Year: 2020
Type: dissertation
Abstract: In order to navigate the complex visual world around us, our visual memory systems must constantly manage trade-offs between capacity and resolution. Determining the optimal balance between the amount of retained information and the precision of such representation is critical to serving the goal at hand. In this dissertation, I demonstrate that this trade-off can be strongly influenced by the statistical structure of visual input. Statistical dependencies can arise either through similarity between simultaneously encoded memory items, or through similar or dissimilar visual interference. This dissertation will provide a systematic examination of how statistical covariance of information, measured by visual similarity, influences visual memory representation. Chapters 2 and 3 investigate how the similarity of visual interference can influence the granularity of what is maintained and ultimately recalled from visual working memory. Specifically, I demonstrate that dissimilar interference disrupts the coarse-grained aspects of representation, while interference of intermediate similarity disrupts the fine-grained aspects. Furthermore, my results also suggest that highly similar interference can be attenuated through surround suppression mechanisms. In Chapter 4, I extend this framework to apply to similarity between items simultaneously held in visual working memory. Finally, in Chapter 5, I investigated the neural underpinnings of how the visual memory system handles large sets of correlated items. By using functional magnetic resonance imaging (fMRI) and multi-voxel pattern analysis, I show that the visual system can aggregate multiple similar items into an ensemble representation as early as early visual cortex. Overall, this dissertation looks to further our understanding of how visual memory is affected by, and can adapt to, the statistical structure of the visual environment.
Source: TSpace (University of Toronto)
Author: Sol Z. Sun
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W3151188283