Figure 4 - available via license: Creative Commons Attribution 4.0 International
Content may be subject to copyright.
Late ERP activity reliably reflects subjective confidence ratings, independently of accuracy and evidence discriminability, across cognitive domains. Grand-averaged stimuluslocked ERP waveforms at electrode P3 for high (red) and low (blue) confidence trials are shown for the A Perception task (day 1), B Knowledge task (day 1), C Perception task (day 2), and D
Source publication
Decision confidence, an internal estimate of how accurate our choices are, is essential for metacognitive self-evaluation and guides behaviour. However, it can be suboptimal and hence understanding the underlying neurocomputational mechanisms is crucial. To do so, it is essential to establish the extent to which both behavioural and neurophysiologi...
Contexts in source publication
Context 1
... used cluster-based permutation analysis (Maris & Oostenveld, 2007) to identify systematic relationships between single-trial ERP amplitude and confidence ratings across all electrodes and post-stimulus time points (0-1s relative to stimulus presentation). Whilst controlling for difficulty level and 1 st -order accuracy, we observed significant relationships between a late component of the evoked potential and confidence ratings in both tasks which replicated across both experimental sessions (Figure 4). Late ERP activity reliably reflects subjective confidence ratings, independently of accuracy and evidence discriminability, across cognitive domains. ...
Context 2
... the perceptual task on day 1, we found that two significant clusters predicted confidence ( Figure 4A): one positive (cluster statistic = 17,693, p = .0015; spanning ~490-1000ms post-stimulus over centroparietal/occipital electrodes with a left parietal maximum (see topography in Figure 4A)) and one negative (cluster statistic = -18,845, p = .001; ...
Context 3
... the perceptual task on day 1, we found that two significant clusters predicted confidence ( Figure 4A): one positive (cluster statistic = 17,693, p = .0015; spanning ~490-1000ms post-stimulus over centroparietal/occipital electrodes with a left parietal maximum (see topography in Figure 4A)) and one negative (cluster statistic = -18,845, p = .001; spanning ~520-1000ms post-stimulus over frontal electrodes with a right maximum). ...
Context 4
... ~520-1000ms post-stimulus over frontal electrodes with a right maximum). We found two highly similar clusters that predicted confidence on the knowledge task on day 1 as well ( Figure 4B): one positive (cluster statistic = 11,124, p = .0025; ~490-960ms over centroparietal electrodes (see Figure 4B topography)) and one negative (cluster statistic = -9,953 p= .0015; ...
Context 5
... found two highly similar clusters that predicted confidence on the knowledge task on day 1 as well ( Figure 4B): one positive (cluster statistic = 11,124, p = .0025; ~490-960ms over centroparietal electrodes (see Figure 4B topography)) and one negative (cluster statistic = -9,953 p= .0015; ~470-890ms over right frontotemporal electrodes). ...
Context 6
... results were largely replicated in the second testing session (Figure 4C-D). On day two, we found a positive (cluster statistic = 14,267, p = .002, ...
Context 7
... results were largely replicated in the second testing session (Figure 4C-D). On day two, we found a positive (cluster statistic = 14,267, p = .002, beginning at ~530ms, over left centroparietal/occipital electrodes, Fig. 4C) and a negative (cluster statistic = -17,997, p = .0005, starting at ~530ms, over right frontotemporal electrodes) cluster in the perceptual task. In the knowledge task (day 2, Fig. 4D), we found a positive cluster (cluster statistic = 4,360, p = .0265, 1101 spanning ~640-810ms, over left centroparietal/occipital electrodes) and a ...
Context 8
... session (Figure 4C-D). On day two, we found a positive (cluster statistic = 14,267, p = .002, beginning at ~530ms, over left centroparietal/occipital electrodes, Fig. 4C) and a negative (cluster statistic = -17,997, p = .0005, starting at ~530ms, over right frontotemporal electrodes) cluster in the perceptual task. In the knowledge task (day 2, Fig. 4D), we found a positive cluster (cluster statistic = 4,360, p = .0265, 1101 spanning ~640-810ms, over left centroparietal/occipital electrodes) and a negative cluster (cluster statistic = -6,387, p = .0125, spanning ~480-930ms, over right frontotemporal ...
Citations
... Consequently, metacognitive ability has been correlated with other stable individual differences, such as brain structure [10][11][12][13] . While metacognitive ability is often assumed to be domain-general and rely on shared neural substrates, this question remains hotly debated [14][15][16][17] . The construct of metacognitive ability is also thought to be different from other constructs such as task skill or bias, so it is often desirable to find metrics of metacognitive ability unrelated to these other constructs 18 . ...
... Guggenmos 41 examined both the split-half reliability and the across-participant correlation between d' and several measures of metacognition (meta-d', M-Ratio, M-Diff, and AUC2) finding surprisingly low reliability and significant correlations with d' for all measures. Relatedly, Kopcanova et al. 14 examined the test-retest reliability of M-Ratio and also found low-reliability values. Another paper developed a new technique to examine dependence on metacognitive bias and found that meta-d' and M-Ratio are not independent of metacognitive bias 28 . ...
... Similar test-retest correlation coefficients were obtained when Pearson correlation was computed instead of ICC (Fig. 6). These results are in line with the findings of Kopcanova et al. 14 and suggest that correlations between measures of metacognition and measures that do not substantially fluctuate on a day-by-day basis (e.g., structural brain measures) are likely to be particularly noisy such that very large sample sizes may be needed to find reliable results. ...
