I thank the reviewers for their input and have implemented amendments in the article in response to many, though not all of their comments. My responses to individual comments of reviewers A and B, and the indications of the corresponding changes in the article, are written below, in italics. Reviewer C has initially indicated his suggestions in the form of direct interventions in the text. I have taken note of these suggestions and have developed them further, but have relegated them to a new, separate section of the article, subtitled 'Auditory Gestalten'. My responses to his subsequent review are below. I have also taken into account suggestions submitted to me from other readers of the article. I have reorganized and partly rewritten the last portion of the text, deleting a section and adding a new 'contemporary work' section, to more clearly delimit it from classical work. Finally, I have added several recent references, have made numerous small stylistic changes throughout the paper, and have slightly edited Figure 3 and, with technical help of the copy-editor, added a bit of pizazz to Figure 6, as explained in the text. Additional changes are described in my reply to Reviewer C.
This review is well organized, balanced, clear and accurate. Nevertheless, I have some suggestions.
1. "Gestalt principles, or gestalt laws, are rules". Traditionally, the term “law” has an epistemological status (see also Wertheimer (1925) "Strukturgesetze" or Metzger "Gesetze des Sehens" etc.) on the contrary “rule” is only a generic term.
Response: Although the German word ‘Gesetz’ does mean ‘law’, I prefer to use this term only for formally expressed regularities with clear and general conditions of application, such as the laws of physics and chemistry. The Gestalt principles have not as yet achieved this status, as they are verbally formulated and their application domain is not always clear, because of the ‘ceteris paribus’ condition. I find that the term ‘rule’ is appropriate to denote this type of regularities.
the Newton's law of gravitation, the Galileo's law of inertia, and the best laws of physics and chemistry are ceteris paribus laws...
2. "(...) are rules of the organization of sensory signals". Using the gestaltic language: (…) organization of the perceptual field. When used properly, "sensory signal" is not empirical reality (immediate experience) but an intellectualization (physical description - see for example Russell, 1914): it is theory. The gestalt laws are primarily empirical principles.
Response: I have dropped ‘sensory signals’ and have changed this to ‘… are rules of the organization of perceptual scenes’. ‘Perceptual field’ is a technical term unfamiliar to the general reader, and I did not want to use it right in the first sentence. Incidentally, claims such as that the proper usage of ‘sensory signals’ involves labeling them as intellectualization or theory, or that empirical reality is the same as immediate experience, while congenial with the views of Berkeley, Mach, early Russell, or philosophical phenomenologists, are not, to my knowledge, widely accepted in the scientific or philosophical community.
the community apart, Wertheimer could not write: "I stand at the window and see a house, trees, sky. Do I have sky, house, and trees? No. I have sensory signals..." because he wrote about objects under observation (immediate experience). In my opinion, it could be useful to distinguish between: 1) the empirical stance of Wertheimer (experimental phenomenology’s empirical perspective is different from the English empiricism. While the former is concerned with present (immediate) experience (Erlebnis), the latter refers to the accumulated effects of past ‘present experiences’ (Erfahrung). Besides, the experimental phenomenology of Wertheimer, Metzger, Michotte, etc. is different from the phenomenalism of Berkeley, Mach, etc.) i.e. the empirical character of the "rules", from 2) theories and explications (that change with the scientific seasons) on the empirical "rules".
3. "How do we accomplish such a remarkable perceptual achievement, given that the visual input is, in a sense, just a spatial distribution of variously colored individual points, and that the auditory input is just a temporally varying complex waveform? The beginnings and the direction of an answer were provided by a group of researchers early in the twentieth century, known as gestalt psychologists." But the Gestaltists "answered" de facto about the perceptual field "we must study the organization of the environmental field" Koffka, 1953, p. 67), not about how the visual system works.
Response: I find that by studying the organization of the perceptual field we at the same time get insights (that is, ‘the beginnings and the direction of an answer’, as I wrote) into how the visual system works to realize such an organization. I don’t think that there is such a large step from purely phenomenological claims such as, e.g. ‘items that are near each other are perceived as grouped’ to claims such as ‘the visual system works in such a way that items that are near each other are perceived as grouped’.
4. "German word meaning 'shape' or 'form'". This seems to me a too strong simplification.
Response: Exact translations between languages are often difficult, and German – English dictionaries offer, as for any word, several English meanings for ‘Gestalt’. However, the main perceptual meanings are ‘shape’ and ‘form’ (these are also my own intuitions, as a non-native but relatively fluent speaker of German). It is true that in the perceptual literature the word Gestalt has acquired more complex connotations, such as’ meaningful perceptual unit’, ‘salient configuration’, and the like, but as an ordinary, everyday German word, ‘Gestalt’ is, in my judgment, quite adequately rendered in English as ‘shape’ or ‘form’.
