High-Throughput Automatic Training System for Odor-Based Learned Behaviors in Head-Fixed Mice

Understanding neuronal mechanisms of learned behaviors requires efficient behavioral assays. We designed a high-throughput automatic training system (HATS) for olfactory behaviors in head-fixed mice. The hardware and software were constructed to enable automatic training with minimal human intervention. The integrated system was composed of customized 3D-printing supporting components, an odor-delivery unit with fast response, Arduino based hardware-controlling and data-acquisition unit. Furthermore, the customized software was designed to enable automatic training in all training phases, including lick-teaching, shaping, and learning. Using HATS, we trained mice to perform delayed non-match to sample (DNMS), delayed paired association (DPA), Go/No-go (GNG), and GNG reversal tasks. These tasks probed cognitive functions including sensory discrimination, working memory, decision making, and cognitive flexibility. Mice reached stable levels of performance within several days in the tasks. HATS enabled an experimenter to train eight mice simultaneously, therefore greatly enhanced the experimental efficiency. Combined with causal perturbation and activity recording techniques, HATS can greatly facilitate our understanding of the neural-circuitry mechanisms underlying learned behaviors.