September 2021
Volume 21, Issue 9
Open Access
Vision Sciences Society Annual Meeting Abstract  |   September 2021
The shape of metacognitive noise confounds metacognitive efficiency with confidence bias
Author Affiliations & Notes
  • Kai Xue
    Georgia Institute of Technology
  • Medha Shekhar
  • Dobromir Rahnev
  • Footnotes
    Acknowledgements  This work was supported by the National Institute of Health (award: R01MH119189) and the Office of Naval Research (award: N00014-20-1-2622).
Journal of Vision September 2021, Vol.21, 2794. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Kai Xue, Medha Shekhar, Dobromir Rahnev; The shape of metacognitive noise confounds metacognitive efficiency with confidence bias. Journal of Vision 2021;21(9):2794.

      Download citation file:

      © ARVO (1962-2015); The Authors (2016-present)

  • Supplements

It is now widely appreciated that confidence ratings are corrupted by metacognitive noise. We recently suggested that the magnitude of the metacognitive noise increases with sensory evidence such that the highest confidence criteria are the noisiest (Shekhar & Rahnev, 2021, Psychological Review). This effect was captured by a new process model of confidence that predicts that increasing one’s confidence – which is equivalent to using lower confidence criteria – should result in higher estimated metacognitive sensitivity. In order to test this predicted relationship, here we developed a new method of simulating a change of confidence by removing the highest or the lowest confidence criterion from existing data. Intuitively, removing the highest (vs. the lowest) confidence criterion leads to the confidence criteria becoming more liberal and thus simulates a situation where subjects make more judgments with high confidence leading to an overall increase in confidence. Since removing the highest (vs. the lowest) confidence criterion removes the criterion that is hypothesized to be the noisiest (vs. least noisy), we predicted that it would lead to higher estimated metacognitive sensitivity. We applied this manipulation to the data from three tasks from the Confidence Database (N > 400 in each) and we used meta-d’ as the measurement for metacognitive sensitivity. Confirming to the model’s prediction, we found that removing the highest (vs. the lowest) confidence criterion leads to an increase in both confidence and metacognitive sensitivity in all three tasks (all p’s < 0.005). These results provide support for the notion that metacognitive noise increases with decision evidence, and point to an important confound between metacognitive sensitivity and confidence bias.


This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.