5. "Gestalt principles aim to elucidate (it would be better "to formulate") the rules (it would be better: laws) according to which the perceptual input (it would be better: perceptual field) is organized into unitary forms."
Response: I have changed ‘elucidate’ into ‘formulate’, and ‘rules’ to ‘regularities’, but I find ‘perceptual input’ adequate.
6. "These principles mainly apply for vision, but there are also analogous aspects in auditory and somatosensory perception." These generalizations are not only analogies: they are the same laws in different domains.
Response: This is an interesting point, but I still think that analogy describes this relation better than identity. It is true, for example, that the principle of proximity does apply directly in the auditory domain, but in audition it mainly involves temporal proximity, whereas in vision it involves spatial proximity, as I elaborate in the new section on auditory Gestalten. Also, what would be, in the auditory domain a principle that is identical to such a principle as 'closure'? Closure involves two-dimensional planar structures, and it is hard to envisage fully corresponding features in audition. I don’t find it implausible that domains as different as vision and audition should involve partly different principles of organization.
Domains are different. This is a pleonasm: but the differences don't involve that generalizations are impossible.
7. In figure 8(c), the elements tend to be perceived as: 1/23/45/6 (against Palmer). Maybe also the shrinkage of a partially occluded space (see Kanizsa) reduced further the distance between the elements 2 and 3 and the elements 4 and 5.
Response: I have removed this figure (I assume the reviewer refers to 11c – there is no 8c) from the article. See below, in comments to reviewer B.
8. I suggest to add the principle of directionality (see Bozzi, 1969 or 1989; or Metzger - the last version of Gesetze des Sehens, 1975, pp. 421-422).
Response: Currently I don’t have access to Metzger (1975) or Bozzi, but I will check at a later point, and update, if needed.
9. "The gestaltists tended to favor the notion that these principles are among the fundamental properties of the perceptual system, providing the basis of our ability to make sense of the sensory signals". This contrasts with the phenomenological orientation of gestaltists. In particular, gestalt principles are expression of a nomic approach and not a functional approach.
Response: My answer, to the extent that I understand this objection, is as follows. My claim quoted above is based, among others, on the following statement by Metzger (1936/2006, p. 180), concerning Gestalt laws: ‘… the fundamental laws of perception are present before the accumulation of this stock of experience, and before such schooling. Those fundamental laws are not profoundly changed by experience, but rather, without the existence and stability of these laws, the store of past experience could neither be collected nor utilized’. I have then contrasted Metzger's claim that Gestalt laws precede learning and enable learning in the first place, with a somewhat elaborated form of Rock's proposal that Gestalt laws are themselves based on learning. The phenomenological approach of the gestaltists does not preclude the legitimacy of asking questions such as where the Gestalt laws come from. It also did not prevent Köhler to ponder the neural counterparts of perceptual experiences, and discuss at length the notion of neuro-phenomenal isomorphism. In sum, although this is an interesting issue to pursue, I don’t find that a close reading of the original Gestalt authors warrants an exclusive stress on a pure phenomenological stance, disallowing any non-phenomenological but perceptually very relevant considerations, such as involving mechanism, origin, development, or neural correlates.
I agree with you: "I don’t find that a close reading of the original Gestalt authors warrants an exclusive stress on a pure [here I prefer "experimental"] phenomenological stance". Laws of observables are an epistemologically independent knowledge field, but they can also be assumed to be the goal of the perceptual system and therefore become propaedeutic knowledge to the study of its functioning.
This is an excellent review. I have a few suggestions that I think will improve the review.
Good gestalt principle section: If one considers wholes rather than parts, as Wertheimer would have done, then Figure 7c is not as "good" as Figure 7a.
Response: I agree with this point, if I understand it correctly. In fact, I think I made the same point in the text, but maybe not clearly enough, so I have now slightly edited the text. The perceived decomposition of 7a into a straight line and a curved line is indeed phenomenally much more salient than the decomposition into the two components indicated in 7c. The question is why this is the case. My point was that this difference in salience cannot be explained by the principle of continuity, because both decompositions involve continuous sub-wholes. Therefore another principle must be invoked, and the point of this example was to introduce the good gestalt principle, which can explain why the first decomposition is preferred.
Past experience principle section: I recommend that the author cite more recent reviews rather than the original Peterson et al. (1991) paper. For example: Peterson, M. A. (1994). Object recognition processes can and do operate before figure-ground organization. Current Directions in Psychological Science, 3, 105-111. or Peterson, M. A. & Skow-Grant, E. (2003). Memory and learning in figure-ground perception. In B. Ross & D. Irwin (Eds.) Cognitive Vision: Psychology of Learning and Motivation, 42, 1-34.
