Schedule of Events | Symposia

Changes in task representation via association is linked to hierarchical task learning

Poster Session B - Sunday, March 30, 2025, 8:00 – 10:00 am EDT, Back Bay Ballroom/Republic Ballroom

WooTek Lee¹ (woo-tek-lee@uiowa.edu), Eliot Hazeltine¹, JIefeng Jiang¹; ¹University of Iowa

Humans have the ability to learn complex tasks efficiently and generalize those learned tasks to new contexts adaptively. One possible mechanism underlying this ability is hierarchical task representation, which refers to the ability to use simpler tasks as building blocks for complex task learning. To investigate neural mechanisms enabling hierarchical task learning, we conducted a two-day fMRI experiment. On the first day, participants first learned four simple tasks (A, B, C, D), each requiring focus on a distinct feature of a visual stimulus. After that, participants performed two complex tasks, which can be learned by associating two simple tasks they had learned (AB, CD). On the second day, participants repeated the same simple tasks phase, and performed two types of complex tasks, one already learned (AB, CD) and the other consisting of simple tasks that were not associated on day 1. We predict that, from day 1 to day 2, neural representations of associated simple tasks will be either more similar (reflecting association) or more distinct (reflecting separation to reduce interference during multi-tasking). Our preliminary behavioral analysis supported associations between simple tasks, with faster RTs when switching between associated simple tasks on day 2 than day 1. Furthermore, representation similarity analysis (RSA) using fMRI data showed increased pattern similarity between associated simple tasks on day 2 than day 1 in orbito-frontal cortex, while medial temporal lobe showed the opposite trend. Overall, it suggests that complex task learning can be achieved by changing neural representations of constituent simple tasks.

Topic Area: LONG-TERM MEMORY: Skill Learning