For orientation contrast, there is no specific adaptation angle, i.e.,
the network has developed in an environment of all possible angles. In
this case, when the surround is of angle 
, the network
response to a stimulus of angle 
is
in which 
and 
has the same meaning as for orientation
adaptation. Again assuming the maximum likelihood, 
,
the stimulus angle 
at which it is perceived as angle 0, is
derived and shown in figure 5 (left). The solid line is the
theoretical curve and the experimental data come from [23]
and their estimated error is 
. The parameter obtained
through 
fit is the strength of decorrelation feedback:
.
We can derive the peak position 
, i.e., the surrounding angle
at which the orientation contrast is most significant,
For 
, one immediately gets 
. This is in
good agreement with experiments, most people experience the maximum
effect of orientation contrast around this angle.
Our theory predicts that the peak position of the surround angle for
orientation contrast should be constant since the orientation tuning
width 
is roughly the same for different human observers and is
not going to change much for different experimental setups. But the
peak value of the perceived angle is not constant since the
decorrelation feedback parameter 
is not necessarily same,
indeed, it could be quite different for different human observers and
different experimental setups.