Response: I have replaced the 1991 reference with the 2003 reference.
In the Non-Classical Principles section: Can you link the discussion of Common Region back to the preceding section on closure by discussing the relationship between common region and closure? The question that occurred to me as I read the review is whether common region should be considered a different principle from closure or an example of closure dominating proximity. In Wertheimer's terms, the later would be an example of whole-properties affecting the perception of subsidiary parts. (See Figure 11a.)
Response: I have shortened this section considerably. See below.
Regarding Figure 11b in the same section: Gillam did some important work on partial closure in grouping: Gillam, B. (1975). The evidence for "closure" in perception. Perception & Psychophysics, 17, 521-524. Peterson and Lampignano examined the role of partial closure in a recent paper: Peterson, M. A., & Lampignano, D. L. (2003). Implicit memory for novel figure-ground displays includes a history of border competition. Journal of Experimental Psychology: Human Perception and Performance, 29, 808-822.
Response: This figure is removed, see below.
The effects in Figure 11c seem quite a bit weaker to me. I can more readily group 1/23/45/6 in 11c than in either 11a or 11b. While the point is well-taken that intervening differently shaped or colored features may play a role in counteracting the proximity principle, and worthy of empirical investigation, it strikes me as premature to publish these ideas in a Scholarpedia piece. Similarly, stating that the same preferred grouping prevails in Figures 12c as in 12a and 12b without experimental evidence strikes me as premature. I wasn't certain it was "the same."
Response: I agree with the criticism that some of my claims in this sections are premature, since they are not accompanied by experimental evidence. To pursue these issues in more detail would be inappropriate for an introductory review paper. Therefore I have decided to remove Figures 11b,c and 12b, c, and much of the accompanying text, retaining only the reference to the two principles and their demonstrations (Figures 11a and 12a, which are now 11a and 11b).
Last sentence in the paragraph on Figure 11: The term "interpunction" is not commonly used; I recommend substituting "punctuation."
Response: This sentence is removed, see above.
Other Principles section: The last sentence in this section is opaque: "The index in Metzger (1936/2006) lists 18 Gestalt laws, but it is not clear whether they all refer to different rules." I recommend elaborating or eliminating.
Response: This sentence is removed. I have elaborated this issue in a little more detail in my review of Metzger (1936/2006).
Unresolved issues section: With respect to the sentence: "The gestalt principles are usually illustrated with rather simple drawings, such as those above. Ideally, it should be possible to apply them to an arbitrarily complex image..." Bill Geisler and his colleagues have done some work applying the Gestalt principles to the complex images of natural scenes.
Response: I have added a Geisler reference in the new 'Contemporary work' section of the paper, but I think that the above statement still applies, and have retained it.
This is an excellent article. I have several small comments, and a broader one at the end.
- Is there a reason you do not use the term "principles of grouping" or something to that effect before you begin listing the principles?
Response: I have now amended and extended the paragraph preceding the beginning of the list, taking also this comment into account.
- I prefer the terms "partition(s)" and "grouping(s)" to "group(s)."
Response: I do use ‘partitions’ in this sense occasionally in the paper, though my main use of this term is to refer to any logically possible aggregations, to contrast them with those that are perceptually realized. As for ‘groupings’, I have now added this term in the introduction, as one among labels for unitary forms, and in the text I use both ‘group’ and ‘grouping’ without distinction. In some cases I find ‘group’ more natural than’ grouping’, such as when I write ‘elements tend to be integrated into groups if they are similar to each other’, whereas in other cases ‘grouping’ may be more appropriate. I don’t think that there is much danger for readers to confuse the Gestaltist notion of ‘group’, as used here, with the completely unrelated mathematical notion with the same name.
- This is not clear: "Each of the six patches is perceived as a visual unit, a figure on a common ground. However, they are also collectively the elements of a higher-order visual unit, the horizontal row. According to Gestalt theory, this integration of individual components into a superordinate whole can be accounted for by the proximity principle: elements tend to be perceived as aggregated into groups if they are near each other." Don't you have a hierarchy here? I.e., you have elements, groupings, whole.
Response: I agree that there is a hierarchy here, but I don’t understand why this fact should make this section unclear. I describe the elements, point out that they are perceived as grouped, and state that this perceptual fact is an instance of a general principle, formulated by the Gestaltists. The intent of this section was not to discuss the notion of perceptual hierarchy, but to motivate the introduction of the principle of proximity.
- It would be nice if you talked about the question of part and wholes, and related it to mereology. For example, Wiegand, O. K. A Formalism Supplementing Cognitive Semantics Based on Mereology. Spatial Cognition & Computation, 2007, 7, 33-59.
