The special issue Quantitative Approaches in Gestalt Perception recently appeared on Vision Research constitutes a commendable tentative to show the state-of-the-art in today’s Gestalt perception, and its relation to potential quantitative measurements and precise mathematical models.
All the papers show the attempt approach Gestalt perception from a quantitative viewpoint in a series of psychophysical and neurophysiological researches. Hawkins et al. (2016), for instance, quantify how different the processing time of Gestalt is compared to a parallel race model, where the parts are processes in isolation, using reaction times on information-processing models. Accordingly, Jäkel et al., argue that characterizing the temporal dynamics of Gestalt phenomena seems like a good way to get a more direct quantitative handle on the underlying perceptual and neural processes. Which are what Spehar and Halim (2016) have done using psychophysics and Sanguinetti, Trujillo, Schnyer, Allen, and Peterson (2016) have also done it using EEG. Overvliet and Sayim (2016), measure the influence of different contexts on haptic discrimination. They study the role of contextual modulation in the quantification of perceptual target-flanker grouping in the performance of haptic domain, that is, in haptic vernier offset discrimination. Kimchi, Yeshurun, Spehar, and Pirkner (2016) study the quantitative effects of different contexts on attentional capture.
In the same line, Hawkind, Houpt, Eidels, and Towsend (2016), made use of Systems Factorial Technology (SFT) “as a quantitative approach for formalizing and rigorously testing predictions made by local and Gestalt theories of features (p. 1, my emphasis) to test local feature of location and the emergent features of Orientation and Proximity in a pair of dots. The procedure involved measuring the response times from stimulus onset. Subjects were asked to press one button (‘no change’) if the probe dots were in the same locations as the reference squares and another button (‘change’) if the probe dots were in a different location. They finally, concluded their results, in conjunction with their modelling tools, favour the Gestalt account of emergent features. Lezama, Randall, Morel, and von Gioi (2016) also proposed an approach for the grouping of dot patterns by the good continuation law. A “quantitative measure of non-accidentalness is proposed, showing good correlation with the visibility of a curve of dots … [a] robust, unsupervised and scale invariant algorithm for the detection of good continuation of dots is derived” (p. 1, my emphasis).
Blusseau, Maiche, Morel, and von Gioi (2016), consonantly, measured the visual salience of alignments by their non-accidentalness. In their view “quantitative approaches are part of the understanding of colour integration and the Gestalt law of good continuation” (p. 1, our emphasis). They, thus present a new quantitative approach based on the ideal observer algorithm, which, in their view is able to detect non-accidental alignments. On the same line, Hock and Schöner (2016) study the grouping affinity of different surfaces in different contexts quantitatively. They suggest a non-linear dynamical system and they sketch a model to account for them.
Im, Zhong, and Halberda (2016) address the challenges of how to model human perceptual grouping in random dot arrays and how perceptual grouping affects human number estimation in these arrays. They introduce a modelling approach relying on a modified k-means clustering algorithm to formally describe human observers’ grouping behavior. Machilsen, Wagemans, and Demeyer (2016), also quantify density cues in grouping displays. They test a number of local density metrics both through their perfeormance as constrained observer models, and through a comparison with a large dataset of human detection trials.
Kimchi, Yeshurun, Spehar and Pirkner (2016), attempting to study the relation between perceptual organization and visual attention, ask the subjects to “indicate, as fast and accurately as possible, whether the upper line of the target was displaced to the right or left” (p. 36, our emphasis) (study 1); “the color of the changed element” (p. 40) (study 2); whether the upper line of the target was displaced to left or the right of the lower line by pressing one of two keys on the keybord” (study 1 and 3) (p. 36, our emphasis). Response times measurements allowed the authors to demonstrate that “when some of the elements in the display are organized by Gestalt factors into a coherent unit ¾ an object ¾ the presence of the object affects performance. Specifically responses were significantly faster when a target was irrelevant to the task at hand not predictive of the target, and was not associated with any unique transient” (p. 47, my emphasis).
