A Comparison of Spiking Statistics in Motion Sensing Neurones of Flies and Monkeys

Abstract

The information processing strategies employed by the brain are constrained by the reliability with which individual neurones encode external stimuli. If repeated presentations of a particular stimulus elicit extremely reliable responses, relatively few neurones are needed to provide an adequate representation of the stimulus. If, on the other hand, responses are highly variable, more neurones and additional processing stages (e.g. averaging) may be necessary to encode the stimulus with equal precision. In their article in this volume, Warzecha and Egelhaaf conclude provocatively that the motion sensing neurone, Hl, of the fly responds to stimuli with considerable reliability, a finding that stands in contrast to the highly variable responses of motion sensing neurones in the middle temporal visual area (MT) of the monkey cerebral cortex. This difference may make a great deal of sense. Invertebrate nervous systems contain far fewer neurones than those of vertebrates; accurate representation of the sensory world with a relatively small number of neurones may therefore be of great importance to invertebrates. Vertebrate nervous systems, on the other hand, may sacrifice fidelity at the single neurone level to obtain the increased computational power afforded by a more densely interconnected nervous system. Because of the explosion of neurone number in the vertebrate central nervous system, fidelity sacrificed at the single neurone level may be preserved by redundant representation and signal averaging (Buracas et al. 1998; Shadlen and Newsome 1998).