Supplementary Material For:

Control of Chromatic Adaptation:
Signals from Separate Cone Classes Interact


Peter B. Delahunt
David H. Brainard
Department of Psychology

University of California, Santa Barbara



Delahunt, P.B. & Brainard, D.H. (2000). Control of chromatic adaptation: signals
from separate cone classes interact. Vision Research, 40, 2885-2903.
Download PDF.

Note about Figure 1 (added December 15, 2007). Figure 1 is not quite what it should have been. Please see here for description. The published (2008) erratum is available here

The following tables provide the matching data for each experiment. The tables are in tab delimited text format. Although the columns may not align when viewed from a browser, after downloading they may be easily read into a spreadsheet. Each table contains the LMS coordinates of the test and match stimuli for the full color and isochromatic images, along with the SEM for each match (computed separately for each cone coordinate).

The following tables provide the matching for the symmetry check condition described in the paper. These matches are not well-suited for testing the independence hypothesis, since the test stimuli are not constrained to be the same for the two contextual images. These data have not been analyzed in detail but are presented here in the interests of completeness.

The following table gives the model fit errors for the Mean, IDD, IDD-Ind, Diagonal, Diagonal-Ind and Null models. The error reported is CIELAB Delta E and the models were fit to minimize this error. All models except for the Diagonal-Ind model are described in the paper. The Diagonal-Ind model is a version of the Diagonal model constrained to satisfy independence. It was derived from the Diagonal model in the same way that the IDD-Ind model was derived from the IDD model.

The following table gives the parameters for the IDD model for the three observers in both image pair conditions. Note that some break points have negative values. In these cases only increment gains apply. These parameters were obtained when the CIELAB Delta E error metric was used to drive the numerical minimization.

The following MATLAB program demonstrates that independence can hold for appearance matches even when threshold experiments suggest that it fails.

Last Modified 09/07/00/00