Animals exhibit a dazzling array of collective behaviors. Recent studies have characterized the neural representations of relevant behavioral variables, like conspecific-location, in such contexts. However, much less is known about the neural mechanisms mediating interactions between individuals in a group, and the sensorimotor transformations responsible for complex group behavior. To advance our understanding of the neural basis of group behaviors, we studied collective motion (‘schooling’) of micro glassfish (Danionella cerebrum). Danionella schooling emerges during the first 8 weeks post-fertilization, concurrently with nervous-system development. This raises the question whether rules governing Danionella movements are simple and ‘reflexive’ (e.g., avoid collisions) or more complex (e.g., follow a specific leader). Rules spanning this range would require markedly different neural circuits for sensorimotor processing. Here, we propose a stochastic model with individual head-direction diffusion and pairwise alignment interactions. By fitting model parameters to behavioral data at different ages, we demonstrate that transitions between distinct stages of schooling: no-schooling, aggregation, and cohesive postural alignment, are driven by a decrease in diffusion, and a dramatic increase in propensity to align with neighbors. Furthermore, this simple model quantitatively accounts for intricate group dynamics: age-dependent distribution of group-sizes, probability of schools merging/splitting, and sorting of mixed-age groups. This suggests that complex Danionella group dynamics do not require sophisticated computations involving individual fish identity and group composition. By contrast, accounting for the distribution of relative spatial positions at ~1 body-length range does require adding a new ‘ingredient’ to the model, pointing to kinematic variables that could be preferentially detected in the developing brain. In summary, by modeling complex, age-dependent collective motion of fish, our study opens a door to understand how neural circuits implement the interactions that underlie group behavior.