CONTEXT, FEEDBACK, AND THE CALIBRATION AND RESOLUTION OF CONFIDENCE IN PERCEPTUAL JUDGMENTS

The effects of variations in the global task difficulty context on jud gmental confidence and confidence calibration were investigated in two experiments requiring perceptual comparisons. In Experiment 1, target judgments of moderate difficulty were embedded in a larger set of mor e difficult (hard context) or less difficult (easy context) judgments. Decisional response time on the target items was longer in the hard c ontext condition, but there was no effect of difficulty context on tar get judgment confidence, accuracy, over/underconfidence, calibration, or resolution. In Experiment 2, each subject was exposed to three leve ls of local judgment difficulty. The global contextual difficulty mani pulation involved varying the frequency with which the hard and easy j udgments appeared, and the presence or absence of trial-by-trial respo nse feedback was manipulated between subjects. As in Experiment 1, con textual difficulty affected decisional response times but Dot mean con fidence ratings or accuracy. However, we found that providing feedback on a globally difficult task improves calibration, but providing feed back on a globally easy task makes calibration worse. Also, resolution (the ability to differentiate correct from incorrect judgments) was f ound to be superior for easy judgments in a difficult context and for difficult judgments in an easy context. We discuss the implications of these findings for research on confidence and confidence calibration.