Inik Kim¹, Tianye Ma¹, Lila Liu¹, Weiwei Zhang¹; ¹University of California, Riverside

Working memory (WM) is a multidimensional cognitive process that involves dynamic activation of information. Although recent research has focused on the noisy WM representations, many studies continue to rely on a unitary measure such as either precision or strength. In the present study we have developed an integrated model with two orthogonal components of WM representations, strength and precision, by modifying the Target Confusability Competition (TCC) model. To test our hypothesis, two experiments were conducted using a Random Dot Motion (RDM) task in which participants remembered and then estimated the direction of moving dots. Experiment 1 manipulated Gaussian bandwidth noise of motion stimuli and showed that increasing noise largely reduced WM precision, but not strength. Experiment 2 manipulated the set size and coherence of RDMs and the results showed that decreasing coherence of RDMs largely reduced WM strength, while precision remained relatively comparable. In summary, these findings demonstrate that WM representations can be computationally and experimentally dissociated into separate dimensions of strength and precision. This work highlights the importance of accurately characterizing the structure and limitation of WM representations.