Investigating the Neural Correlates of Visual Working Memory and Individual Differences in Working Memory Capacity by Tomographical Holo-Hilbert Spectral Analysis

Visual working memory (VWM) is a specific memory system that bridges the gap between our information-rich but short-lived perceptual memory and high-capacity but effortful visual long-term memory. In our daily life, we receive a vast amount of visual information, and almost all of the visual information includes multi-feature objects. The proposed study focuses on investigating the neural oscillatory mechanism of Feature-Binding Visual Working Memory (Binding-VWM) via a change detection task and several simple and complex span tasks. By manipulating the memory load (number of items to be remembered) as well as the extent to which some basic processes (e.g. rehearsal, maintenance, updating, controlled search) operate in these tasks, the proposed study may not only reveal the neural oscillatory mechanism of Binding-VWM, but also shed light on theoscillatory pattern that characterizes individual differences in solving a combination of irrelevant and relevant information in VWM. To better understand the neural oscillatory mechanism underlying this cognitive function, the proposed study employs an up-to-date and sophisticated data-analysis technique - the Holo-Hilbert Spectral Analysis (HHSA) proposed by Norden E. Huang (2016, in press, in which I and one co-PI of this project also share the co-authorship). HHSA has been found to show greater sensitivity than traditional approaches in probing nonlinear characteristics (revealed by amplitude and frequency modulations) for brain signals during both the resting state and task-related cognitive processes, especially those relating to VWM. We will also employ a newly patented method - Dynamic EEG Projected Brain Tomographic Imager (deepBTGI, Liang et al., 2015), which is a source-level extension to HHSA, to investigate Binding-VWM. Our preliminary results show that the amplitude modulation revealed by HHSA and deepBTGI is highly correlated with the capacity of VWM both in the proposed change detection task and an operation span task. More importantly, by employing HHSA with the proposed feature-binding complex span task, the proposed study may reveal individual differences in irrelevant information suppression and relevant information representation in VWM. Therefore, we expect that the working memory capacity as well as the oscillatory pattern of a feature-binding complex span task tend to correlate higher with measures of higher-order cognition than does single-feature complex span task. The proposed study is not only theoretically important because it will bring about advances in data-analysis methods for neural oscillatory mechanism in the field of cognitive neuroscience, but also critical to improving the predictive ability of neural models of visual working memory.