• Research Progress
  • On May 3, PLOS Biology published a research article entitled “The neural system of metacognition accompanying decision-making in the prefrontal cortex” from Xiaohong Wan’s group in State Key Laboratory of Cognitive Neuroscience and Learning, IDG/McGovern Brain Research Institute.
    To introspectively evaluate and revise our own behaviors, cognitions and thoughts is an important approach to improving. We often rely on the feedback from the environments or others to correct our behaviors. It is very useful to improve by self-introspection. The process of monitoring and control of our own cognitive processes is so called “metacognition”, that is, cognition about cognition. The concept and strategies of metacognition have been exploited in development educations. However, it is very limited to understand the neural mechanisms of metacognition. Very recently, it has begun to investigate it in neuroscience. Most of these research works investigated the neural presentations of confidence about perceptual or mnemenic decision-making. For instance, you are very confident that the presented human face or word that you have been just seen, or you feel very uncertain about whether you have seen it or not. The previous research works have found that the neural activities in many brain regions correlated with the confidence levels, even in rats and monkeys. Since the information for perceptual decision-making is also the source of confidence ratings. Hence, some studies have proposed that the decision-making and metacognition processes are coincidental, just like two sides of a coin. However, it has been often observed that the subjective confidence ratings are diverged from the actual performance accuracy. Furthermore, intervention on the brain regions of the prefrontal cortex impaired the confidence ratings, but not the decisions per se. These evidences indicate that metacognition is associated with the prefrontal cortex, separated from those associated with perceptual and mnemenic decision-making. Hence, the neural mechanisms of metacogniton remain to date controversial.
    All of the previous studies used a single decision-making and confidence-rating paradigm. Because the two processes are highly correlated, it is difficult to separate the two processes. In this new study, a novel experimental paradigm-“decision-redecision’ was designed to separate the two processes. After the subjects made the initial decision-making and confidence rating, they were required to make the decision-making and confidence rating again on the same problem. In contrast to the process of initial decision-making, the second decision-making process was more dependent on metacognition, as it is based on the initial decision-making to make revisions. Thus, distinction of the two decision-making processes should isolate the metacognition neural process from the decision-making neural process. Their brains were scanned by magnetic resonance imaging (MRI) during the subjects conducted the perceptual judgment on the random-dot-motion direction or solving the Sudoku puzzle (Figure 1). The brain activities cause signal changes of MRI images of the brain. The brain regions in the prefrontal cortex became much extensively involved during the second decision-making, and the regional neural activities were correlated with the confidence rating on the initial decision-making. The more uncertain about the preceding decision-making, the stronger these neural activities became. The activated regions were identical between the two tasks. Furthermore, the neural activities in the dorsal anterior cingulate cortex (dACC) also correlated with each individual uncertainty sensitivity, an aspect of metacognition abilities. In contrast the neural activities in the frontopolar cortex (FPC) correlated with each individual accuracy improvement, another aspect of metacognition abilities. Thus, the evidence indicates that the functions of metacognition monitoring and metacognition control are dissociated in two separate brain regions.    

    Figure 1. The perceptual decision-making task on judgments of random-dot-motion direction (left) and the rule-based decision-making task on solving the Sudoku puzzles (right).
    Utilizing a new experimental paradigm, the research study successfully separated the neural system of metacognition from the neural system of decision-making, and further separated the subsystem of metacognition monitoring and that of metacognition control. This study is the first time to systematically illustrate the metacognition neural architecture (Figure 2). Further understanding of the neural mechanisms of metacognition will help improve the process of learning in education, reveal the bases of the mental disorders that fail to properly monitor and control own behaviors and thoughts, such as Schizophrenia, finally inspire to fulfill self-conscious artificial intelligence.

    Figure 2. The scheme of the neural architecture of metacognition.

    This study is contributed by Ph.D students of Lirong Qiu, Jie Su and Yinmei Ni (who share the first authors) from Xiaohong Wan’s group, and Dr. Yang Bai and Professor Xiaoli Li, also from State Key Laboratory of Cognitive Neuroscience and Learning. This project was funded by NSFC (No. 31471068) and MOST (No. 2016YFE0129100).
    Lirong Qiu, Jie Su, Yinmei Ni, Yang Bai, Xuesong Zhang, Xiaoli Li, Xiaohong Wan. The neural system of metacognition accompanying decision-making in the prefrontal cortex. PLOS Biology, 2018 16(4):e2004037