Table 1. Population summary of the performance of different LN models of retinal response to colour stimuli.

			High Red Contrast		High Blue Contrast
			Mean	Std. Deviation	Mean	Std. Deviation
Model	LN	Chromatic rotation	0.5201	0.1204	0.5439	0.104
		Achromatic summed	0.483	0.1218	0.5241	0.1021
		2D-fit	0.4993	0.1186	0.5271	0.0921
	GLM	Chromatic rotation	0.5653	0.115	0.5771	0.1253
		Achromatic summed	0.3417	0.2513	0.5153	0.1332

The LN and GLM models that were tested are presented at Figure 3. The stimulus with high blue contrast was a combination of red input (12% contrast) and blue input (24% contrast); the high red contrast stimulus was a combination of red (24% contrast) and blue (12% contrast) – see Figure 1A. The models' performance was quantified for each ganglion cell by the correlation coefficient between the model's output and the average firing rate of the cell. Here we summarize accuracy of the different models for n = 64 ganglions cells, we found that the chromatic rotation LN model is superior over the achromatic summed and 2D-fit LN models.