Comparing brains to deep neural networks
Over the past decade, comparisons between deep neural networks (DNNs) and the human brain have enabled a rigorous investigation of the representations, architectures, and objectives supporting vision and language.
Some of these comparisons validated existing neuroscientific hypotheses across vision and language, while others generated new ones or challenged established views.
Although considerable progress has been made, there is still debate over the validity and interpretability of this framework.
Newly emerging interpretability approaches and cross-modal insights across vision and language promise to enhance rigor and yield unique insights into brain function.
https://www.cell.com/trends/cognitive-sciences/fulltext/S1364-6613(26)00138-5
