On 2015 Oct 15, Lydia Maniatis commented:

There was absolutely no reason for this study, which tests two highly-flammable straw men. That it was deemed worth doing indicates either ignorance of fundamental principles of perception or an inexplicable failure to take them into account.

What is supposedly being tested is the impossible notion that we are able to estimate the physical illumination of a scene directly. Nobody believes this, because a. it is logically impossible and b. there are mountains of evidence against it. The authors don't suggest that any believes it, saying only that a “natural expression of the illumination-estimation hypothesis is that perceived illumination is determined by [actual illumination].” But there is nothing natural about such a claim, logically or empirically. First, there is no conceivable mechanism by which direction and intensity of (often partially-obstructed and/or multiple) light sources (which, furthermore, we never look at directly) falling on each local surface could be estimated. (This applies to the conditions of this experiment as well). Second, if illumination could be directly evaluated, then interpretation of the lightness of surfaces would not require a global, ratio-based approach, but could be estimated based on local luminance info only, by factoring out the independently-estimated illumination. If illumination could be directly evaluated, then a photograph of sunlight and shadow would completely lack this character if placed under a visible light-bulb evenly-illuminating the entire surface. In general, graphic representations of shadows, transparencies, etc. would be impossible. It is a fundamental fact of visual perception that the visual system uses the light reflected from a scene, not the light impinging on it, to estimate reflectance, illumination and everything else. The authors are, naturally aware of this, as reflected in their casual acknowledgment that “good constancy” under changing illumination “depends critically on the scene manipulations (Kraft and Brainard, 1999)” and when they say that the “poor constancy” they found in this experiment might be to the lack of a “number of [stimulus] cues often taken to support constancy.” Direct estimation of the illuminant would not need the support of such “cues” coming from the stimulus itself. No surprise, then, that the proposed “hypothesis” is falsified. Furthermore, the authors are unable to suggest any plausible interpretation of their results.

A second pseudo-hypothesis tested is that simultaneous contrast is a consequence of perceived illumination differences, even though we know that it arises when illumination appears homogeneous. This was also falsified.

A strange suggestion made (and rejected) in the conclusion of the paper is that “one could attempt to understand our results while preserving the elegance of the [our version of the] illumination estimation hypothesis by replacing the physical luminance ...with a perceived quantity.” I'm not sure what this means, but luminance is not perceived. If it is, then what does it look like? On what info would this evaluation be based?

On 2015 Oct 15, Lydia Maniatis commented:

There was absolutely no reason for this study, which tests two highly-flammable straw men. That it was deemed worth doing indicates either ignorance of fundamental principles of perception or an inexplicable failure to take them into account.

What is supposedly being tested is the impossible notion that we are able to estimate the physical illumination of a scene directly. Nobody believes this, because a. it is logically impossible and b. there are mountains of evidence against it. The authors don't suggest that any believes it, saying only that a “natural expression of the illumination-estimation hypothesis is that perceived illumination is determined by [actual illumination].” But there is nothing natural about such a claim, logically or empirically. First, there is no conceivable mechanism by which direction and intensity of (often partially-obstructed and/or multiple) light sources (which, furthermore, we never look at directly) falling on each local surface could be estimated. (This applies to the conditions of this experiment as well). Second, if illumination could be directly evaluated, then interpretation of the lightness of surfaces would not require a global, ratio-based approach, but could be estimated based on local luminance info only, by factoring out the independently-estimated illumination. If illumination could be directly evaluated, then a photograph of sunlight and shadow would completely lack this character if placed under a visible light-bulb evenly-illuminating the entire surface. In general, graphic representations of shadows, transparencies, etc. would be impossible. It is a fundamental fact of visual perception that the visual system uses the light reflected from a scene, not the light impinging on it, to estimate reflectance, illumination and everything else. The authors are, naturally aware of this, as reflected in their casual acknowledgment that “good constancy” under changing illumination “depends critically on the scene manipulations (Kraft and Brainard, 1999)” and when they say that the “poor constancy” they found in this experiment might be to the lack of a “number of [stimulus] cues often taken to support constancy.” Direct estimation of the illuminant would not need the support of such “cues” coming from the stimulus itself. No surprise, then, that the proposed “hypothesis” is falsified. Furthermore, the authors are unable to suggest any plausible interpretation of their results.

A second pseudo-hypothesis tested is that simultaneous contrast is a consequence of perceived illumination differences, even though we know that it arises when illumination appears homogeneous. This was also falsified.

A strange suggestion made (and rejected) in the conclusion of the paper is that “one could attempt to understand our results while preserving the elegance of the [our version of the] illumination estimation hypothesis by replacing the physical luminance ...with a perceived quantity.” I'm not sure what this means, but luminance is not perceived. If it is, then what does it look like? On what info would this evaluation be based?