Fig. 1. Results from Experiment 1: mean credibility ratings (upper panel) and mean guilty ratings (lower panel) as a function of the witness's confidence (high or low) and whether or not the witness was shown to have made a collateral error while testifying. The credibility ratings were made on a scale from 1 to 7. The guilty ratings combine verdict preferences with confidence (-10.5 indicates a decision to acquit, made with the highest confidence, and +10.5 indicates a decision to convict, made with the highest confidence). Because the eyewitness testified against the defendant, lower guilty ratings imply lower credibility of the eyewitness, and higher guilty ratings imply higher credibility of the eyewitness. Error bars include standard errors.