关键词: 视觉工作记忆, 状态模型, 在线态, 离线态, 稳固性优势

Abstract:

Visual working memory (VWM) plays a crucial role in various higher cognitive activities such as problem-solving, learning, and decision-making. The function of VWM extends beyond mere information storage; it also requires the preservation of memory fidelity to ensure the smooth execution of various cognitive activities. State-based models of visual working memory propose a hierarchical structure of memory storage, consisting of both online and offline states. Previous studies have evidenced that memory representations in these two distinct storage states exhibit some resilience to irrelevant distractor and time decay. However, while the properties of memory representations in these two states have been explored independently, it was still unclear whether online and offline memory representations possess comparable resistance to distractor and time decay when exposing to identical cognitive processing.
Thus, this study aimed to compare the robustness of memory representations in the online and offline states, with the goal of determining which storage state had superior resilience. To achieve this, we adopted a sequential presentation memory task, which facilitated the co-existence of both online and offline representations by effectively separating representational storage states. The primary experimental variables were the presence of distractors (Experiment 1) and retention duration (Experiment 2) during memory maintenance involving both states. This design allowed us to examine the effects of these factors on memory representations in the two storage states. A total sample of 115 was recruited in four behavioral experiments. Experiments 1a/1b assessed the effect of distractor on memory representations in the online and offline states, while Experiments 2a/2b examined the effect of time decay. We predicted that, if memory representations in the two states exhibited equal robustness, memory performance in both storage states would remain largely unaffected by the presence of distractors or extended retention duration. Whereas if the representational robustness was weaker in one storage state, then either the presentation of distractor or prolonged retention would result in a significant memory decline in that state.
The results of Experiment 1 collectively showed that when distractor was unexpectedly presented, online memory performance was significantly impaired, while no impairment occurred to offline memory. Experiment 2 jointly revealed that as the retention duration increased from 1s to 3s, significant impairment occurred to online memory only, while offline memory performance was clearly unaffected. These findings indicated that memory representations in the online and offline states differ in their robustness within the identical cognitive processing. Specifically, offline memory had superior resilience compared to the online when it came to resisting both unrelated distractor and time decay.
In summary, during working memory maintenance employing both online and offline storage states, offline memory exhibits a robustness superiority in resisting interference from distractors and temporal decay compared to online memory. These findings contribute to a deeper understanding of multi-state storage mechanisms in working memory, and suggest new directions for future memory research.

Key words: visual working memory, state-based models, online memory, offline memory, robustness superiority