BackgroundThe processing mechanisms of visual working memory (VWM) have been extensively explored in the recent decade. However, how the perceptual information is extracted into VWM remains largely unclear. The current study investigated this issue by testing whether the perceptual information was extracted into VWM via an integrated-object manner so that all the irrelevant information would be extracted (object hypothesis), or via a feature-based manner so that only the target-relevant information would be extracted (feature hypothesis), or via an analogous processing manner as that in visual perception (analogy hypothesis).Methodology/Principal FindingsHigh-discriminable information which is processed at the parallel stage of visual perception and fine-grained information which is processed via focal attention were selected as the representatives of perceptual information. The analogy hypothesis predicted that whereas high-discriminable information is extracted into VWM automatically, fine-grained information will be extracted only if it is task-relevant. By manipulating the information type of the irrelevant dimension in a change-detection task, we found that the performance was affected and the ERP component N270 was enhanced if a change between the probe and the memorized stimulus consisted of irrelevant high-discriminable information, but not if it consisted of irrelevant fine-grained information.Conclusions/SignificanceWe conclude that dissociated extraction mechanisms exist in VWM for information resolved via dissociated processes in visual perception (at least for the information tested in the current study), supporting the analogy hypothesis.

Mainstream theories of visual perception assume that visual working memory (VWM) is critical for integrating online perceptual information and constructing coherent visual experiences in changing environments. Given the dynamic interaction between online perception and VWM, we propose that how visual information is processed during visual perception can directly determine how the information is going to be selected, consolidated, and maintained in VWM. We demonstrate the validity of this hypothesis by investigating what kinds of perceptual information can be stored as integrated objects in VWM. Three criteria for object-based storage are introduced: (a) automatic selection of task-irrelevant features, (b) synchronous consolidation of multiple features, and (c) stable maintenance of feature conjunctions. The results show that the outputs of parallel perception meet all three criteria, as opposed to the outputs of serial attentive processing, which fail all three criteria. These results indicate that (a) perception and VWM are not two sequential processes, but are dynamically intertwined; (b) there are dissociated mechanisms in VWM for storing information identified at different stages of perception; and (c) the integrated object representations in VWM originate from the "preattentive" or "proto" objects created by parallel perception. These results suggest how visual perception, attention, and VWM can be explained by a unified framework.

Visual working memory (VWM) maintains and manipulates a limited set of visual objects being actively used in visual processing. To explore whether and how the fine detailed information is stored in VWM, four experiments have been conducted while recording the contralateral delay activity (CDA), an event-related potential difference wave that reflects the information maintenance in VWM. The type of the remembered information was manipulated by adopting simple objects and complex objects as materials. We found the amplitude of CDA was modulated by object complexity: as the set size of memory array rose from 2 to 4, the amplitude of CDA stopped increasing for maintaining complex objects with detailed information, while continued increasing for storing highly discriminable simple objects. These results suggest that VWM can store the fine detailed information; however it can not store all the fine detailed information from 4 complex objects. It implies that the capacity of VWM is not only characterized by a fixed number of objects, there is at least one stage influenced by the detailed information contained in the objects. These results are further discussed within a two-stage storing model of VWM: different types of perceptual information (highly discriminable features and fine detailed features) are maintained in VWM via two distinctive mechanisms.

Multi-object tracking is an important ability for an autonomous vehicle to safely navigate a traffic scene. Current state-of-the-art follows the tracking-by-detection paradigm where existing tracks are associated with detected objects through some distance metric. The key challenges to increase tracking accuracy lie in data association and track life cycle management. We propose a probabilistic, multi-modal, multiobject tracking system consisting of different trainable modules to provide robust and data-driven tracking results. First, we learn how to fuse features from 2D images and 3D LiDAR point clouds to capture the appearance and geometric information of an object. Second, we propose to learn a metric that combines the Mahalanobis and feature distances when comparing a track and a new detection in data association. And third, we propose to learn when to initialize a track from an unmatched object detection. Through extensive quantitative and qualitative results, we show that our method outperforms current stateof-the-art on the NuScenes Tracking dataset.

Background: The authors compare the effectiveness and safety of endovascular treatment (EVT) versus best medical management (BMM) in strokes attributable to acute basilar artery occlusion (BAO). Methods: The present analysis was based on the ongoing, prospective, multicenter ATTENTION (Endovascular Treatment for Acute Basilar Artery Occlusion) trial registry in China. Our analytic sample comprised 2134 patients recruited at 48 sites between 2017 and 2021 and included 462 patients who received BMM and 1672 patients who received EVT. We performed an inversed probability of treatment weighting analysis. Qualifying patients had to present within 24 hours of estimated BAO. The primary clinical outcome was favorable functional outcome (modified Rankin Scale score, 0–3) at 90 days. We also performed a sensitivity analysis with the propensity score matching–based and the instrumental variable–based analysis. Results: In our primary analysis using the inversed probability of treatment weighting–based analysis, there was a significantly higher rate of favorable outcome at 90 days among EVT patients compared with BMM-treated patients (adjusted relative risk, 1.42 [95% CI, 1.19–1.65]; absolute risk difference, 11.8% [95% CI, 6.9–16.7]). The mortality was significantly lower (adjusted relative risk, 0.78 [95% CI, 0.69–0.88]; absolute risk difference, −10.3% [95% CI, −15.8 to −4.9]) in patients undergoing EVT. Results were generally consistent across the secondary end points. Similar associations were seen in the propensity score matching–based and instrumental variable–based analysis. Conclusions: In this real-world study, EVT was associated with significantly better functional outcomes and survival at 90 days. Well-designed randomized studies comparing EVT with BMM in the acute BAO are needed. Registration: URL: www.chictr.org.cn ; Unique identifier: ChiCTR2000041117.