Distinct Neural Representations across Discrete and Continuous Non-symbolic Ratios

Relative magnitudes, such as ratios and proportions, are crucial to everyday life. According to the ratio processing system (RPS) theory, ratios across different formats share a common neural code. Prior neuroimaging studies have shown that ratios across different visual formats are processed by overlapping brain areas, thus providing support for this hypothesis. However, overlapping activation found from univariate analyses does not necessarily imply common representations. The aim of this study was to probe the key prediction of the RPS theory by investigating the neural representation of ratios depicted in discrete and continuous magnitudes using multivariate fMRI analyses. Thirty participants completed a delayed ratio comparison task on discrete (sets of dots) and continuous (line lengths) magnitudes while in the MRI scanner. Using representational similarity analysis (RSA), we examined the similarity between ratio representations across high-level variations, such as magnitude type (length and numerosity), and low-level variations, such as variations in line orientation and dot size, throughout the brain, notably in parietal areas. Results did not support a common ratio representation across discrete and continuous magnitudes. Instead, ratios in numerosity and length seemed to be encoded in qualitatively distinct ways. Altogether, results suggest that ratios are not represented in a format-independent way in the brain, thus challenging the RPS theory.