Japanese researchers have announced a breakthrough in creating mental images of objects and landscapes from human brain activity using artificial intelligence (AI) technology. Scientists from the National Institutes for Quantum Science and Technology, another national institute, and Osaka University collaborated to generate rough images, including a recognizable leopard with distinct features and objects like an airplane with illuminated red lights on its wings.
Referred to as “brain decoding,” this technology allows the visualization of perceptual contents based on brain activity and holds potential applications in medical and welfare fields. The team’s findings were recently published in the international scientific journal Neural Networks.
Prior studies demonstrated the reconstruction of images perceived by human participants from brain activity measured through functional magnetic resonance imaging (fMRI), though these were limited to specific domains such as alphabetical letters. Building on these methods, the researchers developed a technology that quantifies brain activity, enabling generative AI to draw images through predictive techniques and recreate complex objects.
In their research, participants were exposed to 1,200 images of objects and landscapes, with the relationship between their brain signals and the images analyzed and quantified using fMRI. The same images were then fed into the generative AI to learn their correspondence with brain activity.
The potential applications of this technology extend to the development of communication devices and enhancing understanding of the brain mechanisms involved in hallucinations and dreams, as suggested by the article.
