Making judgments of learning (JOLs) can sometimes modify subsequent memory performance, which is referred to as JOL reactivity. We evaluated two major theoretical explanations of JOL reactivity and used the dual-retrieval model to pinpoint the retrieval processes that are modified by JOLs. The changed-goal hypothesis assumes that JOLs highlight differences in item difficulty and switch learners' goals from mastering all items to focusing more on easier items at the expense of harder ones. It follows that JOLs should impair recall for harder items (negative reactivity). The cue-strengthening hypothesis posits that JOLs strengthen the cues that inform JOLs. Thus, JOLs should enhance recall (positive reactivity) when memory tests are sensitive to those cues. In Experiment 1, we compared the reactivity of item-level JOLs between three types of word pairs that differ in difficulty: identical, strongly related, and weakly related pairs. In Experiment 2, we tested whether prestudy JOLs produced similar reactivity as immediate JOLs for items that differed in difficulty (related versus unrelated word pairs). In Experiment 3, we investigated whether JOL reactivity was moderated by relations among to-be-recalled items (related or unrelated), JOL type (item-level or list-level), and test format (associative or free recall). Across the three experiments, there was converging support for the cue-strengthening hypothesis. Moreover, the dual-retrieval model revealed a consistent process-level pattern that positive JOL reactivity was always accompanied by enhanced direct retrieval of verbatim memories. Under some conditions, JOLs also enhanced reconstructive retrieval and familiarity.