Ouhnana and Kingdom (2016), using a novel method of reverse correlation, aimed to determine the properties of the binding of motion and position. “Observers were instructed to report the motion direction of the frontal plane of the target figure via a key press throughout the trial” (p. 61, our emphasis). Their claim their result suggest “that change-synchrony not common-fate underpins perceptual binding between context and target” (p. 67).
Hazenberg and van Lier (2016) on the other end, measured event-related potentials (ERPs) to study the influence of well-known objects of which the middle part was occluded. They conclude from their study that the interpretation of partly occluded shapes is not solely driven by stimulus structure, but it can also be influenced by knowledge of objects.
Erlikhman and Kellman (2016) studied how minimal formation of spatiotemporal boundary formation ¾ such as contours, shape, and global motion ¾ can produce whole forms and the nature of the computational processes involved.
Kwon, Agrawal, Li and Pizlo (2016) developed a model that aims to find the closed contour represented in the image, according to them, their “approach is practical because finding a globally-optimal solution to a shortest path problem is computationally easy. Our model was tested in four psychophysical experiments” (p. 1).
Keemink and van Rossum (2016) introduce a population model of primary visual cortex, expecting to contribute to a “unified and principled account of the good continuation law on the neural level” (p. 1).
Schmidt and Vancleef (2016) focused their study in contour integration and conceptual similarities between ladders and textures, asking whether ladders and texture processing require feedback from higher visual area while snakes are processes in a fast feedforward sweep. They tested this in a response priming paradigm, where participants responded as quickly and accurately as possible to the orientation of a diagonal contour in a Gabor array (target). They conclude that snakes, ladders, and textures do not share processing characteristics.
On a more processing predictive level, Wilder, Feldman, and Singh (2016) develop a probabilistic model of whole of whole shapes that gives rise to several distinct though interrelated measures of shape complexity. Gershman, Tenenbaum, and Jäkel (2016), likewise, describe a Bayesian theory of vector analysis and show that it can account for classic results from dot motion experiments, as well as new experimental data. Clarke, Ögmen, and Herzog (2016), on their end, offer a computational model for reference-frame with applications to motion perception.
Matin, and Li (2016) propose a multiscale dipole model, which quantifies the effect of the array of points on visually perceived eye level in terms of dipoles of various lengths that activate orientation and size specific neurons in visual cortex.
Dimiccoli (2016) presents a computational model that computes and integrates in a nonlocal fashion several configural cues for automatic figure–ground segregation. Their hypothesis is that the figural status of each pixel is a nonlocal function of several geometric shape properties and it can be estimated without explicitly relying on object boundaries. The methodology is grounded on two elements: multi-directional linear voting and nonlinear diffusion. Their results suggest that figure–ground segregation involves feedback from cells with larger receptive fields in higher visual cortical areas.
Schmidt, Spröte, and Fleming (2016) employed a dot-matching task to study in geometrical detail the effects of rigid transformations on representations of shape and space. They presented an untransformed ‘base shape’ on the left side of the screen and its transformed counterpart on the right (rotated, scaled, or both). On each trial, a dot was superimposed at a given location on the contour (Experiment 1) or within and around the shape (Experiment 2). The participant’s task was to place a dot at the corresponding location on the right side of the screen. By analysing correspondence between responses and physical transformations, they examined for object constancy, causal history, and transformation of space. They found that shape representations are remarkably robust against rotation and scaling.
