Are negative afterimages just filtered perception?

© Riccardo Manzotti, Milan, January 2017.
For any question, please write to riccardo.manzotti@gmail.com.

Hering vs. RGB filtering

The experiment aims to test whether the hue of negative color afterimages is based on Hering's opponent color space or is the background color filtered in the RGB space. The thesis is that negative afterimages are not mental hues opposite to the stimulus. They are the leftovers of the environmental external colors once the stimulus colors have been partially filtered out.

For instance, in contrast with a widespread notion, after a RED stimulus one does not afterimage a GREEN afterimage -- as predicted by Herring-inspired models -- but a CYAN afterimage.

Method

260 subjects were tested with a series of stimuli against different ensuing bakgrounds. The subject were shown two colors (A, B) and then asked to report which one was more similar to what they had just afterimaged. We also tested the effect of different ensuing background hues -- normally afterimages are always tested against a customary grey or white background. In all cases, the subjects reported whether the afterimage was more similar to the hue predicted by traditional Herring-inspired opponent color model or rather to the hue predicted by the proposed model - namely that the afterimage hue is given by

A = B - k S

where A is the afterimage color, B is the ensuingin background hue, k is a scalar, and S is the stimulus color. To avoid any cognitive bias during the experiment, at each trial, the Herring inspired hue and the hue predicted by the suggested model were randomlyswapped. The sequence predicted by the suggested model is ABABABABBA.

Results

The suggested model predicts the actual results with an accuracy of over 90%. In contrast, Herring inspired model fails constantly. Furthermore, the suggest model is able to make predictions based BOTH on the stimulus color AND on the ensuing background color.

The experimental outcome is consistent with the initial hypothesis, namely that afterimages are not mental colors concocted in a mental color space. On the contrary, negative afterimages what one sees when certain color components are, to a certain extent, filtered out.

Materials

Afterimages test experiment dataset

Riccardo Manzotti, www.consciousness.it, December 2016, Milan.