Abstract
If a homogeneous grey patch is surrounded by a dynamic noise background, observers report that the perceptual artificial scotoma fades and is filled in by the dynamic noise from the surround. When the background is switched off, observers report perception of a prolonged patch of twinkling noise in the non-stimulated area (Ramachandran & Gregory, Nature, 1991, 350, 699–702; Hardage & Tyler, Vision Research, 1995, 35, 757–766). These phenomena could be related to active neural processes induced by the surrounding stimulation, which increase the level of internal noise within the visual system. To test this suggestion, we employed the equivalent noise approach in conjunction with the artificial scotoma paradigm.
Observers were presented with Gaussian dynamic noise and a grey patch of 1.5-deg radius centred at the fixation point. They detected a foveal Gabor patch of 4 c/deg (SD 15 min of arc) embedded in Gaussian dynamic noise. Using a 2 interval forced choice method and a staircase procedure, contrast thresholds for detecting the test stimulus were measured in filling-in and after filling-in conditions. The detection threshold increased as the scotoma noise density increased. These functions were shifted to higher contrast levels as the surrounding noise density increased. Observers' performance was analysed by a model for detecting visual patterns. This model took into account sampling efficiency, additive internal noise and multiplicative internal noise components due to the scotoma noise, signal energy and surrounding noise. The results show that in both filling-in and after filling-in conditions, the surrounding noise induces an internal noise component, which approaches the level of the additive internal noise. These findings suggest that dynamic noise may be transferred into internal noise over distances longer than the receptive field size. Thus, we have measured objectively the processes underlying our perception of twinkling noise in the non-stimulated area.