NUMERICAL DISTANCE AND ANCHORING EFFECTS IN JUDGING CONFIDENCE
On half of the trials in a sensory detection task, participants were required to decide whether they were more or less confident of the correctness of their decision than the confidence denoted by a number N ranging from 55 to 95. Subsequently, participants expressed confidence on a scale ranging from 50 to 100. On the other half of the trials, participants simply rendered confidence. Times to compare the internal confidence estimate to the numerical value N revealed numerical distance effects; as the distance between N and their subsequently rendered levels of confidence increased, response times decreased. Analyses of mean confidence ratings revealed anchoring effects, with mean absolute confidence ratings assimilating toward the value of N.