The Attention Of Behavior In The Working Memory With Fmri Study

Working memory and attention are two of the most extensively studied higher cognitive functions. The former means to maintain and manipulate information on line for a short time with limited capacity. The latter allows people to selectively process the external environment or internal memory representations that are most relevant to their goals. Previous researches on attention mainly focused on the attentional selection in outer environment, yet the present article specially introduced a series of studies on the selective attention on the contents stored in human working memory.Despite the relatively large number of studies on visual spatial attention, few studies have considered the attentional selection in working memory (i.e. the selective access to different items held in working memory). In fact, the selective attention to working memory contents is very common in our daily life. For example, mental addition of two three-digit numbers involves first adding the ones, then the tens, and so on. During these steps, some digits are selected for specific roles (e.g., first addend) in the addition operation, while the others are held in working memory, but not used in the current operation. In this example, the first addend at any time is considered to be selected by attention, but those unused in the current operation not.The object switching paradigm is one of the commonly used approaches in studies of attention in working memory. It was originally introduced by Garavan (1998) as the "serial count" task. In this task, two kinds of geometric figures (like squares and circles) are serially presented in a random order at the participant's own pace (the current displaying figure will be replaced by the next one each time the response key being pressed). Subjects in this task need to count the number of times that each of the two figure types appears in a sequential presentation. While the figures being presented one by one, participants have to keep two running counts for each figure type and make corresponding increments each time the figure updates. According to whether the successive stimuli were of the same type, "switch" (successive stimuli were different) and "nonswitch" (successive stimuli were same) updates could be defined. Garavan (1998) showed that the "nonswitch" reaction times were about 500 ms shorter than that of the "switch" condition. He interpreted this difference as the cost for attention shifting from one memory count to the other, and proposed that the capacity of this internal focus of attention is limited to just one item. Garavan's finding provides one of the important bases for the work in this article. Also, different kinds of modified object switching paradigms were adopted in all the experiments introduced here to further study the characteristics of selective attention in working memory.By using the "tri-count" task favored by Zhi-Hao Li, Voigt and Hagendorf, in the first study introduced in the article, the author found that attention switching between working memory items is accompanied by inhibition of the previously attended working memory item, an phenomenon similar to the "backward inhibition" effect found in the task switching studies. The effect is thought to reduce the potential interference from previously attended working memory item on the currently attended one. This finding on the one hand discloses the possible psychological mechanism underlying the prefrontal-cingulate co-activation found in Zhi-Hao Li's and other researchers' fMRI and ERP studies on attention switching within working memory, on the other hand agrees with the viewpoint in Oberauer and Kliegl's (2006) interference model of working memory.The first study investigates the accompanying phenomenon during attention switching within working memory, and lead us a step closer to understanding the attention within working memory. However, another important issue in this area is to find the unit of selective attention. In visual spatial attention studies, there are three main theories, respectively proposing the feature-based, location-based and object-based selection. A rising question is what the unit of attentional selection in working memory should be. Because in our first study, all the working memory items belong to the verbal working memory modality, furthermore each of them only has one feature for selection and ongoing operation, the study cannot answer the selection unit question. Therefore, the author designed the second study to resolve it specially. In the second study introduced in this article, by introducing visuospatial working memory items, the author observed that switching attention between properties of an integrated working memory item was faster than switching between respective properties of different items. This finding supports the object-based attention theory. However, because of the contribution of spatial attention, this binding facilitation may actually not be due to the object-based selection. Thereby, the author conducted an experiment without spatial demand to assure the existence of the object-based attention. The result still showed a binding facilitation. Accordingly, it is suggested that attentional selection within working memory is either location-based or object-based. This conclusion is consistent with the proposals of Ducan, Vecera and Farah, which implicate some similarity of the selection between external and internal attention. Moreover, a broader implication of this study suggests that the region of direct access in Oberauer's concentric model may be a general, supra-modal component.Previous studies on attention switching in working memory suggest that the time to switch attention grows with the number of updating working memory items. These findings imply potential common substrates underlying the working memory storage and the attention switching. By adjusting the working memory load (i.e. number of the counts) and attentional switching condition, and using the functional magnetic resonance imaging (fMRI) technique, the third study of this article explores whether there are overlapping neural mechanisms underlying these two processes. The results show that except the brain activation preferentially to the load and switching factor (right parietal cortex, bilateral occipital cortex), the left dorsolateral prefrontal cortex (BA9/46), the left parietal (BA7/40), and the anterior cingulated gyrus (BA32/24) activate to both the load and switching factor, therefore confirming the sharing mechanisms of the two processes in the cortical level. The study suggests that the central executive in Baddeley's multi-component working memory model may still have a storage function, and provide neuroimaging evidence for Oberauer's and Kliegl's interference model of working memory.The result of Study I in the present article has already been published on Brain Research Bulletin (2006, 69: 214-221, Impact Factor 2005: 2.481); Study II was submitted to Journal of Experimental Psychology: Learning, Memory, and Cognition (Impact Factor 2005: 2.812), then invited to make minor revisions, and now the revision has been submitted. Study III is now in the stage of research paper writing.