To be able to effortlessly learn their native language within just a few years, young children must be equipped with remarkable language learning abilities. The ability to extract and generalize abstract rules between non-adjacent elements in predictive sequences is present very early in life. Previous work using event-related potentials (ERPs) revealed that 3- to 4-month-old infants can discriminate grammatical and ungrammatical dependencies following only brief exposure. In contrast, adults only showed grammar learning when they performed an explicit grammaticality judgment task, but not under passive listening (Friederici, Müller, & Oberecker, 2011; Mueller, Friederici & Männel, 2012). This leads to the question of the differences between infant and adult learning. It has been proposed that the delayed maturation of the prefrontal cortex, supporting cognitive control in the adult brain but not yet fully functional in early development, is a major determinant of infant compared to adult learning (Ramscar & Gitcho, 2007). Thus, in infants, associative learning rooted in temporal cortices may work, unsupervised by cognitive control, more effectively than in adults.

To date, there is no evidence yet on the developmental trajectory from purely associative to more controlled learning, or on the underlying neural bases of different learning processes. In the current ERP studies, we will determine when during children’s first three years of life the switch from associative to more controlled learning occurs. Moreover, by means of optical imaging, we will reveal whether the learning of non-adjacent dependencies follows the same trajectories in the linguistic and non-linguistic domain. For the language domain, we will study grammar learning in Italian as a non-native language, identical to the paradigm used in German 4-month-olds (Friederici, Müller, & Oberecker, 2011). Here, preliminary ERP data suggest that at 3 years of age, children have lost the ability of purely associative learning, as they did not show different brain responses to grammatical and ungrammatical dependencies under passive listening. For the non-linguistic domain, will we use tone sequences that mimic the non-adjacent dependency patterns of the linguistic experiment. Parallel findings in both domains as a function of age would suggest a domain-general change of the learning mechanism during development.