Models from laboratory experiments often fail to capture the complexity of real-world environments. Naturalistic paradigms in cognitive neuroscience aim to develop models suited for these contexts. Advances in artificial neural networks offer new opportunities to model cognition more effectively. This talk examines whether human brain computations align with those of deep neural networks, with a focus on natural language processing in adults and language development in children. Evidence suggests that shared computational principles imply the brain also relies on overparameterized optimization to learn language. However, differences emerge as humans convey novel ideas. Finally, I will discuss ongoing work using deep learning models to simulate how children acquire language and bridge gaps between artificial and biological systems.