Numerical discrimination in domestic horses: a spontaneous binary choice test

While domestic horses can be trained to choose the highest numerosity, spontaneous choice tests led to contrasting results, which in part might have depended upon the adopted experimental protocols. To bring further evidence, we carried out two experiments based on binary choice tests (1:4, 1:2, 2:3, 4:6 apple halves). We tested both horses used for schooling ( N = 25) and semi-wild bred horses ( N = 10), using two different arenas. In both experiments, horses did not choose the highest quantity, irrespective of test difficulty. Horses preferentially chose the stimulus located at their left, while no other tested variable affected their choice. The consistency between the results suggests that the lack of discrimination did not depend on test protocols. The strong preference for a given side and high within-individual repeatability indicate that lateralization may affect the results of spontaneous tests.