On 2015 Nov 13, Lydia Maniatis commented:

The theory and method of this study simply doesn't hold up to scrutiny. Very serious problems include:

Methods

1. The authors claim to be motivated by previous findings of “large individual differences in lightness constancy with respect to changes in surface slant...” Yet they have only two observers per condition, and one of the observers did not complete all of the replications. Large individual differences are an indication that stimuli are ambiguous to observers, who consequently have to make quasi-guesses. In this type of situation there tends to be both intra- and inter-individual variability. Here, the small number of observers does not make either form of variability detectable.

2. They describe having subjected observers to an “induction procedure (Allred & Brainard, 2009; Doerschner, Boyaci & Maloney, 2004) to familiarize them with the difference between surface reflectance...and the amount of light reflected from a surface [luminance]...” Surface reflectance is not perceived, but inferred (the perceptual correlate is lightness), and luminance has no perceptual correlate. The relative luminances of surfaces in context and their geometrical relationships conspire to produce local impressions of both lightness and apparent illumination. You cannot train observers to perceive luminance, which, again has no perceptual correlate (it does not correspond to white, gray, black; virtually any luminance can take on any of these perceptual values).

An example can help make clear why such a “induction procedure” is theoretically and methodologically hollow. This is the video version of the Adelson checkerboard illusion, in which we see a square being moved from apparent shadow to apparent plain view, and changing apparent color. We can watch the video over and over again without thereby learning to see the square as constant in both its luminance and its reflectance.

1. Asking subjects to rate the similarity in lightness of surfaces on a scale of 1 to 30 seems quite unrealistic.

2. The mathematical acrobatics necessitated by the general shoddiness and ambiguity of the methods are moot.

Theory

The authors state that “Changing the orientation of an object's surface with respect to a directional light source affects the luminance of the reflected light [they mean "the luminance of the surface"] even when the light emitted by the illumination sources is held constant.” That's true.

However, the next statement is incomplete and misleading: “The visual system can stabilize lightness with respect to such geometrical effects...” This statement needs to be highly qualified. The visual system can achieve such “lightness constancy” (requiring disambiguation of reflectance and illumination) only if the image conditions support it. An isolated surface (as that used by the authors), whether fronto-parallel or slanted, will not offer much or any such support. Thus, it should not be surprising that the authors found "essentially no lightness constancy with respect to changing the slant of a test surface..." and that "the dissimilarity data were well-accounted for by a one-dimensional perceptual representation..."

While, again, this outcome was highly predictable, they authors are mystified by it. Noting that their results differ from those of Logvinenko and Maloney (2006), they confess that "We do not know the reason for this difference." But as they note, in that study the illumination differences were not only actual, they were (more importantly) APPARENT (a fact that can be ascertained on the basis of observation), because the setting was less spare and more informative. This is something that should have been considered before, not after data collection.

But lightness facts and relevant literature are not of interest to these authors, who dispense with the matter early on, saying: “This form of lightness constancy [the slant surface form?] is not well-understood.” And on to data collection.

On 2015 Nov 13, Lydia Maniatis commented:

The theory and method of this study simply doesn't hold up to scrutiny. Very serious problems include:

Methods

1. The authors claim to be motivated by previous findings of “large individual differences in lightness constancy with respect to changes in surface slant...” Yet they have only two observers per condition, and one of the observers did not complete all of the replications. Large individual differences are an indication that stimuli are ambiguous to observers, who consequently have to make quasi-guesses. In this type of situation there tends to be both intra- and inter-individual variability. Here, the small number of observers does not make either form of variability detectable.

2. They describe having subjected observers to an “induction procedure (Allred & Brainard, 2009; Doerschner, Boyaci & Maloney, 2004) to familiarize them with the difference between surface reflectance...and the amount of light reflected from a surface [luminance]...” Surface reflectance is not perceived, but inferred (the perceptual correlate is lightness), and luminance has no perceptual correlate. The relative luminances of surfaces in context and their geometrical relationships conspire to produce local impressions of both lightness and apparent illumination. You cannot train observers to perceive luminance, which, again has no perceptual correlate (it does not correspond to white, gray, black; virtually any luminance can take on any of these perceptual values).

An example can help make clear why such a “induction procedure” is theoretically and methodologically hollow. This is the video version of the Adelson checkerboard illusion, in which we see a square being moved from apparent shadow to apparent plain view, and changing apparent color. We can watch the video over and over again without thereby learning to see the square as constant in both its luminance and its reflectance.

1. Asking subjects to rate the similarity in lightness of surfaces on a scale of 1 to 30 seems quite unrealistic.

2. The mathematical acrobatics necessitated by the general shoddiness and ambiguity of the methods are moot.

Theory

The authors state that “Changing the orientation of an object's surface with respect to a directional light source affects the luminance of the reflected light [they mean "the luminance of the surface"] even when the light emitted by the illumination sources is held constant.” That's true.

However, the next statement is incomplete and misleading: “The visual system can stabilize lightness with respect to such geometrical effects...” This statement needs to be highly qualified. The visual system can achieve such “lightness constancy” (requiring disambiguation of reflectance and illumination) only if the image conditions support it. An isolated surface (as that used by the authors), whether fronto-parallel or slanted, will not offer much or any such support. Thus, it should not be surprising that the authors found "essentially no lightness constancy with respect to changing the slant of a test surface..." and that "the dissimilarity data were well-accounted for by a one-dimensional perceptual representation..."

While, again, this outcome was highly predictable, they authors are mystified by it. Noting that their results differ from those of Logvinenko and Maloney (2006), they confess that "We do not know the reason for this difference." But as they note, in that study the illumination differences were not only actual, they were (more importantly) APPARENT (a fact that can be ascertained on the basis of observation), because the setting was less spare and more informative. This is something that should have been considered before, not after data collection.

But lightness facts and relevant literature are not of interest to these authors, who dispense with the matter early on, saying: “This form of lightness constancy [the slant surface form?] is not well-understood.” And on to data collection.