Talk:Pinna illusion
<review>message to curators</review>There are several problems with this article that, if corrected, would greatly improve it:
1. It is far too long and detailed for general readers; an article about half this length would be more user-friendly, and have more impact.
2. An explanation of percepts in terms of gestalt principles is not acceptable in modern cognitive neuroscience. People have long since recognized that such “explanations” have little or no scientific basis, although they are of course of historical interest.
3. At the other end of the spectrum, the brief descriptions of percepts in terms of neuronal receptive field properties are equally problematic. The fact is that the processing underlying motion perception is not yet understood, as is widely admitted. The explanations offered in these terms are simply hand waving, and in any event inappropriate for most readers.
4. On the other hand, successful explanations of brightness, color and geometrical percepts on a wholly empirical basis (i.e., predicted by analyses of natural scene databases that serve as proxies for human experience) are not mentioned. This rapidly growing body of evidence argues that there are no illusions except in a trivial sense, or, better put, that all visual percepts are equally illusory. This work done over the last 15 years should be included, either to say why it is wrong (if that is the author’s opinion) or, more plausibly, how the phenomena described here present a further challenge to investigators who are pursuing this way of understanding vision.
Reviewer B
Review of the artice „Pinna illusion“ by Dr. Baingio Pinna
General remarks
This is a very complete overview on the Pinna’s and related apparent motion illusions. Most of the figures are from the peer reviewed work of the author with some recent updates by Pinna and his collaborators. The illustrations and videos are of a high technical and artistic level. The figures show clearly the effects outlined in the text. Unfortunately the article is too long for the intended format. Some of the weaker illusions like the accordion illusion may be omitted. On could also leave out illusions which are not directly related to (apparent) motion. For a person less familiar with the field, the current ordering of sections may be somewhat confusing because seemingly similar stimulus configurations reappear in different sections of the overview with only slight modifications. To facilitate the reading I recommend reordering of the sections according to the configuration of the stimulus elements and the apparent motion effects:
1) Apparent contraction/expansions and rotation of a circular arrays of stimulus elements(the original Pinna illusion with some recent additions)
2) The windmill illusion
3) Apparent translation or divergence/convergence of a rectangular array of stimulus elements without figure-ground configuration
4) Apparent sliding 3D-motion in a figure-ground stimulus configuration (the original Ouchi illusion and related newly found illusions)
Specific remarks
Figures 1 to 6 show nicely some of the most important features of the original Pinna illusion with apparent rotation or contraction/expansion.
Figure 7) is about apparent translation followed again by a figure about apparent rotation. I propose to insert the paragraphs about implicit vs. explicit diagonals (fig. 8a-b), the consequences of asymmetric luminance profiles (figure 16 a and b) and the apparent motion induced by luminance modulation (fig 17a-h) right after figure 6.
Figure 7a-b, d: Perhaps the reader should be warned that in order to perceive consistent sliding motion effects he has to track the black target carefully. Once the gaze strays off the target, the sliding effect is lost.
Figure 7d: When just examining the figure without fixating a particular point a somewhat irregular figure with the square sharing the same implicit diagonal appears as floating before a background with the other squares akin to the Ouchi illusion (figure ground configuration). When tracking the moving target, a different percept arises: a cluster of squares sliding together downwards in the same plane as the other squares (different patches with local motion).
Figure 10: Divergence/convergence: the effect shown on figure is rather weak. The whole paragraph may be omitted.
The accordion illusion (figure 13 a-c): also a rather weak effect which may be omitted.
Sliding motion from edge (figure 15): this is one of the illusions which I failed to perceive but which was very perspicuous to a colleague I asked about.
Conclusions: Reordering of the sections above according to the presumed neuronal mechanism may clarify the final discussion. The discussion about optic flow should lead and not follow the discussion about figure-ground discrimination.
RESPONSES TO THE COMMENTS OF REVIEWER B
I thank the reviewer for the useful suggestions. My responses to individual comments of reviewer B, and the indications of the corresponding changes in the article, are written below, in italics.
GENERAL REMARKS
This is a very complete overview on the Pinna’s and related apparent motion illusions. Most of the figures are from the peer reviewed work of the author with some recent updates by Pinna and his collaborators. The illustrations and videos are of a high technical and artistic level. The figures show clearly the effects outlined in the text. Unfortunately the article is too long for the intended format.
