视觉工作记忆是大脑对视觉信息进行临时存储并处理的系统，是人类实现更复杂更高级的认知功能的基础，并且工作记忆容量能有效预测个体的智商、（未来的）学业成就等。但视觉工作记忆容量（能被成功储存并进行加工的项目的数量）有限。传统视觉工作记忆研究采用的刺激较为简单（如色块、朝向等），得到了人类平均记忆容量为3-4的结论。然而，现实世界中，人们接触的视觉信息更为复杂和多样。若以真实世界的物体作为实验材料，是否会得到类似结果？近年已有研究发现，相比于传统的简单材料，以真实世界的物体作为实验材料，人们反而能记住更多的客体的数量。但潜在的问题是虽然材料是复杂物体，但实验任务考察的记忆精度要求较低。当使用真实世界的复杂物体为材料，在更高的记忆精度要求下，记忆容量又会有何表现？个体在不同的材料种类下的记忆容量是否相同？这一系列问题尚待探究。此外，大量的研究已经发现，视觉工作记忆容量存在较大个体差异。近年来的脑成像研究（特别是脑电和核磁成像技术）也找到了记忆容量个体差异的一些证据，但神经机制尚未完全清晰。 本研究对以上问题进行初步探索和回答。本研究采用来自真实世界的复杂物体作为实验材料（面孔、房屋、车三类材料），设计了一种新的对视觉工作记忆容量的层级测量方法，在不同精度层面（项目水平和视角水平）考察个体的视觉工作记忆容量。具体流程为：每一轮中，1张或3张图片（来自同一类别）依次呈现，经过4秒的延迟，被试从四个选项中选出目标图片。创新点主要是在测试阶段：干扰选项与目标图片的相似性是不同层级，其中一个干扰项与目标图片相似度高（来自同一项目，只是视角不同，被试错选该选项判为角度错误但项目正确）；另外两个干扰项与目标图片相似度低（来自另一项目的两个视角，被试错选这两个选项则判为项目错误）。 行为结果发现，被试在角度水平的正确率显著低于在项目水平的正确率，相应地，视觉工作记忆容量的估计也受到测量精度尺度的影响，一旦考察的记忆精度升高，记忆容量的估计值显著降低。另外，虽然在组水平的视觉工作记忆容量存在材料类别的差异，但视觉工作记忆容量存在稳定的个体差异。同时，借助核磁成像技术记录被试的大脑活动，我们成功在梭状回、海马旁回和腹外侧视觉皮层分别成功定义出材料类别特异的感兴趣区，并找到了对记忆负载敏感的区域作为非类别特异感兴趣区（包括前额叶、顶叶、外侧枕叶皮层等）。更进一步，我们发现在编码和保持阶段，对记忆负载敏感的部分脑区的活动增量能有效预测视觉工作记忆容量的个体差异，为后续的研究提供了一定的神经证据。

Visual working memory is an active system where visual information can be manipulated as well as stored, which underlies the higher cognitive functions of the human brain. And the working memory capacity can predict the individual intelligent scores and (future) education achievements well. However, the visual working memory system shows a severely limited capacity. It’s well received that the average memory capacity of human being ranges from 3 to 4, based on the results of traditional change-detection tasks. But the visual stimuli in daily life are more complicated than the simple laboratory materials (e.g., colors and orientations). So would the behavior results hold on if the real-world objects, rather than the laboratory ones, were used to estimate? Some studies found using the real-world objects results in different working memory capacity estimates. One even reported that compared to simple stimuli, participants can keep more real-world objects. The potential problem, however, is that the memory task demanded a low memory fidelity, the stimuli complicated though. Thus there are a series of open questions? What would the working memory capacity be in the experiment task which demands a high memory precision processing as well as uses real-world complex objects as stimuli? What about different material categories? In addition, it’s supported by majority of studies that the working memory capacity varies across individuals. And researchers have been collecting evidence for individual difference of working memory capacity from behavior to brain activities (mostly based on electroencephalography and functional magnetic imaging) for decades. But the neural basis is not clear yet. Hence, embracing real-world objects as experimental material, a novel test has been designed in this study, aiming to measure human’s visual working memory capacity at different precision level. The materials are three common categories in daily life, including faces, houses and cars. In each trial, one or three pictures were sequentially displayed in the center of screen. Four seconds later, the participants were instructed to pick the target picture from four options. The novelty is that the similarity varies among the three distractors: one distractor is more similar to the target, which is from the same item but a second view. The other two are less similar to the target, which are two views from another item. It’s found that the estimates of capacity differ significant corresponding to different memory fidelity demanding. Participants managed keeping more objects at the item level than the view level. And there is an evident capacity estimate difference across material categories (face, house, car). Meanwhile, the individual difference is stable across categories, resulting a common principal component which could explain 65 percent of the data variance. Combining with fMRI technique, it’s also observed that the brain activities (including PFC, SPL, LOC) during encoding and maintenance period have a close relationship with individual working memory capacity. In summary, this study demonstrates a novel test method of working memory capacity and finds neural evidence for individual difference of visual working memory capacity.