Our recent study on zebrafish brains may enhance our ability to anticipate brain activity.

AI could one day assist neuroscientists in predicting activity in the human brain, similar to how language models can anticipate the next word in a sentence. After over a decade of research on the neural connections in various organisms, Google Research, Harvard University, and HHMI Janelia have developed the Zebrafish Activity Prediction Benchmark (ZAPBench). This dataset and benchmark aim to enhance researchers’ ability to model brain activity in larval zebrafish.

ZAPBench was created from our 4D recordings of the larval zebrafish brain, capturing the activity of 70,000 neurons as they reacted to different virtual reality stimuli such as changing light and water currents. By predicting brain activity in simpler organisms like zebrafish, we are progressing toward a better understanding of more complex brains, including our own. This work ultimately seeks to clarify fundamental human behavior and the neurological issues that can lead to brain diseases.

Explore how advancements in neuroscience are being made through this innovative research.