47 subjects adjusted the extended vertical lines of 20 inverted-T figures to make them appear equal to a horizontal line and rated the confidence in their accuracy after each trial. One group viewed figures of varying sizes, a second group viewed figures of standard size, and a third group viewed figures of standard size but received feedback on their accuracy immediately after completing Trial 5. Except for a significant increase in accuracy on Trial 6 for the Feedback Group, there were no differences in performance among groups. Subjects made the vertical line significantly shorter than the horizontal line on initial trials but their accuracy improved over trials. In contrast, there was no consistent increase in confidence, and several analyses indicated that confidence ratings were unrelated to accuracy. These results suggest that the subjects were unaware of the decrement in illusion that occurred over trials.