Practice Makes Perfect: An Investigation into the Enhancement of Eyewitness Predictive Confidence

Document Type


Degree Name

Doctor of Philosophy (Ph.D)


Psychology and Special Education

Date of Award

Spring 2018


The timing of an eyewitness's confidence assessment has been shown to be crucial to the strength of its relationship with accuracy. Postdictive confidence, which is provided immediately after an identification task (e.g., how confident are you in the person you just identified?), has been shown to have a strong relationship with accuracy. However, external influences have the potential to harm the quality of the postdictive confidence relationship. One such influence is a predictive confidence assessment, which is provided after the witnessed event but before the identification task (e.g., based on your memory for what you witnessed, how confident are you in your ability to make an accurate lineup decision later?). Predictive confidence assessments have a weak relationship with accuracy, and recent research indicates that it can harm the calibration of a subsequent postdictive confidence assessment. One goal of the current study was to replicate that finding. Additionally, the current study borrowed from the metacognitive literature on judgments of learning and retrospective confidence judgments in an effort to improve the predictive confidence assessment relationship with accuracy. Moreover, the current study aimedto diminish the harmful impact of predictive confidence on the postdictive confidence-accuracy (CA) relationship. We used a mock-crime paradigm featuring three different conditions inspired by the metacognitive literature: (a) predictive confidence with standard instructions and postdictive confidence assessments, (b) predictive confidence with modified instructions and postdictive confidence assessments, and (c) postdictive confidence assessment alone. We found no significant differences in either discriminability or the confidence-accuracy relationship among all conditions. The lack of confidence-accuracy relationship for even our control condition (the postdictive alone assessment) suggests a potential limitation in our stimuli construction. Future research could rework the stimuli for this study to further clarify the potential of modified predictive confidence instructions to improve the predictive confidence- accuracy relationship and reduce any harmful impact of predictive confidence on a subsequent postdictive assessment confidence-accuracy relationship.


Curt A. Carlson

Subject Categories

Psychology | Social and Behavioral Sciences