Neurocognitive Mechanisms of Selective Social Learning: ERP Studies

• In Event: 1-086 - Cognitive Mechanisms of Selective Learning

Integrative Statement

Children are selective social learners. Instead of indiscriminately learning from any testimony provided by adults, children appear to learn best from speakers who provide information that is likely to be correct and relevant (see Mills, 2013 for a review). Although we know much about the conditions under which children demonstrate selective social learning, we have a limited understanding of the cognitive mechanisms by which children's selectivity manifests, and whether those mechanisms are the same across situations.

There are two possible cognitive mechanisms that might support selective social learning (Sabbagh & Henderson, 2015). One is that children might simply ignore speakers who provide incorrect or irrelevant information. Another is that children might pay attention to and encode information from these speakers but not form semantic representations. Across two studies, we sought to discriminate between these two potential mechanisms by using event related brain potentials (ERPs).

The first study employed a between-subjects design in which 48 6-year-old children experienced novel word training from either a knowledgeable or an ignorant speaker. Children’s ERPs were subsequently recorded as they heard a recording of the speaker using the novel word, followed by a picture of either the object the word was paired with during training (congruent) or a distractor object that was also present during training (incongruent). We predicted that if children simply ignored ignorant speakers, then when novel word training was from ignorant speakers ERPs would be similar in the incongruent and congruent conditions. In contrast if children paid attention to ignorant speakers but did not form a semantic representation, then they would show an early N200 effect that is elicited by detecting an unusual word-referent pairing, but not show a later N400 effect that is typically elicited by detecting a semantic violation. Results were consistent with this second pattern --- that is when word training was performed by ignorant speakers, children showed a reliable N200 effect but not a reliable N400 effect. Thus, when faced with ignorant speakers, selective learning manifests by blocking the formation of a semantic representation.

Following the same logic, we conducted a second study with a between-subjects design in which 48 6-year-olds were taught novel words for objects they were told came from either Japan and were for children from there, or from "Downtown" and were for children like them. After training, ERPs were recorded in a paradigm similar to the previous study. Here we found that although ERPs recorded in the "Downtown" condition showed both an N200 and N400 effects, ERPs recorded in the "Japan" condition exhibited neither a reliable N200 nor N400 effect. These findings suggest that when children are selective on the basis of relevance, they ignore the speaker altogether.

Together these findings show that the mechanisms that underlie selective social learning may vary with the contextual demands. We will conclude with speculations on the adaptive value of this contextual variability, and discuss how these findings can help us understand the fundamental nature of selective social learning.

Authors

• Mark Sabbagh, Queens University
  Presenting Author

• Haykaz Mangardich, Queen's University
  Non-Presenting Author