Storing temporal sequences of events (i.e., sequence memory) is fundamental to many cognitive functions. However, it is unknown how the sequence order information is maintained and represented in working memory and its behavioral significance, particularly in human subjects. We recorded electroencephalography (EEG) in combination with a temporal response function (TRF) method to dissociate item-specific neuronal reactivations. We demonstrate that serially remembered items are successively reactivated during memory retention. The sequential replay displays two interesting properties compared to the actual sequence. First, the item-by-item reactivation is compressed within a 200 – 400 ms window, suggesting that external events are associated within a plasticity-relevant window to facilitate memory consolidation. Second, the replay is in a temporally reversed order and is strongly related to the recency effect in behavior. This fast-backward replay, previously revealed in rat hippocampus and demonstrated here in human cortical activities, might constitute a general neural mechanism for sequence memory and learning.

DocRED is a widely used dataset for documentlevel relation extraction. In the large-scale annotation, a recommend-revise scheme is adopted to reduce the workload. Within this scheme, annotators are provided with candidate relation instances from distant supervision, and they then manually supplement and remove relational facts based on the recommendations. However, when comparing Do-cRED with a subset relabeled from scratch, we find that this scheme results in a considerable amount of false negative samples and an obvious bias towards popular entities and relations. Furthermore, we observe that the models trained on DocRED have low recall on our relabeled dataset and inherit the same bias in the training data. Through the analysis of annotators' behaviors, we figure out the underlying reason for the problems above: the scheme actually discourages annotators from supplementing adequate instances in the revision phase. We appeal to future research to take into consideration the issues with the recommend-revise scheme when designing new models and annotation schemes. The relabeled dataset is released at https://github. com/AndrewZhe/Revisit-DocRED, to serve as a more reliable test set of document RE models.

26Storing temporal sequences of events (i.e., sequence memory) is fundamental to many 27 cognitive functions. However, how the sequence order information is maintained and 28 represented in working memory and its behavioral significance, particularly in human 29 subjects, remains unknown. Here, we recorded electroencephalography (EEG) in 30 combination with a temporal response function (TRF) method to dissociate item-31 specific neuronal reactivations. We demonstrate that serially remembered items are 32 successively reactivated during memory retention. The sequential replay displays two 33 interesting properties compared to the actual sequence. First, the item-by-item 34 reactivation is compressed within a 200-400 ms window, suggesting that external 35 events are associated within a plasticity-relevant window to facilitate memory 36 consolidation. Second, the replay is in a temporally reversed order and is strongly 37 related to the recency effect in behavior. This fast-backward replay, previously 38 revealed in rat hippocampus and demonstrated here in human cortical activities, might 39 constitute a general neural mechanism for sequence memory and learning. 40 41 Keywords: sequence memory, sequential reactivation, temporal response function 42 (TRF), reverse replay, recency effect 43 44 3 Introduction 45Storing and retrieving temporal sequences of events (i.e., sequence memory), a 46 capacity shared across species, is crucial to many cognitive functions, including 47 speech recognition, movement planning, and episodic memory [1, 2]. For example, 48 we retrieve a stream of serially ordered events when recalling past personal 49 experience, and we memorize numbers in order when trying to call a friend. Swinging 50 the racket to hit an incoming ball in a tennis match similarly involves the planning 51 and controlling of a chain of movement elements over time. 52To accomplish sequence memory, two core components -the content (items) and 53 the ordinal information (temporal order) -of a sequence are vital to be encoded and 54 maintained in working memory. There exists ample evidence that neural responses 55 during the retention interval show a sustained load-dependent enhancement [3][4][5][6][7][8], 56 suggesting that the maintenance of mnemonic contents is implemented by recurrent 57 feedback loops [9], as well as suppression of irrelevant information [6, 10]. That 58 being said, retention of the sequence order information in memory cannot solely rely 59 on an overall enhancement of neural activity and presumably requires the temporally 60 segregated representations of individual items. Theoretical models postulate that 61 sequence memory is mediated by a theta-gamma coupled neuronal oscillatory 62 mechanism [11, 12], such that individual items of the list/sequence, encoded in 63 gamma-band activities, occur at the different phases of a theta-band rhythm. A recent 64 MEG study provides important evidence supporting this theta-gamma coupling model 65 during the memory encoding period [13]. Moreover, by a...