Learning to decompose object-based encoding

Poster Presentation: Saturday, May 17, 2025, 8:30 am – 12:30 pm, Banyan Breezeway
Session: Visual Memory: Objects and features

Yingtao Fu1, Longfei Ju1, Mowei Shen1, Hui Chen1; 1Zhejiang University

Object has been considered as the basic unit for working memory encoding of external inputs. This is typically demonstrated by the involuntary encoding of task-irrelevant features on one object even when only one feature on that object is deemed task-relevant. The current study explored whether the object-based encoding could be decomposed through repetitive learning. Six experiments were conducted to test the working memory trace of task-irrelevant colors when they kept constant throughout the experiment. In Experiments 1 and 2, we found that the automatic encoding of the task-irrelevant color could be eliminated through repetition, providing initial evidence that object-based encoding can be modified through learning. Experiments 3 and 4 revealed that this learning effect diminished when multiple items needed to be learned, suggesting a strict capacity limitation for such learning. Additionally, Experiments 5 and 6 showed that the learning effect is constrained by the load of constant colors, rather than the load of objects containing those colors, indicating that the underlying unit of learning is a specific featural value. In summary, our findings demonstrate that object-based encoding can be further refined through repetitive learning of task-irrelevant features, with the effect being limited to a single featural value. These results highlight both the flexibility and capacity constraints inherent in cognitive selectivity.