Measuring the Approximate Number System

TLDR

The approximate number system (ANS) is proposed to enable non‑symbolic quantity representation across infants, children, and adults, and is thought to support symbolic math learning, yet tasks used to probe it are assumed to tap the same underlying system. The study examined the inter‑relationships among six different ANS measures. The authors compared six distinct ANS tasks to assess whether they measure a common system. Across six ANS tasks, adult participants showed no cross‑task correlations and task‑based acuity estimates were unrelated, revealing methodological problems and challenging the notion of a single underlying system.

Abstract

Recent theories in numerical cognition propose the existence of an approximate number system (ANS) that supports the representation and processing of quantity information without symbols. It has been claimed that this system is present in infants, children, and adults, that it supports learning of symbolic mathematics, and that correctly harnessing the system during tuition will lead to educational benefits. Various experimental tasks have been used to investigate individuals' ANSs, and it has been assumed that these tasks measure the same system. We tested the relationship across six measures of the ANS. Surprisingly, despite typical performance on each task, adult participants' performances across the tasks were not correlated, and estimates of the acuity of individuals' ANSs from different tasks were unrelated. These results highlight methodological issues with tasks typically used to measure the ANS and call into question claims that individuals use a single system to complete all these tasks.