|
How does action influence metacognition? — An exploration based on cognitive models and neural mechanisms
CHENG Xiaorong, QIU Shiming, DING Xianfeng, FAN Zhao
2025, 33 (3):
425-438.
doi: 10.3724/SP.J.1042.2025.0425
Action and metacognition are essential components of cognitive processing. Metacognition reflects an individual's ability to represent, monitor, and regulate cognitive processes, and can be measured through confidence ratings. Actions, on the other hand, serve as critical means for outputting internal cognitive processing, particularly decision-making information. Recent research has shown that various aspects of action—such as response speed, intensity, sequence, conflict, and observation—can influence metacognition. However, the mechanisms through which action influences metacognition remain unclear, especially in terms of the dependency, relationship, and organizational structure between the information used for metacognitive evaluation (i.e., metacognitive evidence) and that used for perceptual decision-making (i.e., perceptual evidence) during this process. From the perspective of cognitive models, post-decision models of metacognition—such as the hierarchical model, the metacognitive Bayesian model, and the Signal Detection Theory (SDT)-based generative model—offer a compelling framework for explaining the empirical findings regarding the influence of action on metacognition. These models propose that the sensory-motor information provided by action impacts metacognition after, rather than before, perceptual decision-making. Even after the perceptual decision has been made, metacognitive evidence continues to accumulate dynamically until the metacognitive evaluation is completed. Notably, while all post-decision models assert that metacognitive evidence differs from perceptual evidence, they vary in how the two types of evidence are related. First, the hierarchical model suggests that perceptual evidence and metacognitive evidence may originate from the same source but differ in quality. More specifically, the processing approach is hierarchical—different types of processing for the same evidence are organized hierarchically, with new processing introducing additional noise (e.g., action-related information), which alters the hierarchical structure of the evidence. Additionally, the processing approach is sequential—early processing results in an objective decision, while later processing inherits the evidence from early stage, evaluates it, and generates a subjective decision. Second, the metacognitive Bayesian model proposes that perceptual decision-making and metacognitive elevation have distinct yet coupled sources of evidence. According to this model, action responses reflect an individual's internal state, which, together with other confidence-related evidence, contributes to the formation of metacognitive elevation. Third, within the framework of the SDT-based generative model, metacognitive elevation is influenced by a trade-off between confidence noise and confidence boost. Confidence noise represents inefficiencies in the metacognitive elevation, which reduces confidence sensitivity. Confidence boost refers to post-perception information (e.g., action information) that is not used in the perceptual decision-making but is utilized in metacognitive elevation. The trade-off between these two factors determines the level of confidence: when the confidence boost is high and noise is low, confidence is high; conversely, when the boost is low and noise is high, confidence is low. From the perspective of neural mechanisms, the ventral stream and dorsal stream are responsible for visual perception (e.g., object feature recognition) and visually guided motor responses (e.g., object grasping), respectively. Notably, since both streams are connected to the prefrontal cortex, perceptual information from the ventral stream and action information from the dorsal stream may be integrated and processed there, forming metacognitive elevation. This integration provides a neural basis for the influence of action on metacognition. More specifically, perceptual and action information may be integrated through brain networks centered on the prefrontal cortex, relying on electrophysiological mechanisms such as β oscillations and α inhibition, with metacognition shaped by attentional regulation. While the cognitive models and neural mechanisms summarized above provide valuable insights into how action influences metacognition, further research could deepen our understanding of this influence by exploring the following directions. First, exploring the boundary conditions under which action influences metacognition. Some studies have shown that actions do not necessarily affect metacognition in all situations, suggesting that influence of action on metacognition may be subject to specific conditions that warrant further exploration. Second, clarifying the true psychological meaning reflected by confidence. One perspective suggests that perceptual confidence reflects the likelihood that a decision is correct, while another argues that confidence reflects the amount of evidence supporting a decision, with relatively less sensitivity to the evidence against it. Distinguishing between these two perspectives can help us better understand which aspect of metacognition is altered by action. Finally, exploring metacognitive performance in individuals with movement disorders, such as those with autism or paraplegic spinal cord injuries. Comparing metacognitive performance in the presence and absence of atypical actions or movement disorders can help identify the necessary conditions and specific mechanisms through which action impacts metacognition.
References |
Related Articles |
Metrics
|