One of the most important aspects of research on metacognition is the measurement of metacognitive ability. However, the properties of existing measures of metacognition have been mostly assumed rather than empirically established. Here I perform a comprehensive empirical assessment of 17 measures of metacognition. First, I develop a method of determining the validity and precision of a measure of metacognition and find that all 17 measures are valid and most show similar levels of precision. Second, I examine how measures of metacognition depend on task performance, response bias, and metacognitive bias, finding only weak dependences on response and metacognitive bias but many strong dependencies on task performance. Third, I find that all measures have very high split-half reliabilities, but most have poor test-retest reliabilities. This comprehensive assessment paints a complex picture: no measure of metacognition is perfect and different measures may be preferable in different experimental contexts.
... Note that due to recently reported limitations of the m-ratio [27] as a reliable and unbiased measure of metacognitive ability, particularly when tasks have under 400 trials [12,28,29], we did not include this measure as one of the primary outcomes. However, for completeness and consistency with previous studies [15][16][17], we report regression results between m-ratio and both age and symptom dimensions in Supplementary Fig. S6. ...
When making decisions in everyday life, we often rely on an internally generated sense of confidence to help us revise and direct future behaviours. For instance, confidence directly informs whether further information should be sought prior to commitment to a final decision. Many studies have shown that aging and both clinical and sub-clinical symptoms of psychopathology are associated with systematic alterations in confidence. However, it remains unknown whether these confidence distortions influence information-seeking behaviour. We investigated this question in a large general population sample (N = 908). Participants completed a battery of psychiatric symptom questionnaires and performed a perceptual decision-making task with confidence ratings in which they were offered the option to seek helpful information (at a cost) before committing to a final decision. Replicating previous findings, an ‘anxious-depression’ (AD) symptom dimension was associated with systematically low confidence, despite no detriment in objective task accuracy. Conversely, a ‘compulsive behaviour and intrusive thoughts’ (CIT) dimension was associated with impaired task accuracy but paradoxical over-confidence. However, neither symptom dimension was significantly associated with an increased or decreased tendency to seek information. Hence, participants scoring highly for AD or CIT did not use the option to information seek any more than average to either increase their confidence (AD) or improve the accuracy of their decisions (CIT). In contrast, older age was associated with impaired accuracy and decreased confidence initially, but increased information seeking behaviour mediated increases in both accuracy and confidence for final decisions. Hence, older adults used the information seeking option to overcome initial deficits in objective performance and to increase their confidence accordingly. The results show an appropriate use of information seeking to overcome perceptual deficits and low confidence in healthy aging which was not present in transdiagnostic psychopathology.
... To investigate whether any time-on-task effects on EEG activity affected brainbehaviour relationships we performed a single-trial multiple regression analysis (for similar approach see Benwell et al., 2017Benwell et al., , 2022Kopčanová et al., 2023). We used a hierarchical twostage estimation (Friston, 2008) to incorporate participant level variability into group level statistics. ...
... In addition to time-on-task related EEG effects, we also found that alpha and beta desynchronisation uniquely predicted single-trial decision confidence independently of accuracy, RTs, stimulus contrast, and trial-order. This is in line with previous studies that showed post-stimulus alpha/beta power correlates with subjective judgements like confidence and perceptual awareness Faivre et al., 2018;Kopčanová et al., 2023) and is unrelated to accuracy Kopčanová et al., 2023). Given the independence of this effect from RTs, it is unlikely it can solely be attributed to motor preparation (Faivre et al., 2018). ...
... In addition to time-on-task related EEG effects, we also found that alpha and beta desynchronisation uniquely predicted single-trial decision confidence independently of accuracy, RTs, stimulus contrast, and trial-order. This is in line with previous studies that showed post-stimulus alpha/beta power correlates with subjective judgements like confidence and perceptual awareness Faivre et al., 2018;Kopčanová et al., 2023) and is unrelated to accuracy Kopčanová et al., 2023). Given the independence of this effect from RTs, it is unlikely it can solely be attributed to motor preparation (Faivre et al., 2018). ...
Fluctuations in oscillatory brain activity have been shown to co-occur with variations in task performance. More recently, part of these fluctuations has been attributed to long-term (>1hr) monotonous trends in the power and frequency of alpha oscillations (8-13 Hz). Here we tested whether these time-on-task changes in EEG activity are limited to activity in the alpha band and whether they are linked to task performance. Thirty-six participants performed 900 trials of a two-alternative forced choice visual discrimination task with confidence ratings. Pre- and post-stimulus spectral power (1-40Hz) and aperiodic (i.e., non-oscillatory) components were compared across blocks of the experimental session and tested for relationships with behavioural performance. We found that time-on-task effects on oscillatory EEG activity were primarily localised within the alpha band, with alpha power increasing and peak alpha frequency decreasing over time, even when controlling for aperiodic contributions. Aperiodic, broadband activity on the other hand did not show time-on-task effects in our data set. Importantly, time-on-task effects in alpha frequency and power explained variability in single-trial reaction times. Moreover, controlling for time-on-task effectively removed the relationships between alpha activity and reaction times. However, time-on-task effects did not affect other EEG signatures of behavioural performance, including post-stimulus predictors of single-trial decision confidence. Therefore, our results dissociate alpha-band brain-behaviour relationships that can be explained away by time-on-task from those that remain after accounting for it - thereby further specifying the potential functional roles of alpha in human visual perception.
