Individual differences in relational learning and analogical reasoning: A computational model of longitudinal change

Leonidas A. A. Doumas, Robert G. Morrison, Lindsey E. Richland

Research output: Contribution to journalArticlepeer-review

Abstract / Description of output

Children's cognitive control and knowledge at school entry predict growth rates in analogical reasoning skill over time; however, the mechanisms by which these factors interact and impact learning are unclear. We propose that inhibitory control (IC) is critical for developing both the relational representations necessary to reason and the ability to use these representations in complex problem solving. We evaluate this hypothesis using computational simulations in a model of analogical thinking, Discovery of Relations by Analogy/Learning and Inference with Schemas and Analogy (DORA/LISA; Doumas et al., 2008). Longitudinal data from children who solved geometric analogy problems repeatedly over 6 months show three distinct learning trajectories though all gained somewhat: analogical reasoners throughout, non-analogical reasoners throughout, and transitional-those who start non-analogical and grew to be analogical. Varying the base level of top-down lateral inhibition in DORA affected the model's ability to learn relational representations, which, in conjunction with inhibition levels used in LISA during reasoning, simulated accuracy rates and error types seen in the three different learning trajectories. These simulations suggest that IC may not only impact reasoning ability but may also shape the ability to acquire relational knowledge given reasoning opportunities.
Original languageEnglish
Article number1235
Pages (from-to)1-14
JournalFrontiers in Psychology
Publication statusPublished - 24 Jul 2018

Keywords / Materials (for Non-textual outputs)

  • analogical reasoning
  • relational knowledge
  • inhibitory control
  • development
  • computational modeling
  • cognitive control


Dive into the research topics of 'Individual differences in relational learning and analogical reasoning: A computational model of longitudinal change'. Together they form a unique fingerprint.

Cite this