“Reading the thoughts” of computers: Decoding brain signals to the speed of perception

Through the use of electrodes in the temporal lobes of patients who were awake, scientists were able to decode brain signals at nearly the speed of perception. Subsequently, analysis of the neural responses of patients in two categories of visual stimuli (images of faces and houses) allowed the researchers to see which images they saw patients and when with accuracy of over 95%.
As stated in a publication of EurekAlert, the research, which was published in PLOS Computational Biology, conducted Rajesh Rao, a neuroscientist at the University of Washington, and the neurosurgeon Jeff Ότζεμαν, UW Medicine, together with his student Kai Miller and colleagues in Southern California and New York.

“We were trying to understand, firstly, how the human brain perceives objects in the temporal lobe, and secondly, how someone could use a computer to extract and predict what someone sees in real time,” explains Rao. The survey involved seven patients with epilepsy from the Harborview Medical Center in Seattle. And the seven were epileptic seizures that were not treated with drugs, and had undergone surgery in which tissues implanted electrodes to identify the “catalysts” that were causing the seizures. The electrodes were connected with a powerful software, which would interpret some important features of the signal, resulting in the export of information that could allow to predict/ assess what he was seeing anyone.
This technique, said Rao, is a step in the direction of the mapping of the brain, from the point of view that could be used to determine in real time which parts of the brain are susceptible to which kinds of information.