Response: There is indeed a natural affinity between the philosophical, logical, semantic etc issues raised by mereology and perceptual issues raised by Gestalt principles. However, there is actually very little cross-talk between the two domains. To my knowledge, vision scientists hardly refer to the mereology literature, and, apparently, vice versa (e.g. see the long literature list for the Mereology entry in the online Stanford Encyclopedia of Philosophy). In particular, I have not, as yet, seen any substantial use of mereological insights for perceptual problems, so that I don’t find that a reference to mereology would be very useful for the intended readers of my short, introductory review.
- In general more use could be made of the competition between grouping principles.
Response: I agree completely, and I have implemented the suggestions put forward in the three comments below, involving Figures 3, 4, and 5 (now renumbered to 4, 5, 6). I have also extended the suggestion by including not only competition but also co-operation between different principles. I have also made similar amendments and additions to Figures 7, 8, and 9 (now renumbered to 8, 9, 10), including the corresponding changes in the text. The renumbering is due to the insertion of new Figure 3, that uses a very simple animation to demonstrate the Common fate principle, which was previously only described in words.
- In Fig. 3, it would be slightly better if the positions of the groupings in the row of six elements were varied a bit (as in Fig 2, b and c). Perhaps set two kinds of similarity in competition with each other, to suggest that the strength of grouping might be measurable.
Response: Implemented and extended, by adding several more examples (now it's Fig. 4).
- In Fig. 4, it might be nice to show that another principle can overcome the continuity principle. E.g., give elements of CXB the same color.
Response: Implemented and extended (now it's Fig. 5).
- Fig. 5. I think the grouping here is quite ambiguous. It's because here you have set up a competition between continuity and closure, and they are in equilibrium. I see a single bow-tie shaped figure: AXBCX(D)A.
This is indeed the figure that I am least satisfied with, but I still find that it serves a pedagogical purpose and fits well in the flow of the argument. However, I have amended the text to point out the ambiguities, and have introduced two more images (now it's Fig. 6).
- Fig. 6 is very nice.
Response: Thanks(now it's Fig. 7).
- The question of good Gestalt has been treated extensively by e.g., Feldman, J. How surprising is a simple pattern? Quantifying "Eureka!" Cognition, 2004, 93, 199-224.
Response: This paper by Feldman is primarily cognitively oriented and has little direct relevance for issues of perceptual organization, so I don’t find that its inclusion as a reference in an introductory review is appropriate.
- On auditory gestalts, see Kubovy, M. & Van Valkenburg, D. Auditory and visual objects. Cognition, 2001, 80, 97-126
Response: Reference added.
- You say, "As formulated by Wertheimer, Gestalt principles involve a 'ceteris paribus' (all other things being equal) clause (Palmer, 1999). That is, each principle is supposed to apply given that the other principles do not apply or are being held constant. In case two (or more) principles apply for the same input, and they favor the same grouping, it will tend to become strengthened; however, if they disagree, usually one wins or the organization of the percept is unclear. Several examples of the domination of one principle over another are presented above. However, although it has been addressed to some extent in the literature, the significant theoretical problem of how to predict which principle will win in which circumstances remains to be worked out in much more detail." This is the topic of Kubovy, M. & van den Berg, M. (2008). The whole is equal to the sum of its parts: A probabilistic model of grouping by proximity and similarity in regular patterns. Psychological Review, 115, 131-154. We have answered this question. Ceteris paribus is dead.
Response: Kubovy & van den Berg is an interesting and relevant paper, and I have added it as a reference in this section. However, the claim that Ceteris paribus is dead seems to me too strong a statement. Their paper deals only with proximity and certain kinds of similarity in certain kind of patterns, and the authors themselves wonder, near the end of the paper, whether their findings would generalize to other pairs of principles. Thus I maintain that the issue of relative importance of various Gestalt principle, in general form, is still open.
- It seems to me that much more progress has been made than the article acknowledges. So perhaps call this article "Classical Gestalt Principles" and ask the author (who is obviously very knowledgeable) to write another, "Contemporary Approaches to GPs."
Response: I agree that more progress has been made, and that the article does not do justice to it. I did try to include a number of contemporary references as pointers, however. I also agree that an entry dealing with contemporary approaches is a good idea. However, there are authors more qualified for that job than myself, people who have actually contributed to the current progress in this field, such as the reviewer himself, or Mary Peterson, or Steve Palmer. I don’t think that changing the title of the present paper is appropriate, though, because in it I did discuss and illustrate more than just the classical principles, including the new findings by Peterson and the new principles suggested by Palmer and colleagues.