Learning and Compression in Human Working Memory

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): An important, yet relatively unexplored aspect of learning and working memory is how, and under what conditions, a perceiver can capitalize on regularities in the environment to remember more information. How can our prior experiences influence how we remember our current experiences? How can such influences be exploited as tools to enhance cognitive processes in those with learning disabilities (e.g., in Autism), or to assist those with memory deficits? To address these issues, this proposal makes use of a powerful form of implicit learning known as visual statistical learning. Previous research has shown that observers are extremely sensitive to regularities in the visual environment (e.g., quickly learning that 'A' is usually followed by 'B'). Surprisingly, this learning is often completely implicit: when asked to explicitly report these regularities observers perform at chance. This implies that visual statistical learning is a powerful process that operates automatically without our intent or conscious control. However, it remains unclear what the benefits of visual statistical learning are for memory processes. In particular, statistical regularities are a form of redundancy, and to the extent that human memory compresses redundant information, learning regularities between objects should enable observers to remember information about more objects. The proposed experiments have three aims: (1) to determine whether visual statistical learning enables observers to compress information and remember more, (2) to determine the "units" of compression, and (3) to determine the "level" at which compression occurs. To investigate these issues, observers will be required to remember simple objects. Over time, some observers will see patterns in the input (e.g., 'A' often occurs with 'B'), while other observers will see random input. To the extent that observers can learn regularities, and compress them to form more efficient memory representations, observers in the patterned group should remember details about more objects. Previous research has not directly explored how visual statistical learning impacts the capacity of working memory, and the proposed studies will provide important insight into the interactions between learning and memory. PUBLIC HEALTH RELEVANCE: An exciting aspect of this proposal is that it promises to inform basic science by studying how statistical learning impacts working memory and cognition, and therefore has potential to increase understanding of mental illnesses with learning or memory related symptoms. For instance, individuals with Autism have certain learning deficits, but it is unknown whether statistical learning, and its interaction with working memory, is impaired in Autism as well. The proposed studies lay the groundwork for future clinical translational research addressing such questions.