Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
4036803 | Vision Research | 2005 | 9 Pages |
The direction of a drifting grating can become more easily identified when a stationary, flickering grating, with the same spatial and temporal frequencies, is added to it. This amplification has been accepted as evidence that motion perception depends on the product of visual signals elicited before and after a target changes position, as computed by a Reichardt detector. However, amplification is also consistent with a model in which direction identification depends on the product of detection probabilities before and after the position shift. In this paper, we compare the Reichardt detector with a model of Probability Multiplication. For 2-frame sequences, similar results are predicted by Probability Multiplication and a Reichardt model, in which the performance-limiting noise is early (i.e. it is added prior to signal multiplication). Many new and previously published results are consistent with these predictions. Other results are documented in which the amplification is too large to be consistent with Probability Multiplication. To explain these latter results, Reichardt detectors must have both early and late noises.