Samet Kaya, Miri Besken
Value-directed remembering (VDR) is typically robust, with people recalling high- over low-value information. The current study examined whether this selectivity survives when value must compete with a demanding conceptual manipulation requiring retrieval and manipulation of information from semantic memory: generating truthful answers versus fabricating lies in response to general-knowledge questions. Across five experiments, participants generated truthful answers or fabricated lies in response to general-knowledge questions tagged as low (1 point) or high value (10 points), provided item-by-item judgments of learning (JOLs), and received a free recall task; Experiments 1-4 used mixed (within-subject) lists, and Experiment 5 used pure (between-subjects) lists. Lie fabrication reliably reduced JOLs yet boosted free recall when mixed with truth-telling; in a pure-list design, lying and truth-telling produced equivalent recall while JOLs still favored truth. Crucially, value increased JOLs but never improved recall in any experiment. A single-paper meta-analysis of JOLs confirmed significant effects of both encoding condition and value, showing that participants can use value in their memory predictions even while fabricating lies. In contrast, the meta-analysis of recall showed a reliable benefit of lie fabrication over truth overall but no effect of high over low value. Thus, the conceptual demands of the encoding task eliminated value-based memory advantages. Taken together, these findings suggest that value cues reliably influence metacognitive judgments but do not necessarily translate into improved memory when encoding involves a concurrent resource-intensive conceptual task These results align with an effortful encoding view, where value guides selective memory only when sufficient cognitive resources remain available, and demanding conceptual tasks can deplete resources, causing value cues to be overlooked.