Episodic memory retrieval as attention turned inward

Vanderbilt University, Nashville TN

Investigators

Abstract

Attention and memory are intricately linked in many kinds of human performance across fields such as education, industry, the military, sports, and everyday life. These processes are major sources of individual differences that are often implicated in psychological and neurological disorders. This project offers new insights into the linkage, focusing on the selective nature of attention and memory retrieval, and testing the conjecture that memory retrieval is perceptual attention turned inward. The research develops new experimental procedures that apply methods for studying perceptual attention to the understanding of memory retrieval. It provides new measures -- of the sharpness of the focus of attention on memory, and the ability to resist distraction from competing memories -- that are not available in current methodologies and which are not addressed in extant theories. The research tests new mathematical theories of selective attention and memory retrieval that specify how memory lists are represented and how retrieval processes can access items in the list selectively. The theories show that attention and memory retrieval rely on the same mathematical principles. Computer simulations based on these principles reproduce critical patterns of human behavior. The primary goal of the research is to develop empirically supported computational models that provide deeper insights into the cognitive and neural basis of the interaction between attention and memory. Understanding the computational principles facilitates integration and comparison with current theories of attention and memory that address other phenomena, like memory encoding or dividing attention between inputs or tasks. The materials, code, and data generated by this project are openly available to everyone. A broader goal of the research is to provide a strong empirical and theoretical basis for applied research in education and the workplace and to provide new theory-based measures of individual differences and clinical disorders. The broadest goal of the research is to disseminate the acquired knowledge to a wide range of audiences, from lay people to scientists, through accessible publications and participation in workshops and summer schools. The beneficiaries included undergraduate students from underrepresented groups, who engage in all aspects of the research process, thus encouraging them to pursue careers in science. This project establishes new empirical connections between attention and memory by adapting the Eriksen flanker task (which is primarily used to study perceptual attention) to measure selective memory retrieval. The new episodic flanker task measures the sharpness of the focus of attention on memory and the ability to resist distraction from flanking items. It requires subjects to remember a list of items (e.g., RFCKGL) and presents a probe display that cues a single item (e.g., ##C###). Subjects are asked to indicate whether the cued item was in the same position in the memory list. The sharpness of their focus on memory is measured by presenting non-matching items from the list that differ in their distance to the probed item (e.g., ##K### vs. ##L###). Nearby items that fall within the focus of attention are harder to reject than distant items outside the focus. Resistance to distraction is measured by presenting flanking letters that match (e.g., RFCKGL) or mismatch (STCPXV) the memory list and subsequently observing how effectively subjects can ignore the flankers. Performance is better when cued and flanking letters “point” to the same response and is worse when they point to opposite responses. Flanking letters that match the target improve performance for “yes” responses but impair performance for “no” responses; flanking letters that mismatch the target impair performance for “yes” responses and improve performance for “no” responses. The size of the improvements and impairments reflects the ability to resist distraction. This research establishes new theoretical connections between attention and memory by adapting computational models of memory retrieval to the episodic flanker task, interpreting their retrieval cues as spotlights of attention focused on memory, and asking whether their spotlights predict the observed distance and compatibility effects. It interprets three memory models (noisy coding, position coding, and item coding) as examples of three classic approaches to selective attention (space-based, object-based, and template-based, respectively), connecting a broad range of memory theory and attention theory. The proposed research claims that retrieval and attention are different names for the same computations, explained by the same underlying mathematics. The research evaluates this claim with a broad range of items, from letters and words to colors and pictures, with a range of timescales that encompass both short-term and long-term memory, and with a broad range of tasks that tap the same representations, relating serial recall and cued recall to cued recognition in the episodic flanker task. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.