A Discrete Component in Visual Working Memory Encoding

Working memory (WM) is a central cognitive bottleneck, which has primarily been attributed to its well-known storage limit. However, relatively little is known about the processing limit during the initial memory encoding stage, which may also constrain various cognitive processes. The present study introduces a novel method using dynamic stimulus presentation and hierarchical Bayesian modeling to quantitatively estimate visual WM encoding speed. Participants performed a delayed-estimation task with two memory items continuously changing color hues in perceptually unnoticeable steps. Across three experiments, the recall errors systematically shifted toward the direction of color change, providing a proxy measure of encoding speed. Importantly, the observed shifts were best characterized by a temporal lag during the encoding of different items, supported by a mixture of two distributions with credibly distinct encoding times. A supplementary model-free analysis further confirmed the discrete encoding component in visual WM for multiple items. These findings shed light on the temporal dynamics of WM encoding processes.