Data for each individual scene produce similar plots to the summary
data shown in Fig. 4 (i.e. the content of the scene has little
effect on the relationship between cone ODs and cell count). A
few general observations can be made. First, for positive OD disparities,
increased peak separation increases the cell count. Second,
where there is a non-zero peak separation, a positive OD disparity
is preferable to a negative OD disparity of the same magnitude.
Third, each plot shows a trough, i.e. a locus of ODs that gives the
lowest cell count for any given peak separation. Fig. 5 allows
clearer visualization of the movement of this trough to more negative
OD disparities as peak separation increases. An important
observation is that the trough for any given peak sensitivity falls
along a locus described by a fixed OD disparity. It is therefore primarily
the OD disparity that influences the cell count (and hence
color vision) and not the absolute values of the OD.
3.3. Effect of cone scaling
The default assumption in our model is that the two longer
wavelength sensitive cones of the anomalous trichromat show
the same 2:1 scaling as do L:M in the normal (see Section 4). Since
we cannot empirically justify this assumption, we also considered
the effect of other scaling ratios. Fixed ratios of 1:1 and 1:2 (for L:L0
or M0
:M) do not change the relationships described in this paper.
We also considered the possibility that there is no mechanism that