关键词: 奖赏预测误差, 事件边界, 时间顺序记忆, 来源记忆, 记忆权衡效应

Abstract:

The human brain automatically segments continuous experiences into discrete events to better remember ongoing experiences in daily life. This automatic process is known as event segmentation. The time points between different events are called event boundaries—they indicate when one event ends, and another begins. Studies have shown that the event boundaries may enhance the item-context source memory of information at the boundaries but impair temporal order memory in across-event information. Notably, previous studies mainly focused on the boundaries caused by changes in the external environment and rarely paid attention to the subjective boundaries caused by changes in an individual’s internal psychological context. Moreover, Rouhani et al. (2020) first confirmed that reward prediction errors (RPE) could be used as event boundaries to influence memory. Additionally, it has been demonstrated that the RPE event boundary reduces the temporal order memory in across-event information. However, the effects of the RPE event boundary on temporal order and source memory and whether the mnemonic trade-off effect exists are not clear. The neurophysiological mechanisms underlying the effects of event boundaries on memory require further examination.

This study used behavioral and event-related potentials (ERP) technology in three experiments to explore the effect of RPE event boundaries on temporal order and source memory, respectively, based on behavioral and neurophysiological mechanisms. Experiment 1 used a neutral scene picture matching value as materials. The value of successive pictures fluctuates around an average value to form an event, and event boundaries denote when there is a significant shift in the value, which is the reward prediction error. The participants were required to complete two memory tests: a temporal order memory test and a source memory test. The source memory test was presented with a neutral scene picture, and participants were asked to choose a value that matched the learning stage from the two alternatives. We aimed to explore the effect of the RPE event boundary on temporal order and source memory. However, no mnemonic trade-off effect was observed. Therefore, whether other factors might influence the mnemonic trade-off effect that exists is not clear. In Experiment 2, RPE was divided into high and low strength. We aimed to explore the effects of different RPE strengths on temporal order and source memory. After we obtained stable results, in Experiment 3, we used ERP technology to explore the N400 and P600 effects under different conditions at the memory retrieval stage in the high RPE condition to examine the detailed mechanism of the effect of event boundary on memory.

The behavioral results showed that the RPE event boundary enhanced only the neutral scene picture-value source memory of information at the boundaries in Experiment 1. High and low RPE event boundaries affect temporal order and source memory differently. The high RPE event boundary enhanced the neutral scene picture-value source memory of information at the boundaries. Further it reduced the temporal order memory of information across-events, which caused the mnemonic trade-off effect between temporal order and source memory in Experiment 2. The ERP results showed that compared to the within-event/non-boundary condition, correctly retrieving information of temporal order and source memory in the across-event/boundary condition induced a larger N400 (350−550ms) effect but did not induce a larger P600 (600−1000ms) effect in Experiment 3. These two memory tests were activated in different brain regions. The temporal order memory in the across-events condition was mainly activated in the anterior region, while the source memory boundary condition was mainly activated in the parietal region.

This study can be summarized as follows. The segmentation strength of the event boundary is an important factor affecting the mnemonic trade-off effect between temporal order and source memory. The mnemonic trade-off effect only occurs when the representation difference between events is sufficiently vast, and the segmentation strength of the boundary is sufficiently high. Furthermore, the N400 component is an important index that reflects the integration and segmentation of episodic memory using event boundaries.

Key words: reward prediction errors, event boundaries, temporal order memory, source memory, mnemonic trade-off effect