Witnessing in real-time firing patterns of all the neurons in the brain sounds like sci-fi, but this has already been achieved in research labs in recent years. The catch is that the brains being studied are in baby transgenic zebrafish, whose brains are tiny (about half a millimeter across), transparent, and express a fluorescent protein that lights up when neurons are activated. While in principle these activity recordings encapsulate everything the fish is thinking and doing, decoding and understanding the activity signals of hundreds of thousands of neurons is still a challenge.A 6-day-old zebrafish is already equipped with a number of useful behavioral responses. For example, they can navigate towards brighter areas of their habitat (phototaxis), follow moving stripes (optomotor response), quickly avoid a looming object (that could be a predator). Sometimes, these different changes in their visual environment lead to similar motor outputs, in that they either swim forward, turn to the left or turn to the right. In terms of the underlying neural circuits, one could conceive of a simple model, where the information from different sensory inputs are processed through separately parts of the brain before arriving at the motor neurons controlling tail movement. Alternatively, in a “sensory convergence” model, the sensory signals may first converge in a brain region that is upstream of the motor neurons controlling the swimming. This could allow for more nuanced behaviors; for example, the fish could possess the “intention” to turn in one direction, independent of its actual swimming.
In a new study, (PDF) Chen et al. set out to understand how a diverse set of visual behaviors are represented in the brain. Utilizing state-of-the-art light-sheet microscopes, the team recorded functional activity of nearly all the neurons in the brains of larval zebrafish, while concurrently presenting a battery of different visual stimuli and recording swims from the tail. They then developed an analysis framework to decipher this wealth of data and determine which model more accurately describes the zebrafish brain. By algorithmically sifting through hundreds of thousands of neurons, they identified a group of neurons in the anterior hindbrain region (analogous to mammalian pons) that play the role of “intention”. These neurons respond to visual patterns based on likely swimming response, but their instructions are not always successfully carried out. These results also suggest that these anterior hindbrain circuits may play a crucial role in decision-making, though further experiments will be needed to confirm this hypothesis.
Although visual stimuli and swimming behavior are important for understanding the larval zebrafish brain, many neurons in the brain exhibit coherent and meaningful activity devoted to other functions. Such ongoing activity patterns are intriguing since they allow speculations about internal mental processes reminiscent of “thinking”. To detect these internal mental processes, the team developed an automated pipeline that selects and groups neurons into clusters based on similarities in their firing activity. Although the algorithm used no information about the anatomical location of the cells, many of the neurons within clusters were restricted to particular brain regions, and robustly found across multiple fish. The neurons within these clusters likely work together in neural circuits to perform a variety of functions. Indeed, the identified clusters included a rich variety of circuit candidates, some known and some previously uncharacterized, including, for example, neuronal populations that may be controlling jaw and gill movement, ensembles in the olfactory bulb, and an eye-movement control circuit.
Taken together, these results demonstrate the exciting potential of whole-brain imaging. The data and analysis pipeline demonstrated here can open doors for future studies to clearly decipher the workings of brain – from single cell activity to whole-brain circuits. The authors have also made their data and code publicly available to promote such future efforts.
by Xiuye Chen and Aaron T Kuan
