On 2016 May 04, Lydia Maniatis commented:

The purpose of the issue of Visual Neuroscience in which this article appeared was to honor the achievements of Davida Teller and also to revisit her critical analysis of the casual use of 'linking propositions.' From the volume intro: “Davida provided critical thinking about criteria for, and approaches to, drawing inferences linking psychophysical phenomena and perception with underlying neural mechanisms (Teller, 1980, 1984; Teller and Pugh, 1983)....It is surprisingly routine today to see claims being made regarding the neural underpinnings of sensory and cognitive phenomena without articulation or even acknowledgment of implicit linking assumptions. With this special issue, Linking Hypotheses in Visual Neuroscience, we seek to...revisit this central challenge in visual neuroscience....”

Unfortunately, Maertens & Shapely's (2013) never make explicit their implicit linking propositions “Linking appearance to neural activity....” When analyzed, these assumptions contain all of the serious problems, including lack of face validity, flagged by Teller (1984). Additional problems include the vagueness of their use of the term “naturalistic,” which is oddly combined with the narrow tailoring of their conclusions to the specific and highly artificial stimuli employed in this study.

Below, I discuss the claims and offer critiques of Maertens & Shapley (2013):

Maertens & Shapely (2013): “The dependence of lightness perception on adjacent and nearby checks could be interpreted as a support for the idea that the neural computations of lightness are strongly influenced by local computations in the primary visual cortex, V1. Several previous studies have shown that responses to visual patterns in V1 resemble lightness perception qualitatively and quantitatively (Kinoshita & Komatsu, 2001; MacEvoy & Paradiso, 2001; Haynes et al., 2004; Paradiso et al., 2006).”

My response: The idea that a particular stimulus situation (“in the contexts we used...with the checkerboard displays”) specifically activates a particular [and fairly peripheral – see below] set of neurons which are then held to be directly responsible for the percept is untenable. How does the visual system decide, a priori, that this or that stimulus will be processed only by these particular neural populations, up to the point of consciousness, and what happens (with respect to perception) to that part of the process when it is supplemented by others? Here is what Teller (1984) had to say about this type of supposition:

First, she refers to the notion that “if psychophysical and physiological data can be manipulated in such a way that they can be plotted on meaningfully similar axes, such that the two graphs have similar shapes, then that physiological phenomenon is a major causal factor in producing that psychophysical phenomenon.” One of the problems with such a linking proposition, accordint to Teller, is “its peripherality: it states a causal or explanatory relationship between the activity of single cells at a peripheral level of the nervous system and a perceptual or behavioral phenomenon, without any accompanying proposal as to how this pattern maintains itself through the system. The proposition includes an implicit appeal to the “nothing mucks it up” proviso (Teller, 1980)....In the absence of any explicit treatment of these problems, [such proposals] would seem to amount to nothing more than a remote homunculus theory: the...stimulus sets up the pattern of activity...that “looks like” [the stimulus] and the homunculus peers down [at it].” Area 17, or V1, is referred to by Teller (1984) as a “relatively peripheral stage of neural processing. In that case, if the properties of Area 17 cells are to be used to explain the elements of our perceptions, a “nothing mucks it up” theory is needed to bridge the gap between the Area 17 cell and the still more central sites...”

Maertens and Shapely (2013) double down on this unviable notion: “Such an interpretation of our data does not preclude that there are significant influences of neural computations beyond V1. There are many phenomena of lightness perception that probably require an explanation in terms of longer distance interactions and higher-order interactions (reviewed in Gilchrist, 2006; Kingdom, 2011). However, in the contexts we used in our experiments, and with the checkerboard displays used (Fig. 1), the nearest-neighbor and next-nearest-neighbor interactions were sufficient to account for most [not all] of the lightness variation (Tables 2 and 3).”

I would add that, given that a small local change in a part of the visual field can alter the structure – including lightness effects – in the entire field, the idea that some stimuli activate only peripheral systems and local interactions and others more global ones is, if possible, even less plausible.

Even in the context of their experiments, the authors walk back their claims about the influence of local contrast, and, again, suggest that different neural populations underlie the perception of different stimuli:

“The data on lightness matches for the transparent condition suggest that more complicated neural computations were employed by the observers when they judged lightness through transparency.” First, I would note that there was no physical transparency in the stimuli – they were all computer generated on a single screen. Second, one must ask, again, at what stage, and why, does the visual system decide to shunt one computer-generated black and white stimulus to be addressed by one peripheral neural process, leading directly to the analogous percept, and another to a different, “more complicated” neural process? Are the activities of V1 supposed to be lost or ignored or altered in some cases, prioritized for conscious consumption in others, and conversely, the effects of higher-level populations shut down, altered, in some and not in others? The decision as to what process to prioritize for what stimulus would necessarily seem to be a high-level decision, because the low-level would not have the “authority” to decide not to send the issue upstairs.

In the end, the authors do nothing but go to a lot of trouble quantify well-known effects in the context of very specific stimuli, from which they draw conclusions that are purely ad hoc, strictly limited to these stimuli, and then construct an ad hoc and casual neural account lacking face validity. They furthermore fail to acknowledge the well-known effects of structure on lightness; by using checkerboards they avoid even basic figure-ground effects. This is a methodological choice which should have been explained and its consequences for drawing conclusions discussed.

The artificiality and of the stimuli makes it even more difficult to understand what the authors mean to imply by the term “naturalistic.” The assertion that stimuli contain a wide range of luminances obviously isn't enough. As mentioned above, the claim that they are in a position to make a generalization about “naturalistic” stimuli in general also conflicts with limiting of their claims to ““in the contexts we used...with the checkerboard displays...”

