Laurielle Kelly Feuzeu Mekoue¹, Vrishak Duriseti¹, Libby Kotei-Fearon¹, Natalie Chace¹, Jeffrey Moher¹; ¹Connecticut College

We examined the relationship between visual search and working memory, particularly focusing on quitting behavior when performing visual search while simultaneously holding information in working memory. Study participants completed an online task with two parts. First, participants were shown one (low-load), two (medium-load), or four (high-load) colored squares. Then, they completed a visual search task where they searched for a vertical blue line target among slanted lines. Targets were present on a randomly selected half of all trials. Finally, they were asked to indicate whether a single square matched the color and location of one of the squares that was previously presented. We hypothesized that when participants have higher working memory loads, they will reach their decision threshold quickly during search, resulting in shorter response times and an increased amount of miss errors. As expected, we observed that as working memory load increased from one to two to four squares, memory accuracy significantly dropped from approximately 90% with 1 square to 70% with 4 squares. Miss rates in search also increased with higher working memory loads, which would be consistent with early quitting. However, on target-absent trials, reaction time increased from 2.44 seconds (s) with 1 square to 2.53 s with 2 squares, then spiking to 2.98 s with 4 squares. This result is not consistent with early quitting with high working memory loads, but may support the idea that increasing memory load generally disrupts search, as more time was taken to search and more errors were made. This is consistent with shared cognitive resources between the two tasks but not consistent with our initial hypothesis that increasing working memory load would decrease search time. This data provides an understanding of how multitasking impairs decision-making by influencing the relationship between working memory, distractor suppression, and visual search.

Acknowledgements: This work was supported by the National Science Foundation Grant BCS-2218384 to JM.