Extended Data Fig. 2: Schematic of perturbation experiment types.

(a) (left) Single fly experiment with whole-brain observation (microscope FOV encompasses all postsynaptic neurons), single neuron stimulation (laser drives filled red neuron), and total rank of stimulation is 1 (only one laser). As a result, all postsynaptic weights from the stimulated neuron are identifiable (blue column of weight matrix filled). (middle) To learn all other effectome weights in this setup would require as many flies as neurons, as each individual neuron is stimulated. (b) Single fly experiment in which all weights are identifiable: whole-brain observation, whole-brain stimulation, and rank of stimulation is equal to the number of neurons in the brain (same number of lasers as neurons). (c) (left) Single fly experiment with partial brain observation (FOV encompasses half of neurons), partial brain stimulation (two neurons red filled), and rank of stimulation is equal to that of the source (2). (middle) To identify all weights of source neurons the experiment is repeated in another fly but with different target neurons. All effectome weights can be identified piecemeal in this manner. (d) Convergence rate to full identifiability of the fly effectome as a function of the fraction of number of experiments over the total number of neurons. Different traces reflect different experimental settings. Columns in legend are three primary ways experiments can vary. Source is the number of neurons that are being stimulated. Target is the number of neurons observed, we assume the source neurons are also observed. Perturbation rank is the dimensionality of the perturbation method. D is the total number of neurons.