Poster #2 - A Sustained Effect of Language on Thought: Relational Language Enhances Spatial Representations via Visual Attention.

• In Event: Poster Session 09

Abstract

Relational representation is foundational for domains that present systems of relational structures, such as mathematics, spatial cognition, analogical and causal reasoning. Yet, it presents a challenge for young children. Past research has suggested that relational language is particularly suited for bootstrapping relational representation in children (Loewenstein & Gentner, 2005). But researchers disagree on the precise mechanism through which language facilitates relational representation (Dessalegn & Landau, 2013). For example, is the effect of language purely “abstract” (or internal), or is it driven by changes in external sensory processes—particularly, the recruitment of visual attention? Furthermore, does the facilitative effect of relational language persist over time even when language is no longer involved in solving a task? We answer these questions in the context of young children’s representation of spatial relations.

Adopted from Dessalegn & Landau (2008), four-year-olds (N = 44) were given a color- location conjunction task in which they were first shown a spatial configuration (e.g., a red bar on the left of a blue bar). After a brief delay, they were asked to choose the same configuration from its mirror reflection (e.g., a red bar on the right of a blue bar). During the first phase, children in the experimental condition (N = 22) heard relational language (e.g., “Red is on the left of the blue”), while those in the control condition (N = 22) heard generic non-relational language (e.g., “Look at this one, look at it closely”). We found a Direct Effect of language, such that children in the experimental condition had higher accuracy than controls (Fig. 1 top), and they also demonstrated differential attention patterns consistent with the Attentional Shift Account of relational representation (Franconeri et al., 2012). Specifically, compared to controls, children in the experimental condition had longer fixations to the side of the square that was the linguistic target (Fig. 1 bottom).

During the second phase of the study, children in the experimental condition were given non-relational language while those in the control condition were given relational language. This design allowed us to test for a Sustained Effect of language. The experimental group, despite no longer hearing relational language, demonstrated the same advantage in accuracy (Fig. 1 top) and differential visual attention (Fig. 1, bottom) relative to controls when they heard non-relational language.

As shown in Fig. 2, time series analyses further revealed that the experimental group who heard relational language online (during the task) and offline (after experiencing relational language in the first phase) differentially attended to the target side—temporally synced with the onset of relational language. In contrast, hearing non-relational language did not lead to such differentiation in visual attention.

Together, these findings suggest that relational language enhances children’s relational representations by guiding their attention, and it produces an offline effect such that changes in behavior and attention patterns persist over time even when the language itself is no longer present. These results have significant implications for learning other systems of relational concepts, such as mathematics and spatial cognition.

Authors

• Abraham Ybarra, University of Colorado Boulder
  Presenting Author

• Miriam A Novack, Northwestern University

• David H. Uttal, Northwestern University

• Steven Franconeri, Northwestern University

• Lei Yuan, University of Colorado Boulder