Domain-general and domain-specific brain activations and networks in visual and auditory working memory
Abstract
Working memory (WM) is a latent cognitive structure that involves active maintenance and manipulation of information for a short time. How items are stored in WM is an important issue that remains controversial. Whereas some researchers hold that different domains recruit different WM storage systems (domain-specific account), some other researchers argue that items of different domains share the same WM storage system (domain-general account). The domain-specific and domain-general theories give rise to distinct predictions of brain activation patterns associated with WM storage. The domain-specific view predicts that different brain regions are involved in WM storage of stimuli from different domains. In contrast, the domain-general view predicts that a common brain region (or regions) is consistently involved in WM storage regardless of stimulus domains. Both predictions are supported by a few empirical findings in the previous literature, and therefore, the domain-specific versus domain-general argument remains unsettled. This dissertation is aimed to provide further neuroscience evidence for the domain-general and domain-specific storage systems through the use of functional magnetic resonance imaging (fMRI). Specifically, this objective is pursued in three related parts. Parts I and II are focused on two new analyses on a previous fMRI data set of visual and auditory WM. In the previous study, our group manipulated WM load to be either pure visual or auditory, or the combination of visual and auditory items in a single experiment. We found that a region in the left anterior intraparietal sulcus (IPS) was sensitive to both visual and auditory memory loads during WM maintenance (Cowan et al., 2011). Visual and auditory WM with the same level of memory load elicited the same level of activation in this brain region. The same activation level in the left anterior IPS, however, does not necessarily imply that the activation pattern in this brain region is the same across domains (e.g., Tamber-Rosenau et al., 2013). In Part I, using the same data set as Experiment 2 in Cowan et al. (2011), the activation pattern in the left anterior IPS was decoded with multivoxel pattern analysis (MVPA). The results showed that the activation pattern in the left anterior IPS was indistinguishable for visual and auditory domains when WM loads were the same across domains, which further supports the argument that a common region in the left anterior IPS supports both visual and auditory WM. In Part II, an exploratory method, constrained principal component analysis (CPCA) was used to explore the domain-specific and domain-general neural networks involved in the same data set (Experiment 2 in Cowan et al., 2011). Analysis revealed evidence of both (1) specific neural networks responsive to either visual or auditory WM, and (2) general neural networks responsive to both visual and auditory WM. Importantly, only a general neural network, which includes a region of the left anterior IPS, was sensitive to WM load during the WM maintenance period, which supports the theory that items from different sensory domains are stored in a unitary WM storage system. Parts I and II were focused on categorical visual and auditory stimuli. However, previous studies found that the estimates of WM capacity for nonverbal sounds were very different from those for categorical visual and auditory stimuli (Li, Cowan, & Saults, 2013). To gain a comprehensive understanding of the brain mechanisms underlying WM maintenance, in Part III we studied the brain activities during a nonverbal auditory WM task. The participants were required to remember 2, 3, 4, 5, or 6 nonverbal sounds and to make a change-detection task after a short delay. The results revealed some evidence that different strategies were used for low and high memory loads and for different stimulus presentation methods. Importantly, the left anterior IPS did not show load-dependent activation across memory loads, indicating that the domain-general system might store abstract, categorical information which was difficult to extract from the nonverbal sounds used in this study. Combining results from all three approaches, the left anterior IPS appears to be part of a neural network for maintenance of abstract, categorical information across stimulus domains. This characteristic of the left anterior IPS supports the existence of a domain-general system for WM storage which stores categorical information.
Degree
Ph. D.
Thesis Department
Rights
OpenAccess.
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 License.