Poster #26 - Children’s Adaptive Strategy Implementation Predicts Matrix Completion Performance

• In Event: 1-169 - Poster Session 04
  In Poster Session: PS 04 Section - Cognitive Processes

Integrative Statement

Relational reasoning, the ability for critical thinking and novel problem solving, is a cognitive skill central to human intelligence that develops across childhood and into adulthood (Ferrer, O’Hare, & Bunge, 2009). Relational reasoning is commonly assessed using matrix completion, in which participants must extract relations, such as size or shape, between items in a 3x3 matrix that is missing one entry and select an item that successfully fulfills the relations in the matrix. Performance on matrix completion in childhood predicts concurrent and future academic achievement and many other positive life outcomes (Gottfredson, 1997). However, little is known about the types of strategies that lead to good or poor matrix completion performance in childhood or whether children alter and adapt strategies in response to more difficult matrix problems. We assessed 6- and 9-year-old children’s (N=40/age) matrix completion performance while tracking eye movements to index strategy use. Similar to previous research in adults, more row-wise and column-wise scanning and lower average frequency in consulting potential answers (Figure 1; Vigneau, Caissie, & Bors, 2006; Hayes, Petrov, & Sederberg, 2011) were associated with better overall performance in 6- and 9-year-olds. Further, strategic indices associated with better performance were implemented significantly more in 9-year-olds than 6-year-olds. These results suggest that successful strategies for matrix completion are similar in children and adults and increase in use across childhood.

To determine whether strategy use led to a correct response on a given trial, we conducted a mixed logistic model assessing the probability of correct response on a given matrix problem if children engaged in row- or column-wise scanning; this model included matrix difficulty as a predictor and allowed random intercepts for participants. The trials with row-/column-wise scanning were significantly more likely to be answered correctly in 6-year-olds; however, this association was not observed in 9-year-olds. An additional mixed model predicting row-/column-wise scanning with matrix difficulty, allowing for random subject intercepts, demonstrated that row-/column-wise scanning was significantly more likely to occur on more difficult matrix problems, suggesting that children adapted their strategy to matrix difficulty. Moreover, adaptive strategy use predicted children’s performance: greater probability of row- or column-wise scanning with increased difficulty was associated with higher accuracy, as indicated by regressing the random coefficients for participants from the mixed model on matrix completion performance (Figure 2). Together, these results suggest that children as young as 6 years old adapt their strategies on matrix completion problems, and across ages, the children who adapt their strategy most to matrix difficulty are those who perform best. These results help to inform the development of children’s relational reasoning ability and highlight a potentially interesting overlap between relational reasoning ability and adaptive strategy use. Further, understanding the development of relational reasoning ability and how children adapt in response to difficulty could provide targets for improving or training relational reasoning and help guide how reasoning skills are assessed in lab and educational settings.

Authors

• Jesse Niebaum, University of Colorado, Boulder
  Presenting Author

• Yuko Munakata, University of Colorado Boulder
  Non-Presenting Author