Neural Correlates of Children's Selective Word Learning

Abstract

Children are selective social learners who resist learning from informants who are either not able or not willing to provide correct information (see Koenig & Sabbagh, 2013; Mills, 2013 for reviews). To date, however, we have a very limited understanding of the underlying cognitive mechanisms by which children’s selective social learning manifests. The few proposals that I know of have been focused on the processes underlying children’s selective word learning. These studies, however, have been restricted to using modified comprehension test questions and word training paradigms to examine the kinds of representations children create for trained novel words. Although informative, findings from these studies do not provide evidence about how, specifically, the word learning process is altered when information about word meanings is provided by knowledgeable versus ignorant sources. To better understand the mechanisms by which children show selective word learning, I recorded children’s event-related brain potentials (ERPs) in response to the presentation of trained novel words. My goal was to assess whether children created semantic representations for novel words that were trained by an ignorant source. To this end, I investigated the N400 component of the ERP, a centro-parietally distributed negative waveform peaking around 400 ms that indexes how meaning-related information is stored in semantic memory. I found that children who were trained novel words by an ignorant source did not integrate the meaning of the novel words into lexical-semantic memory, as evidenced by a tenuous N400 effect, but demonstrated ERP evidence for recognizing the trained novel words. These findings suggest that, when children encounter a word-object link from an ignorant source, they engage a mechanism that specifically disrupts the semantic processes involved in word learning. Furthermore, children who were trained by an ignorant speaker subsequently showed ERP evidence of attenuated attention to novel words relative to children who were trained by knowledgeable speakers. These findings support the idea that children formed a stable attribution of the speaker’s knowledge state and used this information to block their attention to subsequent information by that speaker.