The generality of the experiments in the special issue offer a quantitative model to the understanding of Gestalt phenomena, using one or another psychophysical method. The procedure ranges from button-pushes, response-times, neural correlates, to computational probabilistic processing. Awkwardly, not a single experiment addresses qualitative Gestalt relations, well evidenced by Ehrenfels, Stumpf, Mach, Meinong, and other Gestaltists. Nor the complexity of presentations, necessary for the existence of a given Gestalt quality. Rather, a simplistic approach to gestalt perception appears to be the preferred methodological strategy. It has to be acknowledged that nor times-response, nor button-pushes allow “the process of formation of the intuitive presentation directly from the indirect presentation . . . a process of change, which serves as the foundation for a specific temporal Gestalt quality” (Ehrenfels, 1890, tr. Smith, 1988, p. 104).
Reading the papers, however, one gets the impression of a gap between the theoretical commitment of the Authors (Gestalt perception) and the choice of quantitative methods to be applied to classical stimuli of psychophysics (metric quantities). One reminds of what Metzger (1971), Michotte (1991), Kanizsa (1979; 1991) (among the others) considered to be a Gestalt approach to perception, i.e. having for objects Gestalt qualitative phenomena without leaving the phenomenological domain; without, that is, referring to the underlying psychophysical or neurophysiological processes. In the original Gestalt theory the phenomena to be observed are not “representations of external stimuli”; rather they are internal presentations of active perceptual constructs, co-dependent on, but qualitatively unattainable through a mere transformation of stimuli (Mausfeld, 2010). One may observe that contemporary “Gestalt approaches” means something different, but if so, it has to be stated explicitly, or, more correctly, to name these researches with a different name.
To wit, a few of the Authors in the special issue, Jäkel, Singh, Wichmann, and Herzog seem to be aware of the absence of meaningful parameters in the current quantitative approach to Gestalt. In fact, they hope that in the future there will be mechanistic models of the dynamics of perceptual organization with meaningful parameters that can be fit to behavioural and psychophysical data simultaneously (p. 5). Jäkel and colleagues also acknowledge that “[a]ll the papers [in the special issue] try to be quantitative about Gestalt perception, but the field is certainly still far way from a common theoretical framework” (2016, p. 5).
The main point impeding the field to share a common theoretical framework, as acknowledged, however, will not be reached by adding and further clarify details within the shared approach. It should be noticed, nonetheless, that meaningful parameters are the qualitative aspects of perception, which cannot, by principle, be assessed solely by third-person quantitative measurements and mechanistic models. Certainly Gestalt phenomena can be measured and statistically analysed, but at issue is the very concept of “stimulus” in a Gestalt analysis, and the proper methods to apply as required by the original Gestaltists. The current states of affairs, as presented in the special issue, show excellent researches in psychophysics and neurophysiology, that can be at least correlated to, but not explicative of Gestalt qualitative phenomena. In this respect, the papers evidence how ill equipped vision science seem to be in what concerns conceptual tools to address Gestalt qualities.
There is a methodological problem displayed in this special issue. It is true that, as Jäke et colleagues observe, “Gestalt psychology is often criticized as lacking quantitative measurements and precise mathematical models” (p. 1). The question is what kind of measurement and of models scientists has to adopt, in identify and treat observables of a qualitative nature, rather than reducing them under the psychophysical experimental lens.
To understand the complexity of human perception, both methods, the psychophysical (quantitative), and the phenomenological (qualitative) are relevant, with various scientific achievements to their own credit. Quantitative psychophysical methods (such as reaction times, forced choice response), focus on sensory reduced stimuli, while qualitative methods (such as subjective evaluations in first person account) focus on usually complex stimuli, those to be found in our environment. However, because quantitative and qualitative objects of study differ but are related enough, a complementarity between approaches is strongly encouraged. Particularly because, as Herzogm Thunnel and Ogmen (2016), in the special issue point out, despite the insights of 100 years of Gestalt psychology, vision scientists, often too quickly assume an implicit given for granted idea of isomorphism between the world, neural circuits, and perception, which fails to explain many visual (and other perceptual) phenomena as well. The point is relevant, also in light of the recent efforts made by vision scientists in particular, in order to address the Gestalt question from a wider viewpoint (Albertazzi 2013; Wagemans 2015).
