Researchers Might Have Found A Way To Read Minds Through MRI

Scientists have been long trying to find out what’s going on inside the human brain. Now, researchers at the University of Texas claim to have created a “decoder” algorithm that is capable of reconstructing what a person is thinking just by observing their brain activity with help of a conventional fMRI scanner.

According to a report from The Scientist, this research could pave the way for more sophisticated brain-computer interfaces (BCI) that can help people who are unable to speak or type. The study is yet to be peer-reviewed.

How Can MRI Machines Read Thoughts?

The team of researchers used MRO machines to observe the changes in blood flow in order to decode the broader sentiment of what the subjects were thinking while listening to over 15 hours of radio stories and podcasts.

The researchers fed this data to an algorithm that, as per the team, can associate these changes in the blood flow with the content the subjects were listening to. The promising results showed that the decoder was able to deduce meaning “pretty well,” according to co-author Alexander Huth, a neuroscientist at the University of Texas.

However, the system isn’t perfect yet, as the decoder often got confused between radio and podcast recordings. Simply put, the algorithm "knows what’s happening pretty accurately, but not who is doing the things, " Huth explained.

Few Shortcomings In The Algorithm

The algorithm also failed to what it learned from one subject’s brain scan semantics and use it for other subjects’ scans. However, the decoder managed to deduce a story when the subjects were shown a silent film, which means the algorithm isn’t restricted to spoken language only.

The findings could come in handy for scientists to better understand the functions of different areas of the brain and how they function to interact with the world.

Neuroscientists who weren’t directly involved in the experiment were impressed to see the algorithm decode the brain functions with efficiency. Sam Nastase, a researcher at the Princeton Neuroscience Institute believes the research is "mind-blowing."

Researchers Also Teaching Robots To Laugh

Scientists at Kyoto University are training AI to laugh when speaking to humans. A team of researchers has created a shared-laughter model that allows an AI tool to detect human laughter and accompany them in the laughter. AI tools are yet to have the nuances of human conversations such as jokes and puns, and training them to do so might not be an easy task.