taking an “unbiased” approach (see note below) of modeling the entire fly connectome allows for very quick perturbations of individual neuron’s impact and place in a circuit of interest. you can easily activate one neuron, silence another neuron, or do vast combinatorial experiments to pull out necessity and sufficiency of circuits. the only limit is computational resources i guess. however, the modeling framework provides accurate and testable predictions, saving time and resources when confirming experimentally.
note:
modeling always involves limitations of complexity and assumptions. these should be explicitly stated and explained from the start, so it is important to contextualize results with that in mind. regardless, the simplest model is always the best model