Neural Basis of Number Sense in Larval Zebrafish

Abstract

Number sense, the ability to discriminate discrete quantity, is widespread across animals, yet how this capacity develops remains unknown. Using two-photon light-sheet imaging, we recorded whole-brain activity at single-cell resolution in larval zebrafish (Danio rerio) exposed to controlled visual numerosity stimuli. We discovered that number-selective neurons emerge in a striking developmental sequence: cells tuned to numerosity 1 are already present at 3 days post-fertilization (dpf), and neurons selective for 2, 3, and higher quantities appear in increasing abundance later, at 5 and 7 dpf, accompanied by a relative reduction in 1-tuned cells. The proportion of number-selective cells, summed over all tested numerosities, relative to all identified neurons in the brain, was found to decrease over time. We further showed that a machine-learning decoder based on the activity of the number-selective neurons can predict the number stimulus seen by the animals with accuracies at better than twice chance level. These results reveal how neuronal circuits develop structured numerical codes and provide a framework for studying the emergence of cognitive primitives at cellular resolution.