SPLTRAK Abstract Submission
Studying the neuronal substrates of internal models via a novel closed-loop olfactory task for mice
Priyanka Gupta1, Marie Dussauze1,2, Uri Livneh1, Dinu F Albeanu1,2
1Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, United States
2Watson School of Biological Sciences, Cold Spring Harbor, NY, United States

During behavior, sensation and action operate in closed-loop. Movements shape sensory input, and sensory inputs guide motor commands. Through experience, the brain learns the reciprocal relationship between sensory inputs and movements to build internal models that predict the sensory consequences of upcoming actions (sensorimotor predictions). Comparing internal sensory predictions to actual sensory observations generates prediction errors that can be minimized by learning increasingly accurate models of the world. This exchange of sensory inputs and egocentric expectations is at the core of active perception. To study internal models both at behavioral and circuit-level, we developed a novel behavioral task where head-fixed mice are trained to steer the left-right location of an odor source by controlling a light-weight lever with their forepaws. In this manner, 1) we link a precise motor action to well-defined sensory expectations (odor location) and 2) subsequently violate the learnt expectations via online feedback perturbations in trained animals. Expert mice (6 out of 6 trained mice, training period <2 weeks, >90% accuracy, 400-800 trials/session) showed precise movements that were locked to the instantaneous odor feedback during normal closed-loop coupling. However, when sensory feedback was transiently interrupted (halting of odor source) or distorted (displacement of odor source or change in movement gain), movements were initially guided by each animal’s learnt internal model and further, quickly adapted (within few sniffs in single-trials) in accordance with the instantaneous sensory error. We are currently probing activity in olfactory and motor cortex and the olfactory striatum to understand the sensorimotor transformations that enable this behavior.