Researchers from Weill Cornell Medicine, Cornell Tech, and Cornell’s Ithaca campus recently conducted a study using artificial intelligence (AI) to select and generate visual images to explore how the brain processes what it sees. The study, published in Communications Biology on October 23, involved volunteers looking at images chosen by AI, while their brain activity was recorded using functional magnetic resonance imaging (fMRI).
The research, led by Dr. Amy Kuceyeski and Dr. Mert Sabuncu, aimed to understand how the human visual system responds to different images and whether AI-generated images could be used to study the neuroscience of vision. Using an AI model of the human visual system, the researchers determined that the AI-selected and AI-generated images significantly activated visual processing areas of the brain. This suggests a promising new approach to studying the neuroscience of vision.
The team utilized a dataset of natural images and fMRI responses from human subjects to train an artificial neural network (ANN) to model the human brain’s visual processing system. They then used this model to predict which images across the dataset would maximally activate targeted vision areas of the brain, as well as generating synthetic images to test the same task.
The results indicated that AI-generated images could be useful in probing and improving models of the human visual system. Furthermore, the research demonstrated the potential for individualizing visual system modeling using AI and fMRI methods. The team is currently working on similar experiments using advanced image generation techniques, and they believe their approach could be applied to study other senses such as hearing.
Ultimately, the researchers hope to explore the therapeutic possibilities of this approach by using specific stimuli to alter brain connectivity for potential therapeutic benefits.