Our brain is constantly bombarded with sensory information. Far from being overloaded, the brain is a real expert in managing this stream of information. NeuroSpin researchers from the Institut Frédéric-Joliot have discovered the mechanisms of information filtering. By combining magnetoencephalography (MEG) and machine learning algorithms, they have determined the sequence of neuronal operations that allows the brain to specifically select relevant information. The majority of the information is processed and filtered unconsciously by our brain. Within this stream, the relevant information is selected by a three-step operation and distributed to the associative regions of the brain, in order to be memorized.

The researchers measured the brain activity of 15 participants, while the participants had to identify a "target" image in a stream of 10 images per second. The neurobiologists were thus able to observe three successive operations allowing the participants to process and sort the stream of images:

• Even though about ten images are presented every second, each image is analyzed by the sensory areas of the brain for about half a second. This represents a first phase of automatic processing, which is unconscious and effortless for us.
• When participants are asked to pay attention and memorize a particular image, it is not only the 'target' image that is selected, but all images that are still being processed in the sensory regions. The subject's attention will have the effect of amplifying the neuronal responses induced by these images.
• The third step in processing corresponds to the conscious relationship of the subject. Only one of the selected images induces a prolonged cerebral response involving the parietal and frontal regions. It is this image that the subject will indicate as having perceived.

Bombarded by an ever-increasing amount of information, our brain is still able to manage the surplus of data thanks to an automatic, effortless filtering mechanism and a three-step selection process. Technological advances in brain imaging and information science have provided a tremendous boost to neuroscience research, and this study is a good example.