results indicated that the hierarchical effect first appeared in the P2 component and that the switch effect on the CNV component was modulated by the task hierarchy, reflecting more selective attention given to high hierarchical tasks and higher proactive control during the task-set reconfiguration stage. The target-locked ERP results indicated that task switching induced more negative N2 amplitudes and smaller P3 and SP amplitudes compared to task repetition. The difference wave amplitudes between high hierarchical task switching and repetition were significantly greater for the N2 and SP amplitudes than for the low hierarchical task, reflecting that the process of inhibiting the old task-set and reconfiguring the new response set is more complex, resulting in increased reactive control. These findings provide new evidence for the task-set reconfiguration theory and the hierarchical nature of cognitive control.