We critically review the findings by Wang et al. (2016).
We point out design limitations that undermine the causal claims made.
We emphasize the need to measure change in performance to test causality.
Future directions for research on the causal relationship between the ‘ANS’ and math are outlined.