"We finally found a way to see this complex idea in the signal of MRI," said Marcel Just of Carnegie Mellon University. "The correspondence between thinking and brain activity patterns tells us that How are our ideas built?"
The artificial intelligence system shows that the brain consciousness module is constructed by various subsystems instead of the brain.
Scientists have developed a new "brain reading" AI system that can decode and express complex human thoughts by measuring brain activity. The artificial intelligence system shows that the building blocks of the brain's complex thinking structure are composed of various subsystems of the brain rather than language.
"We finally found a way to allow us to see this complex idea in the MRI signal."
The researchers found that the brain's encoding of 240 complex events, such as the yelling and yelling sentences in the experimental scenario, used 42 practically meaningful components, or seemingly reasonable semantics in the brain's nervous system. These sentences are the correspondence between the way of thinking of the brain and the pattern of brain activity. These characteristics include people, scenes, size, social interactions, and body movements. Researchers say that every kind of information is processed in different brain systems. This is the way the brain processes objective information. By measuring the activity of each brain system, this program can determine what type of ideas the brain system is considering.
The researchers said: “One of the great advancements of the human brain is the ability to incorporate individual concepts into complex ideas, not only thinking about bananas, but also thinking of eating bananas with friends at night.â€
"Our approach overcomes the drawbacks of functional magnetic resonance imaging, which tightly combines the signals generated in brain activity from beginning to end, as if two consecutive words were read in one sentence," said Just.
This advancement has made it possible to decode ideas that contain several concepts. This shows how most people's ideas are composed, "only by constantly adding information.
The researchers used a computational model to evaluate how seven individuals had different neurological responses to the 239 sentences used as materials. Then, the program can decipher the 240th feature. They tested each of the 240 sentences in turn, which is called "cross-validation." The researchers said that the model can predict the characteristics of the first subset of sentences picked out, with an accuracy rate of 87%, although the system has never been predicted before. It can also have other uses: to predict the activation status of a previously invisible sentence, only to show its semantic features.
Trolley Speaker,Bluetooth Trolley Speaker,Portable Speaker Trolley,Trolly Speaker With Mic
GUANGZHOU SOWANGNY ELECTRONIC CO.,LTD , https://www.jerry-power.com