DEVELOPING A PLATFORM FOR SYSTEMS NEUROSCIENCE IN FREELY BEHAVING MARMOSETS

The common marmoset is a highly promising animal model to study neuronal mechanisms supporting visual cognition. Its highly evolved but small and lissencephalic cortex allows quick transferability of cutting-edge systems neuroscience technologies advanced in rodents, such as high-density electrophysiology and optogenetics. Marmosets demonstrate better cognitive task performance when they are allowed to behave freely without head fixation. However, effective methodologies for systems neuroscience in free-moving marmosets are yet to be established. Here, we developed an integrated system that supports flexible task design, adaptive task control, and temporally precise synchronisation with external devices. We combined a liquid crystal screen with a 162 PPI display and 240 Hz/1 ms refresh rate, an infrared touch sensor, multiple high-speed video cameras, Python-based task control, stereo loudspeakers, and a reward pump. We also connected multifunction I/O cards for recording task-related electrical signals. Our platform presented small, fine-grained visual object stimuli with high temporal accuracy, and detected touches by fingers and nails. Multi-angle filming allowed 3-dimensional gaze estimation during task performance. Task events and cameras were synchronised precisely. Our task programmes allowed tailored, stepwise training processes of complex cognitive tasks with on demand and/or automatic feedback and parameter updates. Animals consistently performed several hundred trials per day with stable touching accuracy and consistent reaction time across sessions. Using this platform, we are currently designing psychophysical tasks to dissociate memory-based and perceptual decision processes. By combining our platform with wireless electrophysiology and optogenetics, we would be able to dissect causal neuronal processing for mnemonic and perceptual judgements.