Abstract
Numerosity (the set size of items in a group) is essential for behavior and decision making. Selectively tuned neurons to numerosities were found in animals and humans. In humans, these neurons were shown to be organized in a network of topographic maps. Since many visual features (e.g., circumference, area) change with numerosity, numerosity studies usually use simple and well controlled stimuli (e.g., dots in similar size). Here we challenge the ecological validity of these stimuli and ask whether the numerosity-tuned neural populations within the numerosity maps also respond to the numerosity of items present in natural images? 7 participants were scanned in a 7T MRI scanner where they viewed 6 types of stimuli presented in a randomised block design: (1) natural images with 1-3 main objects; (2) natural images with many objects (mean = 19.42, SD= 8.8); (3) natural images of scenery (vague numerosity); (4) 1-3 dots; (5) 20 dots; (6) 10-42 dots. Participants were asked to respond when the same image was presented repeatedly (1N-back task). No numerosity judgment was required. All participants had previously acquired data of their numerosity maps which was used here to localize the neural populations within the maps that are tuned to numerosities of 1-3. We compared their response to low vs. high numerosities as presented in the natural images and dots conditions. We find significantly higher responses to low vs. high numerosities in the 5 maps covering the occipito-temporal and parietal lobes (p<0.05, Wilcoxon signed rank test, FDR corrected). Only the map in the frontal lobe did not show a significant response to the numerosity of objects in the natural images. Our results reinforce the role of tuned neural populations in numerosity perception, expand the ecological validity of numerosity studies and thus grow our understanding of numerosity perception.