Your Brain on YouTube: fMRIs Reverse Engineer the Visual Experience
The past decade has witnessed a huge boom in the innovations made in film technology, and a research team at UC Berkeley thinks they may have discovered the next hottest thing. Comparing what we see in everyday life as a sort of movie, Gallant and his team of researchers put subjects in a fMRI scan for a few hours while showing them either movies or random videos from YouTube. fMRI, or functional magnetic resonance imaging, is a procedure that uses an MRI to measure changes in the flow of blood throughout your brain. For the purposes of their paper, the researchers focused in on the primary visual cortex, which responds to the specificities in videos, such as colors, edges, and textures. However, the primary visual cortex does not have the capabilities of perceiving and conceptualizing what the objects in the video actually are. As a result, after building a model of their subjects’ brains, the research team decoded the activity in the brain and reconstructed the images from that data.
Their results were surprising. Using hundreds of brain scans, the team was able to reconstruct videos that highly resembled the videos themselves. For clips where the primary object was fairly common and recognizable, the reconstructions were more accurate. When the clips showed were more abstract, such as an ink spot turning into text, the reconstructions were slightly more difficult to decipher. Despite this, Gallant believes that this new technology holds immense potential for application in the clinical sphere. For example, it could facilitate communication with patients who are either locked-in or have various types of neurological diseases. It could also be used to decipher various involuntary subjective mental states, such as hallucination or dreaming. However, the technology also holds obvious limits, as fMRI scans can only be taken over the time course of at least a second, rendering movies made from these scans very blurry and limited in resolution optimization. Nonetheless, researches believe that this could be the first step to directly seeing what our brains see and imagine, and perhaps even decades from now, tapping into the very thoughts and intentions that make us human.
Ellie Shin, History, 2016
Learn more at:
The video linked here shows how a team of UC Berkeley researchers (two neuroscientists, a bioengineer, two…www.edn.com