A pair of studies published this week used artificial intelligence to create images that, when shown to monkeys, were able to target specific brain neurons, giving scientists a clearer understanding of the inner workings of the brain.
In one of the studies, published in Science, researchers at MIT's department of brain and cognitive sciences used an artificial neural network model powered by predictive deep learning to reverse-engineer images meant to activate various populations of neurons. When shown to macaques, the images achieved "considerable and highly selective influence" on the targeted neurons.
The other study, published in Cell, saw scientists from Harvard Medical School's department of neurobiology using both a pre-trained neural network model and a genetic algorithm to allow the macaques' own neuronal responses to create the images. The XDREAM algorithm showed each macaque a series of images, which gradually combined and mutated into completely new images based on how the monkeys' neurons reacted.
"We are seeing that the brain is analyzing the visual scene, and driven by experience, extracting information that is important to the individual over time," Carlos Ponce, MD, PhD, co-lead researcher of the latter study, said in a statement. "The brain is adapting to its environment and encoding ecologically significant information in unpredictable ways."
Both teams suggested that their research could not only lead to deeper understanding of the brain, but also improve the development of AI by more accurately mimicking the inner workings of the human brain.
More articles about AI:
Study: Deep learning can predict reactions to thrombolysis
Machine learning used to detect tuberculosis strains' drug resistance
AI-powered kidney disease diagnostic tool receives FDA 'Breakthrough Device' designation