Different sources of the numerical comparison size effect.

In symbolic number comparison tasks, the numerical distance and size effects are largely influenced by the statistical properties of the stimuli. Here, we tested whether nonsymbolic number comparison is also sensitive to the statistical features of the stimuli. We found that unlike in the symbolic comparison task, in the nonsymbolic comparison task, the distance effect was mainly influenced by the values and not by the statistics of the stimuli. We also found that while the size effect was modified by the frequency of the stimuli, the change of the size effect may not have originated in the difficulty of the task (as predicted by the psychophysical models) but in other parameters of the decision process, such as the adjustment of the decision threshold. Additionally, while the symbolic size effect change is stable throughout the session, the nonsymbolic size effect change gradually decreases. These results confirm that nonsymbolic number comparison is mainly driven by a psychophysical process; however, stimulus frequency has an additional role. The results demonstrate again that symbolic and nonsymbolic numerical stimuli are compared qualitatively differently. Diffusion model parameter recovery also demonstrated that the numerical size effect is not a single effect but can have at least three different generators.