Response: I shortened the length by reducing the blank spaces in each figure and by optimizing the organization of figures and text.
Some of the weaker illusions like the accordion illusion may be omitted.
Response: See my comments below.
On could also leave out illusions which are not directly related to (apparent) motion.
Response: See my comments below.
For a person less familiar with the field, the current ordering of sections may be somewhat confusing because seemingly similar stimulus configurations reappear in different sections of the overview with only slight modifications. To facilitate the reading I recommend reordering of the sections according to the configuration of the stimulus elements and the apparent motion effects: 1) Apparent contraction/expansions and rotation of a circular arrays of stimulus elements(the original Pinna illusion with some recent additions) 2) The windmill illusion 3) Apparent translation or divergence/convergence of a rectangular array of stimulus elements without figure-ground configuration 4) Apparent sliding 3D-motion in a figure-ground stimulus configuration (the original Ouchi illusion and related newly found illusions)
Response: I reordered the sections as suggested but with the windmill illusion at the end preceded by the other illusions due to luminance modulation. I think this organization may be clearer in terms of motion cues and consequent neural mechanism.
SPECIFIC REMARKS
Figure 7) is about apparent translation followed again by a figure about apparent rotation.
Response: I reversed the two paragraphs.
I propose to insert the paragraphs about implicit vs. explicit diagonals (fig. 8a-b), the consequences of asymmetric luminance profiles 0(figure 16 a and b) and the apparent motion induced by luminance modulation (fig 17a-h) right after figure 6.
Response: I followed all the suggestions except the one related to Fig. 17. I think in fact it may be better to insert the luminance modulation effects just before the windmill illusion now shown at the end of the sequence of illusions. This was done to make it easier and clearer in terms of motion cues and then in terms of presumed neural mechanisms the logic organization of the illusions.
Figure 7a-b, d: Perhaps the reader should be warned that in order to perceive consistent sliding motion effects he has to track the black target carefully. Once the gaze strays off the target, the sliding effect is lost.
Response: I included the suggestion within the text.
Figure 7d: When just examining the figure without fixating a particular point a somewhat irregular figure with the square sharing the same implicit diagonal appears as floating before a background with the other squares akin to the Ouchi illusion (figure ground configuration). When tracking the moving target, a different percept arises: a cluster of squares sliding together downwards in the same plane as the other squares (different patches with local motion).
Response: I have incorporated in the text these accurate phenomenal observations.
Figure 10: Divergence/convergence: the effect shown on figure is rather weak. The whole paragraph may be omitted.
Response: OK. Done.
The accordion illusion (figure 13 a-c): also a rather weak effect which may be omitted.
Response: I would like to keep this interesting effect that unfortunately the reviewer did not perceive as strongly as it can be when the size of the figure is larger (see the attachment). To help the reader in perceiving the effect I have enlarged the size of the stimulus and added another figure (13b) where the perceived dynamic deformation of Fig. 13a is represented surrounded by two blue ellipses. The size is in fact one of the crucial factors of the the illusion. I changed the text accordingly.
Sliding motion from edge (figure 15): this is one of the illusions which I failed to perceive but which was very perspicuous to a colleague I asked about.
Response: OK. Unfortunately, also the strength of this effect depends on the size of the stimulus.
Reordering of the sections above according to the presumed neuronal mechanism may clarify the final discussion.
Response: I accepted the suggestions with the exception of the luminance modulation effects followed by the windmill illusion placed at the end. See my previous comments.
The discussion about optic flow should lead and not follow the discussion about figure-ground discrimination.
Response: I rephrased the Conclusion section accordingly.
Reviewer C
Reviewing this article is difficult because it was invited but holds the possibility (presumably) of being rejected; which seems unfair. Given that an invitation has been extended the author should have some latitude in the choices made about what to present but, what standards are a reviewer to apply. Anyway, I'll give this a shot. But, I don't have time to learn all this wiki stuff.
Of course the Pinna illusion is one of the most compelling motion illusions in existence. And Pinna himself is one of the most gifted illusion manufacturers on the planet; the present paper attests to this. That said, I worry that the article does not conform to the intentions of Scholarpedia.
