How long does it take to make a good impression? According to new research from the University of Toronto at Scarborough, the correct answer is about 170 milliseconds or 0.17 seconds. Neuroscientists in Assistant Professor Adrian Nestor’s laboratory at the university recreate images from stored electroencephalography (EEG) data. More than a neuroscience parlor trick, the ability to reconstruct what people see from brain waves has potential implications for medicine, forensics, education, and even history.

Published in eNeuro, the research focuses on spatiotemporal EEG data from face stimuli: the time it takes to create and later reconstruct the image of a face from the data. Nestor and postdoc fellow Dan Nemrodov have used both fMRI (functional magnetic resonance imaging) and EEG data to reconstruct facial images. Based on blood flow, fMRI data capture finer details than EEGs. EEG electrical data, however, have finer temporal resolution than fMRIs, measuring activity in milliseconds rather than full seconds – hence the 170-millisecond facial data storage figure. EEG recording equipment is also more portable and less costly than fMRI machines. In the published study, the researchers recorded EEG data while subjects viewed images of faces. The team then ran the stored data through pattern recognition machine learning algorithms to recreate the images. The study results validate EEG technology for image reconstruction, an outcome with wide potential applications.

We’ve written about brainwave monitoring with wearable EEG devices,  including BrainCo’s Focus1 headband and Kokoon Technology’s sleep-sensing headphones. A step forward to EEG wearables that capture time-based image data differs from the BrainCo and Kokoon wearables in degree, not kind. Applications for visual stimuli recapture span a wide range of instances where people are unable or unwilling to accurately communicate what they see, including people with communications disabilities and eyewitnesses to crimes or other occurrences. Imagine the possibility of reconstructing an infant’s visual world or, with additional development in both brain and forensic sciences, the last images perceived by the recently deceased. Additional implications could include live visual information feeds to the blind or people in remote locations based on reconstructed data-matching extracted from constructed sample sets.