Abstract
The emerging field of NeuroAI represents a groundbreaking intersection of neuroscience and artificial intelligence, aiming to elevate our understanding of brain function and the algorithms we build inspired by it.
Over the past decades, neuroscience has experienced a technological boom, generating immense datasets that detail the complexities of the brain. Concurrently, the AI revolution has brought forth many neural network models, many believed to mimic aspects of brain function.However, the critical question remains: how can we validate these AI models as accurate representations of neural processes? How would we know that a model is good?
In NeuroAI, validation is challenging because the models typically are not supposed to correspond one-to-one with the brain but rather to carry some of their essence, e.g., similar representations or comparable behavior.This talk addresses this issue, focusing on integrating methods in NeuroAI models with neural data to establish a rigorous, mutual understanding of both artificial and biological neural systems.The goal is to advocate for more focused interdisciplinarity in neuroscience, where model development and experimental research are deeply interconnected, that can enhance both fields.