On 2016 May 04, Lydia Maniatis commented:

The purpose of the issue of Visual Neuroscience in which this article appeared was to honor the achievements of Davida Teller and also to revisit her critical analysis of the casual use of 'linking propositions.' From the volume intro: “Davida provided critical thinking about criteria for, and approaches to, drawing inferences linking psychophysical phenomena and perception with underlying neural mechanisms (Teller, 1980, 1984; Teller and Pugh, 1983)....It is surprisingly routine today to see claims being made regarding the neural underpinnings of sensory and cognitive phenomena without articulation or even acknowledgment of implicit linking assumptions. With this special issue, Linking Hypotheses in Visual Neuroscience, we seek to...revisit this central challenge in visual neuroscience....”

Unfortunately, Maertens & Shapely's (2013) never make explicit their implicit linking propositions “Linking appearance to neural activity....” When analyzed, these assumptions contain all of the serious problems, including lack of face validity, flagged by Teller (1984). Additional problems include the vagueness of their use of the term “naturalistic,” which is oddly combined with the narrow tailoring of their conclusions to the specific and highly artificial stimuli employed in this study.

Below, I discuss the claims and offer critiques of Maertens & Shapley (2013):

Maertens & Shapely (2013): “The dependence of lightness perception on adjacent and nearby checks could be interpreted as a support for the idea that the neural computations of lightness are strongly influenced by local computations in the primary visual cortex, V1. Several previous studies have shown that responses to visual patterns in V1 resemble lightness perception qualitatively and quantitatively (Kinoshita & Komatsu, 2001; MacEvoy & Paradiso, 2001; Haynes et al., 2004; Paradiso et al., 2006).”

My response: The idea that a particular stimulus situation (“in the contexts we used...with the checkerboard displays”) specifically activates a particular [and fairly peripheral – see below] set of neurons which are then held to be directly responsible for the percept is untenable. How does the visual system decide, a priori, that this or that stimulus will be processed only by these particular neural populations, up to the point of consciousness, and what happens (with respect to perception) to that part of the process when it is supplemented by others? Here is what Teller (1984) had to say about this type of supposition:

First, she refers to the notion that “if psychophysical and physiological data can be manipulated in such a way that they can be plotted on meaningfully similar axes, such that the two graphs have similar shapes, then that physiological phenomenon is a major causal factor in producing that psychophysical phenomenon.” One of the problems with such a linking proposition, accordint to Teller, is “its peripherality: it states a causal or explanatory relationship between the activity of single cells at a peripheral level of the nervous system and a perceptual or behavioral phenomenon, without any accompanying proposal as to how this pattern maintains itself through the system. The proposition includes an implicit appeal to the “nothing mucks it up” proviso (Teller, 1980)....In the absence of any explicit treatment of these problems, [such proposals] would seem to amount to nothing more than a remote homunculus theory: the...stimulus sets up the pattern of activity...that “looks like” [the stimulus] and the homunculus peers down [at it].” Area 17, or V1, is referred to by Teller (1984) as a “relatively peripheral stage of neural processing. In that case, if the properties of Area 17 cells are to be used to explain the elements of our perceptions, a “nothing mucks it up” theory is needed to bridge the gap between the Area 17 cell and the still more central sites...”

Maertens and Shapely (2013) double down on this unviable notion: “Such an interpretation of our data does not preclude that there are significant influences of neural computations beyond V1. There are many phenomena of lightness perception that probably require an explanation in terms of longer distance interactions and higher-order interactions (reviewed in Gilchrist, 2006; Kingdom, 2011). However, in the contexts we used in our experiments, and with the checkerboard displays used (Fig. 1), the nearest-neighbor and next-nearest-neighbor interactions were sufficient to account for most [not all] of the lightness variation (Tables 2 and 3).”

I would add that, given that a small local change in a part of the visual field can alter the structure – including lightness effects – in the entire field, the idea that some stimuli activate only peripheral systems and local interactions and others more global ones is, if possible, even less plausible.

Even in the context of their experiments, the authors walk back their claims about the influence of local contrast, and, again, suggest that different neural populations underlie the perception of different stimuli:

“The data on lightness matches for the transparent condition suggest that more complicated neural computations were employed by the observers when they judged lightness through transparency.” First, I would note that there was no physical transparency in the stimuli – they were all computer generated on a single screen. Second, one must ask, again, at what stage, and why, does the visual system decide to shunt one computer-generated black and white stimulus to be addressed by one peripheral neural process, leading directly to the analogous percept, and another to a different, “more complicated” neural process? Are the activities of V1 supposed to be lost or ignored or altered in some cases, prioritized for conscious consumption in others, and conversely, the effects of higher-level populations shut down, altered, in some and not in others? The decision as to what process to prioritize for what stimulus would necessarily seem to be a high-level decision, because the low-level would not have the “authority” to decide not to send the issue upstairs.

In the end, the authors do nothing but go to a lot of trouble quantify well-known effects in the context of very specific stimuli, from which they draw conclusions that are purely ad hoc, strictly limited to these stimuli, and then construct an ad hoc and casual neural account lacking face validity. They furthermore fail to acknowledge the well-known effects of structure on lightness; by using checkerboards they avoid even basic figure-ground effects. This is a methodological choice which should have been explained and its consequences for drawing conclusions discussed.

The artificiality and of the stimuli makes it even more difficult to understand what the authors mean to imply by the term “naturalistic.” The assertion that stimuli contain a wide range of luminances obviously isn't enough. As mentioned above, the claim that they are in a position to make a generalization about “naturalistic” stimuli in general also conflicts with limiting of their claims to ““in the contexts we used...with the checkerboard displays...”