Trading places: Can machine learning provide neuroscience insights without causal understanding?
The convergence of machine learning and neuroscience has sparked a debate over whether AI can advance the field without traditional "understanding."
In a piece for The Transmitter, experts weigh in on how the two disciplines are essentially trading places: neuroscience is increasingly focused on prediction, while machine learning is shifting toward causal explanation, News.Az reports, citing The Transmitter.
Large-scale neural foundation models are now capable of generalizing across different species and brain regions, suggesting that machine-learnable rules govern neural population activity.
RECOMMENDED STORIES
- US drops tax claims against Trump in expanded IRS settlement
- Kyiv National University awards Ilham Aliyev an Honorary Doctor title
- The global race for AI sovereignty: Why countries are fighting to control their artificial intelligence future
- Why Spain’s detention of three Canadian police officers is drawing international attention
However, some researchers, such as Anthony Zador, argue that while AI can find structure in vast datasets and automate analysis, true understanding may require "creating" or replicating the brain's computations—a concept rooted in Richard Feynman’s famous quote, "That which I cannot create, I do not understand." Others suggest that the future of the field, or "NeuroAI," must move beyond pure data processing and embrace "embodiment," acknowledging that brain function is inexorably shaped by the body and its interaction with the physical world.
By Leyla Şirinova
Similar news
