Congratulations to Abhra Sarkar (assistant professor, Department of Statistics and Data Sciences, The University of Texas at Austin) on his recent National Science Foundation (NSF) award for the project titled, “Novel Statistical Frameworks for Local Inference in Neuroscience of Learning.” Sarkar will serve as principal investigator, alongside co-principal investigator Bharath Chandrasekaran (professor and vice chair for research, Department of Communication Science and Disorders, School of Health and Rehabilitation Sciences, University of Pittsburgh). Learn about this innovative project by reading the abstract below.
Abstract: Learning to make decisions is important in almost all aspects of our lives. How humans learn is a topic of intense study in many diverse fields including psychology, linguistics, and the neurosciences. Learning to categorize is particularly crucial for humans and other animals in our interactions with our environments — whether it is a friend or a foe, edible or non-edible, the word /bat/ or /hat/ etc. As we learn to categorize information, our brain and behavior fundamentally change. These longitudinal trajectories evolve differently in different individuals and across brain systems. While recording and analyzing human behavior and neural function have become sophisticated, statistical methods for characterizing local similarities and differences in the underlying neural and behavioral process dynamics across individuals and groups are still emerging. This project aims to address this critical gap through the development of novel statistical frameworks for complex behavioral data on learning in adults. This collaborative project will provide trainees with experiences in a broad range of interdisciplinary topics.
A common theme underlying the statistical research undertaken in this project is the assessment of the local dynamic behavior of the learning processes — how similar or different these processes are, including especially how similarly or differently they might be influenced by the associated predictors, across learning stages and brain networks. Through innovative adaptation and amalgamation of classical ideas and recent advances in diverse areas in statistics, including sparsity inducing priors, tensor factorization techniques, functional data analysis methods, locally informative sampling strategies, etc., novel generic statistical frameworks and associated computational machinery for local inference in longitudinal settings will first be developed. These frameworks will then be adapted to specific models for neural learning processes. The principal investigators will use carefully designed learning experiments in one domain of research — how adults learn novel speech categories and theories to validate the results obtained by the new statistical models. While human speech learning is an interesting and well-studied problem on its own, insights gained from this application are generalizable to other domains of studies that involve human learning.
This project is part of a joint initiative through the Division of Mathematical Sciences (DMS) in the Directorate for Mathematical and Physical Sciences (MPS) at the National Science Foundation (NSF), and the National Institute of General Medical Sciences (NIGMS) at the National Institutes of Health (NIH) to support research at the Interface of the Biological and Mathematical Sciences. This program is designed to encourage new collaborations, as well as to support innovative activities by existing teams.
More information can be found at the NSF website: Award Abstract #1953712: Novel Statistical Frameworks for Local Inference in Neuroscience of Learning.
