Neuroscientists at the Sainsbury Wellcome Centre at UCL have conducted research that sheds light on how sensory input is transformed into motor action in the brain across multiple regions. The study, published in Nature, utilized Neuropixels probes to simultaneously record from hundreds of neurons in various brain regions while mice engaged in a decision-making task. The researchers found that decision-making is a global process involving multiple brain areas and is coordinated by learning. This insight could be valuable for artificial intelligence research as it may help in developing more distributed neural networks.
The mice in the study were trained to stand still while observing a visual pattern on a screen. In order to receive a reward, the mice had to lick a spout when they detected a sustained increase in the speed of the visual pattern. The researchers designed the task in such a way that the speed of the movement continuously fluctuated, requiring the mice to constantly pay attention and integrate information to make a decision. By training the mice to stand still, the researchers were able to analyze how neurons track random fluctuations in speed leading up to the action, revealing that no single brain region is responsible for integrating sensory evidence and initiating actions.
Through their recordings, the team observed that in naive animals, information related to the visual stimulus was represented in the visual system and a few midbrain regions. However, in mice that had learned the task, the evidence was integrated throughout the brain. This suggests that learning plays a crucial role in how sensory input is processed and decisions are made. Future studies will focus on tracking neurons over time to understand the learning process better and investigate whether specific brain areas act as causal hubs in establishing the connection between perception and action.
One of the questions raised by the study is how the brain incorporates expectations about when a change in the visual stimulus will occur, ensuring that animals only react when the information is relevant. The researchers plan to explore this further using the dataset they have collected. The study was funded by Wellcome awards and the Sainsbury Wellcome Centre’s Core Grant from the Gatsby Charitable Foundation. Overall, this research provides a comprehensive view of how sensory input is processed in the brain, offering insights that could be valuable for understanding brain function in both animals and artificial intelligence systems.