On 2016 Oct 28, Lydia Maniatis commented:

It seems that the model being proposed by Snow et al (2016) was obsolete even as it was being constructed, as they decided to base it on assumptions of V1 neuron behavior known for decades to be invalid.

The choice is signaled by the authors when they explain in their introduction that they are going to “address primary visual cortex (V1) as a paradigmatic example and focus on orientation adaptation phenomena that are within the classical receptive field (RF).”

I think this is the only mention made in the article of the term “classical receptive field,” but it bears some elaboration.

The obsoleteness of the concept is reflected in quotes from two articles, below.

1. Graham (2011) "Beyond multiple pattern analyzers modeled as linear filters (as classical V1 simple cells): Useful additions of the last 25 years" Vision Research.

“The classical receptive field of a V1 simple cell is very small relative to the distances over which visual perception has to function. Indeed, the classical receptive field is typically composed of only a few inhibitory and excitatory sub-sections…

…non-classical responses of V1 simple cells can occur over a substantially larger area than the classical receptive field. This is one reason that non-classical receptive fields are now frequently invoked in explanations for perceptual phenomena.

…these non-classical receptive fields have been invoked to account for a number of psychophysical phenomena as well

…In addition to the possibility of non-classical suppressive (or facilitatory) effects from outside the classical receptive field, there is also a possibility that the same non-classical effects extend inside the classical receptive field. And perhaps there are different non-classical processes that exist inside the classical receptive field but not outside.”

1. Angelucci and Bullier (2003) Reaching beyond the classical receptive field of V1 neurons: horizontal or feedback axons? (Journal of Physiology)

“It is commonly assumed that the orientation-selective surround field of neurons in primary visual cortex (V1) is due to interactions provided solely by intrinsic long-range horizontal connections. We…conclude that horizontal connections are too slow and cover too little visual field to subserve all the functions of suppressive surrounds of V1 neurons in the macaque monkey. We show that the extent of visual space covered by horizontal connections corresponds to the region of low contrast summation of the receptive field center mechanism. This region encompasses the classically defined receptive field center and the proximal surround. Beyond this region, feedback connections are the most likely substrate for surround suppression. We present evidence that inactivation of higher order areas leads to a major decrease in the strength of the suppressive surround of neurons in lower order areas, supporting the hypothesis that feedback connections play a major role in center–surround interactions.”

Naturally, as the authors point out, the model can’t explain many things (that it should be able to explain):

“We have shown that our modeling approach can explain some classical adaptation effects as well as a more recent phenomenon of equalization. However, clearly the approach has limitations and does not capture the full set of phenomena for adaptation.

First, the model in its present form does not include surround influences and cannot capture disinhibition of the surround nor interesting data on facilitation and attractive shifts of tuning curves (Solomon & Kohn, 2014; Webb et al., 2005; Wissig & Kohn, 2012). It would be interesting to consider extensions of the model to capture both spatial (Coen-Cagli et al., 2012) and temporal contextual influences.”

The model is extremely involved. Evaluating it would require a significant amount of effort and time on the part of colleaugues. SInce even its authors already know its assumptions are false (i.e. lead to false predictions), why should anyone bother? Adding to a failed, ad hoc model is usually not the best way to achieve a better one.

The publication of models based on invalid assumptions and empirically falsified a priori is common in the vision literature and reflects a culture in which piecemeal, ad hoc (with respect to facts and techniques) and logically and/or empirically false models are treated as equivalent to coherent and insightful hypotheses that first have to be tested before we know they’ve failed.

The former, being much easier than the latter, overwhelmingly dominates the literature, which has become almost entirely self-referential and unprogressive, because its “theorists” don’t allow themselves to be challenged by inconvenient facts, but rather are satisfied with avoiding them, occupying themselves instead with mathematical prestidigitations.

On 2016 Oct 28, Lydia Maniatis commented:

It seems that the model being proposed by Snow et al (2016) was obsolete even as it was being constructed, as they decided to base it on assumptions of V1 neuron behavior known for decades to be invalid.

The choice is signaled by the authors when they explain in their introduction that they are going to “address primary visual cortex (V1) as a paradigmatic example and focus on orientation adaptation phenomena that are within the classical receptive field (RF).”

I think this is the only mention made in the article of the term “classical receptive field,” but it bears some elaboration.

The obsoleteness of the concept is reflected in quotes from two articles, below.

1. Graham (2011) "Beyond multiple pattern analyzers modeled as linear filters (as classical V1 simple cells): Useful additions of the last 25 years" Vision Research.

“The classical receptive field of a V1 simple cell is very small relative to the distances over which visual perception has to function. Indeed, the classical receptive field is typically composed of only a few inhibitory and excitatory sub-sections…

…non-classical responses of V1 simple cells can occur over a substantially larger area than the classical receptive field. This is one reason that non-classical receptive fields are now frequently invoked in explanations for perceptual phenomena.

…these non-classical receptive fields have been invoked to account for a number of psychophysical phenomena as well

…In addition to the possibility of non-classical suppressive (or facilitatory) effects from outside the classical receptive field, there is also a possibility that the same non-classical effects extend inside the classical receptive field. And perhaps there are different non-classical processes that exist inside the classical receptive field but not outside.”

1. Angelucci and Bullier (2003) Reaching beyond the classical receptive field of V1 neurons: horizontal or feedback axons? (Journal of Physiology)

“It is commonly assumed that the orientation-selective surround field of neurons in primary visual cortex (V1) is due to interactions provided solely by intrinsic long-range horizontal connections. We…conclude that horizontal connections are too slow and cover too little visual field to subserve all the functions of suppressive surrounds of V1 neurons in the macaque monkey. We show that the extent of visual space covered by horizontal connections corresponds to the region of low contrast summation of the receptive field center mechanism. This region encompasses the classically defined receptive field center and the proximal surround. Beyond this region, feedback connections are the most likely substrate for surround suppression. We present evidence that inactivation of higher order areas leads to a major decrease in the strength of the suppressive surround of neurons in lower order areas, supporting the hypothesis that feedback connections play a major role in center–surround interactions.”

Naturally, as the authors point out, the model can’t explain many things (that it should be able to explain):

“We have shown that our modeling approach can explain some classical adaptation effects as well as a more recent phenomenon of equalization. However, clearly the approach has limitations and does not capture the full set of phenomena for adaptation.

First, the model in its present form does not include surround influences and cannot capture disinhibition of the surround nor interesting data on facilitation and attractive shifts of tuning curves (Solomon & Kohn, 2014; Webb et al., 2005; Wissig & Kohn, 2012). It would be interesting to consider extensions of the model to capture both spatial (Coen-Cagli et al., 2012) and temporal contextual influences.”

The model is extremely involved. Evaluating it would require a significant amount of effort and time on the part of colleaugues. SInce even its authors already know its assumptions are false (i.e. lead to false predictions), why should anyone bother? Adding to a failed, ad hoc model is usually not the best way to achieve a better one.

The publication of models based on invalid assumptions and empirically falsified a priori is common in the vision literature and reflects a culture in which piecemeal, ad hoc (with respect to facts and techniques) and logically and/or empirically false models are treated as equivalent to coherent and insightful hypotheses that first have to be tested before we know they’ve failed.

The former, being much easier than the latter, overwhelmingly dominates the literature, which has become almost entirely self-referential and unprogressive, because its “theorists” don’t allow themselves to be challenged by inconvenient facts, but rather are satisfied with avoiding them, occupying themselves instead with mathematical prestidigitations.