S u m m a r y :
A computer reading brain activity can find out about the particular song to which a person is listening. The new mind-reading study has been published in the journal Scientific Reports.
A Computer Capable of Telepathy
A telepathic computer sounds like something from your usual sci-fi stories, but behold, this one is an actual scientific thing! The brains behind the brain-reading equipment, a team from D’Or Institute for Research and Education, used a Magnetic Resonance (MR) machine to successfully ‘peer’ into the minds of a group of participants to determine the particular song to which each was listening.
A potential application of this technology is to boost brain-computer interfaces to communicate with locked-in syndrome patients.
The study authors also look forward to paving the way for further research on the reconstruction of auditory imagination and inner speech.
Neural Fingerprints of Songs on Human Brains
The participants of the study were to listen to 40 pieces of music of different genres, from classical music to jazz and pop. The MR machine was used to track the neural fingerprints left by each song on the brain of the volunteers. Meanwhile, the computer was working on identifying the specific brain patterns that were forming as a response to each song; the machine considered a number of characteristics pertaining to the music, namely tonality, dynamics, rhythm and timber.
85% Accuracy in Reading Minds
The hypothesis was that the computer would use the brain activity to identify the song the volunteers were hearing. This is a technique known as brain decoding. This was confirmed as the results showed that the computer was 85% accurate in identifying the right song when it was made to choose between two options as to the target song. This is deemed a great success with respect to findings from other studies.
Increasing the Options
Thereafter, when the team increased the number of options available to the computer to 10 (with only one of them being the correct song), it was able to identify the correct song with 74% accuracy.
“Machines will be able to translate our musical thoughts into songs,” says study author Sebastian Hoefle.
Hoefle explains that brain decoding studies constitute an alternative way to gain a better understanding of neural functioning using Artificial Intelligence.
He believes that future research will provide more answers, such as which musical features make some people like a song while others do not, and whether or not our brain is adapted to incline to a certain type of music.