We found that the ability to represent recursion in the visual domain was Roxadustat mw correlated with grammar comprehension, and that this correlation was partially independent from general intelligence. However this effect was not specific to recursion, since grammar comprehension also correlated with embedded iteration. This suggests that grammar comprehension abilities were correlated with a more general ability to represent and process hierarchical structures generated

iteratively, independently of whether these were recursive or not. This result is not completely surprising given that not all syntactic structures in TROG-D are recursive, although all are hierarchical. We also assessed whether there was a more specific correlation between visual recursion and embedded clauses, but found again only a general association with both EIT and VRT. However, it is important to note that TROG-D only includes sentences with one level of embedding, e.g. relative clause (nominative): Der Junge, derdas Pferd jagt, ist dick ‘The boy, who is chasing the horse, is chubby’. Children may potentially use non-recursive representations for these kind of sentences ( Roeper, 2011). Only a task focussed on sentences with several levels of recursive embedding would allow a direct comparison between visual

recursion and syntactic recursion. Despite this limitation, it is interesting that performance on our novel check details visual tasks was correlated with grammar abilities, even when the effects of non-verbal intelligence were taken into account. These correlations could be explained by the existence of shared cognitive resources, independent from non-verbal intelligence, used for the processing of hierarchical structures in both language and visuo-spatial reasoning, or even by the effects of literacy GBA3 (which are partially independent of intelligence) in the processing of hierarchical structures. Interestingly, while individual differences in intelligence predicted VRT and EIT scores both between and within grades, grammatical

comprehension abilities accounted only for differences between grades. Again, this argues in favor of a general age-related maturational influencing the processing of hierarchical structures, occurring between second and fourth grade, which is partially independent from non-verbal intelligence. Furthermore, in our sample, grammar comprehension and non-verbal intelligence were not significantly correlated. Hence, this general maturation process in hierarchical processing cannot be explained solely by the increase of intelligence with age. Future studies with a more comprehensive assessment of grammar (that includes recursion at several levels), and the inclusion of more cognitive tests (assessing cognitive control, attention, etc.) in the experimental procedure could potentially shed more light on a possible relationship between grammar and processing of complex visual structures.