The reader is left with the impression that the basic problems of a Gestalt approach to perception (the nature of qualitative phenomena and their units of measurement, the kind of the correlation among the levels of perceived reality, the methods and the kind of models to be construed to represent them correctly) are still to be answered. It is apparent that a conceptual and methodological clarification are needed, since from a 'classic' gestaltic viewpoint the gist of the issue appears to head in the wrong direction.
Albertazzi, L. (2013). Handbook of experimental phenomenology: visual perception of shape, space and appearance. John Wiley & Sons.
Albertazzi, L. (2015). Philosophical background: phenomenology. In Wagemans, J. (2015). The Oxford handbook of perceptual organization. Oxford University Press, USA.
Blusseau, S., Carboni, A., Maiche, A., Morel, J. M., & von Gioi, R. G. (2016). Measuring the visual salience of alignments by their non-accidentalness. Vision research.
Clarke, A. M., Öğmen, H., & Herzog, M. H. (2016). A computational model for reference-frame synthesis with applications to motion perception. Vision research.
Dimiccoli, M. (2016). Figure–ground segregation: A fully nonlocal approach. Vision research.
Ehrenfels , C. von . (1890). Über Gestaltqualitäten. Vierteljharschrift für wissenschaftliche Philosophie, 14, 242 – 92.
Erlikhman, G., & Kellman, P. J. (2016). Modeling spatiotemporal boundary formation. Vision research.
Gershman, S. J., Tenenbaum, J. B., & Jäkel, F. (2016). Discovering hierarchical motion structure. Vision research.
Hawkins, R. X., Houpt, J. W., Eidels, A., & Townsend, J. T. (2016). Can two dots form a Gestalt? Measuring emergent features with the capacity coefficient. Vision research.
Hazenberg, S. J., & van Lier, R. (2016). Disentangling effects of structure and knowledge in perceiving partly occluded shapes: An ERP study. Vision research.
Herzog, M. H., Thunell, E., & Ögmen, H. (2016). Putting low-level vision into global context: Why vision cannot be reduced to basic circuits. Vision research.
Hock, H. S., & Schöner, G. (2016). Nonlinear dynamics in the perceptual grouping of connected surfaces. Vision research.
Im, H. Y., Zhong, S. H., & Halberda, J. (2016). Grouping by proximity and the visual impression of approximate number in random dot arrays. Vision research.
Jäkel, F., Singh, M., Wichmann, F. A., & Herzog, M. H. (2016). An overview of quantitative approaches in Gestalt perception. Vision Research, 126, 3-8.
Jäkel, F., Singh, M., Wichmann, F. A., & Herzog, M. H. (2016). Quantitative approaches in Gestalt perception. Vision Research.
Kanizsa , G. (1991). Vedere e pensare Bologna: Il Mulino.
Kanizsa, G. (1979). Organization in vision: Essays on Gestalt perception. Praeger Publishers.
Keemink, S. W., & van Rossum, M. C. (2016). A unified account of tilt illusions, association fields, and contour detection based on elastica. Vision research.
Kimchi, R., Yeshurun, Y., Spehar, B., & Pirkner, Y. (2016). Perceptual organization, visual attention, and objecthood. Vision research.
Kwon, T., Agrawal, K., Li, Y., & Pizlo, Z. (2016). Spatially-global integration of closed, fragmented contours by finding the shortest-path in a log-polar representation. Vision research.
Lezama, J., Randall, G., Morel, J. M., & von Gioi, R. G. (2016). Good continuation in dot patterns: A quantitative approach based on local symmetry and non-accidentalness. Vision Research.
Machilsen, B., Wagemans, J., & Demeyer, M. (2016). Quantifying density cues in grouping displays. Vision research.
Matin, L., Matin, E., & Li, W. (2016). Dipole analysis of the influence of linear arrays of points on visually perceived eye level (VPEL). Vision research.
