(A) study on the effects of sequential overwrite and non-uniform allocation of memory resource in working memory performance

Abstract

Working memory capacity is known to be limited since memory performance decreases as the number of items increases. To describe such characteristics, a resource model has been suggested: limited memory resources can be flexibly allocated to each item, and performance increases as more resources are allocated. However, the resource model does not suggest a clear mechanism for the primacy and recency effects, in which the items presented first and last in a sequence are more correctly memorized than other items. In this study, we propose a new model describing the mechanism for sequential allocation of memory resources by introducing the concepts of sequential overwrite and non-uniform allocation. First, to investigate the effects of sequential overwrite and resource allocation by order, we performed a sequential memory task. Our results showed that previous memories are reduced when a new item is encoded. Based on our assumption that the last item is not affected by any overwrite, we estimated the amount of resources using the last item. We found that the amount of resources decreases as the sequential order increases. Therefore, we propose a new model in which the sequential overwrite of decreasing resources can describe the mechanism of sequential memory allocation. Next, we hypothesized that memory performance can be improved if sequential overwrite is reduced by efficient memory allocation. To test our idea, we compared the performance under three memory allocation conditions in which the number of items was (1) uninformed, (2) informed, and (3) falsely informed prior to encoding. Our results showed that informing the number of items improves performance, and our model with less overwrite describes the experimental data well. In summary, we propose that the concept of sequential overwrite and decreasing resource allocation can describe the mechanisms of both memory allocation and pre-allocation in a sequential memory task.