Исправке

The main discovery of the new studies allows us to follow in the human brain activations and on the ground of these activation patterns it can be said how our thoughts are built of.

In comparison to the previous theories, the researchers can not only recognize the regularities in brain activations by thinking but also decode 240 complex events.

According to the study, these patterns have to consist of 42 meaning components.

Each of these 42 neurally plausible features is processed in a separate subsystem of the human brain by measuring the activation in a specific subarea can be recognized the type of thoughts.

The classification model was trained on the data collected from seven adult participants.

The correspondence between the neaurally plausible semantic features and each sentence was computed by using cross-validation method.

On the basis of 239 sentences out of 240, the AI algorithm was able to compute automatically the 240th sentence.

The accuracy in the classification process reached 87%, what shows that the program can recognize successful patterns of an unseen sentence only on the grounds of neutrally semantic features.