Mausfeld R (2010) The perception of material qualities and the internal semantics of the perceptual system. In: Albertazzi L, van Tonder G, Vishwanath D (eds) Perception beyond inference. The information content of perceptual processes. MIT Press, Cambridge, pp 159–200.
Metzger, W. (1971). Ganzheit-Gestalt-Struktur. Lexikon der Psychologie, Bd, 1, 675-682.
Michotte, A. (1991). On phenomenal permanence: Facts and theories. Michotte’s experimental phenomenology of perception, 122-139.
Ouhnana, M., & Kingdom, F. A. (2016). Perceptual-binding in a rotating Necker cube: The effect of context motion and position. Vision research.
Overvliet, K. E., & Sayim, B. (2016). Perceptual grouping determines haptic contextual modulation. Vision research.
Sanguinetti, J. L., Trujillo, L. T., Schnyer, D. M., Allen, J. J., & Peterson, M. A. (2016). Increased alpha band activity indexes inhibitory competition across a border during figure assignment. Vision research.
Schmicking, D. (2010). A toolbox of phenomenological methods. In Handbook of phenomenology and cognitive science (pp. 35-55). Springer Netherlands.
Schmidt, F., & Vancleef, K. (2016). Response priming evidence for feedforward processing of snake contours but not of ladder contours and textures. Vision research.
Schmidt, F., Spröte, P., & Fleming, R. W. (2016). Perception of shape and space across rigid transformations. Vision research.
Spehar, B., & Halim, V. A. (2016). Created unequal: Temporal dynamics of modal and amodal boundary interpolation. Vision research.
Stumpf, C. (1890). Tonpsychologie, Bd 2. J. Barth, Leipzig.
Wagemans, J. (2015). The Oxford handbook of perceptual organization. Oxford University Press, USA.
Wilder, J., Feldman, J., & Singh, M. (2016). The role of shape complexity in the detection of closed contours. Vision research.
In this ingenious paper, Michael Kirchhoff and Daniel Hutto argue that the metaphysical commitments of the hard problem require a firmer foundation that denies the metaphysical distinction between the physical and the phenomenal that stimulate the neurophenomenological method of first- and third-person distinctive and complementary phenomena.
The hard problem results from a fusion between the continental tradition of phenomenology and the sciences of mind and brain. It constitutes the attempt to show how one can coherently imagine phenomenal and physical coming apart, since physical descriptions neither adequately characterize nor capture everything that can apply to phenomenal consciousness (Nagel, 1974; Jackson, 1982; Chalmers, 1996).
Kirchhoff and Hutto start of the paper by explaining in which sense neurophenomenology, as classically formulated by Varela, endorses a form of non-reductionism that assumes, rather than dissolves the hard problem, and, consonantly, cannot close the gap. Building on the radical enactive and embodied approaches to cognitive science (Hutto and Myin, 2012), they intend to make clear how the only way to deal with the hard problem is by denying of the metaphysical distinction between the physical and the phenomenal, generally endorsed by non-reductionists that include neurophenomenologists.
In Varela’s perspective (1996), the problem of consciousness cannot be articulated in a purely third-person cognitive scientific model of theory. Varela’s central problem, in agreement with traditional analytic non-reductionism, emphasizes the irreducible nature of first-person phenomenal experience, which is not exhausted by physical, third-personal description. Therefore, his enactive project endeavours to expand neuroscience to include original phenomenological investigations of experience (Thompson, 2004, p. 383). For non-reductionists, thus, the solution to the hard problem resides in the relation between phenomenal and physical, without the former collapsing into the latter. The neurophenomenological method suggests, in this direction, the promotion of a practical method that combines neuroscientific data and first-personal data, pertaining to subjects’ own conscious experience, as complementary partners in the joint pursuit of a science of consciousness. Drawing on the structure of human experience itself (Varela, 1996), it is hence possible to show how the physical and the phenomenal are integrated.
