An Adaptive Visuomotor Transformation Reservoir Embedded in the Vertebrate Brain
Adaptive visuomotor transformation requires sensorimotor circuits to generate accurate, robust, and flexible outputs, but its underlying cellular mechanism remains poorly understood. Leveraging a zebrafish mesoscopic connectome, we built a biologically constrained spiking neural network of the optic tectum (OT), a conserved vertebrate center for visuomotor transformation, and drove the model with real retinal inputs. Integrating in silico and biological interrogations, we revealed that the OT functions as a biologically structured reservoir complemented by serotonergic systems to implement adaptive visuomotor transformation. Within the tectal reservoir, inhibitory interneurons with layer-matched axonal arborizations suppress task-unrelated pathways to ensure visuomotor accuracy, while excitatory interneurons with deep-layer terminating axons reinforce task-related pathways to enhance noise robustness. Furthermore, visually responsive OT-projecting serotonergic subsystems reweight competing pathways to confer visuomotor flexibility. Thus, our findings delineate how specific interneuron-type-embodied push-pull-like mechanisms coordinate with serotonergic neuromodulation to drive adaptive visuomotor transformation, offering a mechanistic framework for neural computational architectures.