The effect of visual parameters on nonsymbolic numerosity estimation varies depending on the format of stimulus presentation.

The extent to which the approximate number sense is based on the estimation of continuous visual properties has been widely discussed. Some investigators have hypothesized that humans are able to estimate numerosity directly and independently of visual cues. Other investigators have posited that numerosity can be processed only via the estimation of non-numeric visual properties. The latter theory is confirmed by the existence of the congruency effect, that is, greater accuracy in congruent trials where visual properties were positively correlated with numerosity compared with that in incongruent trials. In this study, we tested the assumption that the congruency effect, reflecting the bias in numerosity estimation due to the estimation of visual cues, varies depending on the format of the stimulus presentation and object heterogeneity. The study involved a sample of pupils in Grades 5-9 from Kyrgyzstan (N = 764; 48% girls; mean age = 13.06 years), whereby participants performed a nonsymbolic comparison test in four formats of stimulus presentation: paired/homogeneous, paired/heterogeneous, mixed/homogeneous, and mixed/heterogeneous. Compared arrays of figures might be congruent or incongruent for one visual parameter (convex hull or cumulative area), whereas another visual parameter was held constant for two arrays. The results of generalized linear mixed-effect models demonstrated that the largest congruency effect occurred in a paired format with homogeneous figures. The congruency effect was insignificant in the mixed/heterogeneous format. The results also revealed that the effects of the convex hull and cumulative area varied in different formats of stimulus presentations.