The article is WAY too long. The guidelines suggest an article of 2000 to 3000 words; this one has at least 5000 words, excluding references, and 18 figure, many with multiple frames. Many of the figures are great much of it is not directly related to the original Pinna Illusion.
I imagine that the main thought in a reader's mind would be: I've always wondered what causes that fabulous Pinna Illusion. I find the article weak on this point. The important points are (a) the micropatterns have oriented low-frequency components and (b) these engage low-level direction selective mechanisms which (c) are subject to the aperture problem. The following figure is taken (without permission) from Gurnsey and Pagé (2006, Vision Research, Figure 1). It succinctly summarizes the root cause of the Pinna illusion; at least, a standard account that is generally believed. I would strongly suggest that a figure of this sort be included in the paper. Again, I think it is an account of this sort that interested readers are looking for. I know that Pinna comments about implicit orientations, direction selective mechanisms and the aperture problem, but the ideas just don't come across clearly to me. It's also worth mentioning that a modified version of the illusions designed to optimally stimulate the direction selective mechanisms in question produce a stronger illusion (right panel).
The paragraph, "The role of peripheral vision" has two major problems: (a) "this is likely because..." is really empty (the reason we don't see it at fixation is because the mechanisms at fixation don't see it) and (b) the notion of blurring is introduced before its relevance to the illusion is discussed. (A figure like the one above would help greatly here.)
The general approach of the article is "look and see for yourself"; no data are shown. This can be okay in certain circumstances but, for example, Figure 5 is compared with Figure 4 to make the point that the organization of the display is not critical to the illusion. But, the stimulus in Figure 5 elicits at much weaker illusion (for me). I think it's misleading to categorize stimuli as those that produce or don't produce the illusion, and ignore variations in illusion strength.
The section titled "Neural Mechanisms underlying the Pinna Illusion" is weak, as I said earlier. The references to Furmüler's work are rather oblique; that is, not well explained.
In my view much of the material under "New phenomena related to the Pinna Illusion and originated by implicit orientations" is not appropriate for an article of this sort. I think most of the illusions reflect, in one way or another, the simple fact that stimulating motion energy mechanisms or orientation selective mechanisms can produce some startling effects. For example, are figures 9 and 10 not simply related to the Café wall and Fraser spiral illusion. Isn't Figure 17 just another example of motion energy?
So, my largest concern here is that the Pinna illusion is a symptom of fundamental visual coding principles. The focus on the Pinna illusion and its variants obscures this more fundamental fact. Rather than really elucidating these principles the paper is a compendium of beautiful demos.
As a final point, there are some facts about the Pinna illusion that aren't easily explained in terms of the standard account (Figure 1 above). Gurnsey and Pagé (2006) showed that the Pinna illusion falls apart when the oriented stimuli are broadband and when they are not luminance balanced. This is consistent with my perception of Figure 14. The illusion is strong for dashed lines but not for solid lines. What explains this? Figure 14c is fabulous but doesn't it suggest a role for motion selective end-stopped operators?
In summary, beautiful demos. However, I think readers would appreciate a deeper account of the coding mechanisms that produce this family of illusions.
RESPONSES TO THE COMMENTS OF REVIEWER C
I thank the reviewer for the suggestions that greatly improve this article. The responses to the comments and the corresponding changes in the article are reported below in italics.
REMARKS
The article is WAY too long. The guidelines suggest an article of 2000 to 3000 words; this one has at least 5000 words, excluding references, and 18 figure, many with multiple frames.
Response: That is true. Nevertheless, I noticed that in Scholarpedia there are many other articles that do not fit perfectly the length suggested by the guidelines. Unfortunately, the length of the article increased during the revision process by answering the questions of Reviewers A and B. The References section alone is about 1000 words. It can be reduced but I suggest to leave it as it is for a reason of completeness.
Many of the figures are great but I think some of the material is new (unpublished; sorry if I'm wrong about this) and much of it is not directly related to the original Pinna Illusion.
