Deciphering whether and which mental processes are accessible for metacognitive judgments is a key question to understand higher cognitive functions. Paralleling the crucial role of Reaction times (RT) for unraveling the temporal sequence of mental processes, a comparable chronometric approach can be employed at the second-order level through introspective reaction times (iRT) measures. Although mean iRT correlate with mean RT, suggesting good metacognitive abilities, this would not necessarily imply a direct read-out of the duration of the underlying processes as participants may instead rely on inferences based on other salient, non temporal, cues.

In the present study, two experiments investigated information at the basis of iRT. In visual choice reaction time tasks, participants were asked to report their RT on a visual analog scale after each trial. Thanks to linear regression analyses, we could evidence that trial-by-trial RT and iRT were strongly correlated, indicating a good read-out of RT duration, but also that subjective evaluation was systematically biased by some experimental conditions. In addition, with electromyographic recordings, each single trial RT could be fractionated into premotor and motor times, allowing to investigate the relative contribution of each sub-process to iRT. This revealed that participants access both decision and motor execution durations. Results show that participants can access the duration of their mental processes, but that this read-out can be biased by non-temporal cues. The proposed methodology allows to dissociate the two.