Abstract:

Scene gist refers to the perceptual and conceptual representations about real-world scenes, and the processing of scene category is one form of scene gist recognition. Three theories have been proposed to account for scene gist recognition: local processing theory (LPT), global processing theory (GPT), and dual-processing theory (DPT). LPT held that the recognition of objects and the semantic relationships among the objects (object information) promotes scene gist recognition; but GPT held that scene gist is processed as a whole (spatial information) and there is no need to process the local information; however, according to DPT, LPT and GPT are not absolutely opposite, and the recognition of scene gist relies on simultaneous local and global processing. There has been increasing evidences supporting DPT. However, it is not clear how the two kinds of information are extracted during scene gist recognition. Here we combined the visuospatial working memory task and scene gist discrimination task to examine how visual and spatial working memory affected scene gist recognition. Two experiments were conducted to examine effects of visuospatial working memory loads on scene gist recognition. In Experiment 1, participants were asked to make a judgment as to whether the two scene pictures presented simultaneously on the screen at different time (27 ms, 107 ms or 500 ms) belonged to the same basic-level category or not, while keeping or not keeping four different geometric figures (object working memory load) in working memory. In Experiment 2, the only difference from Experiment 1 was that the participants were required to keep or not keep four different locations (spatial working memory load) while performing the category discrimination task. So in the two experiments, there were two kinds of memory load conditions (non-load vs. four-load), three different presentation times (27, 107, 500 ms) and three kinds of scene matching conditions (natural and natural, natural and manmade, manmade and manmade) crossed and the experiments were within-subject design of 2 (load conditions) by 3 (exposure times) by 3 (scene matching). The dependent factor was the nonparametric discrimination index (A’). The results from the two experiments showed that the main effect of object working memory load was significant, indicating recognition of four-load was lower than that of non-load. But the main effect of spatial working memory load was not significant. The two experiments revealed that the presentation time also affected scene gist recognition. The longer the present time was, the higher recognition was. The main effect of scene matching was also significant, indicating that recognition of natural and manmade was higher than that of “natural and natural” or “manmade and manmade”. These results suggest that scene gist recognition is sensitive to the change of object working memory load but not to the change of spatial working memory load. And the present research also supports the hierarchy of scene gist processing: the processing of superordinate level is prior to the processing of basic level.

摘要：

本研究采用工作记忆任务和场景辨别任务相结合的双任务范式探究客体工作记忆负载 (实验1)和空间工作记忆负载 (实验2) 对场景主旨辨别的影响。结果发现: (1) 与无负载条件相比, 客体负载条件下的非参辨别力指数(A’)明显下降, 而空间负载条件下A’无明显变化; (2) 随着场景图片呈现时间的延长, 场景辨别任务的A’逐渐提高; (3) 对自然场景和人工场景匹配情况下的A’ 明显高于对自然与自然或者人工与人工匹配情况下的A’。由此可见, 场景主旨识别, 特别是基本水平的场景主旨辨别, 与客体工作记忆共享部分认知资源, 而不与/很少与空间工作记忆共享认知资源; 在识别过程中, 上级水平范畴识别优先于基本水平范畴识别, 进一步支持了上级水平范畴优先的主旨层级加工观点。