Although feedback is an important component of computer-based instruction (CBI), the effects of feedback on higher-order learning outcomes are not well understood. Several meta-analyses provide two rules of thumb: any feedback is better than no feedback and feedback with more information is better than feedback with less information. Unfortunately, many of the studies included in these meta-analyses involved only lower-order learning outcomes, where criterion posttest items were taken verbatim from the lesson materials. This review intends to show that multiple-try feedback (MTF), that requires the learner to "try again" on error, may be better than other common forms of feedback in CBI for higher-order learning outcomes. The results of this review indicate that MTF may be slightly superior to the other most common forms of feedback in CBI for higher-order posttest outcomes, while these other forms of feedback will be equivalent to or superior to MTF for lower-order, verbatim outcomes. It is proposed that MTF tends to directly strengthen the activation levels of an individual's semantic meanings of the task content, while KR, KCR, and DF tend to strengthen specific instances or episodes of the content.