In­di­vid­u­als un­con­scious­ly per­ceive their en­vi­ron­ment even be­fore birth. And ba­bies as young as three months can al­ready rec­og­nize in­di­vid­ual words in streams of syl­la­bles. This so-called sta­t­ic learn­ing in in­fan­cy shapes us for life. 

So far, re­search has on­ly been done on the changes that oc­cur in the brain af­ter a sta­t­ic learn­ing process. But the cru­cial ques­tion is where the learn­ing process takes place in the first place and how the brain de­codes these ba­sic units that make up lan­guage and vi­su­al pro­cess­ing. To get to the bot­tom of this phe­nom­e­non, we col­lect­ed, an­a­lyzed and eval­u­at­ed da­ta over five years us­ing in­tracra­nial record­ings. Our study par­tic­i­pants were ex­posed to au­di­to­ry and vi­su­al stim­uli pre­sent­ed in strings of syl­la­bles and frac­tal pat­tern se­quences. What they didn’t know : These da­ta streams con­cealed ba­sic units of words and im­age pairs pre­vi­ous­ly un­known to them.

Us­ing neur­al fre­quen­cy tag­ging, we can iden­ti­fy the ar­eas in the brain that re­spond­ed to these hid­den ba­sic units. We al­so in­ves­ti­gat­ed how dif­fer­ent as­pects of the da­ta streams are en­cod­ed. As a re­sult, some ar­eas of the brain en­code on­ly the sta­tis­ti­cal prop­er­ties of the da­ta streams or track the po­si­tion of in­di­vid­ual el­e­ments with­in the ba­sic units. Oth­er ar­eas of the brain, such as the hip­pocam­pus, en­code the ba­sic units as a whole.

The re­sults in­di­cate that sev­er­al com­pu­ta­tion­al sys­tems ex­ist in par­al­lel. By de­cod­ing this com­pu­ta­tion­al frame­work and pre­cise­ly lo­cat­ing the learn­ing process­es in the brain, the sci­en­tists were able to close a sig­nif­i­cant gap in the re­search sur­round­ing sta­tis­ti­cal learning.