Understanding brain circuit dysfunction in amblyopia using large-scale multimodal recordings in a new visuomotor task.

Some theories about brain function consider the brain as a prediction machine. Based on previous experience, our brains may perform an "autocompletion" function during pretty much every thing we do, and perhaps, think. Here we will study amblyopia, or best known as "lazy eye" in a mouse model to better understand how neural circuits change during this condition, but also how our brain's prediction capacity is changed. In this project, we established a visual virtual reality setup for mice. Head-fixed mice run on a threadmil while the running speed is read out to move through a virtual reality projected on monitors in front of the mouse, similar to how humans navigate through a virtual world in a computer game. In this task, the mice need to find a hidden reward location. Mice learn this task remarkably well and succesfully complete this task in about two weeks of training. To test for how our brain predicts the environment around us, we suddenly stop the visual flow of the virtual world while the mice keep running. This creates a mismatch between the running speed and visual feedback. We also performed the first recordings in what we think is a key brain area to encode predictions of the environment, the retrosplenial cortex (RSC). These show that a substantial amount of excitatory, and different subset of inhibitory neurons become active when such a mismatch occurs, possibly revealing a mechanism to detect mismatch signals to update internal models of the brain that predict the world around us.

We aim to establish the role of predictive processing in neurodevelopmental disorders, specifically in one of the most studied contexts, amblyopia. We use a new visuomotor feedback task across high-visual-acuity preclinical species - cats -, rodents, and humans during large-scale activity readout via best-available methods (functional ultrasound imaging reaching 4 cm depth, mesoscale two-photon imaging, hdEEG) to 1) identify brain regions involved in visuomotor prediction in normal subjects, 2) determine the effect of amblyopia on prediction-related activity patterns, 3) develop functional network models to infer which brain regions and associated brain functions are restored or remain perturbed after amblyopia treatment, 4) validate results via optogenetic perturbation experiments in mice, 5) provide an EEG-based biomarker with high discriminative power across normal vision and different stages of amblyopia.