“…However, it would be unwise to interpret our results as legitimizing the general use of confidence ratings to estimate yes-no ROCs. Our reluctance comes from the fact that confidence judgments often do not behave as expected and/or can change the phenomena being studied (see Benjamin et al, 2013;Brainerd, Nakamura, Reyna, & Holliday, 2017;Kellen & Klauer, 2015;Miyoshi, Kuwahara, & Kawaguchi, 2018). Also relevant is the way in which confidence judgments are requested (e.g., one-step versus two-step procedure; see Moran, Teodorescu, & Usher, 2015;Stephens, Dunn & Hayes, 2019).…”