December 2022
Volume 22, Issue 14
Open Access
Vision Sciences Society Annual Meeting Abstract  |   December 2022
People can evaluate the correctness of their face-identification decisions using comparative confidence judgments
Author Affiliations
  • Geraldine Jeckeln
    University of Texas at Dallas
  • Pascal Mamassian
    CNRS & École Normale Supérieure
  • Alice J. O'Toole
    University of Texas at Dallas
Journal of Vision December 2022, Vol.22, 3341. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Geraldine Jeckeln, Pascal Mamassian, Alice J. O'Toole; People can evaluate the correctness of their face-identification decisions using comparative confidence judgments. Journal of Vision 2022;22(14):3341.

      Download citation file:

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

  • Supplements

Confidence judgments serve as an indicator of identification accuracy in legal practices such as forensic face examination (Phillips et al., 2018). However, little research has focused on the factors that underlie these judgments. Moreover, it is not even clear whether people can evaluate the correctness of their face-identification decisions reliably through confidence reports. Previous studies show that confidence is a good indicator of accuracy when it is based on the strength of the sensory stimulus and when it is assessed with minimal bias (Mamassian, 2016; Wixted & Wells, 2017). Here, we implemented comparative judgments to measure confidence more directly and eliminate the risk of response bias imposed by the commonly used confidence scales (Mamassian, 2020; Phillips et al., 2018). Specifically, we tested participants (N = 58) on a confidence forced-choice task embedded in a face-identification test. On each face-identification trial, participants viewed three face images (two same-identity images, one different-identity image). The task was to select the odd-one-out. Upon completion of two face-identification trials (trial pair), participants compared the two decisions and selected the trial on which they felt more confident. To measure how task difficulty informs confidence decisions, we extracted item-difficulty estimates from an Item Response Theory model fitted to data collected separately. The difference in difficulty between the items in a trial pair predicted the proportion of higher-confidence judgments allocated to the easier item of the pair (R2 = 0.4924, F(1,50)= 50.48, p < .001). Performance was significantly more accurate on higher-confidence trials (M= 0.8853, SE= 0.0138), in comparison to lower-confidence trials (M=0.7924, SE=0.0176), (t(53) = 8.7689, p < .001, 95% CI:[ 0.0717, 0.1142], Cohen’s d=.7994). These findings show that people monitor task difficulty to evaluate the correctness of their face-identification decisions and comparative confidence judgments are good indicators of face-identification accuracy.


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.