On "Directional Harmonic theory", or briefly DHT.
DHT seems to me an intriguing approach of visual phenomena. However, I am not acquainted with this approach. Indeed, after reading the paper I got a gist of the proposed model but no significant insight in the process from input to output that allows me to make predictions on visual phenomena. Thus, in this respect, I am probably an inappropriate referee, but I wonder whether the paper is sufficiently accessible to other readers without reading earlier publications on this matter. I appreciate the presented objections of the author against the various alternative approaches, such as the neural network paradigm, the feature detection hierarchy, dynamic networks, etc.
With respect to the following matter I have a serious comment. I miss an account of more or less recent research of perception that focuses on a salient general feature that these illusory contour phenomena have in common with other phenomena, such as Neon illusions, visual transparency, and pattern completion. This general feature, attended to by Rock (1983, logic of perception), Gregory (1972, Nature) and various others, is that all these illusions are evoked by two rivalling subdivisions of a stimulus. According to one, the illusory subdivision, the stimulus is taken as a superposition of two patterns, each on a different plane. According to the other, the alternative mosaic subdivision, the stimulus is taken as a configuration of adjacent sub-patterns on one plane. If the first subdivision is pregnant (Koffka) or simple and the second complex, the illusion is strong. Otherwise, it is weak or absent. Implicit properties of this rule are the following. If the two superimposed patterns of the first subdivision are arbitrarily oriented and positioned with respect to each other, the illusion is salient, whereas if the sub-patterns of the alternative mosaic subdivision are regular and coincidentally related, the illusion is weak or absent.
On a separate sheet, 8 pairds of patterns are presented that illustrate the idea for completion (no 1,2,3), for transparency (no 4), for the Neon illusion (no 5,6), and for illusory contours (no 7,8). According to experiments, the A patterns show a relatively strong illusory effect whereas the B patterns show a rather weak or absent illusory effect. The simplicity principle for 144 completion patterns is illustrated by Van Lier et al (1994, Perception, 60, 134-143), for transparency by Leeuwenberg (a971, Formal theories of visual perception, Wiley, 25-47), for Neon illusion by Van Tuijl (1979, Perception & Psychophysics, 25, 269-284) and for illusory contours by Van Tuijl (1982, Psychophysical judgement and the process of perception. North-Holland, 114-131). The Neon illusion presupposes a stimulus with two colors, say black and blue. The blue is here indicated by dotted lines and the illusion is a blue spreading within subjective contours.
In my view, the sketched perceptual mechanism underlying these and various other phenomena is obvious and even tenable for various measures of complexity (Simon, 1972, Psychological Review, 79, 369-382). Therefore, my expectation is that DHT should explain its effects, at least, for the few prototypical demonstrations presented here. Otherwise DHT is, to me, just one of the many models that explains a few isolated peripheral phenomena.
A topic of subordinate relevance deals with the distinction between an illusory contour shape and an invisible not-illusory shape represented by a stimulus code. For instance, Figure 8A evokes an illusory triangle indicated in subdivision no 1, and the code of the mosaic subdivision no 2 should represent an invisible triangle being a global schema that specifies the positions of the black sub-patterns. My question is whether DHT makes this distinction, and if so, how it comes across.
Sincerely
