Abstract
When a brief visual target is followed by a second visual stimulus, the perceptual experience is often either of both stimuli superimposed (integration) or of only the second stimulus (backward masking). Recently, Herzog and colleagues (Herzog & Koch, 2001; Herzog, Fahle & Koch, 2001) have shown that a target vernier followed by a grating of non-vernier lines can produce a strikingly different perceptual experience: shine-through. In the shine-through effect, the target is perceived in front of the grating and is brighter and bigger than it would appear without the grating. The shine-through effect is extremely sensitive to spatial stimulus parameters that influence how the elements of the grating group together. We now show that the shine-through effect is a natural by-product of the dynamics of figure-ground segmentation in the 3D-LAMINART neural network model proposed by Grossberg and colleagues (Cao & Grossberg, 2005). Simulations of the model explain how masking by lateral inhibition and contextual grouping leads to a disinhibition of competition between foreground and background depth planes. This disinhibition allows false binocular matches between the target and grating elements to generate a visible representation at the foreground depth plane, which looks like the target and corresponds to the shine-through effect.