Search Papers | Poster Sessions | All Posters
Poster A20 in Poster Session A - Tuesday, August 6, 2024, 4:15 – 6:15 pm, Johnson Ice Rink
High-resolution tracking of internal model formation using automated live-in training
Rifqi O. Affan1 (), Benjamin B. Scott2; 1Graduate Program for Neuroscience, Boston University, 2Department of Psychological and Brain Sciences and Center for Systems Neuroscience, Boston University
The ability to acquire and use an internal world model to plan actions is a hallmark of general intelligence. How this adaptive mechanism develops throughout learning and which neural mechanisms underlie it are unknown. Here we used an automated training system to collect data 24 hours a day, 7 days a week from rats learning and performing a two-step decision-making task designed to dissociate model-based planning from model-free strategies. The live-in system allowed self-paced, around-the-clock training, accelerating learning in the two-step task. Furthermore, this system allowed behavioral data collection at high temporal resolution and revealed how model-based planning emerges during learning. These data indicate that rats rapidly adopted an internal model of the two-step task and exhibited a model-based planning strategy early in training.