Optic flow representation in the optic lobes of Diptera: modeling innervation matrices onto collators and their evolutionary implications

A network model of optic now processing, based on physiological and anatomi cal features of motion-processing neurons, is used to investigate the role of small-field motion detectors emulating T5 cells in producing optic flow selective properties in wide-field collator neurons. The imposition of diff erent connectivities can mimic variations observed in comparative studies o f lobula plate architecture across the Diptera. The results identify two fe atures that are crucial for optic flow selectivity: the broadness of the sp atial patterns of synaptic connections from motion detectors to collators, and the relative contributions of excitatory and inhibitory synaptic output s. If these two aspects of the innervation matrix are balanced appropriatel y, the network's sensitivity to perturbations in physiological properties o f the small-field motion detectors is dramatically reduced, suggesting that sensory systems can evolve robust mechanisms that do not rely upon precise control of network parameters. These results also suggest that alternative lobula plate architectures observed in insects are consistent in allowing optic flow selective properties in wide-field neurons. The implications for the evolution of optic flow selective neurons are discussed.