Continuous visual properties of number influence the formation of novel symbolic representations

Abstract

Numerical symbols are thought to be mapped onto preexisting nonsymbolic representations of number. A growing body of evidence suggests that nonsymbolic numerical processing is significantly influenced by the associated visual properties of continuous quantity (e.g., surface area, density), but their role in the acquisition of novel symbols is unknown. Forty undergraduate students were trained to associate novel abstract symbols with numerical magnitudes. Half of the symbols were associated with nonsymbolic arrays in which total surface area and numerosity were correlated (“congruent”), and the other symbols were associated with arrays in which total surface area was equated across numerosities (“incongruent”). As numbers are represented in multiple formats (words, digits, nonsymbolic arrays), we also tested whether providing auditory nonword labels facilitated symbol learning. Following training, participants engaged in speeded comparisons of the newly learnt symbols. Comparisons were affected by the ratio between the numerosities associated with each symbol, a characteristic marker of numerical processing. Furthermore, comparisons were hardest for large-ratio comparisons of symbols associated with incongruent area and numerosity pairing during learning. In turn, these findings call for the further investigation of visual parameters on the development of numerical cognition.

Keywords:

• Number
• Symbols
• Representation
• Continuous visual properties

Rebecca Merkley is supported by the Clarendon Fund and the Natural Sciences and Engineering Research Council, Canada (NSERC), and Gaia Scerif is supported by a James S. McDonnell Foundation Understanding Human Cognition Award.

Notes

¹For “pure” trials, there was a significant difference between congruent and incongruent trials in reaction time (t(33) = 2.62, p = .013), but it did not quite reach significance for accuracy (t(33) = 2, p = .054). For “mixed” trials, however, there were significant differences between congruent and incongruent trials in both reaction time (t(33) = −9.4, p < .001), and accuracy (t(33) = 3.94, p = .001).