Goal-setting modulates visual working memory performance through alpha power suppression and effort

Poster Presentation: Tuesday, May 20, 2025, 2:45 – 6:45 pm, Pavilion
Session: Visual Memory: Neural mechanism of working memory

Olga Kozlova1, Kirsten Adam1; 1Rice University

Because the capacity of Visual Working Memory (VWM) is strictly limited, previous research has aimed to improve VWM with feedback and monetary incentives. Often, however, the effects of feedback and incentives on VWM have been mixed. We have proposed an “optimal strategy” account to explain when feedback will be effective versus ineffective at improving VWM performance. Participants completed a behavioral whole-report VWM goal task in Experiment 1 (N=100), with EEG and pupillometry in Experiment 2 (N=22). At the beginning of each trial, participants were given performance goals (e.g., “remember 3 items”), and received feedback and monetary bonuses when goals were met. In Experiment 1, optimal VWM performance corresponded to goals that aligned to an individual’s typical VWM capacity (3 items). In contrast, supra-capacity goals harmed VWM performance by increasing lapses of performance. In Experiment 2, we investigated which underlying cognitive processes may support improvements to VWM with optimal goals. Specifically, we predicted that we should observe a “U-shaped” function for cognitive processes altered by goal-setting, with greater engagement for an “optimal goal” of 3 items compared to under- or over-ambitious goals. In contrast, we predicted that we should observe a monotonic increase for cognitive processes related to simple effort (Goal 5 > Goal 3 > Goal 1). First, we found a “U-shaped” pattern for posterior alpha power suppression (p < .044), suggesting that participants successfully maintained more items throughout the delay period when given an optimal goal. In contrast, we found a “monotonic increase” pattern for tonic pupil dilation (p < .005), suggesting that, paradoxically, participants exerted more effort, but remembered fewer items, for the suboptimal Goal 5 condition. Taken together, our results suggest that effective feedback can improve VWM performance, without concurrently increasing overall effort, by encouraging participants to allocate VWM resources to only a subset of items.