The Effect of Labels on Children's Category Learning

The Effect of Labels on Children’s Category Learning Catherine A. Best (best.140@osu.edu) Department of Psychology and the Center for Cognitive Science, The Ohio State University 208G Ohio Stadium East, 1961 Tuttle Park Place, Columbus, OH 43210, USA Christopher W. Robinson (robinson.777@osu.edu) Department of Psychology, The Ohio State University 208F Ohio Stadium East, 1961 Tuttle Park Place, Columbus, OH 43210, USA Vladimir M. Sloutsky (sloutsky.1@osu.edu) Department of Psychology and the Center for Cognitive Science, The Ohio State University 208D Ohio Stadium East, 1961 Tuttle Park Place, Columbus, OH 43210, USA Abstract The effect of language on perception of the visual world is an ongoing debate among researchers. According to one account, labels facilitate categorization by highlighting commonalities among labeled objects. Yet according to another account, early in development labels attenuate categorization by interfering with visual processing. In the current study, 4- year-old children were trained on two contrasting categories that were either labeled or presented in silence. Children were trained to discriminate the categories by associating them with a target object (Experiment 1) or with a target label (Experiment 2). Results demonstrated that children were more likely to learn the visual categories when images were presented in silence than when labeled. Furthermore, there was no evidence that children in the label condition reliably learned the categories, casting doubt on the idea that labels facilitate category learning in children. Keywords: Category Learning; Language; Children Introduction Previous research suggests that labels play a useful role in perceptual and conceptual discrimination of visual information early in development. For example, researchers have shown that infants ranging from 3 to 12 months are often better at learning visual categories when objects are associated with labels than when the same visual stimuli are associated with nonlinguistic sounds (Balaban & Waxman, 1997; Fulkerson & Waxman, 2007; Robinson & Sloutsky, 2007; Ferry, Hespos, & Waxman, 2010). And by 18 months, familiar labels have been shown to facilitate learning of more abstract categories like spatial relations (e.g., Casasola, Bhagwat, & Burke, 2009). In addition, there is neurophysiological evidence suggesting that labels may directly influence how the brain processes visual information. Using EEG recordings, researchers have yielded evidence that 12-month-old infants displayed greater cortical responses (e.g., gamma-band activity) when presented with labeled versus unlabeled objects (Gliga, Vloein, & Csibra, 2010). Finally, labels also influence the category structure that infants learn. For example, while looking at an identical set of visual images, 10-month-old infants hearing only one label associated with all exemplars learned one category; whereas, infants hearing two labels learned two categories (Plunkett, Hu, & Cohen, 2008). Conversely, alternate research with 8- and 12-month- olds suggests that labels can attenuate infants’ learning of visual categories when performance was compared to learning of objects presented in silence (Robinson & Sloutsky, 2007). For pre-linguistic infants, labels (and sounds) have been shown to interfere with the categorization of visual input. What underlying mechanisms can account for the differential effects of labels on early categorization, and do these contradictory effects of labels exist later in development for category learning in childhood? One mechanism that has been proposed to account for effects of labels on category learning is that words facilitate categorization by highlighting the commonalities among labeled entities (Fulkerson & Waxman, 2007; Waxman, 2003). As a result, labeling helps children attend to category-relevant information (Waxman, 2004). An alternative idea is that infants and children have difficulty processing multimodal information, with labels and sounds often attenuating visual processing (Robinson & Sloutsky, 2004; Sloutsky & Napolitano, 2003). Therefore by this account, labels should have no facilitative effect above a silent condition and may even overshadow visual processing (Sloutsky & Robinson, 2008). Still, perhaps with development, labels have more influence on category learning as children become more efficient than infants at processing cross-modal information (Robinson & Sloutsky, 2004) and as they become more familiar with the notion that items belong in categories and labels denote these categories (Gelman & Coley, 1991; Gelman & Markman, 1986). In fact, research on the effect of labels on preschool children’s category learning demonstrates that labels invite children to compare commonalities among category members while assessing both commonalities and differences between item pairs (Namy & Gentner, 2002). To investigate the effect of labels on children’s category learning, the current study presented 4-year-olds with two contrasting categories. In Experiment 1, half of the children were trained on category members with labels and half were trained on category members presented in silence.