On 2016 Sep 24, Lydia Maniatis commented:

Please see additional comments on PubPeer.

On 2016 Feb 14, Lydia Maniatis commented:

The authors begin their article with the following, provocative statement:

"Traditionally, objective studies of perception have measured and explained threshold contrast of a target on a blank background: the contrast required for an arbitrarily selected level of performance. Indeed, most of what we know best about visual processing has come from such studies.1"

The statement is misleading both in its implication that threshold experiments have dominated the study of visual perception and that they have been particularly fruitful. In fact, the "tradition" that has focussed on "explaining threshold contrast on a blank background" has accomplished rather little. It is at a loss to explain the great sensitivity of "threshold contrast" to experimental conditions, a problem acknowledged elsewhere by Pelli (1990 in Vision: Coding and Efficiency, Edited by Colin Blakemore, Cambridge University Press, Cambridge, pp.3-24, 1990): “If the fraction of photons that excite photoreceptors is constant, why should the experimental conditions so dramatically affect the quantum efficiency?” Pelli (1990) proceeds to “present a framework that will allow simple threshold measurements to distinguish between two classes of explanation for the variation in efficiency.” These classes of explanation don't involve the structure of the stimulus, the influence of which on the perception of lightness, illumination, color, orientation and shape is firmly established.

To give a concrete example of the inadequacy of the structure-blind “tradition,” it cannot explain the presence of a perceived luminance difference where no luminance difference exists, as occurs in the case of illusory contours (which are reflected in the behaviour of visual neurons). What is the “signal” here? What role can “noise” play?

It would appear that the entire theoretical and empirical literature on the problems of visual organisation needs to be disappeared (or implicitly deemed not "objective") in order for the authors to make a case for using noise.

On 2016 Feb 14, Lydia Maniatis commented:

The authors begin their article with the following, provocative statement:

"Traditionally, objective studies of perception have measured and explained threshold contrast of a target on a blank background: the contrast required for an arbitrarily selected level of performance. Indeed, most of what we know best about visual processing has come from such studies.1"

The statement is misleading both in its implication that threshold experiments have dominated the study of visual perception and that they have been particularly fruitful. In fact, the "tradition" that has focussed on "explaining threshold contrast on a blank background" has accomplished rather little. It is at a loss to explain the great sensitivity of "threshold contrast" to experimental conditions, a problem acknowledged elsewhere by Pelli (1990 in Vision: Coding and Efficiency, Edited by Colin Blakemore, Cambridge University Press, Cambridge, pp.3-24, 1990): “If the fraction of photons that excite photoreceptors is constant, why should the experimental conditions so dramatically affect the quantum efficiency?” Pelli (1990) proceeds to “present a framework that will allow simple threshold measurements to distinguish between two classes of explanation for the variation in efficiency.” These classes of explanation don't involve the structure of the stimulus, the influence of which on the perception of lightness, illumination, color, orientation and shape is firmly established.

To give a concrete example of the inadequacy of the structure-blind “tradition,” it cannot explain the presence of a perceived luminance difference where no luminance difference exists, as occurs in the case of illusory contours (which are reflected in the behaviour of visual neurons). What is the “signal” here? What role can “noise” play?

It would appear that the entire theoretical and empirical literature on the problems of visual organisation needs to be disappeared (or implicitly deemed not "objective") in order for the authors to make a case for using noise.

On 2016 Sep 24, Lydia Maniatis commented:

Please see additional comments on PubPeer.