Questions we are interested in
Sensorimotor transformation occurs with the corporation among multiple brain areas. Here is one example: what does the brain do to allow a hungry fox to catch a rabbit? First, the little fox is lucky enough to find a rabbit hiding in the woods. Her brain starts to work with multiple threads of neural activities: visual system locates where the rabbit is; then memory system recalls the experience from last hunting and concludes that the chance of success is less than a quarter; however, hyperactivity from hunger center in hypothalamus drives the desire of hunting so badly that she decides to roll the dice; neuromodulatory system, noradrenergic neurons enhance the arousal level and dopaminergic neurons already start to anticipate the rewards; motor system, under the influence of others, has made a plan and finally takes the strike…
We are searching for the basic neural mechanisms underlying sensorimotor transformation: how does each individual brain region process sensory information? How do multiple brain regions exchange processed signals with each other, to produce bound information and generate behaviors? What are the operation principles for the brain at both local and global levels?
Tools we use
To gain a coherent understanding of the brain from local to global, it’s essential to measure neural activities from multiple brain regions simultaneously. We chose larval zebrafish as model animal, because it is the only genetically accessible vertebrate model animal whose brain is small and transparent enough, thus we can image the whole-brain at single cell resolution. Our research paradigm is:
o Performing whole-brain calcium imaging with high spatial and temporal resolution;
o Using virtual reality to promote behaviors during imaging, and capturing the whole-brain activity during sensorimotor transformation;
o Analyzing the big data (hundreds of TB), and identifying the cells whose activities correlate with sensory or motor features;
o Recording the identified cell with in vivo patch clamp, collecting information at synaptic level, and building whole-brain network model, which provides us greater insight into neural mechanisms;
o Model validation with optogenetic or chemogenetic perturbations.
Joy of fish —— specific projects we are studying
In the past, with such research paradigms, we already made discoveries, and some are truly surprising. For example, we found astrocytes, together with neuromodulatory system and inhibitory neurons, mediating “giving up” when behaviors are futile (2019 Cell); we also observed how different sensory inputs converge inside the brain and get categorized (2018 Neuron); we identified a oscillatory circuit guide the animal to organize its behavior when sensory cues are absent (2016 eLife).
We are continuing our research and focusing on binding-problem and decision-making. During daily life, the fish need to bind sensory features to recognize objects and identify preys, siblings, and predators; based on such recognition, they interact with the environment by making decisions for proper swimming patterns, to approach prey, follow the siblings, or avoid the predator.
Sensory-binding and decision-making are important functions we do not fully understand. We are establishing behavior paradigms for each and identifying the critical neural substrates and the underlying neural computation, to uncover the principle operating inside the fish brain.
It’s not easy to understand another brain, even if it's from a fish, as demonstrated by the debate between two Chinese philosophers Zhuangzi and Huizi more than 2000 years ago. They were debating if a fish was happy based on its swimming pattern and verifying the possibility if we can interpret another individual from their behaviors. With the help from newly developed techniques, we will be able to interpret the joy of fish, and establish a comprehensive description from neurons to behavior, and thus understand the neural representation of important brain functions.
Who we collaborate with?
“Progress in science depends on techniques, new discoveries and new ideas, probably in that order.” said Sydney Brenner. Ten years ago, we needed to inject more than a thousand eggs and plenty of good luck, to get the transgenic zebrafish; now the same work can be done within half an hour. However, it still cannot catch up with the pace of molecular tools development ---- when we are close to generating a transgenic fish with new GCaMP, another brighter, faster and more sensitive version of GCaMP has become available already.
It’s hard to make good use of all the available techniques by a single lab. Thus, close collaboration among labs becomes essential, for extending the new techniques to new discoveries, and distilling new ideas from discoveries. In ION there are four labs and one project team focusing on zebrafish, and we will work together to answer these questions: how does the brain give rise to hundreds of functional clusters? How do those clusters connect to each other to form a network? What kind of neural computation is performed by this network? In a more natural environment, how does freely swimming fish use such neural computation? Can we extract the operation principle behind the network and design newer versions of AI? Synergistically, we will achieve a comprehensive understanding of the complicated neural mechanisms from a simple vertebrate’s brain, and such comprehension will be the basis for our exploration of more complicated systems and boost the creation of better AI.