Although the computational processes are indeed similar to Grossberg's model, the significant difference is exactly in the core ideas. In particular, the idea that reciprocal feedback should be an inverse of the corresponding feed-forward transformation, is a very significant statement about the nature of visual processing which is entriely absent from Grossberg's model. This is exactly the "conclusion that has not already been reached by other means".

This is a very interesting and significant concept, because we continue to hear debate in the literature about which cortical processes "precede" or "follow" which others, based on an assumption of some kind of sequential progression from lower to higher representational levels. Even if there were no supporting evidence for it, this concept should at least be exposed to the larger community, if only to dispel the commonly held notion that a hierarchical architecture necessarily suggests a sequential progression.

This concept also offers a more general justification for the processing stages in Grossberg's own model, explaining why the visual system not only detects edges, but completes them again back in a lower level representation, and why those edges then influence the diffusion of the brightess signal between them. The more general concept is the principle of reciprocal action, whereby the function of visual processing is not just to detect pattern in the input, as is commonly assumed, but also to reconstruct or reify the percept on the basis of the patterns detected in it. This is truly a novel and significant concept both for psychology and also for neuroscience, offering a possible explanation for the function of reciprocal pathways between cortical areas not discussed anywhere else in the literature.

This concept is also of great significance to Gestalt theory, firstly because it offers an explicit description of the subjective experience of Gestalt illusions, expressed in objective quantitative terms. More significantly still, the concept of feedback by inverse transformations offers an explanation for why perception performs the kind of reification observed in Gestalt illusions. This concept therefore puts Gestalt theory on a more firm theoretical and computational foundation.

Furthermore, there is substantial evidence for this general principle of perceptual processing, because a great number of diverse illusory phenomena can be accounted for by this general concept, as shown in the paper. This is far more valuable than a specific neurophysiological model like Grossberg's which accounts only for the phenomena which it was designed to account for, but does not generalize to a computational principle which would apply also to other models and other types of perceptual phenomena.

In short, this paper has very significant implications for a great variety of issues in psychology and neuroscience, and I am amazed that this reviewer's conceptual vision is so narrow that he cannot see its significance! There are altogether too many researchers in the field who have an intimate understanding of the details, but no understanding of the greater issues of perception and consciousness!