Stephen Grossberg, Arash Yazdanbakhsh
Through “Neon color spreading” color is spread to where there
is no color! See the following demonstration:
In (a), between
the red bars, the white background can be seen, however in (b), adding the
black flankers causes the spread of weak neon red within an illusory circle
that connects the black bar ends. At first glance, it is hard to see; be
patient for a while till you get the effect, after that it becomes easy to
capture even weaker neon effects.
Grossberg and Mingolla in 1985
developed a neural model that could explain neon color spreading. One of the
problems that now we are concerned with can be depicted as the following:
In (a) neon spread
is visible, however in (b) the effect is blocked. Notably, the geometry of (a)
and (b) are the same, therefore, just different coloring upon the same geometry
let or block the neon effect. How can a laminar cortical
model discriminate between these two
situations?
We call the effect of different painting on neon color
spreading or blockade “contrast relation
constraint”. This constraint can be found in
another phenomenon: Transparency
In (a), just the
bottom square can be seen transparent over the top one, in (b) either of squares
can be seen transparent over the other, and in (c), non
of them. Figures (a), (b), and (c) have same geometry of edges; once again “contrast relation constraint” shows its effect on generating or blocking the transparency. How can a laminar cortical model discriminate between these situations? Is it possible to find a unified laminar circuit to explain both transparency and neon color
spreading generation and blockade?
The following two animations give you a more dynamic sense of:

(A) Neon color spreading, where the neon haze
follows the changing color of the middle cross
(Click on the above icon and make the player
screen size the same as the icon size)

B)
Transparency, where its presence or absence follows the contrast
relation change
(Click on the above icon empty corners and
make
the
player screen size the same as the icon size)
We have presented the results of the model in
a few conferences as listed below:
Grossberg S., & Yazdanbakhsh A. (2003a). Laminar cortical dynamics of 3D surface stratification, transparency, and neon spreading.
3rd annual meeting of Vision Sciences Society, FR43, pp. 77
Grossberg S., & Yazdanbakhsh A. (2003b). Laminar Cortical Mechanisms of 3D Surface Processing.
Society for Neuroscience 33rd annual meeting, 339.5