REC shares a number of theoretical commitments with neurophenomenologists alike emphasize the relational nature of consciousness, rejecting representationalism and neurocentricism. Also, both take a dynamical system theoretical approach to the study of cognitive activity as naturally fit with anti-representationalism.
What Kirchhoff and Hutto disagree with is the fact that “neurophenomenology’s non-reductionism, as reflected in formulations that fortify a schism between first- and third-person phenomena, tacitly accept the terms of the hard problem in a way that makes it impossible to close the epistemic and metaphysical gaps in the way neurophenomenology hopes to” (2016, p. 3). Thus they accentuate that even if neurophenomenology,
it keeps digging the gap between first- and third-personal metaphysics, and so it does noting to deal with the hard problem. According to Bayne (2004), the working hypothesis of neurophenomenology only reflects an epistemic principle. In demanding the equal status between first-person and third-person data, neurophenomenology promotes “an epistemic principle not an explanatory principle”, leaving us wandering in the realm of correlations without providing explanations. Furthermore, Kirchhoff and Hutto point out that reciprocal causality, denoted as bi-directional emergence, although advocating dynamical systems frameworks, presents two problems:
The method: psychophysical identity theory
Kirchhoff and Hutto hence argue for a strict identity thesis, one that acknowledges that differences in description are in fact different ways of describing one and the same physical-phenomenal activity. That is to say, phenomenal and physical data are different ways of getting at features of one and the same phenomenon.
Phenomenal experience, on an REC view is a kind of organismic activity, and as such, it can be given a physical description. They introduce the example of someone holding a book between both hands, and asked, “how are you engaging with the book?”, to which a mechanical description will suffice as a reply. However, when asked “what are you experiencing in the process?”, here the reply certainty includes expectations involved in a certain motor intentionality of active engagement. Thus, Kirchhoff and Hutto tend to agree with the psychophysical identity theory that rejects “the absolute binary distinctions that carve nature at the joints: mind/matter, inner/outer, self/world, subjective/objective, etc.” (Silberstein and Chemero, 2015, p. 186). Setting against a reductive physicalism of the sort that assumes that everything that exists can be fully and adequately described in the vocabulary of physics. The “phenomenal might just be the physical described differently ¾ under a different guise or mode of presentation” (p. 7). To sum up, REC refuses to allow the hard problem to get up running. The main arguments for that being,
The paper concludes that phenomenality should not be thought of as denoting qualitative properties of our experiences, but should be understood as the character of engaging with the world in different ways.
 For a full account of motor intentionality through the predictive engagemnt model, see Bizzari and Hipólito (forthcoming/2017); and also Predictive Brain, special issue by Synthese.
Bayne T. (2004) Closing the gap? Some questions for neurophenomenology. Phenomenology and the Cognitive Sciences 3: 349–364.
Chalmers D. (1996) The conscious mind: In search of a fundamental theory. Oxford University Press, New York.
Hutto D. D. & Myin E. (2012) Radicalizing enactivism: Basic minds without content. MIT Press, Cambridge MA.
Jackson F. (1982) Epiphenomenal qualia. Philosophical Quarterly 32: 127–136.
Nagel T. (1974) What is it like to be a bat? Philosophical Review 83: 435–450.
Silberstein M. & Chemero A. (2015) Extending neutral monism to the hard problem. Journal of Consciousness Studies 22(3–4): 181–194.
Thompson E. (2004) Life and mind: From autopoiesis to neurophenomenology. A tribute to Francisco Varela. Phenomenology and the cognitive Sciences 3(4): 381–398.
Varela F. J. (1996) Neurophenomenology: A methodological remedy for the hard problem. Journal of Consciousness Studies 3(4): 330–349.
As the happy recipient of a grant to be a visiting student here, I really enjoyed meeting everyone and the time I spent here. Go BSMB!
Never mind the gap!