Differences in Children’s Long-Term Memory for Science Concepts Learned Using Structural Alignment

• In Event: 2-084 - Relational Learning in Real-World Contexts: Cognitive, Educational, and Socio-Cultural Factors

Integrative Statement

Previous research has investigated how different types of analogies, such as those using structural alignment, support conceptual development through guiding attention to relevant category information (Christie & Gentner, 2010; Gentner et al., 2011; Goldwater & Gentner, 2015; Orton et al., 2012). Science is one domain in which this type of learning is particularly important. One of the greatest challenges in science learning involves acquiring a thorough understanding of an abstract scientific concept. Structural alignment has been shown to aid children’s learning of science concepts, but primarily after immediate testing only (Vendetti et al., 2015). That is, research has yet to investigate how analogies affect children’s ability to retain concepts over time. The current study addresses this gap in the extant research by examining what types of information children remember and forget about science concepts introduced using structural alignment.
Children between ages 4 and 8 (N = 113, 56 males, Mage=65 months) first completed a Training phase, in which they were taught multiple examples of animal adaptation (e.g. long body parts for reaching food) using structural alignment. Children were prompted to guess the relation between a pair of animals. They were then taught the relation immediately afterwards, and were prompted to repeat the relation they just learned. Following this, children completed the Test phase in one of two randomly assigned conditions. In the Memory condition, children had to recall the same relations with the same animal pairs presented in a different order than the Training phase. In the Generalization condition, children had to generalize the same relations to new animal pairs. Figure 1 illustrates an example of the same animal adaptation in each of the two conditions. This testing occurred either immediately or after a 5-minute delay.
Results showed that children are more likely to correctly identify relations at immediate test rather than delayed test for the Generalization condition (t(55)=-2.548, p=.014). However, this was not true for the Memory condition (t(54)=-1.745, p=.087). Children also correctly identified relations significantly more often when shown the same stimuli than when prompted to generalize to new animal pairs after both immediate (t(56)=-4.275, p<.001) and delayed testing (t(53)=-5.161, p<.001). Figure 2 illustrates the differences in correctly recalled relations across each condition. When examining the type of information children remembered at test, we found that relational similarities (i.e. the function of an adaptation, such as the use of a body part for reaching food) were rapidly forgotten relative to perceptual similarities (i.e. purely visual elements such as long body parts) across all conditions (t(112)=5.641, p<.001), suggesting that children need additional cognitive supports to retain and generalize relations.
These findings shed light on the lower-level cognitive processes that contribute to children’s relational learning of science concepts. In particular, they address the specific types of information children remember and forget during analogical processing. Critically, this work suggests a powerful role of memory mechanisms in analogies and science learning, and suggest that future research is needed to determine which cognitive supports will best improve learning via structural alignment across time.

Authors

• Emma Lazaroff, University of Wisconsin-Madison
  Presenting Author

• Haley Vlach, University of Wisconsin-Madison
  Non-Presenting Author