While vision is one of the most studied systems in the brain, the anatomical organization and underlying computational principles of some basic systems are still not well understood. Among these is visual speed estimation. Behavioral experiments in the honeybee have shown that visual speed estimation plays an integral role in visual odometry, grazing landing and the centering response. This estimate is relatively independent of the spatial frequency and contrast of the image, making it unlikely that a direction selective mechanism based on the Hassenstein-Reichardt model can produce a biologically relevant model of speed estimation. In addition, the absence of conclusive electrophysiological evidence of speed sensitive neurons in the early stages of visual processing has further confounded efforts to identify the neuronal basis of speed estimation and the resulting behaviors. As an alternative, we have previously shown that non-directional motion units show a proportiona l response to image speed over a range of spatial frequencies. Extending this work, we used spatial arrays of these non-directional motion units to model the behaviors mentioned above in simulation. A simulated honeybee navigated through an area lined with two arrays of objects or two gratings at a constant speed. The direction of the bee was specified by a turn angle, determined by collating and comparing the outputs of two separate arrays of non-directional motion units. Using this simple framework, we were able to directly reproduce the centering response. In addition, we were able to take the resulting speed output of the model and integrate it to produce an estimate of distance traveled. Coupling these simulated behavioral results with existing recordings from non-directional motion cells in other insects, this model may elucidate the neuronal underpinnings of speed estimation.
J.P. Dyhr and C. M. Higgins, "Computational modeling of visual speed estimation and related behaviors" (poster), Society for Neuroscience annual meeting, Washington, DC, November 2005.