Opponent color processing ($g_L{Q}_L-g_M{Q}_M$)

A class of post-receptoral computations comparing signals from different cone populations, for example a gain-weighted L–M difference and an S versus L+M comparison. Such comparisons distinguish spectral differences from shared changes in illumination intensity. Receptor counts, signal gains and later appearance judgments need not vary proportionally.

Stimulus directions distinguished by selective adaptation or physiological response nulls under specified conditions, commonly associated with an achromatic-like combination and two cone-opponent combinations. Their usefulness near a particular background does not make them identical to unique-hue directions or sufficient for every suprathreshold appearance judgment.

If a nonzero weight vector combines cone contrasts as $r={\mathbf{w}}^{\mathsf{T}}\mathbf{c}$, its zero-response contrasts satisfy one linear constraint and form a two-dimensional plane through the adapted origin in three-dimensional contrast space. Any two independent vectors in that plane constrain the direction of the normal $\mathbf{w}$, up to scale. Nonlinearities and response noise limit empirical identification.

The illustrative transform $y_0=c_L+c_M$, $y_1=c_L-c_M$, $y_2=c_S-(c_L+c_M)/2$ has determinant $-2$. Solving gives $c_L=(y_0+y_1)/2$, $c_M=(y_0-y_1)/2$ and $c_S=y_2+y_0/2$. Thus it reorganizes three cone coordinates without adding independent spectral information. Biological response weights and nonlinearities are further empirical questions.