College student Ss were given two trials on each of 3 lists of stimulus words with either 2 or 4 digits as response alternatives. On immediate outcome items, E said “Right” or “Wrong” immediately after S's response. On delayed outcome items, E informed S that his response had been “Right” or “Wrong” the next time the item was presented. On no-information items E gave no information on either trial. Delayed “Wrong” led to significantly less response repetition than all other conditions. Differences among other conditions were generally not significant.