Simple speed estimators reproduce MT responses and identify strength of visual illusion

Computational models of vision should not only be able to reproduce experimentally obtained results; such models should also be able to predict the input–output properties of vision. Conventional models of MT neurons are based on the concept of velocity filtering, as proposed by Simoncelli and Heeger (Vis Res 38(5):743–761, 1998). As this report describes, we provide another interpretation of the computational function of MT neurons. An MT neuron can be a simple speed estimator with an upper limitation for correct estimation. Subsequently, we assess whether the MT model can account for illusory perception of “rotating drift patterns,” by which humans perceive illusory rotation (clockwise or counterclockwise rotation) depending on the background luminance. Moreover, to predict whether a pattern causes visual illusion, or not, we generate an enormous set of possible visual patterns as inputs to the MT model: \(8^{8} = 16{,}777{,}216 \). Numerical quantities of model outputs obtained through a computer simulation for 8⁸ inputs were used to estimate human illusory perception. Results of psychophysical experiments demonstrate that the model prediction is consistent with human perception.