Response: All the effects shown have been published and presented in the most important conferences and meetings as reported in the References section. I decided to include other figures for two main reasons: (i) they are related in different ways to the main illusion showing the complex role of directional biases (see Section 4); (ii) they provide a link between the main illusion and other known effects. I think that all the figures presented can demonstrate that through small variations it is possible to go through the most known motion illusions and, as a consequence, that they are strongly related and reducible to simple phenomenal and physical properties. This is an important point for an article of this kind. In fact, not only does it show an illusion isolated from other motion effects and with a specific explanation but also a phenomenon strongly related with many other phenomena useful to shed light within the more general problem of motion perception. Furthermore, the presence of these related effects may improve the impact factor of Scholarpedia.
I imagine that the main question in a reader's mind would be: I’ve always wondered what causes that fabulous Pinna Illusion. I find the article weak on this point. The important points are (a) the micropatterns have oriented low-frequency components and (b) these engage low-level direction selective mechanisms which (c) are subject to the aperture problem. The following figure is taken from Gurnsey and Pagé (2006, Vision Research, Figure 1). It succinctly summarizes the root cause of the Pinna illusion; at least, a standard account that is generally believed. I would strongly suggest that a figure of this sort be included in the paper. Again, I think it is an account of this sort that interested readers are looking for. I know that Pinna comments about implicit orientations, direction selective mechanisms and the aperture problem, but the ideas just don't come across clearly to me. It's also worth mentioning that a modified version of the illusions designed to optimally stimulate the direction selective mechanisms in question produce a strong illusion (right panel).
Response: I have to say thanks to the Reviewer for putting to my attention that I missed to quote the article by Gurnsey & Pagé (2006) that I know very well. In fact, the main explanation reported in the article is exactly the one suggested by the Reviewer and Gurnsey & Pagé. The article by Gurnsey and Pagé explained the illusion by using almost the same words I used. Unfortunately, I missed to include this article quoting only Gurnsey et al. (2002), where this explanation was firstly suggested. In the second quotation of the article by Gurnsey et al. I should have quoted the one by Gurnsey and Pagé. I apologize about that. I have included this reference and added the suggestion of the Reviewer: “The important points useful to explain this illusion are the following: (i) the micropatterns have oriented low-frequency components, (ii) these engage low-level direction selective mechanisms, which (iii) are subject to the aperture problem”. I previously quoted within the article the figure by Gurnsey and Pagè, also suggested by Bayerl & Newmann (2002) and by Morgan (2002). As suggested by the reviewer I included this figure.
The paragraph, the role of peripheral vision has two major problems: (a) "this is likely because..." is really empty (the reason we don't see it at fixation is because the mechanisms at fixation don't see it) and (b) the notion of blurring is introduced before its relevance to the illusion is discussed. (A figure like the one above would help greatly here.)
Response: I changed the text accordingly as follows: “This is because motion detectors in the fovea are not sensitive to this type of stimulus pattern. Furthermore, with peripheral viewing the precise spatial square form of the pattern elements ought to be blurred and the dominant motion cues ought to derive, not from the constituent line sections, but from the entire elements (see Fig. 10).
The general approach of the article is "look and see for yourself"; no data are shown. This can be okay in certain circumstances but, for example, Figure 5 is compared with Figure 4 to make the point that the organization of the display is not critical to the illusion.
Response: I cannot understand this point. However, Fig. 5 is not compared with Fig. 4 within the article. The comparison of Fig. 5 should be made with Fig. 1, which is the starting main figure. Was this the point objected by the Reviewer? I clarified this comparison within the text as follows: “When the squares are randomly shuffled and arranged approximately in concentric circles the apparent rotation persists (see Fig. 5 and compare with Fig. 1). With regard to the “look and see for yourself” aproach, I think that it is perfect and necessary when the main topic is a visual illusion and, more generally, visual perception and it is also perfect for Scholarpedia. In fact, everybody can verify the descriptions and the reported results.
The section titled “Neural Mechanisms underlying the Pinna Illusion” is weak, as I said earlier. The references to Furmüler’s work are rather oblique; that is, not well explained.
Response: I completed the explanation by including the following sentence in Section 2: “More precisely, the interpretation of the motion effect depends on a step where image features such as lines, intersections of lines, black and white edges like those of Fig. 1 and local image movement are derived. These features contain many sources of noise or uncertainty that can cause bias. As a result, the locations of features are perceived erroneously and the appearance of the patterns is altered. Thus, the estimated flow vectors of Fig. 1 are biased in the clockwise and in the counterclockwise directions as can be perceived in the outer and inner ring”.
