Temporal pattern recognition in the human brain: a dual simultaneous processing

Abstract

Pattern recognition is a major scientific topic. Strikingly, while machine learning algorithms are constantly refined, the human brain emerges as an ancestral biological example of such complex procedure. However, how it transforms sequences of single objects into meaningful temporal patterns remains elusive. Using magnetoencephalography (MEG) and magnetic resonance imaging (MRI), we discovered and mathematically modelled an inedited dual simultaneous processing responsible for pattern recognition in the brain. Indeed, while the objects of the temporal pattern were independently elaborated by a local, rapid brain processing, their combination into a meaningful superordinate pattern depended on a concurrent global, slower processing involving a widespread network of sequentially active brain areas. Expanding the established knowledge of neural information flow from low- to high-order brain areas, we revealed a novel brain mechanism based on simultaneous activity in different frequency bands within the same brain regions, highlighting its crucial role underlying complex cognitive functions.