Аннотация:Spatial memory appears to be supported by multiple internal representations – mental images reflecting the topological and metric characteristics of the environment. There are two types of representations: egocentric spatial representations (esr) and allocentric spatial representations (asr) encoding self-to-object and object-to-object spatial information respectively. A lot of studies investigated esr/asr mechanisms using static scenes as stimuli but little data is known about esr/asr of dynamic scenes. We studied the accuracy of esr/asr formation of static and dynamic scenes in working memory. Working memory processing initiated using a dual task methodology. Eight static and eight dynamic scenes consisting of 4 objects located in different 3d positions were constructed and presented using vr hmd technology. Thirty seven participants were tested. Their task was to remember virtual objects and their locations and then to reproduce a memorized scene in a virtual space using esr or asr viewer’s positions. The accuracy of identification and localization in metric and topology units was evaluated. The results showed that the accuracy of object’s identification for dynamic and static scenes was statistically insignificant both for esr (p=0.17) and asr (p=0.09). The metric accuracy of static scenes was higher for esr (p=0.001) and asr (p=0.03). But topological accuracy of dynamic and static scenesWere not differ significantly both for esr (p=0.06) and asr (p=0.57). The results showed the important role of coding topological spatial information during memorizing dynamic scenes in working memory. This is probably closely related to the «biological motion» recognition, which allows the rapid selection of moving objects from the environment and the reconstruction of the perceived objects shape. The study was funded by russian scientific fund project №19-18-00474-p.