Poster #194 - Socio-cognitive Skills are Linked to Infants’ Selective Word Learning

• In Event: 3-206 - Poster Session 13
  In Poster Session: PS 13 Section - Social Cognition

Integrative Statement

Past research has shown that infants engage in selective social learning (SSL), where they can choose the best source to learn from (see reviews by Mills, 2013 and Poulin-Dubois & Brosseau-Liard, 2016). For instance, infants are more likely to learn a novel word from a reliable speaker over an unreliable one (Brooker & Poulin-Dubois, 2013; Koenig & Woodward, 2010; Krogh-Jespersen & Echols, 2012; Luchkina, Sobel, & Morgan, 2018). However, studies measuring infants’ SSL using a within-subject designs are lacking (but see Schmid et al., 2018). Thus, one of the goals of the present study was to investigate infants’ SSL using a within-subject paradigm. In addition to examining what cues infants use to selectively learn from others, it is important to understand how infants engage in SSL. In fact, the psychological mechanisms of infants’ SSL are currently a subject of controversy (Heyes, 2017; Poulin-Dubois, 2017). One side of the debate posits that infants use domain-specific abilities (e.g., theory of mind - ToM) to selectively learn from others, whereas the leaner side of the debate suggests that infants use domain-general abilities (e.g., associative learning). Previous research has found support for a rich interpretation, demonstrating that infants with superior knowledge inference abilities are less likely to trust unreliable sources of information (Crivello et al., 2017; Crivello & Poulin-Dubois, 2018). An additional goal of this study was to investigate whether domain-specific or domain-general abilities guide infants’ SSL. Eighteen-month-olds (n = 57) were exposed to a reliable and an unreliable speaker, followed by a forced-choice word learning paradigm. Infants then completed a knowledge inference task (Liszkowski, Carpenter, & Tomasello, 2008), a false belief task (Southgate, Chevallier, & Csibra, 2010), and an associative learning task (adapted from Bhat, Galloway, & Landa, 2010). Based on previous research, it was hypothesized that infants with superior socio-cognitive abilities would be less likely to learn a new word from the unreliable informant. No relation between infants’ word learning and associative learning was expected. Results revealed that infants were significantly more likely to learn a new word from the reliable speaker (M = .43, SD = .32) compared to the unreliable speaker (M = .33, SD = .28), t(55) = 2.05, p = .045, d = .34. Furthermore, infants who passed the knowledge inference task (M = .22, SD = .22) were less likely to learn a new word from the unreliable speaker compared to infants who failed (M = .41, SD = .28), t(42) = 2.50, p = .02, d = -.77 (see Figure 1). No link was found with infants’ associative learning and false belief abilities. These results suggest that infants who have superior knowledge inference abilities may be better at inferring that the unreliable speaker was ignorant, and therefore, less likely to learn from her. Moreover, infants are not relying on associations they have made when deciding whom to learn from. Taken together, the findings provide additional support for the rich interpretation and suggest that SSL is a precocious ability in infancy.

Authors

• Diane Poulin-Dubois, Concordia University
  Presenting Author

• Cristina Crivello, Concordia University
  Non-Presenting Author

• Shawna Grossman, Concordia University
  Non-Presenting Author