Neural Correlates of Metacognitive Judgment in Perceptual Decision Tasks

Abstract

Metacognitive judgments rely on a distinct neural circuit that monitors the quality of primary perceptual decisions. Using event-related fMRI at 3T, we investigated 42 participants performing a motion discrimination task with confidence ratings. Dorsolateral prefrontal cortex and anterior insula showed parametric increases in activation with confidence levels, independent of decision accuracy. These findings suggest a dissociable neural substrate for metacognitive monitoring.

Introduction

The ability to accurately monitor one's own cognitive performance—metacognition—remains a fundamental puzzle in cognitive neuroscience. While error-monitoring processes have been localized to anterior cingulate cortex and lateral prefrontal regions, the neural mechanisms underlying conscious confidence judgments are poorly understood. Existing lesion and animal studies suggest that prefrontal structures are critical for self-assessment, yet the precise circuits remain unspecified.

The present study used event-related functional magnetic resonance imaging (fMRI) to identify neural correlates of confidence during perceptual decision-making. We hypothesized that distinct regions in prefrontal cortex would track subjective confidence independently of stimulus properties and objective performance. Understanding these mechanisms has implications for models of consciousness, learning, and decision-making under uncertainty.

Method

Participants

Forty-two right-handed participants (mean age 24.3 years, range 19–34; 22 female) with normal or corrected-to-normal vision and no history of neurological disorder were recruited from the local community. Informed consent was obtained in accordance with the institutional review board. Participants were paid $25 CAD for their participation.

Procedure

Participants performed a random dot motion discrimination task while undergoing continuous fMRI scanning on a 3T Siemens Magnetom. On each trial, a circular field of dots (300 dots, 5° visual angle) moved coherently in one of eight directions for 1 second. Coherence levels ranged from 10% to 80%, spanning perceptual difficulty. After each stimulus, participants used a four-button response box to indicate motion direction and then rate their confidence on a four-point scale (1 = low, 4 = high). Each session contained 192 trials. Stimuli were generated with MATLAB and projected via Epson EMP-7850 LCD projector. fMRI data were acquired in 3 × 3 × 3 mm voxels using a T2*-weighted echo-planar sequence (TR = 2 s, TE = 30 ms).

Results

Behavioral performance showed the expected coherence-dependent accuracy curve (F(7,287) = 94.2, p < .001), with mean accuracy ranging from 52% at 10% coherence to 96% at 80% coherence. Confidence ratings and decision accuracy were significantly correlated (r = .62, p < .001), but confidence continued to increase with coherence even when accuracy was held constant (F(3,165) = 18.5, p < .001), indicating that confidence is influenced by stimulus strength beyond error signals.

fMRI activation maps (whole-brain analysis, p < .001 uncorrected, cluster threshold 10 voxels) revealed that right dorsolateral prefrontal cortex (dlPFC; peak MNI coordinates 42, 32, 26) showed a parametric increase in activation as a function of rated confidence (linear trend, t(41) = 4.8, p < .001). Right anterior insular cortex (peak 36, 26, 2) also displayed confidence-dependent activation (t(41) = 4.1, p < .001). These effects remained significant when decision accuracy was entered as a covariate, indicating that dlPFC and insula track subjective confidence independent of objective performance. Posterior cingulate cortex (peak -2, -54, 18) showed the reverse pattern, with greater activation for low-confidence trials (t(41) = -3.5, p < .001).

Discussion

These findings demonstrate a dissociable neural circuit for metacognitive confidence monitoring. The parametric relationship between dlPFC activation and subjective confidence suggests that this region integrates multi-sensory evidence to compute an internal signal of decision reliability. Prior work implicates dlPFC in working memory and executive control; our results extend this literature by showing that dlPFC also supports the representation of confidence in perceptual decisions. The anterior insula's role in confidence aligns with recent evidence linking insular cortex to subjective bodily states and interoceptive awareness, potentially providing a pathway by which decision confidence becomes phenomenologically accessible.

The double dissociation between regions supporting confidence (dlPFC, anterior insula) and those associated with reward processing and default-mode activity (posterior cingulate) suggests that metacognitive monitoring engages a distinct functional system. Future work should investigate whether individual differences in dlPFC structure predict metacognitive accuracy across tasks and whether this neural circuit is disrupted in conditions involving impaired insight or self-awareness. Furthermore, longitudinal designs examining how confidence computations change with practice may reveal the learning dynamics of metacognition.

References

Crick, F., & Koch, C. (1998). Constraints on cortical and thalamic projections: The no-strong-loops rule. Nature, 391(6664), 245–250.
Ferrera, V. P., Yanike, M., & Cassanello, C. R. (2009). Frontal eye field neurons signal changes in decision criteria. Nature Neuroscience, 12(11), 1458–1462.
Friston, K. J., Holmes, A. P., Poline, J. B., Grasby, P. M., Williams, S. C., Frackowiak, R. S., & Turner, R. (1995). Analysis of fMRI time-series revisited. NeuroImage, 2(1), 45–53.
Haynes, J. D., & Rees, G. (2006). Decoding mental states from brain activity in humans. Nature Reviews Neuroscience, 7(7), 523–534.
Kiani, R., & Shadlen, M. N. (2009). Representation of confidence associated with a decision by neurons in the parietal cortex. Science, 324(5929), 759–764.
Luu, P., Tucker, D. M., Derryberry, D., Reed, M., & Poulsen, C. (2003). Electrophysiological responses to errors and feedback in the process of action regulation. Psychological Science, 14(1), 47–53.
Worsley, K. J., & Friston, K. J. (1995). Analysis of fMRI time-series revisited—again. NeuroImage, 2(3), 173–181.