Poster #15 - The Role of our Intuitive Number Sense in Catching Errors in Math

• In Event: Poster Session 13

Abstract

Imagine a problem like 57 + 45 = 202. Although you may not know the exact answer without calculating, you probably "feel" that 202 is too high. Where does this feeling of “error detection” originate? In adults, one theorized source of math-related error detection is the Approximate Number System (ANS; Odic & Starr, 2018; Szkudlarek & Brannon, 2017; Wong & Odic, 2021), a universal perceptual number sense available from birth. Indeed, if the ANS is partly the source of error detection in formal mathematics, then this might explain why individual differences in the ANS predict performance on school-taught math (Szkudlarek & Brannon, 2017; Odic & Starr, 2018, Wong & Odic, 2021).
Previous work has shown that adults with a more precise ANS are better able to catch even small errors in mathematical operations (Wong & Odic, 2021). Our study is the first to adapt this design for 5-9 year-old children. We create a novel paradigm for measuring how well children can identify mistakes made in formal math problems either by themselves or a third party (Study 1). We then examine whether individual and developmental differences in this ability are predicted by variation in the ANS, and whether the ability to catch mistakes mediates the correlation between ANS and formal math tests (Study 2).
Study 1. Five- to nine-year-old children (N = 42) are first shown a novel paradigm to assess each their ability to catch mistakes in math. Children watch videos of a puppet verbally solving math problems, with the puppet making mistakes on half the trials, and are asked to decide if the puppet’s response was correct. At a later time, children are asked to solve the same math questions. In addition, we collect individual differences in the ANS through the PANAMATH task (Halberda et. Al, 2012), and working-memory test (Berch et al., 1999) as a covariate. Much like there is a known asymmetry between when children begin producing versus comprehending language (Clark & Hecht, 2003), we predict that there is a parallel gap for math error detection: children will be better at detecting mistakes made by others, even when they mirror mistakes children themselves made. We additionally predict that the magnitude of this gap will be predicted by differences in the ANS, with children with a better number sense also showing better error detection skills.
Study 2. We tested whether error detection ability is a potential mediator between the ANS and formal math. Children (N = 50) ranging from 5 to 9 years (Mean = 6.9; SD = 1.1) complete four tasks: (1) the third party error detection task from Study 1; (2) the ANS acuity task from Study 1; (3) the working-memory task from Study 1; and (4) a formal math assessment (WIAT-3; Pearson, 2009). While data collection is still ongoing, early results show a robust relationship between error detection and the ANS. Future analyses will be conducted to examine whether the error detection task mediates the relationship between the ANS and formal math.

Authors

• Maria Clara Gil Brandao, University of British Columbia (UBC)
  Presenting Author

• Darko Odic, University of British Columbia (UBC)