2-104 - Abstract Rule Learning from Speech and Non-Speech Sounds in Infancy

Session Type: Paper Symposium

Integrative Statement

The ability to detect higher-order relations between entities – “abstract rules” – underlies many successes of human cognition, most notably the acquisition of language. Thus, understanding the origins of abstract rule learning in infancy is a paramount challenge for developmental psychologists. However, at present, the infant literature offers a confusing picture: On the one hand, there is evidence that rule learning is a domain-general mechanism. On the other, studies have shown that infants learn rules from speech that they cannot learn from other stimuli, such as tones. Here we present new behavioural, neural, and computational research that advances our understanding of the conditions that support rule learning. Talk 1 uses NIRS to show that speech sounds in both artificial and natural languages can facilitate rule learning even at birth. Talk 2 proposes that infants’ unique expectations about the structure of music and language provide an explanation for why, by 7 months, they learn rules from speech but not tones. From a complementary perspective, Talk 3 reveals that infants effectively learn rules from tones when they are led to believe that tones, like speech, are a communicative signal – suggesting that communicative relevance may enable rule learning. Our discussant, who has developed computational models of rule learning, will address the question of what constraints these findings place on such models. Together, the four presentations provide the most rigorous methodological and theoretical contributions to date toward understanding the nature of rule learning in infancy and the conditions that best support it.

Sub Unit

• Language, Communication

Chair

• Brock Ferguson, Northwestern University

Discussant

• Michael C. Frank, Stanford University

Individual Presentations

• Rule learning abilities at birth - Presenting Author: Judit Gervain, Centre National de la Recherche Scientifique

• Rules as hidden causes of surprising input features - Presenting Author: Colin Dawson, Univ of Arizona; LouAnn Gerken, University of Arizona

• Infants learn abstract rules from non-speech communicative signals - Presenting Author: Brock Ferguson, Northwestern University; Casey Lew-Williams, Princeton University