In my view much of the material under “New phenomena related to the Pinna Illusion and originated by implicit orientations” is not appropriate for an article of this sort. I think most of the illusions reflect, in one way or another, the simple fact that stimulating motion energy mechanisms or orientation selective mechanisms can produce some startling effects. For example, are figures 9 and 10 not simply related to the Café wall and Fraser spiral illusion. Isn’t Figure 17 just another example of motion energy?
Response: The Reviewer is right when he/she says that many startling effects shown in the literature on motion effects are based on the stimulation of motion energy mechanisms or orientation selective mechanisms. Nevertheless, I think that the effects related to Fig. 1 demonstrate that such startling phenomena can be obtained not only by stimulating motion energy mechanisms or orientation selective mechanisms but also in other ways as shown in the article and, particularly, in some of the related effetcs (see Figs. 14, 15, 16, 17 and 18, now Figs. 15, 16, 17, 18 and 19). In my opinion Fig. 9, 10 and even Fig. 11 (now Figs. 10, 11 and 12) are not related to the Café Wall and Fraser illusion for three main reasons: (i) The check elements of the figures are different from those used to obtain the Café Wall and Fraser illusions, furthermore, (ii) they do not show any motion illusion as it can be perceived in Fig. 11 c-d and in the figure below (Pinna & Gregory, 2002), which is a mixture between a Café Wall and a Fraser illusion, and (iii) the direction of the illusory tilt of Figs. 9, 10 and 11 (now Figs. 10, 11 and 12) is the opposite of the ones of the Café Wall and Fraser illusions. As regards Fig. 17 (now Fig. 18), I honestly see these conditions as a clear demonstration of the role of explicit orientations in polarizing the field of motion biases induced by the contrast modulation. I think that they are some of the clearest shown in the literature on these kind of illusions.
So, my largest concern here is that the Pinna illusion is a symptom of fundamental visual coding principles. The focus on the Pinna illusion and its variants obscures this more fundamental fact. Rather than really elucidating these principles the paper is a compendium of beautiful demos.
Response: It is my frank opinion that the main effect of Fig. 1 and its variants demonstrate these principles and challenge the explanations based only on one kind of effect. I think indeed that all these effects can stimulate the finding of a unique global explanation based on these fundamental visual coding principles that for example goes beyond oriented low-frequency components.
As a final point, there are some facts about the Pinna illusion that aren’t easily explained in terms of the standard account (Figure 1 above). Gurnsey and Pagé (2006) showed that the Pinna illusion falls apart when the oriented stimuli are broadband and when they are not luminance balanced. This is consistent with my perception of Figure 14. The illusion is strong for dashed lines but not for solid lines. What explains this? Figure 14c is fabulous but doesn’t it suggest a role for motion selective end-stopped operators?
Response: I perceive the illusion shown in Fig. 14 (now Fig. 15) as having the same strength both in dashed and solid lines. The attention plays an important role in these kind of figures. We have several experiments in progress on this topic. I suggest to shift the attention from the dashed to the solid lines both in Figs. 14a and 14b (now Figs. 15a-b). I think that Fig. 14c (now Fig. 15c) cannot be explained by invoking the role of motion selective end-stopped operators, but in the same way as Figs. 14a-b (now Figs. 15a-b) as described in the text. In addition, these figures can shed light on other related motion effects like the boogie-woogie illusion (Cavanagh & Anstis, 2002). The link with other illusions is in fact one of the main topics of this article.
So, my largest concern here is that the Pinna illusion is a symptom of fundamental visual coding principles. The focus on the Pinna illusion and its variants obscures this more fundamental fact. Rather than really elucidating these principles the paper is a compendium of beautiful demos.
Response: I am sorry to contradict the Reviewer, but I think that the related variant phenomena are useful to understand the complexity of the fundamental visual coding principles involved not only in the Pinna illusion or in other known motion effects but more generally in motion perception.
In summary, beautiful demos. However, I think readers would appreciate a deeper account of the coding mechanisms that produce this family of illusions.
Response: I think that whenever suggested by the Reviewer I provided a deeper account of the coding mechanisms involved in these illusions (see previous points). Unfortunately, I do not think that I have enough space to go any deeper into each effect.
