Abstract:

Temporal error monitoring (TEM) refers to the metacognitive process through which individuals detect, evaluate, and regulate deviations in their own temporal judgments. It enables individuals not only to assess whether a timing response is accurate, but also to use internal and external cues to improve later behavior. Although previous studies show that humans can monitor timing errors even without explicit feedback, current findings remain fragmented across behavioral, computational, and neurophysiological levels. The present review therefore examines the metacognitive regulation of TEM and its neural basis, focusing on three mechanisms—self-evaluation, feedback-based regulation, and decision-confidence regulation—and, on this basis, proposes a two-stage, three-pathway integrative model.
The review first organizes the major paradigms used to study TEM. Opt-out tasks index implicit monitoring by asking participants to keep or discard a timing response without external feedback. Self-evaluation tasks assess explicit metacognitive access by requiring judgments of error direction, magnitude, or confidence after a temporal estimate. Reattempt-decision tasks focus on feedback-guided adjustment by examining whether participants choose to retry after receiving performance-related information. Together, these paradigms indicate that TEM is not a single ability, but a coordinated process involving error detection, evaluation, and regulation.
A first major claim of this review is that self-evaluation constitutes the foundational mechanism of TEM. Even without feedback, individuals can detect and correct temporal deviations on the basis of internal representations. This suggests that TEM depends on a metacognitive readout of self-generated timing signals rather than solely on external comparison. Computationally, this process can be understood within a sequential diffusion framework derived from the opposing Poisson drift-diffusion model. Temporal judgments arise from competition among noisy accumulators, and post-response comparisons among them provide information about whether a response was too early or too late, as well as the magnitude of deviation. This account shifts the explanation of temporal self-monitoring from a single internal clock to a distributed evidence-based mechanism.
A second major claim is that feedback-based regulation serves as an important but supplementary route for improving temporal judgments. External feedback provides an objective reference that can compensate for limitations in internal timing signals, yet different forms of feedback differ in effectiveness. Correct/incorrect, absolute-error, and signed-error feedback vary in informational value, with signed-error feedback appearing especially effective because it specifies both direction and magnitude. At the same time, feedback does not simply replace internal monitoring. Its influence depends on whether external information is consistent with the individual’s own evaluation of performance.
A third major claim concerns decision-confidence regulation. Confidence is treated not merely as a retrospective report, but as an estimate of the reliability of one’s temporal judgment. It does not directly encode error direction or magnitude. Instead, it regulates how strongly self-evaluation and feedback influence later decisions and corrections.
Another contribution of the review is the integration of neurophysiological findings. Beta activity is closely related to the generation and maintenance of internal temporal representations, whereas alpha activity is more strongly associated with the readout and evaluation of self-generated timing states. Alpha-beta coupling and coordination between the timing network and the default mode network further indicate that self-evaluation in TEM depends on distributed neural interactions. In feedback-based regulation, feedback-related negativity reflects rapid prediction-error processing, whereas the P3 component is more closely linked to the motivational and evaluative significance of feedback. Findings involving prefrontal and orbitofrontal regions further suggest that confidence has a distinct neural basis and may influence second-order monitoring even when first-order timing performance remains relatively intact.
Based on the above evidence, this review proposes a two-stage, three-pathway model of TEM. In the first stage, internal temporal error representations are formed through temporal encoding, memory, and comparison processes. These representations include both content information, such as error direction and magnitude, and quality-related information, such as uncertainty and representational reliability. In the second stage, three interactive pathways operate on this representation. The self-evaluation pathway extracts endogenous information about temporal deviation from internal representations; the feedback-based pathway uses accumulated feedback history to generate exogenous calibration signals; and the confidence-regulation pathway uses uncertainty-related cues to adjust parameters such as integration weight, learning rate, and decision threshold. The main innovation of this model is that it explicitly distinguishes the generation of temporal errors from their metacognitive regulation, while assigning confidence a modulatory role in the integration process.
In summary, this review reorganizes the TEM literature around three metacognitive regulatory mechanisms, integrates behavioral and neurophysiological evidence within a unified account, and proposes a novel two-stage, three-pathway model to explain how temporal errors are monitored and regulated. These contributions deepen current understanding of the metacognitive architecture of time perception and provide a theoretical basis for future research on development, ecological task design, and clinical application.

Key words: temporal error monitoring, time perception, metacognition, neural mechanisms