Yin-ting Lin¹ (), Andrew B. Leber¹; ¹The Ohio State University

Visual working memory only maintains limited information. One way we can overcome this limitation is to selectively encode information relevant to the ongoing task. Alternatively, we can use strategies to maximize performance, yet relatively fewer studies have investigated this. Here we examined what individuals choose to encode when the available information exceeds their capacity limits. In the choice task, participants viewed six colours and had to report the colour of three items on a continuous wheel. Importantly, we allowed participants to choose which three items they report. Across three experiments, results show that strategy use varied across individuals. Some participants chose to report items that share high colour similarity (colour clustering). Others chose to report items in certain spatial locations (spatial clustering), and many of these participants preferred items from the upper part of the display. Interestingly, there was a strong trade-off between colour and spatial clustering, with many individuals using a dominant strategy. We further examined whether these individual differences reflect idiosyncratic preferences or differences in ability. Specifically, we also asked participants to perform a cued task, where they were pre-cued to report three items in the display. We found that many participants show better performance for some locations, and that participants’ preferred locations in the choice task match the locations with best memory performance in the cued task. Moreover, participants who showed larger benefits for similar versus dissimilar colours were more likely to use colour clustering strategies. Furthermore, we trained a support vector regression model using cued task data and found that participants tend to choose items that yield the best predicted memory performance, suggesting that participants use strategies that maximize their own performance. Overall, our findings suggest that there are large individual differences in encoding strategy, and individuals appear to be using strategies tailored to their abilities.

Acknowledgements: This work was supported by NSF